Astrocyte KDM4A mediates chemokines and drives neutrophil infiltration to aggravate cerebral ischemia and reperfusion injury

Abstract

Neutrophils plays a crucial role in acute ischemic brain injury and have emerged as potential treatment targets to mitigate such injuries. Lysine-specific demethylase 4 A (KDM4A), a member of the histone lysine demethylase family of enzymes involved in transcriptional regulation of gene expression, is upregulated during hypoxic events. However, the exact role of KDM4A in the pathological process of ischemic stroke remains largely unexplored. Our findings reveal that there was an upregulation of KDM4A levels in reactive astrocytes within both stroke mouse models and in vitro oxygen-glucose deprivation/regeneration (OGD/R) models. Using a conditional knockout mouse, we observed that astrocytic Kdm4a knockout regulates neutrophil infiltration and alleviates brain injury following middle cerebral artery occlusion reperfusion. Furthermore, Kdm4a deficiency astrocytes displayed lower chemokine C-X-C motif ligand 1 (CXCL1) level upon OGD/R and decreased neutrophil infiltration in a transwell system. Mechanistically, KDM4A, in cooperation with nuclear factor-kappa B (NF-κB), activates Cxcl1 gene expression by demethylating histone H3 lysine 9 trimethylation at Cxcl1 gene promoters in astrocytes upon OGD/R injury. Our findings suggest that astrocyte KDM4A-mediated Cxcl1 activation contributes to neutrophil infiltration via cooperation with NF-κB, and KDM4A in astrocytes may serve as a potential therapeutic target to modulate neutrophil infiltration after stroke.

Introduction

Stroke is one of the leading causes of death worldwide. ¹ Ischemic stroke, due to inadequate blood flow to a specific brain region, accounts for approximately 87% of all stroke cases. ² Following acute ischemic stroke, the infiltration of neutrophils and other leukocytes that occur within a few hours plays a critical role in neuroinflammation. ³ Manipulating the neutrophils infiltration has emerged as a potential tactic for treating stroke, and therapeutic interventions have shown promise in experimental ischemic stroke mice. ⁴ Despite these advancements, our knowledge of immunomodulation is still limited. The molecular mechanism governing neutrophil infiltration post stroke is complex and still far from understood.

Astrocytes are the most plentiful non-neuronal cell type in the higher mammalian central nervous system (CNS). ⁵ They play critical roles in both the development and damage of the CNS. ⁵ Astrocytes also engage in cross-talk with the infiltrating immune cells in the inflamed brain area, being a major source of proinflammatory chemokines that induce lymphocyte infiltration.^6,7 Evidence has shown that astrocytes have the potential to exert potent proinflammatory functions by producing chemokine including the CXC chemokines ligand 1 (CXCL1), CC-Chemokines ligand 2 and other chemokines, as their primary mode of action post brain injury. ⁸ Nuclear factor-kappa B (NF-κB) is a transcription factor that plays a critical role in the expression of proinflammatory genes including cytokines, chemokines, and adhesion molecules. ⁹ Furthermore, apart from transcription factors, gene transcription is also influenced by histone modification.^{10 –13} However, the precise roles of histone methylases or demethylases in chemokines regulation post ischemic stroke remain obscure.

Lysine (K)-specific demethylase 4 A (KDM4A), an enzyme that catalyzes the demethylation of histone H3 lysine 9 trimethylation (H3K9me3), is a member of the Jumonji domain 2 families. ¹⁴ KDM4A has been found to be highly sensitive to oxygen concentrations and is upregulated under hypoxic conditions.^15,16 The repression of gene transcription of KDM4A involves binding directly to a transcription factor such as the NF-κB suppressor. Diverse physiological or pathological functions of KDM4A have been identified in human subjects and animal models. Recently, KMM4A has been reported to participate in immune cell activation and differentiation. ¹⁷ Inhibition of KDM4A activity is an effective strategy to suppress interleukin-6 production in activated fibroblasts. ¹⁸ However, the role of KDM4A in the context of stroke and its underlying mechanism remain largely unknown.

In the present study, we provide evidence that KDM4A is up-regulated in astrocytes after stroke. Moreover, we find that KDM4A, in cooperation with NF-κB, activates cxcl1 gene expression in astrocytes and promote neutrophil infiltration following stroke. Our data suggest that astrocytic KDM4A is a potential therapeutic target for treating ischemic stroke.

Materials and methods

Reagents and antibodies

Dulbecco’s modified Eagle’s medium (DMEM), Neurobasal medium, B27, L-glutamine, penicillin and streptomycin, collagenase D, and fetal bovine serum (FBS) were purchased from Invitrogen (Carlsbad, CA, USA). Antibodies against CXCL1, interleukin 6 (IL-6), and Tumor necrosis factor-α (TNF-α) were purchased from Cell Signaling Technology (Boston, MA, USA). Antibodies against KDM4A, KDM4B, KDM4C, NF-κB-p65, activator protein-1 (AP1), specificity protein 1 (SP1), nuclear factor of IL-6 (NF-IL6), H3K9me3, neuron-specific nuclear protein (NeuN), glial fibrillary acidic protein (GFAP), anti-β-actin, and horseradish peroxidase-conjugated secondary antibodies for chromatin immunoprecipitation (ChIP), co-immunoprecipitation or Western blot were purchased from Millipore (Carlsbad, CA, USA). RNA extraction and reverse transcription kit and SYBR Premix ExTaq were purchased from Takara (Tokyo, Japan). The ChIP assay kit was purchased from Abcam (Carlsbad, CA, USA). The Terminal deoxynucleotidyl transferase-mediated dUTP nick-end labeling (TUNEL) staining in situ cell death detection kit was obtained from Beyotime Institute of Biotechnology (Nanjing, China). The bichioninic acid (BCA) kit, Hanks balanced salt solution (HBSS), trypsin, and enhanced chemiluminescence were purchased from Thermo Scientific Incorporated (Waltham, MA, USA). The pGV118 lentiviral vectors were prepared by CeneChem Incorporated (Shanghai, China). Glutamate/aspartate transporter-Allophycocyanin (GLAST‐APC) antibody was purchased from Miltenyi Biotec. (Bergisch Gladbach, Germany). Antibodies used for flow cytometry were purchased in Becton, Dickinson and Company (BD) Bioscience (San Jose, CA, USA). Mouse XL Cytokine Array Kit and protease inhibitor were purchased from Research & Diagnostics Systems Incorporated (Minneapolis, MN, USA). Magnetic-activated cell sorting (MACS) kit (astrocyte cell surface antigen-2) and mouse neutrophil isolation kit were purchased from Miltenyi Biotec. (Bergisch Gladbach, Germany). Rapain, collagenase, Deoxyribonuclease (DNase) I, Tosyl Lysyl Chloromethyl Ketone (TLCK) trypsin inhibitor, N-2-hydroxyethylpiperazine-N-ethane-sulphonicacid (HEPES), Triphenyl tetrazolium chloride (TTC), 4, 6-diamidino-2-phenylindole (DAPI), and Evans Blue, Papain, ethylenediaminetetraacetic acid (EDTA) were purchased from Sigma-Aldrich (St Louis, MO, USA). SMARTer Ultra Low Input RNA Kit for sequencing was obtained from Clontech Laboratories Incorporated (Palo Alto, CA, USA).

