Whole-genome and Transcriptome Sequencing of Prostate Cancer Identify New Genetic Alterations Driving Disease Progression

EURURO-7536; No. of Pages 17 E U R O P E A N U R O L O G Y X X X ( 2 0 17 ) X X X – X X X

available at www.sciencedirect.com journal homepage: www.europeanurology.com

Platinum Priority – Prostate Cancer Editorial by XXX on pp. x-y of this issue

Whole-genome and Transcriptome Sequencing of Prostate Cancer Identify New Genetic Alterations Driving Disease Progression Shancheng Ren a,y, Gong-Hong Wei b,y, Dongbing Liu c,d,y, Liguo Wang e,y, Yong Hou c,d,f,y, Shida Zhu c,d,g, Lihua Peng c,d,h, Qin Zhang b, Yanbing Cheng c,g, Hong Su c,d, Xiuqing Zhou c,d, Jibin Zhang c, Fuqiang Li c,d, Hancheng Zheng c, Zhikun Zhao c,d,i,j, Changjun Yin k,x, Zengquan He c, Xin Gao y, Haiyen E. Zhau m, Chia-Yi Chu m, Jason Boyang Wu m, Colin Collins n, Stanislav V. Volik n, Robert Bell n, Jiaoti Huang o, Kui Wu c,d, Danfeng Xu p, Dingwei Ye q, Yongwei Yu r, Lianhui Zhu a, Meng Qiao a, Hang-Mao Lee b, Yuehong Yang b, Yasheng Zhu a, Xiaolei Shi a, Rui Chen a, Yang Wang r, Weidong Xu a, Yanqiong Cheng a, Chuanliang Xu a, Xu Gao a, Tie Zhou a, Bo Yang a, Jianguo Hou a, Li Liu c, Zhensheng Zhang a, Yao Zhu q, Chao Qin k, Pengfei Shao k, Jun Pang y, Leland W.K. Chung m, Jianfeng Xu f,s, Chin-Lee Wu t, Weide Zhong u, Xun Xu c,d, Yingrui Li c, Xiuqing Zhang c, Jian Wang c,v, Huanming Yang c,v, Jun Wang c,w,x,y, Haojie Huang z,z, Yinghao Sun a,z,* a

Department of Urology, Shanghai Changhai Hospital, Second Military Medical University, Shanghai, China; b Biocenter Oulu, Faculty of Biochemistry and

Molecular Medicine, University of Oulu, Oulu, Finland; c BGI-Shenzhen, Shenzhen, China;

d

China National GeneBank-Shenzhen, BGI-Shenzhen, Shenzhen,

China; e Division of Biomedical Statistics and Informatics, Mayo Clinic College of Medicine, Rochester, MN, USA; f State Key Laboratory of Genetic Engineering, School of Life Sciences, Fudan University, Shanghai, China; g Division of Genomics and Bioinformatics, CUHK-BGI Innovation Institute of Trans-Omics, The Chinese University of Hong Kong, Hong Kong, China;

h

BGI Education Center, University of Chinese Academy of Sciences, Shenzhen, China; i School of

Biological Science and Medical Engineering, Southeast University, Nanjing, China; j State Key Laboratory of Bioelectronics, Southeast University, Nanjing, China; k Department of Urology, The First Affiliated Hospital of Nanjing Medical University, Nanjing, China; l Department of Urology, The Third Affiliated Hospital of Sun Yat-Sen University, Guangzhou, China;

m

Uro-Oncology Research Program, Department of Medicine, Samuel Oschin Comprehensive Cancer

Institute, Cedars-Sinai Medical Center, Los Angeles, CA, USA; Columbia, Vancouver, BC, Canada; Angeles, CA, USA;

p

o

n

Vancouver Prostate Centre and Department of Urologic Sciences, University of British

Department of Pathology and Laboratory Medicine, David Geffen School of Medicine, University of California, Los

Department of Urology, Changzheng Hospital, Second Military Medical University, Shanghai, China;

q

Department of Urology, Fudan

University Shanghai Cancer Center, Shanghai, China; r Department of Pathology, Shanghai Changhai Hospital, Second Military Medical University, Shanghai, China; s Program for Personalized Cancer Care, NorthShore University HealthSystem, Evanston, IL, USA; t Department of Pathology, Massachusetts General Hospital, Harvard Medical School, Boston, MA, USA; u Department of Urology, Guangdong Key Laboratory of Clinical Molecular Medicine and Diagnostics, Guangzhou First People's Hospital, Guangzhou Medical University, Guangzhou, China; v James D. Watson Institute of Genome Sciences, Hangzhou, China; w

Department of Biology, University of Copenhagen, Copenhagen, Denmark; x The Novo Nordisk Foundation Center for Basic Metabolic Research, University of

Copenhagen, Copenhagen, Denmark; y King Abdulaziz University, Jeddah, Saudi Arabia; z Department of Biochemistry and Molecular Biology, Mayo Clinic College of Medicine, Rochester, MN, USA

y

These authors contributed equally to this study. These authors are cosenior authors. x Deceased. * Corresponding author. Department of Urology, Shanghai Changhai Hospital, Second Military Medical University, Changhai Road 168, Yangpu District, Shanghai 200433, China. Tel. +86-13601-607755; Fax: +86-21-3505-0006. E-mail address: [email protected] (Y. Sun). z

http://dx.doi.org/10.1016/j.eururo.2017.08.027 0302-2838/© 2017 Published by Elsevier B.V. on behalf of European Association of Urology.

Please cite this article in press as: Ren S, et al. Whole-genome and Transcriptome Sequencing of Prostate Cancer Identify New Genetic Alterations Driving Disease Progression. Eur Urol (2017), http://dx.doi.org/10.1016/j.eururo.2017.08.027

EURURO-7536; No. of Pages 17 2

E U R O P E A N U R O L O G Y X X X ( 2 0 17 ) X X X – X X X

Article info

Abstract

Article history: Accepted August 24, 2017

Background: Global disparities in prostate cancer (PCa) incidence highlight the urgent need to identify genomic abnormalities in prostate tumors in different ethnic populations including Asian men. Objective: To systematically explore the genomic complexity and define disease-driven genetic alterations in PCa. Design, setting, and participants: The study sequenced whole-genome and transcriptome of tumor-benign paired tissues from 65 treatment-naive Chinese PCa patients. Subsequent targeted deep sequencing of 293 PCa-relevant genes was performed in another cohort of 145 prostate tumors. Outcome measurements and statistical analysis: The genomic alteration landscape in PCa was analyzed using an integrated computational pipeline. Relationships with PCa progression and survival were analyzed using nonparametric test, log-rank, and multivariable Cox regression analyses. Results and limitations: We demonstrated an association of high frequency of CHD1 deletion with a low rate of TMPRSS2-ERG fusion and relatively high percentage of mutations in androgen receptor upstream activator genes in Chinese patients. We identified five putative clustered deleted tumor suppressor genes and provided experimental and clinical evidence that PCDH9, deleted/loss in approximately 23% of tumors, functions as a novel tumor suppressor gene with prognostic potential in PCa. Furthermore, axon guidance pathway genes were frequently deregulated, including gain/ amplification of PLXNA1 gene in approximately 17% of tumors. Functional and clinical data analyses showed that increased expression of PLXNA1 promoted prostate tumor growth and independently predicted prostate tumor biochemical recurrence, metastasis, and poor survival in multi-institutional cohorts of patients with PCa. A limitation of this study is that other genetic alterations were not experimentally investigated. Conclusions: There are shared and salient genetic characteristics of PCa in Chinese and Caucasian men. Novel genetic alterations in PCDH9 and PLXNA1 were associated with disease progression. Patient summary: We reported the first large-scale and comprehensive genomic data of prostate cancer from Asian population. Identification of these genetic alterations may help advance prostate cancer diagnosis, prognosis, and treatment. © 2017 Published by Elsevier B.V. on behalf of European Association of Urology.

Associate Editor: James Catto Keywords: Chinese prostate cancer Whole genome sequencing RNA-seq CHD1 deletion TMPRSS2-ERG fusion AR coactivator mutation Clustered deleted tumor suppressor genes Axon guidance pathway Prognosis Personalized medicine

1.

Introduction

Prostate cancer (PCa) is the second most frequently diagnosed cancer and the fifth leading cause of cancer death in men worldwide [1]. The disparity in PCa incidence and death rate is obvious around the globe. The reported incidence and mortality rate of PCa in Asian countries including China were much lower than in Western nations [1,2], with an estimated 60 300 new cases and 26 600 deaths in Chinese men in 2015 [3]. However, the PCa incidence rate increased rapidly in China with an annual percentage change of 12.6% since 2000 [3]. These findings not only highlight the complexity of genomic abnormalities in PCa, but also stress an urgent need of genome-wide molecular and genetic profiling of prostate tumors from different ethnic groups including Asian men. Prostate tumors also show highly variable clinical outcomes. Some patients survive for over 10 yr after diagnosis, but others, particularly those with an aggressive phenotype, only survive for 2–3 yr. Given the dramatic differences in treatment response, great efforts have been made to investigate the genetic and epigenetic heterogeneity and cell signaling defects involved in PCa progression [4]. Many advanced approaches, including target-region sequencing, array-based gene expression, copy number variation (CNV), and whole-genome sequencing of tumor

samples, have been taken to portray the genomic landscape in PCa [5–7]. These works have reported several PCa-related genomic alterations, including the most common TMPRSS2ERG fusion, copy number gains of 8q, copy number losses of 3p, 8p, 10q, 13q, and 17p, as well as complex chains of oncogenic structural DNA rearrangements (chromoplexy). However, the functional consequence of many alterations remains unknown. Recent exome sequencing of prostate tumors revealed specific genetic alterations in coding regions, leading to the identification of frequent mutations in the genes such as SPOP, FOXA1, TP53, and PTEN [8–10], and other genomic alterations in PIK3CA/B, ZBTB16/PLZF, and AR [11]. More recently, The Cancer Genome Atlas (TCGA) described a comprehensive genomic analysis of 333 PCa patients [12], and the International Cancer Genome Consortium performed genomic profiling of 477 localized, nonindolent prostate tumors [13], mainly in the Caucasian population. Together, genome-wide alterations in PCa have been extensively studied in men from Western populations. In contrast, while the incidence rate of PCa has been rising dramatically in cities such as Hong Kong and Shanghai [3], the landscape of genome alterations in Asian PCa cohorts remains incompletely characterized, becoming a major hurdle for comprehensive understanding of the molecular etiology of this fatal disease.

