Protective role of ABCA1 in ischemic preconditioning is mediated by downregulation of miR-33-5p and miR-135-5p

Ischemic preconditioning (IPC) significantly reduces ischemia–reperfusion injury in the brain by inducing ischemic tolerance. Although emerging evidence suggests that microRNAs (miRNAs) contribute to the pathogenesis of brain ischemia and IPC-induced neuroprotection, the role of miRNAs and their underlying mechanisms are still unclear. IPC was induced in male C57BL/6 mice by brief bilateral common carotid artery occlusion. After 24 h, mice underwent transient middle cerebral artery occlusion followed by 3 h of reperfusion. Expression levels of messenger RNAs (mRNAs) and proteins were examined in the ipsilateral cortex, and mimics and inhibitors of selective miRNAs were transfected into Neuro-2a cells before oxygen–glucose deprivation (OGD). Post-IPC miRNA expression profiling identified neuroprotection-associated changes in miRNA expression in the ipsilateral cortex after ischemic stroke. Among them, miR-33-5p and miR-135b-5p were significantly downregulated by IPC. Inhibition of miR-33-5p and miR-135b-5p expression protected Neuro-2a cells from OGD-induced apoptosis. Inhibition of these two miRNAs significantly increased mRNA and protein levels of ATP-binding cassette subfamily A member 1 (ABCA1), and a binding assay showed that these two miRNAs showed specificity for Abca1 mRNA. Overexpression of ABCA1 decreased the Bax/Bcl2 mRNA ratio and activation of caspase-9 and caspase-3, whereas knockdown of ABCA1 expression increased the Bax/Bcl2 mRNA ratio and the percentage of Neuro-2a cells with a loss of mitochondrial membrane potential after OGD-treatment. In conclusion, ABCA1 expression is regulated by miR-33-5p and miR-135b-5p. Increased ABCA1 expression following IPC exerts a protective influence against cerebral ischemia via suppression of a mitochondria-dependent apoptosis pathway.


IPC by bilateral common carotid artery occlusion (BCCAO). Procedures for BCCAO precondition-
ing have been established and published previously 11 . Briefly, mice were anesthetized with isoflurane (1.6-2.0%) and a fiber optic probe was glued to the right parietal bone (2 mm posterior and 5 mm lateral to bregma) and connected to a laser-Doppler flowmeter (Periflux System 5010, Perimed, Sweden). Cerebral blood flow (CBF) was continuously recorded using a computer-based data acquisition system (Perisoft, New South Wales, Australia). IPC was induced by 3 episodes of 1 min occlusion of both common carotid arteries, each followed by 5 min of reperfusion. Wounds were closed and animals were returned to their cages. In the sham-operated mice, the carotid arteries were exposed for 15 min without occlusion (sham). Control animals did not receive any surgical operation (control). Animals were assigned to groups randomly by computerized random number generators.

Transient middle cerebral artery occlusion (MCAO). Time window of BCCAO IPC paradigm for
ischemic stroke was previously defined 12 , therefore, we confirmed the tolerance against ischemic stroke at 24 h after BCCAO IPC in this study. Mice were reanesthetized with isoflurane 24 h after BCCAO (IPC + MCAO group) or sham operation (sham + MCAO group), which were blinded, and the fiber optic probe was reattached to the right parietal bone (2 mm posterior and 5 mm lateral to bregma) and connected to a laser-Doppler flowmeter. CBF was continuously recorded during MCAO and reperfusion periods.
