Single-Cell RNA Sequencing of Peripheral Blood Reveals That Monocytes With High Cathepsin S Expression Aggravate Cerebral Ischemia-Reperfusion Injury

Background: Stroke persists as a major cause of morbidity and mortality worldwide. After a stroke, peripheral immune cells are rapidly activated and then inltrate the central nervous system to cause inammation in the brain. However, it is not clear when and how these peripheral immune cells affect the central inammatory response and whether there are intervention targets that can alleviate ischemia-reperfusion injury. In this study, we collected mouse peripheral blood samples at different time points after stroke for single-cell sequencing to reveal the dynamic changes in peripheral immune cells. Methods: We performed single-cell sequencing on peripheral blood of mice at 1, 3, 7, and 14 days after ischemia-reperfusion to analyze the changes of subpopulations after cerebral ischemia-reperfusion; Real-time PCR, western blot and enzyme ‐ linked immunoabsorbent assay were used to perform mRNA and protein levels verication; Immunoprecipitation veries the interaction of proteins and between junctional adhesion molecule (JAM-A) and Cathepsin S (CTSS) protein, in vitro enzyme digestion and silver staining method to detect the protease digestion effect of CTSS. Results: Peripheral monocyte subpopulations increased signicantly after ischemia-reperfusion. Pseudotime trajectory analysis and gene function analysis further suggested that CTSS may play an important role in regulating monocyte activation and leading to proteolysis. Next, we found that the expression of CTSS was signicantly increased in monocytes after I/R in mice. Then, we used CTSS inhibitors and knockout mouse experiments to prove that inhibiting its expression can signicantly reduce infarct volume and reduce blood–brain barrier (BBB) leakage. In addition, we found that CTSS destroys BBB by binding to JAM family proteins to cause them degradation. Conclusion: Inhibition of Cathepsin S attenuated cerebral ischemia reperfusion injury and Cathepsin S can be used as a novel target for drug intervention after stroke. Interaction interaction


Introduction
Stroke persists as a major cause of morbidity and mortality worldwide [1,2]. After a stroke, innate immunity is rapidly activated to recognize a wide variety of molecular complexes that are perceived as foreign and potentially damaging (danger-associated molecular patterns (DAMPs) [3]. As resident cells in the brain, microglia are activated when receiving DAMP signals; then, these cells undergo morphological changes and secrete a variety of in ammatory factors [4,5]. Moreover, peripheral immune cells can be detected in the brain within a few hours after the onset of stroke. Neutrophils are the rst to in ltrate the central system and promote the development of in ammation in the lesion by releasing a variety of cytokines. Subsequently, lymphocytes are also mobilized to in ltrate into the brain, which together aggravate intracerebral in ammation in the acute phase after stroke [6]. In the subacute phase, brain injury can transform the immune system from an activated state to an inhibitory state. The main features are the decrease in lymphocyte and monocyte activity and the upregulation of the expression of anti-in ammatory factors. However, there remain controversies and knowledge gaps as to when and how peripheral cells change function and about which functions will ultimately be bene cial [7].
In addition, immune therapy for stroke has focused primarily on reducing injury volume and improving functional outcomes. Several drugs targeting the immune system that were effective in preclinical studies have failed in clinical trials. Fingolimod, a sphingosine-1 phosphate receptor (S1PR) modulator that prevents lymphocyte egress from lymph nodes, was approved by the FDA for multiple sclerosis, and there is an ongoing study of ngolimod in stroke [8,9]. Siponimod, another S1PR modulator, also blocks the egress of lymphocytes from lymphoid organs and has been demonstrated to reduce brain in ltration in an experimental stroke model [10]. However, some studies noticed that a large number of T cells in ltrated the brain 14 days after intraluminal middle cerebral artery occlusion (MCAO), and the number continued to increase thereafter [11]. When the egress of activated T cells from lymph nodes was inhibited by treating mice with ngolimod from days 6 to 13 after stroke onset, the number of CD4+ T cells and Treg cells markedly decreased without affecting other immune cell populations, and neurological recovery was delayed [12,13].
Traditionally, peripheral immune cells are thought to play a deleterious role in acute ischemic stroke [14]. Depletion of these cells in the acute phase can alleviate brain injury induced by ischemia. However, recent studies have shown that these cells have anti-in ammatory functions, participate in angiogenesis, phagocytose necrotic neurons, and promote neurovascular repair [15,16]. Therefore, peripheral immune cells play dual roles in ischemic stroke, depending mainly upon the microenvironment and the window of time after stroke.
In this study, we collected mouse peripheral blood samples at different time points after stroke for singlecell sequencing to reveal the dynamic changes in peripheral immune cells. The results showed that the proportion of monocyte subpopulations involved in the protease cleavage reaction signi cantly increased after stroke. In particular, cathepsin S (CTSS), a representative molecule of this cell subgroup, was highly expressed after stroke. We next used CTSS inhibitor and knockout (KO) mouse experiments to prove that inhibiting CTSS expression can signi cantly reduce infarct volume and blood-brain barrier (BBB) leakage, suggesting that CTSS can be used as a novel target for drug intervention after stroke.

