ATM-dependent phosphorylation of MEF2D promotes neuronal survival after DNA damage

Abstract

Mutations in the ataxia telangiectasia mutated (ATM) gene, which encodes a kinase critical for the normal DNA damage response, cause the neurodegenerative disorder ataxia-telangiectasia (AT). The substrates of ATM in the brain are poorly understood. Here we demonstrate that ATM phosphorylates and activates the transcription factor myocyte enhancer factor 2D (MEF2D), which plays a critical role in promoting survival of cerebellar granule cells. ATM associates with MEF2D after DNA damage and phosphorylates the transcription factor at four ATM consensus sites. Knockdown of endogenous MEF2D with a short-hairpin RNA (shRNA) increases sensitivity to etoposide-induced DNA damage and neuronal cell death. Interestingly, substitution of endogenous MEF2D with an shRNA-resistant phosphomimetic MEF2D mutant protects cerebellar granule cells from cell death after DNA damage, whereas an shRNA-resistant nonphosphorylatable MEF2D mutant does not. In vivo, cerebella in Mef2d knock-out mice manifest increased susceptibility to DNA damage. Together, our results show that MEF2D is a substrate for phosphorylation by ATM, thus promoting survival in response to DNA damage. Moreover, dysregulation of the ATM-MEF2D pathway may contribute to neurodegeneration in AT.

Keywords: ATM; DNA damage; MEF2D; ataxia telangiectasia; neuronal survival; phosphorylation.

Figure 1.

ATM phosphorylates MEF2D. A, MEF2D contains conserved domains for DNA binding (at the N terminus), dimerization, and transactivation (at the C terminus). Vertical lines indicate potential ATM-phosphorylatable sites in MEF2D. B, ATM phosphorylates MEF2D in vitro. Cell extracts from HEK293T cells transfected with wt- or kd-ATM cDNAs were immunoprecipitated with anti-Flag M2 antibody. Immunoprecipitants were then incubated with γ-³²P ATP and recombinant GST-p53 or His-tagged MEF2D fusion proteins. C, DNA-damaging agents increase MEF2D phosphorylation in cerebellar granule cells. Cells were exposed to IR, etoposide (Eto), or control conditions. Alkaline phosphatase (AP) was added after immunoprecipitation (IP) in the indicated samples. D, DNA-damaging agents that induce DSBs trigger MEF2D phosphorylation in cultured cerebellar granule cells. Cells were exposed to IR, Eto, staurosporine (STS), UV, or control conditions. E, MEF2D phosphorylation was detected in wt-ATM cells but not in ATM-deficient cells. Cerebellar granule cells cultured from wt and knock-out mice (Atm^−/−) were exposed to etoposide (Eto) or control conditions. Left, ATM-phosphorylated MEF2D proteins were detected by immunoblotting (IB) with anti-phospho-SQ/TQ substrate antibody; total MEF2D protein expression was detected by immunoblotting with anti-MEF2D antibody. Right, Densitometric analysis of immunoblots was performed, and the relative densitometric values are presented as mean ± SEM (n = 3 independent experiments). *p < 0.01 (t test with Bonferroni correction). Densitometric values of bands from untreated ATM-deficient cells were arbitrarily set equal to 1, and other values were normalized to this reference point. F, Irradiation induced MEF2D phosphorylation in brains from wt mice but not from Atm knock-out mice. Adult brains of wt and Atm knock-out (Atm^−/−) mice were exposed in vivo to 10 Gy IR or control conditions (unexposed). Left, ATM-phosphorylated MEF2D proteins were detected by immunoblotting (IB) with anti-phospho-SQ/TQ substrate antibody; total MEF2D protein expression was detected by immunoblotting with anti-MEF2D antibody. Right, Densitometric analysis of immunoblots was performed, and the relative densitometric values are presented as mean ± SEM (n = 3 independent experiments). *p < 0.01 (t test with Bonferroni correction). Densitometric values of bands from brains in untreated Atm knock-out mice were arbitrarily set equal to 1, and other values were normalized to this reference point.

Figure 2.

