Long noncoding RNA ANRIL knockdown attenuates neuroinflammation following ischemic stroke via suppressing the expression of NF-κB in vitro and in vivo

ABSTRACT Objective Increasing evidence suggests that long-noncoding RNAs can exert neuroprotective effects in cerebral ischemia-reperfusion injury. Levels of the long noncoding RNA ANRIL (ANRIL) are reportedly altered in ischemic stroke (IS) patients, but its role in IS requires further clarification. This study was designed to explore the mechanistic function of ANRIL in IS. Methods In vitro, HT22 cells was treated with an oxygen-glucose deprivation/reperfusion (OGD/R). In vivo, brain ischemia/reperfusion was induced by 60-minute transient middle cerebral artery occlusion/reperfusion (MCAO/R) IS model in C57/BL6 mice. Additionally, cells were transfected with si-ANRIL, pcDNA3.1-ANRIL, pcDNA3.1-NF-κB, or appropriate negative controls, and si-ANRIL and pcDNA3.1-NF-κB were administered into the lateral ventricles in MCAO/R model mice. Cell viability and apoptosis were detected via MTT and flow cytometry assays. mRNA and protein expression of NF-κB were detected via qRT-PCR and Western blotting. IL-1β, IL-6, TNF-a, and iNOS levels were detected via ELISA. In addition, infarcted area and neuronal injury were evaluated via TTC, Nissl, and immunofluorescent staining. Results We found that ANRIL knockdown increased cell viability and reduced apoptosis in vitro. Additionally, we found that ANRIL knockdown decreased p-P65, P65, IL-1β, IL-6, TNF-a, and iNOS levels, whereas these effects were reversed by NF-κB overexpression both in vitro and in vivo. Conclusion our results suggest that ANRIL knockdown attenuates neuroinflammation by suppressing the expression of NF-κB both in vitro and vivo model of IS, sugguesting that ANRIL might be a potentially viable therapeutictarget to diminish neuroinflammation in IS patients. GRAPHICAL ABSTRACT


Introduction
Ischemic stroke (IS) is one of the most common causes of death and disability in the world, with more than two million young adults (18-50 years old) suffering from IS episodes each year [1,2]. After a stroke, patients are often faced with aphasia or hemiplegia, leading to significant reductions in their quality of life and mental health.
Inflammation and oxidative stress occurring secondary to IS can also cause irreversible neuronal damage, with unrestrained neuroinflammation having the potential to exacerbate this damage [3,4]. Accumulating evidence indicates that inflammation plays a key role in the pathogenesis of stroke, and as such, it has become a target for therapeutic intervention [5]. The pro-inflammatory cytokines such as interleukin-1β (IL-1β), interleukin-6 (IL-6), and tumor necrosis factor (TNF-α) begin to be produced rapidly within 6 h of ischemia, potentially causing neuronal injury [6]. In addition, these neurons are highly sensitive to irreversible damage caused by ischemia and hypoxia [7]. The therapeutic options available to treat IS are limited, and novel therapies capable of preventing IS-associated neuroinflammatory damage are urgently needed.
Long noncoding RNA over 200 nucleotides in length have been shown to play diverse roles in biological processes, controlling gene expression, mRNA splicing, and other epigenetic processes [8]. Long noncoding RNA have been shown to be highly expressed in the central nervous system [9], with some also being closely tied to IS pathogenesis [10,11]. Long noncoding RNA ANRIL (ANRIL), which is encoded by the chromosome 9p21 region in humans [12]. Previous studies have demonstrated that ANRIL is associated with metabolic disease [13], cancer, and cardiovascular disease [14,15]. ANRIL can additionally influence the pathology of coronary artery disease owing to its ability to modulate microRNA −181b [16]. ANRIL protects H9c2 cells against hypoxia-induced injury by targeting the miR-7-5p/SIRT1 axis [17]. Some prior bioinformatics studies have reported that ANRIL expression is altered in IS patients [18][19][20], but its mechanism role in IS requires further clarification.
The present study was designed to explore the underlying mechanisms whereby ANRIL influences IS-related injury. In vitro, HT22 mice hippocampal neuron-line cells was treated with an oxygen-glucose deprivation/reperfusion (OGD/R). In vivo, brain ischemia/reperfusion was induced by 60-minute transient middle cerebral artery occlusion/reperfusion (MCAO/R) IS model in C57/BL6 mice. Our results suggest that ANRIL knockdown attenuates neuroinflammation by suppressing the expression of NF-κB both in vitro and vivo model of IS.

