Phosphorylation-facilitated sumoylation of MEF2C negatively regulates its transcriptional activity

Background Sumoylation has emerged as an important posttranslational regulatory mechanism for transcription factors and cofactors. Sumoylation of many transcription factors represses their transcriptional activities. The myocyte enhancer factor 2 (MEF2) family of transcription factors plays an important role in regulating gene expression during myogenesis and has been recently shown to be sumoylated. Results Consistent with earlier reports, we show that sumoylation of MEF2C at K391 inhibits its transcriptional activity. Sumoylation of MEF2C does not block its DNA-binding activity. A small C-terminal fragment of MEF2C containing K391, referred to as delta-N2-MEF2C, is efficiently sumoylated and, when targeted to DNA, represses transcription at neighbouring promoters. Because delta-N2-MEF2C lacks the binding site for class II histone deacetylases (HDACs), this result suggests that sumoylation of MEF2C may help to recruit transcriptional repressors other than these HDACs. Intriguingly, we show that phosphorylation of S396 in MEF2C, a residue in close proximity to the major sumoylation site (K391) and known to be phosphorylated in vivo, enhances sumoylation of delta- N2-MEF2C in vitro. The S396A mutation reduces sumoylation of MEF2C in vivo and enhances the transcription activity of MEF2C in reporter assays. Conclusion We propose that phosphorylation of MEF2C at S396 facilitates its sumoylation at K391, which in turn recruits yet unidentified co-repressors to inhibit transcription. Our studies further suggest that sumoylation motifs containing a phosphorylated serine or an acidic residue at the +5 position might be more efficiently sumoylated.


Background
Transcription factors and cofactors orchestrate complex yet precise programs of gene expression that are critical for cell proliferation and differentiation during development. The myocyte enhancer factor 2 (MEF2) family of transcription factors regulates diverse cellular processes in a wide range of cell types [1]. In muscle cells, the MEF2 fam-ily of proteins binds to the promoters of many musclespecific genes and activates their transcription during muscle differentiation [2]. There are four MEF2 proteins, MEF2A, -B, -C, and -D, in vertebrates [1]. They contain a MADS box and an adjacent MEF2 motif at their N-termini that mediate DNA binding, homo-and hetero-dimerization, and cofactor binding [1,3,4]. They also contain a transcriptional activation domain at their C-termini [2]. Homozygous mef2c null mice are embryonic lethal due to defective cardiac myogenesis and morphogenesis [5,6] whereas mef2a null mice die suddenly shortly after birth due to defects in post-natal cardiomyocytes [7]. Therefore, among other functions, MEF2 proteins play essential and distinct roles during cardiac muscle development. In addition, mutations of human MEF2A have been implicated in the pathogenesis of a familial coronary artery disease [8].
The functions of MEF2 proteins are regulated by their direct physical interactions with a large collection of cofactors, including thyroid hormone receptor, MyoD, NFAT, GATA, p300, GRIP1, histone deacetylases (HDACs), and MITR [9][10][11][12][13][14][15][16]. The class II HDACs (HDAC4, -5, -7, and -9) contain a characteristic MEF2-interacting domain at their N-termini [3,4] and an HDAC catalytic domain at their C-termini. They bind to the MADS/MEF2 domain of MEF2 proteins and repress their transcriptional activity [13]. In addition to cofactor binding, MEF2 proteins are also regulated by multiple phosphorylation events [1]. Several kinases, including p38 MAP kinase, ERK5, and CDK5, can phosphorylate MEF2 proteins and regulate their transcriptional activity [17][18][19]. Phosphorylation of MEF2 by p38 occurs in the transcriptional activation domain of MEF2 and stimulates its transcriptional activity [19]. Recently, it has been shown that an alternative splicing event within the last coding exon of MEF2C leads to the formation of two MEF2C isoforms, only one of which contains a 32-residue domain called γ [20]. The MEF2C-(γ-) isoform has higher transcriptional activity than does MEF2C-(γ+), and a Gal4-γ-domain fusion protein represses the basal level transcription of a promoter with Gal4-binding sites [20]. Interestingly, Ser 396 within the γ-domain is phosphorylated and the S396A mutation diminishes the transcriptional repression activity of the γdomain [20]. These results suggest that the γ-domain might recruit transcriptional repressors in a phosphorylation-dependent manner. Because a motif similar to the γdomain is constitutively present in MEF2A, -B, and -D, phosphorylation within this motif might negatively regulate the transcriptional activities of other MEF2 proteins.
We have previously identified human MEF2 proteins as SUMO1 substrates using an in vitro expression cloning (IVEC) strategy [38]. In addition, MEF2C and MEF2D have very recently been shown to be modified by SUMO2 and SUMO3 [39]. Sumoylation of MEF2 is enhanced by class IIa HDACs and by SIRT1-mediated deacetylation of the lysine acceptor for SUMO [39,40]. Here we further confirm the sumoylation of MEF2C in vivo and identify K391 as the major sumoylation site of MEF2C. As compared to the wild-type MEF2C, the MEF2C-K391R mutant has higher transcriptional activity in reporter assays, and is more efficient in promoting the conversion of 10T1/2 cells into myocytes when co-expressed with MyoD. These findings suggest that sumoylation of MEF2C inhibits its transcriptional activity. Sumoylation of MEF2C does not appear to affect its binding to DNA, its interactions with HDACs, or its phosphorylation by p38. Interestingly, K391 is located in the γ-domain and in close proximity to S396, which can be phosphorylated by unknown kinase(s) [20]. When fused to Gal4, a C-terminal fragment of MEF2C containing the γ-domain is sufficient to repress transcription of a promoter that contained Gal4-binding sites. Mutations of either K391 or S396 abolish the transcriptional repression activity of the Gal4-MEF2C fusion protein. Moreover, phosphorylation of S396 enhances the sumoylation of the C-terminal domain of MEF2C in vitro. The S396A mutation reduces the degree of MEF2C sumoylation in vivo. We propose that phosphorylation of S396 facilitates the sumoylation of MEF2C, which in turn recruits yet unidentified transcriptional repressors to inhibit transcription.

