MEF2D sustains activation of effector Foxp3⁺ Tregs during transplant survival and anticancer immunity

The roles played by T regulatory (Treg) cells in the maintenance of immune homeostasis are well established (1). By restricting T cell functions, Tregs restrain the onset of autoimmunity but also dampen anticancer immunity (2). The transcription factor Foxp3 is required but not sufficient to sustain the immunosuppressive properties of Tregs (3). Pioneer factors (4–6) pave the way for Foxp3 expression in concert with a large number of epigenetic regulators that modulate its activities (7–13).

Besides the assembly and regulation of large multiprotein complexes containing Foxp3, the execution and the sustainment of the Treg program requires the activation of other transcription factors that act as genetic modulators (14). The action of these transcription factors may be supportive to Foxp3, like Thpok (15), or may be independent and additive or subtractive to Foxp3 (14). In particular, the coexpression of master transcription factors along with Foxp3 marks some Treg subsets, e.g., colonic Tregs coexpressing RORγt and Foxp3 (16, 17), or follicular Tregs (Tfr) coexpressing Bcl6 and Foxp3 (18, 19). In these various Treg subsets, the sum of the genetic responses triggered by the simultaneous activation of multiple master factors lets them acquire functional nuances more suited to their location (20). In other cases, in response to certain stimuli the activation of transcription factors such as Blimp1 and Irf4 can synergize with Foxp3 and trigger a reinforced Treg program that characterizes so-called effector Tregs (eTregs) (21).

MEF2D belongs to the myocyte enhancer factor-2 (MEF2) family of transcription factors (22). The weak intrinsic transcriptional activity of MEF2D can be potentiated by the binding of coactivators, like p300 (23), or completely blocked or even converted into transcriptional repression by the actions of corepressors, like class IIa histone/protein deacetylases (Hdacs) (24). The great plasticity of MEF2 transcription factors allows the precise modulation of differentiative and adaptive responses (22). Moreover, their binding to tissue-specific cofactors allows MEF2 transcription factors to play key roles in supervising the transcriptional program that underlies heterogeneous processes such as muscle differentiation, ossification, cardiac hypertrophy, synaptogenesis, and lymphoid commitment (22).

MEF2D is known to be highly expressed in T cells (25). In activated T cells, MEF2D sustains the production of IL-2 (26, 27), IL-4, and IFN-γ (28), as a consequence of its release from class IIa Hdacs and Cabin1 inhibition (26–28). In the thymus, MEF2D regulates the negative selection of double-positive T cells by promoting the transcription of Nurr77 in an Hdac7-dependent manner (29). Moreover, the inhibition or the knockout of the MEF2 repressor Hdac9 increases the suppressive properties of Treg cells (30), at least in part by sustaining the metabolic switch to OXPHOS that is characteristic of these CD4⁺Foxp3⁺ cells (31). Conversely, the expression of a gain-of-function mutant of Hdac7 decreases the proportion of Tregs (32).

In this study, we analyzed the still unexplored roles of MEF2D in CD4⁺Foxp3⁺ Treg cells. We focused on MEF2D because it was by far the most highly expressed member of the MEF2 family of paralogs in both human and murine Tregs. We found that MEF2D interacts with Foxp3 and modulates its activities, but also exerts Foxp3-independent functions required for Treg fitness. In addition, depletion of MEF2D in Treg cells dampens the Foxp3 program, especially in supraphysiological conditions that stimulate inflammation, such as transplantation and cancer.

Foxp3 controls the expression and alternative splicing of MEF2D. Human and murine CD4⁺Foxp3⁺ Tregs express high levels of MEF2D (Supplemental Figure 1, A and B, respectively; supplemental material available online with this article; https://doi.org/10.1172/JCI135486DS1), and Foxp3 binds to the Mef2d promoter in both human and murine Tregs (Supplemental Figure 1, C and D). Comparison of MEF2D mRNA and protein levels in T effector (Teff) and Treg cells from C57BL/6 mice shows approximately 2-fold more Mef2d in Treg cells (Figure 1, A and B). We noticed that the genomic region of Mef2d encoding the 2 alternative third exons (3α1 and 3α2, Figure 1C) has chromatin features (H3K27ac and H3K36me3) typical of their active splicing in human Tregs but not in Teff cells (Supplemental Figure 1E). Moreover, the α2 isoform of MEF2D is reported to be expressed not only in muscle cells and neurons, but also in the spleen (33). We confirmed that the slight increase in MEF2D levels observed in freshly isolated Tregs with respect to Teffs is at least in part due to the expression of the α2 isoform (Figure 1, D and E). This increase was stronger after 24 hours of stimulation with anti-CD3/anti-CD28 antibodies and IL-2 (Figure 1, D and E). In these conditions, a slight increase in the α1 transcript can also be observed (Figure 1D), probably as a consequence of higher transcriptional activity on the Mef2d promoter. However, differently from the α2 isoform, the protein levels of the α1 isoform were not increased after stimulation, likely as a consequence of the increased instability of this isoform that selectively undergoes proteasomal degradation (34). Foxp3 is necessary to trigger the expression of the α2 isoform since suppression of Foxp3 led to significant decreases in the mRNA and protein levels of the α2 isoform, while the α1 isoform was minimally affected (Figure 1, G–I). Moreover, prolonged stimulation for 60 hours with anti-CD3/anti-CD28 mAbs and IL-2 stimulated the transcription of both the α1 and α2 isoforms of MEF2D, though with different magnitudes (2 and 10 times, respectively) (Figure 1F). While the α1 isoform is reported to be widespread though all the tissues of the body, the α2 isoform has a restricted pattern of expression and influences differentiative (35, 36) and adaptative (37, 38) responses. Moreover, compared with the α1 isoform, the α2 isoform encodes a stronger transcription factor since it lacks the phosphorylation sites of PKA (36) and CDK4/6 (34) that mediate the inhibition or proteasomal degradation of MEF2D. We conclude that, in contrast to Teffs, Tregs are characterized by an increase in MEF2D levels and by the Foxp3-dependent expression of the α2 isoform of MEF2D.

Figure 1

Foxp3 controls Mef2d transcription and influences MEF2D protein-protein interactions. (A) qPCR results in Treg and Teff cells from WT mice (n = 4, t test). (B) Immunoblots of the indicated proteins in Treg and Teff (Te) cells from WT mice. (C) Diagram of phosphorylation sites retained or lost in MEF2Dα1 and -α2 isoforms. (D) qPCR in freshly isolated WT Teffs and Tregs or cells cultured under activating conditions for 24 hours (1:1 ratio anti-CD3/anti-CD28 mAb–coated beads); n = 6, Tukey’s multiple-comparison test. (E) Immunoblots of the indicated proteins in the same cells as described in D. (F) Densitometric analysis relative to E; n = 2, t test relative to Teffs. (G) qPCR results of Mef2d (pan, α1, α2) expression in freshly isolated WT Teffs and Tregs or cultured under activating conditions for 60 hours in the presence of 5 μM scrambled or Foxp3 antisense (as) oligonucleotide; n = 3, Tukey’s multiple-comparison test. (H) Immunoblots of the indicated proteins in the same cells as described in G. (I) Densitometric analysis relative to H. n = 2, t test relative to scrambled. (J) HEK293T cells were transfected with 1 μg each of tagged constructs encoding Foxp3, MEF2D, and p300; lysates were pulled down with anti-Myc or IgG Ab (1 μg). The membrane was probed with biotinylated anti-Foxp3 mAb, anti-HA for p300, and anti-Myc for MEF2D. (K) HEK293T cells were transfected with 1 μg each of FLAG-Foxp3 and the depicted deletion mutants of MEF2D-GFP. Lysates were pulled down with anti-GFP Ab (1 μg). (L) Lysates from freshly isolated or 24-hour anti-CD3/anti-CD28–stimulated Teffs and Tregs were pulled down with anti-MEF2D or IgG Ab (1 μg). (M) Lysates from freshly isolated Tregs were pulled down with anti-MEF2Dα1 or anti-MEF2Dα2 or IgG Ab (1 μg). (N) HEK293T cells were transfected with 1 μg each of tagged constructs encoding Foxp3, MEF2Dα1, MEF2Dα2, and Hdac9. Lysates were pulled down with anti-MEF2Dα1, -Mα2, or IgG Ab (1 μg). (O) Lysates from freshly isolated Tregs were pulled down with anti-Foxp3 or IgG Abs (1 μg); 1/50 input included in all the IPs. Asterisks in panels K, L, and O indicate heavy chains of IgG Abs used in pull-down experiments. *P < 0.05; **P < 0.01; ***P < 0.005.

