Expression of steroid and peptide hormone receptors, metabolic enzymes and EMT-related genes in prostate tumors in relation to the presence of the TMPRSS2/ERG fusion

Alterations in expression of the androgen receptor (AR) are often associated with development of prostate cancer. It is known already that the AR gene expression is regulated by quite many molecular pathways [1]. Another example of frequent alterations in prostate tumors is formation of gene fusions of androgen dependent gene TMPRSS2 (transmembrane protease, serine 2) with the ETS (E26 transformation-specific) family in particular with ERG (ETS related gene) [2]. Previously, we have shown that the TMPRSS2/ERG fusion is present in prostate adenocarcinoma and even in conventionally normal prostate tissue (CNT) in a group of patients of the Ukrainian population [3]. Therefore, we may speculate that the presence or absence of the gene fusions could be the cause of development of various prostate cancer types with different sensitivity to therapy, recurrence and metastasizing, despite the similar histological characteristics [4].

One of the important characteristics of normal functioning of prostate epithelial cells is sensitivity to steroid and peptide hormones. In the process of cell transformation, tumor cells often lose the sensitivity to hormones and growth factors and also change their metabolism. The AR is a key element of prostate functioning and is involved in malignant transformation. As was shown already, AR signaling plays a primary role in development of androgen resistant and castration-resistant prostate cancer [1]. There are few isoforms of ARs. Some of them are prostate specific. AR expression can change during prostate carcinogenesis. Thus, the overexpression of AR isoform A (1 isof) decreases proliferation but accelerates invasion of prostate cancer cells, compared with overexpression of AR isoform B (2 isof) [5]. Also, it was proposed that formation of the fusion between TMPRSS2 and ERG might be controlled by androgens [6].

In prostate cells, the most potent AR agonist is dihydrotestosterone. This is a metabolite of testosterone, and the reaction of conversion is catalyzed by SRD5A1 (5α-reductase, type 1) and SRD5A2 (5α-reductase, type 2). The latter are expressed at low levels in normal prostate tissues, but upon prostate cancer progression expression of these enzymes is altered [7]. Noteworthy, prostate cancer is a complex pathology and many other hormone receptors and corresponding pathways are involved in tumor development, especially GCR (glucocorticoid receptor, NR3C1 nuclear receptor subfamily 3 group C member 1), IGF1R (insulin like growth factor 1 receptor), ESR1 and ESR2 (estrogen receptors 1 and 2), PRLR (prolactin receptor), VDR (vitamin D receptor) and others.

Of note, GCR and AR share several transcriptional targets [8]. All of the three isoforms of GCR (alpha (A), beta (B) and gamma (G)) are very important in development and progression of prostate cancer [9]. In initiation and also in progression of the prostate cancer the IGF network, including INSR (insulin receptor) — (subtypes INSR A and B), IGF1R and IGF2R plays an important role [10–12].

Both estrogen receptors, alpha (ESRα, ESR1) and beta (ESRβ, ESR2) are associated with deve­lopment of prostate cancer [13]. It was shown, that the increased expression of ESRα is observed upon progression, metastasizing, and in androgen resistant phenotype; ESRα could be involved in regulation of expression of the TMPRSS2-ERG fusion [14].

PRL (prolactin) can induce growth and survival of prostate cancer cells [15]. The PRL expression correlates with the disease severity.

It was shown that vitamin D (calcitriol) influences on prostate cancer cells growth [16]. Furthermore, the TMPRSS2-ERG fusion expression is increased upon activation of VDR and AR. Consequently, expression of TMPRSS2/ERG leads to inactivation of the VDR signaling [17].

We have shown earlier that several genes, regu­lating the epithelial-to-mesenchymal cell transition (EMT), such as CDH1, CDH2, NKX3-1, FN1 and VIM, are expressed differently in prostate tumors [18].

