Endosialin-positive tumor-derived pericytes promote tumor progression through impeding the infiltration of CD8+ T cells in clear cell renal cell carcinoma

Background Immune checkpoint blockade (ICB) therapy can be effective against clear cell renal cell carcinoma (ccRCC), but many patients show no benefit. Tumor-derived pericytes (TDPs) may promote tumor progression by influencing T cells and are an immunotherapy target; however, they may comprise functionally distinct subtypes. We aimed to identify markers of tumor-promoting TDPs and develop TDP-targeting strategies to enhance ICB therapy effectiveness against ccRCC. Methods We analyzed the relationship between endosialin (EN) expression and cytotoxic T-lymphocyte (CTL) infiltration in ccRCC tumor samples using flow cytometry and in a ccRCC-bearing mice inhibited for EN via knockout or antibody-mediated blockade. The function of ENhigh TDPs in CTL infiltration and tumor progression was analyzed using RNA-sequencing (RNA-seq) data from ccRCC tissue-derived TDPs and single-cell RNA-seq (scRNA-seq) data from an online database. The role of EN in TDP proliferation and migration and in CTL infiltration was examined in vitro. Finally, we examined the anti-tumor effect of combined anti-EN and anti-programmed death 1 (PD-1) antibodies in ccRCC-bearing mice. Results High EN expression was associated with low CTL infiltration in ccRCC tissues, and inhibition of EN significantly increased CTL infiltration in ccRCC-bearing mice. RNA-seq and scRNA-seq analyses indicated that high EN expression represented the TDP activation state. EN promoted TDP proliferation and migration and impeded CTL infiltration in vitro. Finally, combined treatment with anti-EN and anti-PD-1 antibodies synergistically enhanced anti-tumor efficacy. Conclusion ENhigh TDPs are in an activated state and inhibit CTL infiltration into ccRCC tissues. Combined treatment with anti-EN and anti-PD-1 antibodies may improve ICB therapy effectiveness against ccRCC. Supplementary Information The online version contains supplementary material available at 10.1007/s00262-023-03372-z.


Introduction
Renal cell carcinoma (RCC) accounts for 3-5% of all malignancies, with 75% being of the clear cell (ccRCC) type [1]. Immune checkpoint blockade (ICB) therapy, such as with antibodies against programmed death 1 (PD-1) and cytotoxic T lymphocyte antigen 4 (CTLA-4), has shown promising effectiveness in patients with ccRCC [2]. However, > 50% of patients show no benefit, and some patients who are initially sensitive to the therapy may develop secondary resistance [3,4]. Therefore, new therapeutic targets are needed in ccRCC.
Tumor-derived pericytes (TDPs) are important cellular components of the tumor microenvironment (TME), with roles in tumor angiogenesis, metastasis, resistance to treatment, and patient mortality [5,6]. Recently, TDPs were shown to be involved in cross-talk with tumor cells, endothelial cells, and immune cells in the TME, resulting in the promotion of tumor progression [7]. Several TDPtargeting strategies have been developed [8]; however, similar to cancer-associated fibroblasts (CAFs), TDPs are heterogeneous and may have different origins [9,10]. Therefore, for precise therapeutic targeting, it is critical to both identify the tumor-promoting TDP subtypes and describe specific markers for these. Endosialin (EN), also known as tumor endothelial marker 1 (TEM1) or CD248, is a transmembrane glycoprotein of the C-type lectin-like receptor family [11] and is highly expressed on CAFs and TDPs [12]. Previously, we found that EN + CAFs promoted hepatocellular carcinoma progression by recruiting and inducing M2 polarization of macrophages [13]; however, the function of EN in TDPs has not yet been elucidated. EN has been shown to regulate platelet-derived growth factor-induced cell proliferation and migration in TDPs [14]; additionally, EN + TDPs may promote tumor cell intravasation, thereby facilitating distant metastasis [15]; however, whether they promote tumor progression by other mechanisms, such as by influencing T cell infiltration, is unknown.
Here, we examined the relationship between EN expression and the infiltration of CD8 + T cells, or cytotoxic T lymphocytes (CTLs), in ccRCC clinical samples. We then examined the effect on CTL infiltration and tumor growth of EN inhibition via knockout or antibody-mediated blockade. We next examined whether EN + TDPs could inhibit CTL infiltration in vitro. Additionally, using knockdown or antibody-mediated blockade, we examined the function of EN in TDP proliferation, migration, and tube formation. Finally, we investigated whether combined treatment with anti-EN and anti-PD-1 antibodies inhibited ccRCC growth in vivo.

