Y chromosome loss in cancer drives growth by evasion of adaptive immunity

Abstract

Loss of the Y chromosome (LOY) is observed in multiple cancer types, including 10-40% of bladder cancers^1-6, but its clinical and biological significance is unknown. Here, using genomic and transcriptomic studies, we report that LOY correlates with poor prognoses in patients with bladder cancer. We performed in-depth studies of naturally occurring LOY mutant bladder cancer cells as well as those with targeted deletion of Y chromosome by CRISPR-Cas9. Y-positive (Y⁺) and Y-negative (Y^-) tumours grew similarly in vitro, whereas Y^- tumours were more aggressive than Y⁺ tumours in immune-competent hosts in a T cell-dependent manner. High-dimensional flow cytometric analyses demonstrated that Y^- tumours promote striking dysfunction or exhaustion of CD8⁺ T cells in the tumour microenvironment. These findings were validated using single-nuclei RNA sequencing and spatial proteomic evaluation of human bladder cancers. Of note, compared with Y⁺ tumours, Y^- tumours exhibited an increased response to anti-PD-1 immune checkpoint blockade therapy in both mice and patients with cancer. Together, these results demonstrate that cancer cells with LOY mutations alter T cell function, promoting T cell exhaustion and sensitizing them to PD-1-targeted immunotherapy. This work provides insights into the basic biology of LOY mutation and potential biomarkers for improving cancer immunotherapy.

Extended Data Fig. 1 |. LOY is associated with a worse clinical outcome for patients with MIBC and NMIBC.

a, Y chromosome genes expressed in normal bladder urothelium that were used to create a Y chromosome gene expression signature. b, Logrank p-values based on stratification by Y chromosome gene expression (normalized FPKM) on TCGA MIBC patient overall survival (OS). Genes resulting in statistically significant OS are plotted in panel c. NE, not expressed. c, Kaplan-Meier plots of OS from TCGA data for males with MIBC and either high or low KDM5D, TBL1Y, UTY (KDM6C), or ZFY expression. d, Kaplan-Meier survival curves stratified by the Y signature score or expression levels for UTY and KDM5D in NMIBC from the E-MTAB-4321 cohort. Survival differences are based on Logrank statistics. e, ChrY gene expression signature scores of TCGA data plotted with respect to extreme downregulation of chromosome Y (EDY, left panel) and Mosaic Alteration Detection for LOY (mLOY, right panel) levels. Statistical significance was determined by Wilcoxon rank-sum test (NoLOY n = 151, LOY n = 90, NoEDY n = 165, EDY n = 76). Boxplots represent the mean with first and third quartile data. Minimum and maximum datapoints are included.

Extended Data Fig. 2 |. Generation of Y+ and Y- BC models.

a, Histogram representation of deferentially regulated genes (DEG) from Y+ vs. Y− MB49 RNAseq data per mouse chromosome. b, qRT-PCR analysis of Uty, Kdm5d, Eifs3y, and Ddx3y expression in MB49 clones isolated from the parental MB49 compared to female murine breast cancer (E0771) and bladder cancer cells (NA13), and testis tissue. Curly brackets indicate the clonal lines used to generate the pooled Y+ and Y− MB49 sublines. c, qRT-PCR analysis of Uty and Kdm5d expression in the pooled Y+ and Y− sublines described in a. n = 3 biological replicates. Data are mean ± s.e.m. d, Bar graph of sequencing depth for each chromosome after performing whole exome sequencing (WES) on DNA from the Parental, Y−, and Y+ MB49 cell lines.

Extended Data Fig. 3 |. LOY has no effect on colony forming ability of BC in vitro.

a, MB49 Y+ and Y− cells were grown in 0.4% agar for two weeks. Colonies were stained with Nitro-BT and quantified using ImageJ. Average colony number and area were determined from those with a diameter that exceeded 100 μm (n = 4 biological replicates). Data representative of three independent experiments. Statistical significance was determined by two-sided unpaired t-test, P-value = 0.722. Data are mean ± s.e.m. b, In vitro cell proliferation (MTT cell viability) over a 6–8-day time course using three sets of genetically engineered MB49 cells: Y+, Y+ Kdm5d KO, Y+ Uty KO (left panel), Y−, Y− Kdm5d OE, Y− Uty OE (middle panel), and CRISPR-Y-Scr vs. CRISPR-Y-KO (right panel). n = 3 biological replicates. Data are mean ± s.e.m. c, qRT-PCR analysis of Uty expression in MB49 clones isolated from the CRISPR-generated Y-KO and Y-Scr MB49 cell lines. Curly brackets indicate the clonal lines used to generate the pooled Y+ Control and Y− KO MB49 sublines. Representative immunofluorescence images of the CRISPR-generated Y-KO and Y-Scr MB49 cell lines. Scale bar, 150 μM.

