Integrated transcriptomic and proteomic profiling reveals the key molecular signatures of brain endothelial reperfusion injury

Abstract Background Reperfusion therapy after ischemic stroke often causes brain microvascular injury. However, the underlying mechanisms are unclear. Methods Transcriptomic and proteomic analyses were performed on human cerebral microvascular endothelial cells following oxygen–glucose deprivation (OGD) or OGD plus recovery (OGD/R) to identify molecules and signaling pathways dysregulated by reperfusion. Major findings were further validated in a mouse model of cerebral ischemia and reperfusion. Results Transcriptomic analysis identified 390 differentially expressed genes (DEGs) between the OGD/R and OGD group. Pathway analysis indicated that these genes were mostly associated with inflammation, including the TNF signaling pathway, TGF‐β signaling pathway, cytokine–cytokine receptor interaction, NOD‐like receptor signaling pathway, and NF‐κB signaling pathway. Proteomic analysis identified 201 differentially expressed proteins (DEPs), which were primarily associated with extracellular matrix destruction and remodeling, impairment of endothelial transport function, and inflammatory responses. Six genes (DUSP1, JUNB, NFKBIA, NR4A1, SERPINE1, and THBS1) were upregulated by OGD/R at both the mRNA and protein levels. In mice with cerebral ischemia and reperfusion, brain TNF signaling pathway was activated by reperfusion, and inhibiting TNF‐α with adalimumab significantly attenuated reperfusion‐induced brain endothelial inflammation. In addition, the protein level of THBS1 was substantially upregulated upon reperfusion in brain endothelial cells and the peri‐endothelial area in mice receiving cerebral ischemia. Conclusion Our study reveals the key molecular signatures of brain endothelial reperfusion injury and provides potential therapeutic targets for the treatment of brain microvascular injury after reperfusion therapy in ischemic stroke.

Ischemic stroke (IS), one of the major diseases that endangers the health and life of elderly individuals, is characterized by high incidence, high mortality, and high rate of disability.The principal treatment of acute IS is to restore the blood supply to the ischemic brain tissue as soon as possible and reduce damage to the penumbra brain tissue.Thrombolysis and intravascular therapy are the main clinical treatment methods for acute IS. 1,2 However, after a period of ischemia, blood reperfusion may also cause damage to the brain, that is, reperfusion injury, 3 although the mechanisms of reperfusion injury are not well understood.[6][7] These complications not only limit the time window for reperfusion therapy, but also affects the treatment efficacy.
The BBB function is determined by the neurovascular unit that is composed of cerebral vascular endothelial cells, pericytes, astrocyte endfeet, and the surrounding basement membrane.BBB is essential for maintaining the homeostasis of brain tissue microenvironment, regulating brain metabolism, and facilitating the normal neuronal function. 8Endothelial cells (ECs) as the main component of the brain vasculature and BBB are the primary target of reperfusion injury.
Therefore, a systematic study of the dysregulated molecules and signaling pathways in brain ECs after cerebral ischemia and reperfusion is fundamental to elucidating the mechanisms of reperfusion injury.
In the current study, we performed transcriptomic analysis by RNA-sequencing (RNA-seq) and proteomic analysis by liquid chromatography-tandem mass spectrometry (LC-MS/MS) with human cerebral microvascular endothelial cells (HCMECs) following oxygenglucose deprivation and recovery (OGD/R).The major findings of the in vitro studies were further validated in a mouse model of cerebral ischemia and reperfusion in vivo.Our data provide a comprehensive molecular atlas of brain EC injury induced by ischemia/reperfusion, which will facilitate the understanding of reperfusion injury and exploring of novel therapeutic approaches by targeting ECs after acute IS.

