Abstract Background The current immunotherapeutic strategies yield limited therapeutic benefits in patients with castration-resistant prostate cancer (CRPC) due to its immunologically “cold” tumor milieu. Pericytes play a pivotal role in facilitating metastatic dissemination and modulating immune response in malignancies. Our investigation is designed to decipher the biological effects of pericytes on CRPC and their interactions with the tumor microenvironment. Methods We leveraged single-cell transcriptomics and immunofluorescence staining to ascertain the presence and spatial distribution of pericytes in prostate cancer (PCa). Subsequently, we thoroughly delineated the phenotypic and functional characteristics of the CRPC-pericytes subpopulation. The clinical and prognostic significance of CRPC-pericytes was assessed by employing several bulk RNA-seq and microarray datasets. Additionally, we examined the association between the abundance of CRPC-pericytes and immunological features in the PCa microenvironment. Furthermore, the RM-1 subcutaneous tumor model was leveraged to assess the synergistic efficacy of platelet-derived growth factor (PDGF) signaling inhibition in conjunction with immunotherapeutic interventions. Results We discerned pericytes according to their marker genes and observed the α-SMA-positive pericytes encircling the vasculature in PCa and adjacent normal tissues. In the CRPC-pericytes subpopulation, a pronounced upregulation of PDGF signaling and angiogenesis was observed, whereas antitumor immunity-related pathways were suppressed. In addition, CRPC-pericytes displayed notably enhanced interactions with endothelial cells, fibroblasts, and myeloid cells, compared to PCa-pericytes. Patients with elevated prevalence of CRPC-pericytes exhibited notably reduced recurrence-free survival and unresponsiveness to immunotherapeutic interventions. Moreover, CRPC-pericytes were positively associated with immunosuppressive properties of the TME. Notably, combinatorial application of PDGFR inhibitor and anti-PD-1 therapy elicited substantial synergistic antitumor effects in murine PCa models. Conclusion Our investigation uncovers a CRPC-pericytes subpopulation implicated in cancer progression and immunosuppression, suggesting that therapies targeting the phenotypic transition of pericytes could act synergistically with immunotherapeutic regimens to improve survival rates in CRPC. Supplementary Information The online version contains supplementary material available at 10.1186/s12935-025-03838-3. Keywords: Castration-resistant prostate cancer, Pericytes, Single-cell transcriptomics, PDGF signaling, Immune microenvironment Introduction In prostate cancer (PCa), activation of the androgen receptor (AR) signaling pathway is a critical determinant in the tumorigenesis and disease progression [[36]1]. Androgen deprivation therapy (ADT) has been acknowledged as a pivotal therapeutic intervention in the clinical management of metastatic or locally advanced PCa [[37]2]. Nevertheless, castration-resistant prostate cancer (CRPC) is a significant challenge in the clinical application of ADT, and patients with CRPC exhibit disease progression and metastatic dissemination despite diminished circulating androgen levels [[38]3]. Recently, immunotherapy has exhibited substantial therapeutic potential across a spectrum of hematological and solid malignancies through augmentation of anti-tumor immune response, which is primarily mediated by T cells [[39]4]. Recently, a series of clinical trials have been conducted to assess the potential of immunotherapeutic interventions for CRPC; however, the findings revealed limited therapeutic efficacy [[40]5–[41]7]. Prostate cancers, classified as immunologically “cold” malignancies, exhibit remarkably diminished density of tumor-infiltrating immune cells and insensitivity to immunotherapeutic agents [[42]8]. Therefore, combinatorial therapeutic strategies integrating immunotherapy with other therapeutic approaches have emerged as innovative research directions in the treatment of CRPC, and preliminary findings from recent clinical trials are encouraging [[43]9]. The burgeoning applications of single-cell transcriptomics and other high-throughput sequencing techniques have demonstrated the complex interplay between the tumor microenvironment (TME) and malignant cells, exerting profound impacts on PCa progression and therapeutic resistance [[44]10]. Therefore, a comprehensive understanding of the TME holds the potential to unveil novel therapeutic targets and pioneer innovative approaches for treatment of PCa. It is well-acknowledged that pericytes function as a critical modulator in vascular biology, influencing stabilization of blood vessels, neovascular growth and maturation, vascular tone control and vessel permeability, mediated through their dynamic communication with endothelial cells [[45]11, [46]12]. Pericytes, recognized as a crucial stromal cell constituent within the TME, have attracted considerable scholarly attention due to their multifaceted roles in oncological processes, including metastasis, angiogenesis, and immune modulation [[47]13]. In a reciprocal manner, the TME can also exert substantial effects on the frequency and functional properties of tumor-associated pericytes. Recent research has uncovered that pericytes with β3-integrin deficiency are capable of augmenting tumorigenesis and the propensity for lung metastases through paracrine secretion of CCL2, which in turn triggers upregulation of MEK1 and ROCK2 signaling cascades within tumor cells [[48]14]. In addition, FAK within pericytes is reported to function as a negative modulator of the Gas6-Axl signaling axis and the production of the secretory protein Cyr61, affecting tumor proliferation and neovascularization via paracrine mechanisms [[49]15]. In nasopharyngeal carcinoma, pericytes have been observed to augment infiltration of macrophages into the tumor milieu and drive their phenotypic switching towards an immunosuppressive state, leading to tumor metastasis [[50]16]. Pericyte-derived CCL5 contributes to insensitivity to the chemotherapeutic agent temozolomide in glioblastoma cells, mediated through upregulation of DNA damage repair signaling [[51]17]. However, the precise modulatory effects of pericytes on the PCa immune microenvironment and their biological roles in CRPC have yet to be fully investigated. In the current research, we conducted an integrative analysis of single-cell transcriptomics data derived from CRPC, primary PCa, and normal prostate tissues to delineate pericyte populations according to the expression levels of the previously characterized marker genes. Immunofluorescence assays were employed to qualitatively validate the expression of the