Please cite this article in press as: Ren S, et al. Whole-genome and Transcriptome Sequencing of Prostate Cancer Identify New Genetic Alterations Driving Disease Progression. Eur Urol (2017), http://dx.doi.org/10.1016/j.eururo.2017.08.027

EURURO-7536; No. of Pages 17 E U R O P E A N U R O L O G Y X X X ( 2 0 17 ) X X X – X X X

3

With the goal to systematically define the genomic alterations specifically occurring in Chinese PCa patients, we performed whole-genome and transcriptome sequencing for 65 tumor and the paired normal tissue samples (discovery cohort) and very deep gene-targeted sequencing for additional 145 tumor-normal pairs (validation cohort). By comparing our data with the TCGA datasets [12], we not only identified common genetic alterations among different ethnic groups such as SPOP, TP53, and PTEN, but also observed different frequencies of genomic alterations specifically associated with Chinese patients, including highly frequent deletion of CHD1 and relatively high mutation rates in androgen receptor (AR) upstream regulator genes including NCOR2. Furthermore, our analysis identified PCDH9 as a novel tumor suppressor gene (TSG) whose deletion is associated with poor survival in metastatic PCa patients. We also identified frequent alterations in axon guidance pathway genes including PLXNA1 whose upregulation and amplification are associated with PCa relapse and survival.

2.3.

2.

Patients and methods

2.5.

2.1.

Patients and samples

Genomic DNA from the 145 paired normal tumor samples of a validation

Gene expression quantification

Trapnell et al [16] (http://cufflinks.cbcb.umd.edu/index.html) was used to quantify the gene expression and identify differential gene expression according to the protocol and manual for each patient. Differential gene expression was identified using the threshold of false discovery rate <0.05. For genes differentially expressed in at least one patient, thresholds of fold change 2 and p value 0.05 were applied for the other patients with a false discovery rate 0.05. Pathway enrichment and correlation network analyses are described in the Supplementary data.

2.4.

Significantly mutated genes analysis

We employed Youn and Simon's [17] method to predict the significance of gene mutations. The model evaluates both functional impact and mutation prevalence. To evaluate the functional impact, a mutation score was assigned base on BLOSUM80 in the following order: missense < inframe indel < mutation in splice sites < frame shift indel = nonsense. The p value was calculated from the background distribution computed by incorporating different background mutation rates of each sample and the observed mutation across samples.

Target sequencing of 293 gene panel in validation cohort

cohort were fragmented to 150–200 bp and subjected to exon capture of Treatment-naive prostate tumor and matched normal tissues were

293 PCa relevant genes (Supplementary Table 3) including the

collected from the radical prostatectomy series at Shanghai Changhai

64 significantly mutated genes (SMGs) detected from our discovery

Hospital and Fudan University Shanghai Cancer Center. The institutional

cohort following the manufacturer's protocols of Agilent. Then the exon

review boards of both hospitals approved the experimental protocols.

captured libraries were sequenced on Illumina HiSeq 2500 according to

Informed consent was obtained from all participants. Hematoxylin and

the manufacturer's instructions. Paired-end 150-bp reads were generat-

eosin (H&E) slides of frozen human tumor tissues and matched normal

ed for each sample.

tissues were examined by a pathologist and another gynecologic pathologist to confirm histological diagnosis and Gleason score. They

2.6.

Tumor cell biology assays

also verified the high-density cancer foci (> 80%) of the selected tumor tissue, and the contamination free of the normal tissues. The frozen blocks

Cell lines and culture, knockdown and overexpression, proliferation,

for DNA/RNA extraction were examined by the pathologist as described

invasion, migration, blotting, and reporter assay procedures are detailed

above, followed by consecutive 10  10 mm cut of tumor section. These

in the Supplementary data.

qualified samples were then used for DNA/RNA isolation. Subsequent sample annotation and preparation were prepared as described [6].

2.7.

Immunohistochemistry

In total, we selected 65 radical prostatectomy specimens with a range of prostate tumor grades and stages from treatment-naive Chinese

Tissue microarrays from Shanghai Changhai Hospital (87 cases), Chinese

patients (Supplementary Table 1) for whole-genome and transcriptome

PCa Consortium (419 cases), and Massachusetts General Hospital

sequencing, and obtained DNA and RNA sequencing data from paired

(213 cases) were used for immunohistochemistry (IHC) staining with

tumor and adjacent benign tissues from these 65 patients (Supplemen-

the antibodies shown in Supplementary Table 4. The detailed protocol is

tary Table 2). DNA and RNA preparation, and library construction are

described in the Supplementary data.

described in the Supplementary data.

2.8. 2.2.

Animal studies

Data preprocessing and reads mapping PC-3 1  106 (with shControl and shPLXNA1, respectively), C4-2 (shControl

Raw reads of DNA and RNA were filtered using an in-house pipeline

and shPLXNA1, respectively), or DU-145 (vector and PLXNA1-overexpres-

based on the following procedure. Reads with sequencing adapters, more

sion, respectively), DU-145 (vector and PCDH9-overexpression, respec-

than 10% N bases or with low quality were removed. Ribosomal RNAs

tively) cells were mixed 1:1 with Matrigel (BD Biosciences, San Jose, CA,

were removed by aligning the RNA reads to a combined reference

USA), and injected subcutaneous into nude mice (4–6 mice in each group).

sequence of ribosomal RNA from Ensembl, University of California, Santa

No blinding was done. For each group, tumor size was measured every 2 d

Cruz, and SILVA database. Clean DNA reads were aligned to the human

using a caliper. Subcutaneous tumor volume was calculated as

reference genome hg19 with Burrows-Wheeler Aligner and then

0.52  length  width. Mice were sacrificed 40–70 d after injection, and

processed with SAMTools [14] to remove the polymerase chain reaction

the tumors were dissected. The wet weights of the tumors were

duplicates. Clean RNA-sequencing reads were aligned to hg19 genome

determined. Tumors were fixed in 4% formaldehyde and embedded in

using Tophat [15]. Supplementary data includes the detailed description

paraffin. Sections were stained with H&E and other markers by IHC.

of the detection of somatic mutation, structure variation, gene fusion,

All animal experiments were approved by our local animal ethics

and CNV, and experimental validation and clustered deleted TSG

committee at Second Military Medical University, and were executed in

identification.

accordance with animal care guidelines.

Please cite this article in press as: Ren S, et al. Whole-genome and Transcriptome Sequencing of Prostate Cancer Identify New Genetic Alterations Driving Disease Progression. Eur Urol (2017), http://dx.doi.org/10.1016/j.eururo.2017.08.027

EURURO-7536; No. of Pages 17 4

E U R O P E A N U R O L O G Y X X X ( 2 0 17 ) X X X – X X X

2.9.

Analysis of clinical PCa data sets

Clinical data set, univariate, and multivariate analyses are detailed in the Supplementary data.

2.10.

Statistical analysis

Statistical calculations were performed using the SPSS software (version 16.0; SPSS Inc., Chicago, IL, USA) and the statistical package, OriginPro 9 (OriginLab, Northampton, MA, USA). Survival curves were calculated by the Kaplan-Meier method and compared with the Log-rank test and Cox regression analysis.

3.

Results

3.1.

The genomic alteration landscape in PCa of Chinese men

In this study, tumor-normal paired samples from 65 Chinese PCa patients were analyzed using whole genome sequencing

(Supplementary Table 1). These samples were treatmentnaive radical prostatectomy specimens. We performed the whole-genome sequencing for three tumor-normal pairs at approximately 96 coverage and for the remaining 62 pairs at approximately 60 coverage. We also performed the whole transcriptome sequencing (RNA-seq) for these patients (Supplementary Table 2). After mapping reads to the human reference genome, we detected somatic point mutations and somatic insertions and deletions (indels; < 50 bp) that were present in tumors and absent in adjacent normal samples (Supplementary data). In total, we identified 146 387 somatic point mutations from the 65 tumors with 117–8728 mutations per tumor genome. We defined 1134 single nucleotide variations (SNVs; 1034 missense, 68 nonsense, and 32 splice site) and 50 indels (25 frameshift, 20 nonframeshift, and 5 located at canonical splice sites) that were predicted to alter the encoded protein sequence (Fig. 1A, Supplementary

Fig. 1 – Somatic mutation landscape of 65 prostate cancer patients detected by whole-genome sequencing. (A) Somatic point mutations. Samples were listed in an order according to the rate of point mutations detected in each sample. (B) Structural variations and genome rearrangements, including interchromosomal translocation (CTX), deletion (DEL), insertion (INS), inversion (INV), intrachromosomal translocation (ITX). (C) A mutation context heat map was constructed from the counts of each mutation type in each mutation context normalized by the background frequency of each trinucleotide in the reference genome. For each mutation type, the 50 base to each mutated base is shown on the vertical axis (from top to bottom, T, G, C, and A) and 30 base on the horizontal axis (from left to right, A, C, G, and T). (D) RNA-sequencing reads were used to classify coding single nucleotide variations (SNVs) into four categories according to their expression patterns. (E) Somatic copy number alterations, including copy number gain and loss. (F) The status of ETS fusions, CHD1 loss/deletion, Gleason Score (GS), and tumor stage in each patient. CNA = copy number alterations; LN = lymph node involvement; NA = not available; WT = wild type.

