Splicing targeting drugs highlight intron retention as an actionable vulnerability in advanced prostate cancer

Abstract

Background: Advanced prostate cancer (PC) is characterized by insensitivity to androgen deprivation therapy and chemotherapy, resulting in poor outcome for most patients. Thus, advanced PC urgently needs novel therapeutic strategies. Mounting evidence points to splicing dysregulation as a hallmark of advanced PC. Moreover, pharmacologic inhibition of the splicing process is emerging as a promising option for this disease.

Method: By using a representative androgen-insensitive PC cell line (22Rv1), we have investigated the genome-wide transcriptomic effects underlying the cytotoxic effects exerted by three splicing-targeting drugs: Pladienolide B, indisulam and THZ531. Bioinformatic analyses were performed to uncover the gene structural features underlying sensitivity to transcriptional and splicing regulation by these treatments. Biological pathways altered by these treatments were annotated by gene ontology analyses and validated by functional experiments in cell models.

Results: Although eliciting similar cytotoxic effects on advanced PC cells, Pladienolide B, indisulam and THZ531 modulate specific transcriptional and splicing signatures. Drug sensitivity is associated with distinct gene structural features, expression levels and cis-acting sequence elements in the regulated exons and introns. Importantly, we identified PC-relevant genes (i.e. EZH2, MDM4) whose drug-induced splicing alteration exerts an impact on cell survival. Moreover, computational analyses uncovered a widespread impact of splicing-targeting drugs on intron retention, with enrichment in genes implicated in pre-mRNA 3'-end processing (i.e. CSTF3, PCF11). Coherently, advanced PC cells displayed high sensitivity to a specific inhibitor of the cleavage and polyadenylation complex, which enhances the effects of chemotherapeutic drugs that are already in use for this cancer.

Conclusions: Our study uncovers intron retention as an actionable vulnerability for advanced PC, which may be exploited to improve therapeutic management of this currently incurable disease.

Keywords: 3’-end mRNA processing; Advanced prostate cancer; Alternative splicing; Intron-retention; Splicing inhibitors; Transcriptomics.

Fig. 1

Alternative-splicing and expression changes elicited in 22Rv1 cells by treatment with Indisulam, Pladienolide B and THZ531. A Schematic representation of the experimental workflow and RNA-seq analysis performed in 22Rv1cells (PladB, pladienolide B). B Bar graphs representing the % of genes of 22Rv1 transcriptome regulated at either alternative splicing (AS, left panel) or gene-expression (GE, right panel) level. C Venn diagrams showing the overlap between genes regulated at AS and GE level by indicated treatment. For GE-regulated genes, either exclusively or in the overlap with AS, % of up- and down-regulated genes are indicated. D, E Venn diagrams showing the overlap for GE (D) and AS (E) regulated genes by the indicated drugs in 22Rv1 cells respect to control. F Bar graph illustrating the GO terms relative to biological processes significantly enriched within the AS regulated genes by all drugs in (E) (analysis performed with Enrichr tool, p-value < 0,05)

Fig. 2

Indisulam, Pladienolide B and THZ531 modulate specific splicing patterns in 22Rv1 cells. A Pie chart showing % of events regulated in indicated splicing pattern upon indicated treatment in 22Rv1 (fold change ≥ 2, p-value ≤ 0.05). B Bar graphs showing the percentages of events for each splicing pattern annotated in FAST-DB (white columns) or regulated in 22Rv1 upon indicated treatment (black columns, modified Fisher’s test). C Pie charts showing percentages of up- (red) or down-regulated (green) events for indicated splicing pattern (ALE, alternative last exon; EC, exon cassette; IR, intron retention) in 22Rv1 upon indicated treatments. D Venn diagram showing the overlap between annotated splicing events regulated by the indicated in treatment in 22Rv1 compared to control cells. E) Pie chart representing distribution of 35 commonly regulated splicing events by the three treatments in 22Rv1 (see panel D) among different splicing patterns. F-I Representative images of RT-PCR analyses for overlapped (F), Indisulam—specific (G), PladB – specific (H) and THZ531- specific AS events in 22Rv1 cells treated or not for 24 h with indicated drugs (Ind, indisulam; PladB, pladienolide Schematic representation for each event is depicted beside the representative agarose gel. Red and green boxes indicate respectively up- and downregulated exons, black arrows indicate primers used for the PCR analysis. Percentage of spicing inclusion (PSI) was evaluated by densitometric analysis, and results are shown below agarose gels (mean ± SD, n = 3, t-test)