Techniques for transient MCAO with an intravascular suture have been described previously 12,13 . Briefly, a 6-0 silicon-coated black monofilament surgical suture (Doccol Cooperation, Redlands, CA, USA) was inserted into the exposed right external carotid artery, advanced into the internal carotid artery, and wedged into the circle of Willis to obstruct the origin of the middle cerebral artery (MCA). The filament was left in place for 30 min and then withdrawn to re-establish CBF. Only animals that exhibited a reduction in CBF > 85% during MCAO and in which CBF recovered by > 80% after 10 min of reperfusion were included in the study. Rectal temperature was maintained at 37.0 ± 0.5 °C using a thermostatically controlled heating pad, both during surgery and during the recovery period until consciousness was regained. After 3 h reperfusion, mice were euthanized, and the ipsilateral cortex of brain was isolated for the experiments because the protective effect of IPC was limited in the cortex 11 . Cell culture. Mouse neuroblastoma cell line Neuro-2a was purchased from American Type Culture Collection (ATCC no. CCL-131; Manassas, VA, USA) and cultured in Dulbecco's modified Eagle's medium (Thermo Fisher Scientific, Waltham, MA, USA) supplemented with 10% fetal bovine serum (FBS, Thermo Fisher Scientific), 100 U/ml penicillin (Thermo Fisher Scientific), and 100 μg/ml streptomycin (Thermo Fisher Scientific) in an atmosphere of 95% humidified air and 5% CO 2 at 37 °C.
In order for IPC in vitro, Neuro-2a cells overnight post-seeding were placed in a hypoxia chamber (5% CO 2 , 1% O 2 and 94% N 2 , 37 °C) for 1 h prior to a main hypoxic insult. After 1 h equilibration, the cells underwent 5 h hypoxic incubation followed 1 h of reperfusion.
Total RNA extraction and miRNA microarray. Total RNA was extracted using the mirVana miRNA isolation kit (Thermo Fisher Scientific) for cultured cells and brain tissues according to the manufacturer's pro- www.nature.com/scientificreports/ tocol. RNA integrity for each sample was confirmed with the Agilent 2100 Bioanalyzer (Agilent Technologies, Santa Clara, CA, USA). miRNA expression profiling was determined using the Agilent 8 × 60 K mouse miRNA microarray (Agilent Technologies) according to the manufacturer's instruction. The arrays were scanned using the Agilent Technologies G2600D SG12494263 (Agilent Technologies) to obtain the hybridization images. The array data export processing and analysis was performed using Agilent Feature Extraction v11.0.1.1 (Agilent Technologies).
To identify differentially expressed miRNAs (DEmiRs), we applied moderated t-statistics based on an empirical Bayesian approach 14 . Significantly upregulated and downregulated miRNAs were described as differentially expressed if the p-values were < 0.05, and the fold change was greater than 1.5 or less than 0.75. Finally, we excluded miRNAs with a low expression level (average log 2 expression level < 4.0) from the list of DEmiRs.
For OGD, the cells were detached from the plates after 24 h transfection and reseeded in a 35 mm culture dish at a density of 5 × 10 5 cells/ml. After 24 h reseeding, the cells were washed with phosphate-buffered saline and replaced with glucose-free RKRB buffer. The cells were then placed in hypoxia chamber (Astec) for 8 h at 5% CO 2 , 1% O 2 and 94% N 2 . After hypoxic treatment, Neuro-2a cells were returned normoxic condition with normal culture media for 1 h for reperfusion.
Cell viability assay. Cell viability assays were performed using Cell Counting Kit-8 (CCK-8; Dojindo Molecular Technologies, Inc., Kumamoto, Japan) according to the manufacturer's instructions. The absorbance at 450 nm was measured using a microplate reader (Molecular Devices LLC, Sunnyvale, CA, USA) after incubation for 2 h at 37 °C. Cell viability was expressed as a percentage of that of the control cells.
Western blot analyses. The ipsilateral cortex and whole cell lysate were prepared in RIPA buffer (Thermo Scientific) supplemented with protease inhibitors (Thermo Scientific). Proteins (50 μg) were resolved using denaturing 6% sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and transferred to polyvinylidene fluoride (PVDF) membranes. Membranes were blocked in 5% skim milk in Tris-buffered saline with 0.1% Tween 20 (TBST) and subsequently incubated overnight at 4 °C with the following primary antibodies: anti-ABCA1 monoclonal antibody (1: Statistical analysis. All data are expressed as the mean ± standard deviation of at least three independent experiments. Statistical analyses were carried out using GraphPad Prism 5 software. The details of each statistical analysis are provided in the figure legends. We considered p-values < 0.05 to be statistically significant.

