Derived Xenografts Highlights the Role of REST in Neuroendocrine Differentiation of Castration- Resistant Prostate Cancer Amilcar Flores-Morales1,2,3, Tobias B

Published OnlineFirst October 1, 2018; DOI: 10.1158/1078-0432.CCR-18-0729

Biology of Human Tumors Clinical Cancer Research Proteogenomic Characterization of Patient- Derived Xenografts Highlights the Role of REST in Neuroendocrine Differentiation of Castration- Resistant Prostate Cancer Amilcar Flores-Morales1,2,3, Tobias B. Bergmann1,2, Charlotte Lavallee1,2, Tanveer S. Batth3, Dong Lin4, Mads Lerdrup5, Stine Friis1,2, Anette Bartels1,2, Gitte Kristensen6, Agnieszka Krzyzanowska7, Hui Xue4, Ladan Fazli4, Klaus H. Hansen5, Martin A. Røder6, Klaus Brasso6, Jose M. Moreira1,2, Anders Bjartell7, Yuzhuo Wang4, Jesper V. Olsen3, Colin C. Collins4, and Diego Iglesias-Gato1,2

Abstract

Background: An increasing number of castration-resistant and elevated in NEPC, while the reduced levels of proteins prostate cancer (CRPC) tumors exhibit neuroendocrine (NE) involved in mitochondrial functions suggested a prevalent features. NE prostate cancer (NEPC) has poor prognosis, and glycolytic metabolism of NEPC tumors. Integration of the its development is poorly understood. REST chromatin bound regions with expression changes Experimental Design: We applied mass spectrometry– indicated a direct role of REST in regulating neuronal gene based proteomics to a unique set of 17 prostate cancer expression in prostate cancer cells. Mechanistically, deple- patient–derived xenografts (PDX) to characterize the effects tion of REST led to cell-cycle arrest in G1, which could be of castration in vivo, and the proteome differences between rescued by p53 knockdown. Finally, the expression of the NEPC and prostate adenocarcinomas. Genome-wide profiling REST-regulated gene secretagogin (SCGN) correlated with of REST-occupied regions in prostate cancer cells was corre- an increased risk of suffering disease relapse after radical lated to the expression changes in vivo to investigate the role of prostatectomy. the transcriptional repressor REST in castration-induced NEPC Conclusions: This study presents the first deep characteri- differentiation. zation of the proteome of NEPC and suggests that concom- Results: An average of 4,881 proteins were identified and itant inhibition of REST and the p53 pathway would promote quantified from each PDX. Proteins related to neurogenesis, NEPC. We also identify SCGN as a novel prognostic marker in cell-cycle regulation, and DNA repair were found upregulated prostate cancer. Clin Cancer Res; 1–14. 2018 AACR.

