Neuronal susceptibility to beta-amyloid toxicity and ischemic injury involves histone deacetylase-2 regulation of endophilin-B1

Abstract

Histone deacetylases (HDACs) catalyze acetyl group removal from histone proteins, leading to altered chromatin structure and gene expression. HDAC2 is highly expressed in adult brain, and HDAC2 levels are elevated in Alzheimer's disease (AD) brain. We previously reported that neuron-specific splice isoforms of Endophilin-B1 (Endo-B1) promote neuronal survival, but are reduced in human AD brain and mouse models of AD and stroke. Here, we demonstrate that HDAC2 suppresses Endo-B1 expression. HDAC2 knockdown or knockout enhances expression of Endo-B1. Conversely, HDAC2 overexpression decreases Endo-B1 expression. We also demonstrate that neurons exposed to beta-amyloid increase HDAC2 and reduce histone H3 acetylation while HDAC2 knockdown prevents Aβ induced loss of histone H3 acetylation, mitochondrial dysfunction, caspase-3 activation, and neuronal death. The protective effect of HDAC2 knockdown was abrogated by Endo-B1 shRNA and in Endo-B1-null neurons, suggesting that HDAC2-induced neurotoxicity is mediated through suppression of Endo-B1. HDAC2 overexpression also modulates neuronal expression of mitofusin2 (Mfn2) and mitochondrial fission factor (MFF), recapitulating the pattern of change observed in AD. HDAC2 knockout mice demonstrate reduced injury in the middle cerebral artery occlusion with reperfusion (MCAO/R) model of cerebral ischemia demonstrating enhanced neuronal survival, minimized loss of Endo-B1, and normalized expression of Mfn2. These findings support the hypothesis that HDAC2 represses Endo-B1, sensitizing neurons to mitochondrial dysfunction and cell death in stroke and AD.

Figure 1

HDAC2 regulates Endo‐B1 expression. A. HDAC2 knockdown increases Endo‐B1 protein levels. Postnatal cortical neurons were infected with lentivirus (10 MOI) expressing shRNA directed against HDAC2 or scrambled control shRNA with no known target for 3 days. Endo‐B1 expression was analyzed by Western blot along with corroborating the efficacy of HDAC2 shRNA. Blots are shown for a single experiment comparing a set of paired samples (control vs. HDAC2 shRNA) run on a gel (the vertical line represents irrelevant lanes that were cut off) and are representative of seven independent experiments. B. The levels of the Endo‐B1a and Endo‐B1b/c bands in HDAC2 shRNA‐treated neurons relative to control shRNA‐treated neurons were determined for individual blots, as represented in A, by densitometry. After normalization to actin, HDAC2 shRNA‐induced changes are presented as averaged “HDAC2 shRNA/control shRNA” ratios (%). **: P < 0.001, ***: P < 0.01 relative to control shRNA (N = 7). C. HDAC2 knockdown does not elevate Endophilin B2 levels in neurons. Wild‐type neurons were infected with control shRNA or HDAC2 shRNA (10 MOI) for 3 days. Extracts were probed for Endophilin B2 (SH3GLB2), and the band intensity was normalized to actin (N = 3). D. Endo‐B1b/c protein is elevated in HDAC2KO (−/−) neurons relative to wild‐type (+/+) neurons in culture as analyzed by Western blot. Blots are shown for a single experiment comparing a set of paired samples (−/− vs. +/+) run on a gel (the vertical line represents irrelevant lanes that were cut off) and are representative of 3 independent experiments. E. The levels of the Endo‐B1a and Endo‐B1b/c bands in wild‐type and HDAC2KO neurons in culture were determined for individual blots, as represented in D, by densitometry. After normalization to actin, the results are presented as averaged “HDAC2KO/HDAC2 wild‐type” ratios (%). ***: P < 0.01 relative to wild‐type neurons (N = 3). F. Endo‐B1b/c protein levels in cortex are elevated in HDAC2KO mice. Cortical tissue from three‐month‐old wild‐type and HDAC2KO animals was dissected out from a 2‐mm coronal brain slice at the level of the bregma and protein lysates were assessed for Endo‐B1b/c and Endo‐B‐1a expression by Western blot. The results are presented as averaged “HDAC2KO/wild‐type” ratios (%). ***: P < 0.001 relative to wild‐type animals (N = 6 animals).

