Specific acetylation of p53 by HDAC inhibition prevents DNA damage-induced apoptosis in neurons

Abstract

Histone deacetylase (HDAC) inhibitors have been used to promote neuronal survival and ameliorate neurological dysfunction in a host of neurodegenerative disease models. The precise molecular mechanisms whereby HDAC inhibitors prevent neuronal death are currently the focus of intensive research. Here we demonstrate that HDAC inhibition prevents DNA damage-induced neurodegeneration by modifying the acetylation pattern of the tumor suppressor p53, which decreases its DNA-binding and transcriptional activation of target genes. Specifically, we identify that acetylation at K382 and K381 prevents p53 from associating with the pro-apoptotic PUMA gene promoter, activating transcription, and inducing apoptosis in mouse primary cortical neurons. Paradoxically, acetylation of p53 at the same lysines in various cancer cell lines leads to the induction of PUMA expression and death. Together, our data provide a molecular understanding of the specific outcomes of HDAC inhibition and suggest that strategies aimed at enhancing p53 acetylation at K381 and K382 might be therapeutically viable for capturing the beneficial effects in the CNS, without compromising tumor suppression.

Figure 1.

Pharmacological HDAC inhibition protects primary cortical neurons from DNA damage-induced death. A, Mouse p53^+/+, p53^−/+, and p53^−/− cortical neurons (E15.5) were treated with or without CPT (10 μm) for 16 h. Neuronal viability was measured by MTT assay. Error bars indicate SD of mean from six independent experiments. **p < 0.01, significant protection compared with p53^+/+ (one-way ANOVA followed by Dunnett's multiple-comparisons test). B, Neuronal viability measured by MTT assay after treatment of mouse primary cortical cultures (E15.5) with CPT (10 μm) with or without TSA (0.67 μm) for 16 h. Error bars indicate SD of mean from six independent experiments. ***p < 0.001, significant difference between TSA- and non-TSA-treated groups; ^###p < 0.001, significant difference between CPT- and non-CPT-treated groups (two-way ANOVA followed by Bonferroni's test). C, Neuronal viability assessed by live/dead staining. Live cells are identified by green fluorescence (calceinAM); dead cells are identified by red fluorescence (ethidium homodimer, EthD-1). D, Neuronal viability measured by MTT assay after treatment of mouse primary cortical cultures (E15.5) with CPT (10 μm) or etoposide (ETP; 10 μm), with or without pan-HDAC inhibitor TSA (0.67 μm) or the class-I HDAC inhibitor sodium butyrate (NaBu; 5 mm) for 16 h. Error bars indicate SD of mean from six independent experiments. ***p < 0.001, significant difference between TSA- and non-TSA-treated groups or NaBu- and non-NaBu-treated groups; ^###p < 0,001, significant difference between CPT- and non-CPT-treated groups or ETP- and non-ETP-treated groups (two-way ANOVA followed by Bonferroni's test). E, Immunoblot analysis of phosphorylated H2AX (γH2AX), total H2AX, phosphorylated ATM (pS1981-ATM), total ATM, phosphorylated p53 (pS15-p53), and total p53 (pan-p53) in neurons treated with CPT (10 μm) with or without TSA (0.67 μm) for 8 h. F, Immunocytochemical analyses of γH2AX (red) and total H2AX (green) in neurons treated with CPT (10 μm) with or without TSA (0.67 μm) for 8 h. Anti-MAP2 (microtubule-associated protein 2; green) or anti-TUBB3 (Tubulin β-3; red) antibodies were used to define neurons. Nuclei were stained with DAPI (blue).

Figure 2.

