Clinical data analysis and single-cell transcriptome atlas of primary RCC

After analysis of 1102 RCC patients receiving radical nephrectomy at the Sun Yat-sen University Cancer Center (SYSUCC), we found the incidence ratio of male to female RCC was about 1.8:1, and the pathological stage of male RCC patients was significantly higher than females (Fig. 1A). After a meta-analysis of five randomized controlled trials of immunotherapy in RCC, we found that male RCC patients had increased benefits from immunotherapy, especially regarding progression-free survival (PFS) (Fig. 1B and Additional file 2: Fig. S1A).

Fig. 1

scRNA-seq profiling of TME of male and female RCC patients. A Sex difference of incidence and clinical stages of 1102 RCC patients who underwent nephrectomy from 1999 to 2020 at SYSUCC. B Meta-analysis of sex-specific pooled hazard ratios in various immunotherapy RCTs. C A schematic representation of scRNA-seq profiling of tumors and adjacent normal kidneys in male and female patients with RCC. D The UMAP plot shows the annotation and color codes for different cell types in tumors and adjacent normal kidney ecosystems. E Heatmap showing the expression of marker genes in the indicated cell types. F, G Histograms and boxplots illustrating the percentage of cell types in different groups. HR: hazard ratio; OS: overall survival; PFS: progression-free survival; RCC: renal cell carcinoma; RCTs: randomized controlled trials; scRNA-seq: single-cell RNA sequencing; SYSUCC: Sun Yat-sen University Cancer Center; UMAP: uniform manifold approximation and projection

We summarized the currently published scRNA-seq of primary RCC from four studies [8, 28,29,30]. After screening, 18 samples (9 tumor and adjacent normal kidney samples) from 9 males and 18 samples (10 tumor samples and 8 adjacent normal kidney samples) from 10 females were included in this study (Additional file 2: Fig. S1B and Additional file 3: Table S1B). The female and male patients included in this study had relatively similar clinical stages (Fig. 1C). The expression of sex-related genes (XIST, RPS4Y1) was used to identify the gender of each patient in this study (Additional file 2: Fig. S1C).

After quality control and removal of the batch effect between samples, 220,156 single cells were clustered and identified into 8 major cell lineages using the uniform manifold approximation and projection (UMAP) method (Fig. 1D and Additional file 2: Fig. S2A, B): epithelial cell, endothelial cell, myeloid cell, fibroblast, T cell, B cell, natural killer cell (NK cell), and proliferating cell. The heatmap of marker genes showed the homogeneity of each major cell lineage (Fig. 1E). Interestingly, compared to the TME of females, a higher infiltration level of T-cells was found in male TME (p < 0.05, Fig. 1G). Each tumor had a relatively uniform cell lineage composition (Additional file 2: Fig. S2C), while the adjacent normal kidney differed widely (Additional file 2: Fig. S2D).

Tumor cells from males were more malignant than females

The above-identified epithelial cells were re-clustered and identified into five cell types: proximal tubules, the loop of Henle, distal tubules, collecting duct, and malignant cells (Fig. 2A, B). The histogram illustrates that the malignant cells were concentrated in the tumor samples (Fig. 2C). The gene set variation analysis (GSVA) based on HALLMARK gene sets indicated a highly activated state of epithelial–mesenchymal transition (EMT), angiogenesis, and transforming growth factor-β (TGF-β) pathways in male malignant cells (Fig. 2D).

Fig. 2

Identification and characterization of epithelial cells and malignant cells in males and females. A UMAP plot representing the subtypes of epithelial cells and malignant cells from male and female samples. B Heatmap showing the expression of marker genes in the epithelial cells and malignant cells. C Histogram showing the percentage of epithelial cells and malignant cells in samples and groups. D GSVA analysis showing the enrichment of specific pathways in malignant cells based on the HALLMARK gene set. E The volcano plot showing DEGs between male (blue dots) and female (red dots) malignant cells. F GSEA of hallmark interferon-γ response and gene ontology antigen presentation and processing via MHC class I signatures in malignant cells between male and female. G The correlation analysis between androgen response score and EMT, angiogenesis and TGF-β score in male malignant cells. DEGs: differential genes expression; EMT: epithelial to mesenchymal transition; GSEA: gene set enrichment analysis; GSVA: gene set variation analysis; TGF-β: transforming growth factor-β; UMAP: uniform manifold approximation and projection

We further examined the differentially expressed genes (DEGs) in malignant cells between males and females (Additional file 3: Table S2A). The expression of SAA1/2, as non-sex-related genes, was significantly higher in male malignant cells (Fig. 2E). The upregulation of SAA1/2 could increase the invasive potential of tumor cells in RCC and lead to massive T-cell infiltration [31, 32], which might explain the sex bias of malignant cells and T-cells infiltration.

