Phosphorylation motifs regulating the stability and function of myocyte enhancer factor 2A.

The phosphorylation status of the myocyte enhancer factor 2 (MEF2) transcriptional regulator is a critical determinant of its tissue-specific functions. However, due to the complexity of its phosphorylation pattern in vivo, a systematic inventory of MEF2A phosphorylation sites in mammalian cells has been difficult to obtain. We employed modern affinity purification techniques, combined with mass spectrometry, to identify several novel MEF2 phosphoacceptor sites. These include an evolutionarily conserved KSP motif, which we show is important in regulating the stability and function of MEF2A. Also, an indirect pathway in which a protein kinase casein kinase 2 phosphoacceptor site is phosphorylated by activation of p38 MAPK signaling was documented. Together, these findings identify several novel aspects of MEF2 regulation that may prove important in the control of gene expression in neuronal and muscle cells.

The phosphorylation status of the myocyte enhancer factor 2 (MEF2) transcriptional regulator is a critical determinant of its tissue-specific functions. However, due to the complexity of its phosphorylation pattern in vivo, a systematic inventory of MEF2A phosphorylation sites in mammalian cells has been difficult to obtain. We employed modern affinity purification techniques, combined with mass spectrometry, to identify several novel MEF2 phosphoacceptor sites. These include an evolutionarily conserved KSP motif, which we show is important in regulating the stability and function of MEF2A. Also, an indirect pathway in which a protein kinase casein kinase 2 phosphoacceptor site is phosphorylated by activation of p38 MAPK signaling was documented. Together, these findings identify several novel aspects of MEF2 regulation that may prove important in the control of gene expression in neuronal and muscle cells.
Myocyte enhancer factor 2 (MEF2) 1 is a transcriptional regulatory complex mediating diverse cellular functions in neurons (1,2), skeletal (reviewed in Ref. 3) and cardiac muscle (4 -6), and T cells (7,8). It is now well established that MEF2 plays a role in the differentiation of these cell types as well as functioning in a protective role against neuronal apoptosis.
To respond to diverse developmental and physiological cues, MEF2 is structurally organized to receive and respond to multiple signals from several intracellular signaling pathways (reviewed in Refs. 3 and 9). In this regard, perhaps the best characterized is the p38 MAPK-MEF2 axis, in mammals (10,11) and in yeast (12), although other kinase-catalyzed cascades mediated by big MAP kinase (13,14), protein kinase C (10), and protein kinase CK2 (15) are known to target MEF2. Moreover, consistent with its role as a signal sensor, putative phosphoacceptor motifs in the carboxyl terminal MEF2 transactivation domain may prove to further modulate MEF2 function in response to extracellular cues.
Given that MEF2, and the biological processes it regulates, are intrinsically governed by MEF2 phosphorylation status, we undertook to systematically document MEF2 phosphorylation patterns in mammalian cells; previous phosphopeptide mapping studies used in vitro phosphorylated MEF2 protein. The purpose thus being to detect physiologically relevant, and possibly novel, in vivo MEF2 phosphorylation sites. To accomplish this we used several state-of-the-art mass spectrometric techniques to detect phosphorylation sites from MEF2 expressed in mammalian cells. To this end, we have made use of a mammalian tandem affinity purification (TAP) method (16,17) for low-abundance nuclear transcription factors that allows purification to homogeneity and provides amounts compatible with mass spectrometric analysis of phosphorylation sites.
In these studies, we have identified two important and novel aspects of MEF2 regulation. One is a highly conserved phosphoacceptor motif that regulates MEF2 stability and function. The second is an indirect pathway of MEF2 regulation by p38 MAPK mediated by the CK2 holoenzyme. These studies on MEF2 regulation thus identify novel aspects of functional MEF2 regulation that will serve to modulate MEF2 controlled gene expression in a variety of cell types.

EXPERIMENTAL PROCEDURES
Materials-Unless otherwise noted, all chemicals were obtained from Sigma-Aldrich. DNA-modifying enzymes were purchased from New England Biolabs (NEB).
Cell Culture and Transfections-COS7 and HeLa cell cultures were maintained in Dulbecco's modified Eagle's medium (Invitrogen) containing penicillin, streptomycin (Invitrogen), and 10% fetal bovine serum (Atlanta Biologicals). Transfections were performed using the calcium phosphate precipitation method.