Animals

Kdm4a^flox/flox mice and GFAP-cyclization recombination (Cre) transgenic mice expressing a GFAP promoter were purchased from Cyagen Incorporated (Suzhou, China). The genotype of KDM4A^flox/flox mice (Fig. S1A) and GFAP-Cre mice (Fig. S1B) were verified by polymerase chain reaction (PCR). The following PCR primer sequences were used for genotyping Kdm4a^flox/flox (forward: 5′-CCA ACA ATCA ACT AAA TAA CACC-3′, reverse: 5′-CAA CTT GTA GTA CAT TTA TTC-3′) and GFAP-Cre (Forward: 5′-GCG GTC TGG CAG TAA AAA CTA TC-3′, Reverse: 5′-CCT TCC AGG GCG CGA GTT GAT AGCT-3′). To generate astrocyte-specific Kdm4a knockout mice (Kdm4a KO), Kdm4a^flox/flox mice were hybridized with GFAP-Cre transgenic mice. Littermate Kdm4a^flox/flox/Cre^- mice were used as controls. Mice were housed at 25 °C under a 12 h light/dark cycle and had free access to food and water. Adult (10–12 weeks old) male mice were used for study. All animal experiments comply with the Animal Research: Reporting of In Vivo Experiments guidelines and are carried out in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals. All animal experimental procedures were approved by the Animal Experiments Ethics Committee of Fourth Military Medical University.

Focal cerebral ischemia

Temporary focal cerebral ischemia was induced by middle cerebral artery occlusion reperfusion (MCAO/R) utilizing the intraluminal thread method as previously reported. ¹⁹ Mice were anesthetized via 2% isoflurane inhalation, the external carotid artery (ECA) and right common carotid artery (CCA) were isolated. The incision near the ECA-CCA branch was inserted into a 6–0 silk suture coated with silicone and advanced about 9–11 mm to the origin of the middle cerebral artery (MCA). Reperfusion was performed 60 min after MCAO. The procedures performed in the sham mice were the same as the model group, except the nylon filament was advanced only about 5 mm excluded occlusion of MCA. The animals’ rectal temperature was kept at 37 °C during and after surgery using a heating pad. To ensure the success of the operation, the degree of cerebral blood flow reduction was monitored using laser speckle contrast imaging (RFLSI III, RWD Incorporated, Shenzhen, China) after the occlusion and reperfusion (Fig. S2).

Neurological deficit scores test and behavioral test

Neurological deficits were assessed using the modified Neurological Severity Score (mNSS) test. ²⁰ The scores are graded from 0 to 18. No neurological deficit is denoted by a score of 0. Mild, moderate, and severe deficits are indicated by scores 1 to 6, 7 to 12, and 13 to 18, respectively.

The foot fault test was performed on an 8 × 10 square grids of 2 × 2 cm. The forelimb and hindlimb were placed on the wire while moving along the grid. Falling or slipping during moving between grid was recorded as a foot fault.

The rotarod test was performed to evaluate motor functional change. ²¹ Briefly, mice were forced to walk on a rotating cylinder (LE8205, Panlab, Barcelona, Spain) with speeds at 40 revolutions/min. Four consecutive trials were tested for each mouse with an interval of 10 min. The falling latencies were recorded via a photobeam circuit.

TTC staining

TTC assay was performed as previously described. ²² Mice were anaesthetized and decapitated 24 h after reperfusion. Brains were dissected, sectioned at a thickness of 2 mm, and then immersed in 1.5% TTC solution at 37 °C for 30 min. The sections were then fixed in 4% paraformaldehyde overnight and photographed by a camera (EOS R, Canon, Tokyo, Japan) equipped with 100 mm macro lens (Canon). Infarct volume was analyzed using Photoshop (Version CC 2019).

Separation of penumbra

The separation of penumbra was performed based on the average infarct volume measured by TTC staining in our pilot experiments. Mice were subjected to deep anesthesia and euthanized with an overdose of isoflurane, the brain tissue was quickly collected and put on ice and the olfactory bulb and cerebellum were removed. Firstly, the brain tissue was cut on a coronal plane 3 mm backward from the top of the frontal lobe into three slices with thickness of 3 mm (section 1), 4 mm (section 2) and 3 mm (section 3), respectively (Fig. S3A). Then, section 2 was taken out. Next, a sagittal cut 1.5 mm far from the midline was conducted from top to bottom on the ipsilateral hemisphere, and a transverse diagonal cut was also made at around the “1 o'clock” position. Thus, the tissue outside the “1 o'clock” position was the infarct core area, and the cortical tissue between sagittal cut and the “1 o'clock” position was the ischemic penumbra (Fig. S3B).

Evans blue staining

The impaired blood brain barrier (BBB) integrity was assessed by Evans Blue leakage in the brain. Evans Blue dye (2%, 4 mL/kg) was intravenously injected via the tail vein 22 h after reperfusion. After 2 h of circulation, the mice were perfused with phosphate buffered saline (PBS). The brains were isolated, froze and cut into 2-mm-thick coronal slices, which were photographed by a camera (EOS R, Canon) equipped with 100 mm macro lens (Canon) to visualize the dye leakage. Then, a quantitative assessment of the dye content in the ischemic hemispheric tissue was performed. The ipsilateral hemisphere was weighed. Brain tissue was then homogenized in a tube containing formamide (5 mL) and incubated in a 60 °C water bath for 72 h. After centrifuging at 300 g for 5 min, the supernatant was collected and added along with standards to 96-well plates. The optical density of the supernatant was measured at λ = 450, 570 nm via a microplate reader (Multiskan MK3, Thermo Scientific Incorporated). Evans Blue content in brain tissue was calculated as the following formula: Evans Blue concentration × formamide (mL)/wet weight (g).

Immunohistochemistry

Mice were anesthetized and perfused with saline, followed by 4% paraformaldehyde. The brain was removed, post-fixed in 4% paraformaldehyde for 12 h, cryoprotected, and embedded as previously described. The brains were cut coronally (20 μm) with a cryostat microtome (Leica biosystem, Weztlar, Germany). The sections were immunohistochemically stained as previously reported. ²³ The sections were incubated in the following antibodies for 12 h at 4 °C: anti-GFAP (1:500 dilution), anti-KDM4A (1:500 dilution), anti-NeuN (1:500 dilution). The sections were then gently washed with phosphate buffer 3 times and incubated in secondary antibody Alexa Fluor 488/594 goat anti-rabbit antibody for 1 h. Finally, the sections were stained with DAPI (1:500 dilution) for 10 min at room temperature. The sections were mounted with a coverslip, observed under an Olympus IX-71 fluorescence microscope (Olympus, Tokyo, Japan), and photographed.