Please cite this article in press as: Ren S, et al. Whole-genome and Transcriptome Sequencing of Prostate Cancer Identify New Genetic Alterations Driving Disease Progression. Eur Urol (2017), http://dx.doi.org/10.1016/j.eururo.2017.08.027

EURURO-7536; No. of Pages 17 E U R O P E A N U R O L O G Y X X X ( 2 0 17 ) X X X – X X X

Tables 5 and 6). The predominant somatic mutation spectrum across the entire genome was C>T nucleotide transition (Fig. 1C, Supplementary Figs. 1A and 1B), consistent with previous observations in PCa [10,18,19], and our analysis of two independent patient cohorts (Supplementary Figs. 1C and 1D). We also performed a non-negative matrix factorization analysis to extract signatures of genome-wide mutations and kataegis events (regional hypermutation) [20], and found that the APOBECrelated signature (including C>T, C>G, and C>A substitutions) [21] contributes to mutations within the regions of kataegis in prostate tumors (Supplementary Figs. 2A–D, Supplementary data), suggesting an involvement of APOBEC activity in the generation of kataegis signature in PCa genomes. We next analyzed transcriptome sequencing data of the cohort and observed the expression of 21 965 genes. Of these, 7327 genes were differentially expressed between tumors and matched normal tissues in at least one patient (Supplementary Table 7). We next compared SNVs detected in coding regions with corresponding transcriptome sequencing reads. We classified these variations into four categories according to their expression patterns, as previously reported [22]: expressed, mutant biased, wildtype biased, and silent gene (Fig. 1D). Except for SNVs from a nonexpressed allele or those that were not sufficiently covered by the transcriptome sequencing reads, these sequencing variations showed a high degree of concordance between the genome and transcriptome, with more than 30% of the variations being covered by at least five reads from the transcriptome data. We also identified 8159 somatic structural variations (SVs) at base-pair resolution (Supplementary Table 8). These SVs comprised 1286 interchromosomal translocations, 2472 intrachromosomal translocations, 2769 deletions, 1627 insertions, and five inversions (Fig. 1B). A total of 4651 events (57.0%) affected coding regions, of which 2274 (27.1%) were chromosomal translocations. We also detected somatic CNVs for each tumor-normal tissue pair in our PCa cohort (Fig. 1E). A total of 125 gene fusions were identified by integrative analysis of whole-genome and whole-transcriptome sequencing data. Four of these gene fusions were recurrent in our cohort, including ETS fusions (6/65 [9.2%], including 4 TMPRSS2-ERG), TTC6MIPOL1 (4/65 [6.2%]), BPTF-LRRC37A3 (3/65 [4.6%]), and WWP2-NFAT5 (2/65 [3.1%]; Supplementary Table 9). The prevalence of ETS family gene fusions (6/65) in our cohort (Fig. 1F) is much lower than that in Caucasian cohort (approximately 50%) as reported previously [23,24]. We confirmed the low prevalence of TMPRSS2–ERG fusion in our cohort by fluorescent in situ hybridization and breakpoint polymerase chain reaction followed by Sanger sequencing (Supplementary Figs. 3A–C). Low frequency of ETS family gene fusions has been reported in other cohorts of Asian PCa patients [25–30]. Thus, our data further confirm that low frequency of ETS fusion represents a salient genetic characteristic of PCa in Asian men.

3.2.

5

Highly frequent deletion of the CHD1 gene

Previous studies in cell culture showed that the chromatin modifier CHD1 is required for the genesis of ERG fusions [31]. This finding provides a plausible explanation for a previous finding in PCa patient samples where CHD1 deletion (CHD1–) almost always coexists with ETS fusion negative (ETS–) [9]. In our discovery cohort, we not only showed that CHD1 deletion and ETS fusion are almost mutually exclusive (Fig. 1F; p = 0.0012, permutation test), but most importantly, we found that CHD1 loss/deletion is approximately in 31% (20/65) of prostate tumors in Chinese men examined (Fig. 1F) and that the deletion rate is almost two times higher than in TCGA patients (majority of them are Caucasian men, CHD1, 54/333 [16%]) [12]. In contrast, only approximately 6% of PCa in Chinese patients had ERG fusions (Supplementary Table 9). Thus, our data not only provide first-hand evidence that the CHD1 gene is frequently deleted in Chinese men, but further extend the previous finding in Caucasian patients that CHD1 deletion inversely correlates with ERG fusions in PCa. Next, we performed mutation co-occurrence and mutual exclusion analysis for important prostate relevant genetic alterations. We confirmed that CHD1 loss/deletion significantly cooccurrs with SPOP mutation (Supplementary Fig. 4). We also observed CHD1 loss/deletion significantly co-occurrs with SPOPL loss/deletion, PCDH9 loss/deletion in our dataset, and significant co-occurrence of PCDH9 and RB1 loss/deletion (Supplementary Fig. 5). In addition, we validated these correlations by analyzing the TCGA published dataset [12]. The large number of patients in TCGA published dataset enable us to discover more genetic alterations correlated with CHD1 loss/deletion (Supplementary Fig. 4). We found in the TCGA published dataset that CHD1 loss/ deletion also significantly co-occurrs with loss/deletion of APC and RB1, respectively, and FOXA1 mutations, as well as mutual exclusive with mutation/loss/deletion of TP53 and PTEN, respectively. 3.3.

Detection of mutations in AR upstream regulator genes in

Chinese patients

Next, we identified a total of 64 SMGs in our discovery cohort (Supplementary Table 10). To validate this finding and explore more potential driver genes in PCa of Chinese patients, we focused on a set of 293 PCa relevant genes (Supplementary Table 3), which include these 64 SMGs and others reported previously [8,9]. We then performed targeted deep sequencing (approximately 243) for the 293 genes using a customized capture array in a validation cohort of 145 matched tumor-normal pairs (Supplementary Tables 11 and 12). While our analysis revealed that most of the SMGs, including SPOP, TP53, ATM, PTEN, and CTNNB1 were commonly mutated in our cohorts of Chinese men and the TCGA cohort of Caucasian patients (Fig. 2A, Supplementary Table 13), we found that alteration frequencies in chromatin remodeling and histone modification genes such

Please cite this article in press as: Ren S, et al. Whole-genome and Transcriptome Sequencing of Prostate Cancer Identify New Genetic Alterations Driving Disease Progression. Eur Urol (2017), http://dx.doi.org/10.1016/j.eururo.2017.08.027

EURURO-7536; No. of Pages 17 6

E U R O P E A N U R O L O G Y X X X ( 2 0 17 ) X X X – X X X

Fig. 2 – Integrative analysis of significantly mutated genes in prostate cancer. (A) The mutation rate of significantly mutated genes was compared between affected tumors in The Cancer Genome Atlas (TCGA) cohort of 333 patients and our cohorts of 210 patients (discovery and validation cohorts). Each row represents a specific gene, and each column indicates a tumor sample. Left panel, mutation frequency of each significantly mutated gene and the matrix of mutations in our cohort. Right panel, results from the TCGA cohort. In the matrix, different color indicates individual mutation types. The upper 12 genes and FOXA1 were predicted to be significantly mutated by MutSigCV in the TCGA cohort. The lower seven in red are androgen receptor (AR) upstream regulator genes with higher mutation prevalence in Chinese tumor samples in comparison to TCGA cohort. Only the tumors with at least one mutation in the listed genes are shown. (B) Mutational aberrations of AR upstream regulator genes and AR are shown. The frequency of high copy number alterations, point mutations, and both aberrations are shown according to the color scales and numbers in each cell.

as MLL3 (13/210 [6.19%]) and transcription coregulator genes such as FOXA1 (13/210 [6.19%]) and NCOR2 (7/210 [3.33%]) were relatively higher in our discovery and validation cohort of clinically localized PCa compared those in the TCGA dataset (Fig. 2A, Supplementary Fig. 6). Among the frequently mutated genes, NCOR2 was found mutated in our cohort (3.33%), but not in the TCGA cohort (Fig. 2A). NCOR2 encodes a transcriptional coregulator existed in a protein complex containing histone deacetylases that modifies chromatin structure and serves as a transcriptional corepressor for the nuclear receptor superfamily of transcription factors [32] including AR in PCa. We also observed frequent NCOR2 expression alterations in our discovery cohort of PCa (Supplementary Table 7). Furthermore, we found that steroidogenic enzyme genes such as CYP11B1 and HSD14B4 were mutated in our cohorts, but little or not in TCGA patients (Fig. 2A). In contrast, we found no apparent alterations of AR in our cohort (Fig. 2B). Thus, while there were no detectable alterations in the AR gene itself, we found that the mutation rate in AR upstream modulator genes was relatively higher in PCa in Chinese men in comparison to that in Caucasian patients.

3.4.

Identification of highly frequent deletion of the adhesion

gene PCDH9

To identify additional PCa relevant genes, we next assessed broader DNA copy number changes in our cohort. Overall, we found very few large regions with significant gain in copy numbers, except for the known amplifications at chromosome 8q (8q13, 8q21, 8q22, 8q22, 8q23 and 8q24) [6,24,33], a region harboring oncogenes such as MYC (8q24.21), PVT1 (8q24.21), and NCOA2 (8q13.3; Fig. 3A, Supplementary Table 14). In contrast, we found frequent large deletions in several genomic regions, including chromosome 5q (eg, 5q21 with CHD1, RGMB, and 5q22.1>q22.2 with APC), 6q, 8p, 10q (10q23.31 with PTEN), 13q (13q14.2 with RB1), and 16q (Fig. 3A, Supplementary Fig. 5), which is consistent with the findings observed previously [6,24,33]. The deletion of chromosome 2q (2q14.3, 2q22.3) was also reported by Taylor et al [6]. Notably, compared with the CNV profiling from Taylor et al [6] and Ross-Adams et al [24] there was no significant deletion of chromosome 21q (21q22) in our cohort. Chromosome 21q (21q22) was frequently deleted in PCa, the deletion of which would lead

Please cite this article in press as: Ren S, et al. Whole-genome and Transcriptome Sequencing of Prostate Cancer Identify New Genetic Alterations Driving Disease Progression. Eur Urol (2017), http://dx.doi.org/10.1016/j.eururo.2017.08.027