MEF2C is sumoylated at K391 in vivo
We and Gregoire et al. had previously shown that the MEF2 family of transcription factors was efficiently sumoylated [38,39]. As shown in Figure 1A, MEF2C was modified in the presence of E1 (AOS1/UBA2), E2 (UBC9), and SUMO1. His 6 -SUMO1 resulted in a further shift in gel mobility as compared to untagged SUMO1 ( Figure 1A). The yeast SUMO isopeptidase, Ulp1, effi-ciently reversed this modification ( Figure 1A). MEF2C contains a sumoylation motif, ΦKXE (Φ, hydrophobic residues; X, any residues) [41,42] in the γ-domain of MEF2C [20] that is highly conserved among other members of the MEF2 family ( Figure 1B). We mutated the lysine residue in this motif (K391) to arginine and observed the MEF2C-K391R is no longer sumoylated in vitro ( Figure 1C). We then examined the sumoylation of MEF2C in HeLa and NIH3T3 cells ( Figure 1D and data not shown). A signifi-MEF2 proteins are sumoylated The results of two independent experiments were averaged with the standard deviation indicated. P value was calculated using student t test. Cell lysates were resolved by SDS-PAGE and blotted with anti-MEF2C or Myc.
1F), but not that of Myc-MEF2C-K391R. These data indicate that MEF2C is sumoylated in vivo and K391 is the major sumoylation site of MEF2C in living cells.