Terminal deoxynucleotidyl transferase-mediated dUTP nick end labeling (TUNEL) staining

Cell apoptosis was assessed by TUNEL staining using an In Situ Cell Death Detection Kit. Brain tissue sections were treated following the procedure of the manufacturer. The nuclei of cells were stained with DAPI. The stained sections were imaged via a confocal fluorescent microscope.

RNA-sequencing analysis

RNA sequencing was performed by Hanheng Biotechnology Corporation. Briefly, astrocytes isolated from the penumbra of Kdm4a KO or Control mice 24 h after MCAO/R. RNA was exacted and amplified using the SMARTer Ultra Low Input RNA Kit for sequencing, then cDNA was fragmented. Paired-end libraries were constructed by using the Ovation Ultralow DR Multiplex System 1-96 (NuGEN Technologies, Puerto San Calros, CA, USA). The libraries were then sequenced on an Illumina HiSeq 2500 system (SY-401-2501) and mapped to the mouse reference genome using TopHat (version 2.0.9). Differential gene expression was determined using Cufflinks (version 2.1.1). Fold expression change ≥2 was considered significance. Gene Ontology analysis was performed using Ingenuity Pathway Analysis software (version 3.1.1).

Flow cytometry

Single-cell suspensions were prepared from brain tissues and stained with fluorochrome-conjugated antibodies. Antibodies were directly labeled with fluorescein isothiocyanate, phycoerythrin, PerCP-Cy5.5, or allophycocyanin. The following reactive antibodies were used, GFAP, cluster of differentiation (CD) 3, CD11b, CD4, CD8, CD19, CD45, lymphocyte antigen 6 complex, locus G (Ly6G), Natural Killer cell (NK) 1.1. Flow cytometry and data analysis were performed with fluorescence-activated cell sorting (FACS) Aria Cell Sorter (BD Bioscience) and FACS C6 (BD Bioscience) analyzing software.

Proteome profiler mouse cytokine array

Twenty-four hours post MCAO/R, astrocytes were purified from the penumbra of Control or Kdm4a KO mice by using the FACS method. Briefly, mice were anesthetized via isoflurane inhalation, and brains were harvested following transcardial perfusion with 0.9% saline. The penumbra region was dissected and then subjected to enzymatic digestion for 20 min at 37 °C. This digestion process involved the use of 24 Unit/mL Papain in HBSS. Following that, an additional 10 min digestion step was carried out after adding 1 mg/mL collagenase D and 0.1 mg/mL DNase I. Enzymatic reactions were neutralized by the addition of 10% FBS. Subsequently, the suspension was gently triturated, followed by filtration over a 100 μm strainer to obtain a single-cell suspension. Cells were separated from myelin and debris by two-phase Percoll gradient centrifugation and washed twice with FACS buffer. Afterward, cells were stained with GLAST‐APC antibody for 30 min at 4 °C and washed twice with FACS buffer. The final pellet was resuspended in 50 µL PBS containing 0.3% EDTA. Cells were sorted using a Beckman Coulter MoFlo Astrios cell sorter. Sorted astrocytes were pooled for subsequent proteome profiler mouse cytokine array. Then cells were lysed with lysis buffer containing protease inhibitor mixture at 4 °C for 30 min. Protein concentrations were determined with a protein BCA kit. Samples were analyzed with a Mouse XL Cytokine Array Kit, according to the manufacturer’s instructions. Immunospots data were analyzed with ImageJ software (Version 1.43c/Java 1.6.0_14).

Astrocyte culture

Primary cultured astrocytes were prepared from the cortex of newborn mice as previously demonstrated. ²⁴ Generally, after carefully removing the meninges, the neopallium was dissected and then digested with 0.25% trypsin at 37 °C for 10 min. After trituration, 1 × 10⁶ cells/mL were cultured in poly-d-lysine coated flasks in the DMEM replenished with 10% FBS. The culture medium was changed at 24 h, and then replaced every 72 h. Astrocytes were purified via vigorous shaking for 24 h at 37 °C. Cell culture purity was determined by GFAP expression detected using immunofluorescent microscopy. After sub-culturing twice, 12–13 day old astrocytes were utilized for further experimentation.

Neuron culture

Primary mouse cortical neuronal cells were isolated from the cortex of 17–18 day embryonic development mice as previously described. ²⁵ The embryonic brain cortex was precisely dissected in HBSS on a 60 mm Petri dish using a stereo microscope (Leica biosystem). The tissue was then transferred to a 15 mL tube and digested in HBSS with 0.25% trypsin for 15 min at 37 °C. Thereafter, cells were mechanically disaggregated by a glass Pasteur pipette in HBSS containing 0.1 mg/mL DNase I. The supernatant was then transferred to a new 15 mL tube, and the non-dissociated tissue was discarded. After trituration twice, the supernatant was centrifuged at 250 g for 5 min. Cell pellets were resuspended in 10 mL Neurobasal medium supplemented 2% B27, 0.5 mM L-glutamine, and 100 Units/mL penicillin and streptomycin. Dissociated neurons were cultured in poly-D-lysine coated plates after counting. 50% of the culture medium was changed every 3 days.

Neutrophil migration assay

Neutrophils were isolated by using a neutrophil isolation kit. The neutrophils were then added to the upper chamber of transwell inserts positioned on 24 well plates pre-seeded with primary astrocytes. Transmigration of neutrophils was then evaluated by using flow cytometry analysis.

Isolation of microglia and astrocytes from adult brain with microbeads

Astrocytes and microglia were isolated from penumbra of Control or Kdm4a KO mice 24 h after MCAO/R. ²⁶ Brains were removed, minced, and enzymatic digested in HBSS containing 50 mg/mL collagenase D, 100 μg/ml TLCK trypsin inhibitor, 0.1 mg/mL DNase I and HEPES 7.2, for 1 h at 25 °C. The homogenate was pushed through a 70 μm strainer. Then cell pellet was resuspended in 30% percoll solution and centrifuged at 1,200 g for 5 min to remove myelin debris. Then cells were washed in PBS and resuspended in MACS buffer. ACSA-2⁺ astrocytes and CD11b⁺ Microglia were isolated using manual MACS sorting. Cells were incubated with FcR blocking reagent prior to labeling with ACSA-2 antibodies for magnetic isolation to prevent non-specific labeling of CD11b⁺ microglia. The purity of astrocytes and microglia was verified by flow cytometry.

Oxygen-glucose deprivation and regeneration (OGD/R)

Oxygen-glucose deprivation followed by regeneration was performed in astrocytes as described previously. ²¹ Briefly, the culture medium was replaced by a glucose-free DMEM buffer, then astrocytes were placed in a hypoxic humidified incubator flushed with a gas mixture (94% N₂/5% CO₂/1% O₂) for 6 h. Then the medium was replaced by a standard culture medium and astrocytes were cultured under normoxia conditions.

ChIP assay

The ChIP assay was performed as previously described. ²⁷ Briefly, cells were cross-linked in formaldehyde for 10 min and subsequently sonicated to shear the chromatin. The diluted lysates were immunoprecipitated with antibodies against KDM4A, NF-κB-p65, AP1, SP1, NF-IL6, and H3K9me3 for the ChIP assay at 4 °C. Target DNA was analyzed by reverse transcription -PCR. Data were normalized to input genomic DNA. Primer pairs used for ChIP assay are described in Table S1.

