HDAC inhibitor protects chronic cerebral hypoperfusion and oxygen‐glucose deprivation injuries via H3K14 and H4K5 acetylation‐mediated BDNF expression

Abstract Vascular dementia (VaD) is the second most common cause of dementia, but the treatment is still lacking. Although many studies have reported that histone deacetylase inhibitors (HDACis) confer protective effects against ischemic and hypoxic injuries, their role in VaD is still uncertain. Previous studies shown, one HDACi protected against cognitive decline in animals with chronic cerebral hypoperfusion (CCH). However, the underlying mechanisms remain elusive. In this study, we tested several 10,11‐dihydro‐5H‐dibenzo[b,f]azepine hydroxamates, which act as HDACis in the CCH model (in vivo), and SH‐SY5Y (neuroblastoma cells) with oxygen‐glucose deprivation (OGD, in vitro). We identified a compound 13, which exhibited the best cell viability under OGD. The compound 13 could increase, in part, the protein levels of brain‐derived neurotrophic factor (BDNF). It increased acetylation status on lysine 14 residue of histone 3 (H3K14) and lysine 5 of histone 4 (H4K5). We further clarified which promoters (I, II, III, IV or IX) could be affected by histone acetylation altered by compound 13. The results of chromatin immunoprecipitation and Q‐PCR analysis indicate that an increase in H3K14 acetylation leads to an increase in the expression of BDNF promoter II, while an increase in H4K5 acetylation results in an increase in the activity of BDNF promoter II and III. Afterwards, these cause an increase in the expression of BDNF exon II, III and coding exon IX. In summary, the HDACi compound 13 may increase BDNF specific isoforms expression to rescue the ischemic and hypoxic injuries through changes of acetylation on histones.


| INTRODUC TI ON
Vascular dementia (VaD) is the most common of cognitive disorders in elderly persons after Alzheimer disease (AD). The causes of VaD are diverse and its pathophysiology is complicated. [1][2][3][4] Significant progress has been achieved in reducing stroke risk, but there remains a lack of effective treatment for VaD after stroke. Although numerous pre-clinical drugs have been tested, currently, only the use of intravenous thrombolytic agents in the ultra-acute stage 5 is approved by the FDA for stroke treatment because the mechanisms underlying chronic ischemia and hypoxia are not fully understood.
Chronic cerebral hypoperfusion (CCH), which is a major cause of VaD, has been shown to involve amyloid accumulation and inflammation. 1 Further clarification of the role in these pathophysiologic changes play in VaD and the development of a related therapy are warranted.
Epigenetic modification plays an important role in the pathophysiology of many diseases. Histone deacetylases (HDACs), along with histone acetyltransferases (HATs), regulate chromatin remodelling and subsequent gene transcription by controlling the status of histone acetylation. Histone deacetylation removes acetyl groups from histones and induces condensed chromatin conformation contributing to the suppression of gene transcription. 6 This phenomenon is involved in diverse physiological processes. In some brain disorders, such as ischemic stroke and AD, HDACs are dysregulated and play a role in the pathophysiology of these diseases. [7][8][9][10][11] This suggests that HDACs might be potential targets for the treatment of brain disorders. Recently, HDAC inhibitors (HDACis) have been used in therapy for acute brain injury and neurodegenerative diseases. 12 The administration of HDACis in the animal model of acute stroke was shown to reduce infarction size, suppress neuroinflammation and improve neurological deficits. It also decreased oedema and improved blood-brain barrier (BBB) disruption by inhibiting NF-κB activation and matrix metallopeptidase expression. 13 Moreover, HDACis could be given either before or even 3-h after ischemic injury. 14 In addition to stroke, a prior study pointed out that pan-HDA-Cis (Trichostatin) could effectively improve outcome in the animal model of multiple sclerosis by mitigating the inflammatory response, demyelination, and nerve or axonal damage. However, the effects of HDACis during post-stoke vascular cognitive impairment still needs to be investigated.
The manipulation of HDACis might serve as a novel therapeutic strategy to promote stroke recovery. For example, in a prospective human study, sodium valproate, which is a non-specific inhibitor of HDAC9 activation, was associated with a reduced risk of recurrent ischemic stroke. 15 Furthermore, knockdown of HDAC2 promoted the recovery of motor dysfunction by enhancing neuronal survival, increased neuroplasticity of surviving neurons, and reduced neuroinflammation.
Meanwhile, overexpression of HDAC2 may worsen stroke-induced functional impairment. 16 All of these findings support the idea that inhibition or enhancement of HDACs exerts a protective or detrimental effect in ischemic diseases, respectively.
Histone acetylation decreases along with ageing. This can cause abnormal synaptic plasticity and subsequent hippocampus-dependent memory decline. 17,18 The dysregulation of HDACs can also disrupt chromatin remodelling, which is associated with AD. 19,20 HDAC can regulate DNA expression and modify some proteins, such as amyloid precursor protein and amyloid-beta; in other words, it has an essential role in AD pathogenesis. 21 These data show that HDAC inhibition is involved in a variety of neuroprotective mechanisms in the ischemic brain and that HDAC inhibition is used as a therapeutic agent for the treatment of post-ischemic brain damage.
HDAC inhibition increases neurotrophic factors, brain-derived neurotrophic factor (BDNF) expression, and dendritic spine density during ethanol withdrawal. 22 This implies that BDNF activation may be beneficial in synaptic plasticity or memory consolidation. BDNF, a secreted protein of the neurotrophin family, binds to two differential receptors, TrkB and p75, which is followed by activation of respective downstream signalling pathways. 23,24 In addition to playing a role in synaptic plasticity, BDNF is critical for long-term potentiation, which makes it critical in learning and memory. One major study indicated that exogenous BDNF expression protects brain slices against oxygen-glucose deprivation (OGD). 25 Furthermore, lipophilic statin pitavastatin treatment exerts its neuroprotection in cultured cerebral neurons after OGD involving the BDNF-TrkB signalling pathway. 26 The regulation of BDNF expression has been proved to be associated with numerous neurodegenerative diseases, mood disorders and ageing but is not fully understood in relation to