Foxp3 influences the assembly of MEF2D complexes. Foxp3 forms multiprotein complexes containing different transcription factors (7, 39). To test whether MEF2D can be part of such complexes, we transfected HEK293T cells with the α1 isoform of MEF2D, p300, and Foxp3. As shown in Figure 1J, MEF2D was stabilized by the coexpression of p300, and it interacted with both p300 and Foxp3, while the coexpression of Foxp3 and p300 did not perturb their binding to MEF2D (Figure 1J). Foxp3 did not bind to the DNA binding domain of MEF2D, as the deletion of this region did not abolish their interaction (Figure 1K). Moreover, the immunoprecipitation (IP) of MEF2D led to co-IP of Foxp3 in both freshly isolated and stimulated Tregs (Figure 1L). MEF2D bound strongly to p300 both in Teffs and in Tregs (resting or activated), while its binding to Hdac3 and Hdac9 was decreased in Tregs compared with Teffs, and respectively severely impaired or abolished after stimulation (Figure 1L). This could also be due to the decreased protein levels of Hdac3 and Hdac9 observed after stimulation (Figure 1L). We hypothesized that the different stoichiometry of MEF2D complexes between Tregs and Teffs could be due to the increased α2 isoform observed in Foxp3⁺ cells. This was the case, because in freshly purified Tregs only the α1 isoform of MEF2D interacted strongly with Hdac9, whereas Foxp3 preferentially interacted with the α2 isoform (Figure 1M). To further assess these interactions, we transfected HEK293T cells with the 2 isoforms of MEF2D plus Foxp3 or Hdac9, and we pulled down MEF2D with isoform-specific Abs (Figure 1N). We found that the α2 isoform of MEF2D interacted strongly with Foxp3 and weakly with Hdac9 compared with the α1 isoform (Figure 1N). In reciprocal studies, IP of Foxp3 led to co-IP of MEF2D (Figure 1O). We conclude that in Tregs, different complexes assemble on MEF2D isoforms in response to Foxp3 expression and exogenous stimulation.

Characterization of Tregs deficient for MEF2D. Two different complexes assemble on MEF2D in Tregs. Only one, mainly associated with the α1 isoform, contains strong repressors like Hdac9. We hypothesized that both of these complexes are required to maintain Treg immune suppressive properties; the first, the repressive one, might be required to switch off some MEF2D genetic programs, while the second one might sustain and promote the expression of Treg-specific genes. To prove this concept, we conditionally deleted Mef2d in Foxp3⁺ Treg cells by crossing Mef2d^fl/fl mice (40) and Foxp3^YFP-Cre mice. The KO allele encodes for a protein with an in-frame deletion of the second exon that is involved in DNA binding and dimerization, leading to the complete loss of function of the protein (40, 41). Mef2d^fl/fl Foxp3^YFP-Cre (hereafter Mef2d^–/–) mice were born at expected Mendelian ratios and appeared to develop normally over 1 year of monitoring. However, tissues from Mef2d^–/– mice were characterized by the presence of mild mononuclear cell infiltrates in the lung and liver upon histologic examination, while they did not show any defects in colon, skin, pancreas, thymus, or other tissues (Supplemental Figure 2). Gene deletion was assessed by purifying Treg cells from WT and Mef2d^–/– mice and confirming MEF2D deletion by immunoblot (Figure 2A). Moreover, as expected, the depletion of a functional MEF2D protein impaired the expression of the MEF2D target gene, Hdac9 (Figure 2A). We further characterized the Mef2d^–/– mice by harvesting secondary lymphoid tissues and assessing their cell populations by flow cytometry. Mef2d^–/– mice were characterized by increased proportions of CD4⁺Foxp3⁺ Tregs within lymph nodes, spleen, and thymus (Figure 2, B and C), and by increased proportions of splenic CD4⁺CD69⁺ cells (Figure 2, D and E). Although no significant alterations were observed in CD8⁺ subpopulations (data not shown), increased proportions of CD4⁺Foxp3^– effector memory cells (CD44^hiCD62L^lo) were observed within lymph nodes, spleen, and thymus (Figure 2, F and G). Nevertheless, the in vitro suppressive functions of Tregs from Mef2d^–/– mice were not significantly impaired compared with WT Tregs in standard Treg assays (Figure 2, H and I), though they did display modestly impaired suppression of CD4⁺ and CD8⁺ cells when using bulk WT splenocytes as responder cells (Figure 2J). Altogether, these features are indicative of a modest inflammatory state in Mef2d^–/– mice but one that is insufficient to provoke significant autoimmunity under basal conditions.

Figure 2

Generation and characterization of mice with conditionally deleted Mef2d in Foxp3⁺ Treg cells. (A) Immunoblots of MEF2Dα1 and -α2 and Hdac9 in freshly isolated Teffs and Tregs obtained from WT and Mef2d^–/– mice, as indicated; β-actin served as loading control. (B and C) Analysis of CD4⁺Foxp3⁺ cells in lymphoid tissues from Mef2d^–/– or WT mice (n = 3); the experiment was repeated 2 times with 3 mice/group with similar results. *P < 0.05, **P < 0.01 by t test. LN, lymph nodes; SP, spleen, THY, thymus. (D–G) Analysis of CD4⁺CD69⁺ and CD4⁺CD44^hiCD62L^lo cells in CD4⁺YFP^– populations in lymphoid tissues from Mef2d^–/– or WT mice (n = 3); the experiment was repeated 2 times with 3 mice/group with similar results. *P < 0.05, **P < 0.01 by t test. (H and I) In vitro Treg suppression assay using pooled (4 mice/group) Tregs from lymph nodes and spleens of Mef2d^–/– or WT mice and WT T cells; the percentage of WT proliferating cells is shown in each panel. The experiment was run in triplicate and repeated at least 3 times, and the results of one representative experiment are shown. (J) AUC ratios (KO/WT, 1-sample t test with theoretical mean = 1) of 3 Treg suppression assays performed on CFSE-labeled WT splenocytes as responder cells. The suppressive effects on CD4⁺ and CD8⁺ responder cells were evaluated after CD4 and CD8 immunostaining.