| Cell line and in vitro OGD/R model
We purchased an immortalized HCMEC line (HCMEC/D3) from Institut Cochin (Paris, France) and cultured it in Dulbecco's modified Eagle's medium (DMEM) containing 10% fetal bovine serum and 1% penicillin.All cultures were maintained in a humidified 5% CO 2 incubator at 37°C and routinely passaged three times after reaching 80%-90% confluency.The cells were divided into a normal control (NC) group, an OGD-6 h group, and different oxygen and glucose recovery (OGD/R) time groups (observation time points: 2.5, 3, 4, 5, 7, 14, and 26 h).Endothelial cell activity decreased by >30% in the OGD-6 h group, which was in line with the requirements of follow-up studies; setting the peak point of cell damage after OGD/R for follow-up studies was conducive to comparing the differences between reperfusion injury and ischemia-hypoxia injury at the molecular level.Three biological replicates were performed for each group of cells, and each biological replicate was performed in three wells.For OGD modeling, the cells were washed three times with phosphate-buffered saline (PBS) to ensure that no residual sugar remained.The cells were then resuspended in DMEM (Solarbio) and placed in an OGD incubator (99% N 2 ) for 6 h.For OGD/R modeling, the cells were removed from the OGD incubator and placed in a conventional oxygen concentration incubator.The sugar/serum-free culture medium was replaced with complete DMEM, and the samples were taken at the predetermined time point for subsequent experiments.Cell Counting Kit-8 (CCK-8; TransGen Biotech) was used to measure cell viabilities, according to the manufacturer's instructions.Briefly, cells were seeded in 96-well plates at a density of 5000 cells/well.Ten microliters of CCK-8 solution were added to each well containing 100 μL of medium.The cells were cultured at 37°C for 2 h.Optical density values were measured at 450 nm using Multiskan GO spectrophotometer (Thermo Scientific).

| RNA extraction and 3′-end sequencing for expression quantification (3′-seq)
Total RNA was extracted using the Direct-zol RNA MiniPrep Kit (Zymo Research).To quantify mRNA expression levels, 3′-seq was performed as previously described. 9Briefly, eukaryotic mRNA was first enriched by Oligo (dT) beads, while prokaryotic mRNA was enriched by removing rRNA using the Ribo-Zero™ Magnetic Kit (Epicenter).The mRNA was then fragmented into short fragments and reverse transcribed into complementary DNA (cDNA) with random primers.Second-strand cDNA was synthesized using DNA polymerase I, RNase H, and dNTPs.Then, the cDNA fragments were purified with a QiaQuick PCR Extraction Kit (Qiagen), endrepaired, modified by poly (A) addition, and ligated to sequencing adapters (Illumina).The ligation products were size-selected

Conclusion:
Our study reveals the key molecular signatures of brain endothelial reperfusion injury and provides potential therapeutic targets for the treatment of brain microvascular injury after reperfusion therapy in ischemic stroke.

K E Y W O R D S
blood-brain barrier, inflammation, ischemic stroke, multi-omics, reperfusion injury by agarose gel electrophoresis, amplified by the polymerase chain reaction (PCR), and sequenced using a Hiseq PE150 System (Illumina).

| RNA-seq data analysis
The raw RNA-seq data were subjected to quality-control analysis (to filter the data and reduce noise) before conducting bioinformatics analysis.Analysis showed that high-quality reads accounted for >98.95% of all reads in each sample (Table S1).Then, short read alignment tool Bowtie2 (version 2.2.9) was used for mapping reads to the ribosome RNA (rRNA) database.Less than 4.20% of reads in each sample (<1.29% in the treatment groups) matched with the rRNA database (Table S2).The rRNA mapped reads were then removed.The remaining reads were mapped to the Ensembl human genome (GRCh38.p13)using TopHat2 (version 2.1.1)and then assembled using Cufflinks (version 2.2.1).At least 85.15% of the sequences were well aligned in each sample (Table S3).Then, the gene expression levels were normalized using the fragments per kilobase of transcripts per million mapped reads (FPKM) method and the edgeR package (version 3.14.0)was utilized to identify the DEGs across groups.DEGs were defined as those showing a foldchange (FC) in expression of >2.0 and an adjusted p value for the false-discovery rate (FDR) of <0.05.The Database for Annotation, Visualization, and Integrated Discovery (DAVID) was used to analyze gene abundances.DAVID is a bioinformatics resource that is used to explain the functions of numerous genes.A list of DEGs from the above group comparisons was uploaded to DAVID (version 6.7) to identify enriched biological functions, such as Gene Ontology (GO) terms and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways.The Gene Set Enrichment Analysis (GSEA) was performed using the gseKEGG function from the R package clusterProfiler (v4.6.0).The enrichment scores were normalized by gene set size, and their statistical significance was assessed by permutation tests (n = 1000).The adjust p value <0.05 were used as the cut-off for significance in all GSEA analysis.

