Extensive germline-somatic interplay contributes to prostate cancer progression through HNF1B co-option of TMPRSS2-ERG

Giannareas, Nikolaos; Zhang, Qin; Yang, Xiayun; Na, Rong; Tian, Yijun; Yang, Yuehong; Ruan, Xiaohao; Huang, Da; Yang, Xiaoqun; Wang, Chaofu; Zhang, Peng; Manninen, Aki; Wang, Liang; Wei, Gong-Hong

doi:10.1038/s41467-022-34994-z

Download PDF

Article
Open access
Published: 28 November 2022

Extensive germline-somatic interplay contributes to prostate cancer progression through HNF1B co-option of TMPRSS2-ERG

Nikolaos Giannareas¹^na1,
Qin Zhang¹^na1,
Xiayun Yang¹^na1,
Rong Na ORCID: orcid.org/0000-0001-7470-5108²^na1,
Yijun Tian³,
Yuehong Yang¹,
Xiaohao Ruan⁴,
Da Huang⁴,
Xiaoqun Yang⁵,
Chaofu Wang⁵,
Peng Zhang⁶,
Aki Manninen ORCID: orcid.org/0000-0002-6263-8101¹,
Liang Wang ORCID: orcid.org/0000-0002-9364-8572³ &
…
Gong-Hong Wei ORCID: orcid.org/0000-0001-6546-9334^1,6

Nature Communications volume 13, Article number: 7320 (2022) Cite this article

4078 Accesses
6 Citations
1 Altmetric
Metrics details

Subjects

Abstract

Genome-wide association studies have identified 270 loci conferring risk for prostate cancer (PCa), yet the underlying biology and clinical impact remain to be investigated. Here we observe an enrichment of transcription factor genes including HNF1B within PCa risk-associated regions. While focused on the 17q12/HNF1B locus, we find a strong eQTL for HNF1B and multiple potential causal variants involved in the regulation of HNF1B expression in PCa. An unbiased genome-wide co-expression analysis reveals PCa-specific somatic TMPRSS2-ERG fusion as a transcriptional mediator of this locus and the HNF1B eQTL signal is ERG fusion status dependent. We investigate the role of HNF1B and find its involvement in several pathways related to cell cycle progression and PCa severity. Furthermore, HNF1B interacts with TMPRSS2-ERG to co-occupy large proportion of genomic regions with a remarkable enrichment of additional PCa risk alleles. We finally show that HNF1B co-opts ERG fusion to mediate mechanistic and biological effects of the PCa risk-associated locus 17p13.3/VPS53/FAM57A/GEMIN4. Taken together, we report an extensive germline-somatic interaction between TMPRSS2-ERG fusion and genetic variations underpinning PCa risk association and progression.

Genetic alterations in the 3q26.31-32 locus confer an aggressive prostate cancer phenotype

Article Open access 14 August 2020

Acquired copy number variation in prostate tumours: a review of common somatic copy number alterations, how they are formed and their clinical utility

Article 09 November 2023

CRISPRi screens reveal a DNA methylation-mediated 3D genome dependent causal mechanism in prostate cancer

Article Open access 19 March 2021

Introduction

Genome-wide association studies (GWASs) have successfully discovered thousands of risk-associated single nucleotide polymorphism (SNP) loci for cancers of prostate and others. While most SNP associations are cancer-specific, roughly one-third of such genomic loci with SNPs associates with multiple types of cancers, namely pleiotropic loci that may have shared mechanisms or hallmarks across cancers¹. Due to the complexity of human genome, the vast majority of GWAS-identified SNPs fell within non-coding genomic regions that have been proven to possess regulatory functions in modulating gene expression through diverse mechanisms^2,3,4,5,6. Growing evidence indicate that GWAS loci are often involved in expression quantitative trait locus (eQTL) conferring cancer risk via altering the DNA-binding affinity of critical transcription factors to causal SNP-containing regulatory elements such as enhancers, representing a major driving force of cancerous gene expression program^2,5,7. However, it remains challenging to define somatic driver transcription factors, target genes and aberrant biological pathways from GWAS discoveries, thereby depicting the molecular mechanisms of germline–somatic interplay and continuum underlying these cancer risk loci.

Prostate cancer (PCa) remains the second most common cancer in men and globally affects millions of individuals. It was estimated that more than 1.41 million men have been diagnosed with PCa and 375,000 PCa-associated deaths occurred worldwide in 2020⁸. PCa has complex etiology and a high estimate of heritability at 57%⁹. Therefore, unraveling PCa risk-associated genetic factors and investigating their underlying mechanisms and biological impacts are expected to greatly inform our understanding of PCa pathogenesis, progression, and clinical management. Comprehensive GWAS analyses have been performed in men with PCa across diverse ancestry groups, and these studies have jointly identified 270 susceptibility loci harboring over 400 SNPs that reach genome-wide significance (P ≤ 5 × 10⁻⁸) in association with PCa risk and aggressiveness^10,11,12,13. Of these PCa risk loci, the 17q12/HNF1B locus variants rs4430796, rs11263763 and rs11651052 have been reproducibly found to be associated with PCa susceptibility^{14,15,16,17,18,19,20,21}. Moreover, SNPs in the 17q12/HNF1B region have also been reported in association with risk of several other cancer types, including pancreatic²², ovarian^23,24, testicular²⁵, and endometrial²⁶ cancers. Yet, the mechanistic effects and underlying biology of the 17q12/HNF1B remain elusive for all of these associated cancer types. HNF1B as a 17q12 locus gene, belongs to HNF1 transcription factor family whose members possess a POU-homeodomain responsible for sequence-specific DNA binding to gene cis-regulatory elements (CRE)^14,27. HNF1B has been reported to play roles in tumourigenesis and is used as a biomarker for clear cell carcinomas of the pancreas²⁸, colorectal cancer²⁹ and endometrial carcinoma³⁰, implying that HNF1B is a plausible causative gene for this pleiotropic association.

Results from our unbiased genome-wide co-expression and ChIP-seq analysis reveal that TMPRSS2-ERG is likely to be an effective transcription regulator of the 17q12/HNF1B PCa risk locus. TMPRSS2-ERG is the most frequent somatic fusion event in PCa, involving a chromosomal rearrangement of ERG transcription factor hijacking the 5′ androgen-responsive regulatory region of TMPRSS2 to form a constitutively activated mutant TMPRSS2-ERG fusion protein^31,32. The frequency of ERG fusion is variable among different ethnic groups, showing the highest frequency (>50%) in Caucasians, followed by African American (20%-30%) and Asian men (<20%)^33,34,35,36. Aberrant ERG fusion transcription factor influences various pathways and biological processes such as androgen receptor (AR) signaling, transforming growth factor beta 1 (TGF-β) signaling and cell invasion^37,38. While the association of TMPRSS2-ERG fusion status with PCa clinical outcomes remains inconclusive³⁹, some reports have demonstrated the role of ERG fusion in PCa cellular growth and tumor progression. Studies have shown that knockdown of ERG inhibits cell proliferation, invasion and xenograft tumor growth of TMPRSS2-ERG-positive PCa cell line VCaP⁴⁰ while ERG overexpression lead to PCa precursor-like lesions in mice⁴¹ and promoted cell invasion in vitro⁴⁰. In addition, cooperation of prostate-specific expression of ERG and genetic activation of the PI3K/AKT pathway or loss of PTEN drive PCa progression in mouse models^42,43. However, given the high rate of TMPRSS2-ERG genomic translocation in PCa, it remains to be investigated whether and how this somatic fusion is involved in germline risk loci for PCa discovered by GWASs.

In this work, employing an unbiased enrichment analysis, we show that the transcription factor genes, such as HNF1B, are greatly enriched nearby known PCa susceptibility loci, implying a hypothesis that an interconnected gene regulatory network by core genes, transcription factors in particular, may explain subtle effects of GWAS-discovered SNPs on complex diseases and traits. We therefore focus on the 17q12/HNF1B locus and seek to identify functional causal variants of this region as well as investigate role of HNF1B in PCa progression and predisposition to aggressive disease, which in turn may be an example for functional study of this cross-cancer pleiotropic genetic association in other types of cancers. We show that TMPRSS2-ERG is a transcription regulator mediating mechanistic effects of the 17q12 locus and control the expression of HNF1B, which in turn regulates a set of cell cycle genes implicating PCa predisposition and progression. We find that HNF1B and TMPRSS2-ERG physically interact with each other, have a remarkable chromatin co-occupancy and cooperatively regulate genes implicated in PCa development. Lastly, we observe that common binding sites for HNF1B and TMPRSS2-ERG can explain more of genetic effects on PCa predisposition than that of their unique binding sites, and present a solid example of HNF1B co-option of TMPRSS2-ERG co-regulating the 17p13.3 PCa risk locus.

Results

Transcription factor genes including HNF1B are markedly enriched in PCa susceptibility loci

Transcription factors direct the chromatin binding to cis-acting regulatory DNA elements (CREs) and play central roles in gene expression networks^44,45. We thereby examined whether the genes encoding transcription factors are more likely to be enriched in PCa susceptibility loci and thus extract nearby genes of each PCa GWAS variant for a statistical examination. This analysis revealed that transcription factor genes are indeed preferentially to be enriched in PCa risk loci (P = 2.49 × 10⁻⁶, hypergeometric distribution test; Fig. 1a), including many with poorly characterized causal roles (Fig. 1b), except for HOXB13, RFX6, and NKX3-1 that have been shown to be causally linked to PCa susceptibility and tumourigenesis^46,47,48,49. Among these functionally uncharacterized PCa risk loci, the 17q12/HNF1B locus with several independent SNPs has been reproducibly found in association with PCa susceptibility^{14,15,16,17,18,19,20,21}, and interestingly the phenome-wide association analysis (PheWAS) using FinnGen cohort data (n = 176,899) observed a specific top-ranked association of the 17q12/HNF1B locus variants with PCa across 2,264 disease endpoints (Fig. 1c). Together with other observations describing 17q12/HNF1B in association with multiple types of cancer^{22,23,24,25,26}, this finding prompted us to delve into the molecular and biological mechanisms as well as the clinical implications of this locus, and the regulation of HNF1B in PCa.

**Fig. 1: Transcription factor genes are highly enriched nearby PCa risk loci and the 17q12 HNF1B is associated with PCa cell growth and tumor progression.**

We first evaluated the expression profiles of HNF1B across 31 cancer types using RNA-seq data of The Cancer Genome Atlas (TCGA) and found that HNF1B was highly expressed in approximately top 25% of cancer types including PCa (Supplementary Fig. 1a). To test the biological relevance of HNF1B in PCa, we performed a cell proliferation assay and found that several PCa cell lines including LNCaP, VCaP, DU145, and PC3 harboring siRNAs against HNF1B displayed reduced cell growth compared to cells harboring control siRNA (Supplementary Fig. 1b–i). In line with this observation, downregulation of HNF1B via lentivirus-mediated short hairpin RNA (shRNA) greatly attenuated cell proliferation and migration in the PCa V16A cell line while ectopic expression of HNF1B elevated cellular proliferation of RWPE1 cells (Fig. 1d–f and Supplementary Fig. 1j). Moreover, the data from a genome-wide CRISPR-mediated loss-of-function screen in the PCa cell line VCaP⁵⁰ demonstrated that HNF1B is a top-ranked essential gene for cell survival (Fig. 1g), further indicating the importance of HNF1B in PCa cell growth and survival.

To investigate the functional and clinical relevance of HNF1B in PCa, we analyzed multiple independent PCa expression profile datasets^{51,52,53,54,55,56}. The results showed that HNF1B expression is greatly elevated in primary and metastatic prostate tumors compared to normal prostate gland (Fig. 1h, i and Supplementary Fig. 1k, l). Moreover, we observed that the mRNA levels of HNF1B are greatly increased in localized or metastatic PCa samples and in tumors of higher grade and Gleason score (Fig. 1j, k), suggesting a potential function for HNF1B in advanced prostate tumors. Together, these findings illustrated a potential role for HNF1B in PCa severity and progression, indicating that HNF1B is a plausible causative gene underlying the effects of the 17q12 PCa susceptibility locus variants.

The 17q12/HNF1B PCa risk locus harbors a single risk-associated gene and multiple causal variants

To identify functional causal variants at HNF1B locus, we screened a total of 13 SNPs highly associated with multiple PCa risk variants at 17q12 using the expression quantitative trait locus (eQTL) data from Wisconsin cohort of 466 prostate normal tissue samples^57,58,59. We first examined the mRNA levels of different HNF1B isoforms and found that HNF1B isoform 1 is dominantly expressed compared to the other two isoforms in prostate tissues (Supplementary Fig. 2a). The eQTL analysis revealed a strong significant association between HNF1B isoform 1 and 11 of the 13 SNPs (Fig. 2a and Supplementary Fig. 2b–o). Remarkably, further cis-eQTL in this Wisconsin cohort revealed the strongest association of HNF1B among all genes within one mega-base window with any of these 11 SNPs (Fig. 2b and Supplementary Data 1). We next performed CRISPR/Cas9-mediated genome editing in the PCa cell lines 22Rv1 and V16A to delete each of the 13 SNP-containing regions and observed great downregulation of HNF1B in most of mixed clones with knockout (KO) of individual SNP-region (Fig. 2c and Supplementary Fig. 2p). To test whether the SNPs residing in HNF1B locus have an enhancer variant-like function in regulating HNF1B, we examined six SNP regions based on the above CRISPR/Cas9-mediated KO screening. Each SNP region with two possible alleles was cloned into upstream of HNF1B promoter in luciferase reporter constructs in both orientations. Enhancer report assays in 22Rv1 and LNCaP cells showed that rs11651052, rs12453443, rs9901746, rs7405696 and rs11263763 regions indicate enhancer-like function to activate luciferase gene expression compared to the HNF1B promoter (Fig. 2d and Supplementary Fig. 2q). Specifically in 22Rv1 cells, the rs9901746 region indicated an orientation-dependent effect on the enhancer activity with 3’−5’ orientation in both alleles. In addition, several SNP regions showed enhancer activity in the presence of specific alleles and, in particular the G and A alleles of rs7405696 and rs11651052, respectively, displayed stronger activity than the C alleles of rs7405696 or the allele G of rs11651052 towards HNF1B promoter in regulating luciferase gene expression (Fig. 2d). To further verify CRISPR/Cas9 screen and the enhancer report assays, we perform genomic deletion of those SNP-enhancer regions through CRISPR/Cas9-mediated KO at single cell levels in the PCa 22Rv1 cells. We observed a profound downregulation of HNF1B in every clone in which SNP-enhancer regions were deleted (Fig. 2e). These results pointed to HNF1B as a PCa risk-associated gene and defined several potential causal variants that are likely to be involved in the regulation of HNF1B expression at 17q12 locus.

