Selective HDAC6 inhibition protects against blood–brain barrier dysfunction after intracerebral hemorrhage

Abstract Backgrounds Blood–brain barrier (BBB) disruption after intracerebral hemorrhage (ICH) significantly induces neurological impairment. Previous studies showed that HDAC6 knockdown or TubA can protect the TNF‐induced endothelial dysfunction. However, the role of HDAC6 inhibition on ICH‐induced BBB disruption remains unknown. Methods Hemin‐induced human brain microvascular endothelial cells (HBMECs) and collagenase‐induced rats were employed to investigated the underlying impact of the HDAC6 inhibition in BBB lesion and neuronal dysfunction after ICH. Results We found a significant decrease in acetylated α‐tubulin during early phase of ICH. Both 25 or 40 mg/kg of TubA could relieve neurological deficits, perihematomal cell apoptosis, and ipsilateral brain edema in ICH animal model. TubA or specific siRNA of HDAC6 inhibited apoptosis and reduced the endothelial permeability of HBMECs. HDAC6 inhibition rescued the degradation of TJ proteins and repaired TJs collapses after ICH induction. Finally, the results suggested that the protective effects on BBB after ICH induction were exerted via upregulating the acetylated α‐tubulin and reducing stress fiber formation. Conclusions Inhibition of HDAC6 expression showed beneficial effects against BBB disruption after experimental ICH, which suggested that HDAC6 could be a novel and promising target for ICH treatment.

][9][10] Pan-and isoform-specific HDACis play roles in different cerebral diseases including hemorrhagic stroke. 11HDAC6 belongs to the class IIb HDACs, mainly locates in cytoplasm and possesses two catalytic domains. 12Tubastatin A (TubA) is the most potent inhibitor of HDAC6 and with over 1000-fold selectivity to HDAC6. 13Notably, genetic or pharmacologic inhibition of HDAC6 has been found to exert neuroprotective effects in ischemia stroke, 14,15 Huntington's disease, 16,17 Alzheimer's Disease, 17 and hemorrhagic stroke. 18,19Besides, researchers found that HDAC6 knockdown by small interfering RNA or TubA can control the dynamics of endothelial barrier integrity in pulmonary edema models. 20As we know, endothelial barrier integrity is essential for maintaining the function of BBB.However, whether pharmacological inhibition or gene interference of HDAC6 improves BBB function after ICH remain unclear.Thus, our study designed to explore the protective roles of HDAC6 inhibition on BBB leakage in ICH model and its related mechanisms.

| Experimental design
Experiment I: Investigate the ICH-induced effects on the expression of acetylated α-tubulin.Rats were randomly distributed into six subgroups: sham group, ICH 6 h, ICH 1d, ICH 2d, ICH 3d, ICH 7 day.
The ipsilateral hemisphere perihematoma tissue were collected for western blot analysis in each group (n = 3/group).Immunofluorescence staining was conducted in ICH 3d group (n = 3/group).
HBMECs were arranged into six groups and treated with different concentrations of hemin for 24 and 48 h to observe cells viability.
Hemin at 100 μM for 24 h was chosen to detect the protein level of acetylated α-tubulin.
Experiment II: Investigate the neuroprotective effects of HDAC6 inhibition after ICH stimulation.Rats were randomly assigned to four groups: (1)  Experiment IV: Probe the potential molecular mechanism of HDAC6 inhibition on early brain injury post-ICH.The subgroups were consistent with experiment II.Thirty-six rats were sacrificed at 3 days post-ICH.The TJ protein expressions (n = 3) were measured by Western blot, the ultrastructure of TJs (n = 3) was observed by TEM and F-actin (n = 3) was evaluated by immunofluorescence assay in each group.The above tests were also performed in vitro.