Extended Data Fig. 4 |. Increased lymphocyte activation in Y+ tumors.

a, Volcano plot of DEGs from bulk RNA isolated from Y+ and Y− MB49 tumors grown in male WT mice. Blue (Y+ tumors) and red (Y− tumors) genes correspond to statistically significant (Benjamini-Hochberg method, P < 0.05) genes that have a | >1 log₂ | fold-change in expression. b, PCA of DEGs described in a. c, Gene ontology (GO) pathway enrichment score plots of statistically significant gene set enrichment analyses (GSEA) using DEGs from a. NES, normalized enrichment score.

Extended Data Fig. 5 |. Comprehensive immune phenotyping of tumor-infiltrating leukocytes (TILs) in Y+ and Y− MB49 tumors.

a, UMAPs demonstrating individual spectral flow cytometry analysis of protein marker expression in CD45⁺ immune cells isolated from Y+ and Y− MB49 tumors grown in WT male mice. b, Heatmap of relative protein expression from immune cells described in a. c, Violin plots of each tumor sample across each cluster from the CD45⁺ immune cell UMAP (see Fig. 3b). d, Violin plot of PD-1 and PD-L1 mean fluorescence intensity in CD45⁺ immune cells from Y+ and Y− MB49 tumors. e, Representative dot plots and percentages of CD8⁺ and CD4⁺ T cells gated on total CD3⁺ T cells from CRISPR-Y-Scr (n = 8) and CRISPR-Y-KO (n = 9), MB49 tumors grown for 22 or 17 days, respectively, in male WT mice (left panels). Percentage of CD8⁺ T cells of total CD3⁺ T cells per tumor sample (right panel). f, Percentage of CD206⁺PDL1⁺ macrophages among total CD11b⁺F4/80⁺ macrophages from Control and Y KO MB49 tumors described in e. Statistics were determined using two-sided unpaired t-tests.

Extended Data Fig. 6 |. GeoMX histological evaluation of infiltrating immune cells in Y− and Y+ MB49 tumors.

a, Table of markers that are functionally categorized for GeoMX evaluation of Y+ and Y− MB49 tumors. b, Representative H&E image (left), immunofluorescence detection of nuclei (blue), cytokeratin (green), CD45⁺ immune cells (red) (middle), and associated computational digital profiling (right) to quantify markers shown in a. Scale bar, 125 μM. c, Quantification (log₂ fold change and P-value) of the markers listed in Y+ versus Y− MB49 tumors (n = 10 tumors per group and three TMA cores per tumor). Data representative of two independent experiments. Statistical significance was determined by two-sided unpaired t-test.

Extended Data Fig. 7 |. Characterization of tumor-infiltrating CD8⁺ T cells after PD-1 pathway blockade.

a–c, Relative spectral flow protein expression (a), sample-level violin plots per cluster (b), and heatmap of individual targets per cluster (c) after 200 μg anti-PD-1 or isotype control IgG treatments for 7 days using CD8⁺ T cells from Y+ and Y− MB49 tumors. d, Representative dot plots and percentages of TOX and/or GZMB-expressing CD8⁺ T cells from CRISPR-Y-Scr and CRISPR-Y-KO MB49 tumors grown in male WT mice after 200 μg anti-PD-1 or isotype control IgG treatments for 7 days. e–f, Percentage of PD1⁺TOX⁺ CD8⁺ T cells (e) and TOX⁻CD44⁺ (top panel) or TOX⁻ICOS⁺ (bottom panel) CD8⁺ T cells (f) from tumor samples described in d. See Fig. 5 for additional method details. Statistical significance was determined by two-sided unpaired t-test. Tests were conducted between isotype controls or between isotype controls and anti-PD1 treatment groups.

Extended Data Fig. 8 |. DDR-related pathways in TCGA Y_low vs. Y_high BC.

a, Heatmap of the indicated pathways and metadata from BC TCGA data. b–c, box plot of tumor neoantigen burden (TNB) per megabase (P = 0.700) (b), and associated pathway enrichment levels (c) from Y_high and Y_low tumors described in Fig. 1a. Statistical significance was determined by Wilcoxon test (Y_low n = 118 and Y_high n = 182). Boxplots represent the mean with first and third quartile data. Minimum and maximum datapoints are included.

Extended Data Fig. 9 |. Defective DDR pathway activation in Y− MB49 cells.

Normalized enrichment scores of statistically significant GSEA GO pathways using DEGs from Y− vs. Y+ MB49 cell cultures. Purple color denotes DNA repair-related pathways enriched in Y− cells.

Extended Data Fig. 10 |. Elevated genomic instability in LOY, Uty KO, and Kdm5d KO MB49 lines.