Animals
Cathepsin S (CTSS) knockout (KO) mouse offspring of breeding pairs on a C57BL/6 background were previously generated by Cyagen Biosciences Inc. (Suzhou, China).C57BL/6 mice were provided by Army Medical University. All mice were raised in a clean environment with a light-controlled room (12 h light and 12 h dark cycle) at a temperature of 25 ± 2°C and free access to food and water. All experiments involving animals were performed in agreement with the guidelines of the National Institutes of Health on the care and use of animals and the Animal Management Committee of the Third Military Medical University. Focal cerebral ischemia model and treatment with CTSS Transient focal cerebral ischemia in mice was induced by intraluminal occlusion of the left middle cerebral artery (MCAO), as previously described [17]. Brie y, male mice (8-10 weeks, 20-25 g) were anesthetized with 1% pentobarbital sodium. A 2-cm length of rounded-tip nylon suture (Jialing, Shanghai, China) was inserted into the internal carotid artery, and then it was advanced to block the left middle cerebral artery. After 90 minutes of MCAO, the mice were allowed to recover for 1-7days. Except for artery occlusion, sham-operated mice underwent the same surgical procedure. The rectal temperature was generally controlled at 37.0±0.5°C during the operation. The CTSS inhibitor LY3000328 (HY-15533, MedChemExpress, NJ, USA)) was injected into young (8-12 weeks) C57 mice by intraperitoneal injection. All mice were randomly divided into two groups: the vehicle group (Veh), which received vehicle (corn oil, 10 mg/kg), and the CTSS-treated group (CTSS), which received CTSS (10 mg/kg) [18]; CTSS was diluted with corn oil.
Single-cell RNA-sequencing Blood was taken from the eyeballs of mice after ischemia-reperfusion (I/R) for 1, 3, and 7 days for peripheral blood single-cell sequencing. Peripheral blood mononuclear cells (PBMCs) were isolated using Ficoll-Paque PLUS (GE healthcare) density gradient centrifugation. Cells were resuspended in RPMI media (Gibco, MA, United States) to obtain a single-cell suspension with high cell viability. Next, cells were stained with a live/death dye (DAPI) and dead cells were removed using uorescence-activated cell sorting (FACS) [19]. Live cells were resuspended in PBS buffer and recounted using AO/PI double staining kit to ensure cell viability again. Finally, cell suspensions were processed for single-cell RNA-sequencing using the 10×-Genomics 3′ v2 kit, as speci ed by the manufacturer's instructions [20]. About 1 × 10 4 cells from each condition were loaded in separate inlets of a 10× Genomics Chromium controller in order to create GEM emulsions. The targeted recovery was 6,000 cells per condition. Emulsions were used to perform reverse transcription, cDNA ampli cation and RNA-sequencing library preparation. Libraries were sequenced on the Illumina HiSeq 4000 platform, using 75 bp paired-end reads and loading one sample per sequencing lane.