Animals
In total, 183 male C57BL/6 mice (20-25 g) were housed under standard conditions in a 25 ± 2°C, facility with 60-70% relative humidity and a 12 h light/dark cycle. Animals had free access to food and water. The Chongqing Medical Animal Experiment Center oversaw all studies, which had additionally been approved by The Ethics Committee of Animal laboratory of Chongqing Medical University (License Number: SYXK YU 2010-001). Animal suffering was minimized wherever possible.

MCAO/R model
We developed an MCAO model of IS based upon previously published methods [21]. Briefly, animals were anesthetized using intraperitoneally injected pentobarbital sodium (40 mg/kg) (sunlidabio, China) and were then placed in a supine position. The skin was then incised, and the right common carotid artery (CCA), internal carotid artery (ICA), and external carotid artery (ECA) were exposed and dissected away from adjacent nerves. A silicone nylon suture (0.21 mm in diameter, Fengteng Biology, China) was then inserted through the CCA into the ICA such that it blocked the middle cerebral artery. After 60 minutes, this blockage was removed and the CCA was ligated at the site of the incision. The wound was then disinfected using iodine and was sutured. Animals were warmed to 37.5 ± 0.5°C during surgery and recovery using heating pads. In sham-operated animals, all surgical procedures were identical to procedure in MCAO model mice, but the silicone nylon suture was not inserted.

Animal treatment groups
For studies of ANRIL expression in MCAO/R mice, 30 mice were randomly assigned to the following 5 groups: Sham, MCAO/R 0 h, MCAO (2 μl, 20 μM, GenePharma, China), a negative control siRNA (si-NC, GenePharma), pcDNA3.1-NF-κB (2 μl, 1 μg/μl, GenePharma), or a negative control pcDNA (pcDNA3.1-NC, GenePharma) were administered into the lateral ventricles (injected at 0.2 uL/min with a mini-pump [RWD, China]) of different groups of mice at 72 h and 24 h prior to ischemia, respectively. Animals in the sham and MCAO model groups were administered ddH 2 O as a control using the same injection conditions. The mini-pump syringe (RWD, Ningbo, China) was inserted into the right lateral ventricle using the following coordinates positon: 1.0 mm lateral and 0.5 mm posterior, and 3.0 mm below the horizontal plane of bregma [22]. Mice hippocampal samples were then collected for analysis 24 h after MCAO/R.

OGD/R treatment
Briefly, HT22 cells were grown for 24 h under normal conditions, after which they were transferred to glucose-free DMEM (Gibco, USA) for 2 h under hypoxic conditions (1.0% O 2 , 93.5% N 2 , 5%CO 2 ) at 37°C. The glucose-free DMEM only contained L-glutamine but without D-glucose, sodium pyruvate or fetal calf serum. Then cells were grown under normal culture conditions for 24 h to simulate reperfusion. For studies of ANRIL expression in OGD/R HT22 cells, we assigned groups as follows: Control, OGD/R/R 0 h, OGD/R 8 h, OGD/R 16 h, and OGD/R 24 h.

Cell viability assay
A Cell Proliferation and Cytotoxicity Assay Kit (Methylthiazolyldiphenyl-tetrazolium bromide, MTT) was used to measure cell viability according to provided directions. Cells were seeded in 96-well plates at a density of 10,000 cells per well. Subsequently, cells were incubated with 20 μl MTT (5 mg/ml, Sigma, USA) at 37°C for an additional 2 h. At the end of this incubation, the culture medium was removed and 150 μl of dimethyl sulfoxide (Sigma) was added to dissolve the formazan crystals in these wells. Absorbance at 490 nm was measured using a fluorescence plate reader (Thermo Scientific, USA).

Apoptosis measurement
Briefly, HT22 cells from different treatment groups were harvested and resuspended at 6 × 10 5 /ml, the percentage of apoptotic cells was analyzed via fluorescence-activated cell sorting following Annexin V-FITC/PI staining according to provided instructions.

Enzyme-linked immunosorbent assay (ELISA)
RIPA buffer (Dingguo, China) was used to prepare proteins from cell lysates and brain homogenates. HT22 cells were dissociated by 100 μl RIPA buffer, which were mixed with complete culture medium of cells for detecting. Hippocampus (18-25 mg) were dissociated by RIPA (10 mg hippocampus: 100 μl RIPA) and diluted up to 1 ml by using PBS for detecting. Levels of IL-1β, IL-6, TNF-α, and iNOS in these samples were then measured using commercial ELISA kits (Beyotime) based on provided directions.

