Student or educator? Sign up with your school email and get 6 months of Pro free.
CoChat

The DOI matches its Crossref record, with a small difference in title or year; 100 of the 101 references it cites were found in a registry or an index; 56 of the 77 sources its claims cite were checked for support.

Genome-scale CRISPR screens identify PTGES3 as a direct modulator of androgen receptor function in advanced prostate cancer

Open access

Abstract

The androgen receptor (AR) is a critical driver of prostate cancer (PCa). Here, to study regulators of AR protein levels and oncogenic activity, we developed a live-cell quantitative endogenous AR fluorescent reporter. Leveraging this AR reporter, we performed genome-scale CRISPRi flow cytometry sorting screens to systematically identify genes that modulate AR protein levels. We identified and validated known AR protein regulators, including HOXB13 and GATA2, and also unexpected top hits including PTGES3-a poorly characterized gene in PCa. PTGES3 repression resulted in loss of AR protein, cell-cycle arrest and cell death in AR-driven PCa models. Clinically, analysis of PCa data demonstrates that PTGES3 expression is associated with AR-directed therapy resistance. Mechanistically, we show PTGES3 binds directly to AR, regulates AR protein stability and is necessary for AR function in the nucleus at AR target genes. PTGES3 represents a potential therapeutic target for overcoming known mechanisms of resistance to existing AR-directed therapies in PCa.

Overview

CoChat Summary

## TL;DR

The paper identifies PTGES3 as a direct regulator of androgen receptor (AR) protein levels and AR-dependent transcription in advanced prostate cancer using an endogenous AR fluorescent reporter and genome-scale CRISPRi screens. PTGES3 loss reduces AR protein, impairs AR function, and delays tumor growth, while higher PTGES3 expression is associated with worse clinical outcomes and AR-directed therapy resistance.

## Key findings

  • A live-cell endogenous AR mNeonGreen2 reporter was built in C42B prostate cancer cells and used for genome-scale CRISPRi flow-sorting screens; the screen recovered known AR regulators such as HOXB13 and GATA2 and nominated PTGES3 as a top hit.
  • PTGES3 repression reduced AR protein in multiple AR-driven models, including C42B, 22RV1, VCaP, and enzalutamide-resistant MR49F cells, and lowered AR splice variant 7 in 22RV1 cells.
  • In clinical samples, PTGES3 protein was present in cytosolic and nuclear compartments in a tissue microarray of 120 PCa biopsies; high nuclear PTGES3 was associated with worse PSA-recurrent free survival, with hazard ratio = 2.655 and log-rank P = 0.0014.
  • In a cohort of 641 prostatectomy samples, PTGES3 overexpression was associated with significantly worse metastasis-free survival in patients who received adjuvant ADT after prostatectomy, and higher tumor PTGES3 expression was correlated with worse overall survival in mCRPC patients treated with first-line AR-targeted therapy.

## Methods

  • The study combined an endogenous AR split-fluorescent reporter with genome-scale CRISPRi screens in two C42BmNG2-AR clonal cell lines; cells were sorted into the top and bottom quartiles of AR fluorescence and sgRNA abundances were sequenced to identify AR regulators.
  • Follow-up assays included qPCR, western blotting, co-IP, proximity ligation, luciferase reporter assays, mass spectrometry, ChIP–qPCR/ChIP–seq, ATAC-seq, and in vivo xenograft experiments in male NSG mice with inducible PTGES3 repression.
  • Clinical association analyses used a 120-sample prostate tumor tissue microarray and public expression data from 641 prostatectomy samples plus mCRPC cohorts; survival was analyzed with Kaplan–Meier, log-rank tests, and Cox proportional hazards regression.

## Limitations

  • The paper states that no statistical methods were used to predetermine sample sizes, and that data distribution was assumed to be normal but not formally tested.
  • The text makes evident that several assays were performed in cell-line and mouse models, so direct clinical efficacy of PTGES3 targeting in patients is not established (evident from the study design).
  • The text also makes evident that some results are associative in patient cohorts rather than causal, for example the PTGES3 expression-survival analyses (evident from the clinical cohort analyses).

Model-written summary. Check it against the paper before relying on it.

Integrity signals

Registry

Crossref · 10.1038/s41588-025-02388-8

Cites

100 of 101 references verified0 citation errors · 1 not checkable

Statistics

Nothing to recompute.