**Fig. 2: Multiple eQTL SNPs at 17q12 locus are potential causal variants and involved in the regulation of *HNF1B* expression.**

Somatic TMPRSS2-ERG fusion as a regulator of the 17q12/HNF1B PCa risk locus

Accumulating evidence indicates that GWAS loci are often involved in key transcription factors acting through genomic regulatory elements like enhancers to drive eQTL gene expression^{2,3,4,5,6,46,60}, and given eQTL genes often display co-expression with the responsible transcription factors in the clinical settings^46,60. We thus performed a genome-wide co-expression analysis of HNF1B in several large clinical PCa datasets to figure out potential transcription factors that are regulating HNF1B via the causal variant containing regions. Intriguingly, ERG emerged as the most positively co-expressed gene with HNF1B in TCGA cohort comprised of 497 primary PCa tumors (Fig. 3a, b and Supplementary Fig. 3a). ERG was also found as a top-ranking gene showing positive expression correlation with HNF1B in another cohort of 264 PCa tumors (Supplementary Fig. 3b). By contrast, ERG exhibited no apparent expression correlation with HNF1B in normal prostate tissues (Supplementary Fig. 3c–e), suggesting that this association is specific to human prostate tumorigenesis. As mentioned above, a common somatic genomic rearrangement involves TMPRSS2 and ERG genes that form an androgen signaling-regulated mutant transcription factor, TMPRSS2-ERG^31,61,62,63. We next assessed whether ERG contributes to the regulation of HNF1B expression by using siRNA-mediated knockdown assay in VCaP cells that harbors a TMPRSS2-ERG fusion³¹ and LNCaP cell line negative to the fusion. This analysis revealed that depletion of TMPRSS2-ERG resulted in decreased mRNA levels of HNF1B in VCaP but not in LNCaP cells (Fig. 3c and Supplementary Fig. 3f).

**Fig. 3: ERG is a responsible transcriptional regulator of the 17q12 locus and regulates HNF1B in PCa cells and clinical tissues in a manner dependent on TMPRSS2-ERG fusion status.**

To understand the regulatory mechanisms at HNF1B locus, we explored a large collection of genome-wide chromatin immunoprecipitation sequencing (ChIP-seq) data derived from PCa⁶⁴. We found an enrichment of epigenetic marks (H3K4me1/2 and H3K27ac) for active enhancers and chromatin binding of ERG transcription factor across or nearby multiple potentially causal SNPs within HNF1B locus (Fig. 3d). We next performed ChIP assays with antibodies to ERG, HNF1B, AR, histone modifications, or with IgG as control. Using quantitative PCR (qPCR) in VCaP with or without androgen treatment, we confirmed specific chromatin enrichment of ERG, HNF1B, AR, H3K4me1, H3K4me2, H3K4me3, and H3K27ac at rs12453443, rs7405696, rs718960, rs11263763, and rs11651052 regions, respectively (Fig. 3e–i). In contrast, similar ChIP assays in fusion-negative LNCaP cells revealed no clear enrichment of ERG in these SNP regions and rather weak enrichment of AR, HNF1B and histone marks (Supplementary Fig. 3g–k). Through a motif analysis at the HNF1B locus, we found four SNPs rs4430796, rs718960, rs8064454 and rs11651052 to some extent residing within DNA-binding motifs of ERG and HNF1B (Supplementary Fig. 3l). Given that ERG hijacks the regulatory element of TMPRSS2 gene to form the most frequent ERG fusion in the clinical setting of PCa³¹, we further investigated in various cohorts whether the expression correlation between ERG and HNF1B is dependent on the TMPRSS2-ERG fusion status. These analyses revealed that ERG and HNF1B show apparent co-expression in TMPRSS2-ERG fusion-positive groups of PCa specimens, but not in fusion-negative groups (Fig. 3j, k and Supplementary Fig. 3m, n).

SNPs are germline inherited while TMPRSS2–ERG fusion is an acquired somatic genomic alteration. Concurrent presence of these germline-somatic features might exert synergistic impact on the target gene HNF1B with a positive eQTL signal in this region. In line with this, we found that none of these 13 SNPs (Fig. 2a) had eQTL association with HNF1B in TCGA cohort of PCa patients (Fig. 3l). When taking into account of TMPRSS2-ERG fusion status and stratifying patients into TMPRSS2-ERG fusion-positive and negative groups, we observed that the eQTL results from the ERG fusion-negative group are non-significant (Fig. 3m and Supplementary Fig. 4), whereas seven SNPs showed significant eQTL signal with HNF1B in the TMPRSS2-ERG fusion-positive PCa tumors (Fig. 3n and Supplementary Fig. 5a–n). Based on these observations, we next asked whether the frequency of the co-occurrence of TMPRSS2-ERG fusion-positive tumors that present the HNF1B locus SNP alleles increase the risk of PCa aggressiveness in the TCGA cohort. We thus performed an enrichment analysis of these SNPs in TMPRSS2-ERG fusion-positive and -negative tumors. However, the results revealed no particular enrichment pattern for the SNP alleles in the PCa patients with TMPRSS2-ERG fusion-positive or -negative tumors (Supplementary Fig. 5o). We next additionally investigated whether genotype at SNPs in HNF1B is associated with TMPRSS2-ERG positive tumors in a Chinese PCa cohort (see “Methods”; Supplementary Table 1). We first performed a case-control study (791 PCa vs 752 non-PCa) and found that, among these 13 SNPs, rs8064454, rs7405696, rs11651052, rs9901746, rs11263763, rs11658063 and rs12453443 were significantly associated with PCa in this cohort (Supplementary Table 2). We next carried out a PCa case-only study (136 ERG positive vs 655 ERG negative) while assuming that positive ERG expression based on immunohistochemistry (IHC) is due to the TMPRSS2-ERG fusion. Despite that we did not observe significant associations of the 13 HNF1B locus SNPs with ERG fusion status, this analysis revealed a moderate but significant association of the three LD SNPs rs79882976 (P = 0.01), rs11651496 (P = 0.01) and rs3744764 (P = 0.02) with ERG expression positive tumors (Supplementary Table 3). We expect larger sample size may enable the detection of more associations between the HNF1B SNP genotypes and TMPRSS2-ERG fusion-positive PCa tumors. Taken together, these results suggest that the effect of these variants on the target gene HNF1B are further orchestrated by the acquired somatic event of TMPRSS2-ERG gene fusion.

HNF1B regulates cell cycle progression pathways implicating PCa severity

To understand the role of HNF1B in PCa, we performed RNA-seq analysis of differentially expressed genes upon depletion of HNF1B with siRNA control or two different siRNAs against HNF1B in the PCa VCaP cells (Fig. 4a). In each experimental group, two biological replicates were included and showed high correlations (Supplementary Fig. 6a–c). This RNA-seq analysis identified 207 significantly upregulated and 132 downregulated genes by HNF1B, respectively (DESeq2, P < 0.05; Fig. 4b). Representative differentially expressed genes were further validated via qPCR after siRNA-mediated knockdown of HNF1B (Supplementary Fig. 6d, e). To investigate potential functional categories of HNF1B knockdown target genes, we performed gene set enrichment analysis (GSEA) and revealed several cell cycle relevant terms highly enriched in HNF1B upregulated genes (Fig. 4c, d), further supporting the above described role of HNF1B in promoting PCa cell proliferation and invasion (Fig. 1d–f and Supplementary Fig. 1b–j). To explore the clinical relevance of HNF1B target genes in PCa, we generated a cell cycle or knock-down signature of HNF1B target genes (see “Methods”). We evaluated the clinical significance in multiple independent cohorts and found that HNF1B cell cycle signature score is positively correlated with the cell cycle progression (CCP) scores⁶⁵ (Fig. 4e, f and Supplementary Fig. 6f–l). Consistently, the HNF1B knock-down signature derived from differentially expressed genes by HNF1B also displays noteworthy positive linear correlation with the CCP scores (Fig. 4g and Supplementary Fig. 6m–t). We next investigated the clinical relevance of HNF1B cell cycle signature with PCa tumor progression and severity. The results indicated that the HNF1B cell cycle signature score is significantly elevated in the metastatic group than primary PCa and normal prostate glands (Fig. 4h–j and Supplementary Fig. 7a, b). Patients with higher levels of HNF1B cell cycle signature score are strongly associated with PCa severity, including Gleason score, lymph node metastasis, tumor stage, and biochemical recurrence (Fig. 4k–m and Supplementary Fig. 7c–j). In addition, we observed that patient group with higher levels of HNF1B cell cycle signature score are associated with shorter overall survival, elevated biochemical relapse and metastasis risk (Fig. 4n–p and Supplementary Fig. 7k, l). To confirm the robustness of the association, we next additionally performed a meta-analysis to systematically review, integrate, and provide an overall interpretation of the association between HNF1B cell cycle signature and patient prognostic outcomes across different studies⁶⁶. The meta-analysis revealed that the elevated levels of HNF1B cell cycle signature were significantly associated with poor overall survival (P = 3.5e–03, lnHR: 0.85; 95% CI: 0.28–1.41) and biochemical recurrence (P = 6.1e–03, lnHR: 0.50; 95% CI: 0.14–0.85) in PCa patients. To validate the effect of the HNF1B cell cycle signature among a set of clinical variables to patient prognosis, we conducted multivariate analysis by incorporating clinical variables including age, Gleason score, PSA, tumor stage, ERG-fusion status, seminal vesical status and extraprostatic extension status in PCa patients using both the continuous and categorical HNF1B cell cycle signature by the median stratification. The multivariate results still showed significant association of the HNF1B cell cycle signature with patient overall survival (Supplementary Fig. 7o) and biochemical recurrence (Supplementary Fig. 7p) among other clinical variables. To further investigate the clinical importance of HNF1B target genes, we derived HNF1B knockdown upregulated signature, consisting of 207 upregulated genes. We found that patients with higher levels of the score are associated with PCa tumor progression and severity (Supplementary Fig. 7q–x). Collectively, these findings indicate a strong association of HNF1B target genes with PCa progression and implicate the role of HNF1B in PCa severity.

**Fig. 4: HNF1B gene signatures correlate with PCa progression in the clinical setting.**

Given that HNF1B is directly regulated by and co-expressed with ERG transcription factor, we examined whether HNF1B targeted genes are also regulated by ERG. Thus, we first performed a co-expression analysis of HNF1B and 339 differentially expressed genes upon HNF1B knockdown across multiple clinical PCa datasets. We found that the fraction of genes co-expressed with HNF1B was higher, showing significant co-expression with ERG (Fig. 4q and Supplementary Fig. 7y). Based on RNA-seq analysis of HNF1B knockdown, we next curated a panel of 25-gene as HNF1B co-expression signature and measured the degree of their expression correlation with ERG in prostate tumors. The results indicated a similar expression pattern for the 25 genes when compared ERG in two large independent cohorts of PCa (Fig. 4r and Supplementary Fig. 7z), further supporting a functional interplay between the PCa susceptibility gene HNF1B and the PCa-specific somatic ERG fusion transcription factor.

HNF1B and ERG display physical interaction, chromatin co-occupancy and cooperative regulation of PCa genes

Given that ERG fusion correlates with and regulates HNF1B expression and is a transcriptional mediator of HNF1B locus variants, we next examined whether HNF1B and ERG may interact physically with each other. To this end we performed co-immunoprecipitation (Co-IP) assays that revealed indeed an interaction between endogenous HNF1B and TMPRSS2-ERG in VCaP cells treated with ETH (ethanol) and DHT (Dihydrotestosterone) (Fig. 5a). We applied nuclease treatment to rule out the possibility that the interaction between ERG and HNF1B was mediated by a DNA/RNA moiety (Supplementary Fig. 8a). In addition, treatment of VCaP cells with enzalutamide showed partial abolishment of the interaction (Fig. 5b) and slight downregulation of ERG and HNF1B (Supplementary Fig. 8b), suggesting possible therapeutic implications. Moreover, our experiments showed protein-protein interaction between ectopically expressed V5-tagged TMPRSS2-ERG and flag-tagged HNF1B (Supplementary Fig. 8c). We found that ERG interacts with the POU domain of HNF1B protein (Fig. 5c, d and Supplementary Fig. 8d). Consistent with the physical cooperation of TMPRSS2-ERG and HNF1B, it was observed that more than 70% (3,632/5,133) of HNF1B genome-wide binding sites were co-occupied by TMPRSS2-ERG (Fig. 5e). To investigate whether HNF1B global chromatin binding sites vary in different cell lines, we compared HNF1B genome-wide binding regions in VCaP cells with the ones previously reported in ERG fusion-negative and androgen-unresponsive cell line DU145⁶⁷. We observed approximately 19% of overlapping peaks for HNF1B between the two cell lines, indicating the chromatin binding profile varies greatly between ERG fusion-positive and negative cell lines (Supplementary Fig. 8e). We also compared the global binding profiles of ERG in VCaP and of HNF1B in DU145 cells and observed nearly 48% of common binding sites (Supplementary Fig. 8f), which is extensively overlapping but rather lower than that of their overlap rate in the same cell line VCaP (Fig. 5e).

