The number of human genes (approximately 25 000) cannot explain or account for the entire human proteome. This observation challenged the traditional concept that one gene is transcribed into a single transcript and subsequently translated into one protein. It is now well established that one of the mechanisms that accounts for discrepancies between the annotated genome and the observable proteome is transcript manipulation via alternative RNA splicing.1 Next-generation RNA sequencing data suggest approximately 95% of all human transcripts with multiple exons undergo alternative splicing.2 The aberrant regulation of alternative splicing processes can lead to cancer-specific defects, conferring on the affected protein either loss-of-function or gain-of-function tumorigenic properties. Such changes can facilitate angiogenesis, invasion, and promote resistance to apoptosis, all of which positively influence cancer progression and metastasis.3,4
We previously undertook a proteomics study to identify and characterize active androgen receptor (AR) complexes to help define novel AR functions involved in prostate tumorigenesis.5 We postulated that the ability of AR to attract different interactors/coactivators or to interact with other pathways could result in differential gene expression; more so, we speculated that prostate cancer (PCa) gain-of-function AR mutants could exploit new avenues of growth-enhancing effects through new aberrant interactions. From this analysis, we identified the AR to interact with several RNA-binding proteins, and several of these RNA-binding proteins are known to regulate alternative RNA splicing.6–10 These RNA-binding/splicing proteins included PTB-associated splicing factor (PSF), splicing factor 3A (SF3A), DEAD (Asp-Glu-Ala-Asp) box helicase 5 (DDX5), DDX17, Src associated in mitosis of 68 kDa (SAM68), KH-type splicing regulatory protein (KHSRP), and several heterogeneous nuclear ribonucleoproteins (hnRNPs).5 Subsequent gene enrichment analysis showed that many of these genes were upregulated at the transcript level in tumor samples and were predictive of survival.5 Pre-mRNA splicing is an involved process of multiple coordinated events, from the formation of the protein splicing complex to the direction of spliceosome to recognize the splice site elements to the catalytic splicing process. Rather than operating post-transcriptionally, there is strong evidence that RNA splicing is an integrated process within transcriptional complex itself (i.e., cotranscription). This was first demonstrated in Drosophila where intron looping and associated ribonucleoprotein complexes of transcripts are tethered to DNA,11,12 which suggested the association of the mRNA splicing complex to DNA transcription sites preceding transcript release. The frequency of cotranscriptional splicing has been examined in purified nascent transcripts through RNA sequencing and array analysis and found high similarity in budding yeast (75%), fruit flies (83%), and human cell line and tissue (85%).13–16 Not all introns are removed by cotranscriptional splicing, specifically terminal introns are least removed via this process.17 Given the proximity of nascent RNA to the DNA axis and chromatin, chromatin immunoprecipitation (ChIP) can be employed to investigate the process of cotranscriptional spliceosome assembly on genes, with the spliceosome interactions to the nascent RNA preserved by the formaldehyde crosslinking and thereby revealing the regions of interactions with the DNA template.18 As a result, it was found that spliceosomal small nuclear ribonucleoproteins (snRNPs) accumulated in distinct patterns along the gene, in a very consistent step-wise pattern.19,20 Furthermore, splice site inspection, spliceosome assembly, and the catalysis of splicing events are also determined by transcription rate,21,22 and fluctuations in RNAPII elongation rates relative to the intron/exon architecture can be directly related to splicing activity.23 In addition, nascent RNA folding and the formation of R-loops within the transcriptional complex can promote or impede elongation.24
Evidence has been previously presented where steroid receptors have previously been shown to be associated with mRNA splicing properties to promote alternative splicing events, but the nature by which such events occurred remained unclear.25–28 Therefore, to further investigate the role for AR in RNA splicing, we examined the functional interaction between AR and the specific RNA splicing factors, SAM68, DDX5, and KHSRP, using a variety of methodologies that directly monitor alternative splicing events. Using well-established molecular tools and also exploring global splicing events in PCa cells, an emphasis was placed on the potential clinical impact of these RNA splicing factors and their splicing events have in PCa disease progression. Our investigations of AR-mediated alternative RNA splicing events suggest a robust and direct effect of AR via its interaction with RNA splicing factors. Through this process, we characterized the influence of AR-directed alternative splicing by identifying a set of genes/exons that are strongly predictive of disease-free outcomes and specifically, has led to the discovery of a unique subtype of lethal PCa.
MATERIALS AND METHODS
LNCaP and PC3 prostate cancer cell lines were obtained from the American Type Culture Collection (ATCC; Rockville, MD, USA). MEF+/+ and MEFSam68−/− cell lines were isolated and initially characterized by Dr. Stéphane Richard (Lady Davis Institute, Montreal, QC, Canada).
The mouse mammary tumor virus (MMTV)-CD44 mini-gene cassette, pCMV-GFP-SAM68-FL (full length, referred to as either SAM68-FL or SAM68), pCMV-GFP-SAM68-440YF (SAM68-440YF or 440YF), and pCMV-GFP-SAM68-G-D (SAM68-G-D or G-D) plasmids were generously provided to us by Dr. Stéphane Richard. The pCMV-YFP-DDX5 plasmid was supplied to us by Dennis Warner (University of Louisville, Louisville, KY, USA). The pcDNA3-FLAG-KHSRP was supplied by Ching-Yi Chen (University of Alabama at Birmingham, Birmingham, AL, USA), and the pCMV-AR45 was provided by Bernard Haendler (Bayer Schering Pharma AG, Berlin, Germany).
Mammalian cell culture, transfections, and hormone treatment
LNCaP or PC3 cell lines were cultured in RPMI 1640 media supplemented with fetal bovine serum (FBS; 11%; Wisent Bio Products, Saint-Jean-Baptiste, QC, Canada), at 37°C with 5% CO2 in plastic culture flasks. Once confluent, 1 ×106 cells were seeded into 6-well plates in the same medium to allow the cells to adhere. Twenty-four hours after plating, the medium was changed to RPMI supplemented with 10% charcoal-dextran stripped FBS (csFBS) and incubated for an additional 24 h. The following day, the medium was changed to fresh RPMI/csFBS for hormone treatment and inhibitor studies. Mibolerone (MB; a synthetic nonmetabolizable androgen) was used in all experiments rather than 5α-dihydrotestosterone (DHT; Sigma, St. Louis, MO, USA) or R1881 allowing for the receptor to maintain its active state for longer periods of time. Hormone treatment of MEF cells was carried out similarly, but in Dulbecco’s modified eagle medium (DMEM; Wisent Bio Products) media.
Transfections and prostate-specific antigen (PSA) enzyme-linked immunosorbent assays (ELISAs)
In brief, LNCaP, PC3, or MEF cell lines were transfected with Lipofectamine™ LTX (Invitrogen, Carlsbad, CA, USA), with 1 µg of DNA per 10 × 106 cells. Cells were plated in T-75 flasks in their respective media, prior to transfections the media was changed to 15 ml of Opti-MEM (Wisent Bio Products) supplemented with 5% csFBS. Twenty-four hours post-transfection, the medium was replaced with RPMI 1640 (LNCaP and PC3) or DMEM (MEF cells) containing 5% csFBS with or without MB. Cells were incubated with 10 nmol l−1 MB for 24 h for RNA analysis and for 4 days for measuring secreted PSA protein production in cell culture supernatants. All PSA stimulations were performed in triplicate with PSA ELISAs performed by Dr. Eleftherios Diamandis’ laboratory (Samuel Lunenfeld Research Institute, Toronto, ON, Canada).
Selective inhibitor studies
LNCaP cells were cultured and transfected as described above. Selective inhibitors (Wortmannin or U0126; EMD Bioscience, Brookfield, WI, USA) we added in the presence or absence of MB, to LNCaP cells as a single dose for 24 h, followed by RNA extraction and reverse transcriptase-PCR (RT-PCR) analysis.