DNA damage and HDAC inhibition differentially regulate the transcription of known p53 target genes. A, Heat map depicting gene expression in mouse primary cortical neurons (E15.5) treated with CPT (10 μm) with or without TSA (0.67 μm) for 12 h. Shown are the 42 putative p53 gene targets that were significantly regulated in at least one condition compared with untreated control (p < 0.05). Expression profiles are displayed in descending order, based on induction level by CPT. B–F, Quantitative RT-PCR of select p53 gene targets in neurons from p53^+/+ and p53^−/− mice treated with CPT (10 μm) with or without TSA (0.67 μm) for 8 h. Error bars indicate SD of mean from five independent experiments. **p < 0.01, *p < 0.05, significant difference compared with untreated control; ^♦♦♦p < 0.01, significant difference compared with CPT alone (one-way ANOVA followed by Bonferroni's test); ^###p < 0.001, significant difference compared with knock-out p53 (two-way ANOVA followed by Bonferroni's test).

Figure 3.

Abrogation of PUMA induction by TSA is sufficient to prevent DNA damage-induced neurodegeneration. A, Immunoblot analysis of PUMAα and Cdkn1a (p21) in neurons treated with CPT (10 μm) with or without TSA (0.67 μm) for 8 h. B, Immunoblot analysis of CytC in cytosolic and mitochondrial fractions from neurons treated with CPT (10 μm) with or without TSA (0.67 μm) for 8 h. COX IV, a mitochondrial marker, was used to assess for fraction purity. C, Immunocytochemical analyses of CytC (red) in neurons treated with CPT (10 μm) with or without TSA (0.67 μm) for 8 h. Anti-MAP2 was used to detect neurons (green). Nuclei were stained with DAPI (blue). D, Immunoblot analysis of Puma to verify genetic knockdown in neurons transfected with plasmids encoding shPuma or shCTRL and treated with or without CPT (10 μm) for 16 h. E, Neuronal viability in D was measured by MTT assay. Error bars indicate SD of mean from three independent experiments. ***p < 0.001, significant protection compared with shCTRL (one-way ANOVA followed by Bonferroni's test); ^###p < 0.001, significant difference between CPT- and non-CPT-treated groups (two-way ANOVA followed by Bonferroni's test).

Figure 4.

HDAC inhibition by TSA abrogates DNA damage-induced binding of p53 to Puma and p21 promoters, but it promotes the recruitment of Sp1 to the p21 promoter. A, Schematic of PUMA gene. The human gene contains three coding exons (exons 2–4) and two noncoding exons (exons 1a and b), which are conserved in mouse. Two p53 response elements (p53RE) are found in the regulatory region of both human and mouse genes. The asterisk indicates PCR primer binding sites. B, C, ChIPs for p53 (B) and acetylated histone H3 (C) in extracts from mouse primary cortical neurons (E15.5) treated with CPT (10 μm) with or without TSA (0.67 μm) for 8 h. Promoter occupancy was determined by quantitative real-time PCR using primers specific for p53RE (defined in A). Error bars indicate SD of mean from three independent experiments. *p < 0.05, significant difference compared with untreated control (one-way ANOVA followed by Bonferroni's test). D, Schematic of p21 (Cdkn1a) gene. The human gene contains two coding exons (exons 2 and 3) and one noncoding exon (exon 1), which are conserved in mouse. Two p53 response elements (p53RE) and six Sp1 binding sites (Sp1-BS) are found in the regulatory region of both human and mouse genes. Asterisks indicate PCR primer binding sites. E, F, ChIPs for p53, Sp1, and acetylated histone H3 in extracts from neurons treated with CPT (10 μm) with or without TSA (0.67 μm) for 8 h. Promoter occupancy was determined by quantitative PCR using primers specific for p53RE or Sp1-BS (defined in D). Error bars indicate SD of mean from three independent experiments. ***p < 0.001, **p < 0.01,*p < 0.05, significant difference compared with untreated control; ^###p < 0.001, significant difference compared with CPT (one-way ANOVA followed by Bonferroni's test).

Figure 5.