| Sample preparation
Fifty micrograms of protein were digested in 8 M urea buffer (8 M urea (U5378, Sigma-Aldrich), 100 mM Tris 8.5 (15,504,020, Thermo Scientific), 10 mM dithiothreitol (D9779, Sigma-Aldrich)) using the filter-aided sample preparation (FASP) protocol.Briefly, proteins were added to a 30 kDa cut-off filter (MRCF0R030, Millipore) and centrifuged at 11,000 rotations/min at 20°C for 15 min.Iodoacetamide (I1149, Sigma-Aldrich) (50 mM) in urea buffer was used to alkylate the proteins at 20°C for 15 min.After a few washes with urea buffer and 50 mM ammonium bicarbonate (ABC; 09830, Sigma-Aldrich) buffer, 100 ng of trypsin (V5280, Promega) in 50 mM ABC buffer was used to digest proteins in a wet chamber overnight at 37°C.Peptides were extracted using 50 mM ABC buffer and acidified with trifluoroacetic acid (TFA; 1,082,620,100, Millipore).After performing the FASP protocol described above, the peptides were separated into 5 fractions (flow through, pH 11, pH 8, pH 5, pH 2).Each sample was measured by LC-MS/MS using a 4 h gradient.

| Proteomic analysis
Tryptic peptides were separated over 240 min using an Easy-nLC 1200 ultra-high-performance LC system, using the following gradient: 2%-7% buffer B (80% acetonitrile (1,000,041,001, Millipore) in 0.1% TFA) over 11 min, 7%-28% buffer B over 200 min, and 28%-36% buffer B over 15 min.Subsequently, the peptides were washed for 5 min while increasing buffer B from 36% to 60% and for 9 min while increasing buffer to 95%.The Easy-nLC 1200 system was connected online to a Fusion Lumos mass spectrometer equipped with a FAMIS Pro interface (Thermo Scientific).Scans were collected in a data-dependent top-speed mode with a dynamic exclusion of 90 s.
Raw data were analyzed using MaxQuant software (version 1.6.0.1) by searching against the Human Fasta Database, with label-free quantification and matches between run functions enabled.The output protein list was analyzed and visualized using the DEP package.

| Middle cerebral artery occlusion/reperfusion (MCAO/R) model and TNFα neutralizing monoclonal antibody (mAb) treatment
Male 8-10-week-old C57BL/6 mice, weighing 20-23 g, were obtained from Beijing Vital River Laboratory Animal Technologies Co., Ltd.Because menstrual period in female animals causes significant changes in estrogen levels and variations in infarct size and neurological outcome, only male mice were used in this study.Animals were housed in a pathogen-free animal facility with 12-h light-dark cycles.All procedures performed on mice were approved and carried out in accordance with the Animal Care and Use Committee of the Shenzhen Institute of Advanced Technology, Chinese Academy of Sciences.Randomized animals were used for the sham-operation and IS models.For the surgical procedures, anesthesia was induced with 4% isoflurane within air in an induction chamber; anesthesia was maintained with 2% isoflurane within air delivered through a face mask (RWD Life Science).A heating pad was used to maintain each mouse's core temperature at 37 ± 0.5°C throughout the surgical procedure.A modified intraluminal-filament model was used to induce middle cerebral artery occlusion (MCAO).
After 60 min of MCAO, reperfusion was established by retracting the filament.The animals had free access to food and water throughout the reperfusion period.To verify the success of ischemia and reperfusion during MCAO surgery, a laser Doppler flowmetry (PERIMED AB) was used to measure the cortical blood flow supplied by the MCA.Mice in which the cerebral blood flow decreased to levels at least 30% below the baseline value after filament insertion were included in the study.
The success of reperfusion was verified by the increase of cerebral blood flow to over 70% of the baseline value after filament retraction.
The animals were divided into two different reperfusion groups (6 and 24 h reperfusion periods; samples were taken at a predetermined time point), a control group (samples taken 1 h after inducing MCAO), and a sham-operation group (n = 3-4 mice per group).TNFα mAb Adalimumab (Abbott, US; 3 mg/kg) or control human IgG (#SP001, Solarbio) was administered by retro-orbital i.v.injection immediately after MCAO at the onset of reperfusion.Neurological deficit scores were evaluated after MCAO by an investigator who was blind to the grouping according to a 5-point Bederson scale: 0, normal motor function; 1, flexion of the torso and the contralateral forelimb upon lifting by the tail; 2, circling to the ipsilateral side but normal posture at rest; 3, leaning to the ipsilateral side at rest; 4, unable to walk spontaneously.
Animals were excluded if no neurological deficits were present after surgery, according to the pre-established exclusion plan.