VaD diseases.
CCH is closely associated with vascular cognitive impairment and VaD. [27][28][29] We have established a mouse model to induce CCH by occlusion of the bilateral common carotid artery (modified BCCAO). This model shows cognitive decline, hippocampal atrophy and chronic neuroinflammation. By contrast, OGD is widely used to study the cell model of stroke and is a simple but useful technique for not only elucidating the molecular mechanisms of ischemic injury but for developing neuroprotective drugs. In this study, we synthesized a series of 10,11-dihydro-5H-dibenzo[b,f]azepine hydroxamates as HDACis. These compounds were designed to confer the structural characteristics of the HDAC, including the zinc-binding domain, linker domain and hydrophobic-capping group. We screened and synthesized these HDACis by using OGD and identified compound 13, which was tested in CCH animals. This study also explored the possible mechanisms of VaD, focusing on the role of BNDF.
Food and water were given ad libitum to all mice throughout the experiments. Animal care and experimental procedures in this study were performed in accordance with the guidelines for the Care and Use of Laboratory Animals from the Ethics Committee of Taipei Medical University.

| BCCAO surgery
The modified-BCCAO surgery was performed by ligating the common carotid artery (CCA), with some modification to a previous procedure. 30 Briefly, a sagittal midline incision (approximately 1 cm in length) was made to expose the parietal skull to measure the baseline CBF using laser doppler flowmetry. Then, a cervical midline neck incision (approximately 1 cm in length) was made, both CCAs were carefully separated, and the right CCA was occluded.
One week later, the procedure was repeated for the left CCA. The steps were similar, except there was transient ligation of the left CCA for 30 minutes by tightening the silk sutures. The CBF was measured to ensure that there was a reduction of CBF by 80%-90% from the baseline. The procedure of sham surgery was similar to BCCAO but involved no ligation of both CCAs. DMSO or compounds were given intraperitoneally (25 mg/mL) once every 2 days for either 1 or 3 months, beginning 2 days after CCA ligation. The mice were grouped as sham-control (n = 12), DMSO with BCCAO (n = 12) and compound 13 or SAHA with BCCAO (n = 12).

| Cell culture
The human neuroblastoma cell line SH-SY5Y was cultured in Dulbecco's modified Eagle medium/F12 (Thermo Fisher Scientific) containing 10% foetal bovine serum at 37°C in a humidified atmosphere containing 95% air and 5% CO 2 . Cells were rendered quiescent by serum starvation for 24 hours before all experiments. Hypoxic injury was induced by hypoxia for 6 hours (ie incubation in OGD medium containing 2.3 mmol/L CaCl 2 , 5.6 mmol/L KCl, 154 mmol/L NaCl, 5 mmol/L Hepes, and 3.6 mmol/L NaHCO 3 [pH 7.4] and under an atmosphere of 5% CO 2 , 95% N 2 and <0.1% O 2 ). The DMSO or HDAC inhibitor was treated with OGD medium. The neutralizing antibody of BDNF (1, 3 or 10 µg/mL) was treated for 24 hours before OGD condition.