Culture of cell lines and isolation of primary stromal cells or pericytes
The human retinal pericyte cell line HRMVP and the mouse RCC cell line Renca were purchased from Aoyinbio Co., Ltd. (Shanghai, China), while the mouse prostate cancer (PCa) cell line RM1 and mouse melanoma cell line B16F10 were purchased from Zqxzbio Co., Ltd. (Shanghai, China). The cells were maintained in Dulbecco's Modified Eagle's Medium (DMEM) supplemented with 10% fetal bovine serum (FBS) (Thermo Fisher Scientific, Waltham, MA, USA; #A3161002C) and 1% penicillin-streptomycin (#15070063, Gibco). Primary stromal cells or pericytes were isolated from tumor samples obtained between May 2021 and June 2021 from patients with ccRCC or lung cancer at Xijing Hospital, Fourth Military Medical University (Xi'an, China), according to previously described protocols [6,16].

Animal experiments
The animal experiments in this study were approved by the Laboratory Animal Welfare and Ethics Committee of Fourth Military Medical University. EN-knockout (EN KO ) C57BL/6 mice were purchased from the Shanghai Model Organisms Center (Shanghai, China; #NM- KO-200094). BALB/c mice (6-8 weeks old, female, body weight 18-23 g) were purchased from the Animal Center of Fourth Military Medical University (Xi'an, China) and maintained in a 12-h light/12h dark cycle with free access to food and water. Each mouse was subcutaneously inoculated in the back with 2 × 10 6 1 3 tumor cells. For the antibody treatment, mice were inoculated with anti-mouse PD-1 antibody (50 mg/kg; Bio X Cell, Lebanon, NH, USA; #BE0033-2) with or without anti-EN antibody (5 mg/kg) every 4 d for a total of four times. Tumor size was measured every 4 d, tumor growth curves were drawn, and the survival of mice was observed. When the tumor diameter reached 2 cm or when severe weakness was observed, mice were euthanized and tumors isolated. The tumor samples were digested into a single-cell suspension, and associated T cells were analyzed using flow cytometry with fluorescently labeled anti-CD3, anti-CD4, anti-CD8, and anti-CD69 antibodies (#100306, #100412, and #100708, Biolegend).

RNA-seq and scRNA-seq analysis
RNA was isolated from TDPs harvested from ccRCC samples from EN high or EN low groups and analyzed using the two-terminal sequencing model of the Illumina HiSeq sequencing platform (Genergy Bio-Technology, Shanghai, China). ScRNA-seq data for ccRCC were obtained from 48 samples from 38 patients [17][18][19][20].
ScRNA-seq dataset analysis was performed using the Seurat package (v4.0.5) in R (v4.1.0) [21]. First, individual gene expression matrices were used to create distinct Seurat objects. Second, strict quality-control procedures were performed: data were filtered to exclude cells that expressed < 200 genes, > 8,000 genes, > 20% mitochondrial genes, or > 0.1% hemoglobin genes; genes expressed in fewer than three cells were also excluded. Third, all Seurat objects were combined into one dataset, which was normalized and scaled using the NormalizeData and ScaleData functions, respectively. The FindVariableFeatures function was used to identify variable genes. Based on the variable genes, principal component analysis (PCA) was conducted using the RunPCA function. The datasets and corrected batch effects from different samples were integrated using the Harmony package (v0.1.0) [22]. Finally, clustering was conducted using the FindNeighbors and FindClusters functions with a resolution of 0.8. Visualization was implemented via t-distributed stochastic neighbor embedding.

Immunohistochemistry (IHC)
IHC was performed using primary antibodies against human EN (#ab204914, Abcam) and CD8 (#70306S, CST). The expressions of EN and CD8 were assessed according to the percentage and intensity of the staining [23,24].

T cell infiltration assay
EN-silenced (si-EN) or control HRMVP cells (7.5 × 10 5 ) were plated in pericyte medium on 8.0-mm-pore Matrigelcoated Transwell filters in the upper 24-mm chamber (Corning, Glendale, AZ, USA) and cultured for 24 h. Meanwhile, human peripheral blood mononuclear cells (PBMCs) were isolated from healthy donors, and T cells were isolated using the Dynabeads FlowComp human CD3 kit (#11365D, Invitrogen). T cells (5 × 10 5 ) in DMEM supplemented with 5% fetal calf serum (FCS) were added to the upper chamber, while DMEM supplemented with 20% FCS was added to the lower chamber. After 24-h incubation, cells in the lower chamber were collected for flow cytometric analysis by using an anti-CD8 antibody (BioLegend; #344717).

