All experiments, including animal studies, were conducted in compliance with federal and state government guidelines and followed approved protocols by the Institutional Biosafety Committee and the Institutional Animal Care and Use Committee (IACUC) at the University of California, Los Angeles.

Animal work

All mice are housed under 12 h:12 h light–dark cycle, with room temperature maintained at 23 °C and relative humidity level of 30–70%. Mouse cages include clean bedding and enrichment materials consistent with IACUC regulations. According to the Animal Research Committee policy on humane treatment and endpoints, mice must be killed if tumours become ulcerated or necrotic, and/or impair normal function. All experiments were terminated before tumours reached this stage.

For experiments described below, animals were housed under the care of the Division of Laboratory Animal Medicine at the University of California, Los Angeles, using protocols approved by the Animal Research Committee (ARC no. 2017-020). Prostates from 3–6-month-old immunocompetent male C57BL/6J mice from Jackson Laboratories were used for primary basal and luminal cell experiments. Mpc1 floxed male mice were of mixed C57BI/6N and C57BI/6J genetic background51. For in vivo UK5099 experiments, 12 million 16D cells were subcutaneously implanted with 100 μl of Matrigel (Corning) into the right flank of NOD-scid-IL2Rgnull male mice through a 25-gauge needle under inhalation anaesthesia with 2–3% isoflurane. The mice were fed with either control chow or chow containing 0.08 mg kg−1 UK5099 (OpenStandard Diet with 15 kcal% Fat with Blue Dye Irradiated (10–20 kGy), Research Diets) until tumours were formed and collected. PDX MDA PCa 203-A and MDA PCa 183-A tumours were obtained from the MD Anderson Cancer Center52. Both 203-A and 183-A PDX models were derived from 58-yr-old males. When these PDX models were originally generated, written, informed consent was obtained from patients before sample acquisition, and all samples were processed according to a protocol approved by the Institutional Review Board of the University of Texas MD Anderson Cancer Center. The studies were conducted in accordance with the Belmont Report and the US Common Rule. Patients were not compensated, and they cannot be identified from data provided in this manuscript. A tumour tissue piece of 50–200 mg was implanted in the right flank of NOD-scid-IL2Rγnull mice subcutaneously through a 5-mm skin incision under inhalation anaesthesia with 2–3% isoflurane. After closing the wound with a surgical clip, 100 ml of Matrigel (Corning 354234) was injected at the implantation site. Carprofen was administered subcutaneously at a dose of 5 mg kg−1 after surgery. The surgical clip was removed 1–2 weeks later. When the tumour had grown larger than 500 mm3, the mouse was euthanized and the tumour was excised and trimmed and then processed for the experiments, re-implanted or cryopreserved. Cryopreservation of the tissue was done in media with 50% FBS, 40% DMEM and 10% dimethylsulfoxide.

The experiments described below were conducted in compliance with European guidelines regarding animal research and ethical protocols (under protocol numbers 671N and 673N) and approved by the local ethical committee for animal welfare, Comité Ethique du Bien-Être Animal (CEBEA). All animals were housed under standard laboratory conditions in a certified animal facility, receiving food and water ad libitum. Prostates used for isolation of multipotent basal cells and basal-derived luminal cells were collected from CD1 mice purchased from the Jackson Laboratory. The experimental mice used were males of mixed background and at P10–P12 age.

Pten floxed and Pten;Rb1 floxed 3-month-old male mice were of mixed C57BL/6:129/Sv:FVB genetic background27 and were housed at Harvard Medical School under IACUC-approved protocols.

Mouse prostate dissociation to single cells

Using a razor blade, individual mouse prostates were mechanically dissociated in dissociation media composed of RPMI 1640 (Gibco) containing 10% FBS (Corning), 1 × penicillin-streptomycin (P/S) (Gibco), 1 mg ml−1 collagenase type I (Gibco), 1 mg ml−1 dispase (Gibco), 0.1 mg ml−1 deoxyribonuclease (Gibco) and 10 μM of the p160ROCK inhibitorY-27632 dihydrochloride (RI) (Tocris Bioscience). When chunks were no longer visible, the samples were incubated at 37 °C on a nutating platform for 1.5 h in 10 ml of dissociation media. After centrifugation at 800g for 5 min, the pellet was washed with 1 × PBS. The cell pellet was resuspended in 2.7 ml of 0.05% Trypsin-EDTA (Gibco) and incubated at 37 °C for 5 min. Trypsin was inactivated with 300 ml of dissociation media. Cells were further dissociated by pipetting with a P-1000 pipette and an 18 G syringe. Cells were passed through a 100-μm cell strainer (Corning).

Staining and sorting cells from mouse prostate for isolation of primary basal and luminal cells

Dissociated cells were stained with directly conjugated primary antibodies rat anti-CD49f-PE (BioLegend 313612, 1:100), rat anti-CD326 (EpCAM)-APC (BioLegend 324207, 1:100), rat anti-CD31-FITC (BioLegend 102405, 1:100), rat anti-CD45-FITC (BioLegend 103108, 1:100), rat anti-Ter119-FITC (BioLegend 116205, 1:100) and rat anti-ESAM-FITC (BioLegend 136205, 1:100) for 20 min on ice. Cells were stained in media containing RPMI 1640 (Gibco), 10% FBS (Corning), 1 × P/S and 10 μM RI. Sorting was performed on a BD FACSAria II (BD Biosciences).