Introduction androgen receptor (AR) blockage and are driven by complex mechanisms that involve genetic, epigenetic, and posttranscrip- Death in prostate cancer patients mostly occurs after the devel- tional changes (1, 2). Extensive studies have been carried out to opment of castration-resistant metastatic disease (CRPC). Castra- identify tumor driving genetic alterations in CRPC, but a unified tion-resistant tumors develop after androgen ablation therapy or genetic model of disease progression is yet to emerge. Amplifi- cation of the AR gene locus or point mutations in the AR gene occurs in 60% of tumors that have been subjected to androgen 1Department of Drug Design and Pharmacology, Faculty of Health and Medical ablation therapy (3). The appearance of ligand-independent AR Sciences, University of Copenhagen, Copenhagen, Denmark. 2The Danish Can- fi 3 splice variants or AR posttranslational modi cation may also cer Society, Copenhagen, Denmark. Novo Nordisk Foundation Center for contribute to continued AR activation following castration (4). Protein Research, Faculty of Health and Medical Sciences, University of Copen- hagen, Copenhagen, Denmark. 4Vancouver Prostate Centre, University of British Other common genetic alterations in AR-expressing CRPC Columbia, Vancouver, British Columbia, Canada. 5Biotech Research and Inno- include heterozygous deletions of PTEN and, to a lesser extent, vation Center, University of Copenhagen, Copenhagen, Denmark. 6Copenhagen p53 inactivation or loss of Rb1 expression (5–8). Prostate Cancer Center, Department of Urology, Rigshospitalet, University of One recurrent cellular feature of CRPC is the presence of 7 Copenhagen, Copenhagen, Denmark. Department of Translational Medicine, carcinoma cells that exhibit a neuroendocrine phenotype. Neu- Division of Urological Cancers, Lund University, Lund, Sweden. roendocrine differentiation (NED) of prostate cells is frequently Note: Supplementary data for this article are available at Clinical Cancer associated with highly proliferative potential of tumors and Research Online (http://clincancerres.aacrjournals.org/). increased chemoresistance, and as a consequence understanding Corresponding Author: Diego Iglesias-Gato, University of Copenhagen, The the development of NED in CRCP can lead to the development of Danish Cancer Society. Strandboulevarden 49Strand, Copenhagen 2100, new therapeutic strategies for CRPC. Neuroendocrine prostate Denmark. Phone: 4535330640; Fax: 4535325001; E-mail: cancer (NEPC) constitutes a distinct class of castration-resistant [email protected] tumors characterized by the expression of neuronal markers such doi: 10.1158/1078-0432.CCR-18-0729 as Chromogranin A (CHGA), Synaptophysin (SYP), and neural 2018 American Association for Cancer Research. cell adhesion molecule (CD56; ref. 9). NEPCs show similarities to

www.aacrjournals.org OF1

Flores-Morales et al.