| Animals
Adult, male SD rats (275~325 g) were obtained from Hunan Slack Jingda Laboratory (Changsha, China) and housed in central laboratory of Hunan Provincial People's Hospital at a fixed temperature (25°C) and relative humidity (60%).All experimental protocols were approved by the Animal Care and ethics review committee of Central South University (IACUC approval No: 2020079).SD rats were fed with free water and food, dwelt in a 12 h light/dark cycle.Assignment and use of rats is clarified in Table S1.

| Rat model of ICH
The ICH procedure was performed via injecting collagenase type IV (Sigma-Aldrich) as previously published with minor modifications. 21,22In simple terms, rats were deeply anesthetized with pentobarbital sodium (40 mg/kg) by intraperitoneal injection.Stereotactically insert a microsyringe (10 μL) into the right striatum across the cranial borehole.Collagenase IV (0.2 U) dissolved in 2 μL of 0.9% saline was injected gradually for 10 min.The sham operation was injected an equal volume of saline only.

| SiRNAs and in vitro transfection
The specific siRNAs against HDAC6 were purchased from Honor-Gene (siG160718025500, Changsha, CHN).We tested three different siRNAs interference efficiencies, and the most effective one was applied to the following research (Sequence: GAAAC AAC CCA GTA CAT GAAT) (Table S2).HDAC6 siRNA or control siRNA was transfected through Lipofectamine 3000 (Invitrogen, L3000-015) for 48 h in accordance with the manufacturer's instructions.

| Assessment of neurological function
The neurological function was assessed at 1 and 3 days post-ICH by Garcia scoring system, 23 which includes six individual tests: voluntary movement, symmetry of limbs, forelimbs extension, body proprioception, climbing ability, and tentacle touch.Each test scores from 0 to 3. All trials conduction, calculation, and evaluation were subjected by two independent trained investigators.

| Blood-brain barrier permeability in vivo
We evaluated blood-brain barrier permeability by brain water content (BWC) measurement and Evans blue (EB) staining.The brain tissue was dissected into 5 sections: contralateral and ipsilateral basal ganglia, contralateral and ipsilateral cortex, and cerebellum.Samples were instantly weighed on a precision electronic autobalance for wet weight (WW), then dehydrated in an oven at 100°C to obtain dry weight (DW).
EB dye (2% in PBS, 4 mL/kg, Sigma-Aldrich) was injected into the femoral vein and circulated for 1 h.Then, animals were perfused transcardially with PBS, and perihematoma tissue were harvested and homogenized in 3 mL formamide (Macklin, F810079) before incubated for 72 h, and then centrifugation at 10,000 rpm for 25 min.The fluorescence intensity of diluted supernatant was measured at 610 nm using a microplate reader and quantified according to a standard curve.

| Endothelial cell permeability in vitro
The transwell assay was employed to detect the permeability of endothelial cell monolayer to FITC-dextran.Briefly, HBMECs were inoculated in the cell chambers.After endothelial cells formed a monolayer, Fluorescein isothiocyanate (FITC)-dextran (Sigma, 46944) was diluted in medium to 10 μg/mL and incubated for 20 min.
Fluorescence intensity at wavelengths of 485 and 520 nm was tested by fluorescence plate reader.

| Immunofluorescence staining
The sections were permeabilized three times with 0.2% Triton X-100 and incubated in sequence with primary antibodies and appropriate conjugated secondary antibodies.All antibodies were same as used in WB assay.In addition, F-Actin was stained with iFluor 647 (ab176759, 1:1000, Abcam) in brain paraffin sections and HBMECs.The images were viewed on a confocal microscope (Nikon, Nikon Eclipse C1).

| TUNEL staining
TUNEL staining was administered to access the death of cells and performed following the manual of the Cell Apoptosis Detection Kit (G1502-50 T, ServiceBio).The TUNEL-positive cells were captured by the fluorescence microscope (Nikon Eclipse C1) and analyzed with Image J software.