Figure 2

HDAC2 overexpression reduces Endo‐B1 levels and promotes caspase‐3 activation. A. Protein extracts from HDAC2KO cortical neurons infected with lentivirus expressing EGFP (5 MOI) or with increasing doses of HDAC2‐expressing lentivirus as indicated for 4 days were probed for active caspase‐3, Endo‐B1, HDAC2, and actin. B. Densitometric quantification of Western blots demonstrates that HDAC2 overexpression significantly suppresses Endo‐B1, both “a” and “b/c” isoforms, and increases active caspase‐3. The results are presented as averaged ratios of HDAC2 relative to EGFP overexpression. ***: P < 0.001 vs. EGFP control (N = 3).

Figure 3

HDAC2 deficiency protects primary cortical neurons from Aβ‐induced cell death. A. Oligomerized Aβ_1‐42 peptide (10 μM) induces caspase‐3 activation, which is suppressed by Endo‐B1c overexpression, but exacerbated by Endo‐B1 knockdown. Protein extracts, obtained 24 h after Aβ_1‐42 treatment, from cultured cortical neurons that were infected as indicated were analyzed for Endo‐B1, activated caspase‐3 and actin by Western blot. Representative of N = 2. B. HDAC2 shRNA reduces caspase‐3 activation induced by Aβ_25‐35. Neurons were infected as indicated for 3 days and treated with Aβ_25‐35 (toxic) or Aβ_35‐25 (control) for 24 h. The level of active caspase‐3 was analyzed by Western blot. *: P < 0.001 compared to Aβ_35‐25/control shRNA (N = 3). N.S.: Not significantly different from Aβ_35‐25/control shRNA (N = 3). C) HDAC2 shRNA suppresses Aβ_25‐35‐induced neuronal death. Dead and viable neurons were counted based on morphology (see methods) in cultures infected and treated as indicated. Data represent the percent change in neuronal death at 24 h after treatment with Aβ_25‐35 with or without preceding HDAC2 knockdown relative to treatment with control shRNA followed by the control Aβ_35‐25 peptide. ***: P < 0.001 compared to Aβ_25‐35 plus control shRNA (N = 3). D. Neurotoxic Aβ_25‐35 elevates HDAC2 protein expression in cultured cortical neurons compared to Aβ_35‐25 (control). Aβ‐induced changes in HDAC2 levels are presented as % changes relative to non‐treated control. *: P < 0.05 (N = 4). E. HDAC2 shRNA prevents Aβ_25‐35‐induced reduction in histone H3 acetylation. Neurons were treated as in (B) and assayed for N‐terminally acetylated histone H3 and actin expression by Western blot. Aβ‐induced changes in acetylated histone H3 levels are presented as % changes relative to non‐treated control. *: P < 0.05, N.S.: not significant (N = 3). F. HDAC2KO cortical neurons show reduced levels of caspase‐3 activation compared with wild‐type neurons when treated with Aβ_25‐35 for 24 h. The blots are representative of 3 independent experiments.