Sumoylation of MEF2C reduces its transcriptional activity
We next used a luciferase reporter assay to examine whether sumoylation of MEF2C regulates its transcriptional activity. The luciferase reporter construct contained three tandem copies of MEF2-binding sites at the promoter region. The transcriptional activity of MEF2C-K391R (the sumoylation-deficient mutant) was about two-fold higher than that of the wild-type MEF2C ( Figure  2A), suggesting that sumoylation of MEF2C inhibits its transcription activity. Overexpression of GFP-SUMO1 downregulates the transcription activity of both MEF2C-WT and K391R ( Figure 2A). Because overexpression of GFP-SUMO1 caused a global increase in the sumoylation of many cellular proteins (data not shown), inhibition of MEF2C-K391R by GFP-SUMO1 overexpression was most likely due to the enhanced sumoylation of other MEF2C regulatory proteins under these conditions. However, we cannot completely rule out the possibility that MEF2C is sumoylated at a second site, the sumoylation of which is below the detection limit of our assay.
We next examined whether sumoylation affected the activation of endogenous MEF2C by the p38MAPK pathway. To do so, we transfected a constitutively active mutant of MKK6 (an upstream activator of p38MAPK) into C2C12 cell lines, together with GFP-SUMO1 or SENP2. Overexpression of GFP-SUMO1 reduced the transcriptional activity of MEF2C whereas overexpression of SENP2 enhanced the activity of MEF2C in the presence of constitutively active MKK6 ( Figure 2B). These results are consistent with the notion that sumoylation of MEF2C inhibits the transcription activity of the endogenous MEF2C stimulated by the p38MAPK pathway. However, it is entirely possible that alteration of the sumoylation levels of other MEF2C regulatory proteins is responsible for the observed effects of GFP-SUMO and SENP2.
Embryonic fibroblast cells can be converted into myoblasts upon overexpression of MyoD and MEF2 [13,43]. We tested whether MEF2C-K391R was more active than MEF2C-WT in collaborating with MyoD to promote the conversion of 10T1/2 cells into myoblasts. We transfected MyoD-and MEF2C-expressing plasmids into 10T1/2 cells and cultured these cells in low-serum media to induce the differentiation of the converted myoblasts into myotubes. On the fifth day after the induction of differentiation, the cells were fixed and stained with an antibody against myosin heavy chain (MHC), a well-established myogenic differentiation marker ( Figure 2C and 2D). Co-expression of MEF2C-WT together with MyoD slightly increased the number of myotubes, as compared to the expression of Sumoylation of MEF2C inhibits its transcriptional activity Figure 3 Sumoylation of MEF2C inhibits its transcriptional activity. (A) Gal4, Gal4-MEF2C-WT, or Gal4-MEF2C-K391R construct was co-transfected with GAL4×5-luciferase reporter and pRL-tk reporter into HeLa cells. Firefly luciferase activities were measured and normalized for transfection efficiency by using Renilla luciferase activities. (B) Gal4-MEF2C-WT or Gal4-MEF2C-K391R construct was co-transfected with GAL4×5-luciferase reporter, pRL-tk reporter, and SENP2, DN-UBC9-Flag, or vector construct into HeLa cells. Firefly luciferase activities were measured and normalized for transfection efficiency by using Renilla luciferase activities and then divided by the luciferase activities of Gal4-MEF2C-WT or Gal4-MEF2C-K391R, respectively, to show the fold differences. (C) Gal4, Gal4-MEF2C-WT, or Gal4-MEF2C-K391R construct was co-transfected with LexA×8-GAL4×5-luciferase reporter, pRL-tk reporter, and LexA-VP16 construct into HeLa cells. Firefly luciferase activities were measured and normalized for transfection efficiency by using Renilla luciferase activities.
MyoD alone ( Figure 2D). Co-expression of MEF2C-K391R with MyoD increased the myoblast conversion rate by two-fold (P < 0.05) ( Figure 2D). The expression levels of MEF2C-WT and MEF2C-K391R were similar ( Figure  2D, left panel). These results suggest that sumoylation of MEF2C down-regulates its transcriptional activity during muscle differentiation.