Figures

Fig. 1 Establishment of an endogenous AR mNG2 fluorescent reporter in PCa cells. a, Diagram of the self-assembling split mNG2 fluorescence protein AR tagging approach. A DNA sequence encoding 16-amino-acid 11th β-strand of the fluorescent protein NeonGreen (mNG2_11) is knocked into the N terminus of AR by cotransfection of a homology-directed repair DNA donor and Cas9 RNPs targeting the AR locus. When the remainder of mNG2 (mNG2_1-10) is expressed in trans in the same cell, it binds noncovalently to mNG2_11, producing a fluorescent mNG2-AR protein complex. b, mNG2 IP followed by western blotting for AR shows formation of mNG2-AR protein complex in mNG2-AR endogenously tagged cells but not in controls. c, C42BmNG2-AR cells were treated with DMSO, 10 nM R1881 (androgen), 10 nM R1881 plus 10 μM Enz (anti-androgen) or 100 nM ARD-61 (AR degrader, PROTAC) for 24 h. Representative confocal microscopic images showed mNG2-AR localization (green). Nuclear fluorescence intensity was quantified using ImageJ. Ten cells per experimental condition were collected from different fields (n = 10; mean ± s.d.). Statistical significance was determined using two-sided t-test (***P < 0.001). d, C42BmNG2-AR cells were treated with DMSO (CTRL, green), 10 nM R1881 (brighter green) or 100 nM ARD-61 (gray) for 24 h and then mNG2-AR fluorescence was analyzed by flow cytometry. e, C42B and C42BmNG2-AR cells were treated with transcription inhibitor 1 μM actinomycin D for 0 h, 8 h or 16 h. RNA was collected to measure AR mRNA levels using real-time PCR (n = 3 as biological replicates; mean ± s.d.; NS, no significant difference by two-way ANOVA analysis). f, C42B and C42BmNG2-AR cells were treated with protein synthesis inhibitor cycloheximide (CHX, 5 μM) with or without proteasome inhibitor 5 μM MG132 for 0 h, 12 h or 24 h. AR protein levels were detected by western blotting. g, C42B and C42BmNG2-AR cells were transfected with androgen response element (ARE)-Luc to enable measurement of AR transcriptional activity. Cells were then treated with DMSO (gray) or 10 nM R1881 (green). ARE-Luc over Renilla was normalized by control (n = 3 as biological replicates; mean ± s.d.). Statistical significance was determined using two-sided t-test (**P < 0.01). h, C42B and C42BmNG2-AR cells were fixed by formaldehyde. ChIP experiments were performed using an IgG (gray) or AR antibody (green). Precipitated DNA fragments were used as templates to amplify the PSA enhancer and TMPRSS2 promoter by real-time PCR (n = 3 as biological replicates; mean ± s.d.). Statistical significance was determined using two-sided t-test (***P < 0.001; **P < 0.01). i, Total RNA from C42B and C42BmNG2-AR cells were collected for RNA-seq. Gene expression values were calculated as ln(TPM + 0.1). Pearson correlation was calculated comparing AR pathway genes57 (n = 22) between the two cell lines. Panel a created with BioRender.com.
Fig. 1 Establishment of an endogenous AR mNG2 fluorescent reporter in PCa cells. a, Diagram of the self-assembling split mNG2 fluorescence protein AR tagging approach. A DNA sequence encoding 16-amino-acid 11th β-strand of the fluorescent protein NeonGreen (mNG2_11) is knocked into the N terminus of AR by cotransfection of a homology-directed repair DNA donor and Cas9 RNPs targeting the AR locus. When the remainder of mNG2 (mNG2_1-10) is expressed in trans in the same cell, it binds noncovalently to mNG2_11, producing a fluorescent mNG2-AR protein complex. b, mNG2 IP followed by western blotting for AR shows formation of mNG2-AR protein complex in mNG2-AR endogenously tagged cells but not in controls. c, C42BmNG2-AR cells were treated with DMSO, 10 nM R1881 (androgen), 10 nM R1881 plus 10 μM Enz (anti-androgen) or 100 nM ARD-61 (AR degrader, PROTAC) for 24 h. Representative confocal microscopic images showed mNG2-AR localization (green). Nuclear fluorescence intensity was quantified using ImageJ. Ten cells per experimental condition were collected from different fields (n = 10; mean ± s.d.). Statistical significance was determined using two-sided t-test (***P < 0.001). d, C42BmNG2-AR cells were treated with DMSO (CTRL, green), 10 nM R1881 (brighter green) or 100 nM ARD-61 (gray) for 24 h and then mNG2-AR fluorescence was analyzed by flow cytometry. e, C42B and C42BmNG2-AR cells were treated with transcription inhibitor 1 μM actinomycin D for 0 h, 8 h or 16 h. RNA was collected to measure AR mRNA levels using real-time PCR (n = 3 as biological replicates; mean ± s.d.; NS, no significant difference by two-way ANOVA analysis). f, C42B and C42BmNG2-AR cells were treated with protein synthesis inhibitor cycloheximide (CHX, 5 μM) with or without proteasome inhibitor 5 μM MG132 for 0 h, 12 h or 24 h. AR protein levels were detected by western blotting. g, C42B and C42BmNG2-AR cells were transfected with androgen response element (ARE)-Luc to enable measurement of AR transcriptional activity. Cells were then treated with DMSO (gray) or 10 nM R1881 (green). ARE-Luc over Renilla was normalized by control (n = 3 as biological replicates; mean ± s.d.). Statistical significance was determined using two-sided t-test (**P < 0.01). h, C42B and C42BmNG2-AR cells were fixed by formaldehyde. ChIP experiments were performed using an IgG (gray) or AR antibody (green). Precipitated DNA fragments were used as templates to amplify the PSA enhancer and TMPRSS2 promoter by real-time PCR (n = 3 as biological replicates; mean ± s.d.). Statistical significance was determined using two-sided t-test (***P < 0.001; **P < 0.01). i, Total RNA from C42B and C42BmNG2-AR cells were collected for RNA-seq. Gene expression values were calculated as ln(TPM + 0.1). Pearson correlation was calculated comparing AR pathway genes57 (n = 22) between the two cell lines. Panel a created with BioRender.com.CC BY · PubMed Central
Fig. 2 Identification of regulators of AR protein levels using a genome-scale CRISPRi screen. a, Schematic of the FACS-based genome-scale CRISPRi screening approach in C42BimNG2-AR cells to identify genes that regulate AR protein levels (Methods). b, A volcano plot of genome-scale CRISPRi screen hit genes (light purple) that regulate AR protein levels58 (Methods). Negative controls are shown in orange. A comparable number of negative control (NC) genes were generated by randomly sampling five nontargeting sgRNAs (with replacement) and analyzed as true genes (n = ~19,000). Empirically derived thresholds (log10 P × phenotype z ≥ 13; dashed lines) were calculated as shown, using the NC gene distribution to derive the background s.d. for z-scoring. Only a limited number of hits were labeled due to the size of the plot. c, A plot showing top positive AR regulator hit (n = 51) protein interaction networks analyzed by STRING. Proteins without protein interactions among any of the top 51 hits were removed. Lines between the proteins indicate a confident interaction. Line thickness indicates the strength of data support. Proteins were also unsupervised clustered, indicated by their color. d, C42BimNG2-AR were infected with sgRNA targeting control or indicated hit genes. After puromycin selection, median fluorescence intensity (MFI) of mNG2-AR fluorescence from the treated cells was measured by flow cytometry. The percentage of AR protein abundance was calculated as AR protein abundance % = (MFIsg Individual gene − MFIC42B)/(MFIsgCTRL − MFIC42B) × 100 + 100 (n = 3 as biological replicates; mean ± s.d.). Statistical significance was determined using two-sided t-test (***P < 0.001; **P < 0.01; NS, not significant). e, C42Bi cells were infected with indicated individual sgRNAs targeting hit genes. Cell lysates were collected after puromycin selection. AR and GAPDH levels were detected by western blotting. Densitometry of the replicate western blots (n = 2) were calculated by ImageJ and normalized; mean ± s.d. are labeled under the lanes. f, C42B cells were treated with 100 nM ARD-61 (ARi), 5 μM ML792 (SAEi), siRNA targeting PTGES3 or 10 μM Slimitasertib (CSNK2A1i) for 24 h. Cell lysates were collected. AR and GAPDH levels were detected by western blotting. Panel a and the STRING plot in c created with BioRender.com. Source data
