DYRK1A phosphorylates MEF2D and decreases its transcriptional activity

Abstract Myocyte enhancer factor 2D (MEF2D) is predominantly expressed in the nucleus and associated with cell growth, differentiation, survival and apoptosis. Previous studies verified that phosphorylation at different amino acids determined MEF2's transcriptional activity which was essential in regulating downstream target genes expression. What regulates phosphorylation of MEF2D and affects its function has not been fully elucidated. Here, we uncovered that dual‐specificity tyrosine phosphorylation regulated kinase 1A (DYRK1A), a kinase critical in Down's syndrome pathogenesis, directly bound to and phosphorylated MEF2D at Ser251 in vitro. Phosphorylation of MEF2D by DYRK1A significantly increased MEF2D protein level but attenuated its transcriptional activity, which resulted in decreased transcriptions of MEF2D target genes. Phosphorylation mutated Ser251A MEF2D exhibited enhanced transcriptional activity compared with wild type MEF2D. MEF2D and DYRK1A were observed co‐localized in HEK293 and U87MG cells. Moreover, DYRK1A‐mediated MEF2D phosphorylation in vitro might influence its nuclear export upon subcellular fractionation, which partially explained the reduction of MEF2D transcriptional activity by DYRK1A. Our results indicated that DYRK1A might be a regulator of MEF2D transcriptional activity and indirectly get involved in regulation of MEF2D target genes.


| INTRODUC TI ON
MEF2D is a member of MEF2 protein family, a transcription factor family containing MEF2A, MEF2B, MEF2C and MEF2D. MEF2 plays critical roles in cell development. 1,2 They participate in many signal pathways relevant to cell division, apoptosis and differentiation, involving MAPK1, 3 CDK5, 4 MAPK7, 5 and AMPK. 6 Many studies manifest MEF2D participating in tumorigenesis and cancer progression. MEF2A and MEF2D are up-regulated during differentiation of mouse embryonal carcinoma P19 cells, while MEF2B is highly expressed in undifferentiated P19 cells. 7 Elevated MEF2D expression is detected in hepatocellular carcinoma clinical specimens, especially in those with poor prognosis. 8 MEF2D is also observed abundant in lung cancer tissues and cell lines comparing with the matched normal tissues and cell lines. Knocking down of MEF2D by miRNA interference suppresses the growth of lung carcinoma. 9 Silencing of MEF2D triggers G2-M arrest in a way associated with direct downregulation of genes related to cell cycle progress in hepatocellular carcinoma. 8,10 Inflammatory conditions increase MEF2D expression in lung cancer cells, which might contribute to cancer development by influencing cancer microenvironment and cell bio-behaviours. 11 Increased MEF2D expression exists in pancreatic cancer tissues compared with adjacent normal tissues. MEF2D regulates cell proliferation, migration and invasion abilities in pancreatic cancer via Akt/GSK-3β signalling pathway. Of note, the increased expression of MEF2D is associated with tumour size, histological differentiation and TNM stage among pancreatic cancer patients. 12 In colorectal cancer tissues, MEF2D is positively correlated with CD31-positive microvascular density and promote tumour angiogenesis in vitro and in vivo, resulting in induction of expression of proangiogenic cytokines. 13 Our previous studies showed that DYRK1A is up-regulated by MEF2D, leading to decreased expression of NFATc2, a substrate of DYRK1A, 14 and a regulator of glioblastoma invasion. 15 MEF2D tumour-associated activity could be suppressed by miR-421 in gliomas. 16 Proliferations and invasion of cervical cancer cells might be regulated by miR-30a which also targets MEF2D. 17 Collectively, expression of MEF2D is vitally important to cancer pathogenesis and progression.
Transcriptional activity of MEF2D could be partially modulated by post-translational modification. Phosphorylation of MEF2D was intensively investigated. Intriguingly, phosphorylation at different amino acid residues has diverse effects on MEF2D transcriptional activity. For instance, phospho-Ser179 induced by MAPK7 increased transactivation of MEF2D in Hela cells. 18 ATM phosphorylates and activates MEF2D, contributes to neuronal survival in response to DNA damage. 19 In contrast, MEF2A and MEF2D were phosphorylated with the induction of apoptosis of cerebellar granule neurons. But sites phosphorylated during apoptosis are functionally distinct. The increased phosphorylation of MEF2A and MEF2D leads to decreased DNA binding ability and reduced transcriptional activity. 20 Removal of depolarization induces hyperphosphorylation of MEF2D at certain serine/threonine residues, following with weakened DNA binding ability and increased susceptibility to caspases. 21 Phospho-Ser444 by Cdk5 inhibits MEF2D transcriptional activity, while mutated MEF2D resistant to Cdk5 phosphorylation restores transcriptional activity and protects primary neurons from apoptosis induced by Cdk5 and neurotoxin. 4,22 Phosphorylation by PKA and GSK3β represses transactivation properties of MEF2D as well. 23,24 As a transcription factor, MEF2D is predominantly expressed in the nucleus.
Early study shows modulation of MEF2D activity by chaperonmediated autophagy. By interacting with Hsc70, MEF2D is observed shuttling from the nucleus to the cytoplasm to undergo degradation in mouse midbrain dopaminergic progenitor cells. 25 MEF2D subcellular localization might also change upon cell stress.
Rotenone, a neurotoxin, induced an up-regulation of MEF2D in the nucleus which might be protective against mitochondrial dysfunction and oxidative stress. 26 MEF2D level in both the cytoplasm and the nucleus is lower in nigral neurons of PD patients than that of AD and the control group, probably associating with α-synuclein aggregates. 27 Given modulation of MEF2 activity is partially determined by phosphorylation at distinct sites, phosphorylation of MEF2D contributing to its export from the nucleus deserves consideration.
Dual-specificity tyrosine-regulated kinase 1A (DYRK1A), abundant in neuronal cells, is widely involved in a variety of diseases including human malignancies, such as haematological and brain cancers. [28][29][30][31] DYRK1A was observed to accumulate in a punctate pattern in the nucleus of transfected cos-7 and HEK293 cells. 32 Identified as regulators of NFAT, DYRK1 and DYRK2 regulate NFAT subcellular localization by directly phosphorylating the conserved serine-proline repeats3 (SP-3) motif of the NFAT regulatory domain, exerting phosphorylation of SP-2 and serine-rich region 1(SRR-1) motifs by GSK3 and CK1. 33 DYRK1A phosphorylates GLI1 at Ser408 and may facilitate GLI1 nuclear localization. 34,35 Cytoplasmic CDKL5 expression is decreased upon phosphorylation at Ser308 by DYRK1A. 36 DYRK1A phosphorylates FKHR at Ser329 and reduces the proportion of FKHR in the nucleus. 37 Obviously, phosphorylation state of protein strongly affects its subcellular localization. Our previous study showed that DYRK1A was degraded by E3 ligase SCF βTRCP . 38 We also proved that DYRK1A phosphorylates NFATc1 and increases its protein stability. 39 Recently, we showed that DYRK1A phosphorylates and increases insulin receptor substrate-1 expression and regulates insulin signalling. 40 We previously demonstrated that MEF2D upregulated expression of DYRK1A through binding to its responsive element in DYRK1A promoter region and enhancing its transcriptional activity. 14 Their expressions were strongly correlated during mice brain development. The kinase activity of DYRK1A may also be induced by MEF2D in glioblastoma cell lines. 14 Therefore, interaction of DYRK1A and MEF2D needs to be further addressed.
Here we demonstrate that DYRK1A significantly increases MEF2D protein expression. DYRK1A and MEF2D are co-localized mainly in the nucleus. DYRK1A directly binds to and phosphorylates MEF2D at Serine 251, reducing its transcriptional activity.
DYRK1A affects MEF2D subcellular localization without affecting its protein stability. These results confirm the interaction between DYRK1A and MEF2D, and highlight their potential roles in glioblastoma development.
Cells were maintained at 37°C in an incubator containing 5% CO 2 .
Harmine (Aladdin) was added into culture medium of U87MG cells for 24 hours with concentration at 10 μmol/L.   Samples were analysed on 10% glycine SDS-PAGE.