Sumoylation of MEF2C recruits transcriptional repressors other than Class II HDACs
We next sought to determine the mechanism by which sumoylation inhibits the transcriptional activity of MEF2C. We first tested whether sumoylation reduced the DNA-binding activity of MEF2C using the chromatin immunoprecipitation (ChIP) assay. Similar amounts of MEF2C-WT or MEF2C-K391R bound to the desmin promoter in a reporter plasmid [see Additional file 1], suggesting that sumoylation does not affect the DNA-binding activity of MEF2C. We next tested whether sumoylation of MEF2C increases its binding to HDACs, such as HDAC4 and HDAC5. MEF2C-WT and MEF2C-K391R bound equally well to HDAC4 or HDAC5 [see Additional file 1], suggesting that sumoylation of MEF2C does not recruit class II HDACs.
A small C-terminal fragment of MEF2C is sufficient to repress transcription in a sumoylation-dependent manner Since sumoylation inhibits the activation of the endogenous MEF2C by the p38MAPK pathway ( Figure 2B), we thus tested whether sumoylation and p38 phosphorylation of MEF2C are antagonistic. However, we observed that sumoylation does not affect the binding affinity of MEF2C toward p38 or phosphorylation of MEF2C by p38 [see Additional file 1]. Likewise, phosphorylation of MEF2C by p38 does not appear to block sumoylation of MEF2C in vivo [see Additional file 1]. Thus, there is no significant interplay between sumoylation and p38 phosphorylation of MEF2C.
To further study how sumoylation reduced the transcriptional activity of MEF2C, we measured the transcriptional activity of Gal4 fusion proteins of MEF2C and MEF2C-K391R, using luciferase reporter assays with a reporter construct that contained Gal4-binding sites. As compared to Gal4-MEF2C-WT, Gal4-MEF2C-K391R was much more active in stimulating transcription ( Figure 3A). Overexpression of DN-UBC9 or SENP2 greatly increased the transcriptional activity of Gal4-MEF2C-WT, but not Gal4-MEF2C-K391R ( Figure 3B). These data indicate that sumoylation also inhibits the transcriptional activity of MEF2C at a promoter that does not contain MEF2C-binding sites, consistent with the fact that sumoylation does not affect the DNA-binding activity of MEF2C.
We next performed luciferase reporter assays with a reporter construct that contained both Gal4-and LexAbinding sites. LexA-VP16 (a fusion protein of the LexA DNA-binding domain and the VP16 transactivation domain) dramatically stimulated the transcription of this reporter gene ( Figure 3C). Co-expression of Gal4-MEF2C-WT, but not Gal4-MEF2C-K391R, greatly reduced the transcriptional activity of LexA-VP16 ( Figure 3C). This suggests that Gal4-MEF2C might recruit transcriptional repressors to this artificial promoter in a manner that is dependent on its sumoylation.
Since sumoylation of MEF2C does not appear to affect its binding to HDAC4 and HDAC5, recruitment of these HDACs is unlikely to be responsible for the transcriptional repression activity of Gal4-MEF2C. To further test this notion, we constructed Gal4-fusion proteins of two C-terminal MEF2C fragments, termed ∆N1 and ∆N2, and their corresponding K391R mutants, and tested their abilities to stimulate/repress transcription in reporter assays ( Figure 4A). MEF2C-∆N1 lacked the N-terminal MADS/ MEF2 domain that is required for HDAC-binding whereas ∆N2 lacked both the MADS/MEF2 domain and the transactivation domain ( Figure 4A). As shown in Figure 4B, Gal4-MEF2C-∆N1-K391R was much more active in promoting transcription of the reporter gene that contained Phosphorylation of MEF2C at S396 and sumoylation of MEF2C at K391 repress transcription through the same pathway Gal4-binding sites at its promoter, suggesting that MEF2C-∆N1 retained the ability to be regulated by SUMO. Furthermore, Gal4-MEF2C-∆N2, but not Gal4-MEF2C-∆N2-K391R repressed the activity of LexA-VP16 to stimulate transcription of the reporter gene whose promoter contained both Gal4-and LexA-binding sites (Figure 4C). Expectedly, both GAL4-MEF2C-∆N1 and Gal4-MEF2C-∆N2 were sumoylated in HeLa cells when co-expressed with GFP-SUMO1 ( Figure 4D). Since MEF2C-∆N1 and MEF2C-∆N2 lack the HDAC-binding domain of MEF2C, these data suggest that sumoylation of MEF2C might mediate the recruitment of transcriptional repressors other than class II HDACs to repress the transcriptional activities of Gal4-MEF2C or LexA-VP16. In addition, MEF2C-∆N2 lacks the p38 phosphorylation sites and yet retains the ability to be regulated by sumoyla- tion. This is consistent with the notion that sumoylation does not repress the transcriptional activity of MEF2C through reducing p38 phosphorylation of MEF2C.