EURURO-7536; No. of Pages 17 E U R O P E A N U R O L O G Y X X X ( 2 0 17 ) X X X – X X X

7

Fig. 3 – Association between PCDH9 deletion and its expression, and prostate cancer (PCa) aggressiveness. (A) Landscape of copy number alteration in Chinese cohort of PCa. (B) Statistically significant copy number alterations were shown on 13q21.31-q21.33 region. The G-score was calculated using GISTIC algorithm. Genomic positions of tumor suppressor genes (TSGs) within PCDH9-DACH1-KLF5-LECT1-OLFM4 cluster were determined. (C,D) PCDH9 expression in prostate tissue samples with different somatic copy number alteration status. Note that PCDH9 deletion was significantly associated with its downregulation in PCa samples. PCDH9_wt, samples with wild type PCDH9; PCDH9_loss, samples with PCDH9 loss; PCDH9_deletion, samples with

Please cite this article in press as: Ren S, et al. Whole-genome and Transcriptome Sequencing of Prostate Cancer Identify New Genetic Alterations Driving Disease Progression. Eur Urol (2017), http://dx.doi.org/10.1016/j.eururo.2017.08.027

EURURO-7536; No. of Pages 17 8

E U R O P E A N U R O L O G Y X X X ( 2 0 17 ) X X X – X X X

to the formation of TMPRSS2-ERG fusion [6,23,24]. This indicated that the low frequency of TMPRSS2-ERG fusion in our cohort might be induced by the lack of interstitial deletion of the intervening genomic region between ERG and TMPRSS2 genes on chromosome 21 (deletion). In addition, we also observed novel frequent large chromosomal deletions in our cohort, which include an approximately 8.6-Mb deletion (24/65 [36.9%]) at 2q22.1-q22.3 containing 12 genes such as SPOPL, an approximately 8-Mb deletion (31/65 [47.7%]) at 13q21.31-q21.33 harboring 10 genes such as PCDH9, and an approximately 0.15-Mb deletion (21/65 [32.3%]) at 16q24.1 encompassing six genes such as FOXF1, FOXC2, and FOXL1 (Supplementary Table 14). We next performed integrative analysis of data from somatic copy-number alteration, corresponding gene expression, and somatic mutations in our cohort, and built a pipeline to identify putative clustered deleted TSGs, which is defined as a cluster of deleted genes located in close proximity or neighboring genes in the same large deletion region. In total, we identified five high-confidence clustered deleted TSGs within seven large copy number deletion regions (Supplementary Tables 15–17). We next focused on PCDH9, as it was the most frequently deleted gene in comparison to others in the large deletion region at 13q21.31-q21.33 (Fig. 3B). In addition, PCDH9 was differentially expressed (rank sum test p < 0.05) in the 65 tumor samples compared with that in the matched normal tissues, whereas RB1 expression showed no difference (p = 0.481; Supplementary Fig. 5B). PCDH9 deletion was significantly correlated with its reduced expression within the same set of tumors (p = 1.74  10–5, t test; Fig. 3C). This concordance was also observed in additional clinical dataset of PCa (Fig. 3D, Supplementary Fig. 7A) [6,12]. Interestingly, we noticed that PCDH9 was downregulated in a subset of tumors without PCDH9 loss/deletion (p = 7.62  10–9, t test; Fig. 3C), despite the observation that the expression level of PCDH9 was much lower in tumors with PCDH9 loss than the ones without copy number loss (p = 7.13  10–4, two-tailed t test; Fig. 3C), suggesting that mechanisms other than genomic deletion may also contribute to the reduced expression of PCDH9 in prostate tumors. 3.5.

PCDH9 is a TSG with prognostic potential in PCa

PCDH9 is a member of the protocadherin family and functions in cell-cell adhesion, neural projection, and synapse formation [34]. Recent evidence showed that loss

of PCDH9 expression was associated with higher histological grade and poor prognosis in glioma [35], implying a tumor suppressor role for PCDH9 in cancer. To determine whether this is the case in PCa, we investigated PCDH9 messenger RNA (mRNA) expression across several independent clinical data sets [6,9,36–39]. The results demonstrated that PCDH9 expression was dramatically downregulated during progression of PCa to the advanced/metastatic stage in multiple cohorts of data sets (Fig. 3E–G, Supplementary Figs. 7B–H). Moreover, PCDH9 downregulation markedly correlated with elevated levels of prostate-specific antigen and high clinical stage of prostate tumors (Fig. 3H, Supplementary Fig. 8). Furthermore, the time to biochemical relapse was significantly shorter in the group of PCa patients with lower PCDH9 expression (Fig. 3I, Supplementary Figs. 9A and 9B), though this association was not observed in additional three PCa datasets by querying the camcAPP, a user-friendly web interface [40], probably due to biological heterogeneity in different patient cohorts. This analysis also showed that PCDH9 was a better predictor than any other individual gene within this region in two clinical PCa datasets [39,41] (Supplementary Table 18). Given that PCDH9 loss/deletion correlates with PCDH9 downregulation (Fig. 3B–D) and that PCDH9 downregulation is detected in advanced tumors (Fig. 3C–D, Supplementary Fig. 7), we explored whether PCDH9 loss/deletion directly correlates with PCa aggressiveness. This analysis revealed that PCDH9 loss/deletion indeed showed to be associated with PCa recurrence [6,34] (Fig. 3J, Supplementary Fig. 9C). Moreover, PCDH9 loss/deletion frequently occurred in metastatic prostate tumors and castrationresistant PCa (CRPC) patients (Fig. 3K, Supplementary Figs. 9D and 9E), and significantly associated with decreased overall survival of metastatic PCa patients (Fig. 3L, Supplementary Figs. 9F and 9G), indicative of the aggressive nature of tumors with PCDH9 copy number loss/deletion. Taken together, these results suggest a prognostic value of PCDH9 loss/deletion and downregulation in PCa and strongly imply a tumor suppressor role of PCDH9 in PCa development and progression. We next examined whether PCDH9 plays a causal role in prostate oncogenesis. We demonstrated that transient knockdown of PCDH9 by small interfering RNA led to increased cell proliferation, migration, and invasion in PCa cell lines LNCaP and DU145, and immortalized prostatic epithelial cell line RWPE-1 (Fig. 4A–C, Supplementary Figs. 10A–E). Accordingly, ectopic expression of PCDH9 significantly attenuated PCa cell proliferation, migration,

PCDH9 deletion. (E–G) PCDH9 messenger RNA expression was significantly downregulated in human primary and metastasis PCa. (H) Decreased PCDH9 expression markedly correlated with elevated serum prostate-specific antigen (PSA) levels. (C–H) The horizontal lines represent the median values. (C, D) The p values were calculated using paired two-tailed t test. Mann-Whitney U-tests were used to assess statistical significance for the comparisons between groups in E–G. (H) The p value was assessed by a Kruskal-Wallis test. (I) Kaplan Meier plots of the risk of biochemical recurrence in patients with high or low expression of PCDH9 in a cohort of human prostate tumors. (J) Kaplan Meier plots of increased risk of biochemical recurrence in prostate tumors with copy number loss/deletion of PCDH9 in one independent cohort of human PCa. (K) Portion of tumors harboring PCDH9 copy number loss was significantly higher in metastasis than in primary PCa. The p value was assessed by two-sided Fisher's exact test. (L) PCDH9 copy number loss/deletion was significantly associated with reduced survival time for the patients with metastatic PCa. (I,J,L) The p values were calculated with a log-rank test, and numbers of patients in each category shown for each 20-mo interval. The p values in I and L were also assessed by Cox regression analysis. TCGA = The Cancer Genome Atlas.

Please cite this article in press as: Ren S, et al. Whole-genome and Transcriptome Sequencing of Prostate Cancer Identify New Genetic Alterations Driving Disease Progression. Eur Urol (2017), http://dx.doi.org/10.1016/j.eururo.2017.08.027

EURURO-7536; No. of Pages 17 E U R O P E A N U R O L O G Y X X X ( 2 0 17 ) X X X – X X X

Fig. 4 – In vitro and in vivo functional validation of PCDH9 as a tumor suppressor in prostate cancer. (A) Cell proliferation was measured by CCK-8 assay (absorbance at 450 nm) in the prostate cell line DU145 in given time points. (B) Quantitation of relative invasion for the cells transfected with control and the efficient small interfering RNAs (siRNAs) against PCDH9. (C) Quantitation of relative migration for the cells transfected with control and the efficient siRNAs against PCDH9. (D) Attenuation of PCa cell proliferation, invasion and migration by ectopic overexpression of PCDH9. (A–D) All experiments were performed in triplicate. Error bars, standard deviation. Results were statistically evaluated with two-tailed t-test. (E) Subcutaneous xenograft assay showed reduced tumor growth of DU145 cells stably expressing PCDH9 compared with control group (n = 6 mice per group; two-tailed t-test). (F) Scatter plot showed tumor weights from individual mice in each group (n = 6 mice per group; two-tailed t-test). (G) Hematoxylin and eosin (H&E) and immunohistochemistry analysis of Ki-67, PCDH9 expression in DU145 xenograft tumors. Representative images from six independent samples are shown. Original magnification, 400T; scale bars: 20 mm. (H) Heat map shows selected differential expression genes identified by genomewide transcriptional profiling of DU145 cells stably expressing PCDH9 (DU145P) and control vector (DU145M) with three replicates, respectively. The color code represents log2 transformed R/G fold ratios. OD = optical density * p < 0.05. ** p < 0.01. *** p < 0.001.