Neurological deficits
At 24 h post-MCAO, mice neurological functions were assessed by a researcher blinded to group assignments using a modified version of a scoring system previously reporte [23]: 0 = no apparent deficits; 1 = difficulty with full extension of the contralateral forelimb; 2 = unable to extend the contralateral forelimb; 3 = mild contralateral circling; 4 = severe circling; 5 = falling to the contralateral side or death. Animals with scores of 0 were considered to have undergone unsuccessful IS modeling, whereas those with a score of 5 were excluded from the study and euthanized due to the severity of their illness.

Infarcted area
At 24 h post-MCAO, hippocampal tissue samples were collected from mice and used to prepare 1.5 mmthick coronal sections. Sections were stained for 10 minutes with 2% TTC (2,3,5-Triphenyltetrazolium chloride, Sigma) at 37°C, after which 4% paraformaldehyde was used to terminate staining. In stained sections, infarcted tissue appeared white whereas all other tissue appeared red. The ImageJ software was used to measure the infarcted area as a proportion of the total area to evaluate infarction severity as follows: infarcted area (%) = [the infarcted area -(the ipsilateral hemisphere area-the intact contralateral hemisphere area)] ×[100/ 2× the intact contralateral hemisphere area] [24].

Nissl staining
Briefly, three mice per treatment group were selected at random, anesthetized using pentobarbital sodium (40 mg/kg), and perfused with 4% paraformaldehyde prior to Nissl staining. Hippocampal CA1 staining was evaluated by a researcher blinded to experimental protocols. The nulear of injured cell shrunk.

Rota-rod test
Murine motor coordination was evaluated using a rota-rod (Ugo Basile, Italy). Briefly, animals were trained on an accelerating (4-40 rpm over 5 minutes) rota-rod cylinder for 3 days prior to MCAO in order to obtain baseline measurements. All animals were able to remain on the rota-rod for at least 160 seconds at baseline, consistent with prior studies [25]. At 24 h post-MCAO, animals were placed on the rota-rod apparatus using the same acceleration conditions as above. A researcher blinded to group experiments then observed animal performance and recorded latency to fall during each trial, with a maximum time of 160 s. Animals were tested three times per day with 10 minutes between trials, and the mean time for each animal was then calculated.

Statistical analysis
All data are expressed as the mean ± standard error of mean (SME). SPSS v21.0 (SPSS Inc., USA) was used for all statistical testing. The experiments were repeated three times in vitro. The number of mice in each group was noted in the legend. Data were compared via one-way ANOVAs with Tukey's post hoc test, Kruskal-wallis test, with p < 0.05 as the significance threshold.

ANRIL expression is elevated in IS model systems in vitro and in vivo
ANRIL expression was upregulated in hippocampal samples at 8, 16, and 24 h post-MCAO compared to Sham controls (Figure 1(a)). Similarly, at 0, 8, 16, and 24 h post-OGD/R, ANRIL expression was significantly increased in HT22 cells compared to untreated cells (Figure 1(b)). These results revealed that ANRIL expression was persistently increased within 24 hours after MCAO. Based on these results, we selected at MCAO/R 24 h time point for subsequent studies both in vitro and in vivo.

ANRIL knockdown reduces OGD/R-induced injury and inflammatory cytokine production in HT22 cells
We confirmed that we were able to successfully knock down or overexpress ANRIL and si-NC, pcDNA3.1-NC had no impact on ANRIL expression in OGD/R-treated cells (Figure 2(a)), whereas did not impact HT22 cells under normal growth conditions (Figure 2(b)). However, we found that ANRIL knockdown increased OGD/R-treated cell viability (Figure 2 (c)) and diminished OGD/R-induced cellular apoptosis (Figure 2(d,e)), while ANRIL overexpression had the opposite effect. In addition, our results showed that ANRIL knockdown decreased NF-κB mRNA levels via qRT-PCR (Figure 2(f)), reduced p-P65 and P65 protein levels via Western blotting (Figure 2(g,h)), and suppressed the production of the proinflammatory cytokines IL-1β, IL-6, TNF-a, and iNOS as assessed via ELISA (Figure 2(i)).