In a present work, we aimed to analyze the expression pattern of a group of the cancer-related genes, depending on the presence or absence of the TMPRSS2/ERG fusion in prostate tumors. Also, we wanted to find the putative correlations between gene expression patterns and clinical and pathological characteristics (CPC) to define the molecular subtypes of prostate cancer.

MATERIALS AND METHODS

A collection of prostate tissues samples. Samples of cancer tissue and CNT (at an opposite side of tumor) were frozen in liquid nitrogen immediately after surgical resection at the National Cancer Institute (Kyiv, Ukraine). Benign prostate tumors (prostate adenoma samples) were collected at the Institute of Urology (Kyiv, Ukraine) after radical prostatectomy and frozen, as described above. The samples were collected in accordance with the Declaration of Helsinki and the guidelines issued by the Ethic Committee of the Institute of Urology, the National Cancer Institute and an Ethic Committee of the Institute of Molecular Biology and Genetics. Experimental studies were conducted on 37 prostate adenocarcinoma samples of different Gleason score and stages; 37 paired CNT samples; 21 samples of benign prostate tumors (ade­nomas). Tumor samples were characterized, according to an International System of Classification of Tumors, based on the tumor-node-metastasis (TNM) and the World Health Organization (WHO) criteria. CPC and the presence/absence of the TMPRSS2/ERG fusion, that we have detected earlier [3] are presented on Table 1.

Total RNA isolation and cDNA synthesis. 50–70 mg of frozen prostate tissues were mashed to a powder in the liquid nitrogen. Total RNA was extracted by TRI-­reagent (SIGMA), according to the manufacturer’s protocol. Total RNA concentration was analyzed by a spectrophotometer (NanoDrop Technologies Inc., USA). The quality of the total RNA was determined in a 1% agarose gel by band intensity of 28S and 18S rRNA (28S/18S ratio). cDNA was synthesized from 1 µg of the total RNA, that was treated with the RNase free DNase I (Thermo Fisher Scientific, USA), using RevertAid H-Minus M-MuLV Reverse Transcriptase (Thermo Fisher Scientific, USA), according to the manufacturer’s protocol.

Quantitative quantitative polymerase chain reaction (qPCR). Relative gene expression (RE) levels of 27 genes (33 transcripts) were detected by qPCR, using Maxima SYBR Green Master mix (Thermo Fisher Scientific, USA) and Bio-Rad CFX96 Real-Time PCR Detection System (USA) under the following conditions: 95 °C — 10 min, following 40 cycles of 95 °C — 15 s, 60 °C — 30 s, elongation 72 °C — 30 s. Primers for the different transcripts of INSR and IGF1R and various isoforms of GCR were as published earlier [9, 19]. Primers for others genes were selected, using qPrimerDepot (primerdepot.nci.nih.gov).

Four reference genes — TBP, HPRT, ALAS1 and TUBA1B — were used for normalization of the gene expression [20]. The two main models (2^-ΔC_T and 2^-ΔΔC_T methods), described earlier [18, 21], were used for the RE level calculation and analysis.

Statistical analysis. The Kolmogorov — Smirnov test was used to analyze the normality of distribution. The Kruskal — Wallis test was used to determine differences by multiple comparison between experimental groups. The Wilcoxon Matched Pairs test was performed to compare RE in prostate adenocarcinoma and paired CNT. RE fold differences in 2^-ΔΔC_T model were considered significant when expression changes were more, than 2 fold. The Fisher exact test was performed to monitor differences between these sample groups [21]. The Benjamini — Hochberg procedure with false discovery rate (FDR) 0.10–0.25 was used when multiple comparisons were performed [22]. The Dunn — Bonferroni post hoc test was performed to determine RE differences between pairs of prostate samples. The Spearman’s rank correlation test was used to find the putative correlations between RE and CPC of prostate tumors and also correlations between RE of investigated genes. The K-Mean clustering was applied for prostate cancer subtyping and also for the specific gene expression profiles, following by the Kruskal — Wallis and Dunn — Bonferroni post hoc tests for detection of inter-cluster differences in RE.