His-tagged Protein Purification-The coding region for human MEF2A was cloned into the EcoRI site of pCDNA 3.1/His (Invitrogen). The coding region for the N-terminal His-MEF2A was then subcloned into the PstI site of pCDNA4/TO/His (Invitrogen). Cells were transfected with 30 g of DNA per 100-mm dish. Typically, 20 plates of cells were used for a single purification. A standard manufacturer's protocol was used for purification using Ni-Agarose resin (Qiagen).
Tandem Affinity Purification-The coding region for human MEF2A was ligated into the EcoRI site of pCDNA4/TO/TAP (described in Ref. 17). Cells were transfected with 30 g of DNA per 100-mm dish. Typically, 5 plates of cells were used for a single purification. The details of mammalian TAP-tagged protein purification are explained in Ref. 17. Briefly, cells were lysed by freeze/thawing. The lysate was passed over IgG resin (Sigma), and the beads were washed. Tagged proteins were eluted by cleaving with TEV protease (Invitrogen) then supplementing with Ca 2ϩ and passed over calmodulin resin (Stratagene) for a second round of purification. Proteins were finally eluted using either 2 mM EGTA or SDS sample buffer and analyzed by SDS-PAGE. Proteins were visualized using Gelcode Blue (Pierce).
In-gel Trypsin Digest-Protein bands were excised, cut into 1 mm 3 pieces, and washed 3 times with 50% acetonitrile/25 mM ammonium bicarbonate for 15 min with shaking. Gel pieces were incubated with 50 mM ammonium bicarbonate ϩ 10 mM dithiothreitol for 30 min at 50°C, washed with acetonitrile, then incubated with 50 mM ammonium bicarbonate ϩ 55 mM iodoacetamide (freshly made) for 20 min in the dark. The gel pieces were washed with acetonitrile, air dried, and rehydrated with 12.5 ng/l trypsin (Promega, sequencing grade) in 50 mM ammonium bicarbonate, then incubated at 37°C overnight. Peptides were extracted once using 3% formic acid, 1 min at 80°C, followed by 20 min of shaking and 1 min of centrifugation.
Mass Spectrometry-Peptides were concentrated prior to analysis using ZipTips TM (Millipore) according to the manufacturer's protocol. Peptides were eluted directly onto a steel target using 10 mg/ml ␣-cyano-4-hydroxycinnamic acid in 65% acetonitrile/0.3% trifluoroacetic acid. To purify phosphopeptides, metal-chelating ZipTips TM (Millipore) were used with copper(II)-sulfate according to the manufacturer's instructions. Peptide fingerprinting was performed on a Voyager DE-STR (Applied Biosystems) using positive ion reflector mode under optimized conditions. Fragmentation analysis was performed using a nanospray ion source on a prototype Q-Star instrument (Sciex). The instrument was optimized and calibrated daily.
Northern Blot Assay-Total RNA was isolated with Trizol reagent (Invitrogen) and resolved on a 1% agarose gel in the presence of 6.2% formaldehyde and then blotted on to Nytran membrane (Schleicher & Schuell) in 20ϫ SSC. The complete cDNA (2.9 kb) of MEF2A, excised from pMT2 vector with EcoRI, and a 316-bp glyceraldehyde 3-phosphate dehydrogenase fragment, derived from exons 5-8 (Ambion) and excised with SacI and HindIII, were used as probes for MEF2A and glyceraldehyde 3-phosphate dehydrogenase, respectively. The probes were labeled with 32 P using Random Primers DNA Labeling System (Invitrogen) and purified on MicroSpin TM G-25 Columns (Amersham Biosciences). Prehybridization (2 h) and hybridization (overnight) were carried out in 50% deionized formamide, 5ϫ Denhardt's, 5ϫ SSC, 5 mM EDTA, 0.1% SDS, and 200 g/ml salmon sperm DNA. Membranes were washed and signals were visualized by exposure to film (BioMax MR, Kodak). Between labeling with MEF2A and glyceraldehyde 3-phosphate dehydrogenase, the blots were stripped for 2 h in 50% deionized formamide, 0.1% SSC, 0.1% SDS preheated to 68°C.