Cell proliferation and migration assay
Cell proliferation was measured using the cell-counting kit 8 (CCK8) assay (Mishubio, Xi'an, China). Si-EN or control cells were seeded into 96-well plates (3 × 10 3 cells/well) and incubated for the indicated time. At the indicated time points, 10 μL of CCK8 reagent was added to each well followed by incubation at 37 °C for 1 h. Plates were then gently shaken in the dark for 5 min before absorbance at 450 nm was measured using a microplate reader.
To measure cell migration, a wound healing assay was performed. First, HRNVP cells were either transfected with siRNA or incubated with anti-EN antibody or control IgG for 45 min. After 48 h, cells were seeded into a six-well plate (5 × 10 5 cells/well) and cultured for 12 h. Next, a scratch was made using a 200-μl pipette tip, and the culture medium was changed to serum-free medium. Images were taken (as time point 0 h) and the plates were returned to the incubator; further images were taken at 12 h, 24 h, and 48 h.

Tube formation assay
Matrigel (BD Biosciences) was added to a 48-well plate (150 μL/well) and incubated at 37 °C for 30 min. Primary stromal cells were suspended by brief exposure to 0.25% trypsin (Invitrogen) and incubated with anti-EN antibody or control IgG in 50% phosphate-buffered saline (PBS)/45% basal media/5% FBS for 45 min, then seeded onto the Matrigel-covered plate. Si-EN and control HRNVP cells or primary pericytes in 300 μL of PBS/media/2% FBS were 1 3 added to the Matrigel-covered plate (21,000 cells/well) and placed in a humidified 37 °C incubator for 16 h. Tubes/networks were imaged using Metamorph Image Analysis Software (Molecular Devices, Sunnyvale, CA).

Statistical analysis
All statistical analyses were performed using IBM SPSS statistical (version 23) or GraphPad Prism 8 software. Results were presented as the mean ± SD. Comparisons were analyzed using two-tailed Student's t test or one-way ANOVA followed by Dunnett's post-hoc test. Differences were considered statistically significant at P < 0.05.

High EN expression negatively correlates with CTL infiltration in patients with ccRCC
Using the GEPIA database, we found a negative correlation between CD8A and endosialin expression (Supplementary Figure 1A). Therefore, we examined the percentage of EN + TDPs in fresh nephrectomy samples from ten patients with ccRCC; the patients' clinical information is shown in Supplementary Table 1. We chose to exclude epithelial cells, endothelial cells and immune cells, retaining as much TDPs as possible for detection. And through identification, we confirmed that these cells were TDPs (Supplementary Figure 2A-B). Results showed that the percentage of EN + TDPs was 20-80% and, using a cut-off value of 40%, we divided patients into EN high and EN low groups (Fig. 1A). We analyzed the T cell infiltration in the samples and found that CTL infiltration was much lower in the EN high group than in the EN low group, while no significant difference was observed in CD4 + T cell infiltration (Fig. 1B). The expression of the T cell exhaustion markers PD-1, CTLA-4, and Tim-3 showed no significant difference between the two groups ( Supplementary Fig. 1B-D). Additionally, no significant difference was observed in FoxP3 + regulatory T cell (Treg) infiltration (Supplementary Figure 1E-F). Using IHC on 80 ccRCC samples, we found that patients with high EN expression exhibited less CTL infiltration, indicating a negative correlation between EN expression and CTL infiltration (Fig. 1C).

Antibody blockade of EN promotes CTL infiltration and inhibits RCC growth in vivo
To study the function of EN in CTL infiltration, we used anti-EN antibody in an RCC mouse model of BALB/c mice subcutaneously inoculated with Renca cells. EN expression significantly decreased in the anti-EN group ( Fig. 2A), tumor growth significantly inhibited, and mouse survival increased (Fig. 2B-C). We next examined T cell infiltration in tumor tissues following anti-EN treatment and found that it increased for both CD8 + and CD4 + T cells compared to the control group (Fig. 2D).
We also found that blocking EN expression resulted in a significant decrease in mature blood vessels. Furthermore, we confirmed that EN was mainly expressed by pericytes in the RCC mouse model (Fig. 2E-F). These results indicate that EN inhibits T cell infiltration in RCC.

ENKO mice exhibit increased CTL infiltration and impaired tumor growth
To determine the function of EN in CTL infiltration in tumors, we inoculated B16F10 and RM-1 cells into EN KO or WT mice. We found that EN expression was lower in the EN KO mice than in the WT mice (Fig. 3A). We also analyzed T cell infiltration in tumor tissues using flow cytometry. In both tumor models, results showed that CTL infiltration was significantly higher in the tumor samples of EN KO mice than in WT mice; no significant differences were observed in CD4 + T cell infiltration (Fig. 3B-C). Additionally, tumor growth in both the melanoma and PCa cells was inhibited in EN KO mice, consistent with the effect of the anti-EN antibody treatment (Fig. 3D-E). These results confirmed that EN impedes CTL infiltration and results in tumor progression.

