Upregulation of the interferon-inducible antiviral gene RSAD2 in neuroendocrine prostate cancer via PVT1 exon 9 dependent and independent pathways

Abstract

PVT1 exon 9 overexpression is a newly uncovered aberration in prostate cancer (PCa). We have previously demonstrated the exon 9 region of PVT1 is overexpressed in some patient PCa tissues and caused development of neuroendocrine prostate cancer (NEPC) in vitro and in vivo. In this study, we focused on elucidating downstream mechanisms induced by PVT1 exon 9 overexpression with the goal of further understanding its role in NEPC development. RNA-seq analysis of a PVT1 exon 9 overexpressing PCa model revealed significant enrichment of genes responsible for inducing inflammatory processes including RSAD2. We observed RSAD2 overexpression in all NEPC models examined whereas PVT1 exon 9 was overexpressed only in a subset of the NEPC models. We identified two distinct pathways in which RSAD2 is overexpressed: one dependent and one independent on PVT1 exon 9 overexpression. Knockdown of RSAD2 suppressed cell proliferation and migration suggestive of its role as a therapeutic target in NEPC. We identified RSAD2 induces increased cell proliferation, colony formation, and may be involved in the transition between CRPC and NEPC. Distinct differences between PVT1 exon 9-dependent and PVT1 exon 9-independent NEPC models include differences in type II interferon signaling and AR modulation. PVT1 exon 9 binds to RSAD2 protein and disruption of binding significantly impedes downstream interferon gamma secretion by PVT1 exon 9-dependent NEPC cells. These novel findings indicate the importance of these two independent pathways in NEPC, the need to identify relevant NEPC patient populations and study strategies for targeting PVT1 exon 9 and/or RSAD2.

Keywords: PVT1; androgen receptor; cell biology; cellular immune response; interferon; long-noncoding RNA; neuroendocrine; prostate cancer.

Figure 1

RNA-seq analysis of PVT1-exon 9 overexpression.A, volcano plot of RNA-seq data comparing RWPE1_ex9 versus RWPE1_ev. Volcano plot was made in RStudio comparing log₁₀(p-value) and log2 Fold change. B, volcano plot of RNA-seq data comparing RWPE1_ex9 versus RWPE1_WT. Volcano plot was made in RStudio comparing log₁₀p-value and log2 Fold change. C, assessment of RSAD2 expression in Weill Cornell Medicine dataset comparing benign (n = 28), hormone-sensitive prostate cancer (n = 66), castrate-resistant prostate cancer (n = 72), and neuroendocrine disease (n = 35). Statistics were provided by collaboration (RShiny application) comparing benign tissue all tumor subtypes. D, cBioPortal analysis of all relevant prostate cancer cases comparing survival between RSAD2 altered and nonaltered disease. Statistics were provided by cBioPortal software where significance is indicated with p-value <0.05.

Figure 2

PVT1 exon 9 overexpression may be caused by alternative promoter site on PVT1 gene.A, RT-qPCR expression analysis of individual PVT1 exons from RWPE1_VPR cells transiently treated with five guides that span the PVT1 canonical promoter region (sequences described in Table S1) (two biological replicates). RPL32 and negative control guides were used as normalization controls. B, RT-qPCR expression analysis of RSAD2 from RWPE1_VPR cells transiently treated with five guides that span the PVT1 canonical promoter region (sequences described in Table S1) (two biological replicates). RPL32 and negative control guides were used as normalization controls. C, RT-qPCR expression analysis of RWPE1_VPR cells transiently treated with two guides that span the PVT1 alternative promoter region assessing each individual exon (two biological replicates). RPL32 and negative control guides were used as normalization controls. D, representative Western blot of RWPE1_VPR cells transiently transfected with two guides which span the PVT1 alternative promoter region assessing RSAD2 expression. Positive control cells (RWPE1_ex9) were also included. GAPDH was used as housekeeping gene. E, RT-qPCR expression analysis of RWPE1_ex9 cells silenced for PVT1 exon 9 (siRNA) and assessing RSAD2 expression. RPL32 was used as normalization control (three biological replicates). Statistics were provided by PRISM software using two-tailed unpaired t test at 95% confidence interval. Representative Western blot of RWPE1_ex9 cells silenced for PVT1 exon 9 (siRNA) and assessing RSAD2 expression. We found loss of RSAD2 expression when PVT1 exon 9 was knocked down. GAPDH was used as housekeeping gene. F, RT-qPCR expression analysis of RWPE1_ex9 cells silenced for RSAD2 (siRNA) and assessing PVT1 exon 9 expression (five biological replicates). RPL32 was used as normalization control.