Site-directed Mutagenesis-The GAL4-MEF2A S255A and S255D mutations were generated using the QuikChange kit (Stratagene) according to the manufacturer's instructions. The primers used were: S255A-GTC ATG CCT ACA AAG GCT CCC CCT CCA CCA G and S255D-GTC ATG CCT ACA AAG GAT CCC CCT CCA CCA G.

Expression and Purification of MEF2A from Mammalian
Cells-Two purification strategies, His 6 and TAP, were employed for obtaining suitable material to probe the phosphorylation status of MEF2A in mammalian cell culture. The His 6 -MEF2A was expressed in HeLa followed by purification using Ni 2ϩ affinity beads. This strategy was capable of purifying amounts of MEF2A that could be visualized using Coomassie Blue staining (Fig. 1A). More importantly, the purified MEF2A was found in two distinct bands, indicating that a fraction of it had been post-translationally modified. This pattern of mobility is in agreement with previous detection of MEF2A from HeLa (18), COS7, and muscle cell cultures (19) by Western immunoblotting, and is in sharp contrast to the single band obtained when MEF2A is purified from bacterial cells (data not shown). Purified MEF2A protein was subsequently used to characterize its endogenous phosphorylation status.
To investigate the effect of p38 MAPK on the phosphorylation status of MEF2A, we employed an additional purification strategy utilizing the TAP scheme. TAP-MEF2A was expressed in COS7 cells in the absence or presence of expression vectors for p38 MAPK and a constitutively active upstream regulator, MKK6. The TAP scheme purified MEF2A to near homogeneity, allowing us to use Coomassie Blue staining to reveal the sizable shift in mobility of MEF2A caused by p38 MAPK (Fig. 1B). This shift has previously only been detected by immunoblotting or radiolabelling. This purification was used to probe for novel p38 phosphorylation sites in MEF2A.
Identification of MEF2A-Proteins were identified by either tryptic mass fingerprinting or CID fragmentation. The MALDI-TOF spectrum for the tryptic digest of MEF2A ( Fig. 2A) contains a number of peaks whose masses match the expected mass of peptides from MEF2A. This confidently identifies the protein as MEF2A. Further confirmation was obtained, using a nanospray-QStar instrument, by CID fragmentation of selected peptides. The fragmentation products reveal short sequences and characteristic ions (Fig. 2B) that could only be derived from specific MEF2A peptides. These two types of analyses were subsequently used for identification of phosphopeptides.
We used a broad range of techniques to characterize the phosphorylation status of MEF2A in mammalian cell culture. This included comparative analysis between tryptic mass fin-FIG. 1. Purification of MEF2A from mammalian cell culture. A, His 6tagged MEF2A was overexpressed in HeLa cells and purified using a Ni 2ϩ affinity beads. The proteins were stained by Coomassie Blue and identified by peptide mass fingerprinting. MEF2A was identified in two of the bands, suggesting it had been post-translationally modified. B, TAP-tagged MEF2A, along with p38 MAPK and its upstream activator, MKK6, were overexpressed in COS7 cells. TAP-MEF2A was purified using the TAP scheme (16). The p38 MAPK caused a significant shift in the mobility of MEF2A as detected by Coomassie Blue staining. gerprints, characteristic 80 Da shifts in mass caused by phosphorylation, CID fragmentation, immobilized metal-affinity capture of phosphopeptides, and sequence homology. The details of these analyses will be discussed below.
A Novel, Endogenous Phosphorylation Site in MEF2A-The MALDI-TOF spectra for tryptic mass fingerprints of MEF2A expressed in mammalian cells were very similar to MEF2A expressed in bacterial cells with one notable exception. A peak at mass 1437.69 Da, corresponding to the amino acids 255-269 (SPPPPGGGNLGMNSR), was significantly lower when MEF2A was expressed and purified from mammalian cells ( Fig. 2A), whereas it was prominent in MEF2A expressed in bacterial cells. This prompted us to investigate what was the cause of this difference. A post-translational modification, such as phosphorylation, was a likely candidate; however, no peak was detected at 80, 160, or 240 Da higher in mass. Careful inspection of the region near 2000 m/z (Fig. 2C) revealed a small peak with an m/z value of 1994.02, corresponding to a MEF2A peptide with one missed trypsin cleavage, amino acids 250 -269 (VMPTKSPPPPGGGNLGMNSR), and a peak 79.94 Da higher in mass (2073.96). Phosphate incorporation adds 79.97 Da. These data indicate that a measurable fraction of the MEF2A protein pool in mammalian cells is phosphorylated at Ser-255. This sample was then analyzed on a nanospray-QStar instrument to obtain CID fragmentation data (Fig. 2D). Fragmentation of the potential phosphopeptide revealed a clear set of y fragments identifying a stretch of 4 proline residues. Additionally, the fragment at m/z 1419.76 can only be explained by a y 15 fragment that has lost H 3 PO 4 (mass loss of 98 Da). These fragments and others confidently identify this peptide as MEF2A 250 -269 phosphorylated at Ser-255.