RNA extraction and RT-PCR
Total RNA was extracted from cell lines using TRIZOL reagent (Invitrogen) following the manufacturers’ instructions. RNA concentration was determined spectrophotometrically, and 5 µg of total RNA was reverse transcribed into first-strand cDNA using the Superscript™ First-Strand Synthesis kit (Invitrogen) with an Oligo(dT) primer. PCRs were carried out using Qiagen HotStar Taq Polymerase (Qiagen, Valencia, CA, USA) on first-strand cDNA. The following primers were used for RT-PCR analysis, CD44 (forward: 5’-GAGGGATCCGCTTCCTGCCC-3’; reverse: 5’-CTCCCGGGCCACCTCCA-3’), β-ACTIN (forward: 5’-ATCTGGCACCACACCTTCTA-3’; reverse: 5’-CGTCATACTCCTGCTTGCTG-3’), and PSA (forward: 5’- CCCACTGCATCAGGAACAAAAGCG-3’; reverse: 5’-GGTGCTCAGGGGTGGCCAC-3’). An equal amount of each PCR product was run on 0.9% agarose gels and visualized by ethidium bromide staining. Inclusion/exclusion ratios of alternative CD44 splicing events, and fold changes in PSA, were determined using ImageJ software (http://rsbweb.nih.gov/ij/).29CD44 exon-inclusion and -exclusion splicing events were performed in at least triplicate (n ≥3) and were displayed via semiquantitative gel analysis to illustrate the shifting splicing events. Differences in exon selection and the determination of exclusion/inclusion ratios of the MMTV-CD44 mini-gene reporter or PSA gene expression were determined using ImageJ software (National Institutes of Health, Bethesda, MD, USA) and statistical tools.
Western blot analysis
Protein concentrations were determined using the BCA protein assay kit (Pierce Biotechnology, Rockford, IL, USA). A total of 20 µg of cleared cell lysate was used for Western blot analysis. Antibodies used for Western blot analysis included anti-β-ACTIN (AC-15; Abcam, Cambridge, MA, USA), anti-AR (N-20; Santa Cruz Biotechnology) and anti-DDX5 (C-20; Santa Cruz Biotechnology), anti-GFP (Roche, Mississauga, ON, Canada), anti-FLAG (Sigma), and anti-SAM68 (courtesy of Dr. Stéphane Richard).30
Tissue microarray (TMA) - protein expression assessment
Tissue samples were collected from archival samples processed at the Department of Pathology, Calgary Laboratory Services (Calgary, AB, Canada). The patient cohorts consisted of a PCa progression cohort (61 patients), which included samples across various disease stages (26 benign, 8 high-grade prostatic intraepithelial neoplasia [HGPIN], 30 localized PCa, and 30 castrate-resistant PCa [CRPC]) assembled on one TMA using 2–10 0.6-mm cores per sample with a total of 320 cores (Supplementary Table 1). Collection of clinical and pathological information from the retrospective cohort did not require written patient consent and was approved by the Health Research Ethics Board of Alberta (HREBA) – Cancer Committee (HREBA.CC-16-0551).
All TMA cores were assigned a diagnosis (i.e., benign, HGPIN, or PCa by a genitourinary pathologist [TAB]). For each sample, at least one available core was evaluated for hematoxylin and eosin (H&E), immunohistochemistry (IHC), and fluorescent in situ hybridization (FISH). Protein expression was assessed semiquantitatively and without a prior knowledge of the clinical information by evaluating the intensity of the expression, using a four-tiered system (0, negative; 1, weak; 2, moderate; and 3, strong), as previously described.31
IHC stain was performed on the Ventana autostainer (SAM68 and DDX5) or by manual staining (AR) on 4-µm thick sections from formalin-fix paraffin-embedded tissue blocks. Prior to the staining, heat-induced antigen retrieval procedure was carried out by vegetable steamer (SAM68 and AR) or pressure cooker (for DDX5) in ethylenediaminetetraacetic acid (EDTA) pH 9.0 buffer (10 mmol l−1 Tris/1 mmol l−1 EDTA) for a cycle of 20 min preheat, 20 min cooking, and 20 min natural cooling. Rabbit α-SAM68 was obtained from Dr. Stéphane Richard,30 rabbit α-DDX5 was purchased from Abcam, and mouse α-AR was purchased from Santa Cruz Biotechnology. Primary antibodies were applied onto slides and incubated for 60 min at 37°C at dilution of 1:100, 1:500, and 1:25 for SAM68, DDX5, and AR, respectively.
A Ventana iView DAB detection kit (Ventana, Tucson, AZ, USA) was used for horseradish peroxidase (HRP) detection when using the Ventana stainer, while MACH 4 HRP-polymer kit from BioCare Medical, LLC (Pacheco, CA, USA) was used for manual staining. Negative controls were performed by omitting the primary antibody and substituting it with normal mouse 1:200 prediluted serum (Ventana).
A semiautomated quantitative image analysis system, ChromaVision ACIS II (ChromaVision Medical Systems, Inc., San Juan Capistrano, CA, USA), was used to evaluate the tissue microarray slides for cytological distribution. The ACIS II device consists of a microscope with a computer-controlled mechanical stage. Proprietary software was used to detect the brown stain intensity of the chromogen used for IHC analysis and compared this value to the blue background counterstain. Theoretical intensity levels ranged from 0 to 255 chromogen intensity units. The correlation coefficient for the ASCIS II was r2 = 0.973, and the reproducibility for the system was previously tested and confirmed by scoring several tissue microarrays on separate occasions and in previous publications.31,32
Exon-microarray splicing expression analysis
LNCaP cells were transfected without or with DDX5 or SAM68; the next day, cells were stimulated without MB or with 10 nmol l−1 MB for 2 h, 6 h, or 18 h. Cells were collected and RNA was extracted using a combination of TRIZOL (Invitrogen) and cleaned RNeasy columns (Qiagen). Gene expression microarray analysis was carried out at CR-CHUM Microarray Facility (Montreal, Canada), using Affymetrix Human Exon 1.0ST array chips (Affymetrix, Santa Clara, CA, USA). Raw expression data were analyzed using open-source software AltAnalyze (http://www.altanalyze.org/),33 with a fold cutoff of 1.5 and P < 0.05. The datasets can be found at GSE176124 (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE176124).
We also utilized more patients’ datasets including GSE21032,34 GSE41408,35 and GSE4669136 to characterize the association of patient disease-free survival outcomes. GSE21032 represents an extensive spectrum of 218 patients with samples that include metastatic, and localized primary with accompany adjacent benign tissue, with 5-year follow-up. GSE41408 is a smaller patients’ sample size of 48, but with expanded clinical data of prostate cancer-specific death (PCSD), which was provided to us by Dr. Joost Boormans (Erasmus Medical Centre, Rotterdam, The Netherlands). GSE46691 consisted of 545 patients’ samples, of which 100 patients have confirmed PCSD.
AR-dependent alternative splicing events are altered through interacting partners: SAM68, DDX5, and KHSRP
We have previously shown the coimmunoprecipitated (co-IP) interaction between AR and SAM68, DDX5, and KHSRP to be RNA-dependent;5 therefore, we selected these proteins to directly investigate their interactions with AR to dictate alternative RNA splicing. To do so, we employed exon mini-gene cassette studies, as they are widely used to characterize alternative splicing events in vitro.6,37,38 These splicing reporter constructs typically yield multiple spliced RNA products that differ in size, allowing one to assess both exon-inclusion vs -exclusion differential splicing products within a single PCR reaction. The mini-gene cassette we opted to use is the MMTV-CD44 mini-gene construct that is transactivated in AR-expressing cells upon androgen exposure. Selection of a CD44 mini-gene cassette is of clinical relevance to PCa because its mRNA has been shown to be alternative spliced in PCa and is linked to disease progression.39–42
Parallel transfection studies were performed in LNCaP (AR positive and hormone sensitive) and PC3 (AR null) PCa cells, where each expression construct (SAM68, DDX5, or KHSRP) was cotransfected with the MMTV-CD44 mini-gene cassette to assess their role in AR-mediated RNA splicing. LNCaP cells transfected with only the MMTV-CD44 mini-gene, the prominent resulting PCR product observed was an exon-exclusion splicing event upon androgen treatment. However, with SAM68 cotransfections, there was a shift toward exon inclusion as observed by the higher molecular weight PCR product which was further pronounced with androgen exposure (Figure 1a). When MMTV-CD44 and SAM68 were cotransfected into PC3 cells, neither androgen-dependent transactivation of the CD44 mini-gene nor alternative splicing (exclusion or inclusion) was observed (Figure 1b). This suggests that the splicing of the mini-gene in LNCaP transfections was dependent on both AR and SAM68.