Promoter reporter assay

Mouse Cxcl1 gene promoter was amplified and cloned into pGL4.15 [luc2P/Hygro] vector. U87 cells were transfected transiently with Cxcl1 gene promoter-driven firefly luciferase (1 μg) in conjunction with a control thymidine kinase promoter-driven renilla luciferase (100 ng). Mutation of the NF-κB binding site in the Cxcl1 gene promoter was performed by PCR (Table S2). Cells were exposed to OGD/R condition and harvested for luciferase activity using the Dual-Luciferase Assay System with a Lumat BL 9507 luminometer (Berthold, Bad Wildbad, Germany). Renilla luciferase activity served as an internal control.

RNA extraction and real-time PCR

RNA was extracted via an RNA extraction kit according to the manufacturer’s instrument. After reverse transcription, the real-time PCR of each gene was performed for three independent experiments with a 7300 Real-time PCR system (Applied Biosystems, Carlsbad, CA, USA) using SYBR Premix ExTaq kit. Primer sequences are listed in Table S3. The relative expression of genes was calculated by the 2^−△△Ct method, with normalization to glyceraldehyde-3-phosphate dehydrogenase levels.

Co-immunoprecipitation

The interaction of KDM4A and NF-κB-p65 proteins was validated by co-immunoprecipitation. Total proteins were extracted in immunoprecipitation lysis buffer supplemented with protease inhibitors. After centrifugation at 3,000 g for 15 min in 4 °C, the supernatant was harvested, and 40 μL was reserved as input. Simply, anti-KDM4A, anti-NF-κB-p65, and anti-IgG antibodies were added into some of the cell lysates and incubated overnight on a shaker at 4 °C. The DNA-protein-IgG complexes were incubated with Protein AG Magnetic Beads at 4 °C for 4 h and washed with low salt buffer, high salt buffer, LiCl buffer, and Tris-ethylenediaminetetraacetic acid buffer. 1×sodium dodecyl sulfate page loading buffer was added to lysates and followed by boiling at 95 °C for 5 min, and the protein expression was analyzed by Western blot analysis.

Western blot

Penumbra tissue of stroke model mice was homogenized in cold lysis buffer containing 1 nM protease inhibitors. Protein concentrations were determined using a BCA kit. Equal amounts of total protein were loaded onto a sodium dodecyl sulfate page gel and transferred electrophoretically to polyvinylidene fluoride membranes (Millipore). The membranes were then blocked in TBS containing 5% nonfat milk for 2 h at room temperature and subsequently immunoblotted with primary antibodies against KDM4A (1:500), KDM4B (1:500), KDM4C (1:500), NF-κB-p65 (1:500) or β-actin (1:1,000) overnight at 4 °C and horseradish peroxidase-conjugated secondary antibodies at room temperature for 1 h. Protein bands were detected by enhanced chemiluminescence.

Statistical analysis

Data are presented as means ± standard deviation. The Shapiro-Wilk test was applied for all the datasets to test for a normal distribution. The choice of statistical calculation method was based on the number of groups to be compared. For comparison between any two groups, a Student’s t-test was performed. Statistical significance for comparisons between more than two independent groups was tested using a one-way analysis of variance (ANOVA), followed by a Dunnett t-test. A two-way ANOVA, followed by a Dunnett t-test, was performed for analyses involving multiple influencing variables. P-values < 0.05 were considered statistically significant.

Results

KDM4A is upregulated in astrocytes after acute brain ischemia

In the present study, we aimed to investigate the potential role of KDM4A post stroke. Firstly, we explored the expression of KDM4A, KDM4B, and KDM4C in MCAO/R mice. We found that only KDM4A expression, while not others, was significantly upregulated in the penumbra 24 h after MCAO/R injury (Figure 1(a) and (b)). Next, to determine which cell type exhibited KDM4A upregulation after stroke, we isolated and cultured neurons, astrocytes, and microglia from the brains of neonatal mice. We then subjected these cells to OGD/R. Real-time PCR analysis showed that astrocytes displayed the most significant Kdm4a upregulation (Figure 1(c)). Western blot analysis corroborated the real-time results, showing that KDM4A expression in astrocytes was remarkably elevated after OGD/R injury (Figure 1(d) and (e)). Additionally, We further examined KDM4A expression in ischemic penumbra by immunofluorescence. KDM4A primarily co-localized with astrocyte marker GFAP in stroke mice (Figure 1(f)). The number of KDM4A⁺ cells dramatically increased 24 h after injury (Figure 1(g)). Collectively, these data suggest that KDM4A is upregulated in the ischemic penumbra and that activated astrocytes are the main source of KDM4A.

Figure 1.

Lysine-specific demethylase 4 (KDM4)A expression is dramatically elevated in astrocytes post stroke. (a) Protein levels of KDM4 isoforms were examined by Western blot analysis from the penumbra of mice that underwent middle cerebral artery occlusion reperfusion (MCAO/R), with mice that underwent sham operation serving as the Control group. KDM4A, KDM4B, KDM4C, and β-actin are presented by the 130 Kilodaltons (KD), 127 KD, 142 KD, and 42 KD band, respectively. Each group (sham or MCAO/R) consisted of 3 bands from one mouse. (b) Band intensity was analyzed with ImageJ software and normalized to β-actin expression. Each data point represents the average intensity of 3 bands. Data are presented as percentages in comparison to the sham group (n = 3 mice per group). (c) Kdm4a transcription level in primary neurons, astrocytes, and microglia were detected by real-time polymerase chain reaction upon oxygen-glucose deprivation/regeneration condition (n = 6 mice per group). (d) KDM4A protein levels were examined by Western blot from astrocytes that underwent oxygen-glucose deprivation for 6 h followed by exposure to normal oxygen condition for 6 h, 12 h and 18 h. (e) Band intensity was analyzed with ImageJ software and normalized to β-actin expression. Data are presented as percentages in comparison to the Normoxia group (n = 3 per group). (f) Representative immunofluorescence staining for astrocytic marker, glial fibrillary acidic protein (GFAP, red) and KDM4A (green) from penumbra of brain tissue after MCAO/R. 4′,6-diamidino-2-phenylindole was used to stain the nuclei (blue). Scale bars, 50 μm and (g) Immunofluorescence intensity of KDM4A was quantified using ImageJ software (n = 4 per group). **P < 0.01.

Generation of mGFAP-Cre;Kdm4a^flox/flox mice for conditional Kdm4a knockout in astrocytes

To gain a better understanding of the role of astroglial KDM4A in the pathological process of cerebral ischemia/reperfusion (I/R), we ablated the Kdm4a gene under the promoter of mice GFAP via the Cre–loxP genetic system (Figure 2(a)). KDM4A expression was detected via Western blot assay in astrocytes isolated from Kdm4a KO and Control new born mice, the results confirmed the success generation of Cre-mediated astroglial Kdm4a knockout mice (Figure 2(b)). Kdm4a KO mice developed normally, showing no symptom of neurological disease, behavioral abnormalities, or infertility. There were also no differences in either gross brain structure or the number of astrocytes (Figure 2(c)) and neurons (Figure 2(d)) compared with Control mice.