Genome instability pathway enrichment scores using RNA-seq data from control and genetically modified MB49 cell lines (Y+ and Y− cells, n = 5 technical replicates. n = 3 for all other cell lines). Two-sided unpaired t-test. Boxplots represent the mean with first and third quartile data. Minimum and maximum datapoints are included.

Fig. 1 |. LOY is associated with a worse prognosis for men with MIBC.

a, Heat map of clinical parameters and metadata for 300 male patients with MIBC from TCGA data. Chr. Y, Y chromosome. b,c, Plot of Y chromosome gene expression (b) and Kaplan–Meier survival curve (c) associated with Y_high (n = 182) and Y_low (n = 118) samples identified in a. CI, confidence interval; HR, hazard ratio. Differences in gene expression and survival were based on Wilcoxon rank-sum test and log-rank statistics, respectively. In box plots, the centre line represents the mean and box edges show first and third quartiles. Minimum and maximum datapoints are included.

Fig. 2 |. LOY and deletion of the Y chromosome genes Kdm5d and Uty promotes bladder tumour growth in an immune-competent host.

a, Proliferation of Y⁺ and Y⁻ MB49 cells in vitro (by MTT cell viability assay) over a six-day time course. Data are mean ± s.e.m. n = 3 biological replicates. OD, optical density. b,c, Tumour volume of Y⁺ and Y⁻ MB49 cells grown subcutaneously in wild-type (WT) C57BL/6 (b; n = 10 mice per group) or Rag2^−/−Il2rg^−/− (c; n = 10 mice per group) male mice. d, Relative mRNA expression (determined by RT–qPCR) of Uty, Kdm5d, Eifs3y and Ddx3y in CRISPR-generated Y-Scr and Y-KO MB49 lines. e, Tumour volume of CRISPR Y-Scr and CRISPR Y-KO MB49 cells grown subcutaneously in wild-type C57BL/6 male mice (n = 10 (Y-Scr) and 8 (Y-KO) mice per group). f, Volcano plot of statistically significant (Benjamini–Hochberg method, P < 0.05) differentially expressed genes (DEGs) from Y⁺ (blue) and Y⁻ (red) MB49 cells that exhibit a positive or negative log₂ fold difference in expression of greater than one. FC, fold change. g, PCA of Y⁺ (blue) and Y^– (red) DEGs described in f. Shaded areas encompass regions where the principal component falls within a 95% confidence internal. h, Left, tumour volume of Y⁺ Uty-KO and Y⁺ Kdm5d-KO cell lines injected subcutaneously into wild-type (n = 10 mice per group) or Rag2^−/−Il2rg^−/− (n = 10 mice per group) male mice. Right, similarly, Y⁻, Uty-overexpression (OE) and Kdm5d-OE cell lines were implanted into wild-type (n = 10 mice per group) or Rag2^−/−Il2rg^−/− (n = 8, 6 and 6 mice per group, respectively) male mice. All mouse experiments are representative of two independent experiments. Repeated measures two-way ANOVA; comparisons were made with the Y⁺ and Y⁻ controls in h. Data are mean ± s.e.m.

Fig. 3 |. Y_low bladder cancer overcomes T cell immunity and endows an immune-suppressive TME.

a, Tumour growth curves of Y⁺ and Y⁻ MB49 cells grown subcutaneously in wild-type (n = 8 mice per group), Ighm^−/− (n = 10 (Y⁺) and 7 (Y⁻) mice per group), Tcrb^−/−/Tcrd^−/− (n = 8 (Y⁺) and 9 (Y⁻) mice per group) or Rag2^−/− (n = 8 (Y⁺) and 10 (Y⁻) mice per group) male mice. Data are representative of two independent experiments. Statistical significance was determined by repeated measures two-way ANOVA. Data are mean ± s.e.m. b, Uniform manifold approximation and projection for dimension reduction (UMAP) of spectral flow cytometry analysis of CD45⁺ immune cell subsets from Y⁺ and Y⁻ MB49 tumours. NK, natural killer cells. c,d, Heat map of UMAP immune cell populations (c) and individual protein markers (d) from Y⁺ and Y⁻ MB49 tumour-infiltrating lymphocytes detailed in b. e, Volcano plot of significantly different (P < 0.05, green) UMAP clusters between Y⁺ and Y⁻ MB49 tumours. Red and blue labels (and circles in c) denote Y⁻ and Y⁺-associated clusters of interest, respectively. FDR, false discovery rate. f, Violin plot of PD-1 and PD-L1 expression in Y⁻ and Y⁺ MB49 CD45⁺ immune cells. Two-sided unpaired t-test. g, Volcano plot of statistically significant (red) GeoMX marker expression between Y⁺ and Y⁻ MB49 tumours. h, Dot plot of mean gene expression (checkpoint molecules and markers of T cell activation and exhaustion) and fraction of CD45⁺ cells from human MIBCs by snRNA-seq. i, Left, contour plot of T cells from human MIBC snRNA-seq overlaid onto an annotated UMAP as described in Methods. Right, fold change enrichment between Y_low and Y_high MIBC T cell subsets. T_ex, exhausted T cell; T_FH, T follicular helper cell; T_H1, T helper 1 cell; T_pex, progenitor exhausted T cell; T_reg, T regulatory cell. j, Stacked bar graph of immune cell subsets and T cell differentiation states from CODEX-stained Y_low and Y_high human MIBC tissue sections. T_RM, resident memory T cell.