ATM potentiates MEF2 activity in cells after exposure to DNA-damaging agents. A, ATM upregulates MEF2D activity in GAL4-dependent luciferase reporter gene assays. HEK293T cells were cotransfected with two reporter constructs, GAL4-luc and Renilla, plus an expression vector encoding GAL4-MEF2D and either wt- or kd-ATM. Cells were exposed to 10 Gy IR or control conditions before measuring luciferase activity (n = 3 independent experiments). *p < 0.01 (t test). Data are mean ± SEM. B, shRNAs directed against ATM decrease ATM expression. Cerebellar granule cells were transfected with shRNA-1, shRNA-2, or scrambled shRNA. Cells were immunostained with anti-ATM antibody, and Hoechst dye 33342 to visualize cell nuclei. Arrows indicate cells transfected with shRNA, and arrowheads and white outlines indicate untransfected cells. Scale bar, 6 μm. C, ATM shRNAs suppress the potentiation of MEF2 activity after exposure to DNA-damaging agents. Cerebellar granule cells were cotransfected with two reporter constructs, MEF2-luc and Renilla, plus shRNA-1, shRNA-2, or scrambled shRNA. Cells were exposed to IR, etoposide (Eto), or control conditions before measuring luciferase activity (n = 3 independent experiments). *p < 0.01 (ANOVA). D, Cerebellar granule cells from Mef2d^−/− mice were cotransfected with two reporter constructs, MEF2-luc and Renilla, and exposed to IR, etoposide (Eto), or control conditions before measuring luciferase activity (n = 3 independent experiments). p > 0.05 (ANOVA). NS, Not statistically significant. Data are mean ± SEM. E, KU55933, a small-molecule inhibitor of ATM, prevented activation of ATM kinase activity after exposure to IR. Cerebellar granule cells were immunostained with anti-phospho-SQ/TQ substrate antibody after exposure to IR in the presence or absence of KU55933. Hoechst dye 33342 was used to visualize cell nuclei. Scale bar, 30 μm. F, KU55933 abrogated potentiation of MEF2 activity after exposure to IR. Cerebellar granule cells were cotransfected with two reporter constructs, MEF2-luc and Renilla, and exposed to IR in the presence or absence of KU55933, following which luciferase activity was measured (n = 3 independent experiments). *p < 0.01 (ANOVA). Data are mean ± SEM.

Figure 3.

ATM complexes with and phosphorylates MEF2D at specific sites. A, Putative ATM-interaction motif in MEF2D. EE/DD/RY doublets are highlighted in red. mMEF2D, Mouse MEF2D; hMEF2D, human MEF2D. B, Coimmunoprecipitation of endogenous ATM with MEF2D after exposure to IR. Left, HEK293T cells were transfected with an expression vector encoding V5-tagged wt-MEF2D. After exposure to10 Gy IR or control conditions, cell extracts were immunoprecipitated (IP) with anti-ATM, anti-V5, or control anti-IgG antibody. Immunoblots (IB) were probed with anti-ATM antibody. Right, Densitometric analysis of immunoblots was performed, and the relative densitometric values are presented as mean ± SEM; n = 3 independent experiments. *p < 0.01 (ANOVA). NS, Not statistically significant. C, Identification of ATM-phosphorylatable sites in MEF2D. HEK293T cells were transfected with plasmids expressing wt or MEF2D mutants (T259A, S275A, S294A, S314A, or T259A/S275A/S294A/S314A), each tagged with V5. Transfected cells were then exposed to 10 Gy IR and immunoprecipitated with anti-V5 antibody. Top, ATM-phosphorylated MEF2D proteins were detected with anti-phospho-SQ/TQ substrate antibody. Total MEF2D protein expression was detected with anti-V5 antibody. Bottom, Densitometric analysis of immunoblots was performed, and the relative densitometric values are presented as mean ± SEM; n = 3 independent experiments. *p < 0.01 (ANOVA). The densitometric values of treated cells that were transfected with wt-MEF2D were arbitrarily set equal to 1, and other values were normalized to this reference point.

Figure 4.