RESULTS

Expression of 17 transcripts (11 genes), representing the receptors and metabolic enzymes and also 16 EMT-related transcripts/genes (3 from them are lncRNAs) were studied in prostate adenocarcinomas, CNT and adenomas.

Earlier, we have shown that the TMPRSS2/ERG fusion was expressed in 21 out of 37 adenocarcinomas [3]. In this group, in 16 paired CNT the TMPRSS2/ERG fusion was detected, and 5 CNT did not show this fusion. Thus, we have 3 groups in a set of the paired adenocarcinomas/CNT: 1) T–/N– group — the TMPRSS2/ERG fusion was not detected neither in adenocarcinomas nor in CNT (n = 16); 2) T+/N+ group — the TMPRSS2/ERG fusion was found in both, cancer and CNT (n = 16); 3) T+/N– group — the TMPRSS2/ERG fusion was present in adenocarcinomas, but not in CNT (n = 5).

The Wilcoxon Matched paired test in the 2-^ΔC_T model showed the differences in RE of 7 genes, when the paired adenocarcinoma (T) and CNT (N) were compared, regardless presence or absence of the TMPRSS2/ERG fusion (Table 2).

The following five genes were upregulated in adenocarcinomas, when T/N pairs with the fusion in both, tumor and CNT were analyzed: ESR1 (p = 0.038), KRT18 (p = 0.007), MKI67 (p = 0.003), MMP9 (p = 0.011) and PCA3 (p = 0.049). In adenocarcinomas without fusion INSR (B isof) (p = 0.039) and HOTAIR (p = 0.027) were expressed at the higher levels, than in the paired CNT. Only one gene, ESR1, showed significant changes in RE in adenocarcinomas with the presence of the fusion, compared with CNT without fusion (p = 0.043).

When the 2^-ΔΔC_T model was used, we found 6 genes with significant differences in RE between adenocarcinomas and CNT (Table 3). Three genes (MMP9, MKI67, and SCHLAP1) where expressed at the higher levels in tumors, compared with CNT (the T+/N+ group) (p < 0.05), two genes (ESR1 and HOTAIR) have shown increased RE in T+/N+ and T–/N– groups (p < 0.05). Only one gene, the PCA3 was significantly increased in T+/N+ and T+/N– groups (p < 0.05).

Hence, the data obtained by the two abovementioned models are only partially overlapping. This could be due to different statistical calculations.

Earlier, we have discussed that CNT isolated from patients with prostate tumors do not represent the normal tissue, therefore they can’t be considered as an adequate control [18]. In order to avoid working with inadequate controls, adenomas were used as the control instead. Noteworthy, the TMPRSS2/ERG fusion was detected in 4 adenomas as well. No differences in the gene expression patterns were found in these 4 adenomas, compared with adenomas without fusion. For further comparison, only the group of adenomas without fusion was analyzed (n = 17). Also, CNT samples without fusion (n = 5) from adenocarcinoma pairs with the fusion were attributed to total CNT fusion negative (N–) group after verification of RE differences in CNT sample groups for all investigated genes.

The Kruskal — Wallis test (with FDR = 0.1) has shown significant differences of RE in 11 out of 33 trans­cripts/genes between 5 investigated groups (T+, T–, N+, N– and A–), while the Dunn — Bonferroni post hoc method of the multiple comparisons has confirmed changes only for 9 transcripts/genes (Table 4, Fig. 1).

Increased RE levels in the adenocarcinoma and CNT groups, compared to the adenoma group was shown for 6 genes 1) ECR1 T–/A– group (p = 0.002), T+/A– (p < 0.001), N–/A– (p = 0.040); 2) KRT18 T+/A– (p = 0.008); 3) MMP9 T–/A– (p = 0.003), N–/A– (p = 0.012); 4) PCA3 T+/A– (p = 0.001), N+/A– (p = 0.001); 5) HOTAIR T–/A– (p = 0.002), and 6) SCHLAP1 T+/A– (p = 0.011). Decreased RE levels in the adenocarcinoma and CNT groups, compared to the adenoma group, was detected for 3 genes: 1) AR (2 isof) T–/A– (p = 0.0172), 2) PRLR T–/A– (p = 0.0088) and 3) SRD5A2 T–/A– (p = 0.0393), T+/A– (p = 0.0034), N+/A– (p = 0.0203).

