NSD2 targeting reverses plasticity and drug resistance in prostate cancer

Abstract

Lineage plasticity is a cancer hallmark that drives disease progression and treatment resistance^1,2. Plasticity is often mediated by epigenetic mechanisms that may be reversible; however, there are few examples of such reversibility. In castration-resistant prostate cancer (CRPC), plasticity mediates resistance to androgen receptor (AR) inhibitors and progression from adenocarcinoma to aggressive subtypes, including neuroendocrine prostate cancer (CRPC-NE)^3-5. Here we show that plasticity-associated treatment resistance in CRPC can be reversed through the inhibition of NSD2, a histone methyltransferase⁶. NSD2 upregulation in CRPC-NE correlates with poor survival outcomes, and NSD2-mediated H3K36 dimethylation regulates enhancers of genes associated with neuroendocrine differentiation. In prostate tumour organoids established from genetically engineered mice⁷ that recapitulate the transdifferentiation to neuroendocrine states, and in human CRPC-NE organoids, CRISPR-mediated targeting of NSD2 reverts CRPC-NE to adenocarcinoma phenotypes. Moreover, a canonical AR program is upregulated and responses to the AR inhibitor enzalutamide are restored. Pharmacological inhibition of NSD2 with a first-in-class small molecule reverses plasticity and synergizes with enzalutamide to suppress growth and promote cell death in human patient-derived organoids of multiple CRPC subtypes in culture and in xenografts. Co-targeting of NSD2 and AR may represent a new therapeutic strategy for lethal forms of CRPC that are currently recalcitrant to treatment.

Fig. 1. Organoids from NPp53 mice recapitulate heterogeneity and neuroendocrine transdifferentiation of CRPC-NE.

a, H&E and immunofluorescence staining of sections from parental tumours and matched NPPO organoid lines established from NPp53 mice at passage 2. NPPO-2 and NPPO-3 are relatively homogeneous neuroendocrine lines, NPPO-1 and NPPO-5 are heterogeneous lines that contain neuroendocrine and AR⁺ non-neuroendocrine cells, and NPPO-4 and NPPO-6 are amphicrine lines with cells that express both AR and neuroendocrine markers. b, Diffusion component (DC) projection of scRNA-seq data from individual NPPO organoid lines analysed by VIPER. Proportions of cells in the three clusters are indicated by bars on the left of each plot. c, DC projection of a composite dataset of all five NPPO organoid lines. d, Left, VIPER-inferred activity for a published NEPC gene signature (CRPC-NE), with colour scale indicating normalized enrichment score (NES). Right, CytoTRACE analysis. e, Activity profiles inferred by VIPER for the indicated proteins in the composite dataset; colour scales indicate NES. f, Dot plot of inferred protein activity in the three clusters. g, Pathway analysis using VIPER-inferred activities, with ten selected enriched pathways shown in the NPPO dataset. h, Schematic of the co-culture assay for neuroendocrine transdifferentiation. i,j, Immunofluorescence analysis of RFP-expressing NPPO-1nonNE and NPPO-1NE organoids cultured separately (i) or together in organoid chimeras (j) for four passages. Arrows indicate cells co-expressing RFP and VIM together with the neuroendocrine markers CHGA and SYP. k, DC projection of scRNA-seq data and VIPER-inferred activity for the indicated proteins from NPPO-1nonNE (top) and co-cultured organoids (bottom) at passage 4. Colour scales correspond to normalized gene counts. Note the presence of RFP⁺ cells in clusters 2 and 3 in the co-cultured organoids. Scale bars, 50 µm (a,i,j).

Fig. 2. NSD2 and H3K36me2 are upregulated in CRPC-NE and correlate with poor patient outcomes.

