A novel CRISPR-engineered prostate cancer cell line defines the AR-V transcriptome and identifies PARP inhibitor sensitivities

Abstract

Resistance to androgen receptor (AR)-targeted therapies in prostate cancer (PC) is a major clinical problem. A key mechanism of treatment resistance in advanced PC is the generation of alternatively spliced forms of the AR termed AR variants (AR-Vs) that are refractory to targeted agents and drive tumour progression. Our understanding of how AR-Vs function is limited due to difficulties in distinguishing their discriminate activities from full-length AR (FL-AR). Here we report the development of a novel CRISPR-derived cell line which is a derivative of CWR22Rv1 cells, called CWR22Rv1-AR-EK, that has lost expression of FL-AR, but retains all endogenous AR-Vs. From this, we show that AR-Vs act unhindered by loss of FL-AR to drive cell growth and expression of androgenic genes. Global transcriptomics demonstrate that AR-Vs drive expression of a cohort of DNA damage response genes and depletion of AR-Vs sensitises cells to ionising radiation. Moreover, we demonstrate that AR-Vs interact with PARP1 and PARP2 and are dependent upon their catalytic function for transcriptional activation. Importantly, PARP blockade compromises expression of AR-V-target genes and reduces growth of CRPC cell lines suggesting a synthetic lethality relationship between AR-Vs and PARP, advocating the use of PARP inhibitors in AR-V positive PC.

Figure 1.

Development and validation of the CWR22Rv1-AR-EK cell line. (A) Diagrammatic representation of the CRISPR strategy utilised to introduce a stop codon into the FL-AR-encoding exon 5 of the AR gene. Sequence of parental and CWR22Rv1-AR-EK AR locus adjacent to PAM site of Cas9/gRNA_2 is shown. (B) Western blotting of either parental CWR22Rv1 cells or CWR22Rv1-AR-EK cells subject to control (siScr), N-terminal AR-targeting (siARex1) or C-terminal-targeting (siARex7) siRNAs for 48 hours, using an N-terminal-binding AR antibody and α-tubulin for loading control. (C) CWR22Rv1 and CWR22Rv1-AR-EK cells were subject to immunoprecipitation (IP) incorporating either N- or C-terminal-binding AR antibodies and resultant immunoprecipitates probed with an N-terminal AR-binding antibody. Input samples were ran alongside IP samples and additionally probed with α-tubulin to demonstrate parity in protein quantities between the IP experimental arms. (D) CWR22Rv1 (left panel) and CWR22Rv1-AR-EK (right panel) cells grown in steroid-depleted media supplemented with and without 10 nM dihydrotestosterone (DHT) were subject to either siScr or siARex7 transfection for 48 hours prior to chromatin immunoprecipitation (ChIP) using C-terminal AR-binding or control (IgG) antibodies and quantitative PCR incorporating primers to the PSA enhancer. VCaP cells treated with 10 nM DHT for 4 hours were used as a positive control for enrichment of FL-AR. Data represents the average of three independent experiments ± SD. Validation of siRNA-mediated FL-AR knockdown was demonstrated by western blotting of CWR22Rv1 chromatin fractions incorporating anti-AR and histone H2B antibodies. (E) Quantitative RT-PCR to compare expression of clinically-relevant AR-Vs in CWR22Rv1 and CWR22Rv1-AR-EK cells grown in serum-containing media was performed. Data represents the average of three independent experiments ± SD.

Figure 2.