High EN expression represents the TDP activation state in ccRCC
To determine how EN inhibits CTL infiltration, we analyzed TDPs isolated from the tumor tissues of four EN high and three EN low group members using RNA-seq (Fig. 4A). Gene Ontology enrichment analysis showed that highly expressed genes in TDPs from EN high patients were related to blood vessel morphogenesis, angiogenesis, cell adhesion, extracellular structure organization, and cell migration, among others, thus indicating that the high EN expression related to TDP activation status (Fig. 4B). Due to the high heterogeneity of TDPs, which is difficult to overcome by RNAseq, we integrated scRNA-seq data from 48 ccRCC samples and analyzed differentially expressed genes in EN high and EN low ccRCC tissues [17][18][19][20]. The results showed that ccRCC tumors contained a higher percentage of TDPs than adjacent normal tissues and that TDPs divided into five subclusters based on different expression signatures (Fig. 4C-D; Supplementary Figure 2A-B). Each subcluster had a unique expression signature, which reflected TDP heterogeneity ( Supplementary Fig. 2C). Subclusters 1, 2, and 3 exhibited high EN expression and were proportionally increased in ccRCC tissues ( Fig. 4E; Supplementary Figure 2D-E). In 1 3 contrast to EN low TDPs, EN high TDPs exhibited activation in almost every signaling pathway, including those of p53, PI3K-Akt-mTOR, glycolysis, and transforming growth factor-beta (TGF-β) (Fig. 4F). These results suggest that high EN expression represents an activated state of TDPs in ccRCC.

EN knockdown inhibits pericyte proliferation and migration and promotes CTL infiltration
Previous studies found that activated TDPs possess strong proliferative and migratory abilities [25,26]. Based on our bioinformatic analysis, we speculated that EN may promote the activation of TDPs. Therefore, we examined pericyte proliferation and migration following EN knockdown or blockade by anti-EN antibody. Results showed that cell proliferation and migration were inhibited in EN KO HRMVP cells (Fig. 5A-B). To corroborate this, we isolated TDPs from the fresh tumor tissue of a lung cancer patient and treated them with anti-EN antibody. The results showed that primary TDP migration was inhibited (Fig. 5C). We also examined the tube formation ability of HRMVP cells and primary TDPs after EN knockdown or anti-EN treatment and found that it was inhibited following both treatments (Fig. 5D-E). Additionally, we examined phosphorylation of ERK1/2 and Akt in EN KO HRMVP cells and found that it was inhibited in both, indicating that the cells were inactive (Supplementary Fig. 3A). Using flow cytometry, we found no increased apoptosis in HRMVP cells following IgG78  Fig. 3B). These results confirmed that EN is involved in maintaining the TDP activation state.
Since we previously found that EN may impede CTL infiltration into tumors, we examined whether EN KO pericytes promote CTL infiltration by culturing EN KO and control HRMVP cells in the upper Transwell chamber and analyzing the infiltrating T cells using flow cytometry. The results showed more infiltration through EN KO HRMVP cells than through the control cells (Fig. 5F). In agreement with The Cancer Genome Atlas, we found that EN and CXCL12 expression positively correlated in  Fig. 3C-E). At high levels, CXCL12 (also called SDF1) acts as a chemorepellent of lymphocytes [27], which partly explains why EN impedes CTLs, though the mechanism remains unclear. Thus, we confirmed that EN inhibits CTL infiltration via pericytes.

Combined anti-EN and anti-PD-1 antibody treatment inhibits RCC growth in vivo
Since EN blockade promoted CTL infiltration, we investigated whether combined anti-EN antibody and ICB treatment with anti-PD-1 could have a stronger anti-tumor effect in RCC-bearing mice (Fig. 6A). Compared to anti-PD-1 antibody alone, tumor growth was significantly inhibited and mice survived longer following combined treatment ( Fig. 6B-C). Examining T cell infiltration and activation in tumor tissues using flow cytometry, we found that both CD4 + and CD8 + T cell infiltration increased following combined treatment and that infiltrating T cells were largely activated, as shown by increased percentages of both CD69 + CD4 + and CD69 + CD8 + T cells (Fig. 6D-F). These results indicate that combined treatment with anti-EN and anti-PD-1 antibodies can synergistically inhibit tumor growth and may overcome resistance to ccRCC ICB therapy.