Phosphorylation of S396 facilitates sumoylation of MEF2C
Recently, S396, a conserved serine residue in the C-terminal γ-domain of MEF2C, has been shown to be phosphorylated [20]. MEF2C-S396A or a MEF2C splicing variant that lacks the γ-domain is much more active in transcriptional reporter assays [20]. Moreover, γ-domain alone when fused to Gal4 can repress the basal transcriptional activity of Gal4, and this repression is diminished by the S396A mutation [20]. These findings indicate that phosphorylation of S396 inhibits the transcriptional activity of MEF2 proteins by recruiting transcriptional repressors. Interestingly, the major sumoylation site of MEF2C, K391, is located in the vicinity of S396 ( Figure 1B and 5A).
We therefore tested whether phosphorylation of S396 affected the sumoylation of MEF2C. We envisioned three possible scenarios (a, b, and c) with respect to the relationship between phosphorylation of S396 and sumoylation at K391 ( Figure 5A). In model a, phosphorylation and sumoylation are independent of each other. In this case, mutations of K391R and S396A are expected to have additive or synergistic effects in reporter assays. In models b and c, sumoylation of K391 is dependent on phosphorylation of S396, or vice versa. Gal4-MEF2C-K391R activated the transcription of a reporter gene that contained Gal4-binding sites much more efficiently than did Gal4-MEF2C-WT ( Figure 5B). The Gal4-MEF2C-K391R,S396A double mutant did not exhibit higher transcriptional activity than did Gal4-MEF2C-K391R ( Figure 5B). Similarly, both Gal4-MEF2C-∆N2-K391R and Gal4-MEF2C-∆N2-K391R,S396A completely lost the ability to repress the transcriptional activity of LexA-VP16 toward a reporter gene that contained both Gal4-and LexA-binding sites ( Figure 5C). These data are consistent with the notion that phosphorylation of S396 and sumoylation of K391 act through the same pathway to repress transcription, and model a is unlikely to be correct. Furthermore, Gal4-MEF2C-S396A was less active in promoting transcription than Gal4-MEF2C-K391R ( Figure 5B). Similarly, Gal4-MEF2C-∆N2-S396A still retained partial activity in repressing the transcriptional activity of LexA-VP16 (Figure 5C). Although MEF2C-S396E (a phospho-mimicking mutant) was slightly more active than the MEF2C-WT for unknown reasons (see Discussion), it was much less active than the S396A mutant ( Figure 5B and 5C). While we cannot rule out model c, our data are more consistent with model b, in that sumoylation of MEF2C is the key event in inhibiting its transcriptional activity and phosphorylation of S396 might facilitate the sumoylation of MEF2C at K391.
To directly test this hypothesis, we expressed and purified GST-MEF2C-∆N2 and the corresponding S396A and S396E mutants from E. coli and examined their phosphorylation by Cdk1/Cyclin B1. CDK1/Cyclin B1 phosphorylated GST-MEF2C-∆N2-WT to a greater extent than the MEF2C-∆N2-S396A mutant ( Figure 6A), suggesting that Cdk1/Cyclin B1 can phosphorylate MEF2C at S396. We then examined the sumoylation of recombinant GST-MEF2C-∆N2-WT, S396A, and S396E proteins in vitro. As shown in Figure 6B, at high concentrations of SUMO reaction components and these substrate proteins, all three GST-MEF2C-∆N2 proteins were efficiently sumoylated in vitro. The S396E mutant was slightly more efficiently sumoylated ( Figure 6B). We next tested whether phosphorylation of MEF2C-∆N2 by Cdk1 affected its sumoylation. Because sumoylation of the S396A mutant was already very efficiently sumoylated at high concentrations of SUMO enzymes, we performed the SUMO reactions with much lower concentrations of SUMO reaction components and substrates. Under these conditions, addition of Cdk1/Cyclin B1 enhanced the sumoylation of GST-MEF2C-∆N2-WT, but not the S396A mutant ( Figure 6C). Furthermore, the S396E mutant was more efficiently sumoylated than WT or S396A proteins in the absence of Cdk1/Cyclin B1, and was not further enhanced by Cdk1/ Cyclin B1 ( Figure 6C). These results suggest that phosphorylation of S396 facilitates the sumoylation of MEF2C-∆N2 in vitro.
We note that the S396A mutation and the K391R/S396A double mutation in the context of the full-length MEF2C appeared to abrogate the transcriptional repression activity of Gal4-MEF2C to similar extents (Fig. 5B). On the other hand, the K391R/S396A double mutation in the context of ∆N2-MEF2C had a significantly larger effect than the S396A mutation in diminishing the transcriptional repression activity of Gal4-∆N2-MEF2C (Fig. 5C). The underlying reasons for this difference is unclear at present. An intriguing possibility is that, as compared to the full-length MEF2C, sumoylation of ∆N2-MEF2C might be less dependent on prior phosphorylation at S396. Indeed, at later timepoints, ∆N2-MEF2C-S396A was as efficiently sumoylated as ∆N2-MEF2C in vitro (Fig. 6B). Unfortunately, the full-length MEF2C did not express well in bacteria, which prohibited us from testing its sumoylation in vitro.