MEF2D-specific and Foxp3-specific signatures are altered in Mef2d^–/– Tregs. The transcriptomes of WT and Mef2d^–/– Tregs were analyzed by RNA sequencing (RNA-seq). Mef2d deletion caused the upregulation of 795 transcripts and the repression of 700 genes (>1.3-fold, P < 0.05) (Figure 3A). Gene set enrichment analysis (GSEA) showed that 4 main signaling pathways were altered in Mef2d^–/– Tregs: (i) DNA repair (Figure 3B and Supplemental Figure 3A), (ii) E2F targets/proliferation (Figure 3C and Supplemental Figure 3B), (iii) PPAR/RA signaling (Figure 3D and Supplemental Figure 3C), and (iv) Treg signature (Figure 3, E and F, and Supplemental Figure 3, D and E). The first 3 were severely impaired in Mef2d^–/– Tregs and belong to the general core responses regulated by MEF2 transcription factors (22). Analysis of the Treg signature (42) showed that key genes induced in Treg versus Teff cells were indeed repressed in Mef2d^–/– Tregs (Figure 3E), and conversely, transcripts normally repressed in Tregs were upregulated in Mef2d^–/– Tregs (Figure 3F). We validated by qPCR (Figure 3H) and immunoblot (Figure 4A) changes in expression the main hubs involved in the identified altered pathways. We confirmed by different methods an impairment in the proliferation of Mef2d^–/– Tregs (Figure 4, B–D and Supplemental Figure 4, A–C) that appears at least partially due to a prolonged quiescence rather than senescence (Figure 4C), even if a moderate impairment in cellular fitness cannot be excluded.

Figure 3

Mef2d deletion affects general and Foxp3-specific responses. (A) A volcano plot showing statistical significance (P value) versus fold change for genes differentially expressed as a result of Mef2d deletion in Foxp3⁺ Treg cells. (B–F) GSEA plots obtained by using the indicated gene sets and the RNA-seq data obtained from the comparison between Mef2d^–/– or WT mice Tregs as data set. (G) qPCR results of the expression of the indicated genes in freshly isolated Mef2d^–/– or WT Teffs (TE) and Tregs or cultured under activating conditions for 24 hours (1:1 ratio of anti-CD3/anti-CD28 mAb–coated beads); n = 5. *P < 0.05, **P < 0.01, ***P < 0.005 by t test between Teffs and Tregs in unstimulated and stimulated groups.

Figure 4

Mef2d deletion perturbs the fitness of Tregs. (A) Immunoblots of the expression of the indicated proteins representing key nodes in the identified perturbed pathways in freshly isolated Teffs and Tregs obtained from Mef2d^–/– or WT mice, as indicated. β-Actin was used as loading control. (B and D) Analysis of CD4⁺Ki67⁺ in CD4⁺Foxp3⁺ populations in lymphoid tissues from Mef2d^–/– or WT mice. n = 3. **P < 0.01 by t test. (C) Time course analysis of Ki67 positivity in CD4⁺Foxp3⁺ populations purified from Mef2d^–/– or WT mice and kept in culture for 3 days with anti-CD3/anti-CD28 mAb–coated beads; n = 4. *P < 0.05, ***P < 0.005 by t test between the 2 groups for each time point.

Mef2d^–/– Tregs lose the hallmarks of eTregs. Foxp3 was expressed in Mef2d^–/– Tregs at levels comparable to those of control mice (Figure 4A). However, Mef2d^–/– Tregs showed reduced levels of 3 genes (Ctla4, Icos, and Il10) that encode proteins centrally involved in the immunosuppressive functions of Tregs (Figure 3G and Figure 4A). High levels of these proteins are a feature of so-called highly suppressive eTregs (43) that are dependent on Blimp1 (21) and Irf4 (44). GSEA performed on the transcriptome of Mef2d^–/– Tregs confirmed their inverse correlation with Blimp1⁺ Tregs (Figure 5, A and B); in particular, key genes induced in Blimp1⁺ cells were significantly repressed in Mef2d^–/– Tregs (Figure 5C). We validated by qPCR (Figure 3G) and flow cytometry (Figure 5, D–F and Supplemental Figure 4F) the downregulation in Mef2d^–/– Tregs of CTLA4, Icos, and IL-10; the latter is especially impressive in stimulated cells (Figure 5G). As a consequence, Teff cells were more active in Mef2d^–/– mice (Figure 2, F and G).

Figure 5

MEF2D is required for the acquisition by Tregs with highly suppressive properties. (A and B) GSEA plots obtained by using the indicated gene sets and the RNA-seq data obtained from the comparison between Mef2d^–/– or WT Tregs as data set. (C) Heatmap of the expression levels (Z scores) of 10 core genes upregulated in Blimp1⁺ Mef2d^–/– or WT Tregs. (D–G) Analysis of CD4⁺Foxp3⁺CTLA4^hi (D), CD4⁺Foxp3⁺Icos^hi (E), and CD4⁺Foxp3⁺IL-10⁺ (F and G) populations in lymphoid tissues from Mef2d^–/– or WT mice, freshly isolated (D and F) or stimulated with PMA/ionomycin for 4 hours (G); the experiment was repeated 2 times, with 3 mice/group, with similar results. *P < 0.05, **P < 0.01 by t test between WT and KO samples. (H) Immunoblots of Blimp1 and Foxp3 in freshly isolated or 24-hour anti-CD3/anti-CD28–stimulated Tregs obtained from Mef2d^–/– or WT mice, as indicated; β-actin was used as loading control. (I and J) Representative Treg suppression assay and pooled data (3 mice/group) using Tregs from lymph nodes and spleens of Mef2d^–/– or WT mice along with the percentage of proliferating cells in each panel; 48 hours before assay, Mef2d^–/– or WT Tregs were isolated and stimulated with 1:1 anti-CD3/anti-CD28–coated beads and 10 U/mL IL-2. *P < 0.05, **P < 0.01 by t test between the 2 groups for each condition.

Blimp1 expression was not altered in Mef2d^–/– cells (Figure 5H), and Mef2d expression was not altered in Blimp1^–/– Tregs (GSE84827, Supplemental Figure 5A) (45). Moreover, Blimp1 does not interact with MEF2D, but does pull down its well-known interactor, Hdac2 (46) (Supplemental Figure 5B). However, Hdac9 is repressed in Blimp1⁺ Tregs (GSE103456, Supplemental Figure 5C) (47) and is induced in Blimp1^–/– Tregs (GSE84827, Supplemental Figure 5A) (45). This suggests that in Blimp1⁺ Tregs, MEF2D is released from its inhibitors. Indeed, many Blimp1 target genes are bound by MEF2 at their proximal promoters (Table 1). We hypothesized that MEF2D acts downstream of Blimp1 and is required to sustain the high repressive properties of eTregs. As expected, the suppressive properties of Foxp3⁺ Mef2d^–/– Tregs were only partially enhanced after their preactivation with anti-CD3/anti-CD28, differently from WT (Figure 5, I and J). This defect may be partially due to a moderately increased susceptibility of Blimp1^–/– Tregs to apoptosis, as seen in vitro (Supplemental Figure 5, D and E). These features, collectively, are typically associated with an impairment of eTreg functions (48). A similar trend was also seen in Tregs, expanded in vivo by IL-2, in Treg assays (Supplemental Figure 5, F–I).