| Measurement of apoptosis in vitro
The flow cytometric analysis was employed to detect the apoptosis of HBMECs.In short, the HBMECs were collected by trypsin digestion and rinsed twice with PBS.The suspension was blended with Annexin V-APC (5 μL) and then added PI (5 μL) to reincubated for 10 min.Finally, the mixture was instantly analyzed by a FACS flow cytometer C6 (FCM; FACSCanto II; BD Biosciences) and the results were analyzed by FlowJo v10.7.1 (Tree Star).

| TEM analysis
We used transmission electron microscope (TEM) to observe the ultrastructure of TJs after ICH induction.All transmission electron microscopy was performed at Electron microscope Centre of Central South University.Brain tissues or HBMECs were prefixed with 2% glutaraldehyde in 0.1 M sodium phosphate buffer (PH 7.4) for 12 h at 4°C.After washing, samples were postfixed for 2 h with 1% osmium tetroxide and then gradually dehydrated in ascending concentrations of ethanol.The samples were then transferred to propylene oxide, embedded in Eponate 12 Resin and cut into 70 nm (Leica UC7).Subsequently, the sections were stained with uranyl acetate.

| Statistical analysis
Normality was evaluated through the Shapiro-Wilk test, and all data were normally distributed.Comparison of multiple groups were analyzed by one-way ANOVA followed by Bonferroni's post-hoc tests.All values are presented as means ± SEM. p < 0.05 was considered statistically significant.Statistical analyses were performed via SPSS software.

| Expression of acetylated α-tubulin after ICH induction in vivo and in vitro
The protein level of acetylated α-tubulin from perihematoma tissue was monitored by WB analysis.Compared with the sham group, acetylated α-tubulin decreased at 1 day post-ICH, reached its lowest level at 3 days and then gradually rebounded at 7 days (Figure 1A).The expression of acetylated α-tubulin in cultured HBMECs also decreased significantly after hemin treatment for 24 h (Figure 1B).Meanwhile, we observed the localization of acetylated α-tubulin in the perihematoma tissue.The expression and distribution of acetylated α-tubulin was further identified by immunohistochemical staining at 3 days post-ICH.As shown in Figure 1C, we barely found acetylatedα-tubulin colocalization with endothelial cells of rat brain at 3 days post-ICH.

| Neuroprotection effects of HDAC6 inhibition in vivo and in vitro
The ICH-induced rats displayed severe neurological impairments at

| HDAC6 inhibition attenuated ICH-induced BBB disruption in vivo and in vitro
After induction of ICH at 1 and 3 days, the BWC in ipsilateral basal ganglia area was significantly higher than sham group (p < 0.001, Figure 3A).Furthermore, TubA treatment did effectively reduce brain edema at doses of 25 mg/kg (vs.Vehicle, p = 0.001, 1 day; p = 0.024, 3 days) and 40 mg/kg (vs.Vehicle, p < 0.001, 1 and 3 days).
The Evans blue analysis showed obvious disruption of BBB both we used markedly alleviated the EB dye extravasation in the ipsilateral hemisphere (vs.vehicle, p < 0.001, 1 and 3 days, Figure 3B), but there was no significant difference between them.
For in vitro experiment, we assessed the effects of HDAC6 inhibition on endothelial permeability after hemin induction.Hemin

| HDAC6 inhibition attenuates the degradation of TJ proteins and the disruption of TJ ultrastructure in vivo and in vitro
Here, we explored the influences of TubA on the expression of two essential TJ proteins (occludin and ZO-1) after ICH induction.Our data indicated that ICH induced notable degradation of occludin and ZO-1, while 40 mg/kg of TubA significantly increased these two TJ proteins (ZO-1: Vehicle vs. TubA 40 mg/kg: p = 0.024; Occludin: Vehicle vs. TubA 40 mg/kg: p = 0.049, Figure 4A).For in vitro experiment, the level of occludin and ZO-1 was significantly up-regulated by TubA treatment in hemin-induced HBMECs (Figure 4C).Consistently, our data also confirmed that specific HDAC6 knockdown by siRNA transfection exerted rescue effects on the degradation of ZO-1 and occludin significantly (p < 0.05).