Fig. 2 Identification of regulators of AR protein levels using a genome-scale CRISPRi screen. a, Schematic of the FACS-based genome-scale CRISPRi screening approach in C42BimNG2-AR cells to identify genes that regulate AR protein levels (Methods). b, A volcano plot of genome-scale CRISPRi screen hit genes (light purple) that regulate AR protein levels58 (Methods). Negative controls are shown in orange. A comparable number of negative control (NC) genes were generated by randomly sampling five nontargeting sgRNAs (with replacement) and analyzed as true genes (n = ~19,000). Empirically derived thresholds (log10 P × phenotype z ≥ 13; dashed lines) were calculated as shown, using the NC gene distribution to derive the background s.d. for z-scoring. Only a limited number of hits were labeled due to the size of the plot. c, A plot showing top positive AR regulator hit (n = 51) protein interaction networks analyzed by STRING. Proteins without protein interactions among any of the top 51 hits were removed. Lines between the proteins indicate a confident interaction. Line thickness indicates the strength of data support. Proteins were also unsupervised clustered, indicated by their color. d, C42BimNG2-AR were infected with sgRNA targeting control or indicated hit genes. After puromycin selection, median fluorescence intensity (MFI) of mNG2-AR fluorescence from the treated cells was measured by flow cytometry. The percentage of AR protein abundance was calculated as AR protein abundance % = (MFIsg Individual gene − MFIC42B)/(MFIsgCTRL − MFIC42B) × 100 + 100 (n = 3 as biological replicates; mean ± s.d.). Statistical significance was determined using two-sided t-test (***P < 0.001; **P < 0.01; NS, not significant). e, C42Bi cells were infected with indicated individual sgRNAs targeting hit genes. Cell lysates were collected after puromycin selection. AR and GAPDH levels were detected by western blotting. Densitometry of the replicate western blots (n = 2) were calculated by ImageJ and normalized; mean ± s.d. are labeled under the lanes. f, C42B cells were treated with 100 nM ARD-61 (ARi), 5 μM ML792 (SAEi), siRNA targeting PTGES3 or 10 μM Slimitasertib (CSNK2A1i) for 24 h. Cell lysates were collected. AR and GAPDH levels were detected by western blotting. Panel a and the STRING plot in c created with BioRender.com. Source dataCC BY · PubMed Central
Fig. 3 PTGES3 is a regulator of AR. a, A plot showing proteomic changes upon repression of PTGES3 relative to controls; 22RV1i cells were infected with sgRNAs targeting CTRL (control) or PTGES3 and then subjected to puromycin selection. Cells were collected for label-free shotgun mass spectrometry (n = 3). Volcano plot showing the DE proteins in sgPTGES3 versus sgCTRL groups. b, A plot showing gene expression changes upon repression of PTGES3 relative to controls. LNCaPi cells were infected with sgRNAs targeting CTRL (nontargeting control) or PTGES3. Total RNA was collected for RNA-seq (n = 2 as biological replicates). DE genes between the two groups were highlighted in red in the volcano plot; top ranking DE genes were labeled. AR is not DE when PTGES3 is knocked down. c, C42B, Enzalutamide resistant cell lines MR49F, 22RV1 and VCaP were treated with siRNA targeting control or PTGES3. AR, PTGES3 and GAPDH levels were detected by western blotting. d, LNCaP cells were treated with siRNA targeting control or PTGES3 for 48 h, then stained with PI/Annexin V. The percentage of the apoptotic population was measured by FACS (n = 3 biological replicates; mean ± s.d.). Statistical significance was determined using two-sided t-test (***P < 0.001). e, LNCaP cells were treated with siRNA targeting control or PTGES3. cleaved PARP, PARP, cleaved Caspase-3, Caspase-3 and GAPDH levels were detected by western blotting. f, A diagram showing the potential dual functions of PTGES3 regulating AR. g, Left: diagrams illustrating the wildtype, enzymatic mutation (Y9N) and HSP90 mutation (W106A) isoforms of PTGES3. Right: LNCaP cells stably expressing a TET-ON inducible PTGES3 wildtype, Y9N mutation and W106A mutation were treated with or without 100 ng ml−1 doxycycline. Flag-PTGES3 and GAPDH levels were detected by western blotting. h, LNCaP (TET-ON PTGES3) expressing inducible flag-tagged PTGES3 WT, Y9N mutation or W106A mutation were treated with 100 ng ml−1 doxycycline. Co-IP assays were performed using flag antibody. GAPDH, Flag-PTGES3 and HSP90 amounts were detected by western blotting. i, LNCaPi cells were infected with sgPTGES3-Blast, then treated with DMSO or 100 ng ml−1 doxycycline to overexpress the PTGES3 WT or Y9N or W106A mutant proteins. AR and GAPDH levels were detected by western blotting. Panels f and g created with BioRender.com. Source data
Fig. 3 PTGES3 is a regulator of AR. a, A plot showing proteomic changes upon repression of PTGES3 relative to controls; 22RV1i cells were infected with sgRNAs targeting CTRL (control) or PTGES3 and then subjected to puromycin selection. Cells were collected for label-free shotgun mass spectrometry (n = 3). Volcano plot showing the DE proteins in sgPTGES3 versus sgCTRL groups. b, A plot showing gene expression changes upon repression of PTGES3 relative to controls. LNCaPi cells were infected with sgRNAs targeting CTRL (nontargeting control) or PTGES3. Total RNA was collected for RNA-seq (n = 2 as biological replicates). DE genes between the two groups were highlighted in red in the volcano plot; top ranking DE genes were labeled. AR is not DE when PTGES3 is knocked down. c, C42B, Enzalutamide resistant cell lines MR49F, 22RV1 and VCaP were treated with siRNA targeting control or PTGES3. AR, PTGES3 and GAPDH levels were detected by western blotting. d, LNCaP cells were treated with siRNA targeting control or PTGES3 for 48 h, then stained with PI/Annexin V. The percentage of the apoptotic population was measured by FACS (n = 3 biological replicates; mean ± s.d.). Statistical significance was determined using two-sided t-test (***P < 0.001). e, LNCaP cells were treated with siRNA targeting control or PTGES3. cleaved PARP, PARP, cleaved Caspase-3, Caspase-3 and GAPDH levels were detected by western blotting. f, A diagram showing the potential dual functions of PTGES3 regulating AR. g, Left: diagrams illustrating the wildtype, enzymatic mutation (Y9N) and HSP90 mutation (W106A) isoforms of PTGES3. Right: LNCaP cells stably expressing a TET-ON inducible PTGES3 wildtype, Y9N mutation and W106A mutation were treated with or without 100 ng ml−1 doxycycline. Flag-PTGES3 and GAPDH levels were detected by western blotting. h, LNCaP (TET-ON PTGES3) expressing inducible flag-tagged PTGES3 WT, Y9N mutation or W106A mutation were treated with 100 ng ml−1 doxycycline. Co-IP assays were performed using flag antibody. GAPDH, Flag-PTGES3 and HSP90 amounts were detected by western blotting. i, LNCaPi cells were infected with sgPTGES3-Blast, then treated with DMSO or 100 ng ml−1 doxycycline to overexpress the PTGES3 WT or Y9N or W106A mutant proteins. AR and GAPDH levels were detected by western blotting. Panels f and g created with BioRender.com. Source dataCC BY · PubMed Central
Fig. 4 AR and PTGES3 interact in the absence and presence of DNA. a, IHC for PTGES3 was performed on human PCa tissue microarrays (n = 120). Tissue cores were scored for intensity of nuclear and cytoplasmic PTGES3 staining. Kaplan–Meier graph depicting relation between PSA recurrence-free survival and nuclear PTGES3 staining in PCa patients. Weak, score 0 or 1; Strong, score 2 or 3. b, LNCaP cells were fractionated to separate the indicated cellular compartments and then the localization of AR, PTGES3 and control proteins was determined by western blot. c, LNCaP cells were fractionated to isolate nuclei and then a co-IP was performed using IgG or PTGES3 antibodies. AR protein levels were detected by western blotting. d, A proximity ligation assay was performed and imaged to measure the AR–PTGES3 interaction in intact LNCaP cells. e, LNCaP cells were transfection with HA-tagged full-length AR (FL), AR NTD, AR NTD and DBD (NTD + DBD), or AR DBD and LBD (LBD + DBD). Co-IP assays were performed using HA antibody. HA–AR construct protein levels were measured by western blotting of input samples (top). Interactions between HA–AR constructs and PTGES3 were detected by western blotting of Co-IP samples. f, Protein domain structures of AR and PTGES3. Disordered domain and regions represented with dashed borders and lines. g, ARΔNTD and Smt3-tagged-PTGES3 interact directly as measured by mass photometry. h, AlphaFold modeling of the ARΔNTD and PTGES3 interaction. i, A fluorescence polarization assay measuring binding of ARΔNTD to fluorescein-labeled double-stranded DNA bearing ARE sequences in the presence and absence of PTGES3. Data presented as mean ± s.d. from n = 4 independent experiments, and compared and analyzed to data in Extended Fig. 7f using a two-sided, two-way ANOVA with Tukey’s multiple comparisons test, with ARΔNTD is the control group (*P < 0.05). +Smt3-PTGES3 versus ARΔNTD, P = 0.0254; PTGES3 versus ARΔNTD, P = 0.0337. j, Mass photometry shows that PTGES3 chaperones ARΔNTD’s binding to DNA (ARE). k, Structural modeling of the AR–PTGES3 interaction on DNA-bound AR50. Source data
Fig. 4 AR and PTGES3 interact in the absence and presence of DNA. a, IHC for PTGES3 was performed on human PCa tissue microarrays (n = 120). Tissue cores were scored for intensity of nuclear and cytoplasmic PTGES3 staining. Kaplan–Meier graph depicting relation between PSA recurrence-free survival and nuclear PTGES3 staining in PCa patients. Weak, score 0 or 1; Strong, score 2 or 3. b, LNCaP cells were fractionated to separate the indicated cellular compartments and then the localization of AR, PTGES3 and control proteins was determined by western blot. c, LNCaP cells were fractionated to isolate nuclei and then a co-IP was performed using IgG or PTGES3 antibodies. AR protein levels were detected by western blotting. d, A proximity ligation assay was performed and imaged to measure the AR–PTGES3 interaction in intact LNCaP cells. e, LNCaP cells were transfection with HA-tagged full-length AR (FL), AR NTD, AR NTD and DBD (NTD + DBD), or AR DBD and LBD (LBD + DBD). Co-IP assays were performed using HA antibody. HA–AR construct protein levels were measured by western blotting of input samples (top). Interactions between HA–AR constructs and PTGES3 were detected by western blotting of Co-IP samples. f, Protein domain structures of AR and PTGES3. Disordered domain and regions represented with dashed borders and lines. g, ARΔNTD and Smt3-tagged-PTGES3 interact directly as measured by mass photometry. h, AlphaFold modeling of the ARΔNTD and PTGES3 interaction. i, A fluorescence polarization assay measuring binding of ARΔNTD to fluorescein-labeled double-stranded DNA bearing ARE sequences in the presence and absence of PTGES3. Data presented as mean ± s.d. from n = 4 independent experiments, and compared and analyzed to data in Extended Fig. 7f using a two-sided, two-way ANOVA with Tukey’s multiple comparisons test, with ARΔNTD is the control group (*P < 0.05). +Smt3-PTGES3 versus ARΔNTD, P = 0.0254; PTGES3 versus ARΔNTD, P = 0.0337. j, Mass photometry shows that PTGES3 chaperones ARΔNTD’s binding to DNA (ARE). k, Structural modeling of the AR–PTGES3 interaction on DNA-bound AR50. Source dataCC BY · PubMed Central