Table 1

Top 10 transcription factors (TFs) binding Blimp1-regulated genes

Conditional Treg deletion of Mef2d affects Tfr and Tfh cells and B cell maturation in germinal centers. MEF2D acts upstream of CTLA4 and downstream of Blimp1 in regulating Treg suppressive properties. As CTLA4 (18) and Blimp1 (19) are key regulators of Tfr functions and B cell responses, we investigated whether Mef2d deletion in Foxp3⁺ cells would affect Tfr and Tfh responses. Similarly to CTLA4 and Blimp1, Mef2d KO showed increased Tfr and Tfh cells (Figure 6, A and B). While the total numbers of B cells, memory B cells, and plasma cells were not altered (Figure 6C), increased numbers of B1 (Figure 6D), follicular type I B (Figure 6E), transitional T3 B (Figure 6H), and germinal center (GC) (Figure 6I) B cells were observed in Mef2d^–/– mice. The rise in follicular type I B cells and the corresponding drop in marginal zone precursors (Figure 6, F and G) observed after MEF2D depletion suggest an enhancement of B cell receptor (BCR) signals (49). The expansion of the GC and the alteration of BCR signaling indicate increased GC reactions. Consistent with that, moderate levels of autoantibodies were found in Mef2d^–/– mice (Figure 6J and Supplemental Figure 5J).

Figure 6

Mef2d deletion dampens Tfr functions partially affecting the Tfh-mediated regulation of B cell maturation. (A and B) Analysis of Tfr (CD4⁺CXCR5⁺PD-1⁺Foxp3⁺) and Tfh (CD4⁺CXCR5⁺PD-1⁺Foxp3^–) populations in lymphoid tissues from Mef2d^–/– or WT mice. n = 5. **P < 0.01 by t test between WT and KO samples. (C) Analysis of B cell (CD19⁺), memory B cell (CD19⁺B220⁺CD62^lo/–FAS^–GL7^–CD138^–), and plasma cell (CD138⁺IgM^–) populations in lymphoid tissues from Mef2d^–/– or WT mice. n = 5; t test between WT and KO samples. (D–I) Analysis of B cell subpopulations in lymphoid tissues from Mef2d^–/– or WT mice. (D) B1 cells (B220^–CD19⁺); (E) follicular B cells (CD19⁺B220⁺CD93^–CD21^loCD23^hi), follicular B type I cells (CD19⁺B220⁺CD93^–CD21^loCD23^hiIgM^lo), follicular B type II cells CD19⁺B220⁺CD93^–CD21^loCD23^hiIgM^hi); (F) marginal zone (MZ) precursor B cells (CD19⁺B220⁺CD93^–CD21^hiCD23^lo) and MZ B cells (CD19⁺B220⁺CD93^–CD21^hiCD23^hi); (H) transitional type 1 (T1) B cells (CD19⁺B220⁺CD93⁺IgM⁺⁺CD23^–), T2 B cells (CD19⁺B220⁺CD93⁺IgM⁺⁺CD23⁺), and T3 B cells (CD19⁺B220⁺CD93⁺IgM^–CD23⁺); (I) GC B cells (CD19⁺B220⁺GL7⁺FAS⁺) and GL7⁺ activated B cells (CD19⁺B220⁺GL7⁺). n = 5. *P < 0.05, **P < 0.01, ***P < 0.001 by t test between WT and KO samples. (J) Autoantibodies detected in the sera of 3 WT and 3 Mef2d^–/– mice, using indirect immunofluorescence.

Mef2d deletion dampens Treg function in vivo. As we predicted that Tregs require MEF2D to acquire the features of highly suppressive eTregs, we established 3 animal models to assess the effects of Mef2d deletion on Tregs in vivo.

First, we tested the strength of Mef2d^–/– Tregs to inhibit homeostatic proliferation of CD4⁺ and CD8⁺ T cells over 28 days following their adoptive transfer into Rag1^–/– mice (30). In comparison with WT Tregs, Mef2d^–/– Tregs were severely impaired in their capability to suppress the proliferation of CD4⁺ Teffs in the lymph nodes (Figure 7A). In addition, upon activation they produced more IFN-γ (Figure 7B and Supplemental Figure 6A) and IL-2 (Figure 7C and Supplemental Figure 6B). No differences were observed in the spleen (Figure 7, A and F and Supplemental Figure 6E), even though Mef2d^–/– Tregs were less activated in both spleen and lymph nodes (Figure 7D and Supplemental Figure 6C). Increased proportions of Mef2d^–/– Tregs were found among the splenocytes but not in lymph nodes (Figure 7E and Supplemental Figure 6D) and this was not due to altered expression of chemokine receptors (Supplemental Figure 6F) or changes in proliferation (Supplemental Figure 4, D and E). Finally, Mef2d^–/– Tregs were found to be even more strongly impaired in their ability to repress CD8⁺ T cell proliferation and/or survival (Figure 7, G and H).

Figure 7

Mef2d^–/– Treg functions are severely impaired in vivo. (A) Twenty-eight days after the adoptive transfer of 0.25 × 10⁶ WT Thy1.1⁺ Teffs and 1.0 × 10⁶ WT or Mef2d^–/– Thy1.2⁺ Tregs into Rag1^–/– mice, splenocytes and lymphocytes were harvested and the total number of Thy1.1⁺ cells determined by flow cytometry; n = 4. **P < 0.01 by t test. (B, C, and F) Analysis of CD4⁺Thy1.1⁺IFN-γ⁺ (B) or IL-2⁺ (C) or Ki67⁺ (F) populations in lymphoid tissues from the same mice as described in A stimulated with PMA/ionomycin for 4 hours; n = 4. *P < 0.05 by t test. (D and E) Analysis of CD4⁺YFP⁺ (D) and CD4⁺YFP⁺CD44^hiCD62L^lo Treg populations (E) in lymphoid tissues from the same mice as described in A. *P < 0.05 by t test. (G and H) Twenty-eight days after the adoptive transfer of 0.25 × 10⁶ WT Thy1.1⁺ Teffs and 1.0 × 10⁶ WT or Mef2d^–/– Thy1.2⁺ Tregs into Rag1^–/– mice, lymphocytes (G) and splenocytes (H) were harvested and the total number of Thy1.1⁺ cells determined by flow cytometry; n = 4. *P < 0.05, **P < 0.01 by Tukey’s multiple-comparison test. (I) In contrast to WT recipients, mice with conditional deletion of Mef2d in their Foxp3⁺ Treg cells acutely rejected cardiac allografts (BALB/c→C57BL/6) despite costimulation blockade with anti-CD154 mAb/donor splenocyte transfusion (DST); n = 5/group. P < 0.01 by Tukey’s multiple-comparison test. (J and K) qPCR results of the expression of the indicated genes in samples collected 2 weeks after cardiac allografting in WT and Mef2d^–/– mice. n = 12 and 9, respectively, for WT and Mef2d^–/–. *P < 0.05, **P < 0.01 by t test.

Second, we performed cardiac allografting using BALB/c donors and WT or Mef2d^–/– C57BL/6 recipients in conjunction with costimulation blockade with anti-CD40L (CD154) mAb plus donor splenocyte transfusion (49). Costimulation blockade induced long-term allograft survival (>100 days) in WT recipients but not in mice with conditional deletion of Mef2d within their Treg cells (P < 0.01) (Figure 7I). The acute rejection observed in Mef2d^–/– mice correlated with CD8⁺ T cell activation and intragraft expression of IFN-γ and granzyme B (GZMB) (Figure 7J) and an impairment of Tregs (Figure 7K). Histologic examination of allografts in Mef2d^–/– recipients collected just before end-stage rejection at 16 days after transplant (post-Tx) showed marked differences from allografts harvested from WT recipients at the same post-Tx interval (Supplemental Figure 7). The latter showed well-preserved myocardium and vessels and small numbers of mononuclear cells palisading within capillaries and occasional interstitial areas. In contrast, allografts in Mef2d^–/– mice showed widespread myocyte necrosis and vascular injury with fibro-intimal proliferation, mixed mononuclear and polymorphonuclear infiltrates, and interstitial hemorrhages. Rejecting allografts also showed vascular deposition of C4d (Supplemental Figure 8). Collectively, these findings in Mef2d^–/– recipients are consistent with increased host cellular and humoral alloresponses.