Meanwhile, ultrastructural changes of TJs in endothelial cells
after ICH induction were primarily detected by TEM.Under normal conditions, it was found that the basal layer was integrate and continuous; the endothelial TJs appeared intact and contiguous.However, BBB ultrastructure was dramatically damaged at 3 day post-ICH.As shown in Figure 4B, some TJs of endothelial cells were severely opened and seemed shorter and blurred.Meanwhile, the basement membrane appeared intermittent, irregular and thinner than sham group.However, TubA treatment (40 mg/kg) group showed the basal membrane of endothelial cells was more regular, smooth, and integrate, and the TJs were nearly intact (Figure 4B).
Although there are breaks between the TJs in low dosage (25 mg/ kg) group, they are longer and more intact than the Vehicle group.
In consistent with the in vivo study, the ultrastructural changes in HBMECs after HDAC6 inhibition were also detected.Compared with the hemin-induced group, basal membranes of endothelial cells in HDAC6 siRNA group and TubA group were found to be more continuous and smoother.Moreover, TJs was longer and appeared as higher electronic density between adjacent cells (Figure 4D).

| HDAC6 inhibition increased the expression of acetylated α-tubulin and prevented subsequent actin stress fiber formation
For in vivo experiment, medium dosage of TubA (25 mg/kg) showed only a tendency to rescue the reduction of acetylated α-tubulin (p = 0.322), while the high dosage of TubA (40 mg/kg) upregulated α-tubulin acetylation after ICH induction significantly (vs.Vehicle, p = 0.003, Figure 5A).For in vitro experiment, the results confirmed that TubA treatment and HDAC6 knockdown both upregulated the expression of acetylated α-tubulin in hemin-induced HBMECs (Figure 6A).Besides, immunofluorescence staining was also performed for each group, and the results were consistent with western blot (Figures 5B and 6B).
Immunostaining results exhibited a distinct increase of phalloidin-positive cells after ICH establishment in vivo and in vitro, which indicated actin stress fiber formation.Moreover, treatment with 25 and 40 mg/kg dosage of TubA both inhibited actin stress fiber formation (Figure 5C).The results of in vitro experiments still confirmed that TubA treatment and siRNA treatment inhibited actin stress fiber formation in hemin-induced HBMECs (Figure 6C).