Figure 2.

Generation and characterization of conditional astrocytic Lysine-specific demethylase 4 A (Kdm4a) knockout (Kdm4a KO) mice. (a) A schematic of crossing mouse glial fibrillary acidic protein (GFAP)-cyclization recombination (Cre) transgenic mice with Kdm4a^loxp/loxp mice to generate Kdm4a KO mice. (b) Primary cultured astrocytes were prepared from the cortex of newborn Kdm4a KO mice, deletion of KDM4A was confirmed by Western blot analysis, with astrocytes prepared from littermate KDM4A^flox/flox/Cre^- mice serving as the Control group. KDM4A and β-actin are presented by the 130 Kilodaltons (KD) and 42 KD band, respectively. Quantitation and statistical analysis of KDM4A band intensity was shown in right panel, band intensity was analyzed with ImageJ software and normalized to β-actin expression. Data are presented as percentages in comparison to the Control group (n = 3 mice per group). (c) Representative immunofluorescence staining for astrocytic marker, glial fibrillary acidic protein (GFAP, green) in brain section of Kdm4a KO and Control mice. 4′,6-diamidino-2-phenylindole (DAPI) was used to stain the nuclei (blue). Scale bar, 20 μm. Quantification of GFAP positive cells in Control and Kdm4a KO mice brain was shown in right panel (n = 4 mice per group) and (d) Representative immunofluorescence staining for neuron marker, neuron-specific nuclear protein (NeuN) in brain section of Kdm4a KO and Control mice. DAPI was used to stain the nuclei (blue). Scale bar, 20 μm. Quantification of NeuN positive cells in Control and Kdm4a KO mice brain was shown in right panel (n = 4 mice per group). **P < 0.01. n.s.: no significant; KO: Kdm4a KO; CTR: Control.

Target knockout of astrocytic Kdm4a reduces acute ischemic brain injury

To explore the potential participation of astrocyte-derived KDM4A in ischemic brain injury, we induced MCAO/R injury in both Control and Kdm4a KO mice. After induction of MCAO/R for 24 h, we first tested the neurological function. Kdm4a KO mice exhibited less impairment in the components of the mNSS (Figure 3(a), also see Fig. S4A) and foot-fault test (Figure 3(b), also see Fig. S4B). Kdm4a KO mice also displayed longer period in rotarod test (Figure 3(c) also see Fig. S4C). We then performed TTC staining to observe the infarct volume 24 h after reperfusion, the results indicated that Kdm4a KO had significantly smaller infarcts than Control mice (Figure 3(d), also see Fig. S5). Furthermore, examination of BBB permeability by Evans blue revealed that the Kdm4a KO mice had significantly smaller Evans blue staining area and less Evans blue extravasation compared to Control (Figure 3(e), also see Fig. S6). Astrogliosis was determined by immunofluorescence staining, results showed that Kdm4a KO mice had smaller number of GFAP⁺ reactive astrocytes in penumbra compared with Control mice 72 h after MCAO/R (Figure 3(f)). Neuronal apoptosis was further assessed by examining TUNEL and NeuN expression following injury. We found that the number of TUNEL/NeuN⁺ apoptotic cells was significantly lower in Kdm4a KO mice when compared to Control after MCAO/R (Figure 3(g)). Taken together, these results suggest that selective knockout of astrocyte KDM4A plays a protective role after ischemic stroke.

Figure 3.

Astrocyte-specific Lysine-specific demethylase 4 A (Kdm4a) knockout reduces brain injury after middle cerebral artery occlusion reperfusion (MCAO/R). (a-c) Neurological deficits evaluated by modified Neurological Severity Score (mNSS) (a), foot-fault test (b) and rotarod test (c) in Control and astrocyte-specific Kdm4a knockout (Kdm4a KO) mice 24 h after MCAO/R. (n = 8 mice per group). (d) Infarct volumes were determined by triphenyl tetrazolium chloride staining 24 h after MCAO/R. Red area indicates the non-infarct tissue, white area indicates the infarct tissue. Infarct volume percentage were quantified by using photoshop CC 2019 software in right panel (n = 6 mice per group). (e) Evans Blue (2%, 4 mL/kg) was intravenously injected via the tail vein 22 h after reperfusion followed by 2 h of circulation. The brains were isolated, freezed and cut into 2-mm-thick coronal slices. The left panel shows the representative staining images from each group. The right panel indicates the quantitative assessment of dye content in the ischemic hemispheric tissue. Details procedure will be found in material and methods section (n = 6 mice per group). (f) Representative immunofluorescence staining for astrocytic marker, glial fibrillary acidic protein (GFAP, red) from penumbra of Control and Kdm4a KO mice after MCAO/R. 4′,6-diamidino-2-phenylindole (DAPI) was used to stain the nuclei (blue). Scale bars, 100 μm for left and middle images. GFAP-positive cells number was quantified by using ImageJ software (right panel) (n = 6 mice per group) and (g) TdT-mediated dUTP nick end labeling (TUNEL, red) and neuron marker, neuron-specific nuclear protein (NeuN, green) staining was used to observe the apoptosis neurons in the penumbra of Control and Kdm4a KO mice after MCAO/R. DAPI was used to stain the nuclei (blue), Scale bars, 100 μm. TUNEL-positive cells number was quantified by using ImageJ software (right panel) (n = 6 mice per group). **P < 0.01. KO: Kdm4a KO; CTR: Control.

Astrocyte KDM4A aggravates neutrophil infiltration after MCAO/R

To further investigate whether KDM4A affects astrocyte function after stroke, we isolated the astrocytes in the penumbra area of Control and Kdm4a KO mice after MCAO/R, and subsequent RNA-Sequencing analyses was performed. We found astrocyte KDM4A affected thousands of genes post-stroke (Figure 4(a)). To gain further insights, we conducted a Gene Ontology analysis of the differentially expressed genes. Then the nature of genes and potential shared pathways were overviewed using the Database for Annotation, Visualization and Integrated Discovery. We listed the top 10 enriched clusters and found several of enriched terms, including immune response and cytokine activity, are involved in leukocyte migration (Figure 4(b)). Therefore, astrocyte KDM4A may play a role in regulating the leukocyte migration after ischemic stroke.

Figure 4.

Effect of Lysine-specific demethylase 4 A (Kdm4a) knockout on the molecular profiles of astrocytes post stroke. (a) Heatmap of gene profiles of astrocytes in Control and astrocyte-specific Kdm4a knockout (Kdm4a KO) mice 24 h after middle cerebral artery occlusion reperfusion (MCAO/R). Red color indicates the upregulated genes, green color indicates the downregulated genes and (b) the most enriched clusters in the comparisons of Control and Kdm4a KO mice after MCAO/R.