pericyte-specific marker α-SMA and to ascertain the spatial distribution of pericytes within PCa and tumor-adjacent tissues. Our comprehensive single-cell analysis uncovered the remarkable alterations between PCa-pericytes and CRPC-pericytes, particularly a substantial upregulation of platelet-derived growth factor (PDGF) signaling in CRPC-pericytes. In addition, we demonstrated that CRPC-pericytes were intricately associated with unfavorable prognosis, insensitivity to immunotherapeutic agents, and immunosuppressive tumor milieu. Furthermore, employing the RM-1 subcutaneous tumor model, we observed that blockade of PDGF signaling exhibited synergistic effects with anti-PD-1 therapy on PCa. Our results indicate that combinatorial administration of PDGFR inhibitors with immunotherapeutic interventions appears to be a potent therapeutic approach for CRPC. Methods Data collection We accessed the scRNA-seq datasets [52]GSE172316 (6 normal prostate samples), [53]GSE181294 (18 treatment-naive PCa samples), and [54]GSE210358 (13 CRPC samples) from the Gene Expression Omnibus (GEO) database. The TCGA-PRAD RNA-seq datasets employed in our present study are publicly accessible within the UCSC Xena platform ([55]https://xenabrowser.net/). The IMvigor210 immunotherapy data were extracted using the “IMvigor210CoreBiologies” package. We acquired microarray data [56]GSE70769, [57]GSE21034, and [58]GSE32269 from the GEO database. RNA-seq data for the immortalized pericytes were obtained from the [59]GSE277698 dataset. Data preprocessing and annotation of cell populations We employed the “Seurat” package (version 5.1.0) for the processing and analyzing the single-cell transcriptomic data. In the process of filtration, we removed single cells from subsequent analysis if the detected gene count was fewer than 400 or exceeded 5000. Additionally, cells exhibiting a mitochondrial gene fraction surpassing 10% were also filtered out. Then, our single-cell data were subjected to normalization and scaling. We eliminated batch effects across various samples through the “harmony” algorithm [[60]18]. To accomplish dimensionality reduction, principal component analysis (PCA) was employed according to the 2000 genes with the highest degree of variability. We executed unsupervised clustering to distinguish distinct cellular clusters, leveraging a resolution parameter set to 0.5. Furthermore, we conducted the manual annotation of the principal cell populations according to the well-documented canonical marker genes: Epithelial cell (EPCAM, KRT8, KRT18, KRT19), Fibroblast (DCN, COL1A1, LUM), Endothelial cell (PLVAP, VWF, CLDN5), Pericyte (PDGFRB, ACTA2, CSPG4, RGS5), T cell (CD2, CD3D, CD3E), Myeloid cell (LYZ, CD68, CD14, AIF1), B cell (CD79A, CD79B, MS4A1), Mast cell (TPSAB1, MS4A2) [[61]19–[62]21]. Differential expression analysis To discern differentially expressed genes (DEGs) between PCa-pericytes and CRPC-pericytes, the FindAllMarkers function was employed to achieve differential expression analysis (parameters: min.pct = 0.25, only.pos = TRUE). The DEGs were identified according to rigorous criteria: avg_log2FC ≥ 1.5 and adjusted p-value < 0.05. Pathway enrichment analysis To decipher the biological functions associated with DEGs in CRPC-pericytes, we employed the “ClusterProfiler” package to conduct GO and KEGG functional enrichment analyses [[63]22]. In addition, we conducted Gene Set Enrichment Analysis (GSEA) to delineate the dynamic changes in signaling pathways of CRPC-pericytes. The “AUCell” package was leveraged to quantitatively determine the activity levels of signaling pathways in individual cells derived from the scRNA-seq data. Single-cell gene regulatory network analysis The activity levels of distinct transcription factors (TFs) were quantitatively assessed by utilizing the pySCENIC tool (version 0.12.0). We leveraged the GRNBoost algorithm to delineate co-expression networks involving TFs' and their corresponding targets. This was followed by a cis-regulatory motif analysis, and identification of regulons was accomplished through the RcisTarget method. We employed the AUCell analysis to accomplish the quantitative assessment of TFs' activity levels. The comparative analysis of TFs activity scores between PCa-pericytes and CRPC-pericytes was conducted by employing the “limma” package. Cell–cell communication analysis We employed the CellChat tool (version 1.6.1) to delineate the intercellular signaling interactions between pericytes and other cell subpopulations. In addition, we conducted a comparative analysis to discern the differential communication signaling patterns between PCa-pericytes and CRPC-pericytes. Abundance and clinical implications of pericytes and CRPC-pericytes The quantification of pericyte abundance within PCa tissues was accomplished through the application of the “xCell” package (version 1.1.0). In addition, we developed a novel signature for CRPC-pericytes based on a selected collection of marker genes that are specific to this pericyte subpopulation. The abundance of CRPC-pericytes was quantitatively assessed through the “ssGSEA” algorithm. We employed Kaplan-Meier analysis to investigate the relationship between the abundance of pericytes, CRPC-pericytes, and survival outcomes in PCa. The optimal cut-off values that categorized patients into high and low abundance groups were ascertained through the surv_cutpoint function from the “survminer” package. Immune microenvironment and immunotherapy response analysis We conducted ESTIMATE analysis to quantitatively determine the stromal and immune scores within the TME of PCa. Additionally, our investigation leveraged the CIBERSORT and ssGSEA to quantify the intratumoral infiltration levels of distinct immune cells. We applied the Tumor Immune Dysfunction and Exclusion (TIDE) scores to quantitatively evaluate the immunosuppressive status and the potential sensitivity to immunotherapeutic interventions in PCa. Tissue microarray and immunofluorescence The human tissue microarray (HProA150PG02), which comprised formalin-fixed and paraffin-embedded PCa tissues and adjacent non-neoplastic tissues, was acquired from Shanghai Outdo Biotech Company (Shanghai, China). Initially, the tissue microarray was subjected to dewaxing, followed by antigen retrieval exposure to heating sodium citrate buffer (pH 6.0). To minimize non-specific protein interactions, we incubated the tissue microarray with a blocking solution that comprised normal goat serum (1%) and Triton X-100 (0.1%) for a duration of one hour. Incubation of the tissue microarray with primary antibodies was achieved under conditions of 4 °C overnight. In the present investigation, we employed the following primary antibodies for immunofluorescence: α-SMA (1:2000, #19245, Cell