Next, we examined the immune escape status of malignant. Compared with females, male malignant cells downregulated gene sets associated with interferon-γ response and antigen processing on major histocompatibility complex (MHC) class I, indicating the immune escape propensity of male tumors (Fig. 2F). Previous studies reported that androgen could promote tumor stemness and angiogenesis and enhance the proliferation, migration, and invasion of RCC [14]. In single-sample gene set enrichment analysis (ssGSEA), we also found that androgen response scores were highly correlated with EMT scores (R = 0.41, p < 0.001), angiogenesis score (R = 0.32, p < 0.001), and TGF-β scores (R = 0.58, p < 0.001) in male malignant cells (Fig. 2G).

Identification and characterization of myeloid cells in different gender

All myeloid cells were extracted and identified into 7 subtypes: type 1 conventional dendritic cells (cDC1), type 2 conventional dendritic cells (cDC2), tumor-associated macrophages (TAM), resident-tissue macrophages (RTM), CD14 monocytes, CD16 monocytes, and DEGs in each cluster were also summarized (Additional file 3: Fig. S3A–C). There was a significantly higher proportion of TAM in the TME compared to the adjacent normal kidney in both males and females, while the proportion of CD16 monocytes was significantly lower (Additional file 3: Fig. S3D, E). However, there was no significant difference in each subtype of myeloid cells between the TME of males and females.

GSVA indicated that female infiltrating macrophages had high activation of several pro-inflammatory signaling pathways in either adjacent normal kidneys (Additional file 2: Fig. S4A) or tumors (Additional file 2: Fig. S4B), including TH2 activation, lymphocyte activation, complement activation, cytokine signaling, and antigen presentation. Then, we separately analyzed the trajectories of macrophages in males and females to investigate their transition states. However, no significant difference was observed in the distribution of macrophages along the pseudotime between males and females in either the TME or adjacent normal kidneys (Additional file 2: Fig. S4C, D).

T/NK cell clustering reveals highly infiltrating and exhausted CD8+ T-cells in male TME of RCC

We performed unsupervised clustering of T and NK cells and identified 9 subtypes: CD4+ naïve T-cells (CD4+ Tn), tissue-resident memory CD4+ T-cells (CD4+ Trm), regulatory T-cells (Treg), effector memory CD8+ T-cells (CD8+ Tem), tissue-resident memory CD8+ T-cells (CD8+ Trm), early exhausted CD8+ T-cells (early CD8+ Texh), terminal exhausted CD8+ T-cells (term CD8+ Texh), natural killer T-cells (NKT cell), and NK cells (Fig. 3A, Additional file 2: Fig. S5A, B). DEGs in each T-cell subtype between males and females showed that the sex-related genes were significantly different in each T-cell subcluster (Additional file 3: Table S3A and Additional file 2: Fig. S3B).

Fig. 3

Characteristics of infiltrating T/NK cells in males and females. A The UMAP plot showing different T/NK cell subtypes, colored and labeled by cell type. B The volcano plot showing DEGs between males (blue dots) and females (red dots) in different T-cells subtypes. C Boxplots illustrating the percentage of infiltrating CD4+ and CD8+ T-cells in the tumor and adjacent normal kidneys of males and females. D Boxplots illustrating the percentage of infiltrating CD8+ T-cell subtypes in tumor and adjacent normal kidneys of males and females. E Differentially enriched pathways were scored per cell by GSVA in tumor-infiltrating CD8+ T-cells between males and females. F MxIF images of male and female tumors demonstrating tumor-infiltrating CD3+CD8+PD1+ T-cells. G The pie charts showing the percentage of CD8+ T-cell infiltration types. H Violin plots demonstrating CD3+, CD3+ CD8+, CD3+ CD8+ PD1+ infiltration levels in MxIF of tumors from males (n = 45) and females (n = 15). DEGs: differential gene expression; GSVA: gene set variation analysis; MxIF: multiplex immunofluorescence; UMAP: uniform manifold approximation and projection

There was a higher infiltration of CD8+ T-cells (CD8+ Tem, CD8+ Trm, early CD8+ Texh, term CD8+ Texh) in male TME as opposed to female TME, while CD4+ T-cells (CD4+ Tn, CD4+ Trm, Treg) were less abundant (Fig. 3C and D). Interestingly, although female and male TME both had a high level of term CD8+ Texh compared to adjacent normal kidneys, male TME had a significantly higher level of term CD8+ Texh infiltration (Fig. 3D). However, there was no significant difference in all CD4+ T-cell subtypes between male and female TME (Additional file 2: Fig. S5C).