a, Immunostaining of H3K36me2 in NPPO-1 organoids. Images are shown with and without co-staining for VIM. Scale bars, 50 µm. b, Scatter plot comparing H3K36me2 immunostaining fluorescence intensity between neuroendocrine (NE) and non-neuroendocrine (Non-NE) cells in three replicate experiments. Data points indicate the mean ± s.d. (n = 55 (NE) or n = 26 (non-NE) cells). Mean fluorescence intensities were compared using unpaired t-tests (two-tailed). c, Dot plot of NSD1, NSD2 and NSD3 expression in a published scRNA-seq dataset. d,e, Western blots of the indicated histone marks and NSD2 and EZH2 proteins in neuroendocrine and non-neuroendocrine organoid lines (for source data, see Supplementary Fig. 1). f, Genome browser view of CUT&Tag signals for H3K36me2, H3K27me3 and H3K27ac together with bulk RNA-seq reads at Chga, Foxa2, Onecut2, Ascl1 and Mycn loci in the indicated organoid lines. Genomic position annotations are shown at the top. g,h, Analysis of NSD2 and H3K36me2 levels in a prostate cancer (PCa) TMA. Violin plots show the percentage of NSD2⁺ cells (g) and H3K36me2^high cells (h) in CHGA⁺ neuroendocrine tumour cells or AR⁺CHGA^– tumour cells in each patient. Data are expressed as median and interquartile (IQR) ranges (primary PCa CHGA^–, n = 33; de novo NEPC CHGA⁺, n = 6; mCRPC CHGA^–, n = 18; CRPC-NE CHGA⁺, n = 6). Welch’s analysis of variance (ANOVA) followed by Dunnett’s multiple comparison test in g; unpaired t-test (two-tailed) in h. i,j, Kaplan–Meier plots of overall survival from the time of CRPC biopsy based on NSD2 gene expression in bulk transcriptomes from independent mCRPC patient cohorts: RMH (i) (n = 28 out of 94) and PCF–SU2C (j) (n = 27 out of 141). The gene expression cut-off was determined using the optimized Maxstat method; P values were calculated using the log-rank test (two-sided).

Fig. 3. NSD2 targeting reverts neuroendocrine differentiation and restores AR expression.

a,c,e, H&E staining and immunofluorescence of sections from NPPO-1NE (a), NPPO-2 (c) and MSKPCa10 (e) organoids cultured in the absence of DHT after CRISPR-mediated knockout of Nsd2 (sgNsd2) or treatment with sgCtrl. BSD, blasticidin (drug-selection marker). Scale bars, 50 µm. b,d, Density plots for VIPER-analysed scRNA-seq data from NPPO-1NE (b) and NPPO-2 (d) organoids after Nsd2 knockout or treatment with sgCtrl. Stacked colour bars at left indicate proportion of cells in each cluster; colour scales at right indicate estimated probability density. f,g, Left, gene set enrichment analysis of pseudo-bulk snRNA-seq data showing enrichment for a canonical AR target signature in NPPO-1NE organoids treated with sgCtrl (f) or sgNsd2 (g). Vertical blue lines (top) indicate the position of genes from the predefined set in the genes ranked from the least expressed to the most expressed in the organoid snRNA-seq sample (bottom). The enrichment score (ES) is shown on the y axis, and the normalized enrichment score (NES) with a nominal P value was determined from 1,000 random permutations of gene labels using permutation tests (one-sided). Right, DC projection of protein activity inferred from snRNA-seq data, showing gene expression enrichment for a canonical AR target signature. Cluster composition is shown as a stacked bar plot on the left; NES values indicating AR target gene enrichment are shown as a colour plot on the right from blue (negative) to red (positive). h, Growth curves for the indicated organoid lines in the absence or presence of DHT. Data points indicate the mean ± s.d. (n = 16 biological replicates). Organoids were cultured for three passages without DHT before treatment with 100 nM DHT or DMSO as a control. Data were analysed using two-way ANOVA and Tukey’s multiple comparison test.

Fig. 4. NSD2 targeting restores enzalutamide responses in neuroendocrine organoids and grafts.

a, Dose–response curves for enzalutamide-treated Nsd2 knockout (sgNsd2) or control (sgCtrl) NPPO-1NE and NPPO-2 organoids, for control (EV, empty vector) or H3.3K36M-transfected NPPO-4 and NPPO-6 organoids, and for NSD2 knockout (sgNSD2) or control (sgCtrl) MSKPCa10 organoids. Data points indicate the mean ± s.d. (n = 3 biological replicates). IC₅₀ values were calculated from dose–response curves by nonlinear regression (curve fit). Dose–response curves were compared by two-way ANOVA. b, Tumour growth curves for the same lines as in a, except treated with enzalutamide or DMSO control in vivo starting at day 14 after subcutaneous grafting in castrated NOD/SCID mice at day 0. Data points indicate the mean ± s.d. (n = 5 (NPPO-1NE, NPPO-2, NPPO-4 and NPPO-6) or n = 6 (MSKPCa10) biological replicates). Analysis was performed using two-way ANOVA and Tukey’s multiple comparison test. c,d, H&E and immunofluorescence analyses of sections from sgCtrl or sgNsd2 NPPO-1NE (c) and sgCtrl or sgNSD2 MSKPCa10 grafts (d). Scale bars, 50 µm.