Similar exon-inclusion events were observed when KHSRP was overexpressed in LNCaP cotransfection experiments with the MMTV-CD44 mini-gene (Figure 1c). Cotransfection of DDX5 overexpression, however, shifted the splicing of the MMTV-CD44 mini-gene cassette toward an exon exclusion event (Figure 1d) and again showed an androgen exposure augmented effect. To highlight the observed splicing events of the mini-gene/splicing factor transfections, several experiments were analyzed and displayed in a histogram in Figure 1e. Significant alternative splicing events were recorded in transfected AR-positive LNCaP cells (CD44 alone, P = 0.0007; SAM68+CD44, P = 0.003; DDX5+CD44, P = 0.01; and KHSRP+CD44, P = 0.006). Whereas, in transfected PC3 cells, we did not observe significant splicing of the CD44 mini-gene cassette (CD44 alone, P = 0.288; and SAM68+CD44, P = 0.16). These cotransfection experiments in LNCaP vs PC3 cells highlight the necessity for AR to confer both androgen-dependent gene activation and facilitate alternative splicing events of the MMTV-CD44 mini-gene cassette. These results, preformed in hormone responsive LNCaP cells, recapitulated previous results of AR cotransfected HEK293 cells.6,37 Moreover, these results also suggest that the participation of AR in specific alternative splicing events, inclusion or exclusion splicing, would depend on the presence of and the interaction with specific splicing factors.
SAM68 influences AR-dependent transcription and splicing activities
SAM68’s role as an RNA splicing factor has been widely characterized.43 Therefore, to implicate a functional role for SAM68 and AR in cotranscriptional splicing, we utilized well-characterized SAM68 loss-of-function mutants: SAM68-440YF, a phosphorylation mutant with impaired nuclear translocation;44 SAM68-G-D, a dominant-negative deficient RNA-binding mutant;30 and SAM68-ΔN (ΔN), an N-terminal deletion null mutant that does not perturb either SAM68 nuclear translocation nor RNA-binding/splicing activity and thus served as a negative control.45 Confocal imaging confirms the expression and the localization patterns of the SAM68 isoforms used; wild type (WT)-SAM68 and SAM68-G-D isoforms are both able to localize to the nucleus of cell, whereas as expected the SAM68-440YF does not (Figure 2a). AR under androgen-stimulated conditions also display distinct nuclear translocation.
By utilizing these SAM68 mutants, we could also examine specific functional properties and facets of hormone-dependent splicing. Using WT-SAM68 and SAM68 mutants, we monitored both splicing and hormone-dependent transcriptional events in LNCaP cells transfected with the MMTV-CD44 mini-gene cassette (Figure 2b). By simultaneously evaluating PSA expression, we could also assess the influence of SAM68 mutants on the transactivation of an endogenous androgen-dependent gene. In these experiments, WT-SAM68 mediated splicing as previously described, and as expected, the SAM68-ΔN mutant did not affect MMTV-CD44 gene expression nor exon inclusion/exclusion ratios of the CD44 mini gene nor androgen-dependent PSA transcription. The expression of SAM68-440Y/F did not hinder PSA expression or hormone-dependent expression from the MMTV-CD44 construct; it did, however, repress selective splicing of the CD44 mini-gene. The expression of the dominant-negative SAM68-G-D mutant entirely abolished androgen-dependent transactivation of PSA and CD44 and subsequent exon-inclusion splicing of the mini-gene. These observations confirm the association of SAM68 and AR cotranscriptional splicing functional interaction.
To substantiate the cotranscriptional splicing function of SAM68 on AR-mediated transactivation, we also employed mouse embryonic fibroblast (MEF) cell lines, in which Sam68 has been knocked out.46 Similarly, to loss-of-function SAM68 mutants, AR transfected Sam68-null MEF cells also failed to illicit hormone-dependent activation of the MMTV-CD44 mini-gene cassette (Figure 2c). Therefore, we can confirm that both AR and SAM68 are needed to confer hormone-dependent gene activation and alternative splicing of the CD44 mini-gene transcript and support the proposition that splicing and transcription are bilateral events from singular complexes, rather than sequential events.
PI3K/AKT inhibition abrogates AR-dependent splicing events
The influence of the PI3K/AKT or MEK/ERK signal transduction pathways on the regulation of alternative splicing has been recently reviewed.47,48 To further dissect the signaling events required for splicing events observed during AR/SAM68 and AR/DDX5 interactions, we blocked the PI3K/AKT or MEK/ERK pathways using the selective chemical inhibitors Wortmannin and U0126, respectively. LNCaP cells were cotransfected with the MMTV-CD44 mini-gene cassette and RNA splicing factors SAM68 or DDX5 (Supplementary Figure 1). Inhibition of PI3K/AKT (Wortmannin) but not MEK/ERK (U0126) blocked the normally observed androgen-dependent exon-exclusion splicing of CD44 in LNCaP cells transfected with either SAM68 or DDX5 expression constructs. It has been shown that AR transcriptional activity is sensitive to both PI3K and MEK pathways;49,50 our data suggest that the PI3K/AKT signal transduction pathway plays the more prominent role in the transcriptional-splicing activities of AR.
Overexpression of RNA splicing factors (SAM68 and DDX5) influences AR protein expression and downstream AR activity
During our analysis of androgen stimulated LNCaP cells, overexpressing SAM68, showed a reduced expression of full-length AR protein vs mock transfected cells (Figure 3a). A similar diminution of full-length AR protein was observed with KHSRP overexpression; on the contrary, with the transfection of DDX5, an increase in AR protein was seen. We then examined whether these observations of changes in AR-FL correlated to decreases in AR transactivation of PSA. Expression analysis showed that, although full-length AR protein expression appeared to be reduced, high levels of PSA transcript (Figure 3b) and protein (Figure 3c) expression were still maintained in SAM68- and KHSRP-overexpressing cells. Increased PSA mRNA and protein expression was recorded, supporting the observation of increased expression of full-length AR in DDX5-overexpressing LNCaP cells, which inversely correlates with the finding of Clark et al.6 where siRNA knockdown of DDX5 was associated with a decrease of AR and PSA expression. Of DDX5 vs mock transfections, although non-significant, a trend of higher PSA protein levels was observed.
Changes in AR expression levels upon overexpression of SAM68 and KHSRP may be the result of a direct regulatory splicing event on AR mRNA itself that would result in another AR isoform that can maintain high hormone-dependent transcriptional activity. Functional AR splicing variants have been characterized.51,52 We used RT-PCR screening to try to identify possible variants that could account for our observed results but could not find an amplicon that would correspond to one of the AR splice variants (data not shown). There is another AR variant, AR45, which utilizes a novel exon 1 transcriptional start53 (Figure 3d) and has been shown to possess intact hormone-dependent gene activation. It is also widely expressed in localized and advanced PCa.54 We also found the AR45 variant to be increased in its expression in our SAM68 and KHSRP transfected LNCaP cells (Figure 3e). We next examined if the cotransfection of the AR45 variant and SAM68 in PC3 cells was still able to promote hormone-dependent exon-inclusion events with the CD44 mini-gene cassette (Figure 3f and 3g). In the absence or reduced expression of full-length AR, the AR45 variant can still maintain hormone-dependent gene activation and facilitate alternative splicing events and could suggest that other AR protein variants, such as the splice variants, could also have RNA splicing activities.
Temporal and spatial differential expression of RNA splicing factors during prostate cancer progression
To determine the clinical association of SAM68 and DDX5 RNA splicing factors during PCa progression, we employed TMAs to document protein expression profiles of SAM68, DDX5, and AR. In prior studies, nearly all “androgen-independent” or “castrate-resistant” prostatic tumors express high levels of AR which are predictive of progressive disease.55,56 Our PCa progression TMA libraries included benign tissue, HGPIN, localized PCa, and CRPC samples (patient demographics in Supplementary Table 1). Qualitative observation of immunohistochemical staining of all three proteins showed increasing expression of all proteins from benign to CRPC stages (Figure 4a). Although prostatic TMAs have been performed for DDX56 and SAM68,37 our TMAs are more informative as they include a wider spectrum of stages and a more thorough level of analysis. This was accomplished by employing digital imaging quantification to determine the relative cellular concentration and cellular distribution of SAM68, DDX5, and AR within the tumor sample. SAM68 displayed an abundant expression, predominantly nuclear, in benign, HGPIN, and localized PCa followed by notable decrease of nuclear expression in CRPC (Figure 4b). Conversely, with DDX5, we observed a different cellular distribution pattern, with a marked cytoplasmic drop-off and an increase in nuclear concentration in CRPC tissues. AR immunostaining showed a trend of increasing cytoplasmic and nuclear concentrations throughout all disease stages. This scenario suggests a dynamic but converse functional relationship between SAM68 and DDX5 with AR. The compartmental nuclear shifting of SAM68 and DDX5 from localized to CRPC disease stages may provide a mechanism for specific splicing proteins and splicing events to be uniquely expressed during transitions from localized to metastatic disease.