Effect of phosphomimetic MEF2D and nonphosphorylatable MEF2D mutations on MEF2 activity. A, Each of the four ATM-phosphorylatable sites in MEF2D contributes to full activation after DNA damage. HEK293T cells were cotransfected with the following expression plasmids: two reporter constructs, MEF2-luc and Renilla, plus wt MEF2D, nonphosphorylatable MEF2D mutant (T259A/S275A/S294A/S314A), MEF2D mutant containing three putative ATM-phosphorylatable sites and a single alanine substitution (T259A, S275A, S294A, or S314A), or pcDNA3 vector control. One day after transfection, cells were exposed to etoposide (Eto) or control (untreated) conditions before measuring luciferase activity (n = 3 independent experiments). *p < 0.01 (ANOVA). NS, Not statistically significant. Data are mean ± SEM. B, Effect of nonphosphorylatable MEF2D mutations on MEF2 activity after serum induction. NIH 3T3 cells were cotransfected with two reporter constructs, MEF2-luc and Renilla, plus wt-MEF2D, nonphosphorylatable MEF2D mutant (T259A/S275A/S294A/S314A), or pcDNA3 vector control. The transfected cells were serum starved for 30 h and then treated with serum for 2 h before preparation of cell lysates for luciferase assays (n = 3 independent experiments). *p < 0.01 (ANOVA). Data are mean ± SEM. C, Effect of phosphomimetic MEF2D mutations on MEF2 activity in HEK293T cells. Cells were cotransfected with two reporter constructs, MEF2-luc and Renilla, plus wt-MEF2D, nonphosphorylatable MEF2D mutant (T259A/S275A/S294A/S314A), phosphomimetic MEF2D mutant (T259D/S275D/S294D/S314D), or pcDNA3 vector control. One day after transfection, transfected cells were exposed to etoposide or control (untreated) conditions before measuring luciferase activity. Basal luciferase activity of the untreated vector control was arbitrarily set to 1, and all other values were normalized to this reference point (n = 3 independent experiments). *p < 0.01 (ANOVA). Data are mean ± SEM. D, Effect of phosphomimetic MEF2D mutations on MEF2 activity after ATM knockdown in cerebellar granule cells. Cells were cotransfected with shRNA-1 and two reporter constructs, MEF2-luc and Renilla, plus wt-MEF2D, nonphosphorylatable MEF2D mutant (T259A/S275A/S294A/S314A), phosphomimetic MEF2D mutant (T259D/S275D/S294D/S314D), or pcDNA3 vector control. Three days after transfection, cells were exposed to etoposide or control (untreated) conditions before measuring luciferase activity. Basal luciferase activity of the untreated vector control was arbitrarily set to 1, and other values were normalized to this reference point (n = 3 independent experiments). *p < 0.01 (ANOVA). Data are mean ± SEM. Protein expression levels of wt and MEF2D mutant constructs are shown to the right of each panel.

Figure 5.

Effect of shRNA on endogenous MEF2D expression in cerebellar granule cells and construction of shRNA-resistant forms of MEF2D cDNAs. A, Cerebellar granule cells from rat were transfected with shRNA-MEF2D-436 or scrambled shRNA. Two days after transfection, cells were immunostained with anti-MEF2D antibody to detect MEF2D and Hoechst dye 33342 to visualize cell nuclei. Arrows indicate cells transfected with shRNA, and arrowheads indicate untransfected cells. Scale bar, 6 μm. B, MEF2D cDNAs bearing silent mutations in the region targeted by shRNA-MEF2D-436 manifest resistance to RNAi. HEK293T cells were transfected with plasmids expressing shRNA-MEF2D-436 plus an shRNA-resistant form of wt-MEF2D (designated wt-MEF2D-1–5). Additionally, cells were transfected with wt-MEF2D plus scrambled shRNA or shRNA-MEF2D-436 as a control. Two days after transfection, cell extracts were immunoprecipitated (IP) with anti-V5 antibody. MEF2D expression was then detected by immunoblotting (IB) with anti-V5 antibody. C, DNA damage induces apoptosis after MEF2D knockdown in rat cerebellar granule cells in vitro. Cerebellar granule cells from rat expressing MEF2D shRNA were more susceptible to etoposide-induced cell death. Cells were cotransfected with the indicated expression plasmids, and after 3 d exposed to etoposide (Eto, 10 μm for 8 h) or control conditions (untreated). The number of GFP-transfected cells was counted for each condition. Hoechst dye 33342 was used to identify condensed nuclei, indicative of apoptotic cell death (n > 300 cells counted in three independent experiments). *p < 0.05 (ANOVA). NS, Not statistically significant. shRNA-R is the shRNA-resistant form. Representative cells expressing MEF2CA or MEF2DN (no treatment) are illustrated. Data are mean ± SEM. NP, Nonphosphorylatable; P, phosphomimetic.