Results
Effects of IPC on apoptosis-associated gene expression in the ischemic cortex. The experimental timelines for sham + MCAO and IPC + MCAO are shown in Fig. 1a. We previously reported ischemic protection of the BCCAO IPC paradigm by showing a smaller infarct volume (~ 48%) in the IPC + MCAO group compared to the control and sham + MCAO groups. Ischemic injury reduction was observed predominantly in the cortex but not in the stratum 11 . Previous studies using animal models of brain ischemia reported neuroprotective effects of IPC through inhibition of apoptosis 15,16 . Therefore, we used RT-qPCR to compare the expression of apoptosis-promoting Bax mRNA and apoptosis-inhibiting Bcl2 mRNA in the IPC + MCAO group (n = 6) to their expressions in the control group (n = 6) and the sham + MCAO group (n = 6). There was no difference in Bax mRNA expression among the three groups; however, the IPC group showed significantly higher Bcl2 mRNA expression compared to both the control and sham + MCAO groups (Fig. 1b). In addition, the IPC + MCAO Bax/Bcl2 mRNA ratio was decreased significantly to approximately 37% and 123% of control and sham + MCAO group ratios, respectively, indicating a neuroprotective effect of IPC against apoptosis (Fig. 1b). Neuronal apoptotic reduction after preconditioning was further evaluated by determining protein expression of cleaved caspase-3 and anti-apoptotic BCL2 in the control (n = 6), sham + MCAO (n = 6), and IPC + MCAO groups (n = 6). Cleaved caspase-3 protein expression was significantly reduced, approximately 5.9-fold, in the IPC + MCAO group compared to the sham + MCAO group, whereas anti-apoptotic BCL2 protein expression was significantly increased, approximately 19.1-fold, in the IPC + MCAO group compared to the sham + MCAO group (Fig. 1c).

Neuroprotective effects by inhibition of miR-33-5p and miR-135b-5p under in vitro ischemic
conditions. To mimic in vitro ischemic-like conditions, Neuro-2a cells were exposed to OGD, following the experimental timeline shown in Fig. 3a for OGD and IPC + OGD. Expression levels of the three miRNAs (miR-33-5p, miR-135b-5p, and miR-551b-3p) were determined under in vitro ischemic conditions using RT-qPCR. The expression of miR-551b-3p was too low to be detected in Neuro-2a cells; however, consistent with the results of our ischemic-stroke mouse model, the expression of both miR-33-5p and miR-135b-5p were significantly downregulated by in vitro IPC stimulation compared to the control or in vitro ischemic-condition levels (Fig. 3b). Furthermore, in vitro IPC stimulation significantly decreased the Bax/Bcl2 mRNA ratios by approximately 169% and 34% compared to the ratios for ischemic conditions and control conditions, respectively (Fig. 3c). Cell viability was significantly increased by in vitro IPC stimulation compared to that of in vitro ischemic-conditions (Fig. 3d). To determine whether downregulation of miR-33-5p and miR-135b-5p protected Neuro-2a cells from ischemia-induced apoptosis, cells were transfected with individual miRNA inhibitors and then assessed for apoptosis after ischemia treatment. Bax/Bcl2 mRNA ratios were significantly reduced in cells transfected with miR-33-5p and miR-135b-5p inhibitors, approximately 24% and 29%, respectively, compared to  www.nature.com/scientificreports/ the ratio in cells transfected with negative-control inhibitor (Fig. 3e). Expression inhibition by both miR-33-5p and miR-135b-5p showed potent protection against ischemia-induced apoptosis.
To further test our hypothesis that miR-33-5p and miR-135b-5p directly regulate protein expression, we assessed ABCA1 expression under hypoxic conditions after treatment with each miRNA inhibitor. Knockdown of miR-33-5p and miR-135b-5p significantly upregulated ABCA1 protein expression (Fig. 4c) and Abca1 mRNA expression (Fig. 4d). These results provide strong evidence that Abca1 mRNA is a target for miR-33-5p and miR-135b-5p, negatively regulating ABCA1 expression. As shown in Fig. 2, miR-33-5p and miR-135b-5p levels were downregulated in the IPC + MCAO group compared to the control and sham + MCAO group. To confirm the role of miR-33-5p and miR-135b-5p in negatively regulating target protein expression in vivo, ABCA1 protein expression was determined using Western-blot analysis in the control (n = 6), sham + MCAO group (n = 6), and IPC + MCAO group (n = 6). Protein expression was significantly higher in the IPC + MCAO group compared to the control and sham + MCAO group (Fig. 4e).