| Infarct size analysis
The brains were sectioned into 2-mm-thick sections and placed in 1% 2,3,5-triphenyltetra-zolium chloride (TTC; #T8877, Sigma-Aldrich) for 10 min at 37°C.The TTC solution was then replaced with 4% paraformaldehyde, and the sections were incubated at room temperature for 1 h.The sections were photographed with a digital camera, and the infarct sizes were measured using ImageJ software (National Institutes of Health) by an investigator who was blind to the grouping.The contribution of post-ischemic edema to the volumes of sites of injury was corrected as described previously. 10The infarct size (%) was calculated as ([volume of the left hemisphere -non-infarct volume of the right hemisphere] / volume of the left hemisphere) × 100%.

| Reverse transcriptase-quantitative polymerase chain reaction (RT-qPCR)
Total RNA was extracted using the Direct-zol RNA MiniPrep Kit (Zymo Research).The RNA was reverse transcribed using iScript Reverse Transcription Supermix for RT-qPCR according to the manufacturer's instructions (Vazyme).The primer sequences are shown in Table S4.RT-qPCR was performed in a StepOnePlus Real-Time PCR System using Power SYBR Green (Roche).Relative RNA-expression levels were calculated using the comparative Ct method and normalized to actin mRNA expression.Data are expressed relative to a calibrator using the 2 −ΔΔCt method.

| Immunofluorescence staining
Frozen sections (10 μm thick) were allowed to dry on adhesion microscope slides (Citotest Scientific Co., Ltd.) at room temperature and then rehydrated in PBS.The sections were blocked in 10% normal goat serum (Thermo Fisher) in PBS + 0.2% Triton X-100 for 1 h at room temperature.The samples were incubated at 4°C with the following primary antibodies in PBS + 5% goat serum +0.2%

| Statistical analysis
Statistical analysis was performed in Graphpad Prism software (ver.7).Data are expressed as the mean ± standard error of mean (SEM), and the number (n) of the samples employed is indicated in the figure legends.The Shapiro-Wilk test was used to assess the normality of the data distribution.Statistical differences among multiple groups were compared using one-way analysis of variance (ANOVA), followed by Tukey's multiple-comparisons test, while two groups were analyzed by 2-tailed Unpaired Student's t-test.Statistical significance was defined as p < 0.05.The RNA-seq data were assessed by a bioinformatics team at Sagene Biotech Co., Ltd., and proteomic analysis was performed by a bioinformatics team at Lum-ingBio Company.

| Transcriptomic analysis of HCMECs subjected to OGD/R
The HCMEC/D3 cells were divided into NC, OGD-6 h, and OGD/R groups and subjected to corresponding treatments.We observed that cells showed the most critical injury at 3 h reperfusion following 6 h of OGD, as determined by assessing cell viabilities in CCK-8 tests (Figure S1).Therefore, we chose the 3 h-OGD/R as a representative time point for reperfusion injury in vitro.
We used FDR of <0.05 and FC of >2.0 as thresholds for assigning significant differences.The results showed that ECs exhibited different transcriptomic characteristics during OGD and OGD/R.
Compared to the NC group, the OGD group had 2478 DEGs with 761 upregulated genes and 1717 downregulated genes.Compared to the OGD, the OGD/R group had 390 DEGs with 301 upregulated and 89 downregulated genes (Figure 1A,B; Data S1).
DEGs-based KEGG analysis showed that 277, 177, and 268 signaling pathways were associated with DEGs when comparing the OGD and NC groups, the OGD/R and OGD groups, and the OGD/R and NC groups, respectively; among them, 13, 23, and 12 signaling pathways were significantly enriched (Q value ≤0.05;Data S2).
The top five pathway terms associated with DEGs between the OGD/R and OGD groups were the TNF signaling pathway, TGFβ signaling pathway, cytokine-cytokine receptor interaction, NODlike receptor signaling pathway, and Salmonella infection (Table 1).
This suggests that dysregulation of these signaling pathways (especially upregulation of the TNF signaling pathway, evidenced by upregulation of TNF, IL-6, LIF, CXCL1, CXCL2, CXCL3, CCL20, NFK-BIA, TNFAIP3, SOCS3, MAP2K3, and other genes; Table 1; Table S5) may play important roles in endothelial reperfusion injury.Furthermore, significantly enriched signaling pathways between the OGD/R and OGD groups were mostly associated with the inflammatory response, including the TNF signaling pathway, TGFβ signaling pathway, cytokine-cytokine receptor interaction, NOD-like receptor signaling pathway, JAK-STAT signaling pathway, MAPK signaling pathway, and NF-κB signaling pathway (Table 1).We further performed whole transcriptome rank-based GSEA between the OGD/R group and the OGD group.We observed enrichment in 17 distinct KEGG signaling pathways in OGD/R group compared with OGD group (p.adjust <0.05; Figure S2 and Data S3).Consistent with the DEGs-based KEGG analysis, we observed significant enrichment in inflammatory responses, such as IL-17 signaling pathway, C-type lectin receptor signaling pathway, cytokine-cytokine receptor interaction, TNF signaling pathway, JAK-STAT signaling pathway, TGFβ signaling pathway, and chemokine signaling pathway (Figure S2).These results collectively suggest that endothelial cell-mediated inflammatory responses are the most prominent mechanism underlying reperfusion injury at the mRNA level.
GO analysis showed that for the DEGs between the OGD/R group and the OGD group, the top three biological processes were cellular processes, single-organism processes, and response to stimuli; the top three terms related to cellular components were cell, cell part, and organelle compartments, and the top three terms related to molecular functions were binding, catalytic activity, and nucleic acid-binding transcription factor activity (Figure S3).