To ascertain the potential importance of this phosphoryla- tion site we compared the sequence of human MEF2A with the other MEF2 family members from various organisms (Fig. 3). The level of sequence homology for this region is remarkable and confirms the high degree of evolutionary conservation of the KSPP phosphorylation motif. The lack of conservation at Thr-253 is in agreement with the fragmentation data, which demonstrates the phosphorylation of Ser-255, not Thr-253.
Phosphorylation of Ser-255 Targets MEF2A for Degradation-Because Ser-255 is in the MEF2 C-terminal transactivation domain, we assessed the function at this site in a GAL4based transcriptional response assay by generating mutations of Ser-255 in a GAL4-MEF2A (residues 99 -507) fusion protein.
These include Ser-255 to aspartic acid (S255D) and Ser-255 to alanine (S255A) mutations. A GAL4 luciferase assay system in COS7 cells was used to determine the activity of these proteins (Fig. 4C). The activity of the S255D mutation was markedly lower than that of either the wild type GAL4-MEF2A or the GAL4-MEF2A (S255A) mutation. A Western blot for MEF2A on cells transfected with these expression vectors indicates decreased full-length protein levels for GAL4-MEF2A (S255D) and evidence of degradation products (Fig. 4B). A Northern blot on these samples demonstrates that transcript expression levels can not account for the differences in GAL4-MEF2A protein levels indicated in Fig. 4B, thus indicating that the decreased protein levels for GAL4-MEF2A (S255D) are likely due to instability of the protein. This is further supported by the detection of at least one large proteolytic fragment by Western analysis (Fig. 4B).
Identification of Novel p38 MAPK Sites in MEF2A-Next, we re-visited the p38 MAPK-catalyzed phosphorylation of MEF2A, because p38 MAPK is a potent physiological activator of MEF2 transactivation potential (10,11). Thr-312 and Thr-319 are known phosphorylation sites important for the increased transcriptional activation of MEF2A by p38 MAPK. Ser-453 and Ser-479 are phosphorylated in vitro but were not important functionally (11). However, previous results indicate that the in vivo phosphorylation of MEF2A by p38 MAPK is possibly more complicated than previously reported (10). A two-dimensional phosphopeptide map of MEF2A expressed in COS7 cells with p38 MAPK and MKK6 showed a very complex pattern of phosphorylation (10) that was different from phosphorylation data derived from in vitro analysis (11). Because this could have important regulatory consequences, we have used mass spectrometry to further dissect and characterize p38 MAPK phosphorylation of MEF2A in mammalian cells.
The tryptic mass fingerprint of MEF2A, co-expressed with p38 MAPK and MKK6, was compared with the lower-mobility band from cells expressing MEF2A alone (Fig. 5). This simple comparison reveals a number of phosphorylated peptides, the evidence for which will now be considered individually.
The peptide of mass 4223.05 (MEF2A 190 -233, note that the fourth carbon isotope, 4226.06, is labeled in Fig. 5) has one methionine residue and two potential p38 MAPK phosphorylation sites. Methionine residues are often partially oxidized due to exposure to air. In the p38 MAPK-treated sample there is a peak at a mass 80 Da higher, indicating phosphorylation. This peak is not present in the sample from cells expressing MEF2A alone. Furthermore, the phosphorylated and non-phosphorylated peptides both show evidence of an oxidized form at a mass 16 Da higher from each respective peak. This confirms that this peptide is the phosphorylated form of MEF2A 190 -233.