Fig. 5. Pharmacological inhibition of NSD2 and AR suppresses growth of human CRPC organoids and grafts.

a, Timeline of treatments in organoids (c,e–i) and xenografts (d,j–l) with NSD2i alone (c–f) or with both NSD2i and enzalutamide (g–l). b, Subtypes and relevant features of human CRPC organoid lines. Met, metastasis; LN, lymph node. c,d, Western blot analyses of histone marks in the indicated organoid lines after treatment with 10 µM NSD2i for 21 days (c) or in WCM1262 xenografts grown for 21 days followed by the indicated NSD2i doses for 5 days (d). Source data are shown in Supplementary Fig. 1. e,f, H&E and immunofluorescence staining of sections from WCM1262 (e) and MSKPCa2 (f) organoids cultured without DHT after 21 days treatment with 3 µM (e) or 1 µM (f) NSD2i. CK8/18, cytokeratin 8 or 18. g, Dose–response curves for cell viability. Organoids were pretreated with NSD2i for 21 days before enzalutamide and NSD2i co-treatment for 5 days. Data points indicate the mean ± s.d. (n = 3 biological replicates). h, Synergy analysis of data in g, with Bliss synergy scores (>10 indicates synergy). P values and 95% confidence intervals (CI) were calculated using bootstrapping F-test (one-sided). i, Dose–response curves for cellular apoptosis. Organoids were pretreated with NSD2i for 21 days before enzalutamide and NSD2i co-treatment for 24–48 h. Data points indicate the mean ± s.d. (n = 3 biological replicates). j, Tumour growth curves for organoids grafted subcutaneously into castrated NOD/SCID mice. Mice were given NSD2i (150 mg kg^–1) or vehicle every day for 14 days, followed by NSD2i and/or enzalutamide as indicated. Data points indicate the mean ± s.d. (n = 6 biological replicates). Analysis was performed using two-way ANOVA and Tukey’s multiple comparison test. k,l, H&E and immunofluorescence staining of sections from xenografts of MSKPCa10 (k) and MSKPCa14 (l). Scale bars, 50 µm (e,f,k,l).

Extended Data Fig. 1. Phenotypes of NPPO organoid lines.

a, Low- and medium-power views of hematoxylin-and-eosin (H&E) stained sections from the indicated organoid lines at passage 2 and corresponding parental tumors. Scale bars, 200 µm. b, Sorting strategy for isolation of NPPO-1NE and NPPO-1nonNE sublines from NPPO-1 organoids. SSC, side scatter; FSC, forward scatter. c, Flow sorting of NPPO organoids to isolate sgCtrl and sgNsd2 transfected cells from the NPPO-1NE and NPPO-2 lines. d, Flow cytometry analysis of H3.3K36M-expressing cells from NPPO-4 and NPPO-6 organoid lines.

Extended Data Fig. 2. Analysis of neuroendocrine and non-neuroendocrine NPPO organoid lines.

a, VIPER-inferred activity of indicated proteins in the composite NPPO organoid dataset. b, Pathway analysis using VIPER inferred activities; 10 selected pathways are also shown in Fig. 1g. c, Western blot of RB1 and phospho-RB1 expression in NPPO organoid lines. d, Epigenomic changes in NPPO-1nonNE, NPPO-7, NPPO-8, and NPPO-9 organoid lines after enzalutamide treatment; these nonNE lines display an AR-low CRPC phenotype. Organoid lines were cultured in the absence of DHT. The levels of NSD2, NSD3, as well as H3K36me2 and H3K27me3 were detected by immunofluorescence staining. AR, Androgen receptor; CHGA, Chromogranin A; SYP, Synaptophysin; VIM, Vimentin. Scale bars, 50 µm.