Fig. 3

Specific sequence features characterize splicing-sensitive cassette exons to RNA splicing inhibitors in 22Rv1 cells. A-F Boxplots showing comparison between up-regulated exons (red boxes in A, B, C) or down-regulated exon (green boxes in D, E, F) by indicated splicing inhibitors (Ind = indisulam, PladB = Pladienolide B, THZ = THZ531) and other not-regulated cassette exons (ref. cassette, blue box) and constitutive exons (ref. constitutive, grey box) for strength of their 3' splice-site (A, D), distance of the branchpoint from the 3' splice-site (B, E) and % of GC content (C, F). Whiskers indicate 1.5 interquartile range and highlighted circles the mean values (*p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001; ns = not significant, Welch’s t test). Number of analysed exons within each group is indicated in Additional File 2: FigureS5.

Fig. 4

Alternative cassette exons in EZH2 and MDM4 genes affects 22Rv1 cells viability. A, B Kaplan Meier curves illustrating the progression (A) and disease free interval (B) of prostate adenocarcinoma (PC) patients of the TCGA cohort, segregated according to median expression levels of the EZH2 gene. C Schematic representation and profile of the RNA-seq reads of the indicated region of EZH2 gene in 22Rv1 cells treated with Indisulam (Ind) or Dmso. Sequence reads (vertical gray lines), exons (blue boxes), and introns (horizontal lines) are shown. Dashed red box highlights the RNA-seq reads along the upregulated exon 12 (red box). D Schematic representation of EZH2 exon 12 alternative cassette exon. Black arrows indicate primers used for the RT-PCR analyses of its inclusion in 22Rv1 treated or not with Ind (3,3 µM, 12 h). Percentage of spicing inclusion (PSI) is shown below the representative agarose gel (densitometric analysis, mean ± SD, n = 3, one-way Anova). NMD, non-sense mediated decay. E Representative Western Blot analysis for EZH2 protein levels in 22Rv1 cells treated or not for 24 h with Ind (3,3 µM). ACTIN was evaluated as loading control. Densitometric analysis of the EZH2/ACTIN ratio is shown below the blots (mean ± SD, n = 4, t-test). F Line graph showing growth rate of 22Rv1 cells treated with the EZH2 inhibitor GSK343 (10 µM) and Indisulam (3,3 µM), evaluated as cell confluence ratio relative to time 0 (t0; mean ± SD, n = 4, two-way Anova). G, H) Kaplan Meier curves illustrating the progression (G) and disease (H) free interval of PC patients of the TCGA cohort, segregated according to median expression levels of the MDM4 gene. I Schematic representation and profile of the RNA-seq reads of the indicated region of MDM4 gene in 22Rv1 cells treated with Pladienolide B (PladB) or Dmso. Sequence reads (vertical gray lines), exons (blue boxes), and introns (horizontal lines) are shown. Dashed green box highlights the RNA-seq reads along the downregulated exons 7 and 10 (green boxes). J Schematic representation of the alternative cassette exon events involving exon 7 and 10 of MDM4 gene. Black arrows indicate primers used for the RT-PCR analyses. K, L RT-PCR analysis of MDM4 exons 7 and 10 inclusion (upper panels) and western blot analysis (lower panels) of MDM4 protein levels in 22Rv1 treated or not with Plad B (10 nM, 6 h) (K) and in 22Rv1 transduced with control or MDM4 exon 7 targeting antisense oligonucleotide (ASO). ACTIN was evaluated as loading control. Results of the densitometric analysis of the PSI are shown below the agarose gel (mean ± SD, n = 3, t-test). Densitometric analysis of the MDM4/ACTIN ratio is shown below the blots (mean ± SD, n = 3, t-test). M) Line graph showing growth rate of 22Rv1 cells transduced with ASO-CTRL or ASO-MDM4 or treated with PladB (1 nM) or Dmso, evaluated as cell confluence ratio relative to t0 (mean ± SD, n = 2, two-way Anova)