Fig. 5 Nuclear PTGES3 facilitates AR-mediated transcription. a, LNCaP cells containing ARE–luciferase reporter were transfected with pCDH-control vector (gray) or pCDH-PTGES3 (blue). Cells were treated with DMSO, 1 nM DHT or 1 nM DHT plus 5 μM apalutamide (APA). Luciferase activities over Renilla were normalized with control (n = 3 as biological replicates; mean ± s.d.). Statistical significance was determined using two-sided t-test (DMSO group *P = 0.0435; DHT group ***P = 0.0002; DHT + APA group **P = 0.0040). b, LNCaP cells were fixed sequentially by EGS and formaldehyde. Dual crosslinking ChIP experiments were performed using indicated antibodies. Precipitated DNA was used as a template to amplify the indicated genomic regions by real-time PCR (n = 3 as biological replicates; mean ± s.d.). Statistical significance was determined using two-sided t-test (**P < 0.01; ***P < 0.001; NS, not significant). c, Venn diagram showing overlaps of AR ChIP–seq peaks in LNCaP(sgCTRL) and LNCaP(sgPTGES3). d, Peak distribution of ChIP-AR peaks in LNCaP(sgCTRL) and LNCaP(sgPTGES3) across different genomic locations. e, ChIP-AR read-density heatmaps of LNCaP(sgCTRL) and LNCaP(sgPTGES3) centered at combined ChIP-AR peaks (n = 38,122). f, The five most enriched known HOMER motifs in the top 2,000 significant ChIP-AR peaks in LNCaP(sgCTRL) and LNCaP(sgPTGES3). g, An IGV screenshot of ChIP-AR enrichment in LNCaP(sgCTRL) and LNCaP(sgPTGES3) at enhancer and promoter sites of KLK3 and KLK2 genes. h, LNCaPi cells were infected with sgCTRL, sgPTGES3 or sgAR. Cells were collected for ATAC-seq (n = 2 as biological replicates). The heatmap shows the differentially accessible ATAC-seq peak regions in sgPTGES3 or sgAR as compared to that in sgCTRL. Each ATAC-seq peak region was scaled to the same size from the start to the end region. Flanking ±500 bp regions are also shown. The top panel shows the corresponding ATAC-seq signal intensity. i, The volcano plot shows the differentially accessible ATAC-seq peak regions upon PTGES3 knockdown as compared to controls. Each dot represents an individual ATAC-seq peak. Differentially accessible ATAC-seq peaks are highlighted in red color (P < 0.05 and log2 Foldchange >1 or <−1). The nearest gene to several representative differentially accessible ATAC-seq peaks have been highlighted. j, A graph depicting the overlap in genomic loci with differentially accessible ATAC-seq peaks upon PTGES3 knockdown that are or are not also bound by AR as measured by ChIP–seq in LNCaP cells52 (Stello2018). The figure shows the percentage of these differentially accessible ATAC-seq peaks that overlapped with the AR ChIP–seq dataset. k, Plots showing ATAC-seq peak coverage in the KLK3 gene body and flanking ±5 kb region. sgCTRL ATAC-seq peaks are colored in blue and sgPTGES3 ATAC-seq peaks in red. The upstream promoter region of KLK3 contains overlapping ATAC-seq peaks and AR Chip–seq peaks52. AR binding motif was identified in this region.
Fig. 5 Nuclear PTGES3 facilitates AR-mediated transcription. a, LNCaP cells containing ARE–luciferase reporter were transfected with pCDH-control vector (gray) or pCDH-PTGES3 (blue). Cells were treated with DMSO, 1 nM DHT or 1 nM DHT plus 5 μM apalutamide (APA). Luciferase activities over Renilla were normalized with control (n = 3 as biological replicates; mean ± s.d.). Statistical significance was determined using two-sided t-test (DMSO group *P = 0.0435; DHT group ***P = 0.0002; DHT + APA group **P = 0.0040). b, LNCaP cells were fixed sequentially by EGS and formaldehyde. Dual crosslinking ChIP experiments were performed using indicated antibodies. Precipitated DNA was used as a template to amplify the indicated genomic regions by real-time PCR (n = 3 as biological replicates; mean ± s.d.). Statistical significance was determined using two-sided t-test (**P < 0.01; ***P < 0.001; NS, not significant). c, Venn diagram showing overlaps of AR ChIP–seq peaks in LNCaP(sgCTRL) and LNCaP(sgPTGES3). d, Peak distribution of ChIP-AR peaks in LNCaP(sgCTRL) and LNCaP(sgPTGES3) across different genomic locations. e, ChIP-AR read-density heatmaps of LNCaP(sgCTRL) and LNCaP(sgPTGES3) centered at combined ChIP-AR peaks (n = 38,122). f, The five most enriched known HOMER motifs in the top 2,000 significant ChIP-AR peaks in LNCaP(sgCTRL) and LNCaP(sgPTGES3). g, An IGV screenshot of ChIP-AR enrichment in LNCaP(sgCTRL) and LNCaP(sgPTGES3) at enhancer and promoter sites of KLK3 and KLK2 genes. h, LNCaPi cells were infected with sgCTRL, sgPTGES3 or sgAR. Cells were collected for ATAC-seq (n = 2 as biological replicates). The heatmap shows the differentially accessible ATAC-seq peak regions in sgPTGES3 or sgAR as compared to that in sgCTRL. Each ATAC-seq peak region was scaled to the same size from the start to the end region. Flanking ±500 bp regions are also shown. The top panel shows the corresponding ATAC-seq signal intensity. i, The volcano plot shows the differentially accessible ATAC-seq peak regions upon PTGES3 knockdown as compared to controls. Each dot represents an individual ATAC-seq peak. Differentially accessible ATAC-seq peaks are highlighted in red color (P < 0.05 and log2 Foldchange >1 or <−1). The nearest gene to several representative differentially accessible ATAC-seq peaks have been highlighted. j, A graph depicting the overlap in genomic loci with differentially accessible ATAC-seq peaks upon PTGES3 knockdown that are or are not also bound by AR as measured by ChIP–seq in LNCaP cells52 (Stello2018). The figure shows the percentage of these differentially accessible ATAC-seq peaks that overlapped with the AR ChIP–seq dataset. k, Plots showing ATAC-seq peak coverage in the KLK3 gene body and flanking ±5 kb region. sgCTRL ATAC-seq peaks are colored in blue and sgPTGES3 ATAC-seq peaks in red. The upstream promoter region of KLK3 contains overlapping ATAC-seq peaks and AR Chip–seq peaks52. AR binding motif was identified in this region.CC BY · PubMed Central
Fig. 6 PTGES3 is a potential therapeutic target for AR-directed therapy resistant PCa. a,b, In a previously established cohort matched for eight clinicopathologic prognostic variables59, high PTGES3 expression was associated with worse MFS in prostatectomy patients who received adjuvant ADT (HR:1.71, log-rank P = 0.0039) (a), but not in prostatectomy patients who did not receive adjuvant ADT (HR:1.24, log-rank P = 0.22) (b). c–h, C42B (c), 22RV1 (d), Enzalutamide (Enz) resistant cell line MR49F cultured with 5 μM Enz (e), VCaP (f), PC3 (g) or DU145 (h) cells were treated with siRNA targeting control (gray) or PTGES3 (blue). Cell viability was measured by WST1 and normalized over control (n = 3 as biological replicates; mean ± s.d.). Statistical significance of siPTGES3-1 versus siControl (compare 1) and siPTGES3-2 versus siControl (compare 2) were determined using two-way ANOVA. In C42B cells (c), 1 **P = 0.0016 and 2 **P = 0.0019; in 22rv1 cells (d), 1 **P = 0.0016 and 2 **P = 0.00196; in MR49F cells (e), 1 **P = 0.0016 and 2 **P = 0.001; in VCaP cells (f), 1 P = 0.0015 and 2 **P = 0.0013; in PC3 cells (g), 1 P = 0.4228 (NS) and 2 P = 0.2013 (NS) and in DU145 cells (h), 1 P = 0.1017 (NS) and 2 P = 0.05265 (NS). i, Mice bearing LNCaP(TET-ON dCas9-KRAB) stably expressing CTRL or PTGES3 sgRNAs were treated with doxycycline when tumors reached ~200 mm3. Average tumor volume was plotted (n = 8; ± s.e.m.) mean and two-way ANOVA was used to measure statistical significance (***P = 0.0002). Source data
Fig. 6 PTGES3 is a potential therapeutic target for AR-directed therapy resistant PCa. a,b, In a previously established cohort matched for eight clinicopathologic prognostic variables59, high PTGES3 expression was associated with worse MFS in prostatectomy patients who received adjuvant ADT (HR:1.71, log-rank P = 0.0039) (a), but not in prostatectomy patients who did not receive adjuvant ADT (HR:1.24, log-rank P = 0.22) (b). c–h, C42B (c), 22RV1 (d), Enzalutamide (Enz) resistant cell line MR49F cultured with 5 μM Enz (e), VCaP (f), PC3 (g) or DU145 (h) cells were treated with siRNA targeting control (gray) or PTGES3 (blue). Cell viability was measured by WST1 and normalized over control (n = 3 as biological replicates; mean ± s.d.). Statistical significance of siPTGES3-1 versus siControl (compare 1) and siPTGES3-2 versus siControl (compare 2) were determined using two-way ANOVA. In C42B cells (c), 1 **P = 0.0016 and 2 **P = 0.0019; in 22rv1 cells (d), 1 **P = 0.0016 and 2 **P = 0.00196; in MR49F cells (e), 1 **P = 0.0016 and 2 **P = 0.001; in VCaP cells (f), 1 P = 0.0015 and 2 **P = 0.0013; in PC3 cells (g), 1 P = 0.4228 (NS) and 2 P = 0.2013 (NS) and in DU145 cells (h), 1 P = 0.1017 (NS) and 2 P = 0.05265 (NS). i, Mice bearing LNCaP(TET-ON dCas9-KRAB) stably expressing CTRL or PTGES3 sgRNAs were treated with doxycycline when tumors reached ~200 mm3. Average tumor volume was plotted (n = 8; ± s.e.m.) mean and two-way ANOVA was used to measure statistical significance (***P = 0.0002). Source dataCC BY · PubMed Central