| Integrative analysis of the transcriptomics and proteomics data
To investigate correlations between mRNA expression and protein expression levels, we performed a combined transcriptomic and proteomic analysis of data from the OGD/R and OGD groups (Figure 1E).
We generated a scatter plot of the mRNA expression levels of known genes (X-axis) versus the corresponding protein expression levels (Yaxis) to compare the protein versus mRNA abundances.The correlation value (r = 0.492), calculated using Spearman's rank correlation coefficient test, suggested little correlation between mRNA expression and protein expression (Figure S5).Integrative analysis showed that six genes (DUSP1, JUNB, NFKBIA, NR4A1, SERPINE1, and THBS1) were differentially expressed at both the mRNA and protein levels, and all of them were upregulated (Figure 1E,F; Data S7).

| TNF signaling pathway was upregulated in the brain of mice receiving MCAO/R
In mice subjected to 1 h-MCAO without reperfusion, infarction had not developed based on TTC staining (Figure 2A,B).Extensive infarction was detected in the cortical and striatum regions in brain tissues at both 6 h (24.0%± 7.9%, p = 0.0115) and 24 h (45.5% ± 4.6%, p < 0.0001) after reperfusion, especially at 24 h post-reperfusion (Figure 2A,B).As we showed previously, the growth of infarct size at 24 h post-reperfusion also leads to increased neurological deficits. 11We next assessed brain microvascular injury by detecting mouse blood IgG extravasation from vessels in the ischemic and infarct areas.The results showed that blood IgG leakage increased significantly in mice after 6 and 24 h of reperfusion (p < 0.0001; Figure 2C-F).These results demonstrate that reperfusion following cerebral ischemia led to severe brain infarction and microvascular injury shown by BBB disruption.
Seventeen DEGs associated with the TNF signaling pathway found by RNA-seq were selected and validated in the mouse MCAO/R model using RT-qPCR and immunofluorescence (IF) analyses (Table 1).We selected genes in this pathway because it was the most significantly enriched pathway.We firstly quantified mRNA expression changes for the 17 DEGs of the TNF signaling pathway by RT-qPCR.Consistent with the RNA-seq data, RT-qPCR data showed that the mRNA levels of most genes (TNF, IL-6, SOCS3, CXCL1, CXCL2, CXCL3, MAP2K3, NFKBIA, LIF, JUN, EDN1, and CSF2) also showed upregulations (Figure 3; Figure S6).As further validation, the protein levels of TNFα, IL-6, and CXCL3 were assessed by IF staining analysis.The results showed that these proteins were significantly upregulated after 24 h reperfusion following 1 h-ischemia (Figure S7), which were consistent with the RT-qPCR results.