Discussion
Many transcription factors and cofactors can be covalently modified by SUMO [21,22,38]. Sumoylation of these factors generally leads to transcriptional repression. Sumoylation of Elk-1 and p300 facilitates the recruitment of HDAC2, and -6, respectively, suggesting that sumoylation might repress transcription through the recruitment of HDACs to promoters [36,37]. In addition to HDACs, Daxx was shown to be another co-repressor that contributes to SUMO-mediated repression of transcription [44,45].
In this study, we show that MEF2C is sumoylated in vitro and in vivo, and sumoylation inhibits its transcriptional activity. Sumoylation of MEF2C does not appear to affect its DNA-binding, HDAC-recruitment, nuclear localization [38], stability (data not shown), or p38 phosphorylation, suggesting that sumoylation does not repress the transcriptional activity of MEF2C through these mechanisms. Moreover, when fused to Gal4, MEF2C-∆N2 that cannot bind to class II HDACs and other transcriptional repressors, including cabin1 [46] and MITR [16], is sufficient to repress the ability of LexA-VP16 to activate transcription from a promoter containing both Gal4-and LexA-binding sites. The K391R mutation abolishes the transcriptional repression activity of Gal4-MEF2C-∆N2, suggesting that sumoylation of MEF2C-∆N2 recruits transcriptional repressors other than class II HDACs. Class I HDACs does not seem to be the repressor recruited after sumoylation since overexpression of class I HDACs failed to repress the transcriptional activity of MEF2C in reporter assays (data not shown). Therefore, sumoylation of MEF2C appears to repress its transcriptional activity through a novel mechanism.

Regulation of the transcriptional activity of MEF2
Transcription activity of myogenic bHLH and MEF2 proteins is tightly regulated during muscle differentiation. Multiple pathways exist to ensure the repression of these transcription factors in dividing myoblasts [47][48][49]. For example, Cdk4/Cyclin D represses the activity of MEF2 proteins through blocking their interactions with the GRIP1 co-activator [14], although it is unclear whether Cdk4/Cyclin D phosphorylates MEF2 proteins directly. In addition, another cyclin-dependent kinase, Cdk5, phosphorylates MEF2 proteins and inhibits their transcriptional activity in neurons [18]. Consistent with recent findings by Gregoire and Yang [39], we observed that sumoylation-deficient MEF2C has higher myogenic activity than wild-type ( Figure 2B and 2C), suggesting that sumoylation might be another important mechanism to actively repress the transcriptional activity of MEF2 proteins in dividing myocytes. Surprisingly, only a very small population of MEF2C is modified by SUMO in C2C12 cells (data not shown). It is unclear how sumoylation of this small population of MEF2C effectively suppresses the activity of cellular MEF2C. On the other hand, this appears to be a recurring theme in the sumoylation of transcriptional factors. One intriguing possibility is that transient sumoylation of these transcriptional factors recruits transcriptional repressors that covalently modify the chromatin at the transcriptional loci, which in turn establishes a relatively long-lived chromatin state that is not permissible for transcription. Alternatively, a "molecular memory" model has been proposed to explain this phenomenon. In this model, a protein molecule that has experienced a sumoylation/desumoylation cycle is proposed to be functionally distinct from one that has never experienced sumoylation [50].