| 3-(4,5-Dimethylthiazol-2-yl)-2,5diphenyltetrazolium bromide (MTT)
Following incubation and treatment with different HDACis under control or H conditions for differential time, cell viability was tested by incubation with MTT reagent (Sigma-Aldrich) at a final concentration of 0.5 mg/mL and a temperature of 37°C for 4 hours. The optical density of the purple MTT formazan product was measured at 570 nm using an enzyme-linked immunosorbent assay reader. The absorbance of cells transfected with NC was regarded as indicating 100% viability.

| Real-time quantitative polymerase chain reaction (qPCR)
The expression of genes was examined through real-time qPCR using G3PDH as internal control. The extracted total RNA was then reverse transcribed into cDNA. The reverse transcripts containing exon (I-IX, II-IX, III-IX, IV-IX or IX) were measured.

| Western blot analysis
Proteins were resolved on the basis of molecular weight through electrophoresis on 8%, 10% and 12% polyacrylamide gels, followed by transfer to a polyvinylidene difluoride membrane. The membrane was blocked and incubated with antibodies for acetylated proteins (ac-H3 or ac-H4), neurotrophic factor (BDNF) or beta-actin. Protein levels were analysed using an enhanced chemiluminescence detection kit (GE Healthcare).

| BDNF promoter chromatin immunoprecipitation assay (ChIP)
A chromatin immunoprecipitation assay was used to study the interaction between intracellular DNA and protein. In this study, we used a ChIP assay kit (Abcam) to survey the relationship between the target gene promoter and the acetylated histone residue. Briefly, after cross-linking the DNA and proteins by adding formaldehyde into the medium, the efficiency of formaldehyde was quenched by adding 0.125 mol/L glycine. Then, the DNA in cells was sheared into the average fragment (size: 200-1000 bp) through lysis buffer and sonication. The antibody for the acetylated protein was used to immunoprecipitate the target DNA, and the DNA was eluted for realtime PCR, ChIP or sequencing.

| HDACi protects SH-SY5Y cells against OGD injury
To examine the possible protection or injury attributable to treat-

| Western blotting analysis identifies BDNF upregulation by HDACi (compound 13) in the neuronal cells under OGD condition and hippocampus of ischemic brain
To test the underlying mechanism of the protective effect of HDACis in cognitive functioning, hippocampal volume (data not shown) and cell viability (Figure 1), we next clarified which molecules were involved in these processes. As can be seen from Figure (Figure 2A,B). We found that the increase in BDNF levels in hypoxic cells with compound 13 was a potential mechanism providing neuroprotective effects against OGD. Therefore, we also considered whether compound 13 can increase the protein levels of BDNF in a CCH brain to protect against VaD. Although compound 13 does not significantly increase BDNF content on the left or right side of the cortex in a CCH mouse ( Figure 2C), it is noteworthy that in the hippocampus, compound 13 significantly increases the protein levels of BDNF. As we know, suberoylanilide hydroxamic acid (SAHA) has a protective effect on ageing animals, and its effect is related to the increment of trophic factors. 32 However, in our CCH animals, SAHA did not increase BDNF content under CCH-induced injury ( Figure 2C,D). The data showed that BDNF, a potent autocrine/ paracrine neurotrophic factor, has a neuroprotective effect, which is induced by HDAC inhibitor, on ischemia/OGD injuries. We examined the possibility that treatment with HDACis might modulate BDNF expression and proved that BDNF expression was significantly up- with the BDNF antibody ( Figure 2E). Thus, BDNF might exert its role in mediating the protective role of HDAC inhibitor under hypoxia.

| HDACi upregulates protein levels of H3K14 or H4K5 acetylation in OGD cells
We further showed that the increased expression of BDNF induced by compound 13 could be because of the activation of the endog-    H3K14 (A, B) and Ac-H4K5 (C, D)) were studied after OGD for 24 h with the Western blotting analysis. β-actin served as the loading control. The densities were normalized to β-actin, and each bar represents the mean ± SEM of four independent experiments (n = 4, *P < 0.05, **P < 0.01 vs indicated group)