Correlations of CPC with RE levels. The Spearman Rank Order Correlations (rs) analysis of CPC characteristics and RE of a set of genes in prostate adenocarcinomas has revealed a number of positive and negative correlations (Table 5A). For example, there is the reverse correlation between the Gleason score and RE of ESR2, VDR and SRD5A2: rs = –0.354, rs = –0.382 (p < 0.05), rs = –0.520 (p < 0.01), respectively. Also, RE of GCR (in AG) and PRL showed the direct correlations with a tumor stage, and 8 genes — AR (1 isof), AR (2 isof), INSR (A isof), IGF1R, IGF1R tr, PRLR, VDR and SRD5A2 showed the negative correlation with the tumor stage. Levels of the PSA in serum correlate negatively with RE of VDR and SRD5A2.

Correlations of RE levels between investigated genes. Investigation of RE correlations in prostate adenocarcinomas have shown 131 significant correlations (from p < 0.0001 to p < 0.05) (Table 5B). Among them 34 have the highest score rs = |0,524–0,936| (from p < 0.0001 to p < 0.05). A maximal number of strong RE correlations showed INSR (A isof) — 7 correlations, AR (1 isof), GCR (in AG), IGF1R, IGF1R tr — 6 correlations. This big number of correlations confirms robust relationships between gene expression profiles and the close connections pathways, where these genes belong to.

Expression profiling of adenocarcinomas. To determine the putative molecular subtypes of the prostate adenocarcinomas, showing the certain gene expression profile, the K-means clustering was performed, with analysis of RE of all of the studied genes and CPC (Gleason score and tumor stages) in the adenocarcinoma group. We found three specific clusters (Fig. 2, Table 6), that included 33 out of 37 cancer samples (89%). These clusters showed the significant differences in RE of 21 out of 33 transcripts. The largest distance is between clusters 1 and 3. All three clusters consist of tumors with the various Gleason scores (6, 7, 9).

The cluster 1 contains 12 samples with the TMPRSS2/ERG fusion. Also, in this group the highest expression of AR, epithelial markers (CDH1, NKX3-1, OCLN) and prostate cancer markers (PSA, PCA3, KRT18, SCHLAP1) is detected.

The cluster 3 contains the tumors with the highest Gleason score and a tumor stage index. By other words, the cluster 3 consists of the most aggressive tumors. This assumption is supported by the RE data. For example, in this group we found the lowest expression of AR, epithelial markers (CDH1, OCLN, NKX3-1), SRD5A2, INSR (A and B) and IGF1R, and the high levels of PRL, lncRNA SCHLAP1 and HOTAIR, and also of mesenchymal markers (VIM, FN1, MMP2). We have to mention, however, that the cluster 3 contains the lowest number of samples with the fusion — only 2 out of 8.

The gene expression profile in cluster 2 has a mixed pattern. For example, several epithelial and luminal markers, such as KRT18, PCA3 and PSA show the lowest expression, and other genes, namely mesenchymal markers CDH2, MMP2, FN1 and VIM are highly expressed.

DISCUSSION

The TMPRSS2-ERG fusion transcript isoform 2 (EF194202.1) was first detected in prostate tumor samples by Lapointe et al. [23]. It is known that formation of this fusion transcript leads to overexpression of the ERG protein, which is involved in the signaling pathways associated with prostate cancer development [24, 25]. We wanted to enlighten the influence of this fusion on expression of some prostate cancer-associated receptors, enzymes and EMT-associated genes. Thus, in paired adenocarcinoma/CNT samples we have found the specific changes in RE in cancers with the fusion for 5 genes, whereas RE alterations for tumors without fusion were found only for 2 genes. The high level of ESR1 in tumors where the fusion was detected was associated with faster cancer progression [14].