Single-cell RNA-seq data analysis
Single-cell RNA-sequencing data were processed using the Cell Ranger Single-Cell Software Suite (version 3.0.2, 10×-Genomics) [20]. Each sample was aligned to the mouse reference genome (mm10) using the Cell ranger, and raw expression data were analysed by R (version 3.5.1). Then, Cells of all datasets were rst analysed for their unique molecular identi er (UMI) and mitochondrial gene counts, and cells with low (<300) or high (>2500) UMI counts or high percentage of mitochondrial genes (>4%) were excluded from further analysis. Data were integrated in a standardized work ow as recommended by the developers of the "Seurat"-package [21], including data normalization, identi cation of variable genes, nding anchors for integration based on variable genes, integration of all datasets, scaling of data, principle component analysis (PCA), and unsupervised clustering with a resolution of 0.7 based on Uniform Manifold Approximation and Projection (UMAP). Cell types were identi ed based on a marker gene panel and on differentially expressed genes (DEGs) in every cluster. DEGs were calculated by Wilcoxon rank sum test with Bonferroni correction for adjusted p-values. As recommended by the "Seurat" developers, data in feature plots, violin plots, heat maps and trajectories demonstrating features that vary across conditions were displayed based on the "RNA"-count slot, and data re ecting the entire dataset were displayed based on the "integrated" dataset. Gene ontology (GO) networks based on DEGs were created using the Functional Annotation Bioinformatics Microarray Analysis (DAVID) [22].

Pseudotime analysis
Cells were ordered into a branched pseudotime trajectory using Monocle 3 and restricting the analysis to the highly variable genes identi ed by Seurat. Monocle was used to test for a signi cant correlation between gene expression and pseudotime in each trajectory [23]. A gene was de ned as signi cantly associated with pseudotime if its estimated q value was lower than 0.01.
Neurological de cit score Neurological de cits after 3 days of reperfusion in each mouse were evaluated with a 5-point scoring system using the Longa score test as previously described [24]. The scoring standards were as follows: 0 = no neurologic de cit; 1 = failure to extend the contralateral forepaw fully; 2 = circling to the contralateral side; 3 = falling over to the contralateral side; and 4 = no spontaneous locomotor activity.

Infarct volume evaluation
After 3 days of reperfusion, eight mice from each group were decapitated, and then we removed their brains to measure infarct volume by TTC staining. The mouse brains were sliced into seven sections on average and exposed to a 2% 2,3,5-triphenyl-2H-tetrazolium chloride (TTC) solution (Sangong Biotech, Shanghai, China) for approximately 15 minutes in a 37°C incubator. The white areas were infarcted tissue, and normal brain tissue was stained red. We used Image-Pro Plus 6.0 to analyze infarct volume according to the following formula: Infarct volume (%) = (right hemisphere volume−uninfarcted volume of left side)/right hemisphere volume×100 [25].

ELISA Analysis
The amount of CTSS in peripheral blood was measured using a Mouse CTSS Quantikine ELISA Kit (CUSABIO, Wuhan, China) according to the manufacturer's instruction.

Western blotting
The detailed protocol was as described in our previous article [26]. In brief, cerebral tissue on the infarct side was extracted, and protein was separated using protein lysate with EDTA-free protease inhibitor (Roche, Germany). Electrophoresis and membrane transfer were performed with 12% concentration SDS-PAGE and polyvinylidene uoride membranes. We used 5% BSA-blocked membranes at room temperature for 2 h and then incubated the membranes with primary antibodies against the following targets overnight at 4°C: β-tubulin (1:1000, Abcam) and CTSS (1:500, Abcam). The membranes were incubated with HRP-conjugated goat anti-mouse secondary antibodies and goat anti-rabbit secondary antibodies for 1 h at room temperature. Finally, we used the enhanced chemiluminescence (ECL) substrate method to visualize the proteins at the membranes and ImageJ software to evaluate the grayscale values. Western blotting-related reagents were obtained from Beyotime.