Regulation of sumoylation of MEF2 by phosphorylation
We have shown that sumoylation of MEF2C appears to be facilitated by phosphorlyation of S396, since S396A mutant is not as efficiently sumoylated in vivo and in vitro as compared to the wild-type MEF2C. Phosphorylation by Cdk1/Cyclin B1 enhances the sumoylation of MEF2C-∆N2, but not that of MEF2C-∆N2-S396A. Surprisingly, though the phospho-mimicking MEF2C-∆N2-S396E mutant was more efficiently sumoylated than the wildtype in vitro, MEF2C-S396E was less efficiently sumoylated than the wild-type in vivo. Consistently, MEF2C-S396E was slightly more active than the wild-type in transcriptional assays. The reason for this discrepancy is unclear at present. One possibility is that phosphorylated and sumoylated MEF2C is a better substrate for SUMO isopeptidases in vivo. Phosphorylation at S396 of the wildtype MEF2C can be reversed by phosphatases, resulting in longer half-life of sumoylated MEF2C. In contrast, the S396E mutant would be refractory to the actions of phosphatases. Even though MEF2C-S396E can be more efficiently sumoylated as compared to the wild-type, it might be also more prone to desumoylation of SUMO isopeptidases, resulting to its lower steady state of sumoylation in vivo.
The kinase that mediates the phosphorylation of S396 in vivo is currently unknown. Because S396 is followed by a proline, the S396 kinase is likely to belong to families of proline-directed kinases, such as cyclin-dependent kinases and MAP kinases. Though Cdk1/Cyclin B1 is capable of phosphorylating MEF2C at S396 in vitro, we do not think that it is the relevant kinase that phosphorylates MEF2C at this site in vivo. Cdk5 was shown to phosphorylate MEF2D at this site in neurons [18]. However, the activity of Cdk5 is strictly dependent on its p35/p39 cofactor that is exclusively expressed in neurons [51]. Furthermore, we did not observe in vitro phosphorylation of MEF2C-∆N2 by Cdk5 alone (data not shown). Thus, yet unidentified kinase(s) mediate phosphorylation of MEF2C at S396 and consequently sumoylation of MEF2C at K391 during muscle differentiation.
There are precedents for phosphorylation-dependent sumoylation. Phosphorylation of HSF1 and PPARγ has recently been shown to facilitate their sumoylation [52,53]. Intriguingly, similar to MEF2C, the phosphorylation sites on HSF1 and PPARγ are both followed by a proline ( Figure 6E). In all three cases, the sumoylation and phosphorylation sites are separated by four residues (Figure 6E). Phosphorylation at +5 position by a prolinedirected kinase is expected to introduce negative charges at this position, which may enhance the binding affinity between UBC9 and the sumoylation motif. An inspection of the known sumoylation sites reveals that a sumoylation site of c-Myb is also followed by an SP motif at the +5 and +6 positions, suggesting that sumoylation of c-Myb at this site might also be regulated by phosphorylation [31]. In addition, several known sumoylation sites contain an acidic residue (D or E) at the +5 position ( Figure 6E). These extended sumoylation consensus motifs that contain acidic residues at +5 position might be more efficiently sumoylated.

Conclusion
The MEF2 family of transcriptional factors regulates cell proliferation, differentiation, and apoptosis of many cell types. The activities of MEF2 proteins are themselves regulated by multiple mechanisms, including sumoylation. We have discovered that phosphorylation of a serine residue in the vicinity of the sumoylation site facilitates sumoylation of MEF2C. This finding also allows us to define extended sumoylation consensus motifs that contain SP at +5/+6 positions or an acidic residue at +5 position.

Cell culture and transfection
HeLa, 10T1/2, and C2C12 cells were grown in Growth Medium (DMEM supplemented with 10% fetal bovine serum, 100 units/ml penicillin, and 0.1 mg/ml streptomycin). Differentiation of MyoD-transfected 10T1/2 cells was induced by substituting Growth Medium with Differentiation Medium (DMEM supplemented with 2% horse serum, 100 units/ml penicillin, and 0.1 mg/ml streptomycin). Cells were plated and transfected in 12-well plates for promoter-luciferase assays, 6-well plates for immunoprecipitations, 10-cm plates for chromatin immunoprecipitation, and 4-well chambered slides for indirect immunofluorescence microscopy. Cells were transfected using the Effectene (Qiagen) or the Lipofectamine 2000 Plus (Invitrogen) reagents according to manufacturer's protocols.