| Inhibiting TNFα significantly attenuated brain endothelial inflammation in mice following cerebral ischemia and reperfusion
In view of our omics results, we found that the inflammatory response was one of the main features of endothelial reperfusion injury following ischemia, and TNF pathway was one of the most enriched inflammatory signaling pathways, so we determined the effect of intravenous injection of adalimumab, a TNFα neutralizing antibody used in clinic, on mice with cerebral ischemia and reperfusion.We first assessed the TNFα downstream inflammation markers NF-κB p65 and VCAM-1.We found that the expression of NF-κB p65 and VCAM-1 in ECs increased gradually with the TA B L E 1 Top ten pathways associated with DEGs in HCMEC/D3 cells in the OGD/R group, as compared with those in the OGD group.prolongation of reperfusion time (data not shown).Subsequently, we found that adalimumab significantly reduced the infarct volume to 28.0% ± 3.5%, compared with the control group (43.0%± 3.2%; t = 3.196, p = 0.0127; Figure 4A,B).Further studies showed that adalimumab significantly reduced the expression of NF-κB p65 (t = 2.750, p = 0.0205; Figure 4C,D) and VCAM-1 (t = 2.390, p = 0.0341; Fig- ure 4E,F) on cerebral ECs in mice with MCAO-1 h and reperfusion-24 h.Our results suggested that inhibition of TNFα is a promising therapeutic approach for reducing brain endothelial inflammation and thus microvascular reperfusion injury after acute IS.

| THBS1 but not PIGT was upregulated in the brain ECs of mice following cerebral ischemia and reperfusion
Next, we validated the proteomics results in the mouse model of cerebral ischemia and reperfusion.We selected two genes THBS1 (encoding Thrombospondin-1 or THBS1) and PIGT (encoding GPI transamidase component PIG-T or PIGT) for further validation, because THBS1 was upregulated by OGD/R both at the mRNA and protein levels in vitro, and PIGT represents the most significantly enriched pathway in the proteomics analysis (Table 2).The protein levels of THBS1 and PIGT in mouse brain were assessed by both IF staining and Western blotting analysis.
The results showed that THBS1 was absent in sham mice or mice receiving only 1 h-MCAO but was significantly upregulated after 6 and 24 h reperfusion following 1 h-MCAO (Figure 5A,B), in line with the proteomics data.Consistent with THBS1 being a secreted adhesive glycoprotein that resides in ECM and cell surface, 12 abundant THBS1 was found in both brain ECs and the surrounding area in mice receiving MCAO/R (Figure 5A).We also measured the protein level of PIGT in mouse brain tissue and found that similar to THBS1, PIGT was not detected in sham mice or mice receiving only 1 h-MCAO and was significantly upregulated after 6 and 24 h reperfusion following 1 h-MCAO (Figure 5A,B).However, IF staining showed that the PIGT signal was mainly located in neurons and diffusely present in brain parenchyma, without clear co-localization with brain ECs (Figure 5A).These results demonstrate that THBS1, but not PIGT, was potentially involved in reperfusion-related brain microvascular injury in vivo.

| DISCUSS ION
The mechanisms underlying reperfusion injury are complex and involve many pathological factors, such as inflammatory responses, mitochondrial dysfunction, free radical production, and Ca 2+ overload. 13,14Previous studies have shown that anti-inflammatory molecules, 15 antioxidants, 16,17 and calcium channel blockers can reduce cerebral reperfusion injury.Targeting mitochondria may also be helpful for preventing and treating reperfusion injury. 14  Previous results showed that an excessive inflammatory response contributes to the pathogenesis of cerebral ischemia reperfusion injury 15,19 and that the inflammatory response is mainly mediated by microglia, astrocytes, and peripheral leukocytes. 20Our results suggest that ECs may also play an important role in mediating inflammation, especially through the TNF signaling pathway.
RT-qPCR was performed using a mouse model of MCAO/R to quantify changes in 17 DEGs associated with the TNF signaling pathway at the mRNA level.By comparing the RT-qPCR and RNAseq data, we found that most genes showed consistent expression changes.However, it must be mentioned that there are limitations in translating results from a human endothelial cell line to a mouse model as we have done in this study.Results may vary when comparing cells in vivo and in vitro because of the differences in environment and mechanisms.Protein expression levels of the TNFα, IL-6, and CXCL3 genes were further verified by IF staining analysis.
Results showed that these proteins were significantly upregulated after ischemia for 1 h and remained at high levels after reperfusion, which was consistent with the RT-qPCR results, suggesting that these genes are important targets for preventing and treating reperfusion injury.Among them, IL-6 is a proinflammatory cytokine and CXCL3 is a chemokine, but their specific roles in cerebral reperfusion injury have not been clarified.Nevertheless, our findings indicate that these genes and their associated pathways may be important for endothelial cell reperfusion injury.
Studies have indicated that dysfunctional TNF pathway signaling plays important roles in the occurrence and development of many diseases, such as inflammatory and autoimmune diseases, cancer, and cardiovascular disease. 21,22TNFα is the most important molecule in the TNF signaling pathway and is considered as a major proinflammatory mediator.Interestingly, in pathological and physiological situations, TNFα serves dual functions. 22TNFα has been reported to be both neurotoxic and neuroprotective in nervous system diseases.Previous data generated using a rat model of MCAO/R showed that TNFα mRNA expression was increased in brain tissue, 23 and in vitro OGR can enhance TNFα secretion from microglia. 246][27][28] However, inhibiting other proinflammatory genes, such as NF-κB, MAPK, and IL-1β, can also reduce cerebral ischemia reperfusion injury in animals. 15Therefore, the critical inducing factors of reperfusion-caused endothelial inflammation remain to be further identified.Here, we firstly showed that inhibition of TNFα with adalimumab can reduce the cerebral infarct volume.