Third, as shown next, we studied the effects of deletion of Mef2d in Tregs on host antitumor immunity.

Mef2d deletion promotes antitumor immunity. We used 3 models involving tumor growth in syngeneic mice; 2 lung cancer cell lines (TC1 and AE.17) were injected subcutaneously, and a hepatocellular carcinoma (HCC) cell line was delivered by injection of the mesenteric vein.

The growth of TC1 cells was profoundly impaired in Mef2d^–/– mice; in 9 of 10 mice the tumors were almost completely cleared (Figure 8A), in conjunction with activation (Figure 8, B and C) and IFN-γ production by CD4⁺ (Figure 8, D and E) and CD8⁺ (Figure 8, F–I) T cells in the tumor-draining lymph nodes. Compared with WT tumor-bearing mice, Mef2d^–/– mice had increased activation of CD4⁺ Teffs (CD44^hiCD62L^lo; Figure 8, B and C) that was inversely correlated with the impairment in Treg functions (IL-10 production, CTLA4 and Ki67 expression; Supplemental Figure 9, A–F). Moreover, Tregs harvested from the tumor-draining lymph nodes showed evidence of transcriptome alterations (Supplemental Figure 9G) consistent with that of Tregs in non–tumor-bearing Mef2d^–/– mice (Figure 3G). The infiltration of Tregs in TC1 tumors was assessed after 8 days, so as to precede shrinking of the tumors. Mef2d^–/– Tregs infiltrated the tumor as well as WT Tregs (Figure 8K). However, tumor-infiltrating CD4⁺ and CD8⁺ T cells were more numerous and more activated in Mef2d^–/– mice (Figure 8, K and L, and Supplemental Figure 9H), likely as a consequence of impaired Treg functions (Supplemental Figure 9, A–F). The increased numbers of HPV/TC1-specific tetramer-positive CD8⁺ T cells in Mef2d^–/– mice suggests that the activation of CD8⁺ T cells was tumor specific (Figure 8J). Similarly to TC1, the growth of AE.17 tumors was impaired in Mef2d^–/– mice (Figure 8M), and again the proportions of activated and IFN-γ–producing CD8⁺ (Figure 8N and Supplemental Figure 9I) and CD4⁺ (Supplemental Figure 9, J and K) T cells were increased within the tumor-draining lymph nodes.

Figure 8

Mef2d deletion promotes anticancer immunity in mice bearing subcutaneous lung cancers. (A) Graphs representing the tumor growth (volume) in 10 WT and 10 Mef2d^–/– mice during a period of 20 days after the subcutaneous injection of 1.2 × 10⁶ TC1 cells. The experiment was repeated twice with similar results (t test between the 2 groups for each time point). (B and C) Analysis of the activation status of CD4⁺Foxp3^– populations in the same samples as A. (D–I) Analysis of CD4⁺Foxp3^–IFN-γ⁺ (D and E), CD8⁺IFN-γ⁺ (F and G), and CD8⁺GZMB⁺ (H and I) populations in single-cell suspensions obtained from the draining lymph nodes and spleens harvested from 3 representative mice injected as in A and stimulated for 4 hours with PMA/ionomycin. n = 3–5, t test. (J) CD8 HPV-TC1–specific tetramer staining (E7) positivity in spleens (SP) and tumor-draining lymph nodes (TDLN) of WT and Mef2d^–/– mice analyzed 10 and 20 days after TC1 cell injection. ***P < 0.001 by t test. (K and L) Analysis of tumor-infiltrating CD4⁺Foxp3⁺ (K) or CD4⁺IFN-γ⁺ (K and L) or CD8⁺IFN-γ⁺ (K and L) populations in tumors harvested from 4 mice per group 8 days after the injection of TC1 cells as in A. n = 4, t test. (M) Graphs representing the tumor growth (volume) in 10 WT and 10 Mef2d^–/– mice during a period of 21 days after the subcutaneous injection of 2.0 × 10⁶ AE.17 cells, t test. (N) Analysis of CD8⁺IFN-γ⁺ populations in single-cell suspensions obtained from the draining lymph nodes and spleens harvested from mice injected with AE.17 cells (n = 10, t test). *P < 0.05, **P < 0.01, ***P < 0.005.

Finally, the injection of HCC cells via the mesenteric vein, after 9 days gave rise to histologically confirmed liver tumors in 7 of 10 WT mice but in only 2 of 10 Mef2d^–/– littermates (Figure 9A). As the tumors differed markedly in the number and dimensions of the lesions, we used α-fetoprotein (AFP) levels in the serum as a read-out of tumor growth (50). In C57BL/6 mice, levels of serum AFP greater than 200 ng/mL correlated with HCC growth (50). Serum AFP levels in WT mice were significantly higher than in Mef2d^–/– mice or in C57BL/6 mice not injected with HCC cells (median values ± SE: C57BL/6 healthy, 60.6 ± 8.1; WT HCC, 204.5 ± 21.5; Mef2d^–/– HCC, 122.3 ± 17.9) (Figure 9B). In Mef2d^–/– mice, conventional CD4⁺ cells were more active (CD44^hiCD62L^lo and CD69⁺, Figure 9, C and D; and Supplemental Figure 10, A and B), produced more IFN-γ (Figure 9E and Supplemental Figure 10C), and were more proliferative (Figure 9F and Supplemental Figure 10D). Conversely, the Treg population was less proliferative (Figure 9G and Supplemental Figure 10E) and was characterized by lower levels of CTLA4 (Figure 9H and Supplemental Figure 10F) and IL-10 (Figure 9I and Supplemental Figure 10G). The alterations in the transcriptome (Figure 9J) and in the proteome (Figure 9K) typical of Mef2d^–/– Tregs were maintained or became even greater in HCC-bearing mice. No significant differences were observed in the CD8⁺ population (Supplemental Figure 10, H and I) or in the production of IL-2 by CD4⁺ cells (Supplemental Figure 10J). Comparable data illustrating the marked inhibitory effects of Mef2d deletion in Tregs on HCC tumor growth were seen on day 21 after injection (Supplemental Figure 11).

Figure 9

Mef2d deletion promotes anticancer immunity in mice bearing HCC. Shown is the analysis on day 9 after tumor cell injections, and additional data at 21 days after injection are shown in Supplemental Figure 11. (A) H&E-stained sections of WT and Mef2d^–/– livers 9 days after the injection of 0.3 × 10⁶ HCC cells into the mesenteric vein. Scale bar: 100 μm. Livers from WT show dense tumors, whereas livers from Mef2d^–/– mice are tumor free. (B) Dot plot representing the AFP serum levels of 3 healthy C57BL/6 mice and 10 WT and 10 Mef2d^–/– mice on day 9 after HCC injection; Tukey’s multiple-comparison test. (C–F) Analysis of the activation status of CD4⁺Foxp3^– and CD8⁺ populations in terms of CD44^hiCD62L^lo (C), CD69 (D), and Ki67 positivity (F) and production of IFN-γ (E) in freshly isolated (C, D, and F) or PMA/ionomycin-stimulated (E) cells obtained from lymphoid tissues of HCC-injected mice. n = 5, t test. (G–I) Analysis of the expression of Ki67 (G), CTLA4 (H), and IL-10 (I) in CD4⁺Foxp3⁺ populations in freshly isolated (G) or PMA/ionomycin-stimulated (H and I) single-cell suspensions obtained from the same samples used for C. n = 5, t test. (J) qPCR results of the expression of the indicated genes in YFP⁺ Tregs obtained from the draining lymph nodes of WT and Mef2d^–/– HCC-injected mice; n = 3–5, t test. (K) Immunoblots of Ki67, Crabp2, Fabp5, and CTLA4 in the same cells described in Figure 9J; β-actin served as loading control. *P < 0.05; **P < 0.01; ***P < 0.005.