Please cite this article in press as: Ren S, et al. Whole-genome and Transcriptome Sequencing of Prostate Cancer Identify New Genetic Alterations Driving Disease Progression. Eur Urol (2017), http://dx.doi.org/10.1016/j.eururo.2017.08.027

9

EURURO-7536; No. of Pages 17 10

E U R O P E A N U R O L O G Y X X X ( 2 0 17 ) X X X – X X X

and invasion (Fig. 4D, Supplementary Fig. 10F). In addition, we observed that knockdown of PCDH9 highly promoted colony formation in RWPE-1 cells (Supplementary Fig. 11). We further evaluated the effect of PCDH9 on tumor growth in vivo using subcutaneous transplantation of DU145 cells with stably overexpressed PCDH9 (DU145-PCDH9) or control (DU145-pReceiver) into nude mice. We found markedly reduced tumor volume and weight in the mice transplanted with DU145-PCDH9 (Fig. 4E and 4F). H&E straining revealed prostate tumors formed in mice (Fig. 4G). IHC analysis showed apparent overexpression of PCDH9 and reduction of proliferation marker Ki-67 [42] in the xenograft tumors derived from DU145-PCDH9 cells (Fig. 4G, Supplementary Fig. 12), suggesting that inhibited growth of PCDH9 overexpression tumors in vivo is due at least in part to reduced cell proliferation. To further understand the role of PCDH9 in PCa, we sought to identify potential downstream target genes of PCDH9. We thus performed genome-wide expression profiling of DU145-PCDH9 and DU145-pReceiver cells (Fig. 4H, Supplementary Table 19). We found that overexpression of PCDH9 downregulated oncogenic drivers of PCa such as MYB and STEAP1, and the highly prostatespecific HOXB13, and the cancer stem cell marker ALDH1A1, and upregulated tumor suppressors including FOXO4, EPHB2, and PBX1, and the epithelial-mesenchymal transition (EMT) marker CDH1 (Fig. 4H, Supplementary Figs. 13A and 13B). We also used the classic PCa cell line C4-2 to validate PCDH9 affected genes, which is consistent with the results from DU145 (Supplementary Fig. 13B). Intriguingly, querying several clinical PCa data sets [6,12,36,38], we observed inverse correlations between PCDH9 and MYB, STEAP1, or HOXB13 expression, and positive correlations between PCDH9 and FOXO4, EPHB2, or PBX1 (Figs. 13C–H, Supplementary Table 19), indicating that PCDH9 may impact these gene expression in the clinical setting. Together, these data suggest that PCDH9 plays a tumor suppressor role in PCa by suppressing oncogenic and enhancing tumor suppressive pathways. 3.6.

Frequently altered axon guidance pathway genes in PCa

We next investigated genes with genomic alterations (somatic mutations, copy number, and structural variations) and changed expression in known and novel pathways that impact PCa progression and aggressiveness. Consistent with previous studies [5], in our cohort we also found frequent alterations in genes present in phosphoinositide 3-kinase, retinoblastoma protein, RAS/RAF, and AR signaling pathways (Fig. 5A, Supplementary Table 20) that are known to be important in PCa [4]. To identify novel progression pathways in PCa, we mapped genes affected by SNVs, indels, CNVs, and SVs to canonical pathways and performed enrichment analysis for genes with different variation types. Notably, the axon guidance pathway genes were consistently found to be significantly altered by different variation types, particularly by SNVs and SVs (p < 0.01, hypergeometric test; Fig. 5B, Supplementary Table 21). We analyzed

genetic data from an independent cohort of PCa [18] and found that the axon guidance pathway was also significantly (p < 0.01, hypergeometric test) altered with SVdisrupted genes (Supplementary Table 21). These analyses suggest a novel link between frequent somatic alterations in axon guidance pathway genes and the pathogenesis of PCa, and a likely role of axon guidance pathway in PCa development. The axon guidance pathway consists of semaphorins, slits, netrins, and ephrins that were originally characterized as the guidance of axons during embryonic development. Recently, this pathway has been found to modulate cancer cell growth, survival, invasion, and angiogenesis [43,44]; however, somatic alterations in axon guidance pathway genes are not fully understood in PCa. Here we found frequent SV alterations in ROBO1 (15.38% [10/65] of patients) and SLIT2 (6.15% [4/65] of patients) involved in SLIT-ROBO signaling (Supplementary Table 22), which are similar to the mutation frequency observed previously in pancreatic cancer [43], suggesting that aberrant SLIT-ROBO signaling may be common in cancer. To examine the related gene expression of these genomic alterations, we next integrated genome rearrangement events in axon guidance pathway genes with corresponding RNA-seq data in our study cohort (Supplementary Fig. 14A). This analysis revealed that low expression of SLIT2 was markedly correlated with the prevalence of SVs in four tumors. Particularly in the CH33 tumor, we observed a high number of RNA-seq reads in the left region of the breakpoint and less reads in the right region. Low expression of SLIT2 protein in tumors was confirmed by IHC (Supplementary Fig. 14B). Given that SLIT2 copy numbers in tumor and normal tissues of sample CH33 were highly accordant, abnormal expression of SLIT2 was likely to be caused by intrachromosomal translocation. Together, whether SV alterations in SLIT-ROBO signaling impacted prostate carcinogenesis through disrupting gene expression warrants further investigation. We also observed frequent gain/amplifications (eg, PLXNA1, PLXNA3, PLXNB2, and PLXNB3) or deletions (eg, DPYSL2 [also termed collapsing response mediator protein-2, CRMP2] and FYN) in semaphorin signaling pathway genes in about 80% of tumors (Fig. 5C, Supplementary Table 22). To assess potential functional impact of these alterations in axon guidance pathway (especially semaphorin signaling) in our cohort, we thus performed a weighted gene expression correlation network analysis [45] of the dysregulated genes in axon guidance pathway and phosphoinositide 3-kinase/AKT, AR, nuclear factor-kB, and WNT signaling pathways. This analysis revealed that genes in the semaphorin signaling pathway showed coexpression pattern genes in the phosphoinositide 3-kinase/AKT pathway and AR signaling (Supplementary Fig. 15A), indicating the potential biological cooperation of these pathways in PCa. Moreover, it has been reported that activation of FYN kinase leads to phosphorylation of CRMP2 by CDK5, which primes further phosphorylation of CRMP2 by GSK3b and reduces affinity of CRMP2 for tubulin heterodimers, microtubule growth, and cell collapse [46,47]. Thus, it can

Please cite this article in press as: Ren S, et al. Whole-genome and Transcriptome Sequencing of Prostate Cancer Identify New Genetic Alterations Driving Disease Progression. Eur Urol (2017), http://dx.doi.org/10.1016/j.eururo.2017.08.027

EURURO-7536; No. of Pages 17 E U R O P E A N U R O L O G Y X X X ( 2 0 17 ) X X X – X X X

11

Fig. 5 – Frequently altered pathways in prostate cancer. (A) Four pathways reported to be important in prostate cancer were commonly altered in our cohort of patients. Overall frequency of alterations (including somatic mutations, copy number variations, structural variation, and up/downregulation of gene expression) is shown for each gene. (B) The top 10 enriched pathways with genes affected by somatic point mutations and top 10 enriched pathways with genes disrupted by structural variations. (C) Genetic alterations in components of the semaphorin signaling pathway. Each row represents a specific gene, and each column represents a tumor sample. Left panel, frequency of alterations in each gene analyzed in our cohort. Right panel, different colors in the matrix indicate individual alteration types as annotated. Tumor classification indicates the different stages of tumors. (D) A hypothetic model depicting the consequence of the deregulation of the semaphoring pathway, tipping from FYN and CRMP2 (DPYSL2) deletioncaused loss of microtubule growth and cell polarity to RAS/phosphoinositide 3-kinase (PI3K)/AKT-mediated cell proliferation and invasion. AR = androgen receptor; GS = Gleason score; SNV = single nucleotide variations; SV = structural variations.

be postulated that loss of FYN may lead to activation of CRMP2 and axon growth, but further deletion of CRMP2 may prohibit microtubule growth, resulting in loss of cell polarity, and cell dedifferentiation (Fig. 5D). 3.7.

High frequent PLXNA1 amplification in PCa

To evaluate the potential role of axon guide pathways in PCa, we focused on PLXNA1, an amplified upstream activator of the pathway (Fig. 5C, Supplementary Fig. 15B). We found that PLXNA1 knockdown attenuated the expression of AR target genes (Supplementary Fig. 15C), and androgeninduced AR transcriptional activity measured by luciferase

assays (Supplementary Fig. 15D) in the tested PCa cell lines. We also observed that knockdown of PLXNA1 inhibited nerve growth factor-induced AKT phosphorylation, while overexpression of PLXNA1 promoted AKT phosphorylation (Supplementary Fig. 15E). In addition, we noted that there was very scant basal expression of PLXNA1 protein in two tested AR-positive CRPC cell lines relative to PC3, an ARnegative CRPC cell model (Supplementary Fig. 15D). Consistent with this, we observed an increased expression of PLXNA1 in LNCaP treated with AR antagonist MDV3100 (Supplementary Fig. 16A). Moreover, the analysis of a large cohort of PCa dataset revealed a negative correlation between AR and PLXNA1 expression (Supplementary

Please cite this article in press as: Ren S, et al. Whole-genome and Transcriptome Sequencing of Prostate Cancer Identify New Genetic Alterations Driving Disease Progression. Eur Urol (2017), http://dx.doi.org/10.1016/j.eururo.2017.08.027

EURURO-7536; No. of Pages 17 12

E U R O P E A N U R O L O G Y X X X ( 2 0 17 ) X X X – X X X

Fig. 16B). In two additional independent large patient cohorts [9,48], we observed higher mRNA levels of PLXNA1 in neuroendocrine CRPC with attenuated AR expression and signaling than that in adenocarcinomas CRPC (Supplementary Fig. 16C and 16D), suggesting an association between AR signaling, transcriptional activity and PLXNA1 expression. Consistent with this, we observed a strong AR binding at PLXNA1 in PCa cells VCaP upon the stimulation of AR signaling in a publicly available AR ChIP-seq data [49] (Supplementary Fig. 16E). Together, these studies further support potential cross-talk between these pathways. Given the known roles of AKT and AR signaling in PCa development [4] and axon guidance pathway in many types of tumors [44], our results imply a previously uncharacterized function of semaphorin signaling gene PLXNA1 in PCa. Notably, PLXNA1 was one of the most frequently altered semaphorin signaling genes with alterations in 23% of our samples by gain/amplification and upregulation (Supplementary Table 22). Accordingly, PLXNA1 amplification was markedly associated with its increased expression (Fig. 6A and 6B) in two independent clinical data sets [12,50], which may lead to altered activity of PLXNA1 in PCa cells. Thus, we decided to investigate the functional roles of PLXNA1 in PCa. 3.8.