Global analysis of androgen-dependent alternative splicing events in prostate cancer: from in vitro studies to clinical analysis
To explore the hypothesis that endogenously cellular AR-regulated genes may also be influenced in their exon selection via SAM68 or DDX5 interactions, we employed an exon array expression platform to document and assess differentially spliced genes under androgen-stimulated conditions and nonstimulated conditions. Our findings with the TMAs suggest that expression of these splicing factors change with disease progression and subsequently could impact on differentially regulated splice/exons genes. Potential prognostic markers for disease progression and outcomes can be defined. Moreover, these alternatively spliced genes and their protein variants could contribute to the cellular pathophysiology and better understanding of PCa.
For our assessment of AR-mediated global alternative splicing events, we used LNCaP cells under different experimental conditions; (1) LNCaP cells treated with or without MB (referred to as the AR condition), (2) LNCaP transfected with GFP-SAM68 with or without MB (SAM68 condition), and (3) LNCaP transfected with YFP-DDX5 with or without MB (DDX5 condition). The MB-stimulated experiments were also carried over a time course of 2 h, 6 h, and 18 h (or overnight, O/N), to ascertain early, mid, and late hormone gene-splicing events. Each experimental condition has both unique and shared set of alternately splice exons: at 2 h (277 exon sets), at 6 h (477 exon sets), and at the O/N time point (175 exon sets; Supplementary Figure 2). The in vitro exon array experimental conditions evaluated expressed genes that displayed androgen sensitivity and as having distinct clinical states, i.e., overexpression of SAM68 or overexpression DDX5, rather than describing them as disease progression profiles originated from the LNCaP condition. To ascribe clinical relevancy, we compared our in vitro alternatively spliced gene sets with a tumor exon array generated from a large clinical PCa dataset (GSE21032).34 This clinical sample set consists of 29 normal control, 127 localized primary, and 16 metastatic disease patients. Being aware of the genetic heterogeneity in clinical datasets, we set the fold-change cutoff for differential exon expression comparing primary or metastatic vs normal tissue at 1.25 (P<0.05). A secondary requirement of the analysis was that we dictated that the spliced gene from our in vitro studies and clinical datasets needed to share the same splicing event(s) (i.e., splice direction). These conditions ensured that the filtered exon gene sets were associated with clinical sampling and not expression artifacts of cell lines.
The matched exon/gene sets are as follows: for the AR experimental condition we identified, 29 exons at 2 h (representing 26 genes), 79 exons at 6 h (62 genes), and 37 exons for the overnight time point (33 genes). For the SAM68 conditions we identified, 32 exons at 2 h (28 genes), 40 exons at 6 h (36 genes), and 53 exons for the overnight time point (47 genes). Finally, for the DDX5 condition, we identified 24 exons at 2 h (21 genes), 35 exons at 6 h (32 genes), and 30 exons for the overnight time point (25 genes; Supplementary Table 2). Initial analysis clearly indicates that there are multiple splicing events occurring within a single gene, and similarly, Database for Annotation, Visualization and Integrated Discovery (DAVID; david.ncifcrf.gov) confirmed that 50%–70% of the genes observed to be spliced in our clinical data analysis are already annotated as showing to be alternatively spliced in PCa disease pathways; examples include BCL2 apoptosis regulator (BCL-2),57 mucin 4 (MUC4),58 MAX interactor 1 (MXI1),59,60 MET proto-oncogene (MET),61,62 Rho guanine nucleotide exchange factor 26 (SGEF),63,64 and myosin light chain kinase (MYLK).65,66
The 37 exons corresponding to AR-O/N time-point could readily segregate normal, primary, and metastatic samples, as illustrated in Figure 5a, suggesting that these splicing events themselves have critical properties with respect to disease transition. Using this clustering analysis, we could also define disease-free outcomes between patient groups and found that the corresponding 37 exon gene set showed significantly (P=0.0212) predictive power of unfavorable disease-free outcome. Moreover, although these exon-gene sets provided significant outcomes values, they were independent of any clinical or pathological criteria (PSA: P = 0.0563, Gleason score: P = 0.1997, tumor stage P=0.9497, and median age: P =0.67; Figure 5b). Of interesting note, the clinical sample clustering of the expression of the 79 AR-6HR exon-gene set segregated the primary tumors into three patient groups. Significant disease-free outcome (P=0.008) was observed between clusters 1 and 3, but not between 1 and 2 or 2 and 3 clusters. This may suggest that patients in cluster 2 may be undergoing a transitional or transformational disease change of their exon expression profile from favorable to unfavorable outcomes, and should be considered for closer clinical monitoring (Supplementary Figure 3). Thus, the results derived from our comparative analysis of global splicing events in LNCaP cells in conjunction with clinical datasets identified pathological splicing programs that correspond with a better degree of confidence for predicting unfavorable disease outcome vs classical clinical criteria (PSA, Gleason score or tumor stage). The observation that we could assess significant disease-free outcomes from our exon gene sets, independent of clinicopathological criteria, highlights the sensitivity of molecular profiling prior to the manifestation of morphological characteristics used to assess disease aggressiveness.
Significant and unfavorable disease-free outcome results were also obtained with the exon-gene set analysis of DDX5 conditions and their correlated clinical sample analysis, DDX5-6 h (P = 0.0498), and DDX5-O/N (P = 0.0017). However, significant disease-free outcomes were not observed for the corresponding time points of the SAM68 exon-gene sets (data not shown). Although we found a large number of genes to be alternatively spliced in our 2 h condition time point, neither gene nor exon gene sets from either experimental condition provided us a significant value of disease-free outcome. This suggests that expression profiles of early androgen-dependent genes and exons, are not major contributors of progression to advanced disease.
Identifying a novel subtype of lethal PCa
Expression profiling of our exon gene sets with PCa patient samples clustered metastatic disease patients very well. Therefore, we aimed to investigate and further validate this aggressive prognostic value of our AR and DDX5 exon gene-sets with another set of clinical samples, specifically of patients with advanced/systemic PCa disease. GSE4140835 and GSE4669136 are two datasets containing patients with systemic disease associated with the progression of disease malignancy to PCSD. GSE46691 included 98 patients and GSE41408 included 8 patients, whose outcomes were associated with PCSD.
These datasets provided an opportunity to test whether these exon-gene sets have a significant prognostic correlation with disease outcomes among patients with the definitive clinical disease feature of PCSD. We, therefore, applied our AR-O/N exon gene set to the expression profiles of these 98 patients with PCSD and identified a segregated group of 9 patients (Figure 6a). The DDX5-O/N gene set was able to segregate the same 9 patients. This group of patients had significantly worse unfavorable outcomes based on both first (P = 0.0105) and second (P = 0.0179) biochemical recurrence (Figure 6b). Similar unfavorable disease-free outcomes from the GSE41408 clinical datasets were observed with 3 of the 8 patients (data not shown). This would suggest that our exon-gene sets have identified a unique subtype of lethal PCa that shows very early and aggressive development. Between AR and DDX5 exon-gene sets, five alternatively spliced exons are shared, which could support the initial hypothesis of a coordinated role of AR-DDX5 interaction in facilitating alternative splicing of a select set of genes.
We have shown the AR to interact with at least three RNA splicing factors, influencing and promoting distinct alternative splicing programs. The DNA-binding domain of AR is a well-characterized zinc finger (ZF)-binding domain motif but also has been demonstrated to act as an RNA-binding domain. The ZF containing transcription factor IIIA (TFIIIA) was discovered to bind 5S ribosomal RNA in Xenopus oocytes.67 Through this binding activity, TFIIIA can regulate the transport and storage of 5S particles. Moreover, the residues of ZF that mediate contacts with RNA overlap almost entirely with the DNA-contact residues in classical ZF:DNA complexes. Thus, the same surface of the classical ZF has been adapted to bind two different target structures in two different contexts. The ability of AR to have binding affinity to RNA has been historically described. Liao et al.68 first showed AR complexes bind to riboncleoprotein particles prepared from rat prostates. This was followed-up by others that demonstrated the AR to have high affinity binding to E. coli rRNA and tRNA, with the receptor also being able to distinguish between ssDNA and RNA.69,70 Others showed that AR had higher affinity binding for select polyribonucleotides or RNA from rat prostate over DNA fragments that encompassed the AR-binding promoter regions.71,72 The authors of these investigations concluded a role of AR-RNA interaction in gene regulation. Moreover, the role of AR participating in RNA binding or RNA metabolic processes such as splicing may very well be applicable to other members of the nuclear/steroid receptor family.26 The discovery of AR’s interaction with RNA splicing factors adds a new dimension in regulating AR-gene expression and illustrates a dynamic bilateral process involving components of AR transcription complex and components of the RNA splicing machinery.