Bulk RNA-seq

These methods apply to Figs. 1b,c,f,g, 3j and 4b and Extended Data Figs. 1b–e, 4f–h, 6d and 8. RNA was extracted from the cells using the RNeasy Mini Kit (QIAGEN) following the manufacturer’s instructions. Libraries for RNA-seq were prepared with the KAPA Stranded mRNA-Seq Kit (Roche). The workflow consists of messenger RNA enrichment, complementary DNA generation, end repair to generate blunt ends, A-tailing, adaptor ligation and PCR amplification. Different adaptors were used for multiplexing samples in one lane. Sequencing was performed on an Illumina HiSeq 3000 for single-end 1 × 50 runs (Figs. 1b,c, 3j and 4b and Extended Data Figs. 1b–e, 2d,e, 4f–h and 6d) and paired-end 2 × 50 runs (Extended Data Fig. 8).

Bulk RNA-seq analysis

These methods apply to Figs. 1b,c,f,g, 3j and 4b and Extended Data Figs. 1b–e, 4f–h and 6d. Sequencing quality metrics were generated during sequencing runs using Illumina Sequencing Analysis Viewer. Demultiplexing was performed with Illumina Bcl2fastq (v.2.19.1.403) software. The reads were mapped by STAR 2.7.9a (ref. 53) and read counts per gene were quantified using the mouse Ensembl GRCm39.105 GTF file. In Partek Flow v.7.0, read counts were normalized by counts per million (CPM) 1.0 × 10−4. All results of differential expression analysis used the statistical analysis tool DESeq2 (v.1.40.2)54. KEGG pathway analysis was performed using DAVID Bioinformatics55,56. GSEA was performed as described previously using GSEA_4.0.3 software57,58.

Significance testing of RNA-seq of UK5099- and sodium butyrate-treated organoids

These methods apply to Extended Data Fig. 8. To measure the RNA abundance, RNA-seq reads were trimmed using fastp (v.0.20.1)59 with default parameters, then mapped to the mouse Ensembl GRCm38-EBI102 using STAR (v.2.7.10a)53. STAR alignment was carried out using default settings with an additional argument to include the minimum length of 10 base pairs (bp) for the chimeric junction segment. Aligned reads were quantified using the rsem-calculate-expression program (v.1.3.3)60 for transcripts per million (TPM) calculation with default settings. We also assessed read-level quality control metrics using FastQC (v.0.11.8).

To test the combination effect of UK5099 and butyrate, we constructed the following two-factor, two-level linear model:

$$Y = _+_\times \mathrm+_\times \mathrm+_\times \mathrm$$

Here, Y refers to the abundance level of a gene, which is log2 transformation of TPM values; \(_\) refers to the basal abundance level of that gene; ‘UK5099’ indicates ‘UK5099-dependent, Butyrate-independent’ abundance changes; ‘Butyrate’ indicates ‘Butyrate-dependent, UK5099-independent’ abundance changes; ‘UK5099:Butyrate’ captures ‘UK5099-dependent, Butyrate-dependent’ abundance changes.

We used R package limma (v.3.17)61 in R (v.4.2.2) to fit each gene in the RNA-seq to the model. The model was adjusted using empirical Bayes moderation for standard error, and the false discovery rate (FDR) was controlled using the Benjamini–Hochberg method62. Genes exhibiting significant changes were identified based on the adjusted P < 0.01 and |log2(Coefficient)| > 1 threshold. Venn diagrams representing the overall and directional effects were generated using the VennDiagram package in R (v.1.7.3)63. The hierarchical clustering heatmap of gene TPM was constructed using R package BoutrosLab.plotting.general (v.7.0.8)64.

Functional enrichment analysis

These methods apply to Extended Data Fig. 8. For genes with differential mRNA abundance calculated based on the coefficient from the general linear model, we ranked the genes according to their log2(Coefficient) from high to low. GSEA was then performed using the R package clusterProfiler (v.3.17)65. Gene ontology enrichment analysis was conducted for both upregulated genes (log2(Coefficient) > 1, −log10(FDR) > 1) and downregulated genes (log2(Coefficient) < −1, −log10(FDR) > 1) using the R package clusterProfiler (v.3.17). The results of both GSEA and gene ontology enrichment analyses were visualized using BoutrosLab.plotting.general (v.7.0.8)64.