ANRIL knockdown reduces OGD/R-induced inflammation by suppressing NF-κB expression
We successfully overexpressed NF-κB in OGD/ R-treated cells, and ANRIL knockdown reduced the expression of NF-κB in OGD/R-treated cell, whereas NF-κB overexpression reversed this effect (Figure 3  (a)). si-ANRIL reduced the expression of ANRIL in OGD/R-treated cell, whereas NF-κB overexpression hadn't effect the the expression of ANRIL (Figure 3 (b)). We found that ANRIL knockdown increased OGD/R-treated cell viability, whereas NF-κB overexpression reversed this effect (Figure 3(c)). ANRIL knockdown was associated with reductions in OGD/ R-induced p-P65 and P65 levels, while NF-κB overexpression also reversed this phenotype in cells via Western blotting (Figure 3(d,e)). ANRIL knockdown also reduced IL-1β, IL-6, TNF-a, and iNOS levels, while NF-κB overexpression restored the production of these cytokines (Figure 3(f)). Immunofluorescence analyses also revealed that ANRIL knockdown suppressed the OGD/R-induced increases in nuclear and cytoplasmic P65 levels (Figure 3(g)), whereas these cytoplasmic levels were decreased by si-ANRIL treatment.

Knockdown of ANRIL protects mice against cerebral ischemia-reperfusion injury
We found that si-ANRIL reduced ANRIL expression successfully in MCAO/R model mice, whereas si-NC had no impact on ANRIL expression in these animals (Figure 4(a)). Neurological deficits scores was elevated in MCAO/R mice, and reduced by si-ANRI (Figure 4  (b)). We observed a significant increase in infarcted area (30.6%) in MCAO/R model animals, while the infarcted area was reduced by 12.9% upon si-ANRIL treatment (Figure 4(c,d)). The number of injured neurons in the CA1 region of MCAO/R group was higher than in sham controls, but these numbers were reduced by si-ANRIL treatment (Figure 4(e,f)). Immunofluorescent NeuN staining also indicated that the survival of hippocampal neurons was significantly enhanced in MCAO/R mice with si-ANRIL treatment (Figure 4(g)).

ANRIL knockdown reduces MCAO/R-induced neuroinflammation in mice by suppressing NF-κB expression
pcDNA3.1-NF-κB was sufficient to increase NF-κB expression in MCAO/R mice, whereas pcDNA-NC had no impact on NF-κB expression ( Figure 5(a)). si-ANRIL significantly reduced ANRIL expression, howerver, NF-κB overexpression had no effect ANRIL expression (Figure 5(b)). ANRIL knockdown significantly reduced NF-κB mRNA expression ( Figure 5(c)) and reduced p-P65 and P65 levels in the hippocampus of MCAO/R mice, whereas these effects were reversed by NF-κB overexpression ( Figure 5(d,e)). Pro-inflammatory IL-1β, IL-6, TNFa, and iNOS levels were increased in MCAO/R mice ( Figure 5(f)), while these levels were decreased by ANRIL knockdown. However, NF-κB overexpression reversed these effects. Neurological deficits scores was elevated in MCAO/R mice, and reduced by si-ANRI, NF-κB overexpression reversed these effects ( Figure 5(g)). We additionally found that fall latency was significantly reduced in MCAO/R model animals relative to sham controls in a rota-rod test, whereas this phenotype was significantly improved in animals in the MCAO/R+ si-ANRIL treatment group, and NF-κB overexpression reversed these effects ( Figure 5(h)).