Plasmids and promoter assays
The coding regions of human SUMO1 (1-97) and SENP2 were amplified from human fetal thymus cDNA library (BD Biosciences) by PCR and cloned into pCS2 mammalian expression vectors that contain N-terminal Myc, HA, or GFP tags. PIASxβ was amplified from pGEX-PIASxβ constructs provided by S. Muller and cloned into the pCS2-HA vector. UBC9 was cloned into a pCS2 vector that contains a C-terminal Flag tag. The pCDNA-MEF2C, pCDNA-Flag-HDAC4, pCDNA-Flag-HDAC5, and pCDNA-Myc-MyoD constructs were obtained from E. Olson. The pCMV5-HA-p38 and pCMV5-MKK6-DD plasmids were gifts from M. Cobb. The dominant-negative UBC9 mutant and various MEF2C mutants were constructed with the QuikChange site-directed mutagenesis kit (Stratagene). The pET11c-hAOS1, pET28b-hUBA2, and pET28b-hUBC9 vectors were gifts from C. Lima and K. Orth.
Promoter assays were performed in triplicates with the dual-luciferase reporter assay system (Promega) according to manufacturer's protocols. Luciferase activity was measured with a Turner Designs luminometer and normalized for transfection efficiency using the activity of Renilla luciferase. The MEF2-responsive promoter activity assays were performed with a pMEF2×3-Luc construct (provided by E. Olson). The Gal4/LexA-promoter activity assay was performed with a pL8G5-Luc plasmid with or without the transfection of a pLexA-VP16 construct.

Immunofluorescence
10T1/2 fibroblast cells transfected with various plasmids were fixed with 4% paraformaldehyde and permeablized with 0.1% Triton-X100 in PBS. An antibody against myosin heavy chain (MHC) was then added at 1:4 dilutions. After washing, fluorescent secondary antibodies (Molecular Probes) were added at 1:500 dilutions. The cells were washed with PBS, counter-stained with DAPI, and viewed using a 20× objective on a Zeiss Axiovert 200 M microscope. Images were acquired using the Intelligent Imaging software and pseudo-colored in Adobe Photoshop.

Sumoylation assays
Plasmids that encode appropriate proteins were in vitro transcribed and translated (IVT) in reticulocyte lysate in the presence of 35 S-methionine and subjected to in vitro sumoylation reactions, which contained 2 µl of IVT product, 2 µg of AOS1-UBA2, 0.5 µg of UBC9, 1 µg of SUMO1, and 1 µl of Energy Mix (150 mM phosphocreatine, 20 mM ATP, 2 mM EGTA, 20 mM MgCl2, adjust pH to 7.7). Reactions were adjusted to a final volume of 10 µl with the XB buffer (10 mM HEPES, pH 7.7, 1 mM MgCl2, 0.1 mM CaCl2, 100 mM KCl, and 50 mM sucrose). Control reactions contained water and XB buffer. After 2 hr at 30°C, reactions were stopped with 10 µl of 2× SDS sample buffer, boiled, and subjected to SDS-PAGE followed by autoradiography.

Protein binding, kinase, and chromatin immunoprecipitation assays
For in vitro protein binding assays, 4 µg of purified His 6 -p38 was incubated with 5 µl of Ni 2+ -NTA beads in 50 µl of Q-A buffer (20 mM Tris, pH 7.7, 100 mM KCl, 1 mM MgCl 2 ) for 1 hr at room temperature. Beads were then incubated with 400 µl of blocking solution (25 mM Tris, pH 8.0, 150 mM NaCl, 2.5 mM KCl, 0.05% Tween-20, 5% dry milk) for 1 hr at RT. 20 µl of in vitro sumoylated MEF2C was added and incubated for another 1 hr at RT. Beads were washed two times with blocking solution without dry milk, eluted in 1× SDS sample buffer, and subjected to SDS-PAGE followed by autoradiography.
Chromatin immunoprecipitation (ChIP) assays were performed with a ChIP assay kit (Upstate) according to the manufacturer's protocol.