Then, we performed GSVA in CD8+ T-cells in both adjacent normal kidneys and the tumor. Cell-toxicity or cell-exhaustion signature pathways were all highly enriched in female adjacent normal kidneys, which indicated a quite active renewal of CD8+ T-cells in females (Additional file 2: Fig. S5D). In the tumor cell-toxicity signature pathways, including chemokines, cytokines, and transporter functions, were highly enriched in females, whereas cell-exhaustion signature pathways were highly enriched in males in tumor-infiltrating CD8+ T-cells (Fig. 3E).

To further validate the highly infiltrating and exhausted CD8+ T-cells in male TME of RCC, we performed MxIF in high-quality tumor samples from RCC patients who had undergone radical nephrectomy (n = 60). We found higher tumor-infiltrating CD3+CD8+ T-cells and CD3+CD8+PD1+ T-cells in males compared to females (Fig. 3F). The majority of RCC (84.4%) in males were the infiltrative type, while there were more excluded (20.0%) and desert (20.0%) types in female RCC (Fig. 3G). The average density of CD3+CD8+ T-cells (p < 0.001) and CD3+CD8+PD1+ T-cells (p < 0.001) was significantly higher in male RCC compared with female RCC (Fig. 3G). After regrouping the previously published MxIF data of RCC [33], we confirmed the high-infiltration level of CD3+CD8+ T-cells in the male TME (Additional file 2: Fig. S5E).

Exhaustion of CD8+ T-cells in RCC by trajectory analysis

We explored the dynamic immune states and gene expression of CD8+ T-cells in tumors and adjacent normal kidneys by inferring the state trajectories using Monocle. Male and female tumors both showed more CD8+ T-cells in terminal pseudotime than adjacent normal kidneys. However, unlike in females, CD8+ T-cells in male RCC peaked at the end-stage of pseudotime (Fig. 4A). Along the pseudotime, the terminally exhausted score, activation dysfunction score, and inhibitory score increased continuously, while the progenitor exhausted score peaked in the middle of pseudotime and then decreased (Fig. 4B). The above findings suggested that the distribution of CD8+ T-cells in pseudotime reflected the level of exhaustion, and tumor-infiltrating CD8+ T-cells in males were mostly exhausted.

Fig. 4

Analysis of infiltrating CD8+ T-cell transition states in male and female samples. A Pseudotime-ordered analysis and density-distribution map of infiltrating CD8+ T-cells in tumors and adjacent normal kidneys of males and females. B Two-dimensional plots showing the change of expression scores for genes related to T-cell exhaustion and dysfunction along with the pseudotime. C Two-dimensional plots showing the dynamic expression of exhaustion and cytotoxicity genes during the CD8+ T-cell transitions along the pseudotime in male (blue) and female (red) samples. D The exhausted status and cytotoxic function of tumor-infiltrating CD8+ T-cells assessed by flow cytometry in male (n = 10) and female tumor (n = 10) samples

Next, we investigated the exhaustion and cytotoxicity of gene changes along with pseudotime in tumor-infiltrating CD8+ T-cells (Fig. 4C). The expression level of exhaustion genes, such as PDCD1, LAG3, and HAVCR2, increased along the pseudotime, while males showed a higher level of expression than females. Along the pseudotime, IFNG increased in males, but at a lesser rate than females, and declined at the end of the period. GrzmB increased to the peak and then decreased along the pseudotime, but the decrease in males was greater than that in females.

High-dimensional flow cytometry (FCM) analysis of available RCC samples (males: 10, females: 10) was performed to validate the protein-level expression of inferred receptors and ligands on CD8+ T-cells (expressing CD45+CD3+CD8+, Additional file 2: Fig. S6A). As shown in Fig. 4D, more exhausted CD8+ T-cells (expressing CD8+ PD1+, p < 0.001) and fewer cytotoxic CD8+ T-cells (expressing CD8+ GrzmB+, p < 0.001) infiltrated in male TME. Besides, tumor-infiltrating CD8+ T-cells in males produced less IFNγ and TNF than in females. As a result, although a higher infiltration level of CD8+ T-cells was found in male TME, those CD8+ T-cells were mainly exhausted and dysfunctional.