MEF2D assembles repressive and activating complexes to maintain Treg immunosuppressive properties. To gain more insight into the mechanisms elicited by MEF2D in sustaining Treg suppressive properties, we performed ChIP using freshly isolated WT and Mef2d^–/– Tregs. We mapped the binding of MEF2D with 3 different Abs (specific for α1, α2, and a pan-Ab to IP both isoforms) on the more relevant genomic loci identified as being under the control of MEF2D. We mapped in the same loci Foxp3 and p300, while the comparison of H3K27ac levels between WT and Mef2d^–/– Tregs provided a read-out of the positive or negative transcriptional activities of the investigated complexes. In the case of Ctla4/Icos we investigated 7 regions that bear MEF2 binding sites and/or were previously pulled down by Foxp3 in ChIP experiments (Supplemental Figure 12A, Figure 10, A–G). Foxp3 significantly bound the proximal promoter (region A) and an intronic region (region D) of Ctla4 (Figure 10, A and D). We did not find significant coassociation of MEF2D in these chromatin positions, while both MEF2Dα1 and MEF2Dα2 precipitated the intronic region C (Figure 10C). A decrease in H3K27ac levels in Mef2d^–/– Tregs was observed in the proximal promoter (region A) and in the intronic region C (Figure 10, A and C); in these positions a significant drop in p300 binding was observed (Figure 10, A and C). MEF2D (pan and MEF2D α2 isoform) and Foxp3 were coassociated at the Icos promoter (region G); this region was characterized by decreased H3K27ac in Mef2d^–/– cells (Figure 10G), while a drop in p300 binding was also observed at the distal promoter (region F; Figure 10F). MEF2D was previously described as a positive regulator of Il10 transcription in T cells (51, 52) and in glial cells (53), through direct binding to the proximal promoter (53). We confirmed binding of MEF2D (α1, α2, and pan) to the proximal promoter of Il10, in association with p300 but in the absence of Foxp3 (Figure 10H). In Mef2d^–/– Tregs, decreased levels of p300 binding and H3K27ac were observed (Figure 10H). MEF2D was associated with the Nr4a1 promoter (Figure 10I), as observed in other contexts (29, 54–57); the same region is bound by Foxp3 (Figure 10I) and p300. In this locus, MEF2D appeared to form a repressive complex, as the KO had increased p300 binding and H3K27ac (Figure 10I). MEF2D was found to be significantly associated with the Hdac9 promoter (Figure 10J), used as a positive control. Finally, MEF2D also repressed the transcription of Itk, without perturbing p300 binding but reducing levels of H3K27ac at the promoter level, probably by promoting the engagement of Foxp3 (Figure 10K). Among the genes bound and regulated by MEF2D in Tregs, some (Hdac9 and Il10) were also under MEF2D control in Teffs (Figure 10L and Supplemental Figure 12, B and C), as well as in non-lymphoid organs and cell types (24, 34, 38, 52, 54, 56). Others, like Icos and Ctla4, were Treg restricted, probably because MEF2D binding requires the establishment of a permissive open chromatin status (Figure 10L and Supplemental Figure 12, B and C), primed by Foxp3.

Figure 10

MEF2D works synergistically, additively, and independently from Foxp3 in regulating the expression of the bound loci. (A–K) Histogram representing the qPCR results obtained in freshly isolated WT and Mef2d^–/– Tregs after ChIP with the indicated Abs. For each investigated genomic locus, a red square indicates the position of the amplified region with respect to the leading TSS and the exons of the associated canonical mRNA isoform (n = 2–5). Black asterisks indicate the significance of the indicated comparisons or with respect to IgG, and red asterisks the significance of the comparison of the same IP between WT and Mef2d^–/– Tregs. *P < 0.05; **P < 0.01; ***P < 0.005 by t test and Tukey’s multiple-comparison test. (L) Heatmap representing the strength of MEF2D and H3K27ac signals for each genomic locus in WT and Mef2d^–/– Tregs with respect to WT Teffs.

In summary, MEF2D established both repressive and activating complexes on relevant loci for Treg homeostasis and immune-suppressive functions. In some cases, its binding was directly assisted by Foxp3 (Icos promoter region G and Nr4a1); in a second group of loci (Ctla4 region C) MEF2D assisted the transcription proficiency stimulated by Foxp3, while in others the MEF2D transcriptional effect was Foxp3 independent (Il10 and Hdac9). The activating or repressive functions of MEF2D depend on the corepressors or coactivators recruited to chromatin. We observed perfect correlation between H3K27ac, p300, and the transcriptional activity in all the examined loci, except for Itk, which appeared to be regulated through an indirect mechanism.

MEF2 transcription factors regulate differentiative and adaptive/stress-related responses (22). In mammals, MEF2D is the most ubiquitously expressed of the 4 paralogs, being abundant in almost all tissues and cell types (22). This wide expression pattern reflects the key roles of MEF2D in regulating general responses involved in maintaining cellular fitness, including the control of cell-cycle progression, mitochondrial activity, and the balance of prosurvival and apoptotic responses (22, 41). However, MEF2D also has tissue- and cell-type specific functions, like the promotion of muscle differentiation, remodeling, regeneration, and the control of neuronal development and synaptogenesis (22, 41). Moreover, MEF2D controls the development of particular cell types, like retina photoreceptors (58). Three mechanisms allow MEF2D to exert context-dependent biological functions: (i) the plasticity of MEF2D-containing complexes and the coexistence in the same cell of MEF2 transcriptional activator and repressor complexes (38), (ii) the ability to bind chromatin away from its traditional consensus sites thanks to MEF2 binding to lineage-specific transcription factors (58), and (iii) the expression of cell-specific MEF2 isoforms that display differential affinity for cofactors (33, 36).

We found that all 3 of these mechanisms are operative in Tregs. The expression of Foxp3 triggers the alternative splicing mechanism that gives rise to the α2 isoform of MEF2D. This isoform is refractory to the binding to class IIa Hdacs and Hdac9, in particular, and has higher affinity for Foxp3. The α1 isoform retains a reduced capability to interact with Foxp3 and a stronger affinity for Hdacs. The copresence of the 2 isoforms sustains the expression of Treg-specific genes, maintaining the balance between the establishment of activating and repressive transcriptional complexes. We observed that in Tregs, MEF2D and Foxp3 can act synergistically (on Icos and Nr4a1 loci), additively (on Ctla4 and Il10), or independently (on Itk and Hdac9). A fine regulation of this balance ensures the maintenance of the highly suppressive properties of Tregs. In particular, Mef2d^–/– Tregs fail to generate eTreg cells. MEF2D appears to be downstream of Blimp1, the main regulator of IL-10⁺ eTregs (21), as the core of Blimp1-regulated genes (47) show evidence of MEF2 binding at their promoters. Moreover, as Blimp1 directly controls only 8% of the eTreg signature (59), we can now identify MEF2D among the transcription factors involved in sustaining the eTreg phenotype.