**Fig. 5: HNF1B physically interacts with ERG and co-opts ERG for chromatin co-occupancy and clinical correlation in PCa.**

We next derived a list of 51 genes as the joint target signature of HNF1B and ERG via an integrated analysis of RNA-seq-defined HNF1B upregulated genes and ChIP-seq profiled genomic co-occupancies of HNF1B and ERG in VCaP (see Methods; Fig. 5f). Notably, HNF1B itself is also a direct target gene of HNF1B and ERG (Fig. 5f) and has several co-binding sites of the two transcription factors (Fig. 5g and Supplementary Fig. 8g). DNA-binding motifs of HNF1B and ERG were frequently observed within the co-binding sites at HNF1B (Supplementary Fig. 8g). Using ChIP-qPCR, we further validated chromatin co-binding of HNF1B and ERG within HNF1B genomic region and three additional representative genes TFF1, RASSF7, and POLR3F (Fig. 5g and Supplementary Fig. 8h–k). We next explored clinical relevance of this HNF1B/ERG joint target signature and examined their correlations with CCP scores. This analysis revealed a significant positive linear correlation in multiple independent PCa cohorts (Fig. 5h, i and Supplementary Fig. 8l–o). Furthermore, patients with higher levels of the signature score is associated with shorter overall survival (Fig. 5j, k).

Taken together, those results indicate additional layers of germline-somatic interplays between HNF1B and TMPRSS2-ERG, including their protein-protein interaction, genome-wide chromatin co-localization and a potential synergistic role for the activation of cis-regulatory elements driving gene expression for PCa progression and prognosis.

HNF1B and ERG co-occupied chromatin regions indicate a greater enrichment of germline variants across PCa risk loci including 17p13.3

Given that ERG fusion is a transcriptional regulator of PCa risk locus in the 17q12/HNF1B region and physically interacts and shares chromatin binding sites with HNF1B, we asked whether HNF1B alone or together with ERG exerts the genetic impacts that explain PCa risk associations. Hence, we performed an enrichment analysis of SNPs in high linkage disequilibrium (LD, R² > 0.8) with PCa GWAS variants in ERG unique, ERG and HNF1B common as well as HNF1B unique binding sites, respectively. Strikingly, we found that PCa risk SNPs are more likely to be enriched in the common binding sites of TMPRSS2-ERG and HNF1B compared to their unique DNA-binding regions (Fig. 6a). We next examined a genotype-gene expression association of these enriched SNPs from three sources of eQTL datasets, including GTEx⁶⁸, PancanQTL⁶⁹ and ncRNA-eQTL⁷⁰, and jointly identified 17 eQTL genes (eGenes) for five PCa risk loci (Fig. 6b). We generated a HNF1B & ERG eGene signature and explored its potential prognostic value in the clinical PCa settings. This analysis showed an elevated risk for biochemical relapse and metastasis in the patient groups with intermediate Gleason score 7 while having higher eGene signature scores (Fig. 6c, d) but not in PCa patients with Gleason score 6 or 8 (Supplementary Fig. 9), suggesting that this eGene signature may serve as molecular stratifier for PCa patients with intermediate risk disease.

**Fig. 6: PCa susceptibility alleles are highly enriched in HNF1B and ERG co-occupied chromatin regions.**

We and others have shown that transcription factors can bind to SNP regions, and often regulate SNP-linked eQTL genes which they are co-expressed^4,46,60. We thus examined the expression pattern of the 17 eGenes in two independent cohorts of 266 or 118 PCa patients^71,72 and found that most of the eGenes indicated high degree of co-expression with HNF1B and ERG, and interestingly, FAM57A together with GEMIN4 and VPS53 at the 17p13.3 PCa risk locus consistently displayed the strongest co-expression with both HNF1B and ERG (Fig. 6e and Supplementary Fig. 10a). We then focused on the 17p13.3/rs684232 locus that has been reproducibly reported to be strongly associated with PCa susceptibility^12,73 whereas the underlying biology and functional mechanisms remain elusive.

In agreement with an enrichment of the 17p13.3 locus variants in HNF1B and TMPRSS2-ERG common binding sites, ChIP-seq profiles in ERG fusion-positive VCaP cells display multiple strong binding peaks of HNF1B and TMPRSS2-ERG at the 17p13.3/rs684232 region (Fig. 6f). These chromatin occupancies were further validated through ChIP-qPCR assays (Fig. 6g and Supplementary Fig. 10b). To examine whether variation at rs684232 directly influenced DNA-binding motifs of any transcription factors, we employed bioinformatics analysis using the enhancer element locator tool⁷⁴ and a computing program namely affinity testing for regulatory SNP (atSNP)⁷⁵ with collected ETS and HNF1B DNA-binding motifs from previous studies^44,76,77. This analysis showed that the rs684232/17p13.3 appears to alter the HNF1B-ERG heterodimer binding sites (lower panel, Fig. 6f). Given that the rs684232/17p13.3 locus SNPs are functionally mediating HNF1B and ERG in transcriptional regulation, we next examined whether these variants are enriched or exclusive to the TMPRSS2-ERG positive tumors in PCa patients. However, the result indicated no specific enrichment of risk alleles the patients with TMPRSS2-ERG positive and –negative tumors (Supplementary Fig. 10c). Based on these findings, we hypothesized that HNF1B and ERG could functionally influence the expression of 17p13.3 locus genes. We thus performed siRNA-mediated knockdown of ERG in VCaP and LNCaP cells treated with DHT and CRISPR-Cas9-mediated genomic deletion of HNF1B in the LNCaP-derived castration resistant cell model V16A⁷⁸, and found downregulation of VPS53, FAM57A and GEMIN4 mRNA levels (Fig. 6h, i). In contrast, we did not observe expression changes of the 17p13.3 locus genes in a successful siRNA-mediated knockdown experiment in the ERG fusion-negative control cell line LNCaP (Supplementary Fig. 10d). Furthermore, through querying multiple clinical PCa datasets, we found a positive expression correlation between FAM57A and ERG or HNF1B in prostate tumors (Fig. 6j, k), but not in adjacent normal prostate tissues (Supplementary Fig. 10e, f), indicating ERG and HNF1B directed transcriptional reprogramming in human PCa tumorigenesis. ERG has been reported to be highly expressed in TMPRSS2-ERG fusion-positive PCa specimens³¹. In line with this, we found that ERG expression levels were significantly higher in ERG fusion-positive group compared to ERG fusion-negative PCa patient group (Supplementary Fig. 10g, h). We next examined whether the expression levels of FAM57A and HNF1B are dependent on ERG fusion status in the clinical PCa cohorts. As expected, the mRNA levels of HNF1B and FAM57A are greatly higher in ERG fusion-positive groups (Fig. 6l, m and Supplementary Fig. 10i, j). Given that a cluster of cis-regulatory elements were reported to regulate elevated expression of TMPRSS2-ERG⁷⁹ and that TMPRSS2-ERG and HNF1B co-bound at a broad H3K27ac-marked super-enhancer region in HNF1B (Fig. 3d) and at the rs684232/17p13.3 region (Fig. 6f), we investigated whether known transcription inhibitors could perturb this transcriptional misregulation circuit that drives the expression of several PCa risk-associated genes. Thus, we treated the PCa cell line V16A with different concentration of BET inhibitor that is known to target a family of BRD proteins highly activated in PCa and playing roles in the control of cell-cycle-associated genes^80,81. The results showed marked reduction of HNF1B, FAM57A, and GEMIN4 expression (Supplementary Fig. 10k), indicating a translational potential of converged somatic ERG fusion, germline loci 17p13.3/rs684232 and 17q12/HNF1B in the same clinical setting.

Multiple causal variants and causative genes at 17p13.3 implicating PCa susceptibility

We next examined an enhancer-like function of the 17p13.3 region harboring the SNPs rs2955626, rs684232 and rs461251 for PCa susceptibility genes VPS53, FAM57A and GEMIN4. Hence, we proceeded to test whether these SNPs could directly alter the promoter activity of VPS53, FAM57A and GEMIN4, respectively. We inserted the SNP-containing regions into the upstream of VPS53, FAM57A or GEMIN4 promoter in pGL4.10-basic vector, except for rs2955626 already within the promoter region of VPS53, and performed enhancer reporter assays in VCaP cells with ETH and DHT treatment. The results showed that SNP-containing regions possess an enhancer activity towards VPS53, FAM57A and GEMIN4. Notably, the rs684232 and rs461251-containing regions have an increased enhancer activity when VCaP cells were treated with DHT compared to that of ETH treatment (Fig. 7a). To investigate whether there are direct chromatin interactions between rs2955626/rs684232/rs461251-containing enhancer and proximal regulatory regions of VPS53, FAM57A and GEMIN4, we performed quantitative chromosome conformation capture assays (3C-qPCR)⁸² with the restriction enzyme HindIII. The constant fragment was designed between the two binding sites of HNF1B and ERG with the SNPs rs2955626, rs684232 and rs461251. We determined its interaction with HindIII-digested chromatin fragments in a 45-kb region covering VPS53, FAM57A and GEMIN4 promoter regions in VCaP cells with ETH or DHT treatment and the lung cancer cell line A549. The results showed that the SNP regions have higher crosslinking frequencies with FAM57A in DHT-treated VCaP compared to ETH-treated VCaP or lung cancer cells A549, suggesting an apparent impact on the observed interactions between the SNP enhancers and FAM57A upon androgen stimulation (Fig. 7b).

**Fig. 7: Effects of the 17p13.3 locus PCa susceptibility alleles on *VPS53*, *FAM57A* and *GEMIN4*.**

To further verify if rs2955626, rs684232 and rs461251 are directly involved in the regulation of VPS53, FAM57A and GEMIN4, we applied the CRISPR/Cas9 genome editing approach in PCa cell line V16A and obtained four independent mixed cell clones with partial depletion of each SNP enhancer region at 17p13.3. The results showed reduced transcriptional levels of VPS53, FAM57A and GEMIN4 upon deletion of the regulatory region harboring rs2955626, rs684232 or rs461251 compared to the parental V16A line (Fig. 7c), suggesting a causal function of rs2955626, rs684232 and rs461251 for the expression of VPS53, FAM57A and GEMIN4.

Finally, to test the biological relevance of the 17p13.3 locus genes, we pursued tumor cellular assays to demonstrate the effect of VPS53, FAM57A and GEMIN4 on PCa cellular phenotypes. The PCa 22Rv1 cells harboring shRNAs against VPS53, FAM57A or GEMIN4 showed markedly attenuated cell growth and viability in comparison with cells harboring control shRNA in the proliferation assays (Fig. 7d–g). Consistent results were obtained via a real-time monitoring of wound-healing assays, indicating that 22Rv1 cells harboring gene-specific shRNAs showed decreased wound closure rates (Fig. 7h–j). Collectively, these results support a mechanistic model of multiple causal variants and causative genes implicating PCa susceptibility and tumor cellular transformation at this 17p13.3 locus.

Discussion

In this study, we revealed an extensive germline-somatic interaction implicating PCa susceptibility and progression through the oncogenic regulatory circuits, consisting of the most frequent PCa-specific somatic genomic alteration TMPRSS2-ERG, and several germline PCa risk locus genes such as HNF1B at 17q12 and VPS53-FAM57A-GEMIN4 at 17p13.3 (Fig. 7k). Our findings demonstrated not only multiple potential causal SNPs and risk CREs within 17q12/HNF1B locus, and the responsibility of TMPRSS2-ERG fusion for transforming molecular and biological effects of 17q12 and regulating HNF1B expression, but also mapped genome-wide binding sites of HNF1B and defined its high rate of chromatin co-occupancy with TMPRSS2-ERG that can explain more of genetic associations discovered by GWASs in PCa. Given that the 17q12/HNF1B has been reported as a cross-cancer pleiotropic genetic risk locus^{14,15,16,17,18,19,20,21,22,23,24,25,26}, our work revealed an understanding of its underlying causation and biological mechanisms implicating in PCa risk prediction and prognosis, while exhibited as an example for the comprehensive evaluation of this locus contributing to risk association and disease progression in other types of cancers.

Here we found HNF1B, together with MYC, FOXA1, HOXB13 and AR, as one of the most essential genes for PCa cell survival and its elevated expression in PCa tumors accompanied with higher grade and Gleason score. Despite the fact that we did not observe significant association between HNF1B expression and patient prognosis, which might be due to the limited sample size in current available data sets, we expect to detect strong association with the availability of large patient cohorts in the future. HNF1B has three isoforms through alternative splicing and has been shown that the two longer isoforms tend to be transcriptional activators, whereas the shortest isoform is a transcriptional repressor⁸³. We proved the relationship between HNF1B isoforms and the 17q12 PCa risk SNPs by which an eQTL analysis indicates a strong association between 11 SNPs and HNF1B isoform 1, the dominant isoform expressed in prostate specimens and other tissues. Moreover, we discovered the role of HNF1B in regulating genes involved in cell cycle progression pathways implicated in PCa progression and clinical severity. It is noteworthy that uncontrolled cell growth and division are a cardinal feature of cancer cells⁸⁴, and accordingly, many inhibitory drugs have been developed in clinical use to target cell cycle progression for cancer therapy⁸⁵. For example, CDK4/6 inhibitors targeted the G1/S phase cell-cycle pathway were recently applied and evaluated in the clinical trial for advanced PCa⁸⁶. Here we show that risk genotypes of the 17q12 PCa risk SNPs are associated with an elevated expression of HNF1B, thereby contributing to PCa cell proliferation and tumor progression. Thus, the findings may raise an important topic for future study through applying cell cycle pathway inhibitors for the PCa patients with high HNF1B expressing tumors.

Despite the fact that TMPRSS2-ERG has been reported as the most frequent genomic rearrangement in PCa and shown profound roles in PCa initiation and development, for instance through cooperating with activated PI3K/AKT pathway or PTEN loss^42,43, all these examples are somatic interaction events. Here our study reported a germline-somatic association of TMPRSS2-ERG with this cross-cancer locus of 17q12/HNF1B, mechanistically and biologically implicating PCa risk and progression. Given that TMPRSS2-ERG fusion is a somatic event specific to PCa^{31,32,33,34,35,36}, additional transcription regulators remain to be uncovered to explore underlying mechanisms of the 17q12/HNF1B locus associated with other types of cancers^{22,23,24,25,26}. Here our results also highlight that the positive correlations of HNF1B expression with the 17q12 SNP genotypes are dependent on ERG fusion status. Correspondingly, we observe a moderate enrichment of three HNF1B SNPs in PCa patients with tumors expressing ERG proteins though there are obvious limitations due to small sample size and IHC-stained ERG expression as a surrogate for ERG fusion. We thus expect larger studies with confirmative ERG fusion status might enable the detection of an apparent enrichment of the genotype of SNPs in HNF1B with ERG fusion-positive tumors in humans in the future. Future studies may also test synergistic effects of 17q12 SNP genotypes and HNF1B expression in predicting PCa clinical outcomes similar as previously reported⁶⁰.

