br Multiple AR Vs have been described
Multiple AR-Vs have been described in PCa models, xeno-grafts, and patient tumors (Lu and Luo, 2013). One of the most common variants, AR variant 7 (ARv7), arises through cryptic exon inclusion. This AR isoform consists of a conserved N-termi-nal activation function-1 (AF-1) domain, a central DNA-binding domain, a partial hinge region and a unique C-terminal 16-amino-acid sequence, in place of the classical LBD. ARv7, unlike the full-length receptor, is continuously localized to the nu-cleus (Hu et al., 2009), and its expression is correlated with androgen-independent cell proliferation and PCa progression (Dehm et al., 2008). ARv7 levels are increased in metastases of CRPC patients (Hornberg et al., 2011; Qu et al., 2015), and ARv7 is present in over 80% of CRPC patient samples in the SU2C PCa cohort (Robinson et al., 2015). Detection of ARv7 in circulating tumor AZD2281 (CTCs) from CRPC patients was associ-ated with resistance to anti-hormonal therapies in a pioneering study (Antonarakis et al., 2014). While some studies confirmed the correlation of ARv7 status with disease progression and overall survival (Antonarakis et al., 2017; Scher et al., 2016; Welti et al., 2016), one follow-up study reported that CTC ARv7 status in patients cannot entirely predict nonresponse to anti-hormonal treatment (Bernemann et al., 2017).
In this article we investigate the genomic function of ARv7 and its potential link with ARfl action by utilizing cistrome and tran-scriptome studies in CRPC cells.
LNCaP95 and 22Rv1 Cell Growth Is Dependent on ARfl and ARv7
A proposed mechanism underlying the reactivation of AR in CRPC is the increased expression of ARv7 (together with ARfl). Utilizing the CRPC cell line LNCaP95, which endogenously ex-presses ARfl and ARv7 and is cultured under hormone-starva-tion conditions, we assessed the dependency of this line for both AR isoforms. We created stable LNCaP95 lines harboring doxycycline (dox)-inducible short hairpin RNAs (shRNAs), selec-
tively silencing the expression of ARfl (shARfl), ARv7 (shARv7), or
a GFP control (shGFP) (Figures 1A and S1A). Knockdown (KD) of either AR isoform was sufficient to decrease LNCaP95 cell pro-liferation in 2D culture (Figure 1B) and in 3D cell culture in a poly(ethylene glycol) diacrylate (PEGda) cryogel matrix (Go¨ppert et al., 2016) compared with the shGFP control (Figures 1C, 1D, and S1B). Silencing of either ARfl or ARv7 in 22Rv1 cells, another CRPC model with high endogenous ARv7 expression, also significantly decreased cell growth (Figures S1C and S1D). Therefore, consistent with previous findings (Dehm et al., 2008; Guo et al., 2009; Hu et al., 2009), we conclude that both ARfl and ARv7 are necessary to support proliferation of CRPC cell line models under castration conditions.
ARv7 Functions as a Transcriptional Repressor in CRPC Cells
To better delineate ARv7 from ARfl function, we next profiled gene expression in the inducible shRNA cells by RNA sequencing (RNA-seq). Using differential expression analysis (DEseq; p < 0.05, fold change >1.5), we observed a higher num-ber of upregulated (n = 300) compared with downregulated genes (n = 129) in response to ARv7 depletion (Figure 2A), sug-gestive of a preferentially repressive ARv7 transcriptional function. In contrast, a slightly higher fraction of genes was downregulated (n = 293) rather than upregulated (n = 242) upon shARfl (Figure 2B). To identify the biological processes associated with these AR isoform-regulated genes, we next employed gene set enrichment analysis (GSEA) (Subramanian et al., 2005). In this analysis some ARfl-activated pathways over-lapped with ARv7-repressed ones, while others were unique, with an unclear discernible pattern (Table S1).
To assess the effect of dihydrotestosterone (DHT) on the AR isoform-regulated transcriptomes, we next performed RNA-seq in the inducible shRNA LNCaP95 lines treated with a range of DHT concentrations for 24 h (Figures 2C, S2A, and S2B; Table S2). We observed that across all DHT concentrations ARv7 re-tains its repressive effect, with significantly regulated genes (p < 0.05) being mostly upregulated after ARv7 KD (with median fold changes >1). Conversely, ARfl retains its activating effect at all DHT concentrations with dysregulated genes being mostly downregulated after ARfl KD (with median fold changes <1). Moreover, we only observed a limited DHT effect, with a stable amplitude for the ARv7-regulated genes (Figures 2C and S2A). In contrast, the distribution of ARfl-regulated genes increased proportionally to DHT increases (Figures 2C and S2B), which agrees with a model of hormone-dependent, ARfl-mediated transcriptional regulation in PCa cells (Wang et al., 2007).