AR-Vs maintain androgenic signalling and bind chromatin in the absence of FL-AR. (A) CWR22Rv1-AR-EK cells grown in steroid-depleted media were subject to control (siScr) or AR-V (siAR-V) depletion for 48 hours with either vehicle, 10 nM DHT or 10 μM enzalutamide (Enz) treatment for the final 24 hours before quantitative RT-PCR analysis to assess PSA and TMPRSS2 expression. Data represents the average of three independent experiments ± SD. Validation of AR-V depletion is shown in the accompanying immunoblot (right panel) incorporating N-terminal-binding AR, AR-V7 and β-actin antibodies. (B) Comparison of PSA, TMPRSS2, UBE2C and FKBP5 in CWR22Rv1 cells grown in steroid-depleted media supplemented with or without 10 nM DHT, and CWR22Rv1-AR-EK and R1-D567 cells grown in steroid-depleted media by quantitative RT-PCR. Data represents the average of three independent experiments ± SD. Accompanying immunoblot (right panel) of the three cell lines grown in the presence and absence of 10 nM DHT shows AR and β-actin levels. (C) CWR22Rv1-AR-EK cells were subject to control (siScr) or AR (siARex1) knockdown for 48 hours before ChIP experiments incorporating either N-terminal AR-binding or control (IgG) antibodies. Data represents the average of three independent experiments ± SD (*, **, ***, **** represent P< 0.05, 0.01, 0.001 and 0.0001, respectively as determined using one-way ANOVA). Accompanying immunoblots (right panel) of CWR22Rv1-AR-EK whole cell lysates (WCL) and chromatin fractions, incorporating AR and histone H2B antibodies, demonstrates successful depletion of AR-Vs in siARex1-transfected cells. (D) Representative bright field (BF) and immunofluorescence images using an anti-AR antibody (left panel) and DAPI counterstain (right panel) in CWR22Rv1-AR-EK cells. Scale bars are 25 μm.

Figure 3.

AR-Vs drive proliferative and survival signals in CWR22Rv1-AR-EK cells. (A) Heatmap of the log transformed normalised expression of genes up- and down-regulated in triplicate CWR22Rv1-AR-EK cells subject to either control (siScr) or AR-V (siARex1) depletion. The data is row-scaled with red and blue representing relative higher and lower expression, respectively. (B) Venn diagrams showing overlap between AR-V-target genes in CWR22Rv1-AR-EK cells and those derived from CWR22Rv1 parental cells depleted of AR-Vs (He et al., 2018 & Jones et al., 2015). (C) Functional annotation of AR-V regulated genes in the CWR22Rv1-AR-EK cell line demonstrates that AR-Vs control cell cycle and mitosis-related pathways. The % of genes identified in each pathway are shown alongside statistical significance of genes featuring in these pathways. (D) CWR22Rv1 and CWR22Rv1-AR-EK cells were grown in steroid-depleted media and subject to transfection with either control (siScr), FL-AR/AR-V-targeting (siARex1) or AR-V-targeting (siAR-V) siRNAs for 96 hours before analysis of cell proliferation by SRB assays. Data represents the average of three independent experiments ± SD (*** and **** represent P< 0.001 and 0.0001, respectively as determined using one-way ANOVA). Lower immunoblotting panels indicate successful depletion of FL-AR and AR-Vs using siARex1 and discriminate knockdown of AR-Vs by siAR-V using an anti-AR antibody. (E) Cells transfected as in (D) were subject to clonogenic assays for 2 weeks before quantification. Representative colony numbers are shown in the left panel. Data in the right panel represents the average of three independent experiments ± SD (** represents P< 0.01 as determined using a two-tailed Student's t-test). Lower panel immunoblot image indicates successful depletion of AR-Vs using an N-terminal AR-binding antibody.

Figure 4.

AR-Vs drive a DNA damage response gene signature to desensitise cells to ionising radiation. (A) Functional annotation demonstrates that AR-V-regulated genes in the CWR22Rv1-AR-EK cell line control the DNA damage response (DDR). The % of genes identified in each pathway are shown alongside statistical significance of genes featuring in these pathways. (B) Heatmap showing log transformed normalised expression of the 41 DDR-associated genes identified in triplicate CWR22Rv1-AR-EK cells transfected with either control (siScr) or AR (siARex1) siRNAs. The data is row-scaled with red and blue representing relative higher and lower expression, respectively. (C) Venn diagram demonstrating overlap of the 41 DDR-associated genes identified in CWR22Rv1-AR-EK cells and those identified in CWR22Rv1 cells depleted of AR-Vs (He et al., 2018 and Jones et al., 2015). (D) CWR22Rv1-AR-EK cells transfected with control (siScr) or AR-V-targeting (siAR-V) siRNAs were treated with and without 2 Gy ionising radiation and then incubated for 24 hours before quantifying γH2AX foci by immunofluorescence. Representative γH2AX/DAPI images are shown. Scale bars are 50 μm. Data in the right panel represents the average of two independent experiments ± SD (*** represents P< 0.001 as determined using a Mann-Whitney test). (E) Cells transfected as in (D) were subject to clonogenic assays for 2 weeks with representative colony numbers shown in the left panel. Colonies were quantified in three independent experiments (error bars represent SD and *, ***, **** represent P< 0.05, 0.001 and 0.0001, respectively, as calculated using two-way ANOVA).