Fig. 4 |. Human Y_low bladder cancers are enriched with CD8⁺ T cells with evidence of exhaustion.

a, Violin plots of macrophages and CD8⁺ T cell immune scores from Y_high and Y_low bladder cancer using urothelial bladder cancer TCGA data (n = 47 per group) and the microenvironment cell populations (MCP) counter method. Two-sided Wilcoxon rank-sum test. b, Y signature score using snRNA-seq data across 21 male naive bladder cancers. Colours denote Y_high (blue; >0.2 mean Y signature score), Y_low (red; <0.2 mean Y signature score) and Y_intermediate (grey; −0.2 to 0.2 mean Y signature score) patient samples. c, Distribution of the indicated non-epithelial immune cell types from samples described in b. d, Normalized expression of the indicated genes per T cell subtype from MIBC samples identified in b. T cells from melanoma and colon cancer samples were used to generate the reference T cell subtypes and gene expression.

Fig. 5 |. Improved response of Y_low bladder cancer to anti-PD-1 ICB therapy.

a, Tumour growth curves of Y⁺ and Y⁻ (top) and CRISPR Y-Scr and CRISPR Y-KO (bottom) treated with 200 μg anti-PD-1 antibody or isotype control IgG. Y⁺ isotype: n = 17 mice; Y⁺ anti-PD-1: n = 17 mice; Y⁻ isotype: n = 14 mice; Y⁻ anti-PD-1: n = 16 mice. Data representative of two independent experiments. CRISPR Y-Scr isotype: n = 11 mice; CRISPR Y-Scr anti-PD-1: n = 11 mice; CRISPR Y-KO isotype: n = 15 mice; CRISPR Y-KO anti-PD-1: n = 15 mice. Repeated measures two-way ANOVA. Data are mean ± s.e.m. NS, not significant. b, UMAPs of CD8⁺ T cell spectral flow cytometry results from Y⁺ and Y⁻ MB49 tumours seven days after treatments described in a. The UMAPs report cluster analyses (top) and cell proportions (bottom). c, UMAP heat maps of individual protein markers for stem-like, activated and exhausted T cells superimposed on the UMAPs in b. d, Volcano plot of statistically significant (green) UMAP clusters between isotype and anti-PD-1-treated Y⁺ and Y⁻ MB49 tumours. e, Left, Kaplan–Meier curves for patients with Y_high and Y_low bladder cancer treated with anti-PD-L1, from the IMvigor210 dataset. Kaplan–Meier curves for patients with bladder cancer stratified by expression of UTY (centre) and KDM5D (right). Survival differences are based on log-rank statistics.

Fig. 6 |. Increased genomic instability in Y_low bladder cancer.

a, Heat map of the indicated pathways and associated metadata from the IMvigor210 anti-PD-L1 clinical trial. Angio, angiogenesis; APM, antigen-processing machinery; CR, complete response; EMT, epithelial-to-mesenchymal transition; F-TBRS, fibroblast TGFβ response signature; GU, genomically unstable; IC, immune cells; Inf, infiltrated; PD, progressive disease; PR, partial response; SCCL, squamous cell carcinoma-like; SD, stable disease; TC, tumour cells; T_eff, T effector cells; UroA, urothelial-like A; UroB, urothelial-like B. PD-L1 immunohistochemistry scores: IC0 (<1%), IC1 (≥1% and <5%) or IC2+ (≥5%). b, Stacked bar graph of PD-L1 expression in immune cells (left) and tumour cells (right) from Y_high and Y_low bladder cancer samples identified in a. Pearson’s Chi-squared test. c, Box plot showing neoantigen burden per megabase in Y_low (n = 67) and Y_high (n = 129) tumours. Wilcoxon rank-sum test. d, Genomic instability pathway enrichment scores from bladder cancer described in a. Wilcoxon rank-sum test (Y_low: n = 33; Y_high: n = 82). BER, base excision repair; FA, Fanconi anaemia; HR, homologous recombination; NER, nucleotide excision repair. In box plots, the centre line represents the mean and box edges show first and third quartiles. Minimum and maximum datapoints are included.