F I G U R E 4
Effect of compound 13 on the BDNF promoter (I, II, III, IV or IX) around H3K14 or H4K5 in the cells exposed to OGD injury. A-J, The expression of BDNF promoters was studied with chromatin IP, followed by qPCR analysis after OGD for 24 h. Fold expression was normalized to input control, and each bar represents the mean ± SEM of four independent experiments (n = 4, *P < 0.05, **P < 0.01 vs indicated group)

| HDACi increases the expression of BDNF exon II, III and IX
We further proved that the increased expression of BDNF P1 and P2 around H3K14 and of BDNF P2 and P3 around H4K5 resulted in the increase of relative transcripts. We designed the forward primer to complement one of the non-coding exon sequences and the reverse primer to complement exon IX. Differential BDNF transcript expression was studied subsequently using qPCR. The data showed that the expression of BDNF exon II-IX, III-IX or IX could be increased ( Figure 5B,C,E) but that of exon I-IX or IV-IX could not in the compound 13 group ( Figure 5A,D).

| D ISCUSS I ON
This is, to our knowledge, the first study to apply HDACis in the treatment of CCH animals and examine the molecular mechanisms in which BDNF is involved. One HDACi in particular, compound 13, increases the acetylation status in histones 3 and 4, which further react with specific promotors of BDNF and up-regulate its protein levels ( Figure 6).
HDACis usually comprise three structural features: a cap, linker and zinc-binding group. We utilized the diphenyl core as the cap, and various linkers to connect the central core to the hydroxamic acid motif, which yielded a series of diphenyl hydroxamates. We synthe-  (I-IX, II-IX, III-IX, IV-IX or  IX) in the cells exposed to OGD injury.
A-E, The expression of BDNF exons was studied with qPCR analysis after OGD for 24 h. The fold expression was normalized to GAPDH, and each bar represents the mean ± SEM of four independent experiments (n = 4, *P < 0.05, **P < 0.01 vs indicated group) HDACis and the target genes. Evidence has indicated that HDAC inhibition works both transcriptionally and non-transcriptionally to increase the expression of neurotrophic factors and stabilize microtubule proteins, respectively, which then increase vesicular transport. 12 Compound 13 was selected because it was superior among HDACis in the OGD experiments ( Figure 1). In this study, we focused on the effects of the transcriptional regulation of HDACis, but we cannot exclude the possibility that compound 13 achieved its effect by regulating non-histone targets, which may include proteins involved in cytoplasmic signalling, hormone receptors, transcription factors or cytoskeletal proteins. 31,[35][36][37] HDACis may be used to treat various brain diseases based on CA3, and dentate gyrus regions and alleviated cognitive deficits in ischemic rats. 48 In vitro data also showed that HDACi-valproic acid and trichostatin A increased neuroprotection and BDNF expression. 49 In short, the functions of BDNF in neurodevelopment, neurite outgrowth, synaptic plasticity and neuroprotection have been well established. In this study, we further confirmed the neuroprotective effects of BDNF through inhibition of BDNF using a BDNFneutralizing antibody ( Figure 2E).  Figure 4B,G,H). We did not exclude the possibility that other lysine residues of histones could be acetylated or that other HDAC activity could be affected, resulting in other BDNF promoter expression. Although BDNF P2 and P3 activity was increased, we found that the expression of BDNF exons II, III and IX were up-regulated ( Figure 5B,C,E). This may have been caused by exon IX, which is a common coding exon that is conjugated by all other exons. Therefore, our findings suggest an intricate regulatory mechanism for the BDNF gene, which provides a new concept for the regulation of BDNF in subcellular locations.
There are some limitations in this study, such as the pharmaceutical therapy of HDACis. This includes questions concerning the capacity of HDACis to pass through the BBB, the effectiveness of HDACis over a given period of time, and the specificity of HDACis.
Because there is a BBB, it is worth determining whether these hydroxamic acid-derived compounds can pass through to the brain.
Our data showed that compound 13 increased the acetylation of H3K14 and H4K5 in a CCH brain, implying that this compound can pass through the BBB. Compound 13 should be administrated once every 2 days for a consecutive 30 or 90 days. As it is a complicated procedure, the development of forms of HDACis should be performed with more comprehensive and long-term studies. Finally, HDACis often inhibit diverse HDAC activity, and they are not specific and may cause side effects. A study to clarify the inhibition of a specific HDAC by HDACis for VaD-related disease is warranted.

ACK N OWLED G EM ENTS
We thank the technical staff at the Small Animal-based Magnetic Resonance Imaging Service, Taipei Medical University for their technological assistance. This work was supported by Taipei Medical University and Shuang Ho Hospital (107TMU-SHH-16, 108TMU-SHH-12).

CO N FLI C T O F I NTE R E S T
The authors declare no potential conflicts of interest with respect to the research, authorship and/or publication of this article.