We also show a physical protein-protein interaction between HNF1B and TMPRSS2-ERG, which in turn co-occupy a large fraction of chromatin regions enriched with various PCa risk-associated non-coding genomic variants, including the SNPs at the 17p13.3 PCa susceptibility locus. Previous studies reported the regulation of VPS53, FAM57A and GEMIN4 within the 17p13.3 regions through regulatory enhancer region harboring SNPs rs2955626 and rs684232 wherein ERG functions as a transcriptional mediator of those CREs^4,87. Here our observations support those findings and freshly pinpoint the involvement of HNF1B in the regulation of VPS53, FAM57A and GEMIN4 through a synergistic cooperation with TMPRSS2-ERG where ChIP-seq data showed occupancy of common binding sites in the SNP-surrounding enhancer regions with rs2955626, rs684232 and rs461251. All the SNP regions showed enhancer activity for each gene at 17p13.3 locus but strikingly, we observed an increased enhancer activity on FAM57A proven by the CRISPR/Cas9-mediated genome editing, and also by enhancer report assay and quantitative analysis via chromosome conformation capture assays. Finally, we tested the biological relevance of VPS53, FAM57A and GEMIN4 expression in PCa cells and demonstrated that knockdown of these genes reduces cell growth and migration. These findings may possess translational application to benefit patients in the future. We show that AR inhibitor enzalutamide attenuates protein-protein interaction between ERG and HNF1B. We also show that BET inhibitor treatment can markedly reduce the expression of HNF1B, FAM57A, and GEMIN4. Together with recent identification of a small molecule selectively inhibiting ERG-positive cancer cells⁸⁸, we expect additive effects among these drugs to inhibit the interaction and expression of ERG and HNF1B or their target genes thereby affecting PCa cell growth in the clinical setting as an important topic for future studies.

In summary, our results provide mechanistic insight into how the PCa risk locus 17q12/HNF1B contributes to disease severity and progression through the germline-somatic interplay between HNF1B and TMPRSS2-ERG with a potential for transcriptionally mediating more genetic variance underpinning PCa susceptibility.

Methods

The presented study complies with all relevant ethical regulations and was approved by the University of Oulu and the Shanghai Jiao Tong University School of Medicine Affiliated Ruijin Hospital. All participating patients provided written informed consent. Patients were not monetarily compensated.

Cell culture

The cell lines used in the work (Supplementary Table 4) are 22Rv1 (CRL-2505, ATCC), LNCaP (CRL-1740, ATCC), VCaP (CRL-2876, ATCC), DU145 (HTB-81, ATCC), PC3 (CRL-1435, ATCC), V16A⁷⁸, A549 (CCL-185, ATCC), RWPE1 (CRL-11609, ATCC), and 293 T (CRL-11268, ATCC). All cell lines were confirmed to be mycoplasma free during our study. As described above, most of cells lines were originally purchased from ATCC (American Type Culture Collection). Cell morphology and growth rate of the cell lines used in this study were similar to previous reports. These cell lines have been authenticated by STR fingerprinting. The cells were cultured under the conditions of 37 °C and 5% CO₂. VCaP and 293 T were grown in Dulbecco’s Modified Eagle’s Medium (DMEM) (11965092, Thermo FisherInvitrogen), for culturing DU145 we used Eagle’s Minimum Essential Medium (EMEM) (30-2003, ATCC). Also, LNCaP, 22Rv1 and V16A were grown in Roswell Park Memorial Institute Medium (RPMI 1640) (R8758, Sigma) and finally A549 and PC3 was grown in F12-K (30-2004, Invitrogen). The cell culture media were supplied with a final concentration of 10% fetal bovine serum (16000044, Thermo Fisher) and 1% of penicillin and streptomycin (15140122, Thermo Fisher)). RWPE1 cells were grown in Keratinocyte-Serum Free Medium. Keratinocyte-SFM Kit including epidermal growth factor (EGF) and bovine pituitary extract (BPE) supplements were purchased from Invitrogen (17005-042, Invitrogen). The VCaP cells were cultured in the charcoal-stripped media wherein activating the androgen receptor signaling through a dihydrotestosterone (DHT) (Olli A. Jänne lab, University of Helsinki) treatment with final concentration of 100 nM for 24 h.

Plasmids and gene cloning

Human cDNA library was used to amplify HNF1B open reading frame (ORF) that was cloned into pLVET-IRES-GFP and pcDNA3.1 vectors (Supplementary Table 9). Wild type ERG was also amplified from the same library and cloned into pcDNA3.1. The cDNA of TMPRSS2-ERG fusion was cloned from VCaP into pcDNA3.1. HNF1B-D (domains of full-length HNF1B NM_000458), HNF1B-POU and HNF1B-T of HNF1B sub-domains were cloned into pcDNA3.1, respectively, to express three recombinant proteins for testing protein interaction between HNF1B and TMPRSS2-ERG at domain levels. Primer sequences, cloning methods and enzymes are shown in Supplementary Tables 5 and 15.

Construction of reporter plasmids

Each enhancer or promoter region was amplified from human genomic DNA and cloned into pGL4.10 [luc2] (E6651, Promega) (Supplementary Table 9) containing a SNP (rs718960, rs7405696, rs11651052, rs9901746, rs11263763 or rs12453443) region. Each of these six SNPs was cloned with two different alleles obtained by site-directed mutagenesis. In addition, three SNP (rs2955626 or rs461251 and rs684232) regions were cloned. The enhancers were cloned into the BamHI site (in both orientations), and the promoters of HNF1B, VPS53, FAM57A or GEMIN4 into the EcoRV/HindIII sites of pGL4.10 [luc2] vector, respectively. Both orientations can facilitate testing enhancer activity of SNP-containing regions regardless of the promoter location. The constructs were transient, reversely transfected into LNCaP or VCaP (treated with DHT or ETH) cells with a Renilla Luciferase control plasmid pGL4.75 [hRluc/CMV] (E6931, Promega) by using X-treme GENE HP DNA Transfection Reagent (06366236001, Roche). The experiments were performed on the 96-well white plates with each well containing 100 μl medium of 3 × 10⁵ 22Rv1 and LNCaP cells/ml or 9 × 10⁵ VCaP cells/ml. After incubation at 5% CO₂ and 37 °C for 48 h, the luciferase activity was measured with Dual-Glo Luciferase Assay System (E2940, Promega). At least three replicate wells were used per construct and the data were statistically analyzed with a two-tailed Student’s t test. Primer sequences, cloning methods and enzymes are shown in Supplementary Tables 5 and 15.

Protein blot analysis

Cell pellet was resuspended in lysis buffer (600 mM Nacl, 1% Triton X-100 in PBS, freshly added 1 x protease inhibitor) and sonicated (Q800R sonicator, Q Sonica). The sample was centrifuged, and the supernatant was collected. The amount of protein was measured with Pierce BCA Protein Assay Kit (23225, Thermo Fisher Scientific) based on the manufacturer’s protocol and 30 μg of protein lysate of each sample was separated by electrophoresis in 7.5% or 12% SDS-PAGE gel and transferred into 0.45 μm Immobilon-P PVDF Membrane (IPVH00010, Millipore) using a Semi-Dry transfer cell (Trans-Blot SD, Bio-Rad). After transfer, the membrane was blocked for minimum 30 min at room temperature using blocking buffer (5% nonfat milk in TBST) while gently shaking. The blocked membrane then was incubated with antibody diluted in blocking buffer (1:1000 (Ab μl: blocking buffer μl): rabbit polyclonal anti-HNF1B, mouse monoclonal anti-HNF-1B and mouse monoclonal anti-FLAG. 1:5000: mouse monoclonal anti-V5, mouse monoclonal anti-V5-HRP, mouse monoclonal anti-ERG, rabbit monoclonal anti-ERG) at 4 °C for 16 h with gentle rotation. After incubation, the membrane was washed three times each 10 min using TBST. Anti-rabbit IgG or anti-mouse IgG was used as secondary antibody (Thermo Fisher) with 1:5000 dilution into blocking buffer and the incubation took place on a rotor at room temperature for 1 h. Afterwards, the membrane was washed three times each with 15 min using TBST. Finally, the membrane was developed with Lumi-Light Western Blotting Substrate (12015200001, Roche) or SuperSignal West Femto Maximum Sensitivity Substrate (34095, Thermo Fisher Scientific) according to the protocol and exposed with Fujifilm LAS-3000 Imager. Original blots are provided in the Source data file. For more information about the antibodies see Supplementary Table 6.

Ectopic expression via transient transfection

293 T cells were used for transient transfection with pcDNA3.1 constructs (Supplementary Tables 9, 14). Mixer A of pcDNA3.1 construct and P3000 reagent (L3000015, Thermo Fisher Scientific) was diluted with Opti-MEM. Mixer B of lipofectamine 3000 reagent (L3000015, Thermo Fisher Scientific) was diluted with Opti-MEM (11058021, Thermo Fisher). We mix A and B which were incubated at room temperature for 15 min and added into 70-80% confluent seeded cells. The cells were incubated 24-48 h before harvesting.

Co-immunoprecipitation

Co-Immunoprecipitation was performed for examining the endogenous interaction of HNF1B with TMPRSS2-ERG in VCaP with DHT treatment. The ectopic interaction of HNF1B or HNF1B domains with ERG cloned into pcDNA3.1 and transfected in 293 T (cloning primers listed in Supplementary Table 5). Cells were harvested and lysed with 0.5 ml cold immunoprecipitation buffer (50 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1% Triton X-100, 10% Glycerol, 1 mM EDTA and 1% protease inhibitor cocktail). Keep Cell lysates on ice for 30 min with few vortex periods in between. Sonicate in 4 °C water bath for 20 s. Centrifuge for 30 min at 4 °C and supernatant was incubated with 30 μl protein-G Magnetic Beads to pre-clear crude cell extract of proteins which can bind non-specifically to the beads at 4 °C for 1 h. Keep the supernatant and add 5 μg of antibody (5 μg: mouse monoclonal anti-V5, rabbit monoclonal IgG and rabbit monoclonal anti-ERG, mouse monoclonal anti-FLAG) (Supplementary Table 6) with incubation at 4 °C overnight. Add 30 μl of fresh protein-G Magnetic Beads and incubate at 4 °C for 6 h followed by washing five times with Immunoprecipitation buffer. Furthermore, resuspend beads in 30 μl of 2 x SDS sample loading buffer and incubate at 95 °C for 5 min and finally use the supernatant on SDS-PAGE gel for electrophoresis separation.

siRNA transfections

Individual set of two siRNAs (Qiagen) against HNF1B or ERG were tested in knockdown efficiency and compared with the siRNA negative-control (Qiagen) by RT-qPCR. For cell proliferation assays, we used the same set of two siRNAs (Qiagen) against HNF1B compared with negative and positive control siRNA (Qiagen). 8 × 10⁵ of VCaP and LNCaP cells were used in reverse transfection for 6-well plate, respectively. siRNA transfection was performed with HiPerFect Transfection Reagent (301705, Qiagen) with a final concentration of 50 nM siRNA (see Supplementary Table 7 for the siRNAs used).

RNA isolation and real time quantitative PCR

RNeasy Mini Kit (74106, QIAGEN) was applied for RNA isolation and RNase-Free DNase (79254, QIAGEN) was used during the isolation to remove DNA. cDNA was synthesized from 2 μg RNA by either the High-Capacity cDNA Reverse Transcription Kit (4368814, Applied Biosystems) or the iScript Reverse Transcription Supermix (1708840, Bio-Rad). After cDNA synthesis we used SYBR Select Master Mix (4472920, Applied Biosystems) and high specificity primers for the quantitative RT-PCR reactions. The results were normalized with beta-actin control and for each gene’s analysis made triplicates. Primers used for RT-qPCR in Supplementary Table 8.

CRISPR/Cas9-mediated genome editing analysis

CRISPR design tool (http://crispr.mit.edu/or crispor.tefor.net) was used to prepare the pair of oligos (sgRNA-top and sgRNA-bottom) attached in Supplementary Table 8. Most of the experiment was performed according to the previous protocol⁸⁹. For annealing process 1 μl sgRNA-top (100 μM) and 1 μl sgRNA-bottom (100 μM) were mixed with 1 x T4 ligation buffer, 1 μl T4 PNK and 6 μl ddH₂O. The oligos were phosphorylated and annealed in a thermocycler at 37 °C for 30 min followed with 95 °C for 5 min; ramp down to 25 °C with 5 °C/min. Then, the annealed oligos were inserted into pSpCas9 (BB)−2A-Puro and the plasmids were transfected in 22Rv1 and V16A cells with 70-80% confluency. 0.6 μg of total amount of Cas9 plasmids designed for the same SNP region, with 1:1 ratio or 1:1:1:1 ratio, which added into cells by using Lipofectamine 3000 in 24-well plate according to the protocol. Medium was changed after 24 h and replaced with medium containing 1 μg/ml puromycin (P9620, Merck). Afterwards, the successfully transfected cells were isolated to single cells by dilution or FACS. The single cells were seeded into 96-well plates and after 2-3 weeks the positive clones were examined further by genotyping and RT-qPCR determination of gene expression.