Figure 4

HDAC2 mediates Aβ toxicity via Endo‐B1. A. Wild type postnatal cortical neurons infected with lentivirus expressing either control or HDAC2 shRNA and treated with Aβ_25‐35 (toxic) or Aβ_35‐25 (control) peptide were processed for Endo‐B1 and actin expression by Western blot. B. Expression levels are presented as average fold changes by normalizing to Aβ_25‐35/control shRNA (set at 1.0) in individual experiments after normalization to actin. Aβ_25‐35 significantly reduced Endo‐B1b/c levels, which was prevented by HDAC2 shRNA pretreatment. *: P < 0.05, N.S.: not significant (N = 4). C. Wild‐type neurons were infected with lentivirus expressing control, HDAC2 and/or Endo‐B1 shRNA and treated with Aβ_25‐35 or control Aβ_35‐25 peptide. Dead neurons were counted 24 h after treatment. Data represents the percent change in neuronal death relative to control shRNA infected cultures treated with Aβ_35‐25. *: P < 0.05, N.S.: not significant (N = 3). D. Endo‐B1 knockout neurons infected with HDAC2 shRNA expressing lentivirus are substantially less protected from Aβ_25‐35‐induced death compared to WT neurons (shown in B). Endo‐B1 knockout neurons were infected with shRNA lentivirus as indicated and then treated with Aβ_25‐35. Data represents the percent change in neuronal death relative to control shRNA‐infected cultures treated with Aβ_35‐25. *: P < 0.05 (N = 3).

Figure 5

Aβ‐induced mitochondrial dysfunction is mediated by HDAC2. A. HDAC2 knockdown mitigates Aβ‐induced mitochondrial depolarization as revealed by JC‐1 fluorescence. Wild‐type cortical neurons were infected for shRNA expression as indicated and treated with Aβ_25‐35 or the control peptide Aβ_35‐25 for 24 h and then labeled with the JC‐1 dye. Higher red/[green + red] fluorescence indicates greater mitochondrial membrane potential. EGFP fluorescence, used as an infection marker for the control and HDAC2 shRNA virus, was very weak relative to JC‐1 green fluorescence and was therefore judged as negligible. The data are presented as the percentage of red relative to total (red + green) fluorescence intensity. *: P < 0.01, N.S.; not significant. (N = 3). B. HDAC2‐induced changes in the expression of mitochondrial fusion/fission proteins. The same blot used in Figrue 2A containing protein extracts from HDAC2KO cortical neurons infected with lentivirus expressing EGFP (5 MOI) or increasing MOI of HDAC2 lentivirus as indicated was additionally probed for Mfn1, Mfn2, Drp1, and Mff. C. Densitometric quantification of Western blots including the representative blot shown in B) performed for the infection condition of EGFP and 5 MOI HDAC2 lentivirus. Both Mff bands seen in the blot (marked by the bracket in B), which reflect alternatively spliced variants, were included for the quantification of Mff. Significant changes were detected for Mfn2 and Mff (*: P < 0.05, **: P < 0.001 vs. EGFP control). No significant change was detected for Mfn1 or Drp1. (N = 3).

Figure 6

Mfn2 and Mff expression are altered in cortical tissue from AD cases. Representative Western blot analysis of protein extracted from the inferior parietal lobule cerebral cortex of patients at Braak V–VI stages (AD) compared to age‐matched control, Braak I‐II patients (control). Expression levels after normalization to actin levels demonstrate statistically significant reduction of Mfn2 and elevation of Mff in AD cases compared to age matched controls. *: P < 0.01 (N = 10 patients per group).

Figure 7

HDAC2 modulates the response to ischemic injury. A. HDAC2KO (−/−) mice show reduced sensitivity to ischemic injury in the MCAO/R model. Infarct volume was determined by TTC staining after 45 minute MCAO followed by 48‐h reperfusion. *: P < 0.05 (N = 7–9 animals per condition). B–D. Ischemia‐induced suppression of Endo‐B1b/c and Mfn2 is not observed in HDAC2KO mice. Penumbra (ipsilateral) and corresponding uninjured (contralateral) tissue from animals subjected to MCAO/R as in A) was dissected out and protein expression was assessed by Western blot for Endo‐B1, Mfn2, and actin. Expression levels are shown after normalization to actin levels. Significant reduction in Endo‐B1b/c (C) and Mfn2 (D) ipsilateral to MCAO/R relative to the contralateral side was detected in wild‐type (WT) animals, but not in HDAC2KO mice. **: P < 0.01 (N = 3 animals per condition).