Extended Data Fig. 3. Immunofluorescence screen for differential levels of epigenetic marks.

a, Immunostaining of indicated epigenetic marks in NPPO-1 organoids. Images are shown in pairs, with and without co-staining for Vimentin (VIM). CHGA, Chromogranin A; SYP, Synaptophysin. Scale bars, 50 µm. b, Scatter plots show quantitation of epigenetic mark levels, comparing fluorescence intensity in neuroendocrine (NE) and non-neuroendocrine (nonNE) cells in three replicate experiments. Data points indicate mean ± s.d. (biological replicates for NE and nonNE cells: H2AK119Ub n = 45, 37; H2BK120Ub n = 32, 32; H3K4me1 n = 74, 36; H3K4me3 n = 40, 23; H3K9me2 n = 65, 28; H3K9me3 n = 31, 27; H3K18Ac n = 30, 25; H3K27Ac n = 50, 24; H3K27me3 n = 34, 22; H3K36me3 n = 41, 26; H3K79me2 n = 39, 23; 5-Methylcytosine n = 38, 27; 5-Hydroxymethylcytosine n = 25, 25). Mean fluorescence intensities were compared by unpaired t-tests (two-tailed); comparisons lacking p-values were not significant. c, Box and whiskers plots show NSD1, NSD2, and NSD3 expression levels (log₁₀ transformed count/million (CPM)) in samples of CRPC-adeno (n = 34) and CRPC-NE (n = 15) based on data from. Data are expressed as median and interquartile (IQR) ranges (n = 15 biological replicates, CRPC-NE; n = 34 biological replicates, CRPC-Adeno); whiskers show Min to Max, and all points are shown. Unpaired t test (two-tailed) was used for comparison between two groups. d,e, Violin plots show NSD1, NSD2, and NSD3 expression in the single-cell RNA-seq dataset from. Pairwise statistical comparisons between subtypes were performed using the non-parametric Mann-Whitney U test (two-sided).

Extended Data Fig. 4. Analysis of histone marks in CRPC-NE organoid lines and patient samples.

a, Western blot analysis of indicated proteins and histone marks in TKO organoid lines with normal, adenocarcinoma, and CRPC-NE phenotypes. For gel source data, see Supplementary Fig. 1. b, Genome browser view of CUT&Tag signals for H3K36me2, H3K27me3, and H3K27ac together with bulk RNA-seq reads at Insm1, Sox11, Syp, and Dnmt3a loci in the indicated organoid lines. Genomic position annotations are shown on top. c, Heatmaps of CUT&Tag signals for the indicated histone marks in genomic regions marked by H3K36me2, comparing four NE organoid lines with four nonNE organoid lines. d, Violin plot with overlaid box plot showing quantitative comparison of H3K36me2, H3K27me3, and H3K27ac CUT&Tag signals of four NE and four nonNE organoid lines, at genomic domains marked by H3K36me2. Data are expressed as median and interquartile (IQR) ranges (n = 4 biological replicates for each histone modification); whiskers show Min to Max. Welch two sample t-test (two-tailed) was used to compare the median of NE and nonNE samples. e, Scatter plot shows correlation between NSD2 expression and a neuroendocrine signature score (based on) in two independent CRPC cohorts (PCF-SU2C (n = 159; r = 0.3, p = 3 × 10⁻⁵) and RMH (n = 95; r = 0.2, p = 0.1)) Spearman correlation test (two-sided) was used for comparison in both cohorts. The weaker correlation in the RMH dataset is likely due to the small number of CRPC-NE patients in this cohort. f,g, Analyses of NSD2 and H3K36me2 levels in a prostate cancer tissue microarray. Shown are five-color overlay images of representative tissue cores and high-power magnification of four-color images. Spatial plots show the enrichment of NSD2⁺ cells (f) and H3K36me2^high cells (g) among CHGA⁺ neuroendocrine tumor cells or among AR⁺/CHGA^– adenocarcinoma. Scale bars, 50 µm.

Extended Data Fig. 5. Analysis of histone marks following NSD2 targeting.