Chromatin immunoprecipitation (ChIP)

ChIP assay was carried out based mainly on previous study⁶². The cells were cross-linked in a final concentration of 1% formaldehyde in medium for 10 min at room temperature with gently shaking. The final concentration of 125 mM glycine was added to stop the reaction and incubated for minimum 5 min with slight shake. Cells were harvested and the pellet was resuspended in hypotonic lysis buffer (20 mM Tris-Cl, pH 8.0, with 10 mM KCl, 10% glycerol, 2 mM DTT, and freshly added cOmplete protease inhibitor cocktail (04693159001, Roche) and incubated up to 1 h on a rotor at 4 °C. Afterwards, the pellet was washed twice with cold PBS and resuspended in SDS lysis buffer (50 mM Tris-HCl, pH 8.1, with 0.5% SDS, 10 mM EDTA, and freshly added cOmplete Protease Inhibitor). Sonication (Q800R sonicator, Q Sonica) was performed as far as the chromatin had size 250-500 bp. Later, 70 μl of Dynabead protein G (10004D, Invitrogen) were washed twice with blocking buffer (0.5% BSA in IP buffer) and incubated with 8 μg of antibody (Rabbit polyclonal anti- HNF1B, Rabbit polyclonal IgG, Rabbit monoclonal IgG, Mouse polyclonal IgG, Rabbit monoclonal anti-ERG, Anti-rabbit Androgen Receptor, H3K4me1, H3K4me2, H3K4me3 and H3K27ac) (examined antibodies in Supplementary Table 6) in 1 ml of 0.5% BSA in IP buffer (20 mM Tris-HCl, pH8.0, with 2 mM EDTA, 150 mM NaCl, 1% Triton X-100, and freshly added Protease inhibitor cocktail) for 10 h at 4 °C on rotor. After incubation the supernatant was removed, and the sonicated chromatin lysate (200-250 μg) was diluted in 1.3 ml of IP buffer and was added into the beads-antibody complex with incubation at 4 °C for at least 12 hrs on rotor. Afterwards, the beads-antibody complex was washed one time with wash buffer I (20 mM Tris-HCl, pH 8.0, with 2 mM EDTA, 0.1%SDS, 1% Triton X-100, and 150 mM NaCl) and once with buffer II (20 mM Tris-HCl pH, 8.0, with 2 mM EDTA, 0.1% SDS, 1% Triton X-100, and 500 mM NaCl), followed by two times of washing with buffer III (10 mM Tris-HCl, pH 8.0, with 1 mM EDTA, 250 mM LiCl, 1% deoxycholate, and 1% NP-40) and two times with buffer IV (10 mM Tris-HCl, pH 8.0, and 1 mM EDTA). 50 μl of extraction buffer (10 mM Tris-HCl, pH 8.0, 1 mM EDTA, and 1% SDS) were added to extract from the beads the DNA-protein complex by incubating and shaking at 65 °C for 20 min (repeat same step with another 50 μl of extraction buffer). Proteinase K (AM2548, Thermo Fisher Scientific) with final concentration 1 mg/ml and NaCl with final concentration 0.3 M were added into the extracted DNA-protein complex and incubated at shaking heat block for 16 h at 65 °C in 1000 rpm to reverse-crosslink of the protein-DNA interactions. DNA was purified with MinElute PCR Purification Kit (28006, QIAGEN) followed by ChIP-qPCR with primers that targeted DNA binding genome sequences (see Supplementary Table 8). ChIP library was prepared according to manufacturer’s protocol TruSeq Sample Preparation Best Practices and Troubleshooting Guide (Illumina). Finally, the sample were sequenced and analyzed.

Quantitative analysis of chromosome conformation capture assay

Quantitative analysis of chromosome conformation capture assay (3C-qPCR) was performed as described in the Hagege et al. protocol⁸². The primers used for these assays are listed in Supplementary Table 8. The cells were trypsinized and resuspended in PBS with 10% FBS. 1 × 10⁷ cells were cross-linked in PBS with 10% FBS and 1% formaldehyde for 10 min at room temperature. To stop the crosslinking reaction, we added 0.57 ml of 2.5 M glycine (ice cold). The pellets of VCaP with treatment and A549 were resuspended in 5 ml cold lysis buffer and incubate for 13 min on ice. Then we centrifuge at 400 g at 4 °C and remove the supernatant and keep the pelleted nuclei, which were collected by centrifugation and used for digestion. We continue on digestion step of sample with HindIII restriction enzyme to digest chromatin DNA and the digestion efficiency was verified. The digested nuclear lysate was used in the ligation step. After ligation, for purification of DNA we increased the volume of sample to dilute DTT presented in the sample with 7 ml distilled water, 1.5 ml of 2 M sodium acetate pH 5.6 and 35 ml ethanol. After washing pellet with 70% ethanol and dry the pellet, and resuspended it in 150 μl of 10 mM Tris pH 7.5. DNA was desalted with centrifuge filters (Microcon DNA Fast Flow) (MRCF0R100, Millipore). For TaqMan qPCR, we used 1 μl of the 3 C sample (100 ng/μl), 5 μl of Quanti tech probe PCR mix (QIAGEN), 1 μl of Taqman probe (1.5 μM), 1 μl of Test + Constant primer (5 μM) and 2 μl distilled H2O. We performed standard curve of each primer using serial dilution of control template, containing amplified fragments across each of 7 HindIII cut sites and mix them together. Values of intercept and slope from the standard curve were used to evaluate the ligation product using the following equation: Value = 10 (Ct-intercept)/slope. These values were finally normalized to ERCC3 (loading control).

Lentiviral constructs, lentivirus production and infection

HNF1B was cloned into the lentivirus plasmid pLVET-IRES-GFP for ectopic expression (see Supplementary Table 5). Two set of shRNA constructs in the pLKO.1-puro vector targeting HNF1B, VPS53, FAM57A or GEMIN4 (Merck) were applied for knockdown assays. More information on the shRNA constructs can be found in Supplementary Table 7. Lentiviral constructs were produced with the third-generation packaging system in human embryonic kidney (HEK) 293 T cells (ATCC, CRL-11268) which were seeded the previous day into 3.5-cm plate in a 70%–80% confluency. At the day of transfection, the medium was replaced with 1 ml low glucose DMEM (Invitrogen) containing 10% FBS, 0.1% penicillin-streptomycin. A mix of four plasmids was made in a ratio 1:1:1:3 in a total amount of 10 μg (pVSVG-envelope plasmid, pMDLg/pRRE-packaging plasmid, pRSV-Rev-packaging plasmid and lentiviral transfer vector) (Supplementary Table 9) and diluted in Opti-MEM with Lipofectamine 2000. 24 h later the medium was replaced with 2 ml fresh medium. After that time, the virus-containing medium was collected every 24 h for 3 d and then was centrifuged at 95 g for 5 min and the supernatant was filtered with 0.45 μm filter unit place on syringe. Then the sample was collected and frozen with liquid nitrogen before stored at −80 °C. For virus transduction into the desired cells seeded 24 h before transduction in 3.5 cm plate, the final concentration of 8 μg/ml polybrene (Sigma) was added in 1.4 ml medium and 0.6 ml lentivirus-containing medium. Then the culture medium of target cells was replaced with the above prepared mix and incubate for 24 h at 37 °C and 5% CO₂. In case of puromycin (Sigma) selection construct, after 24 h the medium was replaced with pre-warmed medium, and 48 h after transduction the medium was changed with fresh medium containing puromycin in a final concentration of 2 μg/ml. Cells without virus transduction were used as control to determine cell survival status upon puromycin selection. For the GFP expression constructs, 48 h after transduction the cells were sorted positively by fluorescence activated cell sorting (FACS) using BD FACS Aria flow cytometer (BD Biosciences).

Cell proliferation assays

The experiments were performed on the 96-well plates with each well containing 100 μl medium of 2 × 10³ V16A, PC3, DU145, RWPE1 or 22Rv1 cells and 8 × 10³ VCaP cells per well, respectively, in an incubation period of 4-6 d with 5% CO₂ and 37 °C. The Cell Proliferation Kit II XTT (11465015001, Roche) was used according to the manufacturer. Cell proliferation was examined at indicated time points by XTT colorimetric assay (absorbance at 450 nm). At least three replicate wells were prepared per condition and the data were statistically analyzed with a two-tailed Student’s t test. The information on critical commercial assays can be available in Supplementary Table 12.

Wound healing assays

Cells were seeded into 96-well imageLock plates with the appropriate culture medium that can allow to grow near 100% confluence. Then we used WoundMaker tool to create homogenous scratch wounds and cells were washed twice with PBS. Culture medium was added into each well. The wound areas of each well were imaged every 2 h for max 180 h using Essen BioScience IncuCyte Live-Cell Imaging System.

Gene expression correlation analysis

We performed the co-expression analysis to evaluate the expression correlation between HNF1B, ERG and FAM57A from multiple independent cohorts with benign and cancerous prostate tissues. The co-expression tests were also applied in scenarios considering TMPRSS2-ERG status. Both Pearson’s product-moment correlation and Spearman’s rank correlation rho methods were applied in all linear expression correlation tests. Genes were ranked according to Pearson coefficient value in a descending order to identify the gene that is most co-expressed with HNF1B in a genome-wise scale.

Survival analysis

Survival analysis was applied to assess the impact of HNF1B cell cycle signature, ERG & HNF1B target gene signature and ERG & HNF1B eGene signature on PCa prognosis and survival in multiple independent cohorts. The survival analyses were performed and visualized as Kaplan-Meier plots using R package “Survival” (v.3.2.3)^90,91. Patients were stratified into two groups based on the median value of the z-score summed signature scores. Function “Surv” was first employed to create the survival models with “time-to-event” and “event status” as input from clinical cohorts. Then signature scores was further followed to fit to the models by function “survfit”. The Cox proportional-hazards model⁹² was applied to investigate the hazard ratio for assessing the association between patients’ survival time and gene expression or signature scores.

Expression quantitative trait loci (eQTL) analysis

To evaluate the associations between genotypes of SNPs and HNF1B expression level, we performed the expression quantitative trait loci (eQTL) analysis by R package “MatrixEQTL” v.2.2⁹³ in Wisconsin and TCGA cohorts^94,95, which comprised of 466 normal and 389 prostate tumor samples, respectively. To examine whether T2E fusion affects the eQTL signal, we matched available T2E information to the existing TCGA cohort and further stratified patients into fusion-positive and -negative PCa tumors consisting of 160 and 228 samples, respectively. The eQTL analysis was applied by fitting a linear regression model between the expression and the genotype data, other parameters were left as default (pvOutputThreshold = 0.05, errorCovariance = numeric ()”). The transcriptional profiling in TCGA cohort was assessed by RNA-Seq. The TCGA cohort was genotyped on Affymetrix SNP array 6. The relevant SNP genomic locations are listed in Supplementary Table 13.

Enrichment analysis of HNF1Β SNPs in ERG-positive PCa tumors

To investigate whether the SNPs were associated with TMPRSS2-ERG fusion-positive tumors in PCa, we performed an independent association study in a Chinese prostate biopsy cohort⁹⁶ and radical prostatectomy cohort. Briefly, a consecutive prostate biopsy cohort and prostatectomy cohort with biospecimen started from October 2017 to December 2021 at a tertiary hospital in Shanghai, China. ERG was regularly stained in biopsy tissue samples via IHC. IHC was performed on the 4-μm-thick FFPE tissue sections using commercially available antibodies against ERG (Agilent Technologies Singapore). Antibody staining was detected using a universal immunoperoxidase polymer method (Envision-kit; Dako, Carpinteria, CA, US). A Dako automated immunohistochemistry system (Dako, Carpinteria, CA, US) was used according to the manufacturer’s protocol. Likewise, the IHC results were independently interpreted by two experienced pathologists: Xiaoqun Yang and Chaofu Wang. Genotyping was performed in most of the samples using Illumina Asian Screening Array. Imputation was performed thereafter⁹⁷. A posterior probability of >0.90 was applied to call genotypes during imputation and the same quality control procedure for excluding genotyped SNPs was applied to imputed SNPs. Genotyping data of HNF1B with a ± 100 kb window was extracted (n = 1,662). SNPs were excluded if they had: (1) genotype call rate < 90% (n = 1,335); (2) minor allele frequency (MAF) <0.01 (n = 53); or (3) p < .05 for the Hardy–Weinberg Equilibrium (HWE) test. The study was approved by the Institutional Review Board (IRB) of Ruijin Hospital, Shanghai, China. A total of 1,543 samples with ERG expression information were found to have genotyping data (October 2017-December 2021). The demographic characteristics of these patients are shown in Supplementary Table 1. Assuming that positive ERG expression based on IHC is due to ERG fusion, 136 ERG-positive PCa cases out of 791 (17.2%) cases were observed in this cohort (30 ERG-positive biopsies negative to PCa to be excluded). This frequency was similar to the reported ERG fusion frequencies in Asian population^33,34,35,36.

RNA-sequencing (RNA-seq) and differential expression analysis

Preparation of RNA samples was made with VCaP cells, which were treated with 100 nM DHT, and reversely transfected with two different siRNAs targeting HNF1B and negative siRNA control and incubated for 72 h at 37 °C each with two biological replicates. For the RNA sequencing in VCaP cells treated with either negative control siRNAs or siRNAs against HNF1B, raw sequence data were first pre-processed with FastQC (v.0.11.4) to assess read quality. SortMeRna was applied to identify and filter rRNA⁹⁸ to limit the rRNA quantity in FastQ files. The filtered data was resubmitted for a QC assessment by FastQC (v.0.11.4) to ensure the validity of the filtering steps. Trimmomatic v.0.39⁹⁹ was employed to process reads for quality trimming and adapter removal with default parameters: TruSeq3-SE.fa:2:30:10 SLIDINGWINDOW: 5:20. A final FastQC (v.0.11.4) run was performed to ensure the success of previous quality control steps. The processed reads were aligned against the human genome assembly hg19 using TopHat2 v.2.1.1¹⁰⁰ with default settings; parameter for library type was set as “fr-firststrand”. HTSeq v.0.11.0 (htseq-count) was employed to quantitate aligned sequencing reads against gene annotation from UCSC and with parameters “-s reverse, –i gene_id”. Differential expression analysis was performed from read count matrix using Bioconductor package DESeq2 v.1.16.1¹⁰¹. Genes with low expressions (<5 cumulative read count across samples) were filtered out before analysis. A threshold of P < 0.05 was applied to generate the differentially expressed gene list. Statistical test was applied to control or treatment to ensure high correlations between biological replicates. Data was normalized using method variance Stabilizing Transformation (VST) and the heatmap presenting differentially expressed genes between siRNA Control and siRNAs HNF1B samples was generated using R package “pheatmap” v.1.0.12. All the relevant software and algorithms are listed in Supplementary Table 10.