ABCA1 prevents OGD-induced mitochondria-dependent apoptotic signaling.
To determine the effect of ABCA1 on OGD-induced neuronal apoptosis, Neuro-2a cells were transiently transfected with Abca1 expression constructs or with Abca1 siRNA (siAbca1) to overexpress or knockdown ABCA1, respectively; cells transfected with a vehicle plasmid construct or non-targeting siRNA (siNC) were used as controls.
After an additional 24 h, cells were subjected to 8 h of hypoxia with glucose-free RKRB buffer and reperfusion for 1 h with normal culture media. Following OGD, Abca1 expression levels were assessed using RT-qPCR: Abca1 mRNA increased ~ 12-fold in Abca1-transfected cells and decreased ~ 70% in siAbca1-transfected cells compared to mock-transfected cells (Fig. 5a). Compared to controls, Bax/Bcl2 mRNA ratios were significantly decreased ~ 47% in Abca1 overexpressed cells and significantly increased ~ 52% in Abca1 knockdown cells following OGD (Fig. 5b), and cell viability was significantly increased ~ 160% in Abca1 overexpressed cells but decreased ~ 31% in Abca1 knockdown cells compared to their controls (Fig. 5c). Caspases are crucial mediators of cellular apoptosis, so we investigated the influence of ABCA1 on caspase -3, -8, and -9 activation following OGD using Western-blot analysis. We found that OGD-induced activation of caspase-3 and -9 was significantly inhibited approximately 75% and 50%, respectively, in Abca1-transfected cells compared to mock-transfected cells, but caspase-8 activity was unchanged (Fig. 5d). In addition, the influence of ABCA1 on MMP was determined using flow cytometry analysis of JC-1 staining. The percentage of cells with reduced mitochondrial membrane potentials (the green fluorescence ratio) increased ~ 20% in siAbca1-transfected cells compared to siNCtransfected cells following OGD, suggesting that ABCA1 attenuated OGD-induced mitochondrial dysfunction (Fig. 5e). These results indicate that ABCA1 protects neuronal cells against OGD-induced apoptosis by inhibiting a mitochondria-dependent apoptosis pathway.
Upregulation of ABCA1 mRNA expression in blood from stroke patients. To examine the role of ABCA1 in human stroke patients, mRNA expression data from peripheral blood samples were collected from the following GEO datasets: acute ischemic stroke (GSE16561) and cardioembolic stroke (GSE58294). Database GSE16561 contained transcription expression profiles from 63 blood samples (39 stroke patients and 24 controls); mRNA expression of ABCA1 was significantly increased in stroke patients compared to controls (Fig. 6a). Database GSE58294 contained transcriptional expression profiles from 92 blood samples (69 stroke patients and 23 controls). The ischemic-stroke samples were collected at three different time points: within 3 h (prior to treatment), 5 h, and 24 h (after treatment) after stroke onset (n = 23). Increased mRNA expression of ABCA1 was found at all time points in the blood samples from stroke patients compared to controls (Fig. 6b), supporting the idea that ABCA1 expression is induced by brain ischemia. Based on these results, we suggest that the induction of ABCA1 may be a protective response to reduce ischemic brain injury.