Signaling Technology, Danvers, MA, USA), CD31 (1:1000, #3528, Cell Signaling Technology). Following incubation with secondary antibodies, DAPI staining was employed to visualize the nuclei within the tissue microarrays. Cell culture We acquired the RM-1 PCa cell line from Pricella Biotechnology Company (Wuhan, China), and cultivation of the RM-1 cells was accomplished in Dulbecco’s Modified Eagle Medium (DMEM, #11965092, ThermoFisher Scientific, Waltham, MA, USA), which was supplemented with 10% fetal bovine serum and 1% penicillin-streptomycin. The cells were maintained under controlled environmental conditions of 5% CO[2] atmosphere at a stable temperature of 37 °C. Animal experiments We acquired C57BL/6J male mice aged 6–8 weeks from Beijing Vital River Laboratory Animal Technology Co., Ltd. (Beijing, China). The experimental protocols involving the animals in our present investigation obtained approval from the Institutional Animal Care and Use Committee of Shenzhen University Medical School. We subcutaneously inoculated a suspension of 1 × 10^6 syngeneic RM-1 cells in 100 µl sterile PBS into the right dorsal region of mice. We allocated the PCa mouse models into four distinct experimental groups in a randomized manner. The administration of anti-PD-1 (10 mg/kg, #HY-[64]P99144, MedChemExpress, Monmouth Junction, NJ, USA) was achieved via intraperitoneal injection at three-day intervals. The administration of the PDGFR inhibitor CP-673,451 (#HY-12050, MedChemExpress) was performed via oral gavage twice daily, with a total daily dosage of 20 mg/kg. In the group subjected to combinatorial interventions, CP-673,451 was orally administered at a daily dosage of 20 mg/kg (b.i.d), and anti-PD-1 (10 mg/kg) was delivered intraperitoneally at three-day intervals. The following formula was employed to ascertain the tumor volume: V = (tumor length) × (tumor width)^2/2. When accomplishing the therapeutic protocol, we humanely euthanized the mice and harvested tumor samples for subsequent analysis. To assess the expression levels of α-SMA, CD31, and PDGFRβ in the subcutaneous tumor tissues, we conducted immunofluorescence staining by utilizing the following antibodies: α-SMA (1:10000, ab7817, Abcam, Cambridge, UK), CD31 (1:1000, ab182981, Abcam), PDGFRβ (1:100, 3169, Cell Signaling Technology). The staining density was subsequently quantified using ImageJ software. To evaluate the density of CD4^+ and CD8^+ T cells in tumor tissues, we employed the following primary antibodies for immunohistochemistry (IHC): CD4 (1:3000, ab183685, Abcam) and CD8 (1:2000, ab209775, Abcam). Statistical analysis Leveraging R (version 4.2.0), GraphPad Prism (version 8.3.0), and SPSS (version 22), we accomplished statistical analyses and visualization for our results. We employed the Spearman’s rank correlation analysis to ascertain the association between CRPC-pericyte abundance and the expression levels of immunosuppressive molecules. We leveraged a parametric Student’s t-test for the data that were normally distributed, and a nonparametric Wilcoxon rank-sum test for the data that were not normally distributed. The threshold for statistical significance in our present research was defined as a p-value of 0.05. Results Identification of pericytes in CRPC, primary PCa, and normal prostate tissues through single-cell transcriptomics and immunofluorescence When accomplishing the filtration, we acquired a total of 138,424 single cells, including 34,567 cells derived from CRPC tissues, 63,583 cells sourced from primary PCa, and 40,274 cells originated from normal prostate tissues. By employing dimensionality reduction and conducting unsupervised clustering, we successfully obtained 26 distinct clusters (Supplementary Fig. [65]1). Subsequently, we accomplished manual annotation and discerned eight major cell populations according to expression profiles of the canonical marker genes (Fig. [66]1A). Pericytes were precisely identified through pronounced upregulation of the well-established pericyte markers, including ACTA2, PDGFRB, CSPG4, and RGS5 (Fig. [67]1B). We found that the relative frequency of pericytes was approximately 3.87% within normal prostate tissues, 5.29% within primary PCa tissues, and 6.63% within CRPC tissues (Fig. [68]1C-E). Immunofluorescence assays of α-SMA (a pericyte marker) and CD31 (also known as PECAM-1, a vascular endothelial cell marker) were conducted on the tissue microarray, and we observed that α-SMA-positive pericytes were encircling CD31-positive blood vessels in both PCa and adjacent normal tissues (Fig. [69]1F). In addition, we found that the ratio of α-SMA^+ pericyte-positive blood vessels was significantly higher in primary PCa tissues than in adjacent normal tissues (Supplementary Fig. [70]2A). Moreover, tumors with high Gleason grade (Gleason score > 7) exhibited higher ratio of α-SMA^+ pericyte-positive blood vessels compared to those with median (Gleason score = 7) or low (Gleason score < 7) grade (Supplementary Figs. [71]2B–C). These findings suggest that pericytes are closely associated with the aggressiveness of prostate cancer. However, a limitation of our study is the absence of human CRPC samples. In the future, it is essential to obtain CRPC samples and compare the ratio of pericyte-positive blood vessels between primary PCa and CRPC. Fig. 1. [72]Fig. 1 [73]Open in a new tab Identification of pericytes in CRPC, primary PCa, and normal prostate tissues via single-cell transcriptomic analysis and immunofluorescence. (A) t-SNE plot of 138,424 single cells originating from 13 CRPC samples, 18 treatment-naive primary PCa samples, and 6 normal prostate samples, color-coded by major cell populations. (B) Dot plot depicting the mean expression levels of marker genes in different cell populations. (C-E) Pie charts displaying the proportional distribution of eight different cell populations in normal prostate (C), primary PCa (D), and CRPC tissues (E). (F) Representative immunofluorescent images of CD31 and α-SMA in the PCa and normal adjacent prostate specimens on a tissue microarray (scale bar: 50 μm, scale bar insets: 20 μm) Pericyte abundance correlates with clinical prognosis and TME in PCa By employing xCell analysis, we quantitatively determined the abundance of pericytes in PCa tissues. PCa patients with high Gleason scores (greater than 7) exhibited markedly increased abundance of pericytes compared to those with intermediate or low Gleason scores (equal to 7 or lower) in the TCGA-PRAD and [74]GSE70769 cohorts (Fig. [75]2A-B). Kaplan-Meier analysis was conducted, and we observed that patients with elevated abundance of pericytes exhibited a markedly diminished recurrence-free survival (RFS) compared to those with low pericyte abundance in the TCGA cohort (Fig. [76]2C). This