PLXNA1 alterations impact PCa progression and prognosis

To characterize the role of PLXNA1 in PCa, we first examined the effect of PLXNA1 knockdown on cellular phenotypes (Supplementary Fig. 17A). We observed that depletion of PLXNA1 greatly attenuated the proliferation, invasion and migration of all tested PCa cell lines (Fig. 6C–E, Supplementary Figs. 17B–E). Consistently, ectopic overexpression of PLXNA1 strikingly promoted the proliferation, invasion and migration of DU145 cells (Fig. 6F, Supplementary Fig. 17F), in line with observed high expression of PLXNA1 in aggressive PCa cell lines, including ARCaPM, and LNCaPRANKL (Fig. 6G). Both cell lines exhibit mesenchymal features with a high metastatic propensity to bone and soft tissue (Fig. 6G), and was therefore used to evaluate the effect of PLXNA1 knockdown on aggressive phenotypes of PCa cells. Given that EMT, stemness, and neuroendocrine phenotypes are associated with aggressive metastatic CRPC [4], we asked whether PLXNA1 knockdown affects the expression of markers associated with these aggressive phenotypes. Strikingly, PLXNA1 knockdown in LNCaPRANKL and ARCaPM cells resulted in reduced expression of the mesenchymal biomarkers vimentin, N-cadherin and fibronectin, and upregulation of E-cadherin, an epithelial marker (Fig. 6H). Moreover, we observed suppression of stem cell markers CD44, CD133, CD49f, OCT4, NANOG, SOX2, and LIN28B, and neuroendocrine markers CgA, SYP, and FOXA2 (Fig. 6H). Consistently, EMT, stemness, and neuroendocrine phenotypes were enhanced by ectopic overexpression of PLXNA1 in tested PCa cell lines (Supplementary Fig. 17G). We further evaluated the effect of PLXNA1 knockdown on tumor growth in vivo using xenograft mouse model. We found greatly decreased tumor volume and weight in the mice transplanted with short hairpin RNA (shRNA)mediated PLXNA1 knockdown cells (Fig. 6I). Consistent

results were obtained when overexpression of PLXNA1 significantly promoted the growth of xenograft tumors (Fig. 6I). Histopathological analysis of PLXNA1 knockdown tumors confirmed a marked suppression of tumor cell proliferation as assessed by Ki-67 immunostaining (Fig. 6J). Moreover, compared with controls, PLXNA1 knockdown tumors indicated reversal of EMT phenotype, characterized by increased E-cadherin staining, reduced expression of Ncadherin, vimentin and fibronectin, and repression of neuroendocrine phenotype in PCa indicated by weaker immunostaining of CgA and Syp (Fig. 6J), which is consistent with our observations in PCa cells with aggressive properties (Fig. 6G). Collectively, these data have established the importance of PLXNA1 for PCa cell growth and metastasis, and suggest a link between high PLXNA1 expression and prostate tumor aggressive phenotype, implying that highly frequent PLXNA1 alterations observed in our study cohort (Fig. 5C, Supplementary Table 22) might impact PCa progression and prognosis in a diagnostic setting. To assess clinical impact of PLXNA1 on human PCa progression, we first examined potential correlation between PLXNA1 expression and disease severity. We found that PLXNA1 expression is greatly upregulated upon PCa progression to metastatic stage (Fig. 7A, Supplementary Fig. 18A–D) in multiple clinical data sets [9,37,51]. Furthermore, PLXNA1 upregulation is significantly correlated with elevated prostate-specific antigen level and high clinical stage of PCa patients [6,52] (Supplementary Fig. 18E–G). Immunostaining of PLXNA1 in an independent cohort of 87 primary PCa samples showed that increased PLXNA1 expression was associated with high Gleason score, advanced tumor stage, and early biochemical recurrence after prostatectomy (Fig. 7B). The latter observation indicates poorer prognosis of PCa patients with higher levels of PLXNA1. Consistent with this finding, by querying from five independent clinical PCa datasets [6,24,38–40] (Supplementary Figs. 18H–L), we also observed higher mRNA levels of PLXNA1 were significantly associated with increased frequency of biochemical relapse in three independent American cohorts, but not in two European datasets [6,38,39] (Supplementary Figs. 18H–J). Whether the different outcome in European cohorts is due to the difference in sample composition (eg, metastasis vs nonmetastasis) or treatment regimen warrants further investigation. To further assess the prognostic potential of PLXNA1 in PCa, we performed immunostaining of PLXNA1 on two independent tissue microarrays of samples with long-term clinical follow-up data from Chinese PCa Consortium (n = 419, 4 institutions, 10-yr follow-up) and Massachusetts General Hospital (n = 213, 20-yr follow-up), respectively. Strikingly, strong PLXNA1 expression showed associations not only with biochemical recurrence but also with metastasis-free and overall survival (Fig. 7C and 7D). Notably, multivariate and univariate regression analyses revealed that PLXNA1 upregulation was an independent prognostic variable of PCa for predicting biochemical recurrence, metastasis-free and overall survival in both cohorts (Fig. 7C and 7D, Supplementary Table 23). We also observed a better prognosis usefulness by adding PLXNA1

Please cite this article in press as: Ren S, et al. Whole-genome and Transcriptome Sequencing of Prostate Cancer Identify New Genetic Alterations Driving Disease Progression. Eur Urol (2017), http://dx.doi.org/10.1016/j.eururo.2017.08.027

EURURO-7536; No. of Pages 17 E U R O P E A N U R O L O G Y X X X ( 2 0 17 ) X X X – X X X

13

Fig. 6 – Functional validation of PLXNA1 in prostate cancer cell growth and tumor progression. (A,B) A significant correlation between PLXNA1 amplification and overexpression was observed in two independent cohorts of human prostate cancer samples. Mann-Whitney U-tests were performed to evaluate statistical significance for the comparisons between groups. (C) Decreased cell proliferation was observed in the tested prostate cell line PC3 via CCK-8 assay at the indicated time point. (D) Quantitation of cell invasion following transfection with nontargeting small interfering RNA (siRNA) or siRNAs against PLXNA1. (E) Quantitation of cell migration following transfection with non-targeting siRNA or PLXNA1 siRNA. (F) Ectopic

Please cite this article in press as: Ren S, et al. Whole-genome and Transcriptome Sequencing of Prostate Cancer Identify New Genetic Alterations Driving Disease Progression. Eur Urol (2017), http://dx.doi.org/10.1016/j.eururo.2017.08.027

EURURO-7536; No. of Pages 17 14

E U R O P E A N U R O L O G Y X X X ( 2 0 17 ) X X X – X X X

expression to the Cancer of the Prostate Risk Assessment Postsurgical score (Supplementary Fig. 19). Together, these analyses suggest that PLXNA1 is a new biomarker that distinguishes aggressive disease. Finally, given that PLXNA1 gain/amplification correlates with its expression and that PLXNA1 expression in turn correlates with PCa aggressiveness, we investigated whether PLXNA1 copy gain directly correlates with relapse status of patients with PCa. This analysis revealed that PLXNA1 copy number gain was highly frequent in metastatic PCa and CRPC patients [6] (Fig. 7E, Supplementary Fig. 18M), and in primary PCa tumor samples with near 50% rate of biochemical relapse by querying the camcAPP database [24,40] (Supplementary Fig. 18N). In addition, our analysis of a TCGA cohort of PCa showed that PCa patients with copy number gain of PLXNA1 had significantly higher risk of biochemical recurrence than the patients having tumors with no copy number changes of PLXNA1 [38] (Fig. 7F). Altogether, our results indicate that PLXNA1 alterations including amplification and overexpression could confer tumor growth and progression, and worse prognosis of the patients with PCa. 4.

Discussion

Emerging evidence indicates that there are remarkable disparities in PCa epidemiology among different ethnic groups [28,53]. However, the underlying molecular mechanisms remain largely unknown. Notably, recent studies consistently show that the incidence of ETS family gene fusions is much lower (ranging from 8% to 21%) in PCa in Asian men compared with the prevalence of approximately 50% in Caucasian patients from Western countries [26– 30]. In agreement with these findings, we detected a low prevalence (9.2%) of ETS fusions in our cohort. Previous studies demonstrate that the tumor suppressor CHD1 is required for ERG rearrangement in PCa [30] and that CHD1 deletion is almost always correlated with ETS-negative PCa [9]. Compared with the frequency (16%) of CHD1 deletion in the TCGA dataset [12], we observed a much higher prevalence (31%) of CHD1 deletion in the Chinese cohort we studied. Thus, our discovery of the inverse correlation of CHD1 deletion and ETS fusions between Chinese and Caucasian populations provides a plausible explanation as to why the rate of ETS fusions is lower in Chinese PCa patients.