a,b, Hematoxylin-and-eosin (H&E) staining and immunofluorescence staining of sections from NPPO-4 (a) and NPPO-6 (b) organoids cultured in the absence of DHT after transfection of the oncohistone H3.3K36M or control (empty vector). AR, Androgen receptor; CHGA, chromogranin A; HA, Hemagglutinin tag; SYP, Synaptophysin; VIM, Vimentin. Scale bars, 50 µm. c, Density plots for VIPER-analyzed scRNA-seq data from NPPO-6 organoids following H3.3K36M expression or control (EV, empty vector). Changes in cluster sizes are quantified in vertical bars at left of each plot. d,e, Western blot analysis of NSD2 and EZH2 proteins (top) and of H3K36me2, H3K36me3, H3K27ac, and H3K27me3 levels (bottom) in control (sgCtrl) and Nsd2 knock-out (sgNsd2) NPPO-1NE and NPPO-2 organoids, in control (EV) and H3.3K36M-transfected NPPO-4 and NPPO-6 organoids, or in control (sgCtrl) and NSD2 knock-out (sgNSD2) MSKPCa10 organoids (source data in Supplementary Fig. 1). f, Violin plots with overlaid box plots comparing H3K36me2, H3K36me3 and H3K27me3 CUT&Tag signals at genomic regions marked by H3K36me2 between sgCtrl and sgNsd2 or between EV and H3.3K36M-transfected NPPO organoids. Data are expressed as median and interquartile (IQR) ranges (n = 20,818 regions tested for each organoid line); whiskers show Min to Max. Wilcoxon rank-sum test (two-tailed) was used. LFC, log₂ fold-change. g, Heatmaps of CUT&Tag signals for the indicated histone marks at genomic domains marked by H3K36me2, comparing sgCtrl and sgNsd2 or EV and H3.3K36M-transfected NPPO organoids. h, Principal Components Analysis (PCA) of H3K36me2 CUT&Tag signals in the indicated NPPO organoid lines. Shaded regions indicate NE (red) and nonNE (green) phenotypes. i, Heatmaps of H3K36me3 CUT&Tag signals at gene body regions (left) and at genomic domains marked by H3K36me2 (right). j, Heatmaps of H3K36me2 CUT&Tag signals in NPPO-1NE organoids in the absence or presence of NSD2i.

Extended Data Fig. 6. Differential activity of genes following NSD2 targeting.

a, Single-cell Gene Set Enrichment Analysis (scGSEA) of gene expression signatures computed as differential between Cluster 3 and Cluster 1, showing enrichment of genes with decreased H3K36me2 levels identified by CUT&Tag analysis in Nsd2 (sgNsd2) knock-out or control (sgCtrl) NPPO-1NE and NPPO-2 organoids, and control (EV, empty vector) or H3.3K36M-transfected NPPO-6 organoids. Plots show enrichment score (ES) on the y-axis and normalized enrichment score (NES) with nominal P-value (p) computed across 1,000 permutations of gene labels using permutation tests (one-sided). b, Heatmap shows transcription factors (TFs) and co-transcription factors (co-TFs) prioritized by VIPER that are differentially active after NSD2 inhibition or H3.3K36M expression. Normalized Enrichment Scores (NES) are colored from blue (negative) to red (positive). c, Heatmaps of CUT&Tag signals for the indicated histone marks at NE-specific enhancers or at nonNE-specific enhancers in the NE organoid lines (NPPO-1NE, NPPO-2, NPPO-4, and NPPO-6) versus nonNE lines (NPPO-1nonNE, NPPO-7, NPPO-8, and NPPO-9). d, Venn diagrams showing differential numbers of NSD2-dependent or NSD2-independent ATAC-seq peaks co-localized with NE or nonNE specific H3K27ac peaks within H3K36me2 domains identified by integration of CUT&Tag and scATAC-seq data. e, Examples of top motifs associated with NE enhancers across all three comparisons, and predicted transcription factors binding to these motifs. f, Gene Ontology enrichment analysis using NE enhancer associated genes, showing enrichment for pathways associated with neural specification.

Extended Data Fig. 7. Organoid response to combined NSD2 targeting and enzalutamide treatment.

a,b, Scatter dot plots show growth of Nsd2 knock-out (sgNsd2) or control (sgCtrl) NPPO-1NE and NPPO-2 organoids (a), and control (EV, empty vector) or H3.3K36M-transfected NPPO-4 and NPPO-6 organoids (b). Experimental values were normalized to DMSO controls and shown as percentage of viable cells. Data points indicate mean ± s.d. (n = 10 biological replicates). Analysis was performed using unpaired t-tests (two-tailed). c-f, Whole-mount images of organoids following Nsd2 knock-out or H3.3K36M expression treated with either DMSO control or 10 µM enzalutamide. All NPPO organoid lines express YFP (green) due to the Cre reporter in the NPp53 mouse model; the NPPO-1NE, NPPO-2 organoids additionally express RFP following sgCtrl or sgNSD2 transfection. g,h, Scatter plots show growth of sgNSD2 or sgCtrl MSKPCa10 organoids treated with 3.5 µM (g) or 10 µM enzalutamide (h) or DMSO control. Data points indicate mean ± s.d. (n = 10 biological replicates). Analysis was performed using unpaired t-tests (two-tailed). i, Whole-mount images of MSKPCa10 organoids following NSD2 knock-out treated with either DMSO control or 10 µM enzalutamide. These organoids express RFP following sgCtrl or sgNSD2 transfection.