| Site-directed mutagenesis
Briefly, primers covering mutated sites were designed to amplify linear fragment. pCMV6-entry-MEF2D and pCMV6-entry-DYRK1A vectors were used as the templates for PCR. PCR products were digested with DpnI at 37°C for 1-2 hour and transfected into DH5α competent cells. Single bacteria colony was amplified in LB media. Plasmid was purified and sent for sequencing.

| Real-time quantitative PCR
Total RNA was isolated from U87MG cells by TRIzol reagent (Sigma).
10-40 cycles of PCR were performed to cover the linear range of the PCR amplification. The real-time quantitative PCR was achieved by ABI 7900HT Fast Real-time PCR system (Applied Biosystems) with SYBR ® Green-based gene expression analysis. A comparative CT method (2 −ΔΔCT ) was used to analyse the gene expression level. The sequences of primers for real-time quantitative PCR were listed in Table 1.

| Luciferase assay
HEK293 cells and U87MG cells were seeded into 48-well plates and transfection was performed next day. Cells were lysed 48 hours after transfection. Dual-luciferase assays were achieved following the protocol supplied by dual-luciferase reporter assay kit (Promega).
Firefly luciferase activity was standardized by renilla luciferase activity and expressed as relative luciferase units.

| Mass spectrometry analysis
Sample Mixture was separated on 10% glycine gel at 120 V for 90 minutes. Upon electrophoresis, proteins were fixed within a polyacrylamide matrix by incubating the entire gel in 5% (vol/vol) acetic acid in 1:1 (vol/vol) water:methanol. Coomassie Brilliant Blue (0.05% in 5% (vol/vol) acetic acid in 1:1 (vol/vol) water:methanol solution) was used for gel staining. After destaining, bands were cut out and in-gel digestion was performed as described previously. 45 Mass spectrometry analysis was performed at Shanghai Institutes for Biological Sciences. M/z values of cleaved peptides were examined and calculated. Mascot was used to identify phosphorylated peptide and their non-phosphorylated counterparts.