The phenomena of AR-dependent cotranscriptional splicing has been described in PCa, where AR has been shown to interact with several proteins and splicing factors that influence alternative splicing.6,37 Evidence through the interaction of AR with cofactor of breast cancer type 1 (BRCA1), COBRA1, has been shown to influence splicing of an androgen sensitive promoter.73 AR and DDX5 were also shown to interact and be recruited to the promoter region of PSA enhancing the transcriptional expression of AR-dependent PSA expression.6 Similarly, SAM68 has been shown interact with AR and is also recruited to PSA promoter.37 Our mini-gene splicing observations are in line with these previous results that show that SAM68 exerts different effects of transcription and splicing events than DDX5, whereas DDX5 cooperated with AR to repress the splicing of variable exons of CD44 and promote exon-exclusion events and SAM68 promoted exon-inclusion events with AR.6,37 Furthermore, our use of the dominant negative SAM68 mutant (Sam68-G/D) supports the role of SAM68 as a direct regulator of AR-dependent gene activation. This can possibly be mediated via a direct interaction of SAM68 with the CREB-binding protein (CBP) transcription factor,74,75 which is a key cofactor of active AR transcriptional complexes.76,77 Recent publications of SAM68-null mice have highlighted the specific expression patterns for SAM68. These studies define SAM68-dependent transcriptionally active stages during spermatogenesis development via its interaction with RNAPII dictating specific gene expression patterns.78,79 Transcriptional-splicing complexes lead to bilateral effects, whereupon transcription factors lend to splicing events and in return splicing factors lead to transcriptional changes.
Prior investigations have examined global RNA splicing in PCa cell lines.80,81 These have demonstrated a diversity of genes undergoing alternative splicing: BCL2L1 (alternative 5’ splice site exon 2),82CD44 (multiple splicing events),41,42 fibroblast growth factor receptor 2 (FGFR2; mutually exclusive exons),83 KLF transcription factor 6 (KLF6; alternative 5’ splice site exon 2),84,85PSA (multiple splicing events),86 ETS transcription factor ERG (ERG; multiple splicing events),87 and vascular endothelial growth factor (VEGF; alternative 5’ splice site exon 8).88–90 Most of the affected proteins demonstrate functions related to neoplastic processes and others are potential indicators for PCa. The prognostic value of alternative splicing events and related patient prognosis have been reported in several cancer types.91–93 Most PCa studies have only focused on studying single-gene alternative splicing events.80,94–96 Most recently, several groups have assessed the prognostic value of global alternative splicing events in PCa. Using The Cancer Genome Atlas (TCGA) database,97 a group identified 44 070 alternative splicing events from 10 381 genes and built a 6-gene prognosis score by correlating the analysis for Percent Spliced In (PSI – a value between 0 to 1 that is commonly used to quantify alternative splicing occurrences) and mRNA expression. The 6 alternatively spliced genes included splicing factor 3b subunit 1 (SF3B1), Rho guanine nucleotide exchange factor 39 (ARHGEF39), cystatin C (CST3), LUC7 like 3 pre-mRNA splicing factor (LUC7L3), ubiquitin associated protein 2 (UBAP2), and SURP and G-patch domain containing 2 (SUGP2). Another investigation assessed expression profiling of 4600 PCa tumors and identified 37 gene sets with the ability to classify three distinct subtypes with one subtype having poor outcomes for patients.98 Again through expression profiling, another group has been able to define bone metastasis in PCa patients into molecular subtypes (MetA, B, or C), with MetA subtype having better prognostic outcomes than MetB.99
A recent publication assessed the induction of alternative splicing events in inducible LNCaP+AR-V7 expressing cells.100 Although the findings did not follow-up with clinical evidence, the investigators found a unique AR-V7 induced transcript splice event for post-GPI attachment to proteins 2 (PGAP2) with a novel exon that occurred in conjunction with a differential transcriptional start site selection. Furthermore, although the investigators found hundreds of differentially expressed genes, they found a low degree of overlap of shared splicing events (25%) between AR-FL and AR-V7. Other publications assessing gene expression profiles in other cell lines have also found a low overlap of gene expression profiles, using AR splice variant expressing cell lines.101,102 Our initial analysis of the in vitro expression data, identified nearly two thousand differentially spliced genes between the experimental conditions, refinement and filtering of these lists against clinical data sets, drastically reduced the numbers to less than one hundred. Therefore, it is unclear whether the presumed functional activity of the expression of the AR splice variant in cell lines is mimicking the clinical scenario and not an artifact of the high protein expression levels. In the CRPC state, most patients have undergone androgen directed therapies and would limit the activity of the full-length AR, however, many mechanisms have been described for full-length AR to maintain its function in a physiologically androgen deprived and recurrent state.103–108 Here we also have shown that AR45 is able to facilitate hormone-dependent exon-selection, but further investigations would be required to determine if AR45 and/or the AR splice variants also participate in clinically significant alternative RNA splicing programs.
Our in vitro results may provide a general discussion of AR facilitation of RNA splicing; however, the observations using clinical sample sets describe the complexity and pathological significance of this activity. The temporal progression of PCa is associated with the changes in gene expression, mediated via spatial transitions of protein complexes facilitating by the dynamic functions of cotranscriptional gene transcription and splicing processes. The inclusion of metastatic samples in our computational analysis of clinical array samples, assisted in segregating our exon-gene sets to define out prognostic indicators for AR and DDX5 groups. Therefore, the lack of SAM68’s association with significant clinical disease progression and outcomes can be partially explained by our TMAs results and the observations of a reduced nuclear expression of SAM68 in CRPC samples, suggesting that it may not have a prominent role in advanced disease stages and therefore unable to segregate those advance disease molecular splicing features. The analysis of the Taylor et al.34 dataset indicates that our exon gene sets segregate with metastatic samples and are predictors for the most aggressive disease manifestations. Therefore, we assessed our exon gene sets using clinical PCa datasets with systemic aggressive disease features,35,36 specifically PCSD. The AR and DDX5 exon gene-sets were able to distinguish a new subtype of lethal PCa. This would be the first characterization of a role of AR in facilitating a molecular process where the regulation of a set of expressed genes requires an RNA splicing functional property as part of lethal PCa subtype. Studies have demonstrated that although most advanced prostatic cancers are “androgen independent”, the AR may still be a very important contributor to the more progressive fatal disorder.34,109,110 This can include the AR splice variants, who are also uniquely enriched in metastatic and CRPC disease states.54
In this work, we presented further evidence for the role of AR participating in alternative splicing through a cooperative transcriptional splicing function through modulation by known RNA splicing factors SAM68 and DDX5. Additionally, we have demonstrated that both AR-specific and DDX5-modulated splicing programs can predict PCa disease-free outcomes with improved prognostic significance than traditional clinicopathological tools. These findings illustrate the role of pathological role of RNA splicing in PCa disease progression by highlighting a distinct subtype of lethal PCa linked to AR’s functionality in alternative RNA splicing.
SS participated in all microarray computational and data analysis. PNG participated in the in vitro experiments, including the transfection expression experiments and mini-gene splicing RT-PCRs. TAB participated in the tissue microarray profiling and analysis. MT participated in the design of the studies and assisted in the revision of the manuscript. MP participated in the design of the studies, confocal microscopy, performance of the transfection experiments for exon microarray analysis, and the drafting and writing of the manuscript. All authors read and approved the final manuscript.
All authors declare no competing interests.
Supplementary Table 1
Patients’ demographics of the tissue microarray study cohort (n=26; localized prostate cancer) (n=29 castration-resistant prostate cancer)
Supplementary Figure 1
(a) PI3K/AKT-dependent signaling pathways regulate AR-mediate RNA splicing and gene transactivation of SAM68 overexpressing LNCaP cells. (b) PI3K/AKT-dependent signaling pathways regulate AR-mediate RNA splicing and gene transactivation of DDX5 overexpressing LNCaP cells. * denotes statistically significant p-value <0.05, ** denotes statistical significance of p-value <0.005 and n.s. denotes not significant. AR: androgen receptor; Wort: Wortmannin; PSA: prostate specific antigen
Supplementary Figure 2
Venn diagram visualizing alternative splicing events in (a) LNCaP exon microarrays and (b) clinical samples.