Gene set enrichment analysis

We applied Gene Set Enrichment Analysis (GSEA) v.4.0.3 to interpret the RNA-Seq results upon knockdown of HNF1B. The pre-ranked gene list was obtained by calculation of data following formula sign (logFC)*-log(p value), and data were sorted in a descending order. GSEAPreranked test¹⁰² was used to test the enrichment of genes with phenotype in Hallmark gene sets. Parameters were set as follows: Enrichment statistic = “weighted”, Max size (exclude larger sets) = 5000, number of permutations =1000. All other parameters were remained as default. The GSEA enrichment plots were generated using R packages “clusterProfiler” v.3.14.3¹⁰³ and “enrichplot” v.1.12.0¹⁰⁴.

Chromatin immunoprecipitation sequencing (ChIP-seq)

The HNF1B ChIP-seq library was sequenced to generate 35-76 bp single-end reads. The HNF1B and ERG replicates were sequenced and produced 150 bp single-end reads. FastQC (v.0.11.4) was applied to assess the quality of raw data and followed by Trimmomatic v.0.39⁹⁹ for quality control. The trimmed reads were mapped into the human genome assembly hg19 using Bowtie2 v.2.4.4¹⁰⁵. MACS2 v. 2.2.7.1¹⁰⁶ was employed for peak calling using default parameters. HOMER v.4.11¹⁰⁷, UCSC, samtools v.1.9¹⁰⁸, bedtools v.2.27.1¹⁰⁹, deepTools v.3.3.2¹¹⁰ and IGV v.2.4.10 tools were used for peak annotation and generating big wiggle and TDF formats. Bioconductor package ChIPseeker v.1.18.0¹¹¹ was applied to perform downstream peak annotation analysis.

Development of the HNF1B/ERG derived signatures

The HNF1B cell cycle signature, composed of 33 genes, was derived from the five top enriched cell cycle related pathways via GSEA, then further being intersected with the 207-upregulated genes from the RNA-Seq upon HNF1B knockdown. We defined the differentially expressed genes from the RNA-Seq as HNF1B knockdown signature. The HNF1B knock-down upregulated signature score was defined as a z-score sum of the 207 HNF1B upregulated genes by RNA-seq measurement. For the HNF1B and ERG direct target gene signature, we first converted the 207 upregulated gene symbols to Entrez IDs. Bedtools v.2.27.1¹⁰⁹ was used to identify common peaks from HNF1B and ERG ChIP-Seq binding signals. Function “annotatePeaks.pl” from HOMER v.4.11 was applied for annotating HNF1B and ERG common peaks. The 207-upregulated gene list and the gene list from HNF1B and ERG common binding peaks were intersected, and thus resulted in a 51-gene list, defined as HNF1B and ERG direct target gene signature. For the eQTL gene (eGene) signature, we screened 13 proxy SNPs enriched in HNF1B and ERG common binding sites from Haploreg v.4.1¹¹². We then set R² ≥ 0.8 as a threshold, which resulted in a total eight proxy SNPs with 17 corresponding eQTL genes. We defined these 17 genes as HNF1B and ERG eGene signature. Signature scores were calculated as weighted sums of normalized expression of the genes from each signature.

Meta-analysis

The pooled HR was calculated by a fixed effect model¹¹³, as the I² statistic was less than 30% or the fixed effects P value for the I² statistic was greater than 0.10, indicating insignificant heterogeneity across studies¹¹⁴. The meta-analysis for investigation of the association between the HNF1B cell cycle signature and patient prognosis across studies was performed using the “metafor” package v.3.4.0¹¹⁵ in R environment v.4.2.0.

Multivariate analysis

We investigated the association of the PCa patient overall survival and biochemical recurrence with the HNF1B cell cycle signature and clinical variables including age, Gleason score, PSA, tumor stage, ERG-fusion status, seminal vesical status and extraprostatic extension status. The Cox proportional hazards model was applied for to investigate the relation between patient prognosis and the HNF1B cell cycle signature together with a set of covariates described above. Samples were stratified into two groups with higher and lower expression by comparing to the median value of the HNF1B cell cycle signature or by the continuous value of the HNF1B cell cycle signature score.

Statistical analysis and data visualization

All statistical analyses were performed using RStudio^116,117 v.1.2.5033 with R environment v.3.6.3 or unless specified. Statistical analyses were applied across normal prostate, tumor and metastatic tissues from multiple cohorts. Mann–Whitney U test was used for gene expression in clinical cohorts with two groups, while Kruskal-Wallis H test was applied for cohorts having three groups or more. R package “Survival” was applied in all Survival analysis. Statistical analyses for all Kaplan-Meier curves were calculated using log-rank test. HNF1B signature scores were calculated from the z-score sum of panels of gene expression levels. For microarray-based expression profiling, we selected gene probes with lowest p values. Circos maps were generated using Circos (v.0.67)¹¹⁸. Asterisks indicate the significance level (*p < 0.05; **p < 0.01; ***p < 0.005). P value <0.05 was considered to be statistically significant.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Data availability

The publically available GWAS data in PCa used in this study was obtained from the GWAS catalog. The publically available RNA-seq or microarray data including Welsh, Yu (GSE6919), Bittner (GSE2109), Wallace (GSE6956), TCGA, MSKCC, SU2C-PCF, Grasso, DKFZ, GSE62872, Fred Hutchinson CRC, SMMU, Broad/Cornel, CPGEA were retrieved from public databases including cBioPortal for Cancer Genomics^95,119, Oncomine database¹²⁰ and GEO database^121,122. The publically available ERG ChIP-seq profiling data in VCaP cells were obtained from the Cistrome Data Browser^64,123. To capture the comprehensive genome-wide map of ERG chromatin binding sites, we retrieved and merged 15 ERG ChIP-seq publically available datasets in VCaP from the Cistrome Data Browser, and obtained a union peak set of ERG in VCaP cells (Supplementary Table 11; GEO: GSM717395, GSM717396, GSM717397, GSM1193658, GSM1328978, GSM1328980, GSM1328981, GSM2086315, GSM2086314, GSM2086313, GSM2086312, GSM2086311, GSM2086310, GSM2086309, GSM353637)^{37,124,125,126,127}. The publically available data used for the eQTL analyses described in this manuscript were obtained from GTEx portal⁶⁸, PancanQTL⁶⁹ and ncRNA-eQTL⁷⁰. The publically available FinnGen data used in this research are available to qualified researchers and detailed documentation is provided on the FinnGen study website. Both raw and processed data from the RNA-seq profiling upon HNF1B depletion and HNF1B ChIP-seq generated in this study have been deposited in the ENA (European Nucleotide Archive) database under accession codes PRJEB46082 and PRJEB46088, respectively. The raw and processed data from the ChIP-seq of HNF1B and ERG replicates generated in this study are available at ENA under accession code PRJEB49662. Source data are provided with this paper. The remaining data supporting the findings of this study are available within the Article, Supplementary Information or Source Data file. A reporting summary for this article is available as a Supplementary Information file. Source data are provided with this paper.