Figure 5.

AR-V activity is controlled by PARP1/2. (A) Venn diagram indicating overlap between the CWR22Rv1-AR-EK DDR gene set and a ‘BRCAness’ gene signature identified in Li et al., 2017. (B) CWR22Rv1-AR-EK cells were treated for 24 h with and without 1 μM rucaparib (Ruc) before qRT-PCR analysis of ‘BRCAness’-associated genes. Data represents two independent experiments ± SD (*P< 0.05 as determined using a two-tailed Student's t-test). (C) CWR22Rv1-AR-EK cells were subject to immunoprecipitation (IP) using either AR or control (IgG) antibodies and resultant immunoprecipitates were immunoblotted with an anti-PARP1/2 antibody. (D) CWR22Rv1-AR-EK and CWR22Rv1 cells were treated with 0.5 and 1 μM olaparib (Olap) for 24 h before quantitative RT-PCR to assess AR-V target gene expression. Data represents the average of three independent experiments ± SD (NS, not significant; *, **P< 0.05 and 0.01, respectively, as determined using a two-tailed Student's t-test). (E) CWR22Rv1-AR-EK and CWR22Rv1 cells were subject to ChIP using either anti-PARP1/2 or control (IgG) antibodies to assess protein enrichment at AR target genes PSA and CCNA2. Data represents the average of two independent experiments ± SD (*, ** P< 0.05, 0.01, respectively, as determined using a two-tailed student T-test). (F) CWR22Rv1-AR-EK cells were treated with and without 1 μM talazoparib (Talaz) for 4 and 8 h before ChIP using AR, PARP and control (IgG) antibodies to assess protein enrichment at AR target genes PSA and CCNA2. Data represents the average of two independent experiments ± SD (NS, not significant; *, ** P< 0.05 and 0.01, respectively, as determined using a two-tailed Student's t-test).

Figure 6.

A feed-forward AR-V-PARP regulatory loop facilitates AR-V activity in CRPC. (A) PARPBP and PARP2 mRNA levels were analysed by quantitative RT-PCR in CWR22Rv1-AR-EK and CWR22Rv1 cells transfected with control (siScr) or AR (siARex1) siRNAs for 48 h; CWR22Rv1 cells were also grown in the presence and absence of 10 μM enzalutamide. Data represents the average of three independent experiments ± SD (*P< 0.05 as determined using a two-tailed Student's t-test). (B) Cellular PARP activity was assessed by immunoblotting in CWR22Rv1-AR-EK and CWR22Rv1 cells depleted of AR (siARexon1) using an anti-PAR antibody. Lysates were also probed for PARP1, AR and α-tubulin antibodies. (C) CWR22Rv1-AR-EK cells were treated with and without either 1 and 10 μM talazoparib (Talaz) or olaparib (Olap) for 96 hours before cell count analysis. Data represents the average of three independent experiments ± SD (**P< 0.01, as determined using a two-tailed Student's t-test). (D) LNCaP cells transduced with control or AR-V7-expressing lentivirus for 24 h and then treated with 1 μM talazoparib (Talaz) for an additional 24 h were subject to quantitative RT-PCR to assess expression of AR-target (upper panel) and DDR-associated (lower panel) genes. Data represents the average of three independent experiments ± SD (**P< 0.01 as determined using a two-tailed Student t-test). Lower immunoblot image demonstrates ectopic expression of AR-V7 in LNCaP cells transduced with AR-V7-expressing lentivirus. (E) Diagrammatic representation of interplay between AR-Vs, the DDR pathway and PARP activity in cells. Expression of AR-V-regulated genes, including those involved in the DDR, is enhanced by PARP1/2. The ability of AR-Vs to up-regulate PARPBP and PARP2 expression, which enhance cellular PARP activity, potentiates the existence of a feed-forward regulatory loop in CRPC.