Supplementary Table 2
AR-SAM68-DDX5-Exon-gene sets, used for clinical analysis
Supplementary Figure 3
AR-6HR PCa progression cluster analysis and disease-free outcomes. AR: androgen receptor; PCa: prostate cancer.
The Disclosure of Potential Conflicts of Interest forms are provided with the online version of the article (https://links.lww.com/AJOA/B14).
The Disclosure of Potential Conflicts of Interest forms are provided with the online version of the article (https://links.lww.com/AJOA/B15).
The Disclosure of Potential Conflicts of Interest forms are provided with the online version of the article (https://links.lww.com/AJOA/B16).
The Disclosure of Potential Conflicts of Interest forms are provided with the online version of the article (https://links.lww.com/AJOA/B17).
The Disclosure of Potential Conflicts of Interest forms are provided with the online version of the article (https://links.lww.com/AJOA/B18).
This work was supported by a research grant from Prostate Cancer Canada #673 to MP. We would like to acknowledge the assistance of Dr. Eleftherios Diamandis and Antonius Soosaipillai at the Samuel Lunenfeld Research Institute-Mount Sinai Hospital in Toronto (ON, Canada) for performing the PSA ELISAs and Gilles Corbiel (The Biochoip Services at the Research Centre-CHUM, University of Montreal, Montreal, QC, Canada) for his bioinformatic services.
Supplementary Information is linked to the online version of the paper on the Asian Journal of Andrology website.
1. Blencowe BJ. Alternative splicing:new insights from global analyses. Cell 2006;126:37–47.
2. Chen M, Manley JL. Mechanisms of alternative splicing regulation:insights from molecular and genomics approaches. Nat Rev Mol Cell Biol 2009;10:741–54.
3. Ghigna C, Valacca C, Biamonti G. Alternative splicing and tumor progression. Curr Genomics 2008;9:556–70.
4. Singh B, Eyras E. The role of alternative splicing in cancer. Transcription 2017;8:91–8.
5. Paliouras M, Zaman N, Lumbroso R, Kapogeorgakis L, Beitel LK, et al. Dynamic rewiring of the androgen receptor
protein interaction network correlates with prostate cancer
clinical outcomes. Integr Biol 2011;10:1020–32.
6. Clark EL, Coulson A, Dalgliesh C, Rajan P, Nicol SM, et al. The RNA helicase p68 is a novel androgen receptor
coactivator involved in splicing and is overexpressed in prostate cancer
. Cancer Res 2008;68:7938–46.
7. Goina E, Skoko N, Pagani F. Binding of DAZAP1 and hnRNPA1/A2 to an exonic splicing silencer in a natural BRCA1 exon 18 mutant. Mol Cell Biol 2008;28:3850–60.
8. Hall MP, Huang S, Black DL. Differentiation-induced colocalization of the KH-type splicing regulatory protein with polypyrimidine tract binding protein and the c-src pre-mRNA. Mol Biol Cell 2004;15:774–86.
9. Han SP, Tang YH, Smith R. Functional diversity of the hnRNPs:past, present and perspectives. Biochem J 2010;430:379–92.
10. Paronetto MP, Achsel T, Massiello A, Chalfant CE, Sette C. The RNA-binding protein Sam68 modulates the alternative splicing of Bcl-x. J Cell Biol 2007;176:929–39.
11. Beyer AL, Osheim YN. Splice site selection, rate of splicing, and alternative splicing on nascent transcripts. Genes Dev 1988;2:754–65.
12. Beyer AL, Bouton AH, Miller OL Jr. Correlation of hnRNP structure and nascent transcript cleavage. Cell 1981;26:155–65.
13. Rodriguez J, Tang CH, Khodor YL, Vodala S, Menet JS, et al. Nascent-Seq analysis of Drosophila
cycling gene expression. Proc Natl Acad Sci U S A 2013;110:E275–84.
14. Khodor YL, Rodriguez J, Abruzzi KC, Tang CH, Marr MT 2nd, et al. Nascent-seq indicates widespread cotranscriptional pre-mRNA splicing in Drosophila
. Genes Dev 2011;25:2502–12.
15. Ameur A, Zaghlool A, Halvardson J, Wetterbom A, Gyllensten U, et al. Total RNA sequencing reveals nascent transcription and widespread co-transcriptional splicing in the human brain. Nat Struct Mol Biol 2011;18:1435–40.
16. Carrillo Oesterreich F, Preibisch S, Neugebauer KM. Global analysis of nascent RNA reveals transcriptional pausing in terminal exons. Mol Cell 2010;40:571–81.
17. Tilgner H, Knowles DG, Johnson R, Davis CA, Chakrabortty S, et al. Deep sequencing of subcellular RNA fractions shows splicing to be predominantly co-transcriptional in the human genome but inefficient for lncRNAs. Genome Res 2012;22:1616–25.
18. Bieberstein NI, Straube K, Neugebauer KM. Chromatin immunoprecipitation approaches to determine co-transcriptional nature of splicing. Methods Mol Biol 2014;1126:315–23.
19. Kotovic KM, Lockshon D, Boric L, Neugebauer KM. Cotranscriptional recruitment of the U1 snRNP to intron-containing genes in yeast. Mol Cell Biol 2003;23:5768–79.
20. Tardiff DF, Rosbash M. Arrested yeast splicing complexes indicate stepwise snRNP recruitment during in vivo
spliceosome assembly. RNA 2006;12:968–79.
21. Fong N, Kim H, Zhou Y, Ji X, Qiu J, et al. Pre-mRNA splicing is facilitated by an optimal RNA polymerase II elongation rate. Genes Dev 2014;28:2663–76.
22. Lacadie SA, Tardiff DF, Kadener S, Rosbash M. In vivo
commitment to yeast cotranscriptional splicing is sensitive to transcription elongation mutants. Genes Dev 2006;20:2055–66.
23. Hodges C, Bintu L, Lubkowska L, Kashlev M, Bustamante C. Nucleosomal fluctuations govern the transcription dynamics of RNA polymerase II. Science 2009;325:626–8.
24. Huertas P, Aguilera A. Cotranscriptionally formed DNA:RNA hybrids mediate transcription elongation impairment and transcription-associated recombination. Molecular Cell 2003;12:711–21.
25. Auboeuf D, Dowhan DH, Kang YK, Larkin K, Lee JW, et al. Differential recruitment of nuclear receptor coactivators may determine alternative RNA splice site choice in target genes. Proc Natl Acad Sci U S A 2004;101:2270–4.
26. Auboeuf D, Honig A, Berget SM, O'Malley BW. Coordinate regulation of transcription and splicing by steroid receptor coregulators. Science 2002;298:416–9.
27. Dowhan DH, Hong EP, Auboeuf D, Dennis AP, Wilson MM, et al. Steroid hormone receptor coactivation and alternative RNA splicing
by U2AF65-related proteins CAPERalpha and CAPERbeta. Mol Cell 2005;17:429–39.
28. Auboeuf D, Dowhan DH, Dutertre M, Martin N, Berget SM, et al. A subset of nuclear receptor coregulators act as coupling proteins during synthesis and maturation of RNA transcripts. Mol Cell Biol 2005;25:5307–16.
29. Abramoff MD, Magalhaes PJ, Ram SJ. Image processing with ImageJ. Biopho Int 2004;11:36–42.
30. Chen T, Boisvert FM, Bazett-Jones DP, Richard S. A role for the GSG domain in localizing Sam68 to novel nuclear structures in cancer cell lines. Mol Biol Cell 1999;10:3015–33.
31. Darnel AD, Behmoaram E, Vollmer RT, Corcos J, Bijian K, et al. Fascin regulates prostate cancer
cell invasion and is associated with metastasis and biochemical failure in prostate cancer
. Clin Cancer Res 2009;15:1376–83.
32. Rubin MA, Bismar TA, Andren O, Mucci L, Kim R, et al. Decreased alpha-methylacyl CoA racemase expression in localized prostate cancer
is associated with an increased rate of biochemical recurrence and cancer-specific death. Cancer Epidemiol Biomarkers Prev 2005;14:1424–32.
33. Emig D, Salomonis N, Baumbach J, Lengauer T, Conklin BR, et al. AltAnalyze and DomainGraph:analyzing and visualizing exon expression data. Nucleic Acids Res 2010;38:W755–62.
34. Taylor BS, Schultz N, Hieronymus H, Gopalan A, Xiao Y, et al. Integrative genomic profiling of human prostate cancer
. Cancer Cell 2010;18:11–22.