association between pericyte abundance and RFS in PCa was further corroborated in the [77]GSE70769 and [78]GSE21034 cohorts (Fig. [79]2D-E). Leveraging ssGSEA analysis, we revealed a pronounced upregulation of angiogenesis, epithelial-mesenchymal transition, NOTCH, hypoxia, and Wnt/β-catenin signaling pathways in patients with elevated pericyte abundance compared to those with lower abundance, whereas the signaling pathways associated with spermatogenesis, protein secretion, androgen response, unfolded protein response, and peroxisome function were notably downregulated in patients with high pericyte abundance (Fig. [80]2F). We found that patients with high pericyte abundance exhibited a marked elevation in the infiltration of regulatory T cells (Tregs), myeloid-derived suppressor cells (MDSCs) and macrophages through the ssGSEA algorithm (Fig. [81]2G). Furthermore, we conducted CIBERSORT analysis and observed that patients with elevated pericyte abundance had a remarkably increased proportion of macrophages and M2-like macrophages compared to those with lower pericyte abundance, but no significant difference in the proportion of M1-like macrophages was found (Fig. [82]2H). Fig. 2. [83]Fig. 2 [84]Open in a new tab The clinical significance of pericytes in PCa. (A-B) Boxplots comparing pericyte scores across three distinct Gleason score-stratified patient groups in the TCGA-PRAD (A) and [85]GSE70769 (B) cohorts. (C-E) Kaplan–Meier curves delineating the differential recurrence-free survival rates in patients stratified by high and low pericyte scores in the TCGA-PRAD (C), [86]GSE70769 (D), and [87]GSE21034 (E) cohorts. (F) Bar plot illustrating the pronouncedly activated signaling pathways in two distinct groups categorized by high and low pericyte scores, with pathway activity scores quantified by ssGSEA. The statistical comparison of pathway activity scores between two distinct groups was accomplished by leveraging the “limma” package, with t-values denoting the statistical measures derived from linear model fitting. (G-H) Boxplots comparing the abundance of tumor-infiltrating immune cells between two distinct groups, quantified through ssGSEA (G) and CIBERSORT (H) algorithms. The assessment of statistical significance was accomplished by utilizing a two-tailed Wilcoxon test. (*p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001.) Transcriptional and functional characteristics of CRPC-pericytes Our analysis discerned a total of 5,764 tumor-associated pericytes, which were categorized into two distinct subpopulations: CRPC-pericytes (2,292 cells) and PCa-pericytes (3,472 cells) (Fig. [88]3A). Through a comparative analysis between two different pericyte subsets, we revealed that 287 DEGs were pronouncedly upregulated in CRPC-pericytes, whereas 70 DEGs were markedly increased in PCa-pericytes (Fig. [89]3B; Supplementary Table [90]1). GO enrichment analysis results uncovered that the CRPC-pericytes DEGs were implicated in the cancer-related biological processes, including extracellular matrix organization, epithelial cell proliferation, TGF-β response, Wnt signaling, and angiogenesis (Fig. [91]3C). Our KEGG analysis indicated that the CRPC-pericytes DEGs were markedly enriched in pathways correlated to proteoglycans in cancer, MAPK signaling, ECM-receptor interaction, PD-1/PD-L1 interaction, EGFR inhibitor resistance, and VEGF signaling (Fig. [92]3D). GSEA results demonstrated that extracellular matrix organization and PDGF signaling pathways were notably upregulated in CRPC-pericytes, whereas cancer immunotherapy by PD-1 blockade and NK cell interactions exhibited substantial downregulation in CRPC-pericytes (Fig. [93]3E). Employing the AUCell algorithm, our results revealed a pronounced activation of PDGF signaling, tyrosine kinase receptor signaling, pathways in cancer and angiogenesis in CRPC-pericytes compared to PCa-pericytes (Fig. [94]3F-G; Supplementary Fig. [95]3). To validate the effects of PDGF signaling activation on pericytes in vitro, we conducted a comprehensive analysis of the impact of recombinant human PDGF-BB on immortalized pericytes using RNA-seq data ([96]GSE277698). Principal component analysis (PCA) revealed that PDGF-BB treatment significantly altered the transcriptome profiles of pericytes (Supplementary Fig. [97]4A). Specifically, 91 genes were significantly upregulated and 77 genes were downregulated in PDGF-BB-treated pericytes compared to the control group (Supplementary Fig. [98]4B). Gene Set Enrichment Analysis (GSEA) demonstrated that signaling pathways associated with the epithelial-mesenchymal transition (EMT), angiogenesis, IL6-JAK-STAT3 signaling, PDGF signaling, and tyrosine kinase receptor signaling were significantly upregulated in pericytes treated with PDGF-BB (Supplementary Fig. [99]4C). Additionally, PDGF-BB treatment significantly increased the activity scores of angiogenesis, PDGF signaling, IL6-JAK-STAT3 signaling, and tyrosine kinase receptor signaling in pericytes, as determined by Gene Set Variation Analysis (GSVA) (Supplementary Fig. [100]4D-E). Fig. 3. [101]Fig. 3 [102]Open in a new tab Phenotypic and functional characteristics of CRPC-pericytes. (A) t-SNE plot displaying 5,764 tumor-associated pericytes from primary PCa and CRPC samples. (B) Scatter plots visualizing the DEGs between CRPC-pericytes and PCa-pericytes according to the average log-fold change and the percentage difference (delta percent) in gene expression. (C) Dot plot illustrating the significant enrichment of GO biological processes for upregulated genes in CRPC-pericytes. (D) Bar plot illustrating the significant enrichment of KEGG functional pathways for upregulated genes in CRPC-pericytes. (E) GSEA uncovering the markedly upregulated and downregulated signaling pathways in CRPC-pericytes. (F) Bar plot illustrating the pronouncedly activated signaling pathways in CRPC-pericytes and PCa-pericytes, with activity scores quantified by AUCell analysis. The statistical comparison of activity scores between two distinct pericyte subpopulations was accomplished by leveraging the “limma” package, with t-values denoting the statistical measures derived from linear model fitting. (G) Boxplots displaying marked upregulation of PDGF signaling, angiogenesis, and cancer-related pathways through AUCell analysis. The assessment of statistical significance was accomplished by utilizing a two-tailed Wilcoxon test. (*p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001.) Transcription factors analysis in CRPC-pericytes and PCa-pericytes Leveraging the pySCENIC tool, we conducted a thorough analysis to decipher the regulatory mechanisms of TFs in CRPC-pericytes and PCa-pericytes (Supplementary Table [103]2). In our analysis, EBF1, SOX6, HAND2, NFATC4, and ALX3 