In the present work we found among adenocarcinomas, CNT and adenomas that the ESR1 and SRD5A2 genes showed altered expression regardless presence of the TMPRSS2-ERG fusion, while AR, MMP9 and HOTAIR were affected only in cases with no fusion, and expression of KRT18, PCA3 and SCHLAP1 were changed in adenocarcinomas with the fusion. Noteworthy, in adenomas we have detected the highest SRD5A2 RE levels. It is known that increased SRD5A2 in adenomas provokes hyperplasia extension through NF-kB and AR isoform 7 conferring 5α-reductase inhibitors resistance [26]. From other hand decreased levels of SRD5A2 in adenocarcinomas is associated with the enhanced cell migration and invasion [7]. Moreover, when SRD5A2 gene was re-introduced, cell migration and invasion was inhibited, due to F-actin reorganization [27].

The high RE of lncRNA SCHLAP1 adenocarcinomas with the fusion predict unfavorable prognosis of disease [28]. The other lncRNA, HOTAIR when it expressed at the high levels in adenocarcinomas without fusion enhances proliferation and invasion at late stages of prostate cancer [29].

Earlier, we could not find a correlation between frequency of the fusion transcript detection and CPC, such as the Gleason score and stage [3], therefore we didn’t analyze TMPRSS2/ERG dependent changes for investigated genes in sample groups with different CPC. Now we found many correlations between CPC and RE of the genes, encoding receptors/enzymes in total group of adenocarcinomas. Eight out of ten significant (p < 0.01 to p < 0.05) correlations were negative, i.e. expression of these genes was decreasing upon cancer progression. Furthermore, large quantity of the RE correlations of investigated genes allow us to perform the clustering of patients with adenocarcinomas. We clustered prostate adenocarcinomas in three groups, based on the RE of 33 transcripts and also CPC characteristics.

Our experimental data on the RE profiles in prostate adenocarcinomas are in concordance with the literature data [1, 2, 4]. It is widely accepted, that high expression of epithelial and luminal markers is usually accompanied by low expression of mesenchymal markers, and that we have observed in cluster 1. Noteworthy, we showed simultaneous high expression of the fusion transcript, PCA3 and NKX3-1 in one cluster. It seems, that the fusion transcript and PCA3 do not influence negatively on expression of the tumor suppression gene NKX3-1 and vice versa, as they belong to different pathways [1, 4]. At the other hand, the oncogenic PCA3 pathway [28, 30], probably, acts in parallel with the ERG pathway [23, 24].

To summarize subtyping data, it is essential to note specific cluster features. Cluster 1, which contains all fusion positive adenocarcinomas, has the most characteristic expression profile namely fusion positive androgen dependent luminal subtype 1. Probably oncogenic pathways in this group are ERG and PCA3 [25, 30] with high sensitivity to androgens, prolactin, IGF, INS stimulation oncogenic signaling.

We suppose that cluster 3 is another luminal prostate cancer subtype most of all it is fusion negative with androgen independent and castration resistant characteristics [30] (fusion negative androgen independent luminal subtype 2). It has molecular characteristic properties as the lowest expression of AR, epithelial markers (CDH1, OCLN, NKX3-1), SRD5A2, INSR (A and B) and IGF1R, high levels of mesenchymal markers (VIM, FN1, MMP2) and lncRNAs SCHLAP1 and HOTAIR. Moreover, increasing RE of HOTAIR may cause the resistance for enzalutamide [29]. It is unique cluster with the highest PRL level, which could promote cancer progression through the PRL/STAT5 signaling pathway [15]. This is could mean prolactin administration of this cluster carcinogenesis.