scRNA-seq

Basal cells were isolated from Ptenfl/fl;Rb1fl/fl mouse prostates and infected with FU-CRW (red fluorescent protein, RFP) lentivirus. Lentiviral spinfections were done by culturing the cells with virus in 200 μl of RPMI 1640 (Gibco) containing 10% FBS (Corning), 1 × (P/S) (Gibco) and 10 μM RI (RPMI, 10% FBS, 1% P/S + RI) plus 8 μg ml−1 polybrene for 30 min at 37 °C, then spinning at 300g for 90 min. After spinfection, growth factor reduced Matrigel (Corning) was added to the cell suspension at a final concentration of 75% before plating into rings in 24-well plates. Organoids were cultured as previously described66 and passaged every 7–10 d. After >4 passages, organoids were treated with vehicle (dimethylsulfoxide) or 10 μM UK5099 for 3 d. Organoids were removed from Matrigel by incubating in Advanced DMEM/F-12 (Gibco) containing 1 mg ml−1 dispase (Gibco) and 10 μM RI for 1 h at 37 °C. After centrifugation at 800g for 5 min, the pellet was washed with 1 × PBS (Gibco). Organoids were resuspended in 800 μl of 0.05% Trypsin-EDTA (Gibco) and incubated at 37 °C for 5 min. The trypsin was quenched with 200 μl of RPMI, 10% FBS, 1% P/S + RI and organoids were pipetted up and down ten times to dissociate to single cells and passed through a 100-μm cell strainer (Corning). Samples were counted using a Countess II Automated Cell Counter (Thermo Fisher Scientific) and haemocytometer for cell concentration and viability using Trypan Blue stain 0.4% (Invitrogen). Cells were loaded to form gel beads in emulsion (GEMs) and barcode individual cells. GEMs were treated according to the manufacturer’s instructions. Single-cell gene expression libraries were created using Chromium Next GEM Single Cell 3′ (v.3.1 Chemistry) (10x Genomics), Chromium Next GEM Chip G Single Cell Kit (10x Genomics) and Single Index Kit T Set A (10x Genomics) according to the manufacturer’s instructions. Paired-end sequencing was done using an Illumina Novaseq 6000 at a sequencing depth of 492,915,641 and 555,876,242 read pairs for vehicle and UK5099 samples, respectively, with read length of 151 for both read 1 and read 2, and with an 8 bp index read for multiplexing. Basecalling was done using Illumina Casava (v.1.7) software. CellRanger (v.5.1) count was used to create an RNA abundance matrix with --expect-cells=1000 and Mus musculus (mm10) from the Ensembl database as a reference genome67. RNA abundance matrices from vehicle- and UK5099-treated samples were loaded into the Seurat (v.3.2.2) R package68. DoubletDcon (v.1.1.2) was used to remove potential doublets69. Additionally, cells were filtered based on the number of genes (≥250), unique molecular identifiers (≥500) and percentage of mitochondrial genes (<20%). After quality control, log2 normalization was performed within each sample using NormalizeData function with default parameters. The top 2,000 variable genes were selected using FindVariableFeatures. The two samples were integrated together with FindIntegrationAnchors and IntegrateData functions which incorporate canonical correlation analysis to align cells with similar transcriptomic patterns across samples. PCA was performed after the integration. The top 20 principal components were used to construct the k-nearest neighbour graph, followed by Louvain algorithm to cluster cells based on similar gene expression patterns. Cell clusters were visualized using t-distributed stochastic neighbour embedding. After, markers for each cluster were determined using FindAllMarkers with average log2 fold change > 0.25 and minimum percentage difference > 0.25. Cell types were determined by comparing canonical markers with cluster-specific markers. After cell type identification, cell type proportions were calculated with the number of cells in each cell type divided by the total number of cells in each sample. To see the effect of UK5099 in the luminal cluster, DotPlot in Seurat was used to visualize the expression of luminal markers, basal markers, glycolytic enzymes, lipid metabolism genes and inflammatory signalling genes.

Cell lysis and western blot

Primary basal and luminal cells were sorted and immediately lysed in RIPA buffer (50 mM Tris-HCl pH 8.0, 150 mM NaCl, 1% NP-40, 0.5% sodium deoxycholate, 0.1% SDS, Fisher Scientific) containing a cOmplete protease inhibitor cocktail tablet (Roche) and Halt Phosphatase Inhibitor (Fisher Scientific). Organoids were removed from Matrigel by incubating in Advanced DMEM/F-12 (Gibco) containing 1 mg ml−1 dispase (Gibco) and 10 μM RI for 1 h at 37 °C. After centrifugation at 800g for 5 min, the pellet was washed with 1 × PBS and immediately lysed in RIPA buffer containing a cOmplete protease inhibitor cocktail tablet and Halt Phosphatase Inhibitor. For tumour lysis, tumour tissue was added to a bead tube (Fisher, 15-340-153) containing 1 ml of RIPA buffer containing a cOmplete protease inhibitor cocktail tablet and Halt Phosphatase Inhibitor on ice. Samples were homogenized for 1 min at maximum speed twice on a bead homogenizer (Fisher). Bead tubes were spun at 17,000g at 4 °C for 10 min. The supernatant was transferred to an Eppendorf tube and spun at 17,000g at 4 °C for 10 min. Each sample was sonicated for 40 s at 20 kHz with a sonic dismembrator (Fisher Scientific) to improve membranous and nuclear protein yield. Samples were run on NuPAGE 4–12% Bis-Tris Gel (Invitrogen) and transferred onto PVDF membranes (Millipore Sigma). Total protein was visualized using SYPRO RUBY protein blot stain (Fisher Scientific) and membranes were blocked in PBS + 0.1% Tween-20 (Fisher Scientific) + 5% milk (Fisher Scientific). Proteins were probed with primary antibodies followed by chromophore-conjugated anti-mouse (Invitrogen A21235, 1:1,000) or anti-rabbit secondary antibodies (Invitrogen A21244, 1:1,000) or HRP-conjugated anti-mouse (Thermo 31430, 1:10,000) or anti-rabbit secondary antibodies (Thermo 31463, 1:10,000) and detected by florescence or HRP chemiluminescence, respectively. Primary antibodies used were anti-Cytokeratin 5 (Biolegend 905504, 1:3,000), anti-Probasin (Santa Cruz sc-393830, 1:1,000), anti-Glut1 (Abcam ab115730, 1:10,000), anti-Glut3 (Abcam ab191071, 1:1,000), anti-Hexokinase 2 (Cell Signaling 28675, 1:1,000), anti-Phosphofructokinase (Abcam ab204131, 1:5000), anti-Pyruvate carboxylase (Abcam ab128952, 1:1,000), anti-Pyruvate dehydrogenase E1 component subunit alpha (Proteintech 18068-1-AP, 1:1,000), anti-Aconitase 2 (Abcam ab110321, 1:1,000), anti-Histone H3 (Cell Signaling 9717S, 1:1,000), anti-Cytokeratin 8 (Biolegend 904804, 1:1,000), anti-p63 (Biolegend 619002, 1:1,000), anti-beta Actin (Fisher MA1-140, 1:15,000), anti-Proliferating cell nuclear antigen (Fisher 13-3900, 1:1,000), anti-Androgen receptor (Abcam ab133273, 1:1,000), anti-Mitochondrial pyruvate carrier 1 (Cell Signaling 14462, 1:1,000), anti-Ki-67 (Abcam ab15580, 1:1,000), anti-cleaved caspase-3 (Cell Signaling 9661L, 1:500), anti-Cytokeratin 18 (Fisher MA5-12104, 1:100), anti-Vinculin (Abcam Ab129002, 1:1,000), anti-Phosphatase and tensin homologue (Cell Signaling 9559, 1:1,000), anti-Retinoblastoma protein 1 (Abcam ab181616, 1:1,000), anti-Acetyl-histone H3 (Lys9) (Cell Signaling 9649, 1:1,000), anti-Pan-acetyl histone H3 (Active Motif 61637, 1:1,000), anti-Histone H4 (Abcam ab10158, 1:1,000), anti-Pan-acetyl histone H4 (Abcam ab177790, 1:1,000), anti-prostate-specific antigen (Cell Signaling 5877, 1:1,000), anti-neuron-specific enolase (Proteintech 66150-1-Ig, 1:3,000), anti-synaptophysin (Cell Signaling 5461, 1:1,000) and anti-Sox2 (Cell Signaling 14962, 1:1,000).