Figure 6.

Mef2d^−/− cerebellar granule cells manifest increased susceptibility to DNA damage. A, Cultured cerebellar granule cells from Mef2d wt and knock-out mice were exposed to etoposide (Eto) or control conditions. Hoechst dye 33342 was used to identify condensed nuclei, indicative of apoptotic cell death (n > 200 cells counted in three independent experiments). *p < 0.05 (ANOVA). B, Cultured cerebellar granule cells from Mef2d wt and knock-out mice were transfected with a GFP expression construct, plus either phosphomimetic MEF2D or vector control. The number of GFP-transfected cells was counted for each condition. Hoechst dye 33342 was used to identify condensed nuclei (n > 200 cells counted in three independent experiments). p > 0.05 (t test). NS, Not statistically significant. C, Cultured cerebellar granule cells from Mef2d wt and knock-out mice were transfected with a GFP expression construct, plus either phosphomimetic MEF2D or vector control, and exposed to Eto. The number of GFP-transfected cells was counted for each condition. Hoechst dye 33342 was used to identify condensed nuclei (n > 200 cells counted in three independent experiments). *p < 0.05 (ANOVA).

Figure 7.

Mef2d^−/− mice are more susceptible to DNA damage induced by IR. A, Brains of wt and Mef2d knock-out mice (Mef2d^−/−) at postnatal day (P) 18 were exposed to 10 Gy IR or control conditions (untreated), and cerebellar brain slices were subsequently labeled with TUNEL. Arrows indicate TUNEL-positive cells. Scale bar, 50 μm. B, Quantification of TUNEL-positive cells in Mef2d^−/− brains shows increased cell death after IR than in wt brains (n > 100 cells counted in three independent experiments). *p < 0.05 (ANOVA). NS, Not statistically significant. Data are mean ± SEM. C, DNA damage increases mitochondrial superoxide formation in wt and Mef2^−/− cerebellar brain slices. Increased fluorescence of the mitochondrial superoxide specific dye, MitoSOX, in brain slices in response to 10 Gy IR. Slices visualized under epifluorescence microscopy. Scale bar, 400 μm.

Figure 8.

MEF2D upregulates Bcl-xL expression after exposure to IR. A, qRT-PCR analysis of Bcl-xL mRNA expression level. Brains of wt and Mef2d knock-out mice (Mef2d^−/−) were exposed to 10 Gy IR or control conditions (untreated), and cerebellar brain mRNAs subsequently prepared for qRT-PCR. Expression of untreated wt samples was arbitrarily set to a value of 100, and other values normalized to this reference point with GAPDH as an internal control (n = 3 independent experiments). p < 0.05 (ANOVA). NS, Not statistically significant. Data are mean ± SEM. B, Western blot analysis of cerebellar brain lysates from wt and Mef2d^−/− mice after 10 Gy IR with Bcl-xL and actin protein expression levels quantified. Expression of untreated wt samples was arbitrarily set to a value of 100, and other values normalized to this reference point with actin as the control (n = 3 independent experiments). *p < 0.01 (ANOVA). NS, Not statistically significant. Data are mean ± SEM. C, Effect of phosphomimetic MEF2D mutations on Bcl-xL promoter activity in Mef2d^−/− cerebellar granule cells. Cells were cotransfected with two reporter constructs, Bcl-xL-luc and Renilla, plus wt-MEF2D, nonphosphorylatable MEF2D mutant (T259A/S275A/S294A/S314A), phosphomimetic MEF2D mutant (T259D/S275D/S294D/S314D), or pcDNA3 vector control. The transfected cells were exposed to 10 Gy IR or control (unexposed) conditions before measuring luciferase activity. Basal luciferase activity of the untreated vector control was arbitrarily set to 1, and other values were normalized to this reference point (n = 3 independent experiments). *p < 0.01 (ANOVA). Data are mean ± SEM.

Figure 9.

Schema for ATM-dependent phosphorylation/activation of MEF2D enhancing neuronal survival in response to DNA damage. ATM is activated by DSBs resulting from DNA-damaging agents. In the face of reparable DNA damage, activation of ATM leads to (1) formation of an ATM–MEF2D complex and (2) resulting phosphorylation/activation of MEF2. Potentiation of MEF2 activity via ATM phosphorylation promotes cell survival after DNA damage.