We then measured the accumulation of leukocytes in the brains of Control and Kdm4a KO mice after MCAO/R by using flow cytometry (Figure 5(a)). Notably, there was no difference between Control and Kdm4a KO mice in the numbers of CD4⁺T cells, CD8⁺ T cells, B cells, macrophages and NK cells. However, Kdm4a KO mice had decreased accumulations of neutrophils in the brain (Figure 5(b)). There was no difference between Control and Kdm4a KO mice in the numbers of leukocyte in the spleen post-stroke (Figure 5(c)). To confirm the decreased infiltration of neutrophils within the penumbra region of Kdm4a KO mice after MCAO/R, we conducted immunofluorescence staining. As expected, a significant decrease in Ly6G-positive cells, indicative of neutrophils, was observed (Figure 5(d)). This significant difference persisted up to 5 days post MCAO/R (Figure 5(e)). Importantly, the spleens of Kdm4a KO and Control mice had comparable numbers of neutrophils post MCAO/R (Figure 5(f)). When depleting neutrophils in Control and Kdm4a KO mice, the mNSS (Figure 5(g)), foot fault (Figure 5(h)), period in rotarod test (Figure 5(i)) and infarct volumes (Figure 5(j)) were similar between two groups after MCAO/R. These collective findings suggest that knockout of astrocyte KDM4A may lead to a reduction in the accumulation of neutrophils in the ischemic brain, potentially contributing to a protective effect against cerebral I/R injury.

Figure 5.

Neutrophil infiltration is attenuated with Lysine-specific demethylase 4 A (Kdm4a) knockout in astrocytes. (a) Gating strategy for subtypes of leukocytes isolated from the penumbra area after middle cerebral artery occlusion reperfusion (MCAO/R). Arrows indicate the parent population that the subsequent plots are gated on. 1. Gating strategy used for identification of microglia (red box, CD11b⁺ CD45^medium) and monocytes (blue box, CD11b^high CD45^high), 2. identification of Natural Killer (NK) 1.1^low (green box) and NK 1.1^high (purple box), 3. identification of NK cells (CD3^low NK 1.1^high), 4. identification of CD4⁺ T cells (CD4^high) and CD8⁺ T cells (CD8^high), 5. identification of macrophages (CD11b^high F4/80^high), 6. identification of neutrophils (CD11b^high Ly6G^high), 7. identification of CD86-expressing microglia (CD86^high) and 8. CD206-expressing microglia (CD206^high). (b) Counts of CD4⁺ T cells, CD8⁺ T cells, NK cells, macrophages, neutrophils, CD86-expressing microglia and CD206-expressing microglia in penumbra area of Control and astrocyte-specific Kdm4a knockout (Kdm4a KO) mice 24 h after middle cerebral artery occlusion reperfusion (MCAO/R) (n = 4 mice per group). (c) Counts of leukocytes subtypes in spleen in Control and Kdm4a KO mice 24 h after MCAO/R (n = 4 mice per group). (d) Representative immunofluorescence staining for neutrophils marker, lymphocyte antigen 6 complex, locus G (Ly6G, red) from penumbra of Control and Kdm4a KO mice after MCAO/R. 4′,6-diamidino-2-phenylindole was used to stain the nuclei (blue). Scale bar, 20 μm. Quantification of neutrophils in Control and Kdm4a KO mice brain was shown in right panel (n = 4 mice per group). (e, f) Neutrophils in brain (e) and spleen (f) were quantified by using flow cytometry in Control and Kdm4a KO mice 1, 3, 5, and 7 days post MCAO/R (n = 4) and (g–j) Peripheral blood neutrophil was depleted via intraperitoneal injection anti-Ly6G monoclonal antibody in Control and Kdm4a KO mice. Stroke outcome were determined by modified Neurological Severity Score (g), foot fault (h), rotarod test (i) and infarct volumes (j) (n = 4 per group). **P < 0.01. n.s.: no significant; SSC: side scattering; KO: Kdm4a KO; CTR: Control; mNSS: modified Neurological Severity Score.

Astrocyte KDM4A-mediated CXCL1 activation promotes neutrophil infiltration

To gain a comprehensive view of astrocyte derived factors regulated by KDM4A after brain ischemia, we adopted a proteome profiler array analysis that measured inflammatory factors from lysates of purified astrocytes at 24 h after MCAO/R in Control and Kdm4a KO mice (Figure 6(a)). Among these factors, the CXCL1, TNF-α and IL-6 were significantly decreased in Kdm4a KO astrocytes (Figure 6(b)). These cytokines expression was confirmed by enzyme-linked immunosorbent assay (ELISA) (Figure 6(c)). We paid special attention to CXCL1, because it is reported to be involved in neutrophils infiltration after stroke. ²⁸ We then verified CXCL1 expression in primary cultured astrocytes exposed to OGD/R. ELISA experiments revealed a significant increase in CXCL1 levels after OGD/R injury, which was attenuated by Kdm4a knockout (Figure 6(d)). Flow cytometry analysis showed that Kdm4a knockout reduced CXCL1 positive cells upon OGD/R condition (Figure 6(e) and (f)). Collectively, these results suggest that knockout of astrocyte Kdm4a may lead to a reduction of CXCL1 in the ischemic brain.

Figure 6.

Lysine-specific demethylase 4 A (Kdm4A) knockout in astrocytes reduces transmigration of neutrophils via regulating neutrophil chemoattractant CXC chemokines ligand 1 (CXCL1). (a) Procedures for the isolation and purification of astrocytes from the stroke mouse brain followed by proteome profiler array analysis. Details procedures will be found in materials and methods section. (b) Heat map shows the cytokine/chemokine expression profiles of the lysates from fluorescence-activated cell sorted astrocytes from astrocyte-specific Kdm4a knockout (Kdm4a KO) and Control mice 24 h post middle cerebral artery occlusion reperfusion (MCAO/R). Results were generated according to clustering of proteome profiler array assays of the listed chemokine normalized to individual reference Control. Red shades represent increased expression of chemokine relative to other cell types. (c) Quantification of enzyme linked immunosorbent assay (ELISA) indicates highly up-regulated CXCL1, interleukin 6 (IL-6) and tumor necrosis factor-α (TNF-α) protein expression level in lysates of astrocytes obtained from Kdm4a KO and Control mice brain at 24 h post MCAO/R (n = 6 mice per group). (d) Quantification of ELISA shows the expression of CXCL1 in primary cultured astrocyte from Kdm4a KO and Control mice upon oxygen-glucose deprivation/regeneration (OGD/R) condition (n = 6 per group). (e) Flow cytometry analysis of CXCL1-Cy3 signal in astrocytes isolated from Kdm4a KO and Control mice 24 h after MCAO/R. All gates were set by using Fluorescence Minus One (FMO) controls. Gray: FMO control; White: positive sample. (f) Quantification of flow cytometry analysis (n = 4 per group). Two thousand cells were evaluated with flow cytometry for each experiment. (g) General schematic of the astrocyte-neutrophil transwell system. (h) Astrocytes isolated from Kdm4a KO and Control mice were exposed to OGD/R condition, neutrophil migration was measured by flow cytometry (n = 4 per group) and (i) Antibodies against CXCL1, IL-6, and TNF-α were added to the conditioned medium, and the migration rate of neutrophils was measured in Transwell system (n = 4 per group). **P < 0.01. FACS: fluorescence-activated cell sorted; KO: Kdm4a KO; CTR: Control.