In fact, while Mef2d^–/– Tregs were only slightly impaired in their suppressive functions in vitro, they were severely impaired in repressing the proliferation of CD4⁺ and CD8⁺ T cells in vivo. Moreover, the activation of CD4⁺ and CD8⁺ T cells after the depletion of Mef2d in Tregs induced allograft rejection and promoted antitumor immunity, as seen in 3 different cancer models (HCC and lung) known for being Treg dependent in vivo (60–62). Interestingly, in HCC and in non–small cell lung carcinoma, as in various tumors, MEF2D expression is increased and reported to be required to sustain tumor growth (63–65). The upregulation of PD-L1 frequently observed in HCC has been correlated to the impairment of CD8-mediated cytotoxicity (66) and could explain the inefficient activation of CD8⁺ lymphocytes observed in this model in Mef2d^–/– mice. As MEF2D was found to control the expression of PD-L1 (67), in these cancers a targeted therapy to inhibit MEF2D functions could have a double benefit of decreasing the malignancy of tumor cells and increasing the immunoreactivity of T cells. Regarding this, small molecules that target the interaction of MEF2 with class IIa Hdacs have been reported (65, 66).

The lack of signs of severe inflammation or autoimmunity in mice depleted of Mef2d in Foxp3⁺ Tregs is in line with the effects of deletion of Blimp1 (45), Icos (67), and Il10 (68). In the latter, no autoimmunity but mild colitis was reported (68), while Prdm1^–/– mice display marks of autoimmunity only in aging mice (59). Moreover, similarly to Mef2d, the ablation of Blimp1, Icos, Il10, or Ctla4 in Tregs does not impact their suppressive properties in vitro, but impairs their activation under stimulating conditions (48) (Figure 5, I and J) and their activities in vivo (45, 67–70), similarly to our Mef2d^–/– mice. MEF2D functions in Tregs are not limited to the maintenance of a suppressive phenotype, as the proliferation of Mef2d^–/– Tregs was also impaired, similarly to Icos^–/– Tregs (71). Moreover, there is indirect evidence of the involvement of MEF2D in the regulation of Treg metabolism (31). Finally, and again similarly to CTLA4 (18) and Blimp1 (19), the depletion of MEF2D in Foxp3⁺ cells affects Tfr functions and impacts the normal maturation of B lymphocytes and the development of GC reactions. As a result of these alterations (50), Mef2d^–/– mice accumulate small quantities of autoantibodies, though not sufficient to induce significant autoimmunity. It is currently unknown whether the unbalanced B cell homeostasis observed in Mef2d^–/– mice may actively contribute to anticancer immunity. Further studies are required to clarify this point.

There are no reports in the literature regarding the association of autoimmunity and MEF2D copy-number loss or genetic deletion, probably because of MEF2D pleiotropy. However, a rare SNP that abrogates the expression of the α2 isoform of MEF2D was recently associated with systemic lupus erythematosus (72), and SNPs associated with multiple sclerosis at the level of enhancer and superenhancers were predicted to alter the binding of MEF2 transcription factors (73). Further investigations are required to unveil a likely direct role played by MEF2 transcription factors in these diseases, given our discoveries in Tregs. Finally, our work underlines the importance played by MEF2D in the regulation of anticancer immunity, offering an alternative interpretation of the conflicting evidence about the contribution of MEF2D to oncogenesis (41).

Mice. We used BALB/c, Rag1^–/– C57BL/6, and CD90.1/B6 mice from The Jackson Laboratory, plus previously described Foxp3^YFP-Cre mice (68) and Mef2d^fl/fl mice (40) that were backcrossed on the C57BL/6 background at least 8 times and used at 6–8 weeks of age, unless specified otherwise. Foxp3^YFP-Cre mice were used as WT controls.

Plasmids and transfections. We purchased Mef2d-FLAG-MYC plasmids from Origene and FLAG-tagged Foxp3 from Addgene. MEF2D-GFP and MEF2D deletion mutants were previously described (38). Hdac9-FLAG was obtained from Ed Seto (George Washington University, Washington, DC) (74). MEF2Dα1 and MEF2Dα2 were obtained from F. Jeffrey Dilworth (Ottawa Hospital Research Institute, Ottawa, Canada) (36). HEK293T cells were grown in DMEM plus 10% FBS and transfected with Lipofectamine 2000 (Invitrogen, 11668027).

Co-IP and Western blotting. HEK293T, Teff, and Treg cells were lysed with a hypotonic buffer (20 mM Tris-HCl, pH 7.5; 2 mM EDTA; 10 mM MgCl₂; 10 mM KCl; and 1% Triton X-100) supplemented with protease inhibitors. For each immunoprecipitation, 1 μg of Ab was used. Ab-antigen complexes were collected with Protein G agarose (Invitrogen, 15920-010). Cell lysates were resolved by SDS-PAGE, transferred to nitrocellulose membranes, and immunoblotted with the following Abs: MEF2D (BD, 610774), Foxp3 (Invitrogen, 700914; eBioscience, 14-4774-82, 13-5773-80), p300 (Invitrogen, PA1-848), FLAG (Cell Signaling Technology [CST], 14793), Myc (CST, 2272), HA (CST, 2367), phospho-histone H2A.X (Ser139; CST, 9718), histone H2A.X (CST, 7631), p53 mAb (DO-1; CST, 18032), β-actin (CST, 3700), Blimp1 (eBioscience, 6D3, 14-5963-82), Hdac3 (Abcam, ab7030), Crabp2 (Santa Cruz Biotechnology, sc-166897), Fabp5 (Santa Cruz Biotechnology, sc-365236), Lmnb1 (Santa Cruz Biotechnology, sc-377000), Ki67 (Abcam, ab15580), p53ac (K120) (Abcam, 78316), and MEF2D (α1 and α2) (36). Secondary HRP-conjugated Abs against mouse (catalog 7076), rat (catalog 7077), and rabbit (catalog 7074) IgG were purchased from CST. Unconjugated anti-CD3 (clone 145-2C11, 553057) and anti-CD28 (clone 37.51, 553294) mAbs used for cell activation were purchased from BD.

Cryopreservation. Single-cell suspensions in CryoStor CS5 cryopreservation media (MilliporeSigma) reagent were frozen at –80°C, stored in liquid nitrogen, and then thawed and transferred quickly into greater than 10 volumes of warm tissue culture medium (DMEM, Thermo Fisher Scientific) for use. Cell number and viability were assessed with a Nexcelom Cellometer Auto2000 and AOPI Staining solution in PBS.

Flow cytometry. Single-cell suspensions from lymph nodes, spleens, or tumors were prepared as previously described (75) and were stained with fluorochrome-conjugated mAbs from BD Biosciences, unless specified otherwise, that were directed against CD4 (Pacific Blue, Invitrogen, MHCD0428), CD8 (Super Bright 645, eBioscience, clone 53-6.7, 64-0081-82), Foxp3 (eFluor 450, eBioscience, clone FJK-16s, 48-5773-82 and PE-Cy5, 15-5773-82), CD62L (PE-Cy7, clone MEL-14, 25-0621-82), IFN-γ (APC, clone XMG1.2, 554413; PE 554412), CD44 (PE-Cyanine5, eBioscience, clone IM7, 15-0441-83), CTLA4 (APC, 17-1522-82), Icos (PE, 12-9949-81), Ki67 (APC700, 561277), IL-2 (PE, eBioscience, clone JES6-5H4, 12-7021-41), CD25 (APC, eBioscience, clone PC61.5, 17-0251-82), CD8a (FITC, 53-6.7), CD4 (FITC, H129.19), NK1.1 (FITC, M1/70, BioLegend), CD11b (FITC, M1/70, BioLegend), CD19 (BV605, 6D5, BioLegend), CD45R/B220 (BV785, RA3-6B2, BioLegend), IgM (PerCP-Cy5.5, II-41, Invitrogen), CD62L (PE, MEL-14, BioLegend), CD95 (PE-CF594, Jo2), GL7 (APC, GL7, BioLegend), CD138 (APC-R700, 281-2), CD93 (PE/Cy7, AA4.1, BioLegend), CD23 (APC Cy7, B3B4, BioLegend), CD21/35 (PB, 7E9, BioLegend), CD19 (FITC, 6D5, BioLegend), CD4 (PE-CF594, RM4-5), CXCR5 (APC, L138D7, BioLegend), PD-1 (PE/Cy7, 29F.1A12, BioLegend), and Foxp3 (PB, FJK-16s, Invitrogen) and acquired on a Cytoflex (Beckman Coulter) flow cytometer.