Discussion
Apoptosis caused by cerebral ischemia/reperfusion (I/R) is a very important process for neuronal cell death 17 . The Bcl2 superfamily includes proapoptotic genes such as Bax and Bad 18 , and anti-apoptotic genes such as Bcl2, Bcl-xL, and Bax 19 . Reduced cellular activation of Bcl2 and Bcl-xL, and increased Bax activation has been reported in neurons injured by ischemia 20,21 , supporting the idea that the ratio of proapoptotic/anti-apoptotic proteins can determine cellular fate after a death signal 22 . The present study demonstrates that IPC has a neuroprotective effect on ischemia by modulating the Bax/Bcl2 mRNA ratio. This ratio was significantly decreased in the IPC + MCAO group compared to the control and sham + MCAO groups through increased Bcl2 mRNA expression without a significant change in Bax expression. Bcl2 acts as a survival factor in the mitochondrial apoptotic pathway by inhibiting caspase activation, inhibiting the production of free radicals, and through the actions of other proapoptotic Bcl2 family members such as Bax and Bad 23,24 . Overexpression of Bcl2 also promoted cortical neuronal survival after focal cerebral ischemia by preventing translocation of apoptosis inducing factor    25 . Consistent with the present results, cyclic AMP response element binding protein (CREB)-mediated Bcl2 overexpression was also shown to be protective in ischemia-preconditioned rats against subsequent I/R exposure 26 .
Several studies have investigated IPC-induced miRNA changes using in vivo and in vitro models to elucidate their functional significance in generating tolerance to cerebral ischemic injury. However, due to different experimental designs, different IPC stimuli, and diverse in vivo or in vitro models, the effects of IPC-induced neuroprotection on miRNAs and their targets are highly heterogenous and inconsistent across studies. In one study investigating miRNA changes with preconditioning, the miR-200 and miR-182 families were both upregulated after IPC following 3 h of reperfusion, resulting in neuroprotection through downregulation of prolyl hydroxylase 2 which is involved in the degradation of hypoxia inducible factors 7 . In contrast, another study reported that upregulation of miR-200c after stroke promoted neuronal cell death through inhibiting expression of its target protein (reelin) that plays a role in synaptogenesis and neuronal migration 27 . A subsequent mouse study has shown that decreased miR-132 expression induced neuroprotection by targeting methyl CpG-binding protein 2 following IPC 28 , but miR-132 overexpression in rats using a viral vector has also been reported to increase neuronal survival following global ischemia and decrease cell death following OGD in vitro 29 .
Previous studies have shown that miR-33-5p (human miR-33a-5p) targets key genes involved in cholesterol transport, such as Abca1 and Abcg1. Inhibition of endogenous miR-33-5p in a mouse model increased ABCA1 expression and led to elevated plasma high-density lipoprotein (HDL), reversal of a cholesterol transport pathway, and prevention of atherosclerosis 30,31 . In addition, inhibition of miR-33-5p lowered the endogenous cortical level of amyloid-β in a mouse model of Alzheimer's disease (AD) via increased ABCA1 expression and APOE lipidation, suggesting a potential therapeutic strategy for AD 32 . Inhibition of miR-33-5p in OGD-treated rat brain microvascular endothelial cells has also been reported to enhance the expression of its other target gene, X-box binding protein 1, improving both survival and angiogenesis through upregulation of the Differentiation antagonizing non-protein coding RNA gene 33 . Neuroprotective effects of miR-135b-5p have also been reported in AD and Parkinson's disease through targeting β-site APP-cleaving enzyme 1 and glycogen synthase kinase 3β, respectively 34,35 . In contrast, the present results demonstrate that inhibition of miR-135b-5p prevented neuronal cell death from OGD-induced apoptosis. This discrepancy is likely the result of a single miRNA targeting different pathways or genes.
In the present study, we demonstrated that both miR-33-5p and miR-135b-5p expression levels were significantly downregulated by IPC, and inhibition of their expression protected neuronal cells from OGD-induced apoptosis (i.e., an in vitro ischemia model). Inhibition of these two miRNAs in neuronal cells significantly increased mRNA and protein levels of ABCA1, and a binding assay showed that these two miRNAs were specific for the 3'UTR of Abca1 mRNA.