were the most specific TFs in tumor-associated pericytes, as determined by their high regulon specificity scores (Fig. [104]4A). We observed that NR2F2, HIC1, YY1, HOXC9, HAND2, GATA6, PRRX2, TBX15, ALX3, and HOXC6 exhibited remarkable upregulation of TFs' activity in CRPC-pericytes compared to PCa-pericytes, whereas DDIT3, NR4A1, ATF4, FOXP1, CHD2, SOX6, MYC, JUNB, BCLAF1, and FOSB exhibited significant activation in PCa-pericytes (Fig. [105]4B-C). Fig. 4. [106]Fig. 4 [107]Open in a new tab Transcription factor analysis and intercellular communication of CRPC-pericytes and PCa-pericytes. (A) Scatter plot illustrating the specificity of regulons in tumor-associated pericytes, as measured by the regulon specificity score. (B) Heatmap displaying the top 10 most activated transcription factors in CRPC-pericytes and PCa-pericytes. (C) Violin plots comparing activity scores of distinct TFs between CRPC-pericytes and PCa-pericytes. (D-E) Dot plots illustrating the comparative analysis of communication probabilities for major ligand-receptor pairs between CRPC-pericytes and PCa-pericytes in interactions with other cell populations. The assessment of statistical significance was accomplished by utilizing a two-tailed Wilcoxon test. (*p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001.) Intercellular communication between tumor-associated pericytes and other cellular components In comparison to primary PCa, CRPC exhibited a pronounced escalation in both the quantity and intensity of intercellular interactions (Supplementary Fig. [108]5). Through CellChat analysis, we revealed a substantial upregulation of VEGFB, FN1, and JAG1 intercellular signaling between CRPC-pericytes and endothelial cells (Fig. [109]4D). In addition, CRPC-pericytes displayed markedly elevated levels of TGFB, JAG1, GAS6, and FN1 signaling during their interactions with fibroblasts and myeloid cells (Fig. [110]4D). As a target of endothelial cells and fibroblasts in the intercellular communication network, CRPC-pericytes exhibited marked upregulation of TNFRSF12, PDGFRB, JAM3, and NOTCH3 ligand-receptor signaling pathways, compared to PCa-pericytes (Fig. [111]4E). Furthermore, the SPP1-related signaling pathways were pronouncedly elevated during the interactions of epithelial cells and myeloid cells with CRPC-pericytes (Fig. [112]4E). The abundance of CRPC-pericytes was associated with prognosis and immunotherapy sensitivity We quantitatively assessed the abundance of CRPC-pericytes through the ssGSEA algorithm based on the CRPC-pericytes gene signature (Supplementary Table [113]3). Employing Kaplan-Meier survival analysis, we observed that patients with higher CRPC-pericytes abundance displayed a notably unfavorable RFS than those with lower CRPC-pericytes abundance in the TCGA-PRAD, [114]GSE21034, and [115]GSE70769 cohorts (Fig. [116]5A-C). Individuals with elevated abundance of CRPC-pericytes displayed a markedly unfavorable overall survival and substantially diminished sensitivity to immunotherapy in the IMvigor210 cohort (Fig. [117]5D-E). Moreover, we observed that patients with higher CRPC-pericytes abundance exhibited a pronounced increase in TIDE scores, suggesting that CRPC-pericytes could play a critical role in immunosuppression and insensitivity to immunotherapeutic agents in PCa (Fig. [118]5F). Furthermore, our results demonstrated that metastatic CRPC had remarkably elevated abundance of CRPC-pericytes in comparison with localized PCa (Fig. [119]5G). Fig. 5. [120]Fig. 5 [121]Open in a new tab Impact of CRPC-pericytes on clinical prognosis and immunotherapy responsiveness. (A-C) Kaplan–Meier curves delineating the differential recurrence-free survival rates in patients with high versus low abundance of CRPC-pericytes in the TCGA-PRAD (A), [122]GSE21034 (B), and [123]GSE70769 (C) cohorts. (D) Kaplan–Meier curves delineating the differential overall survival in patients with high versus low abundance of CRPC-pericytes in the IMvigor210 cohort. (E) Bar plot comparing the objective response rate to immunotherapeutic agents between patients with high and low abundance of CRPC-pericytes in the IMvigor210 cohort. The assessment of statistical significance was accomplished by employing the Chi-square test. (F) Boxplot comparing the tumor immune dysfunction and exclusion (TIDE) scores between patients with high and low CRPC-pericytes. (G) Boxplot comparing the abundance of CRPC-pericytes between localized PCa and metastatic CRPC within the [124]GSE32269 cohort. The assessment of statistical significance was accomplished by utilizing a two-tailed Wilcoxon test. (*p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001.) The correlation between CRPC-pericytes abundance and TME in PCa By employing ESTIMATE analysis, we revealed that patients with elevated CRPC-pericytes abundance had pronouncedly increased stromal scores and immune scores (Fig. [125]6A). The CIBERSORT analysis demonstrated that patients with elevated CRPC-pericytes abundance exhibited a significantly increased frequency of M2-like macrophages and Tregs, whereas the frequency of M1-like macrophages was markedly diminished, compared to those with lower CRPC-pericytes abundance (Fig. [126]6B). In addition, ssGSEA results revealed a substantial increase in the infiltration levels of MDSCs, Tregs, and macrophages in patients with higher abundance of CRPC-pericytes than those with lower abundance (Fig. [127]6C). Furthermore, we observed that patients with elevated CRPC-pericytes abundance displayed a pronounced upregulation of the immunosuppressive molecules PDCD1, LAG3, CTLA4, TIGIT, and HAVCR2 (Fig. [128]6D). The abundance scores of CRPC-pericytes were positively correlated with the expression levels of these five immunosuppressive molecules (Fig. [129]6E). Fig. 6. [130]Fig. 6 [131]Open in a new tab The relationship between CRPC-pericytes abundance and immunosuppressive characteristics of the TME. (A) Boxplots comparing immune score and stromal scores between patients with high and low abundance of CRPC-pericytes. (B-C) Boxplots comparing the infiltration levels of immune cells between patients with high and low abundance of CRPC-pericytes, quantified through CIBERSORT (B) and ssGSEA (C) algorithms. (D) Boxplots comparing the expression levels of immune checkpoint molecules between patients with high and low CRPC-pericytes abundance. (E) Scatter plots displaying a markedly positive correlation between CRPC-pericytes abundance and expression levels of immune checkpoint molecules, as determined by Spearman’s rank correlation coefficient. The assessment of statistical significance was accomplished by utilizing a two-tailed Wilcoxon test. (*p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001.) Synergistic antitumor effects of PDGFR inhibitor