| DISCUSS ION
In the current job, we explored the role of HDAC6 inhibition and its underlying molecular mechanisms in early brain injury after ICH.The major observations are as follows: (1) The expression of acetylated α-tubulin decreased in the early stage of intracerebral hemorrhage.Histone deacetylases (HDACs) can catalyze the hydrolysis of acetyl groups from the lysine residues of histones and non-histones, and play roles in epigenetics and signal modification. 24Accumulating evidences show that HDAC6-targeted treatment is a considerable therapeutic strategy in central nervous system diseases for its neuroprotective and regenerative effects.On one hand, the ubiquitin-binding-ZnF of HDAC6 is involved in clearing the cytotoxic aggregates of misfolded proteins. 25On the other hand, highselective inhibitors or small interfering RNAs targeting HDAC6 exhibited neuroprotective effects in some neurological disease models.For instance, TubA and the shRNA of HDAC6 can ameliorate neuronal necroptosis from oxygen-glucose deprivation/reperfusion (OGDR), 15,25 alleviate cognitive dysfunction in mice with Alzheimer's disease (AD), 17 and prevent axonal loss in mice with Charcot-Marie-Tooth disease. 26Although the neuroprotective effects of HDAC6 inhibition have been extensively studied for many years, it was unclear whether HDAC6 inhibition play a role in ICH.As the main substrate of HDAC6, acetylated α-tubulin indirect represented the activity of HDAC6. 15Plenty of researches have studied the acetylated α-tubulin levels in different neurological diseases.For instance, Wang et al. found remarkable reductions of acetylated α-tubulin in the striatum and cortex at 24 and 72 h post-ischemia. 14Besides, Zhang et al. observed acetylated αtubulin was dramatically decreased in transgenic mice model of AD. 17 Moreover, Zeng et al. observed that α-tubulin acetylation was significantly downregulated after OGDR in N2a cells. 15Recently, Yang et al. found both Dopamine (DA) neuron numbers and acetylated α-tubulin levels decreased significantly at 7-28 days after ICH, and epothilone B (EpoB, a MT-stabilizing agent) could ameliorated DA neuronal damage. 27,28In addition, their team also demonstrated that promoting acetylation of α-tubulin by TubA significantly alleviated axonal injury and mitochondrial dysfunction after ICH. 29 In similarity with their studies, we observed a significant decline in acetylated α-tubulin at 1 day after ICH, suggesting that the down-regulation of acetylated α-tubulin participates in the early stage of ICH pathogenesis.
Then we investigated whether HDAC6 inhibition by TubA or siRNA treatment had neuroprotective effects after ICH induction.
Our results reflect that medium or high doses of TubA (25, 40 mg/ kg) both improved neurological deficits and markedly inhibited cell apoptosis in the perihematomal tissue at 3rd day post-ICH.Meanwhile, we demonstrated that TubA or siRNA can also increase cell viability as well as inhibit apoptosis of Hemin-induced HBMECs.These results demonstrated that genetic or pharmacologic inhibition of HDAC6 exhibited beneficial effects for early brain injury post-ICH.
The primary causes of brain injury post-ICH are impaired BBB integrity and subsequent increased vascular permeability. 30,31BBB disruption can generate brain edema, 32 hematoma expansion, intracranial hypertension, and even midline shift. 5,33Previous researches have shown that the peak of cerebral edema occurs on the 3rd day after ICH 34 and indirectly indicated that the integrity of BBB is fragile at this time point.Hence, the effect of HDAC6 inhibition on BBB disruption after ICH was assessed in subsequent cellular and animal experiment.For in vivo experiment, TubA markedly reduced BBB permeability and ipsilateral brain edema.For in vitro experiment, TubA or HDAC6 knockdown significantly reduced endothelial permeability by measurement of FITC-dextran extravasation.Thus, we preliminarily confirmed that HDAC6 inhibition alleviates the breakdown of BBB after ICH.
BBB is composed of a monolayer of endothelial cells (ECs) structure and numerous cell-cell junctional complexes for maintaining the integrity of BBB.Cell-cell junctional complexes mainly including tight junctions (TJs) and adherens junctions (AJs).
TJs are situated on the apical membrane of ECs and consist of transmembrane proteins and cytoplasmic proteins that connect transmembrane proteins with the cytoskeleton. 35Claudin-5 and occludin are the main transmembrane molecules of tight junctions mediating endothelial cell integrity.Besides, occludin binds to the cytoskeleton via cytoplasmic proteins zonula occludens-1(ZO-1).
Contact in AJs is established mainly through VE-cadherin.They also interact with the cytoskeleton via cytoplasmic anchor proteins such as catenins. 31,36Under physiological conditions, actin in resting endothelium exists in the form of monomeric globular actin (G-actin).Once activated, globular actin aggregates into filamentous actin (F-actin), and this transition induces contractile stress fibers formation. 37Subsequently, part of the TJ proteins demounted at cell-cell contact and internalized due to the tension transmission, 38,39 thus expanding the endothelial gaps and increasing the BBB permeability.Thereby, inhibition of stress fiber formation in endothelial cells, which can promote the stability of cytoskeleton, represents a reasonable treatment strategy for blood-brain barrier protection after ICH. 32 our study, we detected the levels of TJ proteins (ZO-1 and occludin) by western blot and observed the continuity of TJs by TEM to assess microvascular integrity.Previous study has demonstrated that HDACs inhibitors (valproic acid, Vorinostat and sodium butyrate) could up-regulate the specific TJs proteins and this regulation was dependent on protein kinase activity. 40As expected, HDAC6 inhibition by TubA rescued the degradation of ZO-1 and occludin as well as repaired TJs disruption after ICH.Besides, we demonstrated TubA and HDAC6 siRNA reduced stress fibers formation in cellular and animal model of ICH.These evidences suggested that selective HDAC6 inhibition prevented the BBB broke down via inhibition the formation of stress fiber and upregulation of TJ proteins.However, detailed molecular mechanisms of these effects observed by HDAC6 inhibition in ICH models have yet to be determined.
vivo and in vitro model of ICH.Taken all into consideration, we suggested that the effects of HDAC6 inhibition on cytoskeleton may be exerted via upregulating the acetylated α-tubulin and subsequently stabilizing the endothelial microtubule structures.