Figure 3

PVT1 exon 9 and RSAD2 are expressed differentially in NEPC models.A, RWPE1, RWPE1_ev, RWPE1_ex9, 22RV1, T22OH, C22OH, DU145, NCI-H660, and PC3 cell lines assessing PVT1 exon 9 expression. Expression was normalized to RPL32 housekeeping gene (three biological replicates). All statistics (p-values) were performed using PRISM software using two-tailed unpaired t test with 95% confidence interval. B, RT-qPCR expression analysis of RWPE1, RWPE1_ev, RWPE1_ex9, 22RV1, T22OH, C22OH, DU145, and PC3 cell lines assessing RSAD2 expression. Expression was normalized to RPL32 housekeeping gene (three biological replicates). All statistical analysis (including p-values) were performed using PRISM software using two-tailed unpaired t test with 95% confidence interval.

Figure 4

PVT1 exon 9 binds to RSAD2 protein.A, RNA-immunoprecipitation experiment assessing baseline PVT1 exon 9 after RSAD2 pulldown using RT-qPCR in RWPE1, RWPE1_ev, RWPE1_RSAD2, and RWPE1_ex9 cell models (two biological replicates). Fold change of PVT1 exon 9 after RNA-immunoprecipitation, either bound or unbound, was calculated based on initial cell RNA-input expression PVT1 exon 9. PRISM software provided p-values using two-sided unpaired t test at 95% confidence interval. B, RNA-immunoprecipitation experiment assessing baseline PVT1 exon 9 after RSAD2 pulldown using RT-qPCR in RWPE1, C22OH, and PC3 cell models (two biological replicates). Fold change of PVT1 exon 9 after RNA-immunoprecipitation, either bound or unbound, was calculated based on initial cell RNA-input expression PVT1 exon 9. PRISM software was used to perform statistical analysis and determine p-values using two-sided unpaired t test at 95% confidence interval. C, RNA-immunoprecipitation experiment assessing RWPE1_ex9 cells treated with either siControl or siPVT1 exon 9 (siex9 #1, siex9 #2), using two differing siRNAs (three biological replicates). Fold change of PVT1 exon 9 after RNA-immunoprecipitation, either bound or unbound, was calculated based on initial cell RNA-input expression PVT1 exon 9. PRISM software provided p-values using two-sided unpaired t test at 95% confidence interval.

Figure 5

RSAD2 overexpression promotes cell growth in normal prostate epithelial cells.A, validation of RSAD2 overexpression with RNA (three biological replicates) and representative protein in the RWPE1_RSAD2 generated cell line. PRISM software was used to perform statistical analysis and determine p-values using two-sided unpaired t test at 95% confidence interval. B, knockdown of RSAD2 in RWPE1_RSAD2 cell line showed no significant difference in PVT1 exon 9 expression suggesting RSAD2 overexpression does not impact PVT1 exon 9 overexpression and is downstream of this molecular aberration (three biological replicates). PRISM software was used to perform statistical analysis and determine p-values using two-sided unpaired t test at 95% confidence interval. C, cell proliferation assay assessing differences between RWPE1_ev, RWPE1_ex9, and RWPE1_RSAD2. We found similar changes in cell proliferation between RWPE1_ex9, which was previously published (7), and RWPE1_RSAD2 (n = 2). PRISM software was used to perform statistical analysis and determine p-values using two-sided unpaired t test at 95% confidence interval. D, colony formation assay (n = 3) comparing RWPE1_EV to RWPE1_RSAD2 cell line. We found significantly more colonies in the RWPE1_RSAD2 model than RWPE1_EV suggesting its potential for RSAD2 upregulation inducing pathogenic changes. PRISM software was used to perform statistical analysis and determine p-values using two-sided unpaired t test at 95% confidence interval. E, immunofluorescence of RSAD2 was obtained for RWPE1, RWPE1_ex9 (PVT1 exon 9 dependent), and PC3 (PVT1 exon 9 independent) cell lines, 4′,6-diamidino-2-phenylindole staining for nuclei staining and merged image of the two. Images were taken at 40x magnification.