and immunotherapy We employed GSVA to assess the activity of PDGF signaling in TCGA-PRAD bulk RNA-seq data. Importantly, our results uncovered a pronounced activation of PDGF signaling in patients with elevated abundance of CRPC-pericytes in comparison with those with lower abundance (Fig. [132]7A). To assess the association between PDGF signaling activation and the immunosuppressive TME in prostate cancer, we conducted Spearman correlation analyses between the activity scores of the PDGF signaling pathway and the infiltration levels of immune cells in the TCGA-PRAD cohort. Our results revealed a significantly negative correlation between the activity of PDGF signaling and the infiltration of CD8^+ T cells and follicular helper T cells (a subset of CD4^+ T cells) in prostate cancer (Supplementary Fig. [133]6A-B). In addition, the activity of the PDGF signaling pathway was positively correlated with the expression of several immune checkpoint molecules, including CD274, PDCD1, CTLA4, LAG3, TIGIT, and HAVCR2 (Supplementary Fig. [134]6C). Therefore, we leveraged a subcutaneous PCa model to examine the impact of PDGF/PDGFR signaling blockade on tumor progression and its potential synergistic effects with immunotherapy (Fig. [135]7B). The results demonstrated that the monotherapeutic administration of PDGFR inhibitor or anti-PD-1 antibody yielded comparable effects on tumor growth suppression, whereas combinatorial therapeutic approaches involving PDGFR inhibitor and anti-PD-1 elicited a substantially synergistic antitumor response (Fig. [136]7C-E). Given that previous research has uncovered that PDGFR inhibition markedly reduces the abundance of CD31^+ endothelial cells and α-SMA^+ perivascular cells in colorectal cancer [[137]23], we conducted immunofluorescence staining to assess the expression of α-SMA and CD31 in tumor tissues across distinct treatment groups. Compared to control and anti-PD-1 monotherapy groups, treatment with either PDGFR inhibitor monotherapy or the combination of PDGFR inhibitor and anti-PD-1 therapy significantly decreased the expression levels of α-SMA and CD31 (Fig. [138]7F-H). To further confirm the effects of the PDGFR inhibitor, we utilized immunofluorescence staining to assess the protein expression levels of PDGFRβ in tumor tissues across different treatment groups. Our results revealed that both PDGFR inhibitor monotherapy and the combination of PDGFR inhibitor with anti-PD-1 therapy significantly reduced protein expression levels of PDGFRβ (Supplementary Fig. [139]7A-B). This finding is consistent with the observed effects of the PDGFRβ inhibitor on α-SMA and CD31 expression. These results suggest a substantial decrease in the number of pericytes and endothelial cells within the tumor milieu following blockade of the PDGF/PDGFR signaling pathways. Fig. 7. [140]Fig. 7 [141]Open in a new tab Synergistic antitumor effects of PDGFR inhibitor and immunotherapeutic interventions. (A) Boxplots comparing the activity scores of PDGF signaling between patients with high and low abundance of CRPC-pericytes. The assessment of statistical significance was accomplished by utilizing a two-tailed Wilcoxon test. (B) Schematic plot displaying experimental protocols and therapeutic intervention strategies. (C) Gross appearance of RM-1 subcutaneous tumors harvested from murine models. (D-E) Tumor growth curves (D) and tumor weight (E) in murine models with subcutaneous RM-1 allografts across four treatment regimens: control (n = 5), anti-PD-1 monotherapy (n = 5), PDGFR inhibitor monotherapy (n = 5), and combined PDGFR inhibitor and anti-PD-1 therapy (n = 5). (F) Representative images of immunofluorescence staining of α-SMA and CD31 in tumor tissues. Scale bars: 50 μm. (G-H) Quantification and comparison of the α-SMA and CD31 density in RM-1 subcutaneous tumors among four treatment groups (n = 5). The assessment of statistical significance was accomplished by utilizing a two-tailed Student’s t-test. (*p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001.) To assess microvessel density (MVD) and tumor vascular invasiveness across different therapeutic groups, we employed immunohistochemical staining for CD31 to identify blood vessels. MVD was quantified by counting individual vessels at 200× magnification according to previous studies [[142]24–[143]26]. Our results demonstrated that treatment with either PDGFR inhibitor monotherapy or the combination of PDGFR inhibitor and anti-PD-1 therapy significantly reduced MVD compared to the control and anti-PD-1 monotherapy groups (Supplementary Fig. [144]8A-B). Tumor vascular invasion is typically identified by the presence of tumor cells within the lumen of CD31-positive blood vessels [[145]26, [146]27]. Utilizing Fisher’s exact test, we found no statistically significant differences in the proportions of vascular invasion across four distinct therapeutic groups (Supplementary Fig. [147]8C-D). To examine the effects of PDGFR inhibitor and anti-PD-1 therapy on immune infiltration in prostate cancer, we performed IHC staining to assess the infiltration of CD4^+ T cells and CD8^+ T cells. Our results demonstrated that monotherapy with either anti-PD-1 or PDGFR inhibitor significantly increased the number of CD4^+ and CD8^+ T cells compared to the control group (Supplementary Fig. [148]9A-C). Moreover, the combination therapy group exhibited a significantly higher number of CD4^+ and CD8^+ T cells compared to the anti-PD-1 or PDGFR inhibitor monotherapy groups (Supplementary Fig. [149]9A-C). These findings suggest that the inhibition of PDGFR signaling augments the anti-tumor immune response and enhances the efficacy of anti-PD-1 immunotherapy in prostate cancer. Future research is essential to elucidate how PDGFR inhibitors affect the expression of immune checkpoints and the density of inhibitory immune cells in prostate cancer, as well as to uncover the underlying mechanisms. Discussion Pericytes are important cellular components of the TME in prostate cancer and play a pivotal role in angiogenesis. Pericyte density is significantly elevated in high-stage or high-Gleason grade PCa compared to lower-grade tumors and normal tissues [[150]28]. Additionally, the central regions of tumors exhibit significantly higher pericyte density and distinct pericyte phenotypes compared with peripheral regions [[151]28]. As prostate cancer progresses from prostatic intraepithelial neoplasia (PIN) to poorly differentiated adenocarcinomas, pericytes display an abnormal phenotype and weaker attachment to the abluminal surface of vascular endothelial cells [[152]29]. Knockout of NG2 in pericytes has been shown to substantially reduce neovascularization and lymphangiogenesis in prostate cancer (PCa) mouse models [[153]30]. Conversely, a previous