References

Sud, A., Kinnersley, B. & Houlston, R. S. Genome-wide association studies of cancer: Current insights and future perspectives. Nat. Rev. Cancer https://doi.org/10.1038/nrc.2017.82 (2017).
Khurana, E. et al. Role of non-coding sequence variants in cancer. Nat. Rev. Genet. https://doi.org/10.1038/nrg.2015.17 (2016).
Ahmed, M. et al. CRISPRi screens reveal a DNA methylation-mediated 3D genome dependent causal mechanism in prostate cancer. Nat. Commun. (2021) https://doi.org/10.1038/s41467-021-21867-0.
Whitington, T. et al. Gene regulatory mechanisms underpinning prostate cancer susceptibility. Nat. Genet. (2016) https://doi.org/10.1038/ng.3523.
Deplancke, B., Alpern, D. & Gardeux, V. The genetics of transcription factor DNA binding variation. Cell https://doi.org/10.1016/j.cell.2016.07.012 (2016).
Article PubMed Google Scholar
Cowper-Sallari, R. et al. Breast cancer risk-associated SNPs modulate the affinity of chromatin for FOXA1 and alter gene expression. Nat. Genet. https://doi.org/10.1038/ng.2416 (2012).
Article Google Scholar
Demichelis, F. et al. Identification of functionally active, low frequency copy number variants at 15q21.3 and 12q21.31 associated with prostate cancer risk. Proc. Natl. Acad. Sci. USA https://doi.org/10.1073/pnas.1117405109 (2012).
Article PubMed PubMed Central Google Scholar
Sung, H. et al. Global cancer statistics 2020: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. Ca. Cancer J. Clin. https://doi.org/10.3322/caac.21660 (2021).
Article PubMed Google Scholar
Mucci, L. A. et al. Familial risk and heritability of cancer among twins in nordic countries. JAMA - J. Am. Med. Assoc. https://doi.org/10.1001/jama.2015.17703 (2016).
Article Google Scholar
Schumacher, F. R. et al. Association analyses of more than 140,000 men identify 63 new prostate cancer susceptibility loci. Nat. Genet. https://doi.org/10.1038/s41588-018-0142-8 (2018).
Article PubMed PubMed Central Google Scholar
Benafif, S., Kote-Jarai, Z. & Eeles, R. A. A review of prostate cancer Genome-Wide Association Studies (GWAS). Cancer Epidemiol. Biomark. Prev. https://doi.org/10.1158/1055-9965.EPI-16-1046 (2018).
Article Google Scholar
Conti, D. V. et al. Trans-ancestry genome-wide association meta-analysis of prostate cancer identifies new susceptibility loci and informs genetic risk prediction. Nat. Genet. https://doi.org/10.1038/s41588-020-00748-0 (2021).
Article PubMed PubMed Central Google Scholar
Saunders, E. J., Kote‐jarai, Z. & Eeles, R. A. Identification of germline genetic variants that increase prostate cancer risk and influence development of aggressive disease. Cancers https://doi.org/10.3390/cancers13040760 (2021).
Article PubMed PubMed Central Google Scholar
Gudmundsson, J. et al. Two variants on chromosome 17 confer prostate cancer risk, and the one in TCF2 protects against type 2 diabetes. Nat. Genet. https://doi.org/10.1038/ng2062 (2007).
Article PubMed Google Scholar
Gudmundsson, J. et al. Genome-wide association and replication studies identify four variants associated with prostate cancer susceptibility. Nat. Genet. https://doi.org/10.1038/ng.448 (2009).
Article PubMed PubMed Central Google Scholar
Takata, R. et al. Genome-wide association study identifies five new susceptibility loci for prostate cancer in the Japanese population. Nat. Genet. https://doi.org/10.1038/ng.635 (2010).
Article PubMed Google Scholar
Sun, J. et al. Evidence for two independent prostate cancer risk-associated loci in the HNF1B gene at 17q12. Nat. Genet. https://doi.org/10.1038/ng.214 (2008).
Article PubMed PubMed Central Google Scholar
Thomas, G. et al. Multiple loci identified in a genome-wide association study of prostate cancer. Nat. Genet. https://doi.org/10.1038/ng.91 (2008).
Article PubMed PubMed Central Google Scholar
Schumacher, F. R. et al. Genome-wide association study identifies new prostate cancer susceptibility loci. Hum. Mol. Genet. https://doi.org/10.1093/hmg/ddr295 (2011).
Article PubMed PubMed Central Google Scholar
Eeles, R. A. et al. Identification of seven new prostate cancer susceptibility loci through a genome-wide association study. Nat. Genet. https://doi.org/10.1038/ng.450 (2009).
Article PubMed PubMed Central Google Scholar
Eeles, R. A. et al. Multiple newly identified loci associated with prostate cancer susceptibility. Nat. Genet. https://doi.org/10.1038/ng.90 (2008).
Article PubMed PubMed Central Google Scholar
Klein, A. P. et al. Genome-wide meta-analysis identifies five new susceptibility loci for pancreatic cancer. Nat. Commun. https://doi.org/10.1038/s41467-018-02942-5 (2018).
Article PubMed PubMed Central Google Scholar
Pharoah, P. D. P. et al. GWAS meta-analysis and replication identifies three new susceptibility loci for ovarian cancer. Nat. Genet. https://doi.org/10.1038/ng.2564 (2013).
Article PubMed PubMed Central Google Scholar
Shen, H. et al. Epigenetic analysis leads to identification of HNF1B as a subtype-specific susceptibility gene for ovarian cancer. Nat. Commun. https://doi.org/10.1038/ncomms2629 (2013).
Article PubMed Google Scholar
Kristiansen, W. et al. Two new loci and gene sets related to sex determination and cancer progression are associated with susceptibility to testicular germ cell tumor. Hum. Mol. Genet. https://doi.org/10.1093/hmg/ddv129 (2015).
Article PubMed Google Scholar
Spurdle, A. B. et al. Genome-wide association study identifies a common variant associated with risk of endometrial cancer. Nat. Genet. https://doi.org/10.1038/ng.812 (2011).
Article PubMed PubMed Central Google Scholar
Mendel, D. B., Hansen, L. P., Graves, M. K., Conley, P. B. & Crabtree, G. R. HNF-1α and HNF-1β (vHNF-1) share dimerization and homeo domains, but not activation domains, and form heterodimers in vitro. Genes Dev. https://doi.org/10.1101/gad.5.6.1042 (1991).
Article PubMed Google Scholar
Kim, L. et al. Clear cell carcinoma of the pancreas: Histopathologic features and a unique biomarker: Hepatocyte nuclear factor-1β. Mod. Pathol. https://doi.org/10.1038/modpathol.2008.95 (2008).
Article PubMed Google Scholar
Silva, T. D. et al. DNA methylation as an epigenetic biomarker in colorectal cancer. Oncol. Lett. 6, 1687–1692 (2013).
Article PubMed PubMed Central CAS Google Scholar
Yamamoto, S. et al. Immunohistochemical detection of hepatocyte nuclear factor 1β in ovarian and endometrial clear-cell adenocarcinomas and nonneoplastic endometrium. Hum. Pathol. https://doi.org/10.1016/j.humpath.2006.12.018 (2007).
Article PubMed Google Scholar
Tomlins, S. A. et al. Recurrent fusion of TMPRSS2 and ETS transcription factor genes in prostate cancer. Sci. (80-.) https://doi.org/10.1126/science.1117679 (2005).
Article Google Scholar
Hermans, K. G. et al. TMPRSS2:ERG fusion by translocation or interstitial deletion is highly relevant in androgen-dependent prostate cancer, but is bypassed in late-stage androgen receptor-negative prostate cancer. Cancer Res. https://doi.org/10.1158/0008-5472.CAN-06-1871 (2006).
Article PubMed Google Scholar
Sedarsky, J., Degon, M., Srivastava, S. & Dobi, A. Ethnicity and ERG frequency in prostate cancer. Nat. Rev. Urol. https://doi.org/10.1038/nrurol.2017.140 (2018).
Article PubMed Google Scholar
Magi-Galluzzi, C. et al. TMPRSS2-ERG gene fusion prevalence and class are significantly different in prostate cancer of Caucasian, African-American and Japanese patients. Prostate https://doi.org/10.1002/pros.21265 (2011).
Article PubMed Google Scholar
Ren, S. et al. Whole-genome and transcriptome sequencing of prostate cancer identify new genetic alterations driving disease progression [Figure presented]. Eur. Urol. https://doi.org/10.1016/j.eururo.2017.08.027 (2018).
Article PubMed Google Scholar
Li, J. et al. A genomic and epigenomic atlas of prostate cancer in Asian populations. Nature https://doi.org/10.1038/s41586-020-2135-x (2020).
Article PubMed PubMed Central Google Scholar
Yu, J. et al. An integrated network of androgen receptor, polycomb, and TMPRSS2-ERG gene fusions in prostate cancer progression. Cancer Cell https://doi.org/10.1016/j.ccr.2010.03.018 (2010).
Article PubMed PubMed Central Google Scholar
Brase, J. C. et al. TMPRSS2-ERG -specific transcriptional modulation is associated with prostate cancer biomarkers and TGF-β signaling. BMC Cancer https://doi.org/10.1186/1471-2407-11-507 (2011).
Article PubMed PubMed Central Google Scholar
Tian, T. V. et al. Identification of novel TMPRSS2:ERG mechanisms in prostate cancer metastasis: Involvement of MMP9 and PLXNA2. Oncogene https://doi.org/10.1038/onc.2013.176 (2014).
Article PubMed Google Scholar
Wang, J. et al. Pleiotropic biological activities of alternatively spliced TMPRSS2/ERG fusion gene transcripts. Cancer Res. https://doi.org/10.1158/0008-5472.CAN-08-1147 (2008).
Article PubMed PubMed Central Google Scholar
Klezovitch, O. et al. A causal role for ERG in neoplastic transformation of prostate epithelium. Proc. Natl. Acad. Sci. USA https://doi.org/10.1073/pnas.0711711105 (2008).
Article PubMed PubMed Central Google Scholar
King, J. C. et al. Cooperativity of TMPRSS2-ERG with PI3-kinase pathway activation in prostate oncogenesis. Nat. Genet. https://doi.org/10.1038/ng.371 (2009).
Article PubMed PubMed Central Google Scholar
Carver, B. S. et al. Aberrant ERG expression cooperates with loss of PTEN to promote cancer progression in the prostate. Nat. Genet. https://doi.org/10.1038/ng.370 (2009).
Article PubMed PubMed Central Google Scholar
Lambert, S. A. et al. The human transcription factors. Cell https://doi.org/10.1016/j.cell.2018.01.029 (2018).
Article PubMed PubMed Central Google Scholar
Wei, G. H., Liu, D. P. & Liang, C. C. Charting gene regulatory networks: Strategies, challenges and perspectives. Biochemical J. https://doi.org/10.1042/BJ20040311 (2004).
Article Google Scholar
Huang, Q. et al. A prostate cancer susceptibility allele at 6q22 increases RFX6 expression by modulating HOXB13 chromatin binding. Nat. Genet. https://doi.org/10.1038/ng.2862 (2014).
Article PubMed PubMed Central Google Scholar
Akamatsu, S. et al. A functional variant in NKX3.1 associated with prostate cancer susceptibility down-regulates NKX3.1 expression. Hum. Mol. Genet. https://doi.org/10.1093/hmg/ddq350 (2010).
Article PubMed Google Scholar
Papachristodoulou, A. et al. NKX3.1 localization to mitochondria suppresses prostate cancer initiation. Cancer Disco. https://doi.org/10.1158/2159-8290.CD-20-1765 (2021).
Article Google Scholar
Mazrooei, P. et al. Cistrome partitioning reveals convergence of somatic mutations and risk variants on master transcription regulators in primary prostate tumors. Cancer Cell https://doi.org/10.1016/j.ccell.2019.10.005 (2019).
Article PubMed Google Scholar
Aguirre, A. J. et al. Genomic copy number dictates a gene-independent cell response to CRISPR/Cas9 targeting. Cancer Disco. https://doi.org/10.1158/2159-8290.CD-16-0154 (2016).
Article Google Scholar
Welsh, J. B. et al. Analysis of gene expression identifies candidate markers and pharmacological targets in prostate cancer. Cancer Res. 61, 5974–8 (2001).
Yu, Y. P. et al. Gene expression alterations in prostate cancer predicting tumor aggression and preceding development of malignancy. J. Clin. Oncol. https://doi.org/10.1200/JCO.2004.05.158 (2004).
Article PubMed Google Scholar
Ross-Adams, H. et al. Integration of copy number and transcriptomics provides risk stratification in prostate cancer: A discovery and validation cohort study. EBioMedicine https://doi.org/10.1016/j.ebiom.2015.07.017 (2015).
Article PubMed PubMed Central Google Scholar
Wallace, T. A. et al. Tumor immunobiological differences in prostate cancer between African-American and European-American men. Cancer Res. https://doi.org/10.1158/0008-5472.CAN-07-2608 (2008).
Article PubMed PubMed Central Google Scholar
Vanaja, D. K., Cheville, J. C., Iturria, S. J. & Young, C. Y. F. Transcriptional silencing of zinc finger protein 185 identified by expression profiling is associated with prostate cancer progression. Cancer Res. 63, 3877–82 (2003).
Taylor, B. S. et al. Integrative genomic profiling of human prostate cancer. Cancer Cell https://doi.org/10.1016/j.ccr.2010.05.026 (2010).
Article PubMed PubMed Central Google Scholar
Zhang, P. et al. High-throughput screening of prostate cancer risk loci by single nucleotide polymorphisms sequencing. Nat. Commun. https://doi.org/10.1038/s41467-018-04451-x (2018).
Article PubMed PubMed Central Google Scholar
Thibodeau, S. N. et al. Identification of candidate genes for prostate cancer-risk SNPs utilizing a normal prostate tissue eQTL data set. Nat. Commun. https://doi.org/10.1038/ncomms9653 (2015).
Article PubMed Google Scholar
Larson, N. B. et al. Comprehensively evaluating cis-regulatory variation in the human prostate transcriptome by using gene-level allele-specific expression. Am. J. Hum. Genet. https://doi.org/10.1016/j.ajhg.2015.04.015 (2015).
Article PubMed PubMed Central Google Scholar
Gao, P. et al. Biology and clinical implications of the 19q13 aggressive prostate cancer susceptibility locus. Cell https://doi.org/10.1016/j.cell.2018.06.003 (2018).
Article PubMed PubMed Central Google Scholar
Mani, R. S. et al. Induced chromosomal proximity and gene fusions in prostate cancer. Science https://doi.org/10.1126/science.1178124 (2009).
Article PubMed PubMed Central Google Scholar
Wei, G. H. et al. Genome-wide analysis of ETS-family DNA-binding in vitro and in vivo. EMBO J. https://doi.org/10.1038/emboj.2010.106 (2010).
Article PubMed PubMed Central Google Scholar
Wang, Q. et al. A hierarchical network of transcription factors governs androgen receptor-dependent prostate cancer growth. Mol. Cell https://doi.org/10.1016/j.molcel.2007.05.041 (2007).
Article PubMed PubMed Central Google Scholar
Zheng, R. et al. Cistrome Data Browser: Expanded datasets and new tools for gene regulatory analysis. Nucleic Acids Res. https://doi.org/10.1093/nar/gky1094 (2019).
Article PubMed PubMed Central Google Scholar
Cuzick, J. et al. Prognostic value of an RNA expression signature derived from cell cycle proliferation genes in patients with prostate cancer: A retrospective study. Lancet Oncol. https://doi.org/10.1016/S1470-2045(10)70295-3 (2011).
Article PubMed PubMed Central Google Scholar
Tierney, J. F., Stewart, L. A., Ghersi, D., Burdett, S. & Sydes, M. R. Practical methods for incorporating summary time-to-event data into meta-analysis. Trials https://doi.org/10.1186/1745-6215-8-16 (2007).
Article PubMed PubMed Central Google Scholar
Wang, J. et al. HNF1B-mediated repression of SLUG is suppressed by EZH2 in aggressive prostate cancer. Oncogene https://doi.org/10.1038/s41388-019-1065-2 (2020).
Article PubMed PubMed Central Google Scholar
Lonsdale, J. et al. The genotype-tissue expression (GTEx) project. Nat. Genet. https://doi.org/10.1038/ng.2653 (2013).
Article Google Scholar
Gong, J. et al. PancanQTL: Systematic identification of cis -eQTLs and trans -eQTLs in 33 cancer types. Nucleic Acids Res. https://doi.org/10.1093/nar/gkx861 (2018).
Article PubMed PubMed Central Google Scholar
Li, J. et al. NcRNA-eQTL: A database to systematically evaluate the effects of SNPs on non-coding RNA expression across cancer types. Nucleic Acids Res. https://doi.org/10.1093/nar/gkz711 (2020).
Article PubMed PubMed Central Google Scholar
Abida, W. et al. Genomic correlates of clinical outcome in advanced prostate cancer. Proc. Natl. Acad. Sci. USA https://doi.org/10.1073/pnas.1902651116 (2019).
Article PubMed PubMed Central Google Scholar
Gerhauser, C. et al. Molecular evolution of early-onset prostate cancer identifies molecular risk markers and clinical trajectories. Cancer Cell https://doi.org/10.1016/j.ccell.2018.10.016 (2018).
Article PubMed PubMed Central Google Scholar
Eeles, R. A. et al. Identification of 23 new prostate cancer susceptibility loci using the iCOGS custom genotyping array. Nat. Genet. https://doi.org/10.1038/ng.2560 (2013).
Article PubMed PubMed Central Google Scholar
Hallikas, O. et al. Genome-wide prediction of mammalian enhancers based on analysis of transcription-factor binding affinity. Cell https://doi.org/10.1016/j.cell.2005.10.042 (2006).
Article PubMed Google Scholar
Zuo, C., Shin, S. & Keleş, S. AtSNP: Transcription factor binding affinity testing for regulatory SNP detection. Bioinformatics https://doi.org/10.1093/bioinformatics/btv328 (2015).
Article PubMed PubMed Central Google Scholar
Jolma, A. et al. DNA-binding specificities of human transcription factors. Cell https://doi.org/10.1016/j.cell.2012.12.009 (2013).
Article PubMed Google Scholar
Diaferia, G. R. et al. Dissection of transcriptional and cis ‐regulatory control of differentiation in human pancreatic cancer. EMBO J. https://doi.org/10.15252/embj.201592404 (2016).
Article PubMed PubMed Central Google Scholar
Hua, J. T. et al. Risk SNP-mediated promoter-enhancer switching drives prostate cancer through lncRNA PCAT19. Cell https://doi.org/10.1016/j.cell.2018.06.014 (2018).
Article PubMed PubMed Central Google Scholar
Kron, K. J. et al. TMPRSS2-ERG fusion co-opts master transcription factors and activates NOTCH signaling in primary prostate cancer. Nat. Genet. https://doi.org/10.1038/ng.3930 (2017).
Article PubMed Google Scholar
Segura, M. F. et al. BRD4 sustains melanoma proliferation and represents a new target for epigenetic therapy. Cancer Res. https://doi.org/10.1158/0008-5472.CAN-13-0122-T (2013).
Article PubMed PubMed Central Google Scholar
LeRoy, G., Rickards, B. & Flint, S. J. The Double Bromodomain Proteins Brd2 and Brd3 Couple Histone Acetylation to Transcription. Mol. Cell https://doi.org/10.1016/j.molcel.2008.01.018 (2008).
Article PubMed PubMed Central Google Scholar
Hagege, H. et al. Quantitative analysis ofchromosome conformation capture assays (3c-qpcr). Nat. Protoc. https://doi.org/10.1038/nprot.2007.243 (2007).
Bach, I. & Yaniv, M. More potent transcriptional activators or a transdominant inhibitor of the HNF1 homeoprotein family are generated by alternative RNA processing. EMBO J. https://doi.org/10.1002/j.1460-2075.1993.tb06107.x (1993).
Article PubMed PubMed Central Google Scholar
Hanahan, D. & Weinberg, R. A. Hallmarks of cancer: The next generation. Cell https://doi.org/10.1016/j.cell.2011.02.013 (2011).
Article PubMed Google Scholar
Sherr, C. J. & Bartek, J. Cell cycle-targeted cancer therapies. Annu. Rev. Cancer Biol. https://doi.org/10.1146/annurev-cancerbio-040716-075628 (2017).
Article Google Scholar
de Kouchkovsky, I. et al. A phase Ib/II study of the CDK 4/6 inhibitor ribociclib in combination with docetaxel plus prednisone in metastatic castration-resistant prostate cancer. Clin. Cancer Res. J. Am. Assoc. Cancer Res. https://doi.org/10.1158/1078-0432.CCR-21-4302 (2022).
Article Google Scholar
Ramanand, S. G. et al. The landscape of RNA polymerase II–associated chromatin interactions in prostate cancer. J. Clin. Invest. https://doi.org/10.1172/JCI134260 (2020).
Article PubMed PubMed Central Google Scholar
Mohamed, A. A. et al. Identification of a small molecule that selectively inhibits ERG-positive cancer cell growth. Cancer Res. https://doi.org/10.1158/0008-5472.CAN-17-2949 (2018).
Article PubMed PubMed Central Google Scholar
Ran, F. A. et al. Genome engineering using the CRISPR-Cas9 system. Nat. Protoc. https://doi.org/10.1038/nprot.2013.143 (2013).
Article PubMed PubMed Central Google Scholar
Therneau, T. M. survival: A Package for Survival Analysis in R. R Packag. version 2.38 (2020).
Therneau, T. M. & Grambsch, P. M. The Cox Model BT—Modeling Survival Data: Extending the Cox Model. in Statistics for Biology and Health (2000).
Cox, D. R. Regression Models and Life-Tables. J. R. Stat. Soc. Ser. B https://doi.org/10.1111/j.2517-6161.1972.tb00899.x (1972).
Article Google Scholar
Shabalin, A. A. Matrix eQTL: Ultra fast eQTL analysis via large matrix operations. Bioinformatics https://doi.org/10.1093/bioinformatics/bts163 (2012).
Article PubMed PubMed Central Google Scholar
Abeshouse, A. et al. The molecular taxonomy of primary prostate cancer. Cell https://doi.org/10.1016/j.cell.2015.10.025 (2015).
Article Google Scholar
Cerami, E. et al. The cBio Cancer Genomics Portal: An open platform for exploring multidimensional cancer genomics data. Cancer Disco. https://doi.org/10.1158/2159-8290.CD-12-0095 (2012).
Article Google Scholar
Huang, D. et al. Genetic polymorphisms at 19q13.33 are associated with [−2]proPSA (p2PSA) levels and provide additional predictive value to prostate health index for prostate cancer. Prostate https://doi.org/10.1002/pros.24192 (2021).
Article PubMed PubMed Central Google Scholar
Das, S. et al. Next-generation genotype imputation service and methods. Nat. Genet. https://doi.org/10.1038/ng.3656 (2016).
Article PubMed PubMed Central Google Scholar
Kopylova, E., Noé, L. & Touzet, H. SortMeRNA: Fast and accurate filtering of ribosomal RNAs in metatranscriptomic data. Bioinformatics https://doi.org/10.1093/bioinformatics/bts611 (2012).
Article PubMed Google Scholar
Bolger, A. M., Lohse, M. & Usadel, B. Trimmomatic: A flexible trimmer for Illumina sequence data. Bioinformatics https://doi.org/10.1093/bioinformatics/btu170 (2014).
Article PubMed PubMed Central Google Scholar
Kim, D. et al. TopHat2: Accurate alignment of transcriptomes in the presence of insertions, deletions and gene fusions. Genome Biol. https://doi.org/10.1186/gb-2013-14-4-r36 (2013).
Article PubMed PubMed Central Google Scholar
Love, M. I., Huber, W. & Anders, S. Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol. https://doi.org/10.1186/s13059-014-0550-8 (2014).
Article PubMed PubMed Central Google Scholar
Subramanian, A. et al. Gene set enrichment analysis: A knowledge-based approach for interpreting genome-wide expression profiles. Proc. Natl. Acad. Sci. USA https://doi.org/10.1073/pnas.0506580102 (2005).
Article PubMed PubMed Central Google Scholar
Yu, G., Wang, L. G., Han, Y. & He, Q. Y. ClusterProfiler: An R package for comparing biological themes among gene clusters. Omi. A J. Integr. Biol. https://doi.org/10.1089/omi.2011.0118 (2012).
Article Google Scholar
Yu, G. enrichplot: Visualization of Functional Enrichment Result. https://yulab-smu.top/biomedical-knowledge-mining-book/ (2021).
Langmead, B. & Salzberg, S. L. Fast gapped-read alignment with Bowtie 2. Nat. Methods https://doi.org/10.1038/nmeth.1923 (2012).
Article PubMed PubMed Central Google Scholar
Model-based analysis ofChIP-Seq (MACS). Genome Biol. 9, R137 https://doi.org/10.1186/gb-2008-9-9-r137 (2008).
Heinz, S. et al. Simple combinations of lineage-determining transcription factors prime cis-regulatory elements required for macrophage and B cell identities. Mol. Cell https://doi.org/10.1016/j.molcel.2010.05.004 (2010).
Article PubMed PubMed Central Google Scholar
Danecek, P. et al. Twelve years of SAMtools and BCFtools. Gigascience 10, (2021).
Quinlan, A. R. & Hall, I. M. BEDTools: A flexible suite of utilities for comparing genomic features. Bioinformatics https://doi.org/10.1093/bioinformatics/btq033 (2010).
Article PubMed PubMed Central Google Scholar
Ramírez, F. et al. deepTools2: a next generation web server for deep-sequencing data analysis. Nucleic Acids Res. 44, W160–W165 (2016).
Article PubMed PubMed Central Google Scholar
Yu, G., Wang, L. G. & He, Q. Y. ChIP seeker: An R/Bioconductor package for ChIP peak annotation, comparison and visualization. Bioinformatics https://doi.org/10.1093/bioinformatics/btv145 (2015).
Article PubMed PubMed Central Google Scholar
Ward, L. D. & Kellis, M. HaploReg: a resource for exploring chromatin states, conservation, and regulatory motif alterations within sets of genetically linked variants. Nucleic Acids Res. 40, D930–D934 (2012).
Article PubMed CAS Google Scholar
Mantel, N. & Haenszel, W. Statistical aspects of the analysis of data from retrospective studies of disease. J. Natl Cancer Inst. https://doi.org/10.1093/jnci/22.4.719 (1959).
Article PubMed Google Scholar
DerSimonian, R. & Laird, N. Meta-analysis in clinical trials. Control. Clin. Trials https://doi.org/10.1016/0197-2456(86)90046-2 (1986).
Article PubMed Google Scholar
Viechtbauer, W. Conducting Meta-analysis in R with the metafor package. J. Stat. Softw. 36, 1–48 (2010).
The R Development Core Team. R: A language and environment for statistical computing. ISBN 3-900051-07-0 (2017).
Rstudio Team. RStudio: Integrated development for R. RStudio, Inc., Boston MA. RStudio at (2019).
Krzywinski, M. et al. Circos: An information aesthetic for comparative genomics. Genome Res. https://doi.org/10.1101/gr.092759.109 (2009).
Article PubMed PubMed Central Google Scholar
Gao, J. et al. Integrative analysis of complex cancer genomics and clinical profiles using the cBioPortal. Sci. Signal. https://doi.org/10.1126/scisignal.2004088 (2013).
Article PubMed PubMed Central Google Scholar
Rhodes, D. R. et al. ONCOMINE: A Cancer Microarray Database and Integrated Data-Mining Platform. Neoplasia https://doi.org/10.1016/s1476-5586(04)80047-2 (2004).
Article PubMed PubMed Central Google Scholar
Edgar, R., Domrachev, M. & Lash, A. E. Gene Expression Omnibus: NCBI gene expression and hybridization array data repository. Nucleic Acids Res. https://doi.org/10.1093/nar/30.1.207 (2002).
Article PubMed PubMed Central Google Scholar
Barrett, T. et al. NCBI GEO: Archive for functional genomics data sets - Update. Nucleic Acids Res. https://doi.org/10.1093/nar/gks1193 (2013).
Article PubMed PubMed Central Google Scholar
Mei, S. et al. Cistrome Data Browser: A data portal for ChIP-Seq and chromatin accessibility data in human and mouse. Nucleic Acids Res. https://doi.org/10.1093/nar/gkw983 (2017).
Article PubMed PubMed Central Google Scholar
Chng, K. R. et al. A transcriptional repressor co-regulatory network governing androgen response in prostate cancers. EMBO J. https://doi.org/10.1038/emboj.2012.112 (2012).
Article PubMed PubMed Central Google Scholar
Sharma, N. L. et al. The ETS family member GABPα modulates androgen receptor signalling and mediates an aggressive phenotype in prostate cancer. Nucleic Acids Res. https://doi.org/10.1093/nar/gku281 (2014).
Article PubMed PubMed Central Google Scholar
Asangani, I. A. et al. Therapeutic targeting of BET bromodomain proteins in castration-resistant prostate cancer. Nature https://doi.org/10.1038/nature13229 (2014).
Article PubMed PubMed Central Google Scholar
Mounir, Z. et al. ERG signaling in prostate cancer is driven through PRMT5-dependent methylation of the androgen receptor. Elife https://doi.org/10.7554/eLife.13964 (2016).
Article PubMed PubMed Central Google Scholar