RNA scope in situ hybridization
Brain tissue was xed with PBS and 4% paraformaldehyde and then made into frozen sections. mRNA expression levels were detected using the mouse gene-speci c probes CTSS and C1q according to the manufacturer's instructions for RNAscope® detection (Advanced Cell Diagnostics,CA, United States). Confocal microscopy and ImageJ software were used for image acquisition and analysis.

TUNEL staining
Apoptotic cells were demonstrated Using the In Situ Cell Death Detection Kit, Fluorescein (Roche,54421700), brain tissues were performed following the protocol speci ed by the manufacturer.

Immunoprecipitation
Combined JAM-A antibody with magnetic beads, then lysed the brain tissue of fresh wild-type mice and I/R3 days and incubated with JAM-A magnetic bead complex overnight at 4°C. Speci c operation steps was following the protocol speci ed by the manufacturer (Thermo Fisher Scienti c, 88288).

Vascular permeability quantitation
As described in detail in the study [28], in brief, after 3 days of reperfusion, Alexa 488-uorescein thiocarbamoyl dextran (Sigma, Carlsbad, CA ,United States) was injected into wild-type (WT) or CTSS KO mice through the tail vein. After 10 minutes, brain tissues were removed, sliced into 10-µm sections and stained with uorescent antibodies against CD31 (1:100, Santa Cruz Biotechnology) to visualize mouse vascular morphology. Slices were imaged with a Leica confocal uorescence microscope as described above.

In vitro cleavage assays Cleavage of recombinant protein in vitro
We obtained recombinant inactive CTSS from (R&D Systems). Inactive CTSS was activated in Silver stain These assays were described in detail previously [29]. We obtained Quick Silver Staining Kit from Beyitime(P0017S). After the electrophoresis is completed, take the gel and put it in about 100ml of xing solution, and shake it on a shaker at room temperature for 20 minutes at a shaking speed of 60-70 rpm. Fixing for 2h to reduce the background. Discard the xative, add 100ml of 30% ethanol, and shake on a shaker at room temperature for 10 minutes at a shaking speed of 60-70 rpm.Discard the original solution, add 200ml Milli-Q grade pure water or double distilled water, shake on a shaker at room temperature for 1 minute at a shaking speed of 60-70 rpm and reduplicate once. Discard water, add 100ml silver solution (1X), shake on a shaker at room temperature for 10 minutes at a shaking speed of 60-70 rpm. Discard the original solution, add 100ml Milli-Q grade pure water or double distilled water, and shake on a shaker at room temperature for 1-1.5 minutes at a shaking speed of 60-70 rpm.Discard the water, add 100ml of silver-stained color developing solution, shake for 3-10 minutes at room temperature on a shaker, until an ideal expected protein band appears, and the shaking speed is 60-70rpm. Discard the silver staining solution, add 100ml silver staining stop solution (1X), and shake on a shaker at room temperature for 10 minutes at a shaking speed of 60-70 rpm. It is normal for gas to be produced when it is terminated, and the gas produced is carbon dioxide. Discard the silver dye stop solution, add 100ml Milli-Q grade pure water or double distilled water, shake on a shaker at room temperature for 2-5 minutes at a shaking speed of 60-70rpm.

Isolation of cells and ow cytometry analysis
The brain tissues of the control and ischemic reperfusion 3 days were broken up and incubated with papain and DNase for 30 minutes at 37 • C. Horse serum was used to terminate the digestion, and the tissue was pipetted into a single-cell suspension. Then, the myelin sheath and tissue debris were removed with a solution (3 ml DMEM, 1 ml Percoll, 4 ml D-PBS), and the red blood cells were lysed; The single cells obtained were blocked with 10% FBS at room temperature, and then the corresponding primary antibodies (CD45) were added; nally, the samples were washed and analyzed by ow cytometry.

Statistical analysis
Data was shown as percentage or the mean ± SEM and using student's t-tests to assess the statistical differences between two groups. Comparisons among multiple samples were evaluated by one-way ANOVA. Statistical signi cance was set at P<0.05.