Discussion
ANRIL has previously been studied as a biomarker associated with IS risk [26]. We observed that ANRIL expression was elevated in the hippocampus of MCAO/R mice, consistent with previous reports [18,19]. Our results revealed that ANRIL knockdown reduced OGD/R-induced injury and associated inflammation, whereas ANRIL overexpression aggravated these phenotypes. ANRIL knockdown suppressed neuroinflammation by suppressing NF-κB expression both in vitro and in vivo. A previous study demonstrated that knockdown of circ-ANRIL improved OGD/R-induced cell damage, apoptosis, and inflammatory responses by inhibiting the NF-κB pathway in OGD/R-induced human brain microvascular endothelial cells [27]. Our data and this study together thus confirmed the protective effect of ANRIL knockdown in IS. However, another study reported that the overexpression ANRIL alleviated OGD-induced cell injury, as evidenced by increased cell viability and reduced apoptotic cell death, through a mechanism dependent upon the negative regulation of Mcl-1 expression in the rat adrenal medulla-derived pheochromocytoma PC-12 cell line [28]. While there is thus substantial evidence that ANRIL plays a role in IS, the mechanismft basis for this role requires further clarification. Our results provide further support for a model wherein ANRIL alleviates inflammation and thus influences IS pathophysiology.
It has been shown that NF-κB expression is increased in patients with acute ischemic stroke [29,30]. Enhanced NF-κB expression and activation in MCAO/R mice results in the uncontrolled release of the downstream pro-inflammatory cytokines such as IL-1β, IL-6, TNF-α, and iNOS, which play vital roles in neuronal injury [31,32]. Suppressing NF-κB activity is thus critical as a means of defending the neural parenchyma against brain ischemia. Our results indicated that ANRIL knockdown suppressed neuroinflammation by reducing NF-κB expression in cell lines and animal models of IS.
The NF-κB transcription factor can be composed of homo-or hetero-dimers of the p50, p52, p65, Rela-B, and c-Rel proteins, with p65 being essential for NF-κB functionality. We therefore measured p65 levels as a readout for overall NF-κB activity. Under steadystate conditions, this transcription factor is normally found in the cytoplasm bound to its inhibitor protein IκB. The phosphorylation and degradation of IκB ultimately result in NF-κB activation, in turn driving the upregulation of damaging pro-inflammatory factors that aggravate IS-induced neuronal damage [31,33] . We detected significantly elevated total NF-κB levels in OGD/R-treated HT22 cells and in hippocampal tissues from MCAO/R mice. These increases were likely attributable to a combination of hypoxia-and hypoglycemia-induced de novo NF-κB transcription and the release of NF-κB from its interaction with IκB. However, as ANRIL is a long noncoding RNA, it is only able to directly interact with NF-κB at the mRNA level and cannot influence this protein directly, preventing it from disrupting NF-κB protein activity following IκB dissociation. As ANRIL might influence total NF-κB expression by reducing de novo NF-κB transcription of mRNA NF-κB, and because translation is governed by cytoplasmic ribosomes, we therefore studied total NF-κB levels in our samples without analyzing the subcellular localization of this transcription factor via Western blotting. Additionally, cellular immunofluorescent staining results indicated that si-ANRIL was able to reduce cytoplasmic NF-κB levels in HT22 cells after OGD/R.
Our results revealed that ANRIL was upregulated in OGD/R-treated HT22 cells. However, ANRIL (e, f) Nissl staining was used to visualize neuronal loss, n = 3, magnification: × 400, black arrow: normal cell, red arrow: injured cell whose nucleu shrunk. (g) Immunofluorescent analysis of mice hippocampus tissue using a NeuN antibody, magnification: × 400. (a, d, and f), One-way ANOVA with Tukey's post hoc test; B, Kruskal-wallis test; * P < 0.05 vs. Sham; # P < 0.05 vs. MCAO/R group. overexpression alone in cells cultured under normal conditions did not induce cellular injury. The reasons might be that overexpressed ANRIL need to interact with other damaging mechanisms induced by OGD/R as a means of driving enhanced cellular damage.
IS is associated with damage to the cortical regions of the brain, which contain high levels of astrocytes and microglia. Following IS, these cell types produce large volumes of pro-inflammatory cytokines and chemokines [5,34]. It is possible that ANRIL may play distinct roles in cortical astrocytes and microglia in MCAO/R model, but future research will be required to support such a hypothesis.
Long noncoding RNAs are capable of promoting or suppressing the expression of specific target genes through a range of mechanisms including transcription factor recruitment, regulation of mRNA splicing, chromatin modification, the formation of heterogeneous nuclear ribonucleoprotein complexes, or via directly interacting with certain RNA or DNA targets through complementary base-pairing [35]. In the present study, we found that ANRIL knockdown reduced NF-κB expression both in OGD/R-treated cells and in MCAO/R mice, however, overexpressed NF-κB did not impact ANRIL expression levels in OGD/ R-treated cells or in MCAO/R-treated mice, which indicated that ANRIL regulates NF-κB expression at the mRNA level but is not reciprocally regulated by NF-κB in vitro and in vivo. In the future, we will explore the mechanistic basis for this ANRILmediated regulation of NF-κB and will establish the cause of ANRIL upregulation in IS.

Conclusion
Our results indicate that ANRIL plays a functionally important role in the pathology of cerebral ischemia reperfusion injury. ANRIL knockdown can suppress the expression of NF-κB and consequent neuroinflammation in the hippocampus of MCAO/R mice, suggesting that ANRIL might be a potentially viable therapeutic target to diminish neuroinflammation in IS patients.