Treg suppression assays. For in vitro studies, 5 × 10⁴ CD4⁺CD25^–YFP^– T cells and CD4⁺CD25⁺YFP⁺ Tregs from Foxp3^YFP-Cre and Mef2d^–/– mice isolated using a CD4⁺CD25⁺ Treg isolation kit (Miltenyi Biotec, 130-091-041) or cell sorted were added to 96-well plates. Equal numbers of CFSE-labeled CD4⁺CD25^– T cells and γ-irradiated antigen-presenting cells, isolated using a CD90.1 kit (Miltenyi Biotec, 130-049-101), plus anti-CD3 mAb (1 μg/mL), were cultured for 72 hours. After 72 hours, proliferation of Teff cells was determined by flow and analysis of CellTrace Violet (Thermo Fisher Scientific) dilution. The same assay was performed by using Tregs isolated from mice treated with IL-2/anti–IL-2 complexes (76), or by preactivating Tregs for 2 days with anti-CD3/anti-CD28 beads (1:1) and IL-2 (10 U/mL) (48). For Treg assay performed by using WT splenocytes as responder cells, WT splenocytes were CFSE labeled and cultured in the presence of WT and Mef2d^–/– Tregs for 3 days under CD3 stimulation (anti-CD3 mAb, 1 μg/mL).

For in vivo homeostatic proliferation assay, 0.2 × 10⁶ CD4⁺CD25^–Thy1.1⁺ or 0.2 × 10⁶ CD8⁺Thy1.1⁺ and 1.0 × 10⁶ Tregs were injected via the tail vein into Rag1^–/– mice. After 1 or 4 weeks, lymph nodes and spleens were harvested and stained with Thy1.1-PE and CD4–Pacific Blue or CD8–Super Bright 645, and the numbers of Thy1.1⁺ T cells determined using a Cytoflex flow cytometer.

Cardiac transplantation. Heterotopic cardiac allografts were performed using BALB/c donors and WT or Mef2d^–/– recipients on the C57BL/6 background (77). At the time of engraftment, recipients were treated i.v. with anti-CD154 mAb (200 μg) plus 5 × 10⁶ donor splenocytes (49). Allograft survival was monitored by palpation of ventricular contractions and confirmed by histological evaluation.

Detection of autoantibodies. Sera of 3 WT and Mef2d^–/– mice were tested for autoantibodies by indirect immunofluorescence. Nonfixed 4- to 5-μm cryosections of WT tissues were incubated for 1 hour at room temperature with serial dilutions (from 1:1 up to 1:80) of sera, washed in PBS, and incubated for 30 minutes at room temperature with FITC-labeled goat anti-mouse secondary Ab (Jackson Immunoresearch Lab, 115-545-062; diluted in PBS, 1:200). After washings in PBS, slides were mounted and evaluated by specialist in autoimmune clinical diagnostics in human autoimmune diseases. Precharacterized sera from mice with known autoantibodies of corresponding specificity were used as positive controls, while pooled sera from WT B6 mice and secondary Abs only were used as negative controls. All positive sera were analyzed a second time to evaluate individual titers.

Immunoperoxidase. Formalin-fixed, paraffin-embedded sections of allografts harvested at 14 days post-Tx were stained by immunoperoxidase with rabbit Abs directed against CD3 (A0452, Agilent), C4d (12-5000, American Research Products), or rabbit IgG control Ab (I-1000, Vector Labs) (49).

ChIP assays. Teffs or Tregs were fixed for 15 minutes with 1% formaldehyde and fragmented by sonication. Chromatin was immunoprecipitated as previously described (38) using Abs against H3K27ac (2 μg, Abcam, ab4729), MEF2D (5 μg, BD, 610774), MEF2Dα1 and MEF2Dα2 (36) (5 μg), p300 (3 μg, Invitrogen, PA1-848), and Foxp3 (10 μg, Invitrogen, PA1-806). The immunoprecipitated DNA was purified and analyzed by qPCR (SYBR Green, Applied Biosystems).

RNA-seq and real-time qPCR. RNA was isolated using RNeasy kits (QIAGEN), and RNA integrity and quantity were analyzed by NanoDrop ND-1000 and Nanochip 2100 Bioanalyzer (Agilent Technologies). Library preparation, RNA-seq, genome mapping, and analysis were performed by Novogene on the Illumina Platform PE150; data were deposited in the NCBI’s Gene Expression Omnibus (GEO) database (accession GSE139480). The edgeR package (https://bioconductor.org/packages/release/bioc/html/edgeR.html) was used to identify the differentially expressed genes (P value < 0.05 and fold change > 1.3). GSEA (78) was performed to identify altered signaling pathways, as previously described (38). Expression of individual genes was verified by qPCR. RNA was reverse transcribed to cDNA (Applied Biosystems) and RT-qPCR was performed using TaqMan primer and probe sets; data were normalized to endogenous 18S rRNA, and relative expression was determined by the formula 2^–ΔCT.

Cell lines and tumor models. The murine lung adenocarcinoma cells, TC1 (79), were provided by Yvonne Paterson (University of Pennsylvania). The murine AE17.ova mesothelioma cell line (80) was provided by Delia Nelson, University of Western Australia, Perth, Australia. The murine HCC cell line, Dt81-Hepa1-6, was developed by Marc Bilodeau (Centre Hospitalier de l’Université de Montréal, Montreal, Canada) (81). Lung cancer cells were grown in RPMI, and HCC cells were grown in DMEM/F-12 supplemented with 10% FBS, 2 mM glutamine, and 5 μg/mL penicillin/streptomycin. For lung tumor cells, each mouse was shaved on their right flank and injected s.c. with 1.2 × 10⁶ TC1 or 2 × 10⁶ AE17 tumor cells, whereas 0.3 × 10⁶ HCC cells were injected directly into the mesenteric vein. Tumor volume was determined by the formula V = (W² × L)/2. Serum AFP was measured by using a mouse alpha-Fetoprotein/AFP Quantikine ELISA Kit (MAFP00, R&D Systems).

Statistics. Data were analyzed using GraphPad Prism 8.0. Data are presented as mean ± SD unless specified otherwise. Measurements between 2 groups were done with a 2-tailed Student’s t test if data were normally distributed or unpaired Mann-Whitney U test when the populations were not normally distributed. Comparison of multiple samples was performed by 1-way ANOVA with corresponding Tukey’s multiple-comparison test. Graft survival was evaluated with Kaplan-Meier followed by log-rank test. AUC for suppression assays was calculated as previously described (82); AUC for KO samples was expressed as ratio to WT, then ratios of 3 experiments were combined and tested for significance (1-sample t test with theoretical mean = 1). P less than 0.05 was considered significant: *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 primers used are listed in Table 2.

Table 2

Primers used in this study