Single-cell transcriptome pro ling of peripheral blood immune cells in mice
To explore the dynamic changes in peripheral blood immune cells in mice after stroke and the relationship with ischemia, we collected fresh peripheral blood samples derived from mice with MCAO for 1, 7, 14 days and the control mice for single-cell sequencing ( Figure 1A).After ltering out cells with low quality, we obtained transcriptome datasets from 36,905 cells with an average of 9,000 cells for each sample at each time point. The cell clusters were annotated with expression of canonical marker genes. Major cell types comprising PBMCs could be well captured by scRNA-seq, including CD4+ T cells (CD4), CD8+ T cells (CD8), Treg (FOXP3), B cells (CD19), monocytes (CD68), natural killer (NK) cells (KLRB1), and proliferative cells (MKI67) ( Figure 1B). We found that compared with controls, MCAO-treatment mice showed an increased percentage of monocytes, while a decreased percentages of B cells. Other types of peripheral blood cells (proliferative cells, CD8+ T cells, Treg cells) account for less than 10% of the total cells, so their dynamic changes are not clear in this experiment ( Figure 1C).
In addition, we found 28 clusters representing different cell types using t-distributed stochastic neighbor embedding (t-SNE) analysis ( Figure 1D). The top 10 featured genes of each cluster were displayed in the heat map ( Figure S1 and Table S1). We further calculated the proportion of each cell subgroup at different ischemia-reperfusion time points ( Figure 1E). Then we divide it into four categories, which indicated that the proportion of cell subpopulations continues to rise (group 1) or continues to decline (group 2) with the prolonged perfusion time, etc ( Figure 1F), among them, the 2, 9, and 10 cell subgroups in group 1 showed an increasing trend ( Figure 1F). In particular, clusters 2 and 10 belong to monocyte subgroups, which further con rms that monocytes and their subgroups continue to increase their expression after ischemia-reperfusion.

Pseudotime analysis reveals the dynamic changes of cell subpopulations
To understand the hypothetical developmental relationships that might exist within the monocyte and macrophage clusters, we performed trajectory analysis on clusters 2, 7, 10, and 18 using the Monocle algorithm. Two branch points were determined based on changes in monocytes gene expression and this was plotted in pseudotime. Clusters were superimposed on the monocle pseudotime plot and revealed that cluster 18 and 2 fell towards the beginning while cluster 7 on the tail of pseudotime (Figure 2A-C). Interestingly, the distribution of cluster 10 is located in the entire pseudo-time trajectory, suggesting that cluster 10 may play a key role in the differentiation process of monocytes ( Figure 2C). Therefore, we checked the expression levels of the characteristic genes (including APOE, CSF1R and CTSS) in cluster 10 in the pseudo-time trajectory. We found that the expression levels of these genes were signi cantly increased ( Figure 2D), suggested that the them may be the key molecules in the activation process of monocytes after ischemia-reperfusion. On the contrary, the expression of characteristic genes in cluster 2 (including CCL6, CCR2, FN1 and CHIL3) gradually decreases with the pseudo-time trajectory ( Figure 2D), indicating that their effect in promoting monocyte activation was limited.

Gene functional analysis of each cell cluster
According to the previous data (Figure 2), we found that cluster 2 and 10 may play the key role in regulating the activation of monocytes. Thus, we further performed GO analysis on these subpopulations. Through biological process and molecular function analysis, we found that the biological functions of cluster 2 mainly focus on in ammation and toll-like receptor pathways ( Figure 3A). In order to obtain the core genes that regulate the biological functions of the cluster 2, we merged the data and take the intersection. It was found that S100a8 and serum amyloid A-3 protein (Saa3) were the key regulatory genes in cluster 2 ( Figure 3B). However, we excluded them in the follow-up study because the function of these two genes in stroke have been widely reported.
Then, we performed gene function analysis on the cluster 10. It was showed that the biological functions of this subgroup were mainly enriched for defense response, in ammatory response, proteoglycan binding and collagen binding ( Figure 3C). We have also obtained the core gene that regulate cluster 10 by combining the results of biological process and molecular function analysis ( Figure 3D). We found that Cathepsin S (CTSS) was the only molecule that can regulate in ammation and bind with collagen at the same time. Since the activation of in ammatory cells and the degradation of collagen in the blood-brain barrier after stroke are important pathological features, we speculated that CTSS may act on these two pathways to affect the outcome of stroke.
Besides, we also analyzed the molecular functions of other clusters and found that each group has its own different functions ( Figure S2). In group 2, there are mainly T cell subgroups (including cluster 3,13 and 14) whose molecular functions were concentrated on ribosomal composition and RNA binding, suggesting that these molecules may be involved in transcriptional regulation of T cells,The main gene function enrichment in group 3 and 4 were related to the binding of antigen and MHC, and the activation of chemokines. The genes involved in these functions and their regulatory mechanisms remain to be resolved.

