Comprehensive analysis of oncogenic signatures and consequent repurposed drugs in <i>TMPRSS2:ERG</i> fusion‐positive prostate cancer

Abstract

Dear Editor, TMPRSS2:ERG (TE) fusion occurs in approximately 50% of all prostate cancer cases.1 However, details about altered signaling or the difference of gene expression regarding potential therapeutic targets between TE fusion-positive and negative group is yet to be fully investigated. In this study, we investigated the landscape of molecular signaling and curated potential therapeutic targets in TE fusion-positive prostate cancers using The Cancer Genome Atlas data. Firstly, we identified 3870 genes in coordination with ERG in RNA expression and nine cancer-related pathways specifically altered in TE fusion-positive prostate cancer patients. Secondly, we deduced repositionable 55 drugs targeting for TE fusion-positive prostate cancer from network analysis. Finally, we provided experimental data for six drugs obtained from our in silico analysis and showed sensitivity specific for TE fusion-positive prostate cancer cell line. This study is designed as shown in the overview (Figure 1): First, after getting RNA-seq data and clinical information from broad global data assembly centers firehose (GDAC, https://gdac.broadinstitute.org/), we selected the genes correlated with ERG using Pearson correlation test (|R| > 0.3) with the RNA expression level of each gene (Figure S1). Second, pathway analysis was performed using ConsensusPathDB (CPDB, http://consensuspathdb.org/)2 and pathways with key altered genes were visualized (Figures 2 and S2-S4.). Third, potential actionable drugs were inferred through network analysis using Clinical Interpretation of Variants in Cancer (CIViC) database (https://civicdb.org/)3 and ERG correlated gene list (Figure 3A). Finally, among the actionable drugs inferred, seven drugs were selected, and drug sensitiveness was tested using TE fusion-positive and negative prostate cancer cell lines (Figure 3B). In overrepresentation analysis for elucidation of target cellular pathways in TE fusion-positive prostate cancer patients, we identified nine altered signaling pathways, including Wnt signaling, Androgen receptor (AR) signaling, gene expression signaling, VEGFA-VEGFR2 signaling, p53 signaling, NOTCH1 signaling, TGF-beta signaling, p53-independent G1/S DNA damage checkpoint, and insulin signaling (Figures 2 and S2-S4), and confirmed it in validation set except p53-independent G1/S DNA damage checkpoint due to low incidence (Figure S5). Among them, various kinds of HDACs (HDAC1, HDAC2, HDAC4, HDAC6, and HDAC7) were found to be participating in five signaling pathways (Figures 2 and S2-S3), suggesting that ERG upregulation alters HDAC1,2,4,6 and 7, which can play key roles in prostate cancer signaling. Indeed, we found that HDAC1 showed the best correlation with ERG in RNA expression (R = 0.82). These results are supported by previous study that ERG is known to form ESET (ERG-associated protein with a SET domain) with HDAC1, and ESET is related with pluripotency and de-differentiation which is important signaling in cancer.4, 5 We also identified well established cancer-specific genes (Table S1) that were associated with TE fusion-positive cases. In case of AR signaling (Figure 2A), 33 of cancer-related genes are altered including EP300, CREBBP, CDKN1A, AKT1, CCND1, CDK6, and TMPRSS2 genes in TE-positive group. As consistent with our analysis, different androgen profiles were observed in TE fusion-positive patients.6 In addition, clinical impact of androgen has been introduced, specifically, androgen deprivation therapy showed survival benefit in TE fusion positive prostate.7 Wnt signaling was also confirmed to be altered in pathway analysis (Figure 2B). In this pathway, 14 of cancer-related genes are altered including TSC2, CDH1, TCF7L2, AKT1, CDK6, and CCND1 genes. In addition, alteration of TGF-beta, NOTCH1, VEGFA-VEGFR2, gene expression, and insulin signaling were also listed to be associated with ERG expression (Figures 2 and S2-S3). Intriguingly, although various molecular pathways and associated genes are significantly altered in RNA-level (Figure 2), clinical phenotype is not significantly different between the TE fusion-positive and the negative group (Table S2). These results indicate that even if the patient's clinical phenotype is similar, the molecular subtype is different at the molecular level, and therefore the drug target is significantly different, indicating that understanding molecular features in each patient is crucial for cancer therapy. Next, we examined therapeutic targets and potential actionable drugs through the pathway analysis based on RNA expression and the annotation with drug-target database. In this analysis, RNA expression of 28 genes, which could be potential therapeutic targets, was observed to have different levels of alteration between TE fusion-positive and negative groups, and they were involved in various signaling pathway (Table 1). Interestingly, in gene-drug network analysis, 14 drugs were found to be related with multiple genes among 28 genes which were up- or downregulated in TE fusion-positive group (Figure 3A). To prove whether our in silico analysis method is supporting the idea of drug repurposing, we randomly selected seven drugs and performed cell viability test in VCap cells, TE fusion-positive cells and 22RV1 cells, TE fusion-negative cells. Dasatinib (targeting ABL1), imatinib (targeting ABL1, PDGFRB, and BCL2L11), and olaparib (targeting CBLC, CDK12, and ATM) effectively reduced viability of VCap cells compared to 22RV1 as expected because the expression of these genes targeted by dasatinib, imatinib, and olaparib was positively correlated with ERG expression (Figures 3A and 3B). Gefitinib, targeting IGF1R and ERBB2 inhibited viability of VCap cells compared to 22RV1 cells, and it is thought to be via inhibition of upregulated IGF1R rather than downregulated ERBB2 (Figure 3B). In addition, we tested effect of everolimus, tamoxifen, and cisplatin. Everolimus, tamoxifen, and cisplatin target genes whose expressions are both positively and negatively correlated with ERG expression. Everolimus and tamoxifen decreased Vcap cell viability compared to 22RV1 (Figure 3B). Although cisplatin is one of classical chemotherapeutic agent based on NCCN guideline,8 cisplatin does not seem to be specifically effective in TE fusion-positive prostate cancer cell line in our in vitro study (Figure 3B). Cisplatin could target both BIRC7 and ABCB1, and its effect could be offset. We need to further explore the effect of drugs in each target genes and signaling pathway to precisely understand underlying mechanism of each drug. But at least, we here suggest that our systematic in silico analysis is proper approach for drug repurposing study. Taken together, we provided the portrait of cellular signaling pathways and prioritized therapeutic targets correlated with ERG expression in TE fusion-positive prostate cancer. We believe that this study will further advance precision medicine in prostate cancer treatment. This study was supported by a VHS Medical Center Research Grant, Republic of Korea (VHSMC 20049), grant number: 18-2018-023 from the SNUBH Research Fund, the National Research Foundation of Korea (NRF) funded by Ministry of Science and ICT (MSIT), (grant number: NRF-2020R1F1A1074659), and the Basic Science Research Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Education (grant numbers: NRF-2018R1D1A1A02086141 and NRF-2019R1A6A10073079) by the Korea government. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.

Topics & Concepts