study has indicated that reduced pericyte coverage is closely associated with enhanced tumor invasion and the development of androgen-independent prostate cancer [[154]31]. These findings highlight the phenotypic and functional heterogeneity of pericytes across different pathological stages and conditions in prostate cancer. Therefore, it is essential to elucidate the biological characteristics of pericytes in PCa and CRPC. Recently, a burgeoning body of research has revealed the pivotal role of pericytes in modulating malignant progression and metastatic dissemination of cancers, particularly through paracrine interactions with neoplastic cells or endothelial cells [[155]32–[156]35]. However, the current scholarly efforts have not thoroughly elucidated the impact of tumor-associated pericytes on clinical outcomes and immunological characteristics of the TME in PCa. In this research, we have successfully ascertained the presence and spatial positioning of tumor-associated pericytes in the TME of PCa, and assessed their clinical significance. An increased prevalence of tumor-associated pericytes has been found to positively correlate with the escalated risk of disease progression and the propensity for biochemical recurrence in PCa. In line with our findings, it has been reported that patients with elevated pericyte signature scores display substantially diminished RFS in ovarian cancer [[157]33]. Within the context of colorectal cancer, the density of tumor-infiltrating pericytes escalates with the tumor pathological staging, and elevated pericyte infiltration is predictive of unfavorable clinical outcomes [[158]36]. In our current investigation, single-cell transcriptomic analyses have meticulously delineated the functional and phenotypic heterogeneity among tumor-associated pericytes in PCa and CRPC, highlighting the paramount biological roles of the CRPC-pericytes subpopulation in disease progression. It is noteworthy that PDGF signaling and angiogenesis were prominently activated in the CRPC-pericytes subpopulation, whereas signaling pathways implicated in immunotherapy and T cell activation were considerably suppressed. Similarly, a recent scholarly endeavor, leveraging single-cell transcriptomics, discerned a previously unreported pericyte subpopulation with elevated transcriptional activity of TCF21, which is capable of modulating ECM remodeling in the tumor milieu, eventually facilitating liver metastatic dissemination and contributing to unfavorable survival rates in colorectal cancer [[159]37]. Tumor-associated pericyte-derived TCAF2 notably augments the paracrine secretion of Wnt5a and subsequently activates Wnt5a/STAT3 signaling cascades in cancer cells, which in turn induces EMT and promotes metastatic spread in CRC [[160]38]. In renal cell carcinoma, single-cell analysis has uncovered a particular subpopulation of pericytes co-expressing PDGFR-β and GPR91, which play a paramount role in tumor progression and unresponsiveness to tyrosine kinase inhibitors, mediated through upregulated secretion of methionine and maintenance of cancer stemness [[161]39]. Similarly, our data indicate that the CRPC-pericytes subpopulation is intricately implicated in the risk of relapse and the efficacy of immunotherapeutic agents in PCa. A recent investigation has revealed that pericytes with elevated RGS5 expression contribute to resistance to adoptive immunotherapy in melanoma through inducing a diminished frequency of T cells and an escalated proportion of M2-like macrophages in the TME [[162]40]. Pericyte stem cells expressing CD106 have been documented to induce an immunosuppressive TME, which is characterized by augmented recruitment of MDSCs and diminished infiltration of dendritic cells and macrophages, eventually facilitating insensitivity to immunotherapeutic interventions in pancreatic cancer [[163]41]. Intercellular communication analysis has revealed markedly increased interactions between CRPC-pericytes and other cellular constituents in the TME, encompassing endothelial cells, fibroblasts, and myeloid cells. Recently, tumor-associated pericytes have been reported to recruit and remarkably activate Axl signaling cascades in endothelial progenitor cells via extracellular vesicles derived Gas6, leading to neovascularization and resistance to antiangiogenic interventions in CRC [[164]42]. In addition, tumor-associated pericytes, characterized by upregulated glycolytic activity, play a pivotal role in vascular dysfunction through modulating pericyte contractility and the tube formation capabilities of endothelial cells, eventually affecting the delivery and efficacy of chemotherapeutic agents [[165]43]. Accumulating evidence supports the notion that pericytes play a critical role in the modulation of immune response through their secretion of a spectrum of cytokines [[166]44]. Our data demonstrated that CRPC-pericytes were intricately associated with markedly elevated density of immunosuppressive cells and a pronounced upregulation of immune checkpoint molecules. The activation of PDGFRβ signaling cascades in tumor-associated pericytes has been observed to enhance the metastatic dissemination of cancer through upregulating the production of IL-33, which in turn facilitates macrophage recruitment and elicits a shift towards the M2 phenotype [[167]45]. In gliomas, CD90-positive pericytes have been implicated in the suppression of T cell-mediated immune responses and the reduction of CD8^+ T cell density, mediated through markedly increased production of TGF-β, HGF, and HLA-G [[168]46]. Our data reveal that monotherapy with PDGFR inhibitors effectively restrains PCa progression. In addition, the combination of PDGFR inhibitor and anti-PD-1 immunotherapeutic interventions displays potent synergistic effects. In line with our findings, a recent study demonstrated that targeted inhibition of PDGFR signaling can effectively reshape the immunosuppressive tumor milieu and augment antitumor immune responses, potentiating the responsiveness to immunotherapeutic agents [[169]47]. The blockade of PDGFR signaling cascades has been demonstrated to facilitate overcoming resistance of ovarian cancer to chemotherapeutic agents through the downregulation of HIF-1α in tumor cells [[170]48]. Furthermore, our results indicate that blockade of PDGF/PDGFR signaling substantially decreases the density of endothelial cells and pericytes, thereby attenuating angiogenesis and impeding tumor progression. Consistently, a previous study has demonstrated that PDGFRβ antibody therapy can notably reduce the number of endothelial cells and perivascular cells in colorectal cancer [[171]23]. The PDGFR inhibitor (CP-673451) effectively suppressed tumor proliferation across various cancer xenograft models and