We assume that cluster 2 is mixed stem-like androgen dependent subtype. The lowest expression of some epithelial and luminal markers KRT18, PCA3, PSA and high expression for mesenchymal markers CDH2, MMP2, and tendency to RE growth of FN1, and VIM are characteristics of stem-like (basal) prostate cancer, in spite of high AR, CDH1, NKX3-1 RE. The highest RE levels of ESR1, SRD5A2, INSR B, PRLR and lncRNA HOTAIR give to this cluster peculiar carcinogenic property.

CONCLUSIONS

We have analyzed RE of 33 transcripts from 27 genes to find alterations in prostate tumors, depending on the presence or absence of the TMPRSS2/ERG fusion. The significant differences of RE (p < 0.05) for 7 genes were detected, when compared adenocarcinomas and corresponding CNTs, using the 2-^ΔC_T model. Five genes (ESR1, KRT18, MKI67, MMP9, PCA3) showed differential expression, when the paired samples with compared that were bearing the fusion; and only two genes (INSR (B isof) and HOTAIR) — when samples did not expressed the fusion product. When the 2^-ΔΔC_T model was used, the number of the differentially expressed genes were six (MMP9, MKI67, PCA3, SCHLAP1) and two (ESR1, HOTAIR) when the tissues expressed the fusion or regardless the presence of the fusion, respectively.

When adenomas, CNT and adenocarcinomas were compared, the KRT18, PCA3 and SCHLAP1 genes showed significant differences in RE in adenocarcinomas with the fusion. In adenocarcinomas without the fusion, such properties were shown by the AR (2 isof), MMP9, PRLR the HOTAIR genes. The ESR1 and SRD5A2 gene expression was altered in both types of adenocarcinomas.

Using the statistical analysis, we created three clusters of adenocarcinomas, based on gene RE and CPC characteristic. One of clusters was represented by adenocarcinomas with the TMPRSS2/ERG fusion. Further experiments are needed to confirm these data in a larger patient cohort.