Apoptosis assay

Cell culture media and wash media were collected and pooled with quenched trypsin-containing media containing cells and apoptosis analysis was performed using an apoptosis detection kit (BioLegend, 640922) according to manufacturer instructions. Flow cytometry was performed to quantify the percentage of annexin V−, 7-AAD− cells.

Primary cell metabolic profiling and nutrient tracing

Twelve-well plates were coated with a 1/80 dilution of growth factor reduced Matrigel (Corning) in RPMI 1640 (Gibco) to enhance cell attachment. The 1/80 Matrigel coating was aspirated before primary basal and luminal cells were seeded at a density of 200,000 cells per well and 140,000 cells per well, respectively. Cells were cultured overnight in mouse organoid media66 containing [U-13C]glucose (Cambridge Isotope Laboratories). Before metabolite extraction, tracer-containing medium was aspirated and cells were washed with cold 150 mM ammonium acetate pH 7.3. Metabolite extractions were performed by adding 500 μl of cold 80% methanol to each well and removing cells using a cell scraper. The cell suspension was transferred to an Eppendorf tube and 10 μl of 1 mM norvaline (Sigma) was added as an internal standard. Each sample was vortexed for 30 s and centrifuged at 17,000g for 5 min at 1 °C. Then, 420 μl of the supernatant was transferred to an ABC vial (Fisher Scientific) and evaporated using an EZ-2Elite evaporator (Genevac). Samples were stored at −80 °C before analysis.

The liquid chromatography separation using an Ion Chromatography System (ICS) 5000 (Thermo Scientific) was performed on a Dionex IonPac AS11-HC-4μm anion exchange column. The gradient was 5–95 mM KOH over 13 min, followed by 5 min at 95 mM, before re-equilibration to 5 mM. Other liquid chromatography parameters: flow rate 350 µl min−1, column temperature 35 °C, injection volume 5 μl. The Q Exactive mass spectrometer (Thermo Scientific) was operated in negative ion mode for detection of metabolites using a resolution of 70,000 at m/z 200 and a scan range of 70–900 m/z. Data were extracted using Tracefinder 3.1 (Thermo Scientific). Metabolites were identified based on accurate mass (±5 ppm) and previously established retention times of pure standards.

Normalization was performed by resuspending the cell pellet in 300 μl of lysis solution (0.1 M NaCl, 20 mM Tris-HCl, 0.1% SDS, 5 mM EDTA in distilled water). Samples were syringed with a 25 G needle to reduce viscosity and 50 μl of each sample was transferred to a 96-well black-wall, clear-bottom tissue culture plate (Corning). We added 50 μl of lysis solution to one well for a blank reading. Then, 100 μl of 5 μg ml−1 Hoechst 33342 (Invitrogen) in distilled water was added to each well and 96-well plates were incubated for 30 min in the dark at 37 °C before measurement of DNA-based florescence using a Tecan Infinite M1000 plate reader with 355-nm excitation and 465-nm emission. The blank reading was subtracted from each absorbance value to calculate relative cell amount.