Androgen was involved in the dysfunction and exhaustion of CD8+ T-cells in male RCC

ssGSEA were used to evaluate the terminally exhausted score, activation dysfunction score, and androgen response score of each CD8+ T cell in male tumor samples. Interestingly, the androgen response score was highly associated with the terminally exhausted score and activation dysfunction score in CD8+ T-cells, which indicated that androgen might contribute to the dysfunction of exhaustion of CD8+ T-cells in males RCC (Fig. 5A). Human CD8+ T-cells isolated from PBMC were cultured with androgen to further verify these observations. We found an increase in the percentage of CD8+PD1+ T-cells and a decrease in the percentage of CD8+ GrzmB+ T-cells (Fig. 5B). Moreover, androgen can significantly inhibit the secretion of IFNγ and TNFα in CD8+ T-cells. In CD8+ T-cells toxicity assays, CD8+ T-cells were isolated from OT-I mice and co-cultured with Renca-OVA cells. FCM showed that androgen could significantly reduce CD8+ T-induced Renca-OVA cell apoptosis (Fig. 5C). IHC (CD8 and PD1) and ELISA (androgen) were performed on 42 patients (32 males and 12 females) with RCC who received immunotherapy. The results showed that male RCC patients had higher androgen levels and more CD8+PD1+ T-cells. The androgen levels were significantly associated with the percentage of CD8+PD1+ T-cells (R2 = 0.53, p < 0.0001, Fig. 5D). Additionally, we found that higher serum androgen was significantly associated with a worse prognosis in male RCC patients receiving immunotherapy (Fig. 5E).

Fig. 5

Androgen contributes to the dysfunction and exhaustion of CD8+ T-cells. A Correlation analysis between T-cells status score and androgen response score. B The FCM of human CD8+ T-cells in androgen culture. C CD8+ T-cells toxicity of OT-I mice in androgen culture. D IHC and ELISA analysis of 44 RCC patients receiving immunotherapy in SYSUCC (males: 32; females: 12). E Prognostic analysis of male RCC patients in different serum androgen after immunotherapy in SYSUCC. ELISA: enzyme-linked immunosorbent assay; FCM: flow cytometry; IHC: immunohistochemistry; HR: hazard ratio. RCC: renal cell carcinoma; SYSUCC: Sun Yat-sen University Cancer Center

Androgen receptor inhibitors (ARi) combined with anti-PD1 enhanced the efficacy of immunotherapy in vivo

Considering the persistence of androgen in males, a potential therapeutic drug was required to reduce the androgen-induced dysfunction and exhaustion of CD8+ T-cells. We designed a series of mouse experiments to verify the effects of androgen and ARi in RCC (Fig. 6A). We found that the RCC mice grew larger after tumor formation in male mice than in female mice (Fig. 6B). However, castration in male mice inhibited tumor growth, while androgen administration in female mice did the opposite. Enzalutamide, as one of ARi, could block the effect of androgen on cells. We found that both ARi and anti-PD1 could inhibit tumor growth in male mice, but only a combination of the two demonstrated the greatest inhibition on tumor growth (Fig. 6C).

Fig. 6

Androgen receptor inhibitors can enhance immunotherapy efficacy. A Flowchart of animal experiment in this study. B, C The tumor growth curves and tumor weight showing the effect of androgen and ENZ on tumor growth. D, E The expression of cytotoxicity and exhaustion markers of CD8+ T-cells were evaluated in mouse tumors by IHC. F Schematic representation showing the mechanism of androgen and androgen receptor inhibitors in RCC. ENZ: enzalutamide; IHC: immunohistochemistry; RCC: renal cell carcinoma

Then, we evaluated the expression of cytotoxicity and exhaustion markers of CD8+ T-cells using IHC in each mice group (Fig. 6D, Additional file 2: Fig. S7A, B). We found that male RCC had a higher H-score of exhaustion (PD1) and lower H-score of cytotoxicity (GrzmB, IFN-γ, and TNF-α) than females (Fig. 6E). However, castration decreased the H-score of PD1 and increased the H-score of GrzmB, IFN-γ, and TNF-α. In immunotherapy drug-sensitivity experiments, ARi combined with anti-PD1 maximized the cytotoxicity of CD8+ T-cells and minimized the exhaustion of CD8+ T-cells (Fig. 6E). As shown in the mechanism diagram, androgen led to the dysfunction and exhaustion of CD8+ T-cells in the TME of male RCC, while ARi activated CD8+ T-cells and enhanced the efficacy of immunotherapy (Fig. 6F).