Download references

Acknowledgements

We want to acknowledge the participants and investigators of the FinnGen study. This work was supported by the grants from the National Natural Science Foundation of China (82073082), the Jane and Aatos Erkko Foundation, the Finnish Cancer Foundation, the Sigrid Juseliuksen Saatio, and the Fudan University Recruit Funding. The work was also supported by a grant from National Institute of Health (1R01CA250018) to L.W. Lentiviral vector and virus preparations were done at the Biocenter Oulu Virus Core Laboratory. Linux High-Performance Computing servers were provided by the CSC – IT CENTER FOR SCIENCE LTD, and also supported by the Medical Research Data Center of Fudan University.

Author information

These authors contributed equally: Nikolaos Giannareas, Qin Zhang, Xiayun Yang, Rong Na.

Authors and Affiliations

Disease Networks Research Unit, Faculty of Biochemistry and Molecular Medicine & Biocenter Oulu, University of Oulu, Oulu, Finland
Nikolaos Giannareas, Qin Zhang, Xiayun Yang, Yuehong Yang, Aki Manninen & Gong-Hong Wei
Division of Urology, Department of Surgery, Li Ka Shing Faculty of Medicine, the University of Hong Kong, Hong Kong, China
Rong Na
Department of Tumour Biology, H. Lee Moffitt Cancer Center and Research Institute, Tampa, FL, USA
Yijun Tian & Liang Wang
Department of Urology, Ruijin Hospital, Shanghai Jiaotong University School of Medicine, Shanghai, China
Xiaohao Ruan & Da Huang
Department of Pathology, Ruijin Hospital, Shanghai Jiaotong University School of Medicine, Shanghai, China
Xiaoqun Yang & Chaofu Wang
Fudan University Shanghai Cancer Center & MOE Key Laboratory of Metabolism and Molecular Medicine and Department of Biochemistry and Molecular Biology of School of Basic Medical Sciences, Shanghai Medical College of Fudan University, Shanghai, China
Peng Zhang & Gong-Hong Wei

Authors

Nikolaos Giannareas
View author publications
You can also search for this author in PubMed Google Scholar
Qin Zhang
View author publications
You can also search for this author in PubMed Google Scholar
Xiayun Yang
View author publications
You can also search for this author in PubMed Google Scholar
Rong Na
View author publications
You can also search for this author in PubMed Google Scholar
Yijun Tian
View author publications
You can also search for this author in PubMed Google Scholar
Yuehong Yang
View author publications
You can also search for this author in PubMed Google Scholar
Xiaohao Ruan
View author publications
You can also search for this author in PubMed Google Scholar
Da Huang
View author publications
You can also search for this author in PubMed Google Scholar
Xiaoqun Yang
View author publications
You can also search for this author in PubMed Google Scholar
Chaofu Wang
View author publications
You can also search for this author in PubMed Google Scholar
Peng Zhang
View author publications
You can also search for this author in PubMed Google Scholar
Aki Manninen
View author publications
You can also search for this author in PubMed Google Scholar
Liang Wang
View author publications
You can also search for this author in PubMed Google Scholar
Gong-Hong Wei
View author publications
You can also search for this author in PubMed Google Scholar

Contributions

Conceptualization, G.-H.W.; N.G. designed and performed most of the experiments and analyzed the data. Q.Z. designed, performed, and interpreted most of the bioinformatics analysis. N.G., Q.Z., and G.-H.W. prepared figures. X.Y.Y. performed ChIP-seq assays and together with N.G. for ChIP-qPCR experiments. R.N. provided SNP genotype data in Chinese cohorts and performed the association study. Methodology, N.G., Q.Z., X.Y.Y., R.N., Y.T., Y.Y., X.R., D.H., X.Q.Y., C.W., P.Z., A.M., L.W., and G.-H.W.; Software, Q.Z., R.N., G.-H.W.; Validation, N.G., X.Y.Y., Y.Y.; Formal analysis, Q.Z., G.-H.W.; Investigation, N.G., Q.Z., X.Y.Y., R.N., Y.Y., and G.-H.W.; Resources, R.N., A.M., L.W., G.-H.W.; Data curation, N.G., Q.Z., G.-H.W.; Writing – original draft, N.G., Q.Z., G.-H.W.; Writing – review & editing, N.G., Q.Z., G.-H.W. with inputs from all authors; Visualization, Q.Z., G.-H.W.; Supervision, G.-H.W.; Project administration, R.N., Y.Y., A.M., L.W., G.-H.W.; Funding acquisition, G.-H.W.

Corresponding author

Correspondence to Gong-Hong Wei.

Ethics declarations

Competing interests

The authors declare no competing interests.

Peer review

Peer review information

Nature Communications thanks Victor Jin and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. Peer reviewer reports are available.

Additional information

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary information

Supplementary Information

Peer Review File

Description of Additional Supplementary Files

Supplementary Data 1

Reporting Summary

Source data

Source Data

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.

Reprints and permissions

About this article

Cite this article

Giannareas, N., Zhang, Q., Yang, X. et al. Extensive germline-somatic interplay contributes to prostate cancer progression through HNF1B co-option of TMPRSS2-ERG. Nat Commun 13, 7320 (2022). https://doi.org/10.1038/s41467-022-34994-z

Download citation

Received: 18 June 2021
Accepted: 15 November 2022
Published: 28 November 2022
DOI: https://doi.org/10.1038/s41467-022-34994-z

This article is cited by

GATA2 co-opts TGFβ1/SMAD4 oncogenic signaling and inherited variants at 6q22 to modulate prostate cancer progression
- Xiayun Yang
- Qin Zhang
- Gong-Hong Wei
Journal of Experimental & Clinical Cancer Research (2023)

Comments

By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.