Extended Data Fig. 1 Phenotypical characterization of C42BmNG2-AR cell lines. a, Exome sequencing results of C42BmNG2-AR showed a homogenous knock-in of mNG2. Left: mNG2_11 Knock-in design indicating sequencing location and the reads count from the exosome sequencing. Right: IGV visualization of the sequence alignment around the ATG of AR. b, C42BmNG2-AR cells were treated with siRNA targeting control (grey) or AR (blue). RNA was collected, AR mRNA levels over 18 s rRNA were measured by real-time PCR then normalized to siCTRL (n = 3 as biological replicates; Mean ± SD). Statistical significance was determined using two-sided t-test (**P < 0.01). c, C42BmNG2-AR cells were treated with siRNA targeting control (green) or AR (black). Cells were collected for flow cytometry, mNG2-AR levels were indicated by the fluorescence intensity. Median Fluorescence Intensity (MFI) was calculated (n = 3; Mean ± SD). Statistical significance was determined using two-sided t-test (***P < 0.001). d, C42B (grey) and C42BmNG2-AR cells (blue) were treated with DMSO (solid line) or 50 nM ARD-61 (dashed line). Cell viability was measured by WST1 and normalized over control (n = 3 as biological replicates; Mean ± SD). Statistical significance between C42B and C42BmNG2-AR cells under DMSO or ARD-61 treatment was determined using two-way ANOVA (**P < 0.01; ns=not significant). e, Total RNA from C42B and C42BmNG2-AR cells were collected for RNA-seq. Gene expression values were calculated as ln(TPM + 0.1). Pearson correlation was calculated comparing all genes (n = 19127) between the two cell lines. Source data
Extended Data Fig. 1 Phenotypical characterization of C42BmNG2-AR cell lines. a, Exome sequencing results of C42BmNG2-AR showed a homogenous knock-in of mNG2. Left: mNG2_11 Knock-in design indicating sequencing location and the reads count from the exosome sequencing. Right: IGV visualization of the sequence alignment around the ATG of AR. b, C42BmNG2-AR cells were treated with siRNA targeting control (grey) or AR (blue). RNA was collected, AR mRNA levels over 18 s rRNA were measured by real-time PCR then normalized to siCTRL (n = 3 as biological replicates; Mean ± SD). Statistical significance was determined using two-sided t-test (**P < 0.01). c, C42BmNG2-AR cells were treated with siRNA targeting control (green) or AR (black). Cells were collected for flow cytometry, mNG2-AR levels were indicated by the fluorescence intensity. Median Fluorescence Intensity (MFI) was calculated (n = 3; Mean ± SD). Statistical significance was determined using two-sided t-test (***P < 0.001). d, C42B (grey) and C42BmNG2-AR cells (blue) were treated with DMSO (solid line) or 50 nM ARD-61 (dashed line). Cell viability was measured by WST1 and normalized over control (n = 3 as biological replicates; Mean ± SD). Statistical significance between C42B and C42BmNG2-AR cells under DMSO or ARD-61 treatment was determined using two-way ANOVA (**P < 0.01; ns=not significant). e, Total RNA from C42B and C42BmNG2-AR cells were collected for RNA-seq. Gene expression values were calculated as ln(TPM + 0.1). Pearson correlation was calculated comparing all genes (n = 19127) between the two cell lines. Source dataCC BY · PubMed Central
Extended Data Fig. 2 Validation of the CRISPRi activity of C42BimNG2-AR. a-b, C42BimNG2-AR cells were infected with sgRNAs targeting CTRL, P53 (a) or AR (b). RNA was collected; P53 or AR mRNA levels over 18 s rRNA were measured by real-time PCR (n = 3 as biological replicates; Mean ± SD). Statistical significance was determined using two-sided t-test (***P < 0.001, **P < 0.01). c, C42BimNG2-AR cells were infected with sgRNAs targeting control (green) or AR (blue). Cells were collected for flow cytometry; mNG2-AR levels were indicated by the fluorescence intensity. d, C42B cells were infected with increasing amounts of CRISPRi v2 library virus, and BFP-positive percentages were measured by flow cytometry 3 days post-infection to determine the optimal viral dose targeting 25% transduction. Two biological replicates were shown in the plot. e, Counts of essential and non-essential sgRNAs at post-infection day 7 (T0) are shown as box-and-whisker plots. The central line denotes the median, the box spans the interquartile range, and open circles represent outliers as defined by the R base boxplot function (values > 1.5× outside the interquartile range). Counts of non-essential sgRNAs (n = 94,070) and essential genes (n = 10,465, defined by DepMap: https://depmap.org/portal/) were compared using a two-sided Mann–Whitney test (P = 0.8489), showing no significant difference between essential and non-essential guides. f, A competitive growth-based assay of the C42BimNG2-AR cells infected with sgRNA (mCherry + ) targeting CTRL, AR, or commonly essential genes (HSPA9 and AARS) as a positive control. The relative fraction of sgRNA in the cell population was measured by flow cytometer starting from day 3 post-infection. Statistical significance between each targeting sgRNA and sgCTRL were determined using two-way ANOVA (**P < 0.01; ***P < 0.001). Source data
Extended Data Fig. 2 Validation of the CRISPRi activity of C42BimNG2-AR. a-b, C42BimNG2-AR cells were infected with sgRNAs targeting CTRL, P53 (a) or AR (b). RNA was collected; P53 or AR mRNA levels over 18 s rRNA were measured by real-time PCR (n = 3 as biological replicates; Mean ± SD). Statistical significance was determined using two-sided t-test (***P < 0.001, **P < 0.01). c, C42BimNG2-AR cells were infected with sgRNAs targeting control (green) or AR (blue). Cells were collected for flow cytometry; mNG2-AR levels were indicated by the fluorescence intensity. d, C42B cells were infected with increasing amounts of CRISPRi v2 library virus, and BFP-positive percentages were measured by flow cytometry 3 days post-infection to determine the optimal viral dose targeting 25% transduction. Two biological replicates were shown in the plot. e, Counts of essential and non-essential sgRNAs at post-infection day 7 (T0) are shown as box-and-whisker plots. The central line denotes the median, the box spans the interquartile range, and open circles represent outliers as defined by the R base boxplot function (values > 1.5× outside the interquartile range). Counts of non-essential sgRNAs (n = 94,070) and essential genes (n = 10,465, defined by DepMap: https://depmap.org/portal/) were compared using a two-sided Mann–Whitney test (P = 0.8489), showing no significant difference between essential and non-essential guides. f, A competitive growth-based assay of the C42BimNG2-AR cells infected with sgRNA (mCherry + ) targeting CTRL, AR, or commonly essential genes (HSPA9 and AARS) as a positive control. The relative fraction of sgRNA in the cell population was measured by flow cytometer starting from day 3 post-infection. Statistical significance between each targeting sgRNA and sgCTRL were determined using two-way ANOVA (**P < 0.01; ***P < 0.001). Source dataCC BY · PubMed Central
Extended Data Fig. 3 Extended validation of hits from the genome-scale CRISPRi screening. a, Clone 2 C42BimNG2-AR were infected with sgRNAs targeting control or indicated hit genes. After puromycin selection, Median Fluorescence Intensity (MFI) of mNG2-AR fluorescence from the treated cells was measured by flow cytometry. The percentage of AR expression was calculated as AR expression%=(MFIsg Individual gene-MFIC42B)/(MFIsgCTRL-MFIC42B)*100 + 100 (n = 3 as biological replicates; Mean ± SD). Statistical significance was determined using two-sided t-test (***P < 0.001; **P < 0.01; ns=not significant). b, C42Bi were infected with sgRNAs targeting control or indicated hit genes. mRNA levels of the indicated genes over 18 s rRNA were measured by real-time PCR (n = 3 as biological replicates; Mean ± SD). Statistical significance was determined using two-sided t-test (***P < 0.001; NA=undetectable). c, C42Bi cells were infected with indicated individual sgRNAs. Cell lysates were collected after puromycin selection. AR and GAPDH levels were detected by western blotting. Each western blot experiment was performed twice to determine reproducibility. d, LNCaPi cells were infected with indicated individual sgRNAs. Cell lysates were collected after puromycin selection. AR and GAPDH levels were detected by western blotting. Each western blot experiment was performed twice to determine reproducibility. Source data