attenuated PDGF-BB-induced angiogenesis [[172]49]. The current research encounters a series of limitations that remain to be addressed in further investigation. Firstly, it is critical to culture primary tumor-associated pericytes isolated from PCa and CRPC tissues to decipher the exact mechanisms underlying the interplay between CRPC-pericytes and the TME in the future. Additionally, the application of conditional transgenic mice is essential for the in vivo assessment of the effects exerted by CRPC-pericytes on disease progression and sensitivity to immunotherapeutic interventions in CRPC. Furthermore, we need additional prospective clinical cohorts to ascertain the prognostic significance of CRPC-pericytes. In conclusion, leveraging single-cell transcriptomics, our investigation has discerned a particular CRPC-pericyte subset that is implicated in relapse and unresponsiveness to immunotherapeutic interventions. Notably, CRPC-pericytes exhibiting upregulated PDGF signaling are remarkably correlated with the immunosuppressive characteristics of the tumor milieu in PCa. Our data indicate that the synergistic application of therapies targeting CRPC-pericytes, in combination with immunotherapeutic interventions, could constitute a potentially effective therapeutic regimen for improving clinical outcomes in CRPC. Electronic supplementary material Below is the link to the electronic supplementary material. [173]12935_2025_3838_MOESM1_ESM.pdf^ (1.7MB, pdf) Supplementary Material 1: Supplementary Figure 1. Unsupervised clustering results of single-cell transcriptome data. (A) t-SNE plot of 138,424 single cells, color-coded by 26 distinct clusters. (B) t-SNE plot of 138,424 single cells, color-coded by three groups [174]12935_2025_3838_MOESM2_ESM.pdf^ (37.9MB, pdf) Supplementary Material 2: Supplementary Figure 2. Association between the ratio of pericyte-positive blood vessels and clinical-pathological characteristics in prostate cancer. (A) Comparison of the ratio of α-SMA+ pericyte-positive blood vessels in prostate cancer tissues versus normal adjacent tissues. (B) Representative immunofluorescent images of CD31 and α-SMA in prostate cancer tissues across three Gleason grade groups (scale bar: 50 μm). (C) Comparison of the ratio of α-SMA+ pericyte-positive blood vessels among three Gleason grade groups in prostate cancer. The assessment of statistical significance was accomplished by utilizing a two-tailed Student’s t-test. (*p <0.05; **p < 0.01.) [175]12935_2025_3838_MOESM3_ESM.pdf^ (1,020.2KB, pdf) Supplementary Material 3: Supplementary Figure 3. Remarkably activated signaling pathways in CRPC-pericytes, as evaluated through AUCell analysis [176]12935_2025_3838_MOESM4_ESM.pdf^ (1.3MB, pdf) Supplementary Material 4: Supplementary Figure 4. The effects of PDGF signaling activation on human immortalized pericytes. (A) PCA scatter plot comparing transcriptome profiles of immortalized pericytes treated with PDGF-BB versus control pericytes. (B) Volcano plot depicting differentially expressed genes (DEGs) between immortalized pericytes treated with PDGF-BB and vehicle. (C) GSEA analysis highlighting significantly upregulated signaling pathways in immortalized pericytes treated with PDGF-BB. (D) Heatmap showing the activity levels of signaling pathways in immortalized pericytes, as measured by GSVA. (E) Comparison of GSVA activity scores for signaling pathways between immortalized pericytes treated with PDGF-BB and vehicle [177]12935_2025_3838_MOESM5_ESM.pdf^ (1.3MB, pdf) Supplementary Material 5: Comparative analyses of intercellular interactions between primary PCa and CRPC. (A) The number and strength of intercellular communications in primary PCa and CRPC. (B) Network plots displaying the number of intercellular interactions between different cell populations in primary PCa and CRPC [178]12935_2025_3838_MOESM6_ESM.pdf^ (3MB, pdf) Supplementary Material 6: Supplementary Figure 6. Association between PDGF signaling activation and the immunosuppressive tumor microenvironment in prostate cancer. (A-B) Spearman's rank correlation analysis revealing significantly negative correlations between the activity score of the PDGF signaling pathway and the infiltration of CD8+ T cells (A) and follicular helper T cells (B). (C) Spearman's rank correlation analysis displaying a markedly positive correlation between the activity scores of the PDGF signaling pathway and the expression levels of immune checkpoint molecules [179]12935_2025_3838_MOESM7_ESM.pdf^ (19.6MB, pdf) Supplementary Material 7: Supplementary Figure 7. Expression levels of PDGFR? in tumor tissues across four treatment groups. (A) Representative immunofluorescent images of PDGFRβ in tumor tissues from four distinct treatment groups (scale bar: 50 μm). (B) Quantification and comparison of the PDGFRβ density in tumor tissues among four different treatment groups. Statistical significance was determined using a two-tailed Student’s t-test. (*p < 0.05, **p < 0.01, ***p < 0.001) [180]12935_2025_3838_MOESM8_ESM.pdf^ (45.7MB, pdf) Supplementary Material 8: Supplementary Figure 8. Microvessel density and tumor vascular invasion across four treatment groups. (A) Representative images of IHC staining for CD31 in four distinct treatment groups. Microvessel density (MVD) was evaluated using CD31 immunostaining. (B) Comparison of MVD among the four treatment groups. (C) Representative images of tumor tissues with positive and negative vascular invasion. (D) Comparison of the proportion of positive vascular invasion in tumor tissues across the four groups, analyzed using Fisher’s exact test. Statistical significance was determined using a two-tailed Student’s t-test. (*p < 0.05, **p < 0.01, ***p < 0.001) [181]12935_2025_3838_MOESM9_ESM.pdf^ (42.5MB, pdf) Supplementary Material 9: Supplementary Figure 9. Infiltration of CD4+ and CD8+ T cells in tumor tissues across four treatment groups. (A) Representative images of IHC staining for CD4 and CD8 in four different therapeutic groups (Scale bars: 100 μm). (B-C) Comparison of CD4+ T cell density (B) and CD8+ T cell density (C) across four different treatment groups. Statistical significance was assessed using a two-tailed Student’s t test. (*p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001.) [182]12935_2025_3838_MOESM10_ESM.csv^ (23.9KB, csv) Supplementary Material 10: Supplementary Table 1. The DEGs notably upregulated in CRPC-pericytes and PCa-pericytes. [183]12935_2025_3838_MOESM11_ESM.csv^ (2.1MB, csv) Supplementary Material 11: Supplementary Table 2. The transcription factors and their target genes, identified by pySCENIC analysis. [184]12935_2025_3838_MOESM12_ESM.csv^ (1.5KB, csv) Supplementary Material 12: Supplementary Table 3. The signature genes of CRPC-pericytes. Acknowledgements