Discussion
Despite being immunogenic, RCC inhibits anti-tumor immunity, partially by upregulating immune checkpoint expression in the TME [3]. ICB is recommended as first-line therapy, and combination therapies involving ICB are now standard for patients with advanced RCC [28]. However, a substantial proportion of patients do not benefit from such approaches [4].
The factors that determine RCC responsiveness to ICB therapy have been subject to much research. Braun et al. found that CTL-infiltrated tumors have less PBRM1 mutations but are enriched for chromosomal losses of 9p21.3, explaining why PD-1 blockade is resisted in patients with high CTL infiltration [29]. Peng et al. reported that CTLs were proportionally reduced in recurrent RCC and were significantly associated with weaker immunotherapy responses [30]. Many combination therapies have been developed and applied in clinical trials to overcome ICB therapy resistance [31,32]. Although these have dramatically improved the outcomes of patients with metastatic RCC, most patients either have primary resistance or acquire resistance to these therapies [4]. Identifying targets for RCC combination therapies is urgently required.
TDPs are important components of the TME, playing key roles in tumor progression. Recently, TDPs were found Fig. 6 Combined anti-EN and anti-PD-1 antibody treatment inhibits RCC growth in vivo A Schematic image to show the process of combined treatment. Isolated RCC tissues after mice were sacrificed after treatment. B Tumor growth curves after treatment. C Survival curve of the RCC beating mice after treatment. D Flow cytometry analysis of intratumoral T cells in different groups after treatment. E-F Quantification of the flow cytometry data to show the intratumoral T cells (CD3 + CD8 + , CD3 + CD4 + ) and active T cells (CD8 + CD69 + , CD4 + CD69 + ) in different groups after treatment. G Graphic diagram to describe the function of endosialin, which inhibits the activation of TDPs and impede the infiltration of CD8 + T cells into tumor tissues. The number of samples in each group of all the above experiments was 8 1 3 to regulate T cell function in the TME; e.g., glioma TDPs inhibited T cell proliferation by releasing prostaglandin E2, serum human leukocyte antigen G, hepatocyte growth factor, and TGF-β and were negatively associated with T cell infiltration [33]. Elsewhere, glioblastoma (GBM) TDPs strongly secreted anti-inflammatory cytokines and immunosuppressive molecules and have reduced co-stimulator expression, suppressing CD4 + T cell responses and IL-2 production in vitro [34]; additionally, GBM TDPs upregulated chaperone-mediated autophagy to enhance the expression of TGF-β and IL-10, which inhibit T cell function [35]. TDPs are, thus, an attractive target for immunotherapy [36].
TDPs are heterogeneous and may comprise functionally distinct subtypes. For precise targeting, it is important to identify specific markers of tumor-promoting TDPs. Regulator of G-protein signaling 5 (RGS5) inhibits TDPs maturation, and its knockout results in enhancing CTL infiltration and extending the survival of tumor-bearing mice [37]. Additionally, TDPs negatively influence CD4 + T cell activation and proliferation in an RGS5-and IL-6-dependent manner [25]. Other markers for tumor-promoting TDPs are still needed.
EN is expressed as a transmembrane glycoprotein specifically in CAFs and TDPs and is an immunotherapy target [38]. However, its function in TDPs requires elucidation. It might be required for TDP maturation and tumor vascularization, as EN KO mice have more small vessels and less large vessels than WT mice [39]. Other tumor-promoting mechanisms of EN + TDPs, such as through CTL infiltration, remain unknown.
Previously, we identified high EN expression in ccRCC which correlated with an immunosuppressive TME, such as increased Treg infiltration and immune checkpoints, and poor prognosis [40,41]; however, the mechanism of this remained largely obscure. Here, we found that high EN expression correlated with low CTL infiltration in ccRCC tissues. In RCC-bearing mice, antibody blockade of EN promoted CTL infiltration and inhibited RCC growth. Analysis of both RNA-seq data of clinical ccRCC samples and scRNA-seq data of ccRCC revealed that EN high TDPs represented an activated state. We then showed that EN knockdown or antibody blockade inhibited pericyte proliferation, migration, and tube formation ability and promoted CTL infiltration through pericytes in vitro. Finally, we showed that combined treatment with anti-EN and anti-PD-1 antibodies inhibited RCC growth in vivo.
In conclusion, our study demonstrated that high EN expression was exhibited in ccRCC and was representative of activated TDPs, which may inhibit CTL infiltration and thereby promote ccRCC progression. EN may provide a target for ccRCC treatment, and combined treatment of anti-EN and anti-PD-1 antibodies may enhance ICB therapy effectiveness in ccRCC. 1 3