Extended Data Fig. 3 Extended validation of hits from the genome-scale CRISPRi screening. a, Clone 2 C42BimNG2-AR were infected with sgRNAs targeting control or indicated hit genes. After puromycin selection, Median Fluorescence Intensity (MFI) of mNG2-AR fluorescence from the treated cells was measured by flow cytometry. The percentage of AR expression was calculated as AR expression%=(MFIsg Individual gene-MFIC42B)/(MFIsgCTRL-MFIC42B)*100 + 100 (n = 3 as biological replicates; Mean ± SD). Statistical significance was determined using two-sided t-test (***P < 0.001; **P < 0.01; ns=not significant). b, C42Bi were infected with sgRNAs targeting control or indicated hit genes. mRNA levels of the indicated genes over 18 s rRNA were measured by real-time PCR (n = 3 as biological replicates; Mean ± SD). Statistical significance was determined using two-sided t-test (***P < 0.001; NA=undetectable). c, C42Bi cells were infected with indicated individual sgRNAs. Cell lysates were collected after puromycin selection. AR and GAPDH levels were detected by western blotting. Each western blot experiment was performed twice to determine reproducibility. d, LNCaPi cells were infected with indicated individual sgRNAs. Cell lysates were collected after puromycin selection. AR and GAPDH levels were detected by western blotting. Each western blot experiment was performed twice to determine reproducibility. Source dataCC BY · PubMed Central
Extended Data Fig. 4 PTGES3 is required for AR protein levels and AR activity but does not affect AR mRNA levels, mRNA half-life or protein half-life. a, Analysis of proteomic effects of PTGES3 knockdown versus control for select AR target genes. Mass spectrometry was performed in Fig. 3a and focused results are shown here as a heatmap. b, LNCaPi were infected with sgRNA targeting control or PTGES3. Total RNA was collected for RNA-seq. Gene Set Enrichment Analysis (GSEA) was performed. AR targeted genes are the top rank downregulated gene set. c, sgPTGES3 vs sgCTRL RNA-seq was performed as described in b. The RNA levels of individual Hallmark Androgen response genes were shown in the heatmap. d, LNCaPi were infected with sgRNAs targeting CTRL or PTGES3. AR mRNA levels over 18 s rRNA were measured by real-time PCR (n = 3 as biological replicates; Mean ± SD; ns indicates no significant difference by two-sided t-test analysis). e. LNCaPi cells infected with sgCTRL or sgPTGES3 were treated with 1uM Actinmycin D for 0, 8 or 16 h. RNA was collected to measure the AR mRNA levels using real-time PCR (n = 3 as biological replicates; Mean ± SEM; ns indicates no significant difference by two-way ANOVA analysis). f. LNCaPi cells infected with sgCTRL or sgPTGES3 were treated with protein synthesis inhibitor Cycloheximide (CHX, 5 μM) for 0, 12, 24 h. AR protein levels were detected by western blotting. g, 22RV1 cells were treated with siRNA targeting control or PTGES3. AR, ARV7, and GAPDH levels were detected by western blotting. Source data
Extended Data Fig. 4 PTGES3 is required for AR protein levels and AR activity but does not affect AR mRNA levels, mRNA half-life or protein half-life. a, Analysis of proteomic effects of PTGES3 knockdown versus control for select AR target genes. Mass spectrometry was performed in Fig. 3a and focused results are shown here as a heatmap. b, LNCaPi were infected with sgRNA targeting control or PTGES3. Total RNA was collected for RNA-seq. Gene Set Enrichment Analysis (GSEA) was performed. AR targeted genes are the top rank downregulated gene set. c, sgPTGES3 vs sgCTRL RNA-seq was performed as described in b. The RNA levels of individual Hallmark Androgen response genes were shown in the heatmap. d, LNCaPi were infected with sgRNAs targeting CTRL or PTGES3. AR mRNA levels over 18 s rRNA were measured by real-time PCR (n = 3 as biological replicates; Mean ± SD; ns indicates no significant difference by two-sided t-test analysis). e. LNCaPi cells infected with sgCTRL or sgPTGES3 were treated with 1uM Actinmycin D for 0, 8 or 16 h. RNA was collected to measure the AR mRNA levels using real-time PCR (n = 3 as biological replicates; Mean ± SEM; ns indicates no significant difference by two-way ANOVA analysis). f. LNCaPi cells infected with sgCTRL or sgPTGES3 were treated with protein synthesis inhibitor Cycloheximide (CHX, 5 μM) for 0, 12, 24 h. AR protein levels were detected by western blotting. g, 22RV1 cells were treated with siRNA targeting control or PTGES3. AR, ARV7, and GAPDH levels were detected by western blotting. Source dataCC BY · PubMed Central
Extended Data Fig. 5 Dual functions of PTGES3 potentially regulating AR protein levels. a, LNCaPi were infected with sgRNA targeting CTRL, FKBP4, HSP90 or PTGES3 and then selected with puromycin. Cell lysates were collected. AR and GAPDH levels were detected by western blotting. b-c, LNCaPi were infected with sgRNA targeting CTRL, PTGES, PTGES2 or PTGES3 and then selected with puromycin. RNA or cell lysates were collected. Indicated genes mRNA levels over 18 s rRNA were measured by real-time PCR (b; n = 3 as biological replicates; Mean ± SD). Statistical significance was determined using two-sided t-test (***P < 0.001). AR, PTGES3 and GAPDH levels were detected by western blotting (c). Each western blot experiment was performed twice to determine reproducibility. d, LNCaPi cells were infected with sgCTRL-Blast or sgPTGES3-Blast and selected, then treated with DMSO or 100 ng/ml doxycycline to overexpress the PTGES3 WT or Y9N or W106A mutant proteins. AR, PTEGS3 and GAPDH levels were detected by western blotting. e, LNCaPi cells infected with sgCTRL or sgPTGES3 were treated with proteasome inhibitor 5 μM MG132 for 0, 12, 24 h. AR, PTGES3 and GAPDH protein levels were detected by western blotting. f, LNCaPi cells infected with sgCTRL or sgPTGES3 were treated with lysosomal inhibitor 1 μM Bafilomycin A1 for 0, 12, 24 h. AR, PTGES3 and GAPDH protein levels were detected by western blotting. g, LNCaPi cells infected with sgCTRL or sgPTGES3 were treated DMSO, 10 nM R1881 (androgen), or 10 μM Enzalutamide (Enz, anti-androgen). AR, PTGES3 and GAPDH protein levels were detected by western blotting. Source data
Extended Data Fig. 5 Dual functions of PTGES3 potentially regulating AR protein levels. a, LNCaPi were infected with sgRNA targeting CTRL, FKBP4, HSP90 or PTGES3 and then selected with puromycin. Cell lysates were collected. AR and GAPDH levels were detected by western blotting. b-c, LNCaPi were infected with sgRNA targeting CTRL, PTGES, PTGES2 or PTGES3 and then selected with puromycin. RNA or cell lysates were collected. Indicated genes mRNA levels over 18 s rRNA were measured by real-time PCR (b; n = 3 as biological replicates; Mean ± SD). Statistical significance was determined using two-sided t-test (***P < 0.001). AR, PTGES3 and GAPDH levels were detected by western blotting (c). Each western blot experiment was performed twice to determine reproducibility. d, LNCaPi cells were infected with sgCTRL-Blast or sgPTGES3-Blast and selected, then treated with DMSO or 100 ng/ml doxycycline to overexpress the PTGES3 WT or Y9N or W106A mutant proteins. AR, PTEGS3 and GAPDH levels were detected by western blotting. e, LNCaPi cells infected with sgCTRL or sgPTGES3 were treated with proteasome inhibitor 5 μM MG132 for 0, 12, 24 h. AR, PTGES3 and GAPDH protein levels were detected by western blotting. f, LNCaPi cells infected with sgCTRL or sgPTGES3 were treated with lysosomal inhibitor 1 μM Bafilomycin A1 for 0, 12, 24 h. AR, PTGES3 and GAPDH protein levels were detected by western blotting. g, LNCaPi cells infected with sgCTRL or sgPTGES3 were treated DMSO, 10 nM R1881 (androgen), or 10 μM Enzalutamide (Enz, anti-androgen). AR, PTGES3 and GAPDH protein levels were detected by western blotting. Source dataCC BY · PubMed Central
Extended Data Fig. 6 PTGES3 has a nuclear localization. a, Representative immunostainings against PTGES3 in PCa patient samples. b, Cell fractions from LNCaP cells were immunoblotted with indicated antibodies. Source data
Extended Data Fig. 6 PTGES3 has a nuclear localization. a, Representative immunostainings against PTGES3 in PCa patient samples. b, Cell fractions from LNCaP cells were immunoblotted with indicated antibodies. Source dataCC BY · PubMed Central
Extended Data Fig. 7 PTGES3 interaction with AR modulates AR’s ability to bind DNA. a-b, Protein gels of final recombinant AR and PTGES3 protein preparations used in DNA binding assays. Sypro Ruby-stained 4-12% SDS-PAGE. c, Structural modeling of the AR ligand binding domain67 (PDB: 1XOW) onto the PTGES3-HSP90 client maturation complex43 (PDB: 7KRJ). d-e, Fluorescence polarization binding curves corresponding to data presenting in Fig. 4i showing ARΔNTD binding to ARE (d) or canonical ARE (e) double-stranded DNA. f, Fluorescence polarization measuring binding of ARΔNTD to fluorescein-labeled double-stranded DNA bearing canonical ARE sequences. Data presented as mean ± SD from n = 4 independent experiments, and compared and analyzed to data in Fig. 4i using a two-sided, two-way ANOVA with Tukey’s multiple comparisons test, n.s. (not significant), P > 0.05, where ARΔNTD is the control group. g-h, Electrophoretic mobility shift assays (g) and quantitation (h) measuring binding of ARDBD and PTGES3 to 36 base pair ARE DNA assayed in e,f. Experiments and gels were processed in parallel. Data presented as mean ± SD from n = 3 independent experiments. i, AlphaFold modeling of the dimeric ARFL and PTGES3 interaction on ARE double-stranded DNA. Source data