Extended Data Fig. 8. Graft response to combined NSD2 targeting and enzalutamide treatment.

a-c, Representative whole-mount images of grafts following subcutaneous implantation of NPPO-1NE and NPPO-2 control or Nsd2 knock-out organoids (a), NPPO-4 and NPPO-6 control or H3.3K36M-transfected organoids (b), and MSKPCa10 control or NSD2 knock-out organoids (c) into castrated NOD/SCID immunodeficient mice treated with DMSO control or enzalutamide. d-f, H&E and immunofluorescence analysis of sections from NPPO-2 (d), NPPO-4 (e), and NPPO-6 (f) grafts. CHGA, Chromogranin A; HA, Hemagglutinin tag. Scale bars, 50 µm. g-j, (left) Gene Set Enrichment Analysis (GSEA) on pseudo-bulk from snRNA-seq data showing enrichment for a canonical AR target signature in the NPPO-2 sgCtrl (g) and NPPO-2 sgNsd2 (h) organoids, as well as in the NPPO-6 EV (i) and NPPO-6 H3.3K36M (j) organoids. Vertical blue lines (top) indicate the position of genes from the predefined set within the genes ranked from the least expressed to the most expressed in the organoid snRNA-seq sample (bottom). Plot show the enrichment score (ES) on the y-axis, and normalized enrichment score (NES) with nominal P-value (p) computed across 1,000 permutations of gene labels using permutation tests (one-sided). (right) Diffusion component projection of protein activity inferred from snRNA-seq data. Shown are cluster composition on the left as stacked bar plots and gene expression enrichment for a canonical AR target signature in the NPPO-2 sgCtrl (g) and NPPO-2 sgNsd2 (h) organoids, as well as in the NPPO-6 EV (i) and NPPO-6 H3.3K36M (j) organoids. Normalized Enrichment Scores (NES) are colored from blue (negative) to red (positive).

Extended Data Fig. 9. Synthesis and activities of a novel small molecule inhibitor of NSD2.