In vivo basal-to-luminal differentiation RNA-seqCell preparation from postnatal prostates

Prostate tissue of mice at P10–P12 was microdissected under a stereoscope to separate the different lobes. The ventral lobe was used to further separate (by cutting) the tips from the main ducts. The ventral lobes of 20 mice at P10–P12 were used. Tissues were collected in 24-well plates and chopped. Minced tissues were digested in 5 mg ml−1 Collagenase Type I (Sigma-Aldrich, diluted in HBSS) for 2 h at 37 °C under agitation. Physical dissociation using a P-1000 pipette was performed every 20 min throughout the enzymatic digestion. Collagenase activity was blocked by adding EDTA (5 mM) for 2 min, followed by 0.25% Trypsin-EDTA for 5 min. Cells were rinsed in HBSS supplemented with 10% FBS and the cell suspensions were filtered through 70-μm cell strainers (BD Bioscience), followed by two successive washes in HBSS supplemented with 2% FBS.

Cell labelling, flow cytometry and sorting from postnatal prostates

Samples were incubated in 200 μl of PBS supplemented with 2% FBS with fluorochrome-conjugated antibodies for 30 min on ice protected from light, with shaking every 10 min. Antibodies were washed with 2% FBS/PBS and cells were resuspended in 2.5 mg ml−1 DAPI (Invitrogen, D1306) before analysis. The following antibodies were used: PE-conjugated anti-CD45 (rat, clone 30-F11, dilution 1:100, BD Biosciences Cat. no. 553081), PE-conjugated anti-CD31 (rat, clone MEC 13.3, dilution 1:100, BD Biosciences Cat. no. 553373), PE-conjugated anti-CD140a (rat, clone APA5, dilution 1:100, BD Biosciences Cat. no. 624049), APC-conjugated anti-CD49f (rat, clone GoH3, dilution 1:100, eBioscience Cat. no. 17-0495), APC-Cy7-conjugated anti-EpCAM (rat, clone G8.8, dilution 1:100, BioLegend Cat. no. 118218). Living cells were selected by forward and side scatter, doublets discriminating and DAPI dye exclusion. CD45+, CD31+ and CD140a+ cells were excluded (Lin+) before analysis of the epithelial basal (EpCamhigh/CD49fhigh) and luminal (EpCamhigh/CD49flow) cells. FACS and analysis were performed using FACSAria and LSRFortessa, using FACSDiva software (BD Bioscience). Sorted cells were collected in lysis buffer for RNA extraction (RLT buffer, QIAGEN). The following samples were collected in replicates from P10–P12 prostates: 1,702 basal cells and 1,626 luminal cells from the tips.

RNA extraction and RNA-seq

These methods apply to Fig. 2b and Extended Data Fig. 3a. RNA extraction from FACS isolated cells was performed using RNeasy micro kit (QIAGEN) according to the manufacturer’s recommendations. Before sequencing, the quality of RNA was evaluated by Bioanalyzer 2100 (Agilent). Indexed cDNA libraries were obtained using the Ovation Solo RNA-seq Systems (NuGen) following the manufacturer’s recommendations. The multiplexed libraries (11 pM/18 pM) were loaded on flow cells and sequences were produced using a NovaSeq 6000 S2 Reagent Kit (200 cycles from Novaseq 6000 System, Illumina) on a NovaSeq 6000 System (Illumina). Reads were mapped against the mouse reference genome (Grcm38/mm10) using STAR software to generate read alignments for each sample. Annotations for Mus_musculus.GRCm38.87.gtf were obtained from ftp.Ensembl.org. After assembling transcripts, gene level counts were obtained using HTseq and normalized to 20 million aligned reads. Genes with individual sample expression levels lower than 10 and replicate average abundance levels lower than 20 were filtered out. The fold changes of average gene abundance for the replicates were used to calculate the level of differential gene abundance between different cell populations. Genes with a fold change greater than or equal to 2 were considered upregulated and those with a fold change lower than or equal to 0.5 were considered downregulated.

Organoid culture of primary basal- and luminal-derived mouse organoids

Basal cells were plated at a density of 1,000 cells per well and luminal cells were plated at a density of 20,000 cells per well. Growth factor reduced Matrigel (Corning) was added to the cell suspension at a final concentration of 75% before plating into rings in 24-well plates.

Viral vectors

A Cre recombinase cassette was inserted into a red fluorescent protein-expressing FU-CRW lentivirus vector70 by restriction digestion and ligation at the EcoRI site to make FU-Cre-CRW. Insertion of the Cre cassette was confirmed by DNA sequencing. Concentrated viral preps of FU-Cre-CRW and FU-CRW were made by the University of California, Los Angeles Vector Core facility and the Cre recombinase activity was validated by infecting HEK 293T cells transduced with a Cre-reporter plasmid (Addgene no. 62732)71.