| Cycloheximide (CHX) pulse-chase assay
CHX pulse-chase assay was performed as previously described. 38 Briefly, HEK293 cells were transfected with MEF2D and MEF2D mutant vectors, respectively. Twelve hours after transfection, HEK293 cells were seeded in 6-well plates. Thirty-six hours after transfection, cells were treated with 150 μg/mL CHX and harvested every 12 hours for Western blotting analysis.

| Statistical analysis
Data are presented as mean ± standard deviation (SD) from three independent experiments. For immunoblotting, one representative picture was shown. Quantifications from three independent experiments were defined with blots density by ImageJ software. The data were evaluated analysed for statistical significance with analysis of variance or non-parametric analysis by Prism 7 (GraphPad Software, Inc, San Diego, CA, USA). Differences were classified as significant at P < .05.

| DYRK1A phosphorylates MEF2D and increases MEF2D expression
DYRK1A plays crucial roles in brain function and tumorigenesis.
Biological activity of DYRK1A is generally ascribed to the phosphorylation of substrates. We find that overexpression of DYRK1A in HEK293 cells leads to a significant increase of MEF2D protein but not mRNA ( Figure 1A Figure 1E). Silencing DYRK1A by siRNA leads to decreased MEF2D protein to 62.58 ± 6.6% compared with the control ( Figure 1G). Harmine, a specific inhibitor of DYRK1, evidently decreases MEF2D protein level to 52 ± 5.6% compared with the control, which further verifies the induction of MEF2D protein by DYRK1A ( Figure 1I

| DYRK1A directly interacts with MEF2D
Since MEF2D expression is up-regulated by DYRK1A, the interaction between them needs to be further addressed. pCMV6-entry-

MEF2D is transfected into HEK293 cells. Co-immunoprecipitation
(Co-IP) by anti-flag antibody shows that MEF2D directly pulls down DYRK1A (Figure 2A). Co-IP is also applied to HEK293 cells and DYRK1A are co-localized in U87MG cells as well ( Figure 2E).
Results above provide strong evidence that DYRK1A and MEF2D might physically interact with each other.

| DYRK1A is responsible for phosphorylation of MEF2D at Ser251
Considering possible phosphorylation of MEF2D by DYRK1A, phosphorylation sites on MEF2D need to be identified. Recombinant protein purification combined with in vitro kinase assay was applied to obtain phosphorylated MEF2D. The band was clearly detected by Coomassie Brilliant Blue staining on SDS-PAGE ( Figure 3A).
The gel was fixed and the gel location of MEF2D band was cut for mass spectrometry (MS) analysis. Analysis by Mascot showed that MEF2D was phosphorylated at Ser251 upon DYRK1A based on m/z of five different peptides after in-gel digestion. The phosphorylation was not detected from control sample ( Figure 3B-D).

| Phosphorylation by DYRK1A inhibits transcriptional activity of MEF2D
Accumulative studies have demonstrated that phosphorylation of MEF2D could affect its transcriptional activity. 18 Figure 4C). This result is also confirmed by qPCR in U87MG cells.
MEF2D-S251A markedly increases HDAC9 and ZEB1 mRNA expression compared with wild type MEF2D while MEF2D-S251D modestly abolishes the regulation enhancement ( Figure 4D,E). These results suggest that transcriptional activity of MEF2D is significantly inhibited by DYRK1A through phosphorylation at Ser251.   20 In our study, we detect higher luciferase activity of MEF2DS251A and lower luciferase activity of MEF2DS251D compared with wild type MEF2D in dual-luciferase assay. It proves Ser251 is vital to MEF2D transcriptional activity. The activity of MEF2D was regulated by chaperone-mediated autophagy. 25 Phosphorylation by

| D ISCUSS I ON
Cdk5 is sufficient to promote degradation of MEF2D through activation of caspase-3. 22 No significant change to the caspase-dependent cleavage of MEF2D by DYRK1A was detected (data not shown).
Cytoplasmic MEF2D was detected in subcellular fractionation assay in the presence of DYRK1A. MEF2D is present in rodent neuronal mitochondria and enhances complex I activity by promoting ND6 transcription. 53  Aberrant expression of DYRK1A is detected in glioblastomas tissue.
Inhibition of DYRK1A promotes EGFR degradation in primary glioblastoma cell lines and neural progenitor cells, sharply reducing the self-renewal capacity of normal and tumorigenic cells. 29 Another study found that DYRK1A and DYRK1B kinases phosphorylate ID2 on threonine 27 (Thr27), leading to HIF2α destabilization, loss of glioma stemness and inhibition of tumour growth. 30 As DYRK1A is a potential therapeutic target to glioblastoma growth, more work needs to be achieved to profoundly elucidate underlying mechanism.

ACK N OWLED G EM ENTS
This study was supported by grants from Nature Science Foundation of China (81371226).

CO N FLI C T O F I NTE R E S T
The authors declare that they have no conflicts of interest with the contents of this article.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are openly available in this article.