REFERENCES

1. Zhou Y, Bolton E, Jones J. Androgens and androgen receptor signaling in prostate tumorigenesis. J Mol Endocrinol 2015; 54: 15–29.
2. Cancer Genome Atlas Research Network. The Molecular Taxonomy of Primary Prostate Cancer. Cell 2015; 163: 1011–25.
3. Mevs L, Gerashchenko G, Rosenberg E, et al. Detection of prostate specific ETS fusion transcripts in cancer samples. Biopolym Cell 2017; 33: 256–67.
4. Shtivelman E, Beer T, Evans C. Molecular pathways and targets in prostate cancer. Oncotarget 2014; 5: 7217–59.
5. Zeng R, Liu Z, Sun Y, et al. Differential expression and function of AR isoforms in prostate cancer. Oncol Rep 2012; 27: 492–8.
6. Lin C, Yang L, Tanasa B, et al. Nuclear receptor-induced chromosomal proximity and DNA breaks underlie specific translocations in cancer. Cell 2009; 139: 1069–83.
7. Audet-Walsh Е, Yee T, Tam I, et al. Inverse regulation of DHT synthesis enzymes 5α-reductase types 1 and 2 by the androgen receptor in prostate cancer. Endocrinol 2017; 158: 1015–21.
8. Isikbay M, Otto K, Kregel S, et al. Glucocorticoid receptor activity contributes to resistance to androgen-targeted therapy in prostate cancer. Horm Cancer 2014; 5: 72–89.
9. Kach J, Long T, Selman P, et al. Selective glucocorticoid receptor modulators (SGRMs) delay castrate-resistant prostate cancer growth. Mol Cancer Ther 2017; 16: 1680–92.
10. Sharon M, Pozniak Y, Geiger T, et al. TMPRSS2-ERG fusion protein regulates insulin-like growth factor-1 receptor (IGF1R) gene expression in prostate cancer: involvement of transcription factor Sp1. Oncotarget 2016; 7: 51375–92.
11. Samani A, Yakar S, Le Roith D, et al. The role of the IGF system in cancer growth and metastasis: overview and recent insights. Endocr Rev 2007; 28: 20–47.
12. Heidegger I, Kern J, Ofer P, et al. Oncogenic functions of IGF1R and INSR in prostate cancer include enhanced tumor growth, cell migration and angiogenesis. Oncotarget 2014; 5: 2723–35.
13. Jia M, Dahlman-Wright K, Gustafsson J. Estrogen receptor alpha and beta in health and disease. Best Pract Res Clin Endocrinol Metab 2015; 29: 557–68.
14. Setlur S, Mertz K, Hoshida Y, et al. Estrogen-dependent signaling in a molecularly distinct subclass of aggressive prostate cancer. J Natl Cancer Inst 2008; 100: 815–25.
15. Sackmann-Sala L, Goffin V. Prolactin-induced prostate tumorigenesis. Adv Exp Med Biol 2015; 846: 221–42
16. Kambale P, Haldar D, Kabi B, et al. Study of vitamin D receptor gene polymorphism (FokI, TaqI and ApaI) among prostate cancer patients in North India. J Clin Diagn Res 2017; 11: BC05–8.
17. Roberts J, Martin R, Piyarathna D, et al. Vitamin D receptor activation reduces VCaP xenograft tumor growth and counteracts ERG activity despite induction of TMPRSS2:ERG. Oncotarget 2017; 8: 44447–64.
18. Gerashchenko G, Mankovska O, Dmitriev A, et al. Epithelial-mesenchymal transition related gene expression in prostate tumours. Biopolym Cell 2017; 33: 335–55.
19. Flannery C, Rowzee A, Choe G, et al. Development of a quantitative PCR assay for detection of human insulin-like growth factor receptor and insulin receptor isoforms. Endocrinol 2016; 157: 1702–8.
20. Schmidt U, Fuessel S, Koch R. Quantitative multi-gene expression profiling of primary prostate cancer. Prostate 2006; 66: 1521–34.
21. Rosenberg E, Gerashchenko G, Hryshchenko N, et al. Expression of cancer-associated genes in prostate tumors. Exp Oncol 2017; 39: 131–7.
22. Benjamini Y, Hochberg Y. Controlling the false discovery rate: a practical and powerful approach to multiple testing. J Royal Statist Society 1995; 57: 289–300.
23. Lapointe J, Kim Y, Miller M, et al. A variant TMPRSS2 isoform and ERG fusion product in prostate cancer with implications for molecular diagnosis. Mod Pathol 2007; 20: 467–73.
24. Luo J, Liu S, Zuo Z, et al. Discovery and classification of fusion transcripts in prostate cancer and normal prostate tissue. Amer J Pathol 2015; 185: 1834–45.
25. Feng F, Brenner C, Hussain M, et al. Molecular pathways: targeting ETS gene fusions in cancer. Clin Cancer Res 2014; 20: 4442–8.
26. Austin D, Strand D, Love H, et al. NF-κB and androgen receptor variant 7 induce expression of SRD5A isoforms and confer 5ARI resistance. Prostate 2016; 76: 1004–18.
27. Aggarwal S, Singh M, Kumar A, et al. SRD5A2 gene expression inhibits cell migration and invasion in prostate cancer cell line via F-actin reorganization. Mol Cell Biochem 2015; 408: 15–23.
28. Smolle M, Bauernhofer T, Pummer K, et al. Current insights into long non-coding RNAs (LncRNAs) in prostate cancer. Int J Mol Sci 2017; 22: E473–90.
29. Zhang A, Zhao J, Kim J, et al. LncRNA HOTAIR enhances the androgen-receptor-mediated transcriptional program and drives castration-resistant prostate cancer. Cell Rep 2015; 13: 209–21.
30. Teixeira A, Marchit S, Emmanuel Dias-Neto E, et al. Going viral? Linking the etiology of human prostate cancer to the PCA3 long noncoding RNA and oncogenic viruses. EMBO Mol Med 2017; 9: 1327–30.

No Comments » Add comments