a, Chemical structure of NSD2i used in this work. b, Strategy for synthesis of NSD2i, adapted with slight modifications from. In brief, commercially available Boc protected chiral amino ketone (1) was reduced by sodium borohydride in methanol to provide the diastereomeric mixture of alcohols (2). The Boc group in 2 was then removed under mild PTSA conditions at 60 °C to provide, after basic work up, piperidine 3 as a free base. Compound 3 was then condensed with fluoropyridine (4) in N-methyl pyrrolidinone at 100 °C for 12 h to give ester 5 in more than 65% yield for the 3 steps. Ester 5 was in turn reduced to the corresponding alcohol (6), before transformation to the respective chloride (7), which set the stage for reaction with N,N-Bis(Boc) adenine under basic conditions to get to advanced intermediate (8) in 40% yield over 3 steps. Removal of the Boc groups using 50% TFA in DCM (v:v), and hydrogenolysis of the benzyl group in (9) provided compound 10 in 30% yield over 2 steps. Preparative chiral resolution of amino alcohol (10) provided the desired R,R-compound in 28% yield. c,d, NSD2i IC₅₀ values for the indicated lysine methyltransferases (KMTs). Data are presented as mean ± s.e.m. (n = 3 independent replicates). e, Diffusion component projections of single-cell RNA-seq data from NPPO-1NE organoids treated with DMSO (control) or NSD2i. Changes in cluster sizes are quantified in vertical bars at left of each plot. f, Gene Set Enrichment Analysis (GSEA) on pseudo-bulk from snRNA-seq data showing enrichment of a canonical AR target signature in the DMSO and NSD2i-treated NPPO-1NE organoids. Vertical blue lines (top) indicate the position of genes from the predefined set within the genes ranked from the least expressed to the most expressed in the organoid snRNA-seq sample (bottom). Plot show the enrichment score (ES) on the y-axis, and normalized enrichment score (NES) with nominal P-value (p) computed across 1,000 permutations of gene labels using permutation tests (one-sided). g, Western blots of AR expression in nonNE organoid lines following 1 µM enzalutamide or combined 1 µM enzalutamide and 1 µM NSD2i treatment for 40 days. h, Box and whiskers plots show percentage of viable cells in the four treatment conditions utilized in the indicated human CRPC organoid lines, combining all enzalutamide concentrations. Data are expressed as median and interquartile (IQR) ranges (n = 24 biological replicates); whiskers show Min to Max, and all points are shown. Analysis was performed using one-way ANOVA and Tukey’s multiple comparison test. i, Dose-response curves for cell viability following NSD2i and enzalutamide treatment of NPPO-1NE, NPPO-2, NPPO-4, and NPPO-6 mouse CRPC-NE organoids. Organoids were pre-treated with 1 µM NSD2i for 12 days prior to enzalutamide and NSD2i treatment for 5 days. IC₅₀ values were calculated by nonlinear regression (curve fit). Data points indicate mean ± s.d. (n = 3 biological replicates). Dose-response curves were compared by two-way ANOVA. j, Dose-response curves for cell viability following NSD2i and enzalutamide treatment of control and Nsd2 knock-out NPPO-1NE, NPPO-2, and MSKPCa10 organoid lines. Organoids were pre-treated with 1 µM NSD2i for 12 days (NPPO-1NE and NPPO-2) or with 0.3 µM, 1 µM, 3 µM, or 10 µM NSD2i for 21 days (MSKPCa10) prior to enzalutamide and NSD2i treatment for 5 days. Data points indicate mean ± s.d. (n = 3 biological replicates). k,l, Dose-response curves for cell viability following NSD2i and enzalutamide treatment of the human CRPC-NE organoid line WCM154 (k) or EZH2 inhibitor PF-06821497 (mevrometostat) and enzalutamide treatment of the human CRPC-NE organoid line MSKPCa10 (l). Organoids were pre-treated with the indicated concentrations of NSD2i (k) or PF-06821497 (l) for 21 days prior to enzalutamide and NSD2i (k) or PF-06821497 (l) treatment for 5 days. Data points indicate mean ± s.d. (n = 3 biological replicates). Synergy analysis is shown below with Bliss synergy score (> 10 indicates synergy). P-values (p) and 95% confidence intervals (CI) were calculated using bootstrapping F-test (one-sided). No synergy was detected in l. m, Box and whiskers plots show percentage of apoptotic cells in the four treatment conditions as in (h) in the indicated human CRPC organoid lines. Data are expressed as median and interquartile (IQR) ranges (n = 18 biological replicates); whiskers show Min to Max, and all points are shown. Analysis was performed using one-way ANOVA and Tukey’s multiple comparison test. n, Dose-response curve for cellular apoptosis following NSD2i and enzalutamide treatment of WCM154. Organoids were pre-treated with the indicated concentrations of NSD2i for 21 days prior to enzalutamide and NSD2i co-treatment for 24–48 h. Data points indicate mean ± s.d. (n = 3 biological replicates).

Extended Data Fig. 10. Effects of the novel NSD2i on in vivo xenografts of CRPC organoids.

a, Overall body weights of mice utilized in xenograft experiments. Data points indicate mean ± s.d. (n = 6 mice for each experiment). No significant differences were observed between treatment groups. b, Representative whole-mount images of grafts following subcutaneous implantation of the indicated organoid lines and growth for 56 days. Grafts were treated with 150 mg/kg NSD2i and 10 mg/kg enzalutamide following the timeline shown in Fig. 5a. Images are not shown for the MSKPCa2 grafts following NSD2i treatment as they were too small to be recovered. c, Hematoxylin-eosin (H&E) stained sections of grafts for MSKPCa10, MSKPCa14, and WCM1262 following the indicated treatments. Note the extensive necrosis and/or fibrosis observed following combined NSD2i and enzalutamide treatment. Scale bars, 50 µm. d, H&E and immunofluorescence staining of sections from WCM1262 grafts following the indicated treatments. CC3, cleaved caspase 3. Scale bars, 50 µm. e, Box and whiskers plots show percentage of proliferating cells (left) and apoptotic cells (right) in the four treatment conditions in MSKPCa10, MSKPCa14, and WCM1262 grafts. Each dot represents one xenograft. Data are expressed as median and interquartile (IQR) ranges (n = 6 biological replicates); whiskers show Min to Max, and all points are shown. Analysis was performed using one-way ANOVA and Tukey’s multiple comparison test.