Figure 6

RSAD2 knockdown leads to loss of cell viability and colony formation.A, MTT cell viability assay comparing silenced scrambled control versus siRSAD2 treatment in C22OH, DU145, and PC3 cell lines (three biological replicates). PRISM software was used for p-value using two-sided t test at 95% confidence interval. B, wound healing assay assessing whether knockdown of RSAD2 leads to decreased wound healing in RWPE1_ex9 compared to RWPE1_ev cell line (two biological replicates). PRISM software was used to perform statistical analysis and determine p-values using two-sided unpaired t test at 95% confidence interval.

Figure 7

RSAD2 upregulation stimulates type II interferon signaling mechanisms.A, normalization of the type I and II interferon gamma secretion of RWPE1_ex9, RWPE1_RSAD2, and RWPE1_EV cell lines using ELISA assay (two biological replicates). B, normalization of the interferon gamma secretion of RWPE1_WT, RWPE1_EV, RWPE1_ex9, C22OH, NCI-H660, DU145, and PC3 cell lines using ELISA assay (two biological replicates). PRISM software provided statistics using two-sided t test at 95% confidence interval. C, linear regression analysis of interferon gamma secretion and RT-qPCR expression of PVT1 exon 9 from all individual cell lines tested (representative of three biological replicates). PRISM software was used for linear regression analysis, using best fit-line, and provided the statistics and correlation coefficient. D, linear regression analysis of interferon gamma secretion and RT-qPCR expression of RSAD2 from all individual cell lines tested. PRISM software was used for linear regression analysis, using best fit-line, and provided p-value and correlation coefficient. E, knockdown of IRF3 and cGAS in RWPE1_RSAD2 cell line (two biological replicates) does not cause any significant change in RSAD2 mRNA expression. Representative Western blot of RWPE1_RSAD2 cell line with scrambled control, IRF3, or cGAS knockdown and showed no difference in RSAD2 protein expression in any condition. PRISM software was used for p-value using two-sided t test at 95% confidence interval. F, normalized secretion of interferon gamma using ELISA assay of silenced scrambled control versus siPVT1 exon 9 knockdown in RWPE1_ex9 cells using either siRNA #1 or siRNA #2 (two biological replicates). PRISM software provided p-values using two-way ANOVA with alpha 0.05. G, interferon gamma secretion was significantly increased at 72 h in the RWPE1_ex9 model and was increased somewhat in the RWPE1_RSAD2 to a lesser degree (two biological replicates). These results are consistent with our initial findings (Fig. 7A) which show substantially increased interferon gamma secretion in the RWPE1_ex9 model. PRISM software provided p-values using two-way ANOVA with alpha 0.05. H, normalized secretion of interferon gamma using ELISA assay of silenced scrambled control versus siRSAD2 knockdown RWPE1_ex9, C22OH, and DU145 cells (three biological replicates). PRISM software provided p-values using two-sided student's t test at 95% confidence interval.

Figure 8

PVT1 exon 9 contributes to AR suppression in NEPC models.A, RT-qPCR analysis of silenced scrambled control versus siPVT1 exon 9 in C22OH and DU145 cell lines (three biological replicates). Expression was normalized to RPL32 housekeeping gene. PRISM software was used for p-value using two-sided t test at 95% confidence interval. B–C, representative Western blotting of C22OH and DU145 assessing AR expression between silenced scrambled control and siRSAD2 as well as housekeeping genes GAPDH. D, cell viability assay assessing knockdown of either scrambled control, PVT1 exon 9, or RSAD2 alone or in combination with AR (three biological replicates). E, ELISA analysis of lysates collected from siControl or siAR (48 h timepoint) assessing interferon gamma secretion (two biological replicates) was found to not be significantly altered in either condition. PRISM software was used for p-value using two-sided t test at 95% confidence interval. F, RT-qPCR analysis of RWPE1_ex9 cell line treated with 100 μg/ml of exogenous interferon gamma and assessed for AR expression using RT-qPCR (three biological replicates). Expression was normalized to RPL32 housekeeping gene. PRISM software was used for p-value using two-sided t test at 95% confidence interval. G-H, RWPE1_ex9 and C22OH cell lines were treated with increasing doses of enzalutamide (10 nM–10 μM) for 72 h with either siControl, siRSAD2, or siPVT1 exon 9. Cells were processed for proliferation compared to normal control using MTT reagent (three biological replicates). All statistics were provided by PRISM software.

Figure 9

Schematic of neuroendocrine prostate cancer subsets which overexpress RSAD2 alone or in conjunction with PVT1 exon 9. The image of neuroendocrine prostate cancer (NEPC) was reused from Pal et al., 2019 (Genes) (7).