A recent study found that the parasympathetic nervous system and sympathetic nervous system played key roles in triggering PCa and influencing metastasis [54]. Nerves of the sympathetic nervous system and parasympathetic nervous system promote tumor growth by producing norepinephrine and acetylcholine, which activate a signaling pathway within stromal cells in tumor microenvironment [54]. However, the mechanism by which nervous system directly affects PCa cell growth remains largely unknown. Axon guidance pathways are critical for neural development. It has been shown previously that activation of PLXNA2 by semaphoring 3A leads to activation of the FYN-CDK5 pathways, which not only causes reduced microtubule growth in the distal end of axon and growth cone collapse by inducing GSK3b-mediated phosphorylation of CRMP2 and lost affinity of CRMP2 for tubulin heterodimers, but also promotes inhibition of the RAC-PAK pathway [46]. Notably, we found that both FYN and CRMP2 genes were frequently deleted in our cohort and TCGA patients. These findings suggest that coordinated deletion of FYN and CRMP2 signaling may lead to a loss of cell polarity and dedifferentiation of prostatic cells, and further investigation of this concept is warranted. Additionally, our weighted gene expression correlation network analysis showed that amplification of PLXNA1 and deletion of the FYN-CRMP2 signaling axis correlates with deregulation of the PI3K/AKT and AR pathways, activation of which was validated in PLXNA1-overexpressed PCa cells. Interestingly, our analyses revealed potential involvement of axon guidance pathway in PCa progression. We identified frequent and diverse somatic aberrations in genetic components of the axon guidance pathway. Our experimental evidence from gene and protein expression, in vitro and in vivo assessments indicate that axon guidance pathway gene PLXNA1 is involved in PCa aggressiveness. Patients with higher level of PLXNA1 expression in PCa are at markedly higher risk of having biochemical relapse and decreased metastasis-free and overall survival in multiple independent cohorts. Given that invasion- and metastasis-inhibiting and antiangiogenic agents such as anti-PLXNB1 and UnclSema3E (binds to PLXND1) antibodies have been reported [55], PLXNA1 is expressed on the cell surface and may therefore be easily targeted by these agents. Further studies will aim to define PLXNA1 as a potential marker to riskstratify patients with PCa and to investigate PLXNA1 protein as an effective therapeutic target to treat advanced PCa.

overexpression of PLXNA1 dramatically promoted PCa cell proliferation, invasion and migration. (C–F) All experiments were performed in triplicate. Error bars, standard deviation. Results were statistically examined by two-tailed t-test. (G) Immunohistochemistry analysis of PLXNA1 expression in human prostate cancer cell lines ARCaPM and LNCaPRANK. PLXNA1 expression in murine xenograft tumors induced by injecting these cells either subcutaneously or orthotopically and bone metastases. (H) Reversal of EMT and decreased expression of the stem cell marker and neuroendocrine phenotypes in LNCaPRANKL and ARCaPM following siRNA-mediated knockdown of PLXNA1. The quantitative reverse transcription-polymerase chain reaction results were shown as mean  standard deviation. of triplicate values for each sample. (I) Growth curves for subcutaneous xenografts derived from C4-2 (n = 6 mice per group) and PC-3 cells (n = 6 mice per group) stably expressing a short hairpin RNA (shRNA) against PLXNA1 (shPLXNA1) or scrambled shRNA (shControl). Bottom panel indicates growth curves for subcutaneous xenografts (n = 4 mice per group) derived from DU145 with stable expression of PLXNA1 or control vector. Scatter plots (right) show tumor weights from individual animals in each group. (J) Hematoxylin and eosin staining and immunohistochemistry analysis of Ki-67, PLXNA1, E-cadherin, N-cadherin, vimentin, fibronectin, SYN, NSE, and CgA expression in PC-3 cell-derived xenograft tumors. Representative images from six separate tumors are shown. Original magnification, 400T; scale bar 20 mm. OD = optical density; TCGA = The Cancer Genome Atlas. * p < 0.05. ** p < 0.01. *** p < 0.001.

Please cite this article in press as: Ren S, et al. Whole-genome and Transcriptome Sequencing of Prostate Cancer Identify New Genetic Alterations Driving Disease Progression. Eur Urol (2017), http://dx.doi.org/10.1016/j.eururo.2017.08.027

EURURO-7536; No. of Pages 17 E U R O P E A N U R O L O G Y X X X ( 2 0 17 ) X X X – X X X

15

Fig. 7 – PLXNA1 upregulation in human prostate cancer (PCa) correlates with tumor progression and the risk of biochemical relapse and reduced survival. (A) PLXNA1 messenger RNA expression is strikingly upregulated in metastatic prostate tumors from a clinical PCa dataset. (B) Association of PLXNA1 overexpression with Gleason grade, tumor stage, and biochemical recurrence. The p values were assessed using chi-square test. Kaplan-Meier curves show the risk of biochemical recurrence after radical prostatectomy for the patient groups stratified by PLXNA1 protein expression levels determined by immunostaining. The p value was determined using log-rank test. (C,D) Kaplan-Meier analysis of the risk of biochemical recurrence and

Please cite this article in press as: Ren S, et al. Whole-genome and Transcriptome Sequencing of Prostate Cancer Identify New Genetic Alterations Driving Disease Progression. Eur Urol (2017), http://dx.doi.org/10.1016/j.eururo.2017.08.027

EURURO-7536; No. of Pages 17 16

5.

E U R O P E A N U R O L O G Y X X X ( 2 0 17 ) X X X – X X X

Conclusions

Innovative Research Team Program (NO. 2009010016, Yang); Academy of Finland (284618 & 279760, Wei), University of Oulu Strategic funds, Jane

In summary, we report the first comprehensive mutational landscape of PCa in Chinese men, which will provide an invaluable resource for in depth comparison of genomic alterations in PCa across ethnic groups. Our integrated analysis of whole-genome and transcriptome data from the same patient reveals potential disease relevant mutations and copy number alterations in AR pathway genes, cell adhesion molecules such as PCDH9, and axon guidance pathway genes such as amplification of PLXNA1 and deletion of FYN and CRMP2. While further functional assessment of these genetic alterations is warranted, our findings contribute to the understanding, prediction, prognosis, and precise treatment of PCa in men with vast ethnic disparity. Author contributions: Yinghao Sun had full access to all the data in the study

and Aatos Erkko Foundation, and Finnish Cancer Foundation grants (Wei). Acknowledgments: This manuscript is dedicated to the memory of our wonderful colleague, the excellent research scientist, Prof. Changjun Yin in recognition of his immense contributions to Chinese Urology as well as this project. We thank Dr. Laurie Goodman with the help in revision and editing of the manuscript, Professor Michael Dean and Professor Qiang Pan-Hammarström for valuable suggestions. We thank all the lab members for helpful discussion. Whole Genome and whole transcriptome sequencing data has been deposited in The European Genomephenome Archive (EGAS00001000888).

Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j. eururo.2017.08.027.

and takes responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Sun, Ren, Wei, Hou. Acquisition of data: Ren, Wei, Liu, Wang, Yin, Gao, Xu, Ye, Xu, Gao, Zhou, Yang, Hou, Zhang, Zhu, Qin, Shao, Pang, Huang, Sun. Analysis and interpretation of data: Ren, Wei, Liu, Wang, Hou, Liu, Cheng, Zhang, Lee, Yang, Yu, Zhu, Qiao, Zhu, Shi, Chen, Wang, Xu, Cheng, Zhau, Chu, Wu, Wang, Peng, Zhou, Zhang, Su, Zhao, Yin, He, Wu, Li, Zheng, Collins, Volik, Bell, Huang, Xu, Huang. Drafting of the manuscript: Ren, Wei, Liu, Hou, Huang. Critical revision of the manuscript for important intellectual content: Ren, Wei, Liu, Wang, Wu, Zhong, Huang, Sun. Statistical analysis: Ren, Wei, Wang, Liu, Xu, Huang. Obtaining funding: Ren, Wei, Yang, Wang, Sun. Administrative, technical, or material support: Ren, Wei, Liu, Hou, Huang, Zhang, Peng, Zhou, Zhang, Su. Supervision: Ren, Wei, Xu, Li, Zhang, Wang, Yang, Wang, Huang, Sun. Other: None. Financial disclosures: Yinghao Sun certifies that all conflicts of interest, including specific financial interests and relationships and affiliations relevant to the subject matter or materials discussed in the manuscript (eg, employment/affiliation, grants or funding, consultancies, honoraria, stock ownership or options, expert testimony, royalties, or patents filed, received, or pending), are the following: None.

References [1] Torre LA, Bray F, Siegel RL, Ferlay J, Lortet-Tieulent J, Jemal A. Global cancer statistics, 2012. CA Cancer J Clin 2015;65:87–108. [2] Center MM, Jemal A, Lortet-Tieulent J, et al. International variation in prostate cancer incidence and mortality rates. Eur Urol 2012;61:1079–92. [3] Chen W, Zheng R, Baade PD, et al. Cancer statistics in China, 2015. CA Cancer J Clin 2016;66:115–32. [4] Shen MM, Abate-Shen C. Molecular genetics of PCa: new prospects for old challenges. Genes Dev 2010;24:1967–2000. [5] Tomlins SA, Rhodes DR, Perner S, et al. Recurrent fusion of TMPRSS2 and ETS transcription factor genes in PCa. Science 2005;310:644–8. [6] Taylor BS, Schultz N, Hieronymus H, et al. Integrative genomic profiling of human PCa. Cancer Cell 2010;18:11–22. [7] Baca SC, Prandi D, Lawrence MS, et al. Punctuated evolution of PCa genomes. Cell 2013;153:666–77. [8] Barbieri CE, Baca SC, Lawrence MS, et al. Exome sequencing identifies recurrent SPOP, FOXA1, and MED12 mutations in PCa. Nat Genet 2012;44:685–9. [9] Grasso CS, Wu YM, Robinson DR, et al. The mutational landscape of lethal castration-resistant PCa. Nature 2012;487:239–43. [10] Kumar A, White TA, MacKenzie AP, et al. Exome sequencing identifies a spectrum of mutation frequencies in advanced and lethal prostate cancers. Proc Natl Acad Sci U S A 2011;108:17087–92.

Funding/Support and role of the sponsor: This work was supported by the Program for Changjiang Scholars and Innovative Research Team in University scheme of the Ministry of Education of China (NO.IRT1111, Sun); National Basic Research Program of China (2012CB518300, Sun; NO.2011CB8092, 2011CB809203, Wang); National High Technology

[11] Robinson D, Van Allen EM, Wu YM, et al. Integrative clinical genomics of advanced PCa. Cell 2015;161:1215–28. [12] Cancer Genome Atlas Research Network. The molecular taxonomy of primary PCa. Cell 2015;163:1011–25. [13] Fraser M, Sabelnykova YV, Yamaguchi TN, et al. Genomic hallmarks

Research and Development Program of China (NO.2012AA02A201,

of localized, non-indolent prostate cancer. Nature 2017;541:359–64.