3. 2 |
Proteomic analysis of HCMECs subjected to OGD/R By using a threshold of a 1.5-FC and a p value of <0.05, 201 differentially expressed proteins (DEPs) between the OGD/R and OGD group were identified, including 96 upregulated and 105 downregulated proteins (Figure 1C,D; Data S4).DEPs-based KEGG enrichment analysis showed that the top five significantly enriched pathways were glycosaminoglycan biosynthesis-chondroitin sulfate/dermatan sulfate, extracellular matrix (ECM)-receptor interaction, ATP-binding cassette (ABC) transporters, Staphylococcus aureus infection, and vitamin digestion and absorption (Table 2; Data S5), suggesting their potential roles in reperfusion injury.Further analysis of the top ten enriched pathways between OGD/R and OGD demonstrated that pathways related to ECM destruction (evidenced by downregulation of COL4A1, LAMA4, ITGA1, ITGA7, ITGB3, and HSPG2) and remodeling (evidenced by upregulation of PIGT and THBS1) were the most significantly enriched pathways.Therefore, our results show that destruction and remodeling of the ECM is an important feature of endothelial reperfusion injury at the protein level.In addition, reperfusion injury also resulted in impaired endothelial transport functions, shown by downregulation of ABC transporters (ABCD1, ABCA3, ABCB7) and multivitamin transporter SLC5A6.Consistent to the results of RNAseq, proteomic analysis also revealed an inflammatory response in brain ECs after reperfusion injury, demonstrated by upregulation of proteins involved in Staphylococcus aureus infection and cytosolic DNA-sensing pathway.Whole transcriptome rank-based GSEA analysis also reveals a strong enrichment in pathways related to ECM destruction (ECM-receptor interaction pathway) and inflammatory response (cell adhesion molecules pathway), shown in Figure S2 and Data S6.We annotated 151 DEPs to 1010 GO function entries.The most enriched biological processes were cellular processes, biological regulation, metabolic processes, and regulation of biological processes; the most enriched cellular component GO terms were cell, cell part, and organelle compartments; the most enriched molecular function GO terms were binding, catalytic activity, enzyme regulator activity, and transporter activity (Figure S4).

F I G U R E 1
Integrative transcriptomic and proteomic analysis of HCMEC/D3 cells subjected to OGD/R.(A) Volcano plots of DEGs for the transcriptomic analysis.Red dots represent upregulated genes, and green dots represent downregulated genes.(B) Bar graph demonstrating the number of upregulated and downregulated genes.DEGs were defined as those with an FDR of <0.05 and an absolute FC of >2.0.(C) Volcano plots of DEPs for the proteomic analysis.Red dots represent upregulated proteins, and green dots represent downregulated proteins.(D) Bar graph demonstrating the number of upregulated and downregulated proteins.DEPs were defined as those with a p value of <0.05 and an absolute FC of ≥1.5.(E) Quadrant map of DEGs and DEPs.The red dots represent co-upregulated mRNAs and proteins, and the green dots represent co-downregulated mRNAs and proteins.(F) An alphabetical list of co-upregulated genes at both the mRNA and protein level.DEG, differentially expressed gene; DEP, differentially expressed protein; FDR, false-discovery rate; HCMEC/D3, a type of human cerebral microvascular endothelial cells; NC, normal control; OGD/R, oxygen-glucose deprivation and recovery.