35. Boormans JL, Korsten H, Ziel-van der Made AJ, van Leenders GJ, de Vos CV, et al. Identification of TDRD1
as a direct target gene of ERG in primary prostate cancer
. Int J Cancer 2013;133:335–45.
36. Erho N, Crisan A, Vergara IA, Mitra AP, Ghadessi M, et al. Discovery and validation of a prostate cancer
genomic classifier that predicts early metastasis following radical prostatectomy. PLoS One 2013;8:e66855.
37. Rajan P, Gaughan L, Dalgliesh C, El-Sherif A, Robson CN, et al. The RNA-binding and adaptor protein Sam68 modulates signal-dependent splicing and transcriptional activity of the androgen receptor
. J Pathol 2008;215:67–77.
38. Matter N, Herrlich P, Konig H. Signal-dependent regulation of splicing via phosphorylation of Sam68. Nature 2002;420:691–5.
39. De Marzo AM, Bradshaw C, Sauvageot J, Epstein JI, Miller GJ. CD44 and CD44v6 downregulation in clinical prostatic carcinoma:relation to Gleason grade and cytoarchitecture. Prostate 1998;34:162–8.
40. Ghatak S, Hascall VC, Markwald RR, Misra S. Stromal hyaluronan interaction with epithelial CD44 variants promotes prostate cancer
invasiveness by augmenting expression and function of hepatocyte growth factor and androgen receptor
. J Biol Chem 2010;285:19821–32.
41. Iczkowski KA, Omara-Opyene AL, Shah GV. The predominant CD44 splice variant in prostate cancer
binds fibronectin, and calcitonin stimulates its expression. Anticancer Res 2006;26:2863–72.
42. Noordzij MA, van Steenbrugge GJ, Verkaik NS, Schroder FH, van der Kwast TH. The prognostic value of CD44 isoforms in prostate cancer
patients treated by radical prostatectomy. Clin Cancer Res 1997;3:805–15.
43. Sanchez-Jimenez F, Sanchez-Margalet V. Role of Sam68 in post-transcriptional gene regulation. Int J Mol Sci 2013;14:23402–19.
44. Lukong KE, Larocque D, Tyner AL, Richard S. Tyrosine phosphorylation of sam68 by breast tumor kinase regulates intranuclear localization and cell cycle progression. J Biol Chem 2005;280:38639–47.
45. Chen T, Damaj BB, Herrera C, Lasko P, Richard S. Self-association of the single-KH-domain family members Sam68, GRP33, GLD-1, and Qk1:role of the KH domain. Mol Cell Biol 1997;17:5707–18.
46. Huot ME, Brown CM, Lamarche-Vane N, Richard S. An adaptor role for cytoplasmic Sam68 in modulating Src activity during cell polarization. Mol Cell Biol 2009;29:1933–43.
47. Castellano E, Downward J. Role of RAS in the regulation of PI 3-kinase. Curr Top Microbiol Immunol 2010;346:143–69.
48. Bielli P, Busa R, Paronetto MP, Sette C. The RNA-binding protein Sam68 is a multifunctional player in human cancer. Endocr Relat Cancer 2011;18:R91–102.
49. Paliouras M, Diamandis EP. Intracellular signaling pathways regulate hormone-dependent kallikrein gene expression. Tumour Biol 2008;29:63–75.
50. Paliouras M, Diamandis EP. An AKT activity threshold regulates androgen-dependent and androgen-independent PSA expression in prostate cancer
cell lines. Biol Chem 2008;389:773–80.
51. Zhu X, Daffada AA, Chan CM, Dowsett M. Identification of an exon 3 deletion splice variant androgen receptor
mRNA in human breast cancer. Int J Cancer 1997;72:574–80.
52. Guo Z, Yang X, Sun F, Jiang R, Linn DE, et al. A novel androgen receptor
splice variant is up-regulated during prostate cancer
progression and promotes androgen depletion-resistant growth. Cancer Res 2009;69:2305–13.
53. Ahrens-Fath I, Politz O, Geserick C, Haendler B. Androgen receptor
function is modulated by the tissue-specific AR45 variant. FEBS J 2005;272:74–84.
54. Robinson D, Van Allen EM, Wu YM, Schultz N, Lonigro RJ, et al. Integrative clinical genomics of advanced prostate cancer
. Cell 2015;161:1215–28.
55. Culig Z, Hobisch A, Bartsch G, Klocker H. Androgen receptor
–An update of mechanisms of action in prostate cancer
. Urol Res 2000;28:211–9.
56. Hobisch A, Culig Z, Radmayr C, Bartsch G, Klocker H, et al. Distant metastases from prostatic carcinoma express androgen receptor
protein. Cancer Res 1995;55:3068–72.
57. Mercatante DR, Bortner CD, Cidlowski JA, Kole R. Modification of alternative splicing of Bcl-x pre-mRNA in prostate and breast cancer cells. analysis of apoptosis and cell death. J Biol Chem 2001;276:16411–7.
58. Moniaux N, Escande F, Batra SK, Porchet N, Laine A, et al. Alternative splicing generates a family of putative secreted and membrane-associated MUC4 mucins. Eur J Biochem 2000;267:4536–44.
59. Long Q, Xu J, Osunkoya AO, Sannigrahi S, Johnson BA, et al. Global transcriptome analysis of formalin-fixed prostate cancer
specimens identifies biomarkers of disease recurrence. Cancer Res 2014;74:3228–37.
60. Prochownik EV, Eagle Grove L, Deubler D, Zhu XL, Stephenson RA, et al. Commonly occurring loss and mutation of the MXI1
gene in prostate cancer
. Genes Chromosomes Cancer 1998;22:295–304.
61. Torres-Martin M, Lassaletta L, San-Roman-Montero J, De Campos JM, Isla A, et al. Microarray analysis of gene expression in vestibular schwannomas reveals SPP1/MET signaling pathway and androgen receptor
deregulation. Int J Oncol 2013;42:848–62.
62. Ashida S, Furihata M, Katagiri T, Tamura K, Anazawa Y, et al. Expression of novel molecules, MICAL2-PV (MICAL2 prostate cancer
variants), increases with high Gleason score and prostate cancer
progression. Clin Cancer Res 2006;12:2767–73.
63. Wang H, Wu R, Yu L, Wu F, Li S, et al. SGEF is overexpressed in prostate cancer
and contributes to prostate cancer
progression. Oncol Rep 2012;28:1468–74.
64. Qi H, Fournier A, Grenier J, Fillion C, Labrie Y, et al. Isolation of the novel human guanine nucleotide exchange factor Src homology 3 domain-containing guanine nucleotide exchange factor (SGEF) and of C-terminal SGEF, an N-terminally truncated form of SGEF, the expression of which is regulated by androgen in prostate cancer
cells. Endocrinology 2003;144:1742–52.
65. Wang XY, Hao JW, Zhou RJ, Zhang XS, Yan TZ, et al. Meta-analysis of gene expression data identifies causal genes for prostate cancer
. Asian Pac J Cancer Prev 2013;14:457–61.
66. Fujita A, Gomes LR, Sato JR, Yamaguchi R, Thomaz CE, et al. Multivariate gene expression analysis reveals functional connectivity changes between normal/tumoral prostates. BMC Syst Biol 2008;2:106.
67. Miller J, McLachlan AD, Klug A. Repetitive zinc-binding domains in the protein transcription factor IIIA from Xenopus oocytes. EMBO J 1985;4:1609–14.
68. Liao S, Liang T, Tymoczko JL. Ribonucleoprotein binding of steroid-“receptor”complexes. Nat New Biol 1973;241:211–3.
69. Lin SY, Ohno S. The binding of androgen receptor
to DNA and RNA. Biochim Biophy Acta 1981;654:181–6.
70. Lin S, Ohno S. The interactions of androgen receptor
with poly(A)-containing RNA and polyribonucleotides. Eur J Biochem 1982;124:283–7.
71. Mulder E, Vrij AA, Brinkmann AO. DNA and ribonucleotide binding characteristics of two forms of the androgen receptor
from rat prostates. Biochem Biophys Res Commun 1983;114:1147–53.
72. Mulder E, Vrij AA, Brinkmann AO, Van der Molen HJ, Parker MG. Interaction of rat prostate androgen receptors with polynucleotides, RNA, DNA and cloned DNA fragments. Biochim Biophy Acta 1984;781:121–9.
73. Sun J, Blair AL, Aiyar SE, Li R. Cofactor of BRCA1 modulates androgen-dependent transcription and alternative splicing. J Steroid Biochem Mol Biol 2007;107:131–9.