| CON CLUS IONS
Our research suggested that HDAC6 contributes, or at least partly, to the early pathophysiological process of brain injury after ICH.Besides, pharmacological inhibition or genetic inhibition of HDAC6 exerted protective effects on BBB.Furthermore, the neuroprotective effects were exhibited via upregulating the acetylated α-tubulin and inhibiting stress fiber formation.In all, HDAC6 may be a promising target to fight against ICH-induced BBB disruption.

1 and 3
days post-ICH by modified Garcia test (vehicle vs. sham, p < 0.001, Figure 2A).Administration of TubA with high dose (40 mg/kg) were significant effective for the improvement of neurological impairments from day 1 to day 3 post-ICH, while medium F I G U R E 1 Time-dependent trends of acetylation α-tubulin protein levels in ICH rats and hemin-induced HBMECs.(A) Western blot and quantification analysis of acetylated α-tubulin in ICH animal model.Values are indicated by means ± SEM; versus sham, *p < 0.05; **p < 0.01; ***p < 0.001.(B) Western blot and quantification analysis of acetylated α-tubulin in hemin-induced HBMECs model.Values are indicated by means ± SEM; versus control, *p < 0.05; **p < 0.01; ***p < 0.001.(C) Double immunostaining of acetylated α-tubulin with von Willebrand Factor (VWF).Scale bar = 20 μm.dose TubA (25 mg/kg) could improve neurological deficits only on day 3. Consistent with the behavioral results, the quantity of TUNEL-positive cells in TubA-treated group were obviously reduced compared with ICH group (vehicle vs. TubA 25 mg/kg, p = 0.001, vehicle vs. TubA 40 mg/kg, p < 0.001, Figure 2C).In vitro, CCK-8 was conducted to assess the effect of HDAC6 inhibition on HBMECs viability.As expected, hemin-induced cells viability was significantly improved while treated with TubA or specific HDAC6 knockdown by siRNA transfection (Figure 2B).Correspondingly, the apoptosis rate of hemin-induced HBMECs was ameliorated significantly by siRNA treatment or TubA treatment (p < 0.001, Figure 2D).

| 9 of 13 PENG
led to high permeability of HBMECs to FITC-dextran, suggesting et al.endothelial dysfunction.On contrast, TubA treatment and specific siRNA treatment significantly reduced the hemin-induced endothelial permeability (p < 0.001, Figure3C).

( 2 )
HDAC6 inhibition by TubA or siRNA treatment inhibited cell apoptosis after ICH.(3) HDAC6 inhibition by TubA or siRNA treatment rescued the degradation of TJ proteins (ZO-1 and occludin) and reduced BBB permeability in ICH animal as well as cellular model.(4) TubA treatment increased α-tubulin acetylation and inhibited actin stress fiber formation after ICH in ICH animal as well as cellular model.In general, these findings indicate that HDAC6 inhibition might be a promising treatment for protecting the BBB after ICH.