F I G U R E 2
injury, clinical studies have failed to demonstrate positive outcomes, and only a few drugs, such as edaravone and butylphthalide,17,18 are currently in clinical use.The low success rate of drug development for acute IS may reflect the facts that the mechanisms of reperfusion injury are still not fully clear.As mentioned above, damage to ECs is the main pathological basis of reperfusion injury.Thus, gaining insight into the molecular signatures of brain ECs during reperfusion injury will aid in developing multi-pathway and multi-target intervention strategies against reperfusion injury.In this study, we analyzed the transcriptomic and proteomic changes in HCMECs during reperfusion injury.Pathway analysis of the transcriptomics data showed that the TNF signaling pathway was the most significantly enriched pathway after OGD/R versus OGD, suggesting that upregulation of the pathway is an important Mouse model of MCAO/R and assessment of brain microvascular injury.(A) TTC staining of mouse brain tissues after 0, 6, or 24 h of reperfusion following 1 h of MCAO induction (n = 6 mice per group).Infarcted brain tissues were visualized by TTC staining.(B) Quantification of infarct sizes, as determined by TTC staining.(C,D) Immunofluorescence staining and quantification of endogenous plasma IgG (green) extravasation from brain vessels (red) of mice after 0, 6, or 24 h of reperfusion following 1 h of MCAO induction or a sham operation (n = 6 mice per group).Scale bar, 100 μm.(E,F) Western blotting and quantification of endogenous plasma IgG extravasation in brain parenchyma of mice after 0, 6, or 24 h of reperfusion following 1 h of MCAO induction or a sham operation (n = 6 mice per group).*p < 0.05, ***p < 0.001, ****p < 0.0001 by one-way ANOVA followed by Tukey's multiple-comparisons test.LC, light chain; MCAO/R, middle cerebral artery occlusion/reperfusion; TTC, 2,3,5-triphenyltetrazolium chloride.cause of endothelial reperfusion injury.In addition, the top significantly changed signaling pathways were mostly associated with the endothelial cell-mediated inflammatory responses, including TNF signaling pathway, TGFβ signaling pathway, cytokine-cytokine receptor interaction, NOD-like receptor signaling pathway, JAK-STAT signaling pathway, MAPK signaling pathway, and NF-κB signaling pathway.

Furthermore, we showed
that inhibition of TNFα significantly reduces the inflammatory responses in cerebral ECs, evidenced by attenuated activation of NF-κB p65 and expression of VCAM-1, highlighting a critical role of TNFα in reperfusion-induced endothelial inflammation.The results of our study suggest that targeting inflammatory signaling pathway members, especially those in the TNF signaling pathway, may be an important way to alleviate endothelial reperfusion injury and BBB damage.

F I G U R E 3
Consistent upregulation of the TNF signaling pathway in both HCMEC/D3 cells and a mouse model of MCAO/R.The left part of the histograms shows the RNA-seq data of HCMEC/D3 cells (#p < 0.05, n = 3 biological replicates per group), and the right part of the histograms shows the relative mRNA levels in MCAO/R mouse brain tissues determined by RT-qPCR (*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001 by one-way ANOVA followed by Tukey's multiple-comparisons test, n = 3 mice per group).HCMEC/D3, a type of human cerebral microvascular endothelial cells; MCAO/R, middle cerebral artery occlusion/reperfusion; TNF, tumor necrosis factor.Proteomic analysis showed that proteins associated with ECM-receptor interaction, ABC transporters, and the PI3K-AKT signaling pathway were significantly enriched, suggesting that the associated proteins may also play important roles in reperfusion injury.Our results show that destruction and remodeling of endothelial ECM is an important feature of endothelial reperfusion injury, such as downregulation of the basement membrane proteins collagen IV, integrin, and laminin, and upregulation of PIGT F I G U R E 4 TNF mAb (adalimumab) attenuated endothelial inflammation in mice following cerebral ischemia and reperfusion.(A) TTC staining of coronal brain tissue slices after 1 h-MCAO and 24 h reperfusion.Infarcted areas were visualized as white in TTC staining.(B) Quantifications of infarct volume as determined by TTC staining in mice treated with control human IgG, or ADM (3 mg/kg) after 1 h-MCAO and 24 h reperfusion (n = 5 mice per group).(C-F) Co-immunofluorescence staining and quantifications for endothelial NF-κB p65 or VCAM-1 (green) with CD31 (red) in infarcted brain tissues of mice with the treatment of control IgG or ADM (n = 3 mice per group).Protein fluorescence signal densities were quantified and normalized to the CD31 signal area.Scale bar, 100 μm.*p < 0.05 by unpaired t test.ADM, adalimumab; Ctrl, control; MCAO, middle cerebral artery occlusion.
Top ten pathways associated with DEPs in HCMEC/D3 cells in the OGD/R group, as compared with the OGD group.