CTSS expression increases after stroke
To verify the results of single-cell sequencing, we detected the expression of CTSS in peripheral blood of mice by ELISA, and found that its expression increased after MCAO in C57BL/6N mice,which is consistent with our single-cell sequencing results. (Figure 4A). Since the peripheral immune cells (monocytes, neutrophils, lymphocytes) would aggravate the neuroin ammation when they enter brain after the BBB is destroyed, the expression of CTSS could be evaluated in the central nervous system.Quantitative real-time PCR revealed highly increased CTSS mRNA in the ischemic brain 3 days to 7 days after stroke and gradually returned to baseline at 14 days, western blot analysis also indicated the protein expression increased after MCAO in mice ( Figure 4B and C).These results suggested that the expression of CTSS was signi cantly increased mainly in the acute and subacute phases.
Our single-cell sequencing results showed that CTSS was mainly derived from peripheral blood monocyte subpopulations, thus,we assessed the location of CTSS in the mouse brain and found that CTSS was colocalized with C1q (a marker of microglia/macrophages) via in situ hybridization analysis in mouse brains ( Figure 4D and E), which was consistent with a previous article reported that CTSS displayed tissue speci c distribution, selectively expressed on Antigen Presenting Cells (APCs),including monocytes/ macrophages, microglia and other cells. [30,31] In addition, we obtained from the Gene Expression Omnibus database (GSE16561) that the expression of CTSS in the peripheral blood of patients with stroke was obviously increased ( Figure 4F). These results con rmed that the expression of CTSS was signi cantly increased after stroke, which may be a potential target for intervention.

Ablation of CTSS ameliorates cerebral ischemic damage
We rst veri ed the knockout e ciency of CTSS-KO mice by quantitative real-time PCR (Figure 5A), and the results showed that the mRNA level of CTSS was signi cantly suppressed (close to 80%). Then, to determine the impact of CTSS on cerebral ischemic injury, male CTSS KO mice and their WT littermates were subjected to 90 minutes of ischemia followed by reperfusion for 3 days. TTC and Longa scores were respectively used to evaluate the area of cerebral infarction and neurological de cit in mice. The brain infarct areas of MCAO(TTC) CTSS KO mice were smaller than those of WT mice ( relative to the contralateral hemisphere; P < 0.05) (Figure 5B), and the neurological function score was also less than that of WT mice ( Figure 5C). Compared with WT mice, CTSS KO mice exhibited lower levels of neuronal apoptosis after MCAO. (Figure 5D). Furthermore, BBB disruption is also an important feature for neurological disorders after stroke [32]. We found that ischemia-induced BBB leakage after stroke was obviously decreased in CTSS KO mice via the reduced permeability of thiocarbamoyl dextrans compared with that in WT controls ( Figure 5E).
According to our data analysis in Figure 2, CTSS may be related to the activation and function of monocytes. Besides, some articles reported CTSS is involved in immune responses through major histocompatibility complex class II antigen presentation and inactivate key innate immunity proteins, such as β-defensins 2,3 and secretory leukocyte protease inhibitor [33].Thus, we speculated CTSS can reglute neuroin ammatory in mice after MCAO. We found that immune cell in ltration caused by ischemia was signi cantly reduced by the decrease in CD45 in the CTSS KO mice by ow cytometry ( Figure 5F). In addition, quantitative real-time PCR revealed the in ammatory factors IL-1β, IL-6, and TNFα were signi cantly reduced after ischemia reperfusion ( Figure 5G-I) which was consistent with the previous study reported that CTSS promotes the release of in ammatory factors IL-1 [34] from macrophages and microglia [35]. These results showed that knockout of CTSS can relieve neuroin ammation after ischemia in mice.