CPT-induced DNA damage and HDAC inhibition by TSA differentially modulate p53 acetylation. A, Schematic of p53 protein. Human p53 contains 393 amino acids, organized into three functional domains, that are conserved in mouse. The N terminus includes the transactivation domain (TAD) and the proline-rich domain (PRD). The central region contains the DNA-binding domain (DBD). The C-terminal region includes the tetramerization domain of p53 (4D), the C-terminal regulatory domain (CTD), and nuclear localization and export signals (NLS and NES). Lysines in the C-terminal region can be acetylated, affecting p53 protein stability and function. B, Immunoblot analysis of acetylated K320-p53, acetylated K373-p53, acetylated K381-p53, and acetylated K382-p53 (respectively, K317, K370, K376, and K379 in mouse) in mouse primary cortical neurons treated with CPT (10 μm) with or without TSA (0.67 μm) for 8 h. C, Immunocytochemical analyses of phosphorylated p53 (pS15), acetylated K320-p53, acetylated K381-p53, and acetylated K382-p53 (AcK320, AcK381, AcK382; green) in neurons treated with CPT (10 μm) with or without TSA (0.67 μm) for 8 h. TUBB3 (Tubulin β-3; red) was used to define neurons. Nuclei were stained with DAPI (blue). D, Scheme showing the lysine-to-arginine (deacetylation-mimic) and lysine-to-glutamine (acetylation-mimic) human p53 mutants generated and used in this study. The circle represents lysine substitution with arginine or glutamine. E, Immunoblot analysis of human p53 shows expression of wild-type p53 (WT), DNA-binding mutant p53 (R175H), or the different p53 acetylation mutants in mouse cortical neurons. eGFP was used as an electroporation control.

Figure 6.

Acetylation of p53 on lysines 381 and 382 abrogates its ability to induce apoptosis in neurons. A, Neuronal viability measured by MTT assay of cortical neurons prepared from p53^−/− mice (E15.5) 48 h after electroporation with plasmid DNA encoding the DNA-binding mutant (R175H), WT, or different acetylation mutants of p53. Neurons expressing WT p53 were treated with or without TSA (0.67 μm). eGFP was used as an electroporation control. Error bars indicate SD of mean from three independent experiments. ***p < 0.001, **p < 0.01, *p < 0.05, significant difference compared with eGFP; ^###p < 0.001, ^##p < 0.01, ^#p < 0.05, significant difference compared with wild-type p53 (WT) (one-way ANOVA followed by Bonferroni's test). B, Viability of p53-null neurons assessed by live/dead staining under the conditions described in A. Live cells are identified by green fluorescence (calceinAM), whereas dead cells are identified by red fluorescence (ethidium homodimer, EthD-1). C, Viability of p53^−/− neurons measured by live/dead staining after electroporation with plasmid DNA encoding DNA-binding mutant (R175H), WT, or K381/382Q acetylation-mimic of p53 and treated with CPT (10 μm). eGFP was used as a control.

Figure 7.

Acetylation of lysines 381 and 382 abrogates the transactivating activity of p53 in neurons. A, Luciferase reporter assay of plasmid DNA encoding a fragment of the human PUMA proximal promoter (PUMA Frag1-Luc) in p53^−/− primary cortical neurons. Promoter reporter plasmid was cotransfected with DNA-binding mutant (R175H), WT, or different acetylation mutants of p53. WT p53-transfected neurons were treated with or without TSA (0.67 μm). Error bars indicate SD of mean from five independent experiments. ***p < 0.001, significant difference compared with R175H; ^###p < 0.001, ^#p < 0.05, significant difference compared with wild-type p53 (WT) (one-way ANOVA followed by Bonferroni's test). B, C, Chromatin immunoprecipitations for human p53 in extracts from mouse primary cortical neurons electroporated with p53 plasmids: R175H, WT, or the acetyl-mimics K381Q and K382Q. Promoter occupancy was determined by quantitative PCR using primers specific for p53RE (defined in Fig. 4A). Error bars indicate SD of mean from three independent experiments. ***p < 0.001; **p < 0.01; *p < 0.05, significant difference compared with R175H (one-way ANOVA followed by Dunnett's multiple-comparisons test). D, E, Quantitative RT-PCR of Puma (D) and p21 (E) expression in neurons electroporated with p53 plasmids: R175H, WT or the acetyl-mimics K381Q and K382Q. Error bars indicate SD of mean from three independent experiments. *p < 0.05, significant difference compared with R175H (one-way ANOVA followed by Dunnett's multiple-comparisons test).