A similar analysis reveals a phosphorylated peptide at mass 2714.32 (MEF2A 475-498). Again, this peptide is 80 Da higher than the non-phosphorylated peptide and is absent from the non-treated sample. Interestingly, in the sample from cells expressing MEF2A alone, the non-phosphorylated peptide is also absent. Occasionally, phosphopeptides lose their phosphate group, either in solution or during ionization. Because this peptide is actually a missed tryptic cleavage of Arg-492, there is the possibility that phosphorylation of this peptide interfered with the trypsin digestion, prior to the loss of phosphate.
The MEF2A 95-114 peptide has a mass of 2247.99 whereas its phosphorylated form is found at 2327.96. The intensity of the latter peak is quite high compared with the former peak. However, it is difficult to quantify these peptides based on peak intensity, as the energy required to ionize a phosphopeptide may differ from its non-phosphorylated form.
In addition to peptide mass fingerprinting, we employed CID fragmentation to sequence putative phosphopeptides. Furthermore, the efficiency of nanospray ionization employed in this instrument differs from that of the MALDI source used for peptide mass fingerprinting. This form of ionization was particularly suitable for detecting the MEF2A phosphopeptide 404 -413 (doubly charged m/z of 617.28). Subsequent CID fragmentation of this peptide confirmed the presence of phosphorylation (loss of H 3 PO 4 from the precursor ion and several fragment ions) at Ser-408 (Fig. 6).
A p38 MAPK-induced Phosphopeptide with no MAPK Consensus-Of particular interest is the MEF2A peptide 283-300. This region of MEF2A is alternatively spliced, and this particular splice variant is abundant in muscle and nerve cells. Phosphorylation of this peptide is indicated by the 80 Da shift in mass to 2182.95; however, this peptide does not contain a proline-directed MAPK consensus site (Fig. 7A). Further evidence that this peptide is phosphorylated is provided by the oxidation of the two methionine residues in this peptide. The phosphorylated peptide shows evidence of two oxidized methionine residues, in agreement with the two oxidized methionine residues seen for the non-phosphorylated peptide. Additional evidence that at least a fraction of this peptide is phosphorylated is provided by the use of a copper immobilized metalaffinity capture purification. MEF2A tryptic peptides were fractionated using a copper(II)-sulfate-treated metal chelating ZipTip TM . Phosphorylated peptides have a higher affinity for Cu 2ϩ than non-phosphorylated peptides. The selectively bound peptides were eluted, concentrated on a C-18 ZipTip TM , and analyzed by MALDI-TOF. The presence of the m/z 2182.92 peptide in this eluate confirms the phosphorylation status of the MEF2A 283-300 peptide (Fig. 7B). Sequence analysis of this peptide reveals a strong consensus motif that is targeted by protein kinase CK2. Because phosphorylation of this peptide is dependent on p38 MAPK activation, cross-talk between p38 and protein kinase CK2 in MEF2A phosphorylation is implicated by these data (see "Discussion").

DISCUSSION
In these studies we have used state-of-the-art mass spectrometric techniques to asses the in vivo phosphorylation pattern of the MEF2A transcriptional regulator. This approach has yielded novel information concerning the regulation of MEF2: first, the identification of a novel phosphoacceptor site that regulates MEF2 stability; second, the identification of previously uncharacterized p38 MAPK phosphoacceptor sites; and third, the discovery of a protein kinase CK2 consensus phosphoacceptor site that requires p38 MAPK activity to be phosphorylated. These studies constitute several novel aspects of MEF2 regulation by reversible phosphorylation.
Phosphorylation of Ser-255 in MEF2A by endogenous kinases is detected in several cell types (C2C12, COS, and HeLa), and this phosphorylation is independent of the activity of p38 MAPK (Fig. 5). This sequence is highly conserved in members of the MEF2 family across numerous species, highlighting the importance of this site in the regulation of MEF2 proteins. The conserved sequence consists of a KSP motif, which is a known phosphoacceptor site for several kinases. This includes MAPK members (ERK1, ERK2, and stress-activated protein kinase) (20,21), cdc2-like kinases (including CDK5) (22,23), and glycogen synthase kinase 3 (24). The presence of these proteins in both muscle and nerve cells is suggestive of a link between these pathways and the functional consequences of phosphorylation of this site in MEF2.