Generation of Mpc1-KO, Pten SKO and Pten;Rb1 DKO organoids

Basal cells were isolated from Mpc1fl/fl, Ptenfl/fl and Ptenfl/fl;Rb1fl/fl mouse prostates and infected with RFP (FU-CRW) or Cre-RFP (FU-Cre-CRW) lentivirus. Lentiviral spinfections were done by culturing the cells with virus in 200 μl of RPMI 1640 (Gibco) containing 10% FBS (Corning), 1 × P/S and 10 μM RI (RPMI, 10% FBS, 1% P/S + RI) plus 8 μg ml−1 polybrene for 30 min at 37 °C then spinning at 300g for 90 min. After spinfection, growth factor reduced Matrigel (Corning) was added to the cell suspension at a final concentration of 75% before plating into rings in 24-well plates. After 1 week of culture, organoids were dissociated to single cells. Organoids were removed from Matrigel by incubating in Advanced DMEM/F-12 (Gibco) containing 1 mg ml−1 dispase (Gibco) and 10 μM RI for 1 h at 37 °C. After centrifugation at 800g for 5 min, the pellet was washed with 1 × PBS. Organoids were resuspended in 800 μl of 0.05% Trypsin-EDTA (Gibco) and incubated at 37 °C for 5 min. The trypsin was quenched with 200 μl of RPMI, 10% FBS, 1% P/S + RI and organoids were pipetted up and down ten times to dissociate to single cells and passed through a 100-μm cell strainer (Corning). After centrifugation at 800g for 5 min, the pellet was washed with 1 × PBS and resuspended in RPMI, 10% FBS, 1% P/S + RI. RFP-positive cells were isolated by FACS. All prostate organoids were cultured based on established protocols66,72. Single organoids were imaged on a light microscope and organoid diameter was measured as a readout of organoid size.

Organoid metabolic profiling and nutrient tracing

For glucose tracer analysis experiments, 17.5 mM [U-13C]glucose (Cambridge Isotope Laboratories) was added to glucose-free SILAC Advanced DMEM/F-12 Flex Media (Fisher Scientific). Arginine, lysine and alanine were also added back to the SILAC base media at the same concentrations found in Advanced DMEM/F-12 (Fisher Scientific). Organoids were grown in mouse organoid media made with the SILAC base media. For lactate tracer analysis experiments, organoids were cultured with 20 mM [U-13C]Lactate (Cambridge Isotope Laboratories, CLM-1579-0.5) for 24 h before metabolite extraction. To extract metabolites, tracer-containing medium was aspirated. Organoids were repeatedly blasted with cold 150 mM ammonium acetate pH 7.3 using a P-1000 pipette until the Matrigel ring was dislodged. The suspension was transferred to an Eppendorf tube and centrifuged at 800g for 5 min at 1 °C. The supernatant was aspirated and 500 μl of cold 80% methanol was added to the organoid pellet. We added 10 μl of 1 mM norvaline (Sigma) as an internal standard. Each sample was vortexed for 30 s and centrifuged at 17,000g for 5 min at 1 °C. We transferred 420 μl of the supernatant to an ABC vial (Fisher Scientific) and evaporated using an EZ-2Elite evaporator (Genevac). Samples were stored at −80 °C before analysis.

Dried metabolites were resuspended in 50% acetonitrile:water and one-tenth was loaded onto a Luna 3μm NH2 100A (150 × 2.0 mm2) column (Phenomenex). The chromatographic separation was performed on a Vanquish Flex (Thermo Scientific) with mobile phases A (5 mM NH4AcO pH 9.9) and B (acetonitrile) and a flow rate of 200 μl min−1. A linear gradient from 15% A to 95% A over 18 min was followed by 9 min of isocratic flow at 95% A and re-equilibration to 15% A. Metabolites were detected with a Thermo Scientific Q Exactive mass spectrometer run with polarity switching (+3.5 kV/−3.5 kV) in full scan mode with an m/z range of 70–975 and 70,000 resolution. TraceFinder 4.1 (Thermo Scientific) was used to quantify the targeted metabolites by area under the curve using expected retention time and accurate mass measurements (<5 ppm). For labelled datasets, relative amounts of metabolites were calculated by summing up the values for all isotopologues of a given metabolite. Metabolite isotopologue distributions were corrected for natural C-13 abundance.

Normalization was performed by resuspending the cell pellet in 300 μl of lysis solution (0.1 M NaCl, 20 mM Tris-HCl, 0.1% SDS, 5 mM EDTA, 500 μg ml−1 Proteinase K (Fisher Scientific) in distilled water). Samples were syringed with a 25 G needle to reduce viscosity and 50 μl of each sample was transferred to a 96-well black-wall, clear-bottom tissue culture plate (Corning). We added 50 μl of lysis solution to one well for a blank reading. Then, 100 μl of 5 μg ml−1 Hoechst 33342 (Invitrogen) in distilled water was added to each well and 96-well plates were incubated for 30 min in the dark at 37 °C before measurement of DNA-based florescence using a Tecan Infinite M1000 plate reader with 355-nm excitation and 465-nm emission. The blank reading was subtracted from each absorbance value to calculate relative cell amount.

PCA

PCA was performed using metabolite abundance and fractional contribution data in the Python programming language (v.3.9.12). Data were processed using the NumPy (v.1.22.4), pandas (v.1.4.2) and scikit-learn (v.1.0.2) libraries and visualized using the Matplotlib library (v.3.5.1). Feature scaling was done along the metabolite dimension using the StandardScaler class from scikit-learn which employs z-score normalization. The 95% confidence ellipses were generated with a script provided by Matplotlib.