Cathepsin S promotes BBB destruction through junctional protein cleavage
We found in previous experiments that inhibiting CTSS can alleviate stroke damage, but the speci c molecular mechanism of CTSS is still unclear. So we predicted that there are multiple interacting proteins with CTSS through the interacting protein database, including HLAs, CD72, TLR9, etc ( Figure 6A). Among them, junctional adhesion molecules (JAMs) are one of the components of tight junction proteins and are critical for maintaining the integrity of the BBB. Then we con rmed the interaction between JAM-A and CTSS through CO-IP experiments of brain tissue, indicating that JAM is very likely to be one of the substrates of CTSS. Moreover, its interaction is enhanced after cerebral ischemia, which may be due to the increase of CTSS in the brain after stroke ( Figure 6B).In tumors, CTSS can increase tumor brain metastasis by degrading JAMs of the blood-brain barrier [29], thus we con rmed CTSS can increase the permeability of the blood-brain barrier after stroke by degrading JAMs in mice.
Cathepsin S is a lysosomal cysteine protease that exerts proteolysis activity across a wide range of pH in macrophages [33].Thus we veri ed that JAMs are substrates of CTSS and are active under conditions of acidic pH (4.5), neutral pH (6.0) and alkaline pH (9.8) via in vitro enzyme digestion experiments ( Figure   6C). JAM-A and JAM-B were cleaved into small fragments of approximately 25 kD by CTSS recombinant protein under neutral and acidic conditions and slightly cleaved under alkaline conditions JAM-C could not be cleaved under neutral and acidic conditions but was slightly cleaved under alkaline conditions. This is not consistent with the previous article JAM-A and JAM-B can be degraded under acidic conditions, but only JAM-B can be cut under neutral conditions. It may be due to the different conditions of tumor and stroke, and it was also discovered for the rst time that the JAM family can be slightly cleaved by CTSS under alkaline conditions. [29] However, the cleavage of JAM was inhibited by a CTSS inhibitor, LY3000328,which was consistent with our expectations. These results indicated that CTSS inhibitors attenuate the destruction of the BBB by inhibiting the cleavage of JAM after stroke.
Pharmacological inhibition of CTSS alleviates cerebral ischemic damage in mice All mice, including the inhibitor group (intraperitoneal injection of CTSS inhibitor, dissolved in corn oil) and control group (intraperitoneal injection of corn oil), were subjected to MCAO for 90 minutes and reperfusion for 3 days. The CTSS inhibitor signi cantly decreased infarct size and neurological de cits ( Figure 7A and B). The permeability of the BBB in mice after MCAO was also reduced in the inhibitor group ( Figure 7C). These results suggest that pharmacological inhibition and de ciency of CTSS alleviates ischemic injury in mice.