The extracellular-regulated kinases (ERK1 and ERK2) are abundant in nerve cells (25) and have been implicated in the myogenic program (26 -30). The activity of ERK is biphasic, with higher activity in myoblasts and later stage myotubes (26,28). In the early stages of the myogenic program, ERK has an inhibitory effect on the expression of myogenic regulators and muscle-specific proteins (28,29). In later stages of differentiation, ERK appears to enhance the formation of mature myotubes (26,27) and induce hypertrophy (30). Thus, phosphorylation-induced degradation of MEF2 proteins by ERK would explain the early inhibitory effect that ERK has on the myogenic program.
The CDK5 protein is found predominantly in nervous tissue, although it has also been detected in muscle and other nonneural tissues. It has been implicated in migration, actin dynamics, microtubule transport, cell adhesion, axon guidance, synaptic structure and plasticity, membrane transport, and myogenesis (31). In muscle cells CDK5 can be detected throughout the myogenic program. The level of expression and activity of CDK5 is increased during myogenesis. In addition, proliferating myoblasts show a predominantly cytoplasmic localization of CDK5, whereas differentiating cells have an increase in nuclear levels of CDK5 (32). The possibility exists that CDK5 has a role in regulating the activity of MEF2 proteins by direct phosphorylation, although how this would pertain to regulation of MEF2 by degradation would require further investigation.
The ubiquitous phosphorylation of Ser-255, coupled with its role in the degradation of MEF2, points tantalizingly toward a role for glycogen synthase kinase 3. Originally thought of as an enzyme solely involved in glycogen metabolism, this ubiquitous kinase is now known to be involved in numerous cellular signaling functions (33). Overexpression of active glycogen synthase kinase 3 induces apoptosis in neuronal cells by phosphorylation of downstream targets (34). In addition, glycogen synthase kinase 3 is known to phosphorylate several proteins, which are subsequently targeted for degradation (35,36). It is interesting to note that MEF2 has recently been identified as protecting neuronal cells from apoptosis (2,37,38) and that it is also targeted for degradation by phosphorylation (39).  (11,15). These sites and their likely kinase are indicated. Important features, such as alternative splicing (87-132 and 288 -295) and the p38 docking site, are also indicated.
Whether Ser-255 is the targeted residue that links MEF2 degradation and neuronal apoptosis will be of great interest.
MEF2A has been convincingly identified as a target for p38 MAPK. Phosphorylation of MEF2A by p38 MAPK strongly increases the transcriptional activity of MEF2A, and this increase in activity has been mapped to a couple of key residues (Thr-312 and Thr-319) (10,11). However, in vivo phosphorylation of MEF2A by p38 MAPK produces a complex pattern of phosphorylation that cannot be entirely explained by these two key residues. The possibility, therefore, remains that other sites within MEF2A are involved in p38 MAPK signaling. Although the tryptic phosphopeptide containing Thr-312 and Thr-319 was too large to detect by mass spectrometry, the data from this study conclusively identify several new phosphorylation sites, induced by p38 MAPK activity (Fig. 8).
Most notable among these p38 MAPK-induced phosphorylation sites is the confirmation of phosphorylation at a consensus protein kinase CK2 site, Ser-289. This region of MEF2A (and the homologous region of MEF2C) is known to be alternatively spliced in different tissues (40,41), suggesting a possible tissue-specific functional role for this region. Its sequence is highly suggestive of a protein kinase CK2 site ((S/T)XX(D/E)) that, until now, had not been confirmed as a phosphoacceptor site. The data presented here demonstrates that Ser-289 is likely a target of protein kinase CK2, but more interestingly, it suggests a link between p38 MAPK activity and the phosphorylation of MEF2A by protein kinase CK2. In fact, a previous report indicates that the CK2 holoenzyme (␣ and ␤ subunits) can be activated specifically by p38 MAPK (42), indicating the possibility that CK2 and p38 MAPK may cooperatively dock with MEF2A.
MEF2 proteins are involved in a number of different cellular processes including proliferation, differentiation, apoptosis, and hypertrophy. These disparate roles for MEF2 are partly explained by the regulated activation of specific signaling pathways. Here we provide conclusive evidence for the phosphorylation of MEF2A by a number of signal-dependant kinases. These data concerning reversible phosphorylation of MEF2A will be a fundamental aspect of understanding the diverse function of MEF2 proteins.