To assess whether astrocyte KDM4A influences neutrophils infiltration via CXCL1. The astrocytes were exposed to OGD/R and then cultured for another 4 h in a Transwell system with purified isolated neutrophils by a Transwell membrane (Figure 6(g)). The proportions of neutrophils that had migrated into the lower Transwell chamber were significantly decreased, followed by depletion of Kdm4a in astrocytes (Figure 6(h)). We then sequentially added antibodies against IL-6, TNF-α and CXCL1 into the astrocyte-conditioned medium to block the action of each cytokine. Only by blocking CXCL1 did the proportions of neutrophils in the Kdm4a KO and Control groups become equal (Figure 6(i)). These data indicate that astrocyte KDM4A-mediated CXCL1 upregulation promotes neutrophils infiltration post-stroke.

NF-κB is required for KDM4A mediated in cxcl1 gene activation in astrocytes upon OGD/R injury

CXCL1 expression is regulated by several transcription factors, including NF-κB, activator protein 1 (AP1), Specificity protein 1 (SP1), Nuclear factor for IL-6 expression (NF-IL6). To further unveil the signaling pathway KDM4A regulated CXCL1, promoter occupancy of KDM4A and these transcription factors was examined by ChIP assay in primary culture of astrocytes. Proteins recruited to the cxcl1 promoter was quantified using oligonucleotide primers encompassing the transcription factors binding site via RT-PCR. The results showed that KDM4A and NF-κB-p65 but not AP1, SP1 and NF-IL6, significantly bound to cxcl1 gene promoters upon OGD/R injury (Figure 7(a)). Thus, we speculate that KDM4A maybe a coactivator of NF-κB regulating cxcl1 gene during OGD/R.

Figure 7.

Lysine-specific demethylase 4 A (KDM4A) and Nuclear factor-kappa B (NF-κB) are recruited to C-X-C motif ligand 1 (Cxcl1) gene promoters upon oxygen-glucose deprivation/regeneration (OGD/R) injury. (a) Chromatin immunoprecipitation (ChIP) assay examined the occupancy of KDM4A, NF-κB, activator protein-1 (AP1), specificity protein 1 (SP1), and nuclear factor of IL-6 (NF-IL6) in Cxcl1 promoter upon normoxia or OGD/R condition (n = 4 per group). (b–c) Lysates of astrocytes precipitated with the antibody against nuclear factor-kappa B (NF-κB) (b) or KDM4A (c), and immunoblotted with indicated antibodies. KDM4A, phospho-NFκB p65 (Ser536) (NF-κB-p65), immunoglobin G (IgG) are presented by the 130 Kilodaltons (KD), 65 KD, and 52 KD band, respectively. (d) A Cxcl1 gene promoter driven-firefly luciferase reporter was transfected in conjunction with a Control Renilla luciferase expression vector into U87 cells, respectively. The NF-κB binding site in Cxcl1 gene promoter was mutated. Reporter activities are expressed as percentage of activation relative to Renilla luciferase activity (n = 4 per group) and (e) ChIP assays were performed in astrocyte-specific Kdm4a knockout (Kdm4a KO) and Control astrocytes upon normoxia and OGD/R condition using the anti- histone H3 lysine 9 trimethylation (H3K9me3) antibodies (n = 4 per group). **P < 0.01. WT: wild-type; WB: Western blot; IP: immunoprecipitation; KO: Kdm4a KO; CTR: Control.

To verify this hypothesis, we investigated the interaction of NF-κB-p65 and KDM4A in the primary culture of astrocytes upon OGD/R condition. The results showed that immunoprecipitation of NF-κB-p65 pulled down endogenous KDM4A (Figure 7(b)), or vice versa (Figure 7(c)). Then the putative NF-κB binding site in the cxcl1 gene was identified using the PROMO TRANFACT program. To determine whether the intact NF-κB binding site is necessary for OGD/R-induced activation of the cxcl1 gene, we conducted promoter luciferase reporter assays. The results showed that promoter activities of cxcl1 genes were elevated by OGD/R injury. However, no significant elevated promoter activity was observed for the NF-κB binding site-mutated cxcl1 gene promoters, indicating that NF-κB binding site is required for the elevation of the promoter activities of cxcl1 gene by OGD/R injury (Figure 7(d)).

KDM4A was reported to demethylate H3K9me3. ¹⁴ The methylation level of H3K9 at the cxcl1 gene promoters was investigated using anti-H3K9me3 antibody. A decreased level of trimethylated H3K9 at cxcl1 promoters was observed in astrocytes when subjected to OGD/R injury. In the Kdm4a KO astrocytes, no significant change was observed in the level of trimethylated H3K9 upon OGD/R injury (Figure 7(e)). Taken together, these results suggest that KDM4A, in cooperation with NF-κB, activates cxcl1 gene expression via the demethylation of H3K9me3 at the cxcl1 promoters in astrocytes during OGD/R injury.

Discussion

Cerebral I/R is one of the leading causes of brain injuries and deaths. Therefore, it is of urgent demand to explore its molecular mechanisms. Research on epigenetic mechanisms in cerebral I/R is still in its early stages, but it holds promise for a better understanding of the pathology. Previous studies have shown that the DNA methylation level elevated after cerebral I/R, and this increase correlates with more severe brain injuries in various animal stroke models. ²⁹

At post-translational level, histone methylation proteins have been researched in animal models of stroke. One study found dysregulation of KDM4 families at various time intervals post internal carotid artery (ICA) occlusion, leading to reduced H3K9me2 levels in the penumbra region 24 h post-occlusion. ³⁰ In this study, we observed that KDM4A was prominently upregulated in astrocytes after stroke, where it played a critical role in neutrophil infiltration. We found that KDM4A activated cxcl1 gene expression in astrocytes in cooperation with NF-κB. These findings highlight previously undiscovered functions of KDM4A in astrocytes and provide new insights into the mechanism of cerebral I/R.

Our results demonstrated that KDM4A was expressed at low levels in normal brain but was elevated post MCAO/R; KDM4A was specifically increased in astrocytes under OGD/R condition. To investigate whether the increased KDM4A in astrocytes plays a role in the pathological process after cerebral I/R, we used conditional knockout mice. To overcome the lack of specific methods to explore the function of astrocytes in pathophysiological processes, a genetic tool that specifically manipulated astrocytes by invoking the expression of Cre recombinase driven by GFAP promoter was used in this study. GFAP Cre mice was used in our study to achieve selective genetic knockout of astrocytes.