74. Babic I, Jakymiw A, Fujita DJ. The RNA binding protein Sam68 is acetylated in tumor cell lines, and its acetylation correlates with enhanced RNA binding activity. Oncogene 2004;23:3781–9.
75. Hong W, Resnick RJ, Rakowski C, Shalloway D, Taylor SJ, et al. Physical and functional interaction between the transcriptional cofactor CBP and the KH domain protein Sam68. Mol Cancer Res 2002;1:48–55.
76. Fronsdal K, Engedal N, Slagsvold T, Saatcioglu F. CREB binding protein is a coactivator for the androgen receptor
and mediates cross-talk with AP-1. J Biol Chem 1998;273:31853–9.
77. Aarnisalo P, Palvimo JJ, Janne OA. CREB-binding protein in androgen receptor
-mediated signaling. Proc Natl Acad Sci U S A 1998;95:2122–7.
78. Paronetto MP, Messina V, Barchi M, Geremia R, Richard S, et al. Sam68 marks the transcriptionally active stages of spermatogenesis and modulates alternative splicing in male germ cells. Nucleic Acids Res 2011;39:4961–74.
79. Paronetto MP, Messina V, Bianchi E, Barchi M, Vogel G, et al. Sam68 regulates translation of target mRNAs in male germ cells, necessary for mouse spermatogenesis. J Cell Biol 2009;185:235–49.
80. Shah K, Gagliano T, Garland L, O'Hanlon T, Bortolotti D, et al. Androgen receptor
signaling regulates the transcriptome of prostate cancer
cells by modulating global alternative splicing. Oncogene 2020;39:6172–89.
81. Germain L, Lafront C, Beaudette J, Karthik Poluri RT, Weidmann C, et al. Alternative splicing regulation by the androgen receptor
in prostate cancer
cells. J Steroid Biochem Mol Biol 2020;202:105710.
82. Mercatante DR, Mohler JL, Kole R. Cellular response to an antisense-mediated shift of Bcl-x pre-mRNA splicing and antineoplastic agents. J Biol Chem 2002;277:49374–82.
83. Gnanapragasam VJ, Robinson MC, Marsh C, Robson CN, Hamdy FC, et al. FGF8 isoform b expression in human prostate cancer
. Br J Cancer 2003;88:1432–8.
84. Hartel M, Narla G, Wente MN, Giese NA, Martignoni ME, et al. Increased alternative splicing of the KLF6 tumour suppressor gene correlates with prognosis and tumour grade in patients with pancreatic cancer. Eur J Cancer 2008;44:1895–903.
85. Narla G, Difeo A, Reeves HL, Schaid DJ, Hirshfeld J, et al. A germline DNA polymorphism enhances alternative splicing of the KLF6 tumor suppressor gene and is associated with increased prostate cancer
risk. Cancer Res 2005;65:1213–22.
86. Heuze-Vourc'h N, Leblond V, Courty Y. Complex alternative splicing of the hKLK3 gene coding for the tumor marker PSA (prostate-specific-antigen). Eu J Biochem 2003;270:706–14.
87. Wang J, Cai Y, Yu W, Ren C, Spencer DM, et al. Pleiotropic biological activities of alternatively spliced TMPRSS2/ERG fusion gene transcripts. Cancer Res 2008;68:8516–24.
88. Manetti M, Guiducci S, Romano E, Ceccarelli C, Bellando-Randone S, et al. Overexpression of VEGF165b, an inhibitory splice variant of vascular endothelial growth factor, leads to insufficient angiogenesis in patients with systemic sclerosis. Circ Res 2011;109:e14–26.
89. Woolard J, Wang WY, Bevan HS, Qiu Y, Morbidelli L, et al. VEGF165b, an inhibitory vascular endothelial growth factor splice variant:mechanism of action, in vivo
effect on angiogenesis and endogenous protein expression. Cancer Res 2004;64:7822–35.
90. Bates DO, Cui TG, Doughty JM, Winkler M, Sugiono M, et al. VEGF165b, an inhibitory splice variant of vascular endothelial growth factor, is down-regulated in renal cell carcinoma. Cancer Res 2002;62:4123–31.
91. Li Y, Sun N, Lu Z, Sun S, Huang J, et al. Prognostic alternative mRNA splicing signature in non-small cell lung cancer. Cancer Lett 2017;393:40–51.
92. Liu J, Li H, Shen S, Sun L, Yuan Y, et al. Alternative splicing events implicated in carcinogenesis and prognosis of colorectal cancer. J Cancer 2018;9:1754–64.
93. He RQ, Zhou XG, Yi QY, Deng CW, Gao JM, et al. Prognostic signature of alternative splicing events in bladder urothelial carcinoma based on spliceseq data from 317 cases. Cell Physiol Biochem 2018;48:1355–68.
94. Narla G, DiFeo A, Yao S, Banno A, Hod E, et al. Targeted inhibition of the KLF6 splice variant, KLF6 SV1, suppresses prostate cancer
cell growth and spread. Cancer Res 2005;65:5761–8.
95. Hu Y, Dobi A, Sreenath T, Cook C, Tadase AY, et al. Delineation of TMPRSS2-ERG
splice variants in prostate cancer
. Clin Cancer Res 2008;14:4719–25.
96. Hu R, Dunn TA, Wei S, Isharwal S, Veltri RW, et al. Ligand-independent androgen receptor
variants derived from splicing of cryptic exons signify hormone-refractory prostate cancer
. Cancer Res 2009;69:16–22.
97. Cao ZX, Xiao GA, Zhang W, Ji J, Ye C, et al. Comprehensive investigation of alternative splicing and development of a prognostic risk score for prostate cancer
based on six-gene signatures. J Cancer 2019;10:5585–96.
98. You S, Knudsen BS, Erho N, Alshalalfa M, Takhar M, et al. Integrated classification of prostate cancer
reveals a novel luminal subtype with poor outcome. Cancer Res 2016;76:4948–58.
99. Thysell E, Vidman L, Ylitalo EB, Jernberg E, Crnalic S, et al. Gene expression profiles define molecular subtypes of prostate cancer
bone metastases with different outcomes and morphology traceable back to the primary tumor. Mol Oncol 2019;13:1763–77.
100. Rana M, Dong J, Robertson MJ, Basil P, Coarfa C, et al. Androgen receptor
and its splice variant, AR-V7, differentially induce mRNA splicing in prostate cancer
cells. Sci Rep 2021;11:1393.
101. Chen Z, Wu D, Thomas-Ahner JM, Lu C, Zhao P, et al. Diverse AR-V7 cistromes in castration-resistant prostate cancer
are governed by HoxB13. Proc Natl Acad Sci U S A 2018;115:6810–5.
102. Hu R, Lu C, Mostaghel EA, Yegnasubramanian S, Gurel M, et al. Distinct transcriptional programs mediated by the ligand-dependent full-length androgen receptor
and its splice variants in castration-resistant prostate cancer
. Cancer Res 2012;72:3457–62.
103. King CJ, Woodward J, Schwartzman J, Coleman DJ, Lisac R, et al. Integrative molecular network analysis identifies emergent enzalutamide resistance mechanisms in prostate cancer
. Oncotarget 2017;8:111084–95.
104. Chandrasekar T, Yang JC, Gao AC, Evans CP. Mechanisms of resistance in castration-resistant prostate cancer
(CRPC). Transl Androl Urol 2015;4:365–80.
105. Einstein DJ, Arai S, Balk SP. Targeting the androgen receptor
and overcoming resistance in prostate cancer
. Curr Opin Oncol 2019;31:175–82.
106. Mollica V, Di Nunno V, Cimadamore A, Lopez-Beltran A, Cheng L, et al. Molecular mechanisms related to hormone inhibition resistance in prostate cancer
. Cells 2019;8:43.
107. Banerjee PP, Banerjee S, Brown TR, Zirkin BR. Androgen action in prostate function and disease. Am J Clin Exp Urol 2018;6:62–77.
108. Huang Y, Jiang X, Liang X, Jiang G. Molecular and cellular mechanisms of castration resistant prostate cancer
. Oncol Lett 2018;15:6063–76.
109. Isaacs JT, Isaacs WB. Androgen receptor
outwits prostate cancer
drugs. Nat Med 2004;10:26–7.
110. Taplin ME, Balk SP. Androgen receptor
:a key molecule in the progression of prostate cancer
to hormone independence. J Cell Biochem 2004;91:483–90.