Extended Data Fig. 7 PTGES3 interaction with AR modulates AR’s ability to bind DNA. a-b, Protein gels of final recombinant AR and PTGES3 protein preparations used in DNA binding assays. Sypro Ruby-stained 4-12% SDS-PAGE. c, Structural modeling of the AR ligand binding domain67 (PDB: 1XOW) onto the PTGES3-HSP90 client maturation complex43 (PDB: 7KRJ). d-e, Fluorescence polarization binding curves corresponding to data presenting in Fig. 4i showing ARΔNTD binding to ARE (d) or canonical ARE (e) double-stranded DNA. f, Fluorescence polarization measuring binding of ARΔNTD to fluorescein-labeled double-stranded DNA bearing canonical ARE sequences. Data presented as mean ± SD from n = 4 independent experiments, and compared and analyzed to data in Fig. 4i using a two-sided, two-way ANOVA with Tukey’s multiple comparisons test, n.s. (not significant), P > 0.05, where ARΔNTD is the control group. g-h, Electrophoretic mobility shift assays (g) and quantitation (h) measuring binding of ARDBD and PTGES3 to 36 base pair ARE DNA assayed in e,f. Experiments and gels were processed in parallel. Data presented as mean ± SD from n = 3 independent experiments. i, AlphaFold modeling of the dimeric ARFL and PTGES3 interaction on ARE double-stranded DNA. Source dataCC BY · PubMed Central
Extended Data Fig. 8 PTGES3 potentially interacts with histone acetyltransferase KAT2A. a, LNCaPi cells were infected with sgCTRL or sgPTGES3 and selected with puromycin. KAT2A, P300, FOXA1, PTGES3, and GAPDH protein levels were detected by western blotting. b, IP experiments were performed with LNCaP cell lysate using an IgG or PTGES3 antibody. KAT2A, P300, FOXA1, and PTGES3 protein levels were detected by western blotting. c, LNCaP cells containing ARE-luciferase reporter were transfected with pCDH-PTGES3. Cells were treated with siRNA targeting control or KAT2A in the condition of 1 nM DHT. Luciferase activities over Renilla were normalized with control (n = 3 as biological replicates; Mean ± SD). Two sided t-test was used to determine statistical significance (***P < 0.001). d. LNCaP cells were fixed sequentially by EGS and formaldehyde. Dual cross-linking ChIP experiments were performed using indicated antibodies. Precipitated DNA was used as a template to amplify the indicated genomic regions by real-time PCR (n = 3 as biological replicates; Mean ± SD). LKB1 serves as a positive Dual X ChIP control which indirectly binds to the P21 promoter (non-ARE) by interacting with P53. The P21 promoter levels in LKB1 group were compared to the IgG group, two sided t-test was used to determine statistical significance (***P < 0.001). Source data
Extended Data Fig. 8 PTGES3 potentially interacts with histone acetyltransferase KAT2A. a, LNCaPi cells were infected with sgCTRL or sgPTGES3 and selected with puromycin. KAT2A, P300, FOXA1, PTGES3, and GAPDH protein levels were detected by western blotting. b, IP experiments were performed with LNCaP cell lysate using an IgG or PTGES3 antibody. KAT2A, P300, FOXA1, and PTGES3 protein levels were detected by western blotting. c, LNCaP cells containing ARE-luciferase reporter were transfected with pCDH-PTGES3. Cells were treated with siRNA targeting control or KAT2A in the condition of 1 nM DHT. Luciferase activities over Renilla were normalized with control (n = 3 as biological replicates; Mean ± SD). Two sided t-test was used to determine statistical significance (***P < 0.001). d. LNCaP cells were fixed sequentially by EGS and formaldehyde. Dual cross-linking ChIP experiments were performed using indicated antibodies. Precipitated DNA was used as a template to amplify the indicated genomic regions by real-time PCR (n = 3 as biological replicates; Mean ± SD). LKB1 serves as a positive Dual X ChIP control which indirectly binds to the P21 promoter (non-ARE) by interacting with P53. The P21 promoter levels in LKB1 group were compared to the IgG group, two sided t-test was used to determine statistical significance (***P < 0.001). Source dataCC BY · PubMed Central
Extended Data Fig. 9 PTGES3 promotes the chromatin opening status of AR binding to AREs. a, LNCaPi cells were infected with sgCTRL or sgAR and selected with puromycin. Cells were collected for ATAC-Sequencing. The volcano plot shows the differentially accessible ATAC-seq peak regions in sgAR as compared to that in sgCTRL. Each dot represents an individual ATAC-seq peak. Differentially accessible ATAC-seq peaks are highlighted in red color (pvalue < 0.05 and log2 Foldchange > 1 or <-1). Nearest gene to several representative differentially accessible ATAC-seq peaks have been highlighted. b, The volcano plot shows the differentially accessible ATAC-seq peak regions in sgPTGES3 as compared to that in sgAR. c, The differentially accessible ATAC-seq peaks (in sgPTGES3 as compared to that in sgCTRL) were mapped to the promoter or enhancer regions. The nearest genes to the regions were selected. AR target genes from the two groups were shown in the heat map. d, Total RNA from LNCaP(sgCTRL) and LNCaP(sgPTGES3) cells were collected for RNA-seq. Gene expression values were calculated as ln(TPM + 0.1). Pearson correlation was calculated comparing expression levels of HSF1 target genes between the two cell lines. Source data
Extended Data Fig. 9 PTGES3 promotes the chromatin opening status of AR binding to AREs. a, LNCaPi cells were infected with sgCTRL or sgAR and selected with puromycin. Cells were collected for ATAC-Sequencing. The volcano plot shows the differentially accessible ATAC-seq peak regions in sgAR as compared to that in sgCTRL. Each dot represents an individual ATAC-seq peak. Differentially accessible ATAC-seq peaks are highlighted in red color (pvalue < 0.05 and log2 Foldchange > 1 or <-1). Nearest gene to several representative differentially accessible ATAC-seq peaks have been highlighted. b, The volcano plot shows the differentially accessible ATAC-seq peak regions in sgPTGES3 as compared to that in sgAR. c, The differentially accessible ATAC-seq peaks (in sgPTGES3 as compared to that in sgCTRL) were mapped to the promoter or enhancer regions. The nearest genes to the regions were selected. AR target genes from the two groups were shown in the heat map. d, Total RNA from LNCaP(sgCTRL) and LNCaP(sgPTGES3) cells were collected for RNA-seq. Gene expression values were calculated as ln(TPM + 0.1). Pearson correlation was calculated comparing expression levels of HSF1 target genes between the two cell lines. Source dataCC BY · PubMed Central
Extended Data Fig. 10 High PTGES3 expression predicts a poor outcome in advanced PCa patients treated with first line AR targeted therapy. a, Kaplan-Meier analysis of mCRPC patients treated with a first-line AR targeted therapy (abiraterone acetate, enzalutamide, or apalutamide) shows a worse overall survival in patients with high PTGES3 expression compared to low expression. b, PC3 or DU145 cells were treated with siRNA targeting control. were collected for western blot using AR, PTGES3, and GAPDH antibodies. c, A histogram of pan-cancer essentiality CHRONOS scores from the Cancer Dependency map showing the distribution of PTGES3 essentiality in 1070 cell lines. PTGES3 is considered essential for cell proliferation or survival in 85 cell lines ( ~ 10% of all cell lines) with scores less than -0.5. This result indicates PTGES3 is not required for the survival or growth of the vast majority of non-prostate cancer or AR independent cell lines. d-g, CRISPRi sgRNA level screen phenotypes performed in MiaPaCa2(c), H358(d), K562(e), and HL60(f) cells were plotted as log2 values for PTGES3, for PSMA1 which is a common essential gene and for 5 randomly selected non-targeting sgRNA controls present in the sgRNA library. h, Tumor tissues from Fig. 4i were collected for western blot using AR and GAPDH antibodies. Source data
Extended Data Fig. 10 High PTGES3 expression predicts a poor outcome in advanced PCa patients treated with first line AR targeted therapy. a, Kaplan-Meier analysis of mCRPC patients treated with a first-line AR targeted therapy (abiraterone acetate, enzalutamide, or apalutamide) shows a worse overall survival in patients with high PTGES3 expression compared to low expression. b, PC3 or DU145 cells were treated with siRNA targeting control. were collected for western blot using AR, PTGES3, and GAPDH antibodies. c, A histogram of pan-cancer essentiality CHRONOS scores from the Cancer Dependency map showing the distribution of PTGES3 essentiality in 1070 cell lines. PTGES3 is considered essential for cell proliferation or survival in 85 cell lines ( ~ 10% of all cell lines) with scores less than -0.5. This result indicates PTGES3 is not required for the survival or growth of the vast majority of non-prostate cancer or AR independent cell lines. d-g, CRISPRi sgRNA level screen phenotypes performed in MiaPaCa2(c), H358(d), K562(e), and HL60(f) cells were plotted as log2 values for PTGES3, for PSMA1 which is a common essential gene and for 5 randomly selected non-targeting sgRNA controls present in the sgRNA library. h, Tumor tissues from Fig. 4i were collected for western blot using AR and GAPDH antibodies. Source dataCC BY · PubMed Central