Discussion
Single-cell RNA sequencing (scRNA-seq) has been widely used to characterize the dynamics of various cells to identify disease-related cell subgroups [36,37]. Recent studies have revealed through peripheral blood single-cell sequencing technology that T cell subsets are involved in the regulation of AD and aging [38,39], but there is still a lack of research on peripheral blood changes after stroke. According to the changes in the proportion of each cell subpopulation at different times of ischemia-reperfusion (I/R), we further divided the cells into four categories and rst analyzed the continuously rising cell subpopulation in this study. The results showed that peripheral monocyte subpopulations increased signi cantly after I/R, which was consistent with previous reports. In particular, through pseudo-time trajectory analysis, we have obtained the key cell subpopulations that regulate monocyte activation, and this subpopulation was signi cantly enriched in the protease cleavage pathway through GO analysis. These results enhanced the understanding of the biological functions of monocytes after I/R. Furthermore, we found that CTSS, a key molecule involved in the regulation of protease cleavage, was signi cantly highly expressed in monocytes after I/R. CTSS, as a lysosomal enzyme, is expressed in a wide variety of immune cells. CTSS plays a signi cant role in various intracellular and extracellular processes, including proteolysis and major histocompatibility complex (MHC) class II-mediated immune responses [40][41][42]. Dysregulated expression and activity of CTSS is linked to the pathogenesis of multiple diseases, including a number of conditions affecting the lungs, liver and heart [43][44][45]. Unlike other members of the lysosomal cathepsin family that require an acidic pH, CTSS has potent endoproteolytic activity in a broad range of pHs, [46] which indicates that CTSS is proteolytically active at the neutral pH found in the extracellular microenvironment. Previous research found that CTSS e ciently cleaved the junctional adhesion molecule (JAM) family members JAM-A, JAM-B and JAM-C at pH 4.5, the acidic pH of the lysosome, and maintained robust cleavage of JAM-B speci cally at pH 6.0, the acidi ed pericellular pH measured in solid tumors. They also found that CTSS speci cally mediates BBB transmigration through proteolytic processing of the JAM to accelerate breast-to-brain metastasis [29,32].. This is similar to our results, con rming that CTSS destroys the BBB by degrading JAM family proteins over a broad pH range.
JAM family proteins are essential for maintaining the structural integrity of the BBB [47,48]. A large number of previous studies have found that after stroke, metalloproteinase (MMP) family proteins (speci cally MMP-2, MMP-7 and MMP-9), which are released from in ltrated immune cells, disrupt the BBB by degrading JAMs and other extracellular matrix proteins [32] [49]. Inhibition of peripheral immune cell in ltration through blood replacement or ngolimod treatment can effectively reduce the expression of MMP proteins, thereby reducing I/R injury [10]. However, the protease cleavage activity of MMP proteins was sensitive to the pH value [50][51][52][53]. Whether the change in pH value during ischemia and reperfusion has an effect on MMP activity needs to be further evaluated. In this study, we found that CTSS degrades JAM family proteins over a broad pH range. Additionally, either the use of CTSS KO mouse models or the use of CTSS inhibitors can prevent the destruction of the BBB caused by JAM degradation, suggesting that CTSS is an important treatment target for I/R.
Although we found that CTSS may be an important molecule that aggravates the destruction of the BBB by peripheral immune cells, there is still a large amount of single-cell sequencing data that urgently needs to be deciphered. For example, the cluster 2 and 9 subgroups, which are highly expressed throughout the acute and chronic phases, have stroke-associated functions related to oxidative stress, Toll receptors and SH3/SH2 receptors. The cluster 1 and 4 subgroups increased in the acute phase but began to decrease after 7 days of ischemia, indicating that they are likely to participate in the recovery of stroke. Therefore, it is necessary to further verify that these groups of cells are involved in acute ischemia and chronic repair.

Conclusions
In summary, through single-cell sequencing at different time points in cerebral I/R mice, we found that cluster 10 was involved in the enzyme digestion process. In this group, CTSS was obviously highly expressed after stroke, and CTSS inhibitors or KO in mice alleviated the destruction of the BBB by inhibiting JAM cleavage and neuroin ammation. We conclude that CTSS inhibitors may serve as new targets for clinical brain ischemia protection therapy.  <p>Circulating and cerebral CTSS are increased in mice after MCAO. C57BL/6N mice were subjected to sham operation or MCAO for 90 minutes followed by 1, 3, 7, and 14 days of reperfusion. (A) Serum CTSS levels by ELISA after MCAO in mice ( n = 6). (B) Representative immunoblots and quanti cation of CTSS (relative to GAPDH) in the ischemic ipsilateral brain at various time points after MCAO ( n = 4). (C) The mRNA abundance of CTSS in the ischemic ipsilateral brain normalized to that of the GAPDH gene (n =5).