Fig. 5

Pervasive intron retention is induced by splicing inhibitors in 22Rv1 cells. A For each indicated drug, the left pie chart illustrates the percentages of exonic (light blue) and intronic (purple) regulated events and the right pie chart highlights the percentages of up- (red) and downregulated (green) events within the subset of intronic events. B Box plots showing the expression levels (log2 FPKM) of regulated (light blue bar) and unaffected genes (reference, grey bar) by intron-retention in 22Rv1 cells treated with indicated drugs. C Curve graphs showing distribution of the upregulated introns (red line) by indicated drugs within the transcription unit of their hosting genes (5′ → 3’ direction). D-G Boxplots showing comparison between up-regulated introns (red boxes) by indicated splicing inhibitors (Ind = indisulam, Pb = Pladienolide B, THZ = THZ531) and other not-regulated intron-retention events (ref. intron, blue boxes) and properly spliced-introns (ref. constitutive, grey box) for indicated features. In B and D-G Whiskers indicate 1.5 interquartile range and highlighted circles the mean values (Welch’s t test). Number of analysed exons within each groups is indicated in Additional File 2: Fig S7.

Fig. 6

Splicing targeting drugs affects 3’-end processing genes through intron retention and alternative last exon selection. A Venn diagram showing the overlap of regulated alternative last exon (ALE) and intron-retention (IR) events regulated by Indisulam (Ind), Pladienolide B (PladB) and THZ531 in 22Rv1 cells, according to our RNA-seq data. B, C Bar graph illustrating for genes regulated by IR and ALE by at least two drugs (see panel A) annotation of significantly enriched pathways in the Reactome database (B) and putatively regulatory transcription factors according to query of indicated ChiP-seq databases (C) (analysis performed with Enrichr tool, p-value < 0.05). D Heatmap showing the gene-expression fold-change for indicated genes of the cleavage and polyadenylation (CPA) machinery of 3’-end mRNA: CPA specificity factor (CPSF), cleavage factor I (CFI) and II (CFII), cleavage stimulation factor (CSTF) and other genes according to RNA-seq data of 22Rv1 treated with indicated drugs with respect to DMSO treated cells. Circles inscribed in a square indicate regulation by an ALE or IR event. E Schematic visualization in the EASANA database of the RNA-seq reads profile for CSTF3 (upper panel) and PCF11 (lower panel) genes in 22Rv1 cells treated with either Ind, Plad B, THZ531 with respect to the DMSO. P-value and log2 of the gene expression fold-change with respect to DMSO are indicated. F) qPCR analysis of CSTF3 and PCF11 full length transcripts in 22Rv1 cells treated (for 16 h) with indicated drugs compared to control. RPL34 was evaluated as loading control (mean ± SD, n = 3 one-way Anova). G, H Representative Western Blot analysis for CSTF3 (G) and PCF11 (H) in 22Rv1 cells treated for 24 h (G) or 16 h (H) with indicated drugs. ACTIN was evaluated as loading control. I) Bar graph showing the results of the densitometric analysis of the ratio of expression of indicated proteins with respect to ACTIN (mean ± SD, n = 3, one-way Anova)

Fig. 7

JTE-607 inhibitor exerts anti-tumoral activity in CRPC cells. A Dose–response curve illustrating the growth inhibitory effects of JTE-607 treatment on 22Rv1 cells. Half maximal inhibitory concentration (IC₅₀) for the two cell lines is indicated (mean ± SD, n = 2). B Line graph showing the cell death ratio of LNCaP and 22Rv1 cells treated with 25 µM JTE-607or Dmso at indicated time point (mean ± SD, n = 3, two-way Anova). C Violin plot showing the expression levels of CPSF3 and PCF11 genes in primary prostate adenocarcinoma (PC, n = 429) or normal prostate tissue (n = 43) (TCGA dataset, median value is highlighted, Welch Anova test). D, E Kaplan Meier graphs of the progression free interval (D) and disease free interval (E) survival probability of PC patients in the TCGA database classified according to the expression levels of CSTF3/PCF11 genes. F Representative image of Incucyte® Cytotox green stained 22Rv1 at 48 h of treatment with indicated drugs (JTE-607 6 µM; CPT, cisplatin 1 µM, DCX, docetaxel 3 nM). Violet boundaries highlight cells positive for cytotox green dye. Nuclight Rapid NIR (blue color) stains cells nuclei. G, H Line graphs showing either the cell death ratio of 22Rv1 treated with either 6uM JTE-607 and CPT 1 µM (G, mean ± SD n = 3) or DCX (3 nM) (H, mean ± SD n = 2) alone or in combination with respect to vehicle (Dmso) treated cells (two-way Anova)