Verification

Claims the paper makes

23 Valid · 0 Not valid · 6 Not checked · 91 Own results and interpretations

50 more claims · 4 with their cited sources checked

Create a free account to see all 120 claims and every source verdict.

See all claims — free account
  • Probably not a substantial claim. 36
  • It may state two separate claims as one. 33
  • The quote only partly states this claim. 1

Create a free account to see these claims.

Citations

Cites

Show all 95 references

Cited by

See all 10

Metadata

Authors
Li, H., Melnyk, J. E., Fu, B. X. H., Shrestha, R., Zhang, M., Sjöström, M., Feng, S., Anderson, J. A., Han, W., Chesner, L. N., Shin, H. J., Farsh, T., Suarez, H. J., Nath, S., Chou, J., Das, R., Egusa, E. A., Calvert, M., Kishishita, A., Barpanda, A., Zhu, J., Maheshwari, A., Chen, W. S., Alshalalfa, M., Winters, A., Hua, J. T., Liu, T., Davicioni, E., Wiita, A. P., Stohr, B. A., Siddiqui, J., Huang, B., Small, E. J., Shokat, K. M., Nelson, P. S., Quigley, D. A., Wasmuth, E. V., Gilbert, L. A., & Feng, F. Y.
Affiliations
  • University of California, San Francisco (US)
  • Emory University (US)
  • Fred Hutch Cancer Center (US)
  • The University of Texas at San Antonio Health Science Center (US)
  • University of Michigan (US)
  • Howard Hughes Medical Institute (US)
Corresponding author
  • Li, Haolong
  • Gilbert, Luke A.
Venue
Nature Genetics
Published
2025-11-05
Volume
57
Issue
12
Pages
3027-3038
ISSN
  • 1061-4036 (ISSN-L)
  • 1546-1718
OpenAlex
W4415913528
Cited by
10
References
95
Citations by year
2026: 10
FWCI
6.51
Type
journal article
Language
en
Open access
Licence
Locations
  • Nature Genetics (publishedVersion, cc-by)
  • bioRxiv (Cold Spring Harbor Laboratory) (acceptedVersion)
  • PubMed (publishedVersion)
  • PubMed Central (publishedVersion, cc-by)
  • PubMed Central (submittedVersion)
  • Lund University Publications (Lund University) (submittedVersion)
Funders
  • U.S. Department of Defense (HT94252310462, HT94252410252, W81XWH2010136, W81XWH2210833, W81XWH2110046, W81XWH)
  • Howard Hughes Medical Institute
  • U.S. Department of Health and Human Services (1R01CA221969-01, R01GM131641, CA230516-02S1, 1F32CA236347-01, CA204602, 1R01CA230516-01, 1R01CA244550, R01GM124334, P50 CA186786, 1R01CA227025)
  • Prostate Cancer Foundation (SU2C-AACR-DT0812, P50CA097186, 21CHAL06, P30 CA015704, 17CHAL06, 21YOUN12, CA-097186, PC230420)
  • V Foundation for Cancer Research (V2024-016)
  • Cancer Prevention and Research Institute of Texas
  • Cancerfonden
  • Vetenskapsrådet (2018-00382, 2018-)
  • Svenska Läkaresällskapet
  • National Institutes of Health (1f32ca236347-01, 1s10od016229-01, W81XWH, K99-R00, CA015704, CA-097186, K99/R00 CA204602)
  • University of California, San Francisco
  • Stand Up To Cancer (SU2C-AACR-DT0812, DT0812)
  • National Cancer Institute (CA097186, K99CA204602, R01CA221969, DP2CA239597, R01CA227025, CA015704, K99‐R00, F32 CA236347, R01CA244550, K99/R00 CA204602, R01CA230516, CA186786, R01CA234715, R00CA204602, R01CA266452, R01CA290875, P30CA015704, P50CA097186, P50CA186786, P50CA275741)
  • National Institute of General Medical Sciences (R01GM124334, P41GM103481, R00GM140264, R01GM131641)
  • Department of Medicine, University of California, San Francisco
  • DOD Prostate Cancer Research Program
Topics
  • Life Sciences › Biochemistry, Genetics and Molecular Biology › Cancer Research › Prostate Cancer Treatment and Research
  • Life Sciences › Biochemistry, Genetics and Molecular Biology › Molecular Biology › Chromatin Remodeling and Cancer
  • Health Sciences › Medicine › Urology › Prostate Cancer Diagnosis and Treatment
Keywords
PTGES3, androgen receptor, advanced prostate cancer, CRISPRi screen, cell cycle arrest, cell death, therapeutic target, HOXB13, GATA2

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

The claim verdicts this report counts are experimental: their accuracy is not yet measured.

A language model drew the claims from the paper on 2026-10-10, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).

A second model labelled each claim on 2026-10-10 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).

Cited sources are checked with a language model’s help (gpt-6-sol).

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review

How to cite

Cite CoChat for the checks on this page, and the paper for its own findings.

This page
CoChat. (2026). Citation check: Genome-scale CRISPR screens identify PTGES3 as a direct modulator of androgen receptor function in advanced prostate cancer. https://app.cochat.ai/papers/10.1038/s41588-025-02388-8
The paper
Li, H., Melnyk, J. E., Fu, B. X. H., Shrestha, R., Zhang, M., Sjöström, M., Feng, S., Anderson, J. A., Han, W., Chesner, L. N., Shin, H. J., Farsh, T., Suarez, H. J., Nath, S., Chou, J., Das, R., Egusa, E. A., Calvert, M., Kishishita, A., . . . Feng, F. Y. (2025). Genome-scale CRISPR screens identify PTGES3 as a direct modulator of androgen receptor function in advanced prostate cancer. Nature Genetics. https://doi.org/10.1038/s41588-025-02388-8

Data sources

Bibliographic data: Semantic Scholar, OpenAlex, Crossref.

Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).

Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.