Intracellular flow cytometry

Organoids were removed from Matrigel by incubating in Advanced DMEM/F-12 (Gibco) containing 1 mg ml−1 dispase (Gibco) and 10 μM RI for 1 h at 37 °C. After centrifugation at 800g for 5 min, the pellet was washed with 1 × PBS. Organoids were resuspended in 800 μl of 0.05% Trypsin-EDTA (Gibco) and incubated at 37 °C for 5 min. The trypsin was quenched with 200 μl of RPMI, 10% FBS, 1% P/S + RI and organoids were pipetted up and down ten times to dissociate to single cells and passed through a 100-μm cell strainer (Corning). Dissociated cells from mouse prostate organoids were washed with PBS and fixed in 1 ml of 2% paraformaldehyde made from 16% paraformaldehyde (Electron Microscopy Sciences) in PBS for 15 min on ice. For experiments including EpCAM surface staining, cells were stained with EpCAM-APC/Cy7 (BioLegend 118218, 1:100) in RPMI, 10% FBS, 1% P/S + RI for 15 min before fixation. Cells were then washed with PBS and permeabilized in 1 ml of permeabilization buffer (0.1% Saponin (Sigma-Aldrich), 5% FBS (Corning) in PBS) for 15 min at room temperature in the dark. Cells were resuspended in 100 ml of permeabilization buffer and stained with rabbit anti-cytokeratin 5-Alexa Fluor 647 (Abcam Ab193895, 1:100) and rabbit anti-cytokeratin 8-Alexa Fluor 488 (Abcam Ab192467, 1:100) for 20 min at room temperature in the dark. Cells were washed with permeabilization buffer and resuspended in PBS for analysis on a BD FACS Canto (BD Biosciences).

Organoid immunofluorescence

Organoids were removed from Matrigel by incubating in Advanced DMEM/F-12 (Gibco) containing 1 mg ml−1 dispase (Gibco) and 10 μM RI for 1 h at 37 °C. After centrifugation at 800g for 5 min, the pellet was washed with 1 × PBS three times. Organoids were then fixed in 4% paraformaldehyde in PBS for 15 min. After fixation, organoids were washed with PBS three times. Organoids were then blocked in 2% donkey serum in 0.25% Triton X-100 for 1 h. Organoids were washed once with PBS and stained with anti-Cytokeratin 8 (Biolegend 904804, 1:500) antibody and anti-p63 (Biolegend 619002, 1:500) antibody in 0.5% BSA, 0.25% Triton X-100 at 4 °C overnight. Organoids were then washed with PBS three times, with the last wash lasting 6 h. Secondary antibody staining was performed overnight at 4 °C using goat anti-rabbit IgG-AlexaFluor647 (Thermo Fisher 21245, 1:1,000) and goat anti-mouse IgG-AlexaFluor488 (Thermo Fisher 11001, 1:1,000) in 0.5% BSA, 0.25% Triton X-100 with one drop of NucBlue. Organoids were washed with PBS three times and placed in PBS + 0.1% Tween-20 until imaging on a Nikon Ti-E Fluorescence Motorized DIC Microscope (Nikon) with RCM1 confocal box (Confocal.nl) using Nikon NIS Elements Imaging Software and Nikon CFI Apo LWD Lambda S 20XC WI objective, material number MRD77200.

Cell lines

Cell lines were routinely tested for mycoplasma and authentication by short tandem repeat analysis (Laragen). Tissue culture plates were coated with 0.01% (v/v) poly-l-lysine (Sigma, P4832) diluted 1/20 in distilled water and washed with PBS to enhance cell attachment. 16D cells were received from Dr Amina Zoubeidi and cultured in RPMI base media (Gibco) + 10% FBS (v/v) + 1 × P/S. LuCaP35 cells were received from Dr Eva Corey and Dr Peter Nelson and cultured in DMEM base media (Gibco, 11965-092) + 10% FBS (v/v) + 1 × P/S and 1 × GlutaMAX. LAPC4 cells were received from Dr Rob Reiter and cultured in IMDM (Gibco, 31980-030) + 5% FBS (v/v) + 1 × P/S. UK5099 treatment was performed by adding 10 μM or 30 μM UK5099 (Sigma, PZ0160) every 48 h.

Histone extractions

Histone extractions were performed using a histone extraction kit (Abcam, Ab113476) according to manufacturer instructions.

ATAC-seq

Cells were collected and frozen in culture media containing FBS and 5% dimethylsulfoxide. Cryopreserved cells were sent to Active Motif to perform the ATAC-seq assay. The cells were then thawed in a 37 °C water bath, pelleted, washed with cold PBS and tagmented as previously described73, with some modifications74. Briefly, cell pellets were resuspended in lysis buffer, pelleted and tagmented using the enzyme and buffer provided in the Nextera Library Prep Kit (Illumina). Tagmented DNA was then purified using the MinElute PCR purification kit (Qiagen), amplified with ten cycles of PCR and purified using Agencourt AMPure SPRI beads (Beckman Coulter). The resulting material was quantified using the KAPA Library Quantification Kit for Illumina platforms (KAPA Biosystems) and sequenced with PE42 sequencing on the NovaSeq 6000 sequencer (Illumina). Reads were aligned using the BWA algorithm (v.0.7.12; mem mode; default settings). Duplicate reads were removed; only reads mapping as matched pairs and uniquely mapped reads (mapping quality ≥ 1) were used for further analysis. Alignments were extended in silico at their 3′-ends to a length of 200 bp and assigned to bins 32 nucleotides in size along the genome. The resulting histograms (genomic ‘signal maps’) were stored in bigWig files. Peaks were identified using the MACS 2.1.0 algorithm at a cutoff of P value 1 × 10−7, without control file, and with the –nomodel option. Peaks that were on the ENCODE blacklist of known false ChIP–seq peaks were removed. Signal maps and peak locations were used as input data to Active Motif’s proprietary analysis program, which creates Excel tables containing detailed information on sample comparison, peak metrics, peak locations and gene annotations. For differential analysis, reads were counted in all merged peak regions (using Subread), and the replicates for each condition were compared using DESeq2 (v.1.24.0)54.