Astrocytes constitute over 50% of the brain’s volume and are the most abundant cell types in CNS. Therapeutic strategies targeting astrocytes have been shown to influence stroke outcomes through multiple mechanisms. For instance, previous research showed that the activation of cannabinoid type 1 receptors in astrocyte protects neurons from cerebral I/R injury via upregulating extracellular glutamate. ³¹ Astrocytes also actively participate in the formation and maintenance of the BBB, ³² and interventions targeting astrocytes have been an effective method to alleviate BBB injury caused by stroke. Additionally, targeting astrocytes is also an effective strategy to increase cerebral blood flow, which is important for tolerance in response to ischemia insult. ³³ Astrocytes also engage in cross-talk with CNS-infiltrating leukocyte by releasing proinflammatory cytokines and chemokines. ³⁴ Previous study has shown that astrocytes exert potent proinflammatory functions by secreting inflammatory factors including monocyte chemotactic protein-1, IL-1β, IL-6 as their primary mode of action after stroke. ³⁵ These researches suggest that astrocytes play a crucial role in the migration of leukocytes after brain ischemia. According to the gene array data, the terms affected by astrocyte KDM4A focus on leukocytes infiltration. This indicates that astrocyte KDM4A upregulation is involved in pathological immune responses following stroke injury.

The leukocytes infiltration affects all stages of ischemic stroke. In our research, neutrophils displayed the most significant change in Kdm4a KO mice post-stroke. Neutrophils are pivotal immune components after stroke and play critical role in neuroinflammation. ³⁶ Neutrophils are among the first cell types to contribute to BBB disruption and neuronal apoptosis. Neutrophil infiltration into the ischemic brain begins at about 30 min after a stroke and peaks at 24–48 h. These infiltrating neutrophils aggravate tissue damage by generating cytokines, reactive oxygen species, and metalloproteases. Limiting neutrophil recruitment to the brain is an effective strategy to reduce brain injury. Consequently, Kdm4a deficiency in astrocytes may ameliorate stroke injury by reducing neutrophil infiltration. To investigate it further, we spatially isolated neutrophils and astrocytes from both Control and Kdm4a KO mice that were subjected to OGD/R conditions. After addition of antibody against CXCL1, the migration index of neutrophils was similar in both two groups. Based on these findings, we reasoned that KDM4A regulate the neutrophil infiltration might through CXCL1 in astrocytes.

A previous study has reported that KDM4A activates the Il-8 gene expression via the association of NF-κB with the Il-8 gene promoter. ³⁷ In addition, NF-κB was known to activate Cxcl1 gene expression. ³⁸ Our findings of NF-κB dependent Cxcl1 activation in astrocytes are consistent with previous studies reporting NF-κB pathway activation in astrocytes after cerebral I/R. ³⁹ Furthermore, silencing of IkappaB kinase reduced CXCL1 expression in astrocytes upon OGD/R condition. ⁴⁰ Apart from transcription factors, gene transcription also relies on the state of histone modifications including acetylation, phosphorylation and methylation. ¹⁰ While previous studies found that histone deacetylases play a critical role in the outcome of ischemic stroke by regulating neuroplasticity-related genes via NF-κB,^{41 –43} the role of histone methylation in NF-κB regulated gene transcription is still largely unknown. Increases in H3K9me3 have been reported in various cell types exposed to hypoxia with subsequent regulation of gene expression. ⁴⁴ On this basis, we reasoned that an intact NF-κB binding site is required for the KDM4A induced Cxcl1 expression. In our study, we reported for the first time that KDM4A, in cooperation with NF-κB, activates Cxcl1 gene expression via the demethylation of H3K9me3 at the Cxcl1 gene promoters in astrocytes upon OGD/R injury. The link of KDM4A-NF-κB-Cxcl1 pathway to ischemic induced neutrophil infiltration requires further investigation in future studies. Interestingly, NF-κB also activates histone lysine demethylase gene expression, ⁴⁵ indicating that NF-κB acts as both a transcriptional activator of KDM4A and a KDM4A-interacting transcription factor on downstream target genes.

Previous studies have shown that inhibiting NF-κB in astrocytes reduces the severity of ischemic damage following stroke. ⁴⁶ Our study revealed that KDM4A regulates CXCL1 via NF-κB. NF-κB also plays a role in the regulation of multiple proinflammatory mediators that are upregulated following stoke, contributing to post stroke inflammation.^47,48 Thus, further research on the effect of KDM4A on other proinflammatory mediators may help flesh out the mechanism underlying its regulation of neutrophil infiltration following stroke. In addition, NF-κB is involved in events that impact neuron survival, such as excitotoxic and oxidative events. Thus, we cannot rule out the possibility that KDM4A may influence other pathological events besides inflammation. Further research is needed to elucidate the role of KDM4A/NF-κB in the process of cerebral I/R.

There are a few limitations to the current work that should be considered when interpreting the results. Firstly, the use of the constitutive GFAP Cre mouse line to generate the Kdm4a conditional knockout presents potential pitfalls. In embryonic brain, GFAP is also expressed by non-astrocytes, such as some progenitor cell during early development. ⁴⁹ The GFAP-positive progenitor cells then produce neurons and oligodendrocytes throughout the CNS. ⁵⁰ Thus, we cannot exclude the possibility that some proportion of neurons and oligodendrocytes also experienced Kdm4a gene deletions, this impact needs to be carefully considered. Moreover, we cannot exclude developmental effects of the constitutive astrocytic Kdm4a knockout, which could mediate compensatory outcomes in Kdm4a KO mice. Thus, whether acute pharmacological inhibition of KDM4A in astrocytes or inducible astrocytic Kdm4a knockout has a similar effect on stroke outcome need to be further investigated.

Another limitation is that only adult male mice were used in our study. Sexual dimorphism was well-documented in the pathophysiology, incidence and severity of stroke. ⁵¹ It is important to recognize the role of sex differences in cerebrovascular reactivity and brain micro-vessel functions, which can influence the susceptibility to stroke.^52,53 Previous research found that specific gene knockout of focal adhesion kinase gene in astrocytes reduced infarct volume and IL-6 expression only in female mice. ⁵⁴ Thus, we cannot exclude the possibility that sex bias might have affected the outcomes in our Kdm4a KO mice. Investigating the sex differences in the stroke outcome caused by astrocytic Kdm4a knockout would require further research, though it is beyond the scope of this paper.

Moreover, our evaluation of stroke outcome is limited to the acute phase. However, cerebral I/R injury triggers a sequence of pathological cascades which may persist for days. Astrocytes are important contributors to responses that promote long term recovery through changes leading to the generation of the glial scar as well as changes that can promote facilitate recovery of neuronal function and increase neuronal plasticity. ⁵⁵ In our study, in astrocytes acutely isolated from the penumbra region, Kdm4a deficiency caused a large change in the expression of genes that contribute to many cellular functions. For a certain proportion of genes, the expression may persistently increase or decrease over an extended. Thus, the long-term effect of astrocytic Kdm4a Knockout on stroke outcome and its underlying mechanism warrants further research.

In conclusion, our in vivo and in vitro investigations have demonstrated that astrocyte KDM4A upregulation contributes to neutrophil infiltration following ischemic stroke partly through activation of Cxcl1 expression via NF-κB. As such, astrocyte KDM4A may be a potential therapeutic target for regulating neutrophil infiltration following stroke and should be further explored.