ABCA1 is an integral cell-membrane protein that plays a major role in HDL formation by exporting cholesterol and phospholipids from cells and solubilizing these lipids by apolipoprotein 36 . There is also evidence that ABCA1 has anti-inflammatory activity and other beneficial effects on ischemic injury. Brain evaluation in ABCA1-knockout mice has shown increased blood-brain barrier leakage, white matter/axonal damage, and functional deficits after stroke compared to ABCA1-expressing control mice 37 . ABCA1 overexpression in a mouse stroke model induced by GW3905, a liver-X receptor (LXR) agonist, increased blood HDL, both gray and white matter densities, oligodendrocyte progenitor cell number, and also improved functional outcomes after distal MCAO 38 . In a mouse model of I/R using acute elevation of intraocular pressure, another LXR agonist (TO901317) induced ABCA1 and reduced retinal ganglion cell apoptosis by promoting the nuclear translocation of annexin A1, an anti-inflammatory factor 39 . In the present study, we found that ABCA1 expression was enhanced in both preconditioned ischemic neurons and the brain compared to neurons and brain without preconditioning. We further demonstrated that overexpression of ABCA1 decreased the Bax/Bcl2 mRNA ratio and activation of caspase-9 and caspase-3, whereas knockdown of ABCA1 expression increased the Bax/Bcl2 mRNA ratio and the percentage of Neuro-2a cells with MMP loss after OGD-treatment, supporting the idea of Figure 5. ABCA1 suppresses a mitochondria-dependent apoptosis pathway. (a-c) Neuro-2a cells were transiently transfected with either vehicle or Abca1, siNC, and siAbca1 expression constructs. For OGD, cells were detached from plates 24 h after transfection and reseeded. Twenty-four hours after reseeding, cells were subjected to 8 h of OGD and 1 h of reperfusion. mRNA expression of Abca1, Bax, and Bcl2 was measured by RT-qPCR (n = 3). Cell viability was determined by CCK-8 assay. (d) Caspase-3, -8, and -9 activation was assessed by Western-blot analysis. Representative results are illustrated, and values from densitometric analyses are shown after normalization to α-tubulin from the controls (n = 3). Full-length blots are presented in Supplementary Fig. 4 (e) Mitochondrial membrane potentials were determined by flow cytometry analysis using JC-1 staining. Representative results are illustrated, and values are reported mean ± standard deviation (n = 4). Statistical analyses were performed using t-tests (* = p < 0.05, ** = p < 0.01, *** = p < 0.001). Abbreviations: Abca1, ATP-binding cassette subfamily A member 1; siNC, non-targeting control small interfering ribonucleic acid; siAbca1, Abca1 siRNA; OGD, oxygen-glucose deprivation; mRNA, messenger ribonucleic acid; RT-qPCR, reverse transcription-quantitative polymerase chain reaction; Bax, Bcl2 associated X; Bcl2, B-cell lymphoma 2; JC-1, 5,5′,6,6′-tetrachloro-1,1′,3,3′-tetraethyl-benzamidazolocarbocyanin iodide. www.nature.com/scientificreports/ a protective role for ABCA1 via apoptosis-pathway blocking in mitochondria. However, the precise molecular mechanism underlying how ABCA1 modulates apoptotic gene expression and MMP under ischemic conditions is not clear and remains to be investigated.
In human blood, the expression of ABCA1 mRNA was significantly upregulated in stroke patients compared to non-stroke controls, suggesting that increased ABCA1 expression may contribute to cell protection after ischemic injury and prevent further cellular damage.
In conclusion, increased ABCA1 expression following IPC exerts a protective effect against cerebral ischemia via suppression of a mitochondria-dependent apoptosis pathway. The regulatory mechanisms for both miR-33-5p and miR-135b-5p on ABCA1 expression may provide potential therapeutic targets for treating stroke.
Received: 16 October 2020; Accepted: 3 June 2021 Figure 6. Upregulated ABCA1 mRNA expression in stroke-patient blood. Human blood mRNA expression levels of ABCA1 obtained from the Gene Expression Omnibus datasets GSE16561 and GSE58294 are shown in (a) and (b), respectively. Data are shown as box-and-whisker plots (from minimum to maximum). Statistical analyses were performed using t-tests (a, *** = p < 0.001) or a one-way analysis of variance followed by Dunnett's multiple comparison test anchored to control data (b, *** = p < 0.001). Abbreviations: Abca1, ATP-binding cassette subfamily A member 1; mRNA, messenger ribonucleic acid.