HOMER transcription factor motif analysisIdentification of sites with differential ATAC-seq signal

After identifying merged regions as part of the standard analysis pipeline, the DESeq2 software was run on the unnormalized BAM files (without duplicates). In brief, the DESeq2 software generates normalized counts specifically for the merged regions, and the shrunken log2 fold change and adjusted P values for each merged region are calculated. For the subsequent steps of the analysis, we consider any region as differential if the adjusted P value is less than 0.1.

HOMER-based motif analysis

BED files listing the significantly increased (‘DESeq2_Up_difpeaks.bed’) and decreased (‘DESeq2_Down_difpeaks.bed’) regions were generated for each comparison. Each BED file was then sorted by the shrunken log2 fold change and the 2,500 regions with the largest absolute fold changes were selected. We then performed HOMER motif analysis (findMotifsGenome.pl) on the 200-bp sequence centred around the midpoint of the differential region (+100 bp, −100 bp). During this analysis, common repeats are masked as this can affect the discovery of de novo motifs. The analysis identifies motifs that are enriched across all sequences; individual peak regions are not annotated with specific motifs.

PDX enzalutamide sensitivity assay

Using a razor blade, MDA PCa 203-A PDX and MDA PCa 183-A PDX tumours were mechanically dissociated in dissociation media composed of RPMI 1640 (Gibco) containing 10% FBS (Corning), 1 × P/S, 1 mg ml−1 collagenase type I (Gibco), 1 mg ml−1 dispase (Gibco), 0.1 mg ml−1 deoxyribonuclease (Gibco) and 10 μM RI. When large chunks were no longer visible, the samples were incubated at 37 °C on a nutating platform for 15 min in 20 ml of dissociation media. After centrifugation at 800g for 5 min, the pellet was washed with 1 × PBS (Gibco). The cell pellet was resuspended in Advanced DMEM/F-12 and passed through a 100-μm cell strainer (Corning). After centrifugation at 800g for 5 min, the pellet was resuspended in human organoid media and plated in 75% growth factor reduced Matrigel (Corning) based on established protocols66. After 7 d of culture with vehicle, 10 μM UK5099 (Sigma, PZ0160) or 20 mM sodium lactate (Sigma, L7022-5G), organoids were removed from Matrigel by incubating in Advanced DMEM/F-12 containing 1 mg ml−1 dispase and 10 μM of the p160ROCK inhibitor Y-27632 dihydrochloride for 1 h at 37 °C. After centrifugation at 800g for 5 min, the pellet was washed with 1 × PBS. Organoids were then plated into rings in a 96-well black-wall, clear-bottom plate (Fisher, 07-200-588) in 75% growth factor reduced Matrigel with or without 10 μM enzalutamide (Selleck Chemicals, S1250). After 5 d of culture, a CellTiter-Glo assay (Promega, G7571) was performed according to manufacturer instructions and relative luminescence signal was quantified on a Tecan Infinite M1000 plate reader.

Cell line enzalutamide sensitivity assays

After 7 d of culture with vehicle, 10 μM UK5099 (Sigma, PZ0160), 30 μM UK5099 or 20 mM sodium lactate (Sigma, L7022-5G), cells were plated into a 96-well black-wall, clear-bottom plate (Fisher, 07-200-588) with or without 10 μM enzalutamide (Selleck Chemicals, S1250). The 96-well plate was coated with 0.01% (v/v) poly-l-lysine (Sigma, P4832) diluted 1/20 in distilled water and washed with PBS before plating cells to enhance cell attachment. After 2 d of culture, a CellTiter-Glo assay (Promega, G7571) was performed according to manufacturer instructions and relative luminescence signal was quantified on a Tecan Infinite M1000 plate reader. For the 5-ethynyl-2′-deoxyuridine-based (EdU) cell cycle assay, cells were seeded at 30% confluence and cultured in 6-well dishes for 72 h before cell cycle analysis. Media changes were performed 48 h after plating. After 72 h of culture, cell cycle analysis was performed using an EdU kit (Thermo Fisher Scientific, C10635) according to the specified protocol. EdU labelling was performed for 2 h. For experiments that contained small-molecule inhibitors, fresh inhibitor(s) were adding during each media change. Flow cytometry analysis identified the percentage EdU-positive.

Statistics and reproducibility

Prism v.8.3.0 (GraphPad) was used to generate graphs and perform statistical analyses. All in vitro experiments shown were repeated at least three times with similar results obtained, and representative data are shown unless otherwise indicated. No statistical method was used to predetermine sample size but our sample sizes are similar to those reported in previous publications21. Data distribution was assumed to be normal but this was not formally tested. No data were excluded from the analyses. Data collection and analysis were not performed blind to the conditions of the experiments. For animal experiments, mice were randomly divided into cages. For in vitro experiments, samples were not randomized as this was not relevant for the individual assays.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.