Figure 8.

Acetylation of lysines 381 and 382 promotes the pro-apoptotic functions of p53 in proliferating tumor cell lines. A, HCT116, DLD1, and HT22 cells were transfected with an equal amount of plasmid DNA encoding wild-type human p53 (WT), DNA-binding mutant p53 (R175H), or the p53 acetylation mutants K373Q and K381/382Q. Immunoblot analysis was performed on total cell lysates 24 h after electroporation using antibodies specific for human p53. Plasmid DNA encoding eGFP was used as a transfection control. B, Luciferase reporter assay of plasmid DNA encoding a fragment of the human PUMA proximal promoter (PUMA Frag1-Luc) in human colorectal carcinoma cell lines HCT116 and DLD-1 and in mouse hippocampal cell line HT22. Promoter reporter plasmid was cotransfected with DNA-binding mutant p53 (R175H), wild-type p53 (WT), K373Q, or K381/382Q acetylation mutants of p53. Error bars indicate SD of mean from three independent experiments. ***p < 0.001, significant difference compared with R175H; ^###p < 0.001, ^##p < 0.01, significant difference compared with WT (one-way ANOVA followed by Bonferroni's test). C, Cell viability measured by MTT assay of HCT116, DLD-1, and HT22 cultures 24 h after transfection with R175H, WT, K373Q, or K381/382Q acetylation mutants of p53. Error bars indicate SD of mean from three independent experiments. ***p < 0.001, **p < 0.01, *p < 0.05, significant difference compared with R175H; ^##p < 0.01, significant difference compared with WT (one-way ANOVA followed by Bonferroni's test). D, HCT116 cell death assessed by ethidium homodimer staining (EthD-1; red) 24 h after transfection with R175H, WT, K373Q, or K381/382Q acetylation mutants of p53.

Figure 9.

HDAC inhibition does not prevent PUMA induction and DNA damage-induced death in HCT116 cells. A, Immunoblot analysis of pan-p53, phosphorylated p53 (pS15-p53), and acetylated p53 (AcK381-p53 and AcK382-p53) in extracts from HCT116 cells treated with CPT (20 μm) with or without TSA (1.3 μm) for 8 h. B, Quantitative RT-PCR of PUMA expression in HCT116 cells treated with CPT (20 μm) with or without TSA (1.3 μm) for 8 h. Error bars indicate SD of mean from three independent experiments. ***p < 0.001, significant difference compared with untreated control; ^##p < 0.01, significant difference compared with CPT (one-way ANOVA followed by Bonferroni's test). C, HCT116 cell viability measured by MTT assay after treatment with CPT (20 μm) with or without TSA (1.3 μm). Error bars indicate SD of mean from three independent experiments. ***p < 0.001, **p < 0.01, significant difference compared with untreated control; ^#p < 0.05, significant difference compared with CPT (one-way ANOVA followed by Bonferroni's test). D, HCT116 cell death assessed by ethidium homodimer staining (EthD-1; red) after treatment with CPT (20 μm) with or without TSA (1.3 μm).

Figure 10.

The acetylation pattern of p53 determines the fate of neurons in response to DNA damage and HDAC inhibition. A, In unstressed neurons, p53 C terminus is in a steady state of acetylation, resulting from the balance of the dose and activity of histone acetyltransferases, zinc-dependent HDACs, and NAD-dependent histone deacetylases (SIRTs). B, DNA damage induced by camptothecin triggers p53 acetylation at lysine 373, which, along with p53 stabilization and phosphorylation at serine 15, promotes p53 binding to the pro-apoptotic PUMA promoter and leads to neuronal death. C, When DNA damage occurs in the presence of an HDAC inhibitor, DNA damage-induced lysine 373 acetylation occurs as in B, and lysines 320, 381, and 382 are acetylated. This acetylation pattern abrogates the ability of p53 to bind to the PUMA promoter and induce apoptosis in neurons but not in cancer cells.