Zhang); National Natural Science Foundation of China (81472397, Ren);

[14] Li H, Handsaker B, Wysoker A, et al. The sequence Alignment/Map

Shanghai Pujiang Program (12PJD008, Ren); PCa Foundation Young Investigator Award (Ren); Shanghai Municipal Health and Family

format and SAMtools. Bioinformatics 2009;25:2078–9. [15] Langmead B, Trapnell C, Pop M, Salzberg SL. Ultrafast and memory-

Planning Commission Outstanding Young Investigator (XYQ2013077,

efficient alignment of short DNA sequences to the human genome.

Ren); Shanghai Municipal Education Commission (Ren); Guangdong

Genome Biol 2009;10:R25.

reduced progression/metastasis-free and overall survival in prostate tumors with high or low protein levels of PLXNA1 in two independent cohorts of PCa cases. (C,D) The p values were assessed by a log-rank test and Cox regression analysis, respectively. (C) Numbers of patients in each category were indicated for each 20-mo interval or (D) 2-yr interval. (E) Portion of tumors harboring PLXNA1 copy number amplification was significantly higher in metastasis than in primary PCa. The p value was determined by two-sided Fisher's exact test. (F) Kaplan-Meier analysis of biochemical recurrence in two patient groups carrying the tumors with PLXNA1 copy gain (copy number as 1) or without PLXNA1 gain (copy number as 0). Note that, PLXNA1 copy gain is significantly associated with early arrival of PCa biochemical relapse. The p value was evaluated by log-rank test. CPCC = Chinese Prostate Cancer Consortium; GS = Gleason score; MGH = Massachusetts General Hospital; PSA = prostate-specific antigen; TCGA = The Cancer Genome Atlas.

Please cite this article in press as: Ren S, et al. Whole-genome and Transcriptome Sequencing of Prostate Cancer Identify New Genetic Alterations Driving Disease Progression. Eur Urol (2017), http://dx.doi.org/10.1016/j.eururo.2017.08.027

EURURO-7536; No. of Pages 17 E U R O P E A N U R O L O G Y X X X ( 2 0 17 ) X X X – X X X

[16] Trapnell C, Williams BA, Pertea G, et al. Transcript assembly and

17

[37] Tamura K, Furihata M, Tsunoda T, et al. Molecular features of

quantification by RNA-seq reveals unannotated transcripts and

hormone-refractory PCa cells by genome-wide gene expression

isoform switching during cell differentiation. Nat Biotechnol

profiles. Cancer Res 2007;67:5117–25.

2010;28:511–5. [17] Youn A, Simon R. Identifying cancer driver genes in tumor genome sequencing studies. Bioinformatics 2011;27:175–81. [18] Weischenfeldt J, Simon R, Feuerbach L, et al. Integrative genomic

[38] Cerami E, Gao J, Dogrusoz U, et al. The cBio cancer genomics portal: an open platform for exploring multidimensional cancer genomics data. Cancer Discov 2012;2:401–4. [39] Glinsky GV, Glinskii AB, Stephenson AJ, Hoffman RM, Gerald WL.

analyses reveal an androgen-driven somatic alteration landscape in

Gene expression profiling predicts clinical outcome of PCa. J Clin

early-onset PCa. Cancer Cell 2013;23:159–70.

Invest 2004;113:913–23.

[19] Berger MF, Lawrence MS, Demichelis F, et al. The genomic complexity of primary human PCa. Nature 2011;470:214–20. [20] Nik-Zainal S, Alexandrov LB, Wedge DC, et al. Mutational processes molding the genomes of 21 breast cancers. Cell 2012;149:979–93. [21] Robbins CM, Tembe WA, Baker A, et al. Copy number and targeted mutational analysis reveals novel somatic events in metastatic prostate tumors. Genome Res 2011;21:47–55.

[40] Dunning MJ, Vowler SL, Lalonde E, et al. Mining human prostate cancer datasets: The “camcAPP” Shiny App. EBioMedicine 2017;17: 5–6. [41] Setlur SR, Mertz KD, Hoshida Y, et al. Estrogen-dependent signaling in a molecularly distinct subclass of aggressive prostate cancer. J Natl Cancer Inst 2008;100:815–25. [42] Ramos-Montoya A, Lamb AD, Russell R, et al. HES6 drives a critical

[22] Govindan R, Ding L, Griffith M, et al. Genomic landscape of non-

AR transcriptional programme to induce castration-resistant pros-

small cell lung cancer in smokers and never-smokers. Cell

tate cancer through activation of an E2F1-mediated cell cycle

2012;150:1121–34. [23] Rubin MA, Maher CA, Chinnaiyan AM. Common gene rearrangements in PCa. J Clin Oncol 2011;29:3659–68. [24] Ross-Adams H, Lamb AD, Dunning MJ, et al. Integration of copy

network. EMBO Mol Med 2014;6:651–61. [43] Biankin AV, Waddell N, Kassahn KS, et al. Pancreatic cancer genomes reveal aberrations in axon guidance pathway genes. Nature 2012;491:399–405.

number and transcriptomics provides risk stratification in prostate

[44] Neufeld G, Kessler O. The semaphorins: Versatile regulators of

cancer: a discovery and validation cohort study. EBioMedicine

tumour progression and tumour angiogenesis. Nat Rev Cancer

2015;2:1133–44. [25] Boyd LK, Mao X, Lu YJ. The complexity of PCa: genomic alterations and heterogeneity. Nat Rev Urol 2012;9:652–64. [26] Lee K, Chae JY, Kwak C, Ku JH, Moon KC. TMPRSS2-ERG gene fusion and clinicopathologic characteristics of Korean PCa patients. Urology 2010;76:1268.e7–13. [27] Magi-Galluzzi C, Tsusuki T, Elson P, et al. TMPRSS2-ERG gene fusion

2008;8:632–45. [45] Eisen MB, Spellman PT, Brown PO, Botstein D. Cluster analysis and display of genome-wide expression patterns. Proc Natl Acad Sci U S A 1998;95:14863–8. [46] Sasaki Y, Cheng C, Uchida Y, et al. Fyn and Cdk5 mediate semaphorin-3A signaling, which is involved in regulation of dendrite orientation in cerebral cortex. Neuron 2002;35:907–20.

prevalence and class are significantly different in PCa of Caucasian,

[47] Nasarre P, Gemmill RM, Drabkin HA. The emerging role of class-3

African-American, and Japanese patients. Prostate 2011;71:489–97.

semaphorins and their neuropilin receptors in oncology. Onco

[28] Mao X, Yu Y, Boyd LK, et al. Distinct genomic alterations in prostate

Targets Ther 2014;7:1663–87.

cancers in Chinese and western populations suggest alternative

[48] Beltran H, Prandi D, Mosquera JM, et al. Divergent clonal

pathways of prostate carcinogenesis. Cancer Res 2010;70:5207–12.

evolution of castration-resistant neuroendocrine PCa. Nat Med

[29] Miyagi Y, Sasaki T, Fujinami K, et al. ETS family-associated gene fusions in Japanese PCa: analysis of 194 radical prostatectomy samples. Mod Pathol 2010;23:1492–8. [30] Ren S, Peng Z, Mao JH, et al. RNA-seq analysis of PCa in the Chinese population identifies recurrent gene fusions, cancer-associated

2016;22:298–305. [49] Massie CE, Lynch A, Ramos-Montoya A, et al. The androgen receptor fuels prostate cancer by regulating central metabolism and biosynthesis. EMBO J 2011;30:2719–33. [50] Kumar A, Coleman I, Morrissey C, et al. Substantial interindividual

long noncoding RNAs and aberrant alternative splicings. Cell Res

and limited intraindividual genomic diversity among tumors from

2012;22:806–21.

men with metastatic PCa. Nat Med 2016;22:36–978.

[31] Burkhardt L, Fuchs S, Krohn A, et al. CHD1 is a 5q21 tumor suppressor required for ERG rearrangement in PCa. Cancer Res 2013;73:2795–805. [32] Chen JD, Evans RM. A transcriptional co-repressor that interacts with nuclear hormone receptors. Nature 1995;377:454–7.

[51] Chandran UR, Ma C, Dhir R, et al. Gene expression profiles of PCa reveal involvement of multiple molecular pathways in the metastatic process. BMC Cancer 2007;7:64. [52] Vanaja DK, Cheville JC, Iturria SJ, Young CY. Transcriptional

[33] Lalonde E, Ishkanian AS, Sykes J, et al. Tumour genomic and

silencing of zinc finger protein 185 identified by expression

microenvironmental heterogeneity for integrated prediction of 5-

profiling is associated with PCa progression. Cancer Res 2003;63:

year biochemical recurrence of prostate cancer: a retrospective cohort study. Lancet Oncol 2014;15:1521–32. [34] Morishita H, Yagi T. Protocadherin family: diversity, structure, and function. Curr Opin Cell Biol 2007;19:584–92. [35] Wang C, Yu G, Liu J, et al. Downregulation of PCDH9 predicts prognosis for patients with glioma. J Clin Neurosci 2012;19:541–5. [36] Penney KL, Sinnott JA, Tyekucheva S, et al. Association of PCa risk

3877–82. [53] Zhau HE, Li Q, Chung LW. Interracial differences in PCa progression among patients from the united states, china and japan. Asian J Androl 2013;15:705–7. [54] Magnon C, Hall SJ, Lin J, et al. Autonomic nerve development contributes to PCa progression. Science 2013;341:1236361. [55] Rehman M, Tamagnone L. Semaphorins in cancer: Biological mech-

variants with gene expression in normal and tumor tissue. Cancer

anisms and therapeutic approaches. Semin Cell Dev Biol 2013;24:

Epidemiol Biomarkers Prev 2015;24:255–60.

179–89.

Please cite this article in press as: Ren S, et al. Whole-genome and Transcriptome Sequencing of Prostate Cancer Identify New Genetic Alterations Driving Disease Progression. Eur Urol (2017), http://dx.doi.org/10.1016/j.eururo.2017.08.027