Radical prostatectomy tissue microarray Translational Relevance The study cohort is composed of a consecutive series of men The number of castration-resistant prostate tumors exhibit- (n ¼ 336) with clinically localized prostate cancer who underwent ing neuroendocrine features is increasing, due in part to the radical prostatectomy (RP) with curative intent from January 1, improvement of drugs targeting the androgen receptor. Neu- 2002, until December 31, 2005, at the Department of Urology, roendocrine prostate cancer has poor prognosis due to the Rigshospitalet, Copenhagen, Denmark. This study was approved limited efficacy of currently available therapies. In this study, by the Danish National Committee on Health Research Ethics for we provide the first deep characterization of the proteome of the Capital Region (Journal no.: H-6-2014-111). Biochemical castration-resistant neuroendocrine prostate cancer. Proteins failure (BF) was defined as PSA 0.2 mg/L. The collection of are the ultimate targets of most anticancer drugs; therefore, our data was approved by The Danish Data Protection Agency work constitutes a valuable resource for the development of (file#2006-1-6256). The validation cohort (Malmo)€ was previ- novel compounds targeting proteins with increased expression ously described (18, 21). All studies were conducted following the in these tumors. Moreover, we provide evidence that early principles of the Declaration of Helsinki. All participants gave expression of the neuron-specific protein secretagogin in early written or verbal consent. Verbal consent was documented by the stages of prostate cancer correlates with increased risk of physician in the patient journal. biochemical recurrence after radical prostatectomy and could be accounted for in patients managed with conservative pro- IHC and statistics tocols such as active surveillance. TMA sections were deparaffinized in xylene and rehydrated through graded ethanol- Antigen retrieval performed at pH6. IHC staining was performed using anti SCGN antibody (HPA006641, Sigma-Aldrich, diluted 1:250). The SCGN staining was quantified small-cell carcinoma of the prostate (SCCP), which accounts for by scoring the intensity and the percentage of stained tumor cells 0.5% to 2% of all primary prostatic malignancies (10). Patients (0 ¼ negative or positive only in scattered neuroendocrine pros- diagnosed with SCCP have poor prognosis, with a median sur- tate cells; 1 ¼ weak staining in less than 30% for the tumor cells; vival time ranging from 5 to 17.5 months and the 5-year survival 2 ¼ intense staining in less than 20% of the cells on a localized rate of 14.3% (11). Moreover, in patients with locally advanced focus; 3 ¼ moderated staining in more than 50% of the cells or tumors treated with androgen ablation therapy (alone or in a intense staining in less than 50%; 4 ¼ intense staining in more coadjuvant setting), and in patients with CRPC, the extent of NED than 50% of the cells). Of all cores obtained from the same has been shown to be an independent prognostic factor (12). specimen, the highest expression was selected. Scores 2 were The molecular mechanisms driving the development of NE considered as high SCGN expression. The association between tumors after castration therapy are yet to be fully understood. The SCGN expression and clinicopathologic variables was analyzed loss of AR expression, the identification of recurrent mutations in using c2 test or Fisher exact test for categorical variables and the tumor suppressor genes such as Rb1 and p53, and the amplifi- Mann–Whitney U test for continuous variables. The median time cation of genes such as Aurora Kinase A (AURKA) and MYCN are of follow-up was calculated using the reverse Kaplan–Meier thought to play important roles in NE transdifferentiation and the method (22). Follow-up for BF was calculated until the latest aggressiveness of prostate cancer (13–17). Additionally, we PSA measurement, whereas time to death was calculated until the recently demonstrated that inactivation of the transcriptional latest follow-up date. Cumulative incidences of study endpoints repressor REST in prostate cancer cells results in the enhanced were analyzed using the Aalen–Johansen method for competing expression of a cluster of neuronal-specific genes including the NE risks. Death before BF was treated as competing event when markers CHGA and SYP. AR signaling enhances REST stability analyzing risk of BF. Other causes of mortality were treated as through the inhibition of the E3 ubiquitin ligase b-TrcP, which competing events when analyzing risk of prostate cancer–specific might explain why androgen ablation therapy can, in some cases, death. Gray test was used to assess differences in the cumulative promote NE transdifferentiation (18). Recently, REST has been incidences between biomarker subgroups (23). suggested as regulator of the epithelial-to-mesenchymal transi- Univariate and multivariate cause-specific Cox proportional tion of hormone-refractory prostate cancer (19). Whether REST hazard regression models were performed for risk of BF contributes to other aspects of the NE phenotype in prostate and prostate cancer–specific death. All tests were two-sided, and cancer, other than regulating the expression of neuron-specific P < 0.05 was considered to be statistically significant. Statistical genes, has not been studied in depth. analyses were performed using SPSS (software version 22; IBM), R In this study, we characterized the proteome changes occurring (R Development Core Team, Vienna, Austria), or GraphPad during the differentiation from adenocarcinoma to NEPC in PRISM. patient-derived xenograft (PDX) models and investigated the role For the evaluation of the public MCKCC cohort (24), high of REST in this process by matching in vivo expression changes to SCGN tumors were defined as those with expression above the REST-occupied chromatin regions. Functional studies also mean expression level of SCGN in nonmalignant neighboring revealed an interplay between REST and the tumor suppressor tissue plus 2 times the standard deviation. p53 in regulating cell-cycle progression of prostate cancer cells. Chromatin immunoprecipitation analysis of REST-occupied Materials and Methods regions Patient-derived xenografts LNCaP cells were grown in normal medium to 80% confluency. Detailed information about the tissues of origin and generation Cells were crosslinked in 1% cold formaldehyde for 10 minutes of patient-derived primary xenografts has been published else- and the fixation was stopped with 0.125 M glycine for 5 minutes. where (20). Cells were harvested in SDS buffer and then pelleted further to be

OF2 Clin Cancer Res; 2018 Clinical Cancer Research

The Proteome of Neuroendocrine Prostate Cancer

resuspended in a cold IP-buffer supplemented with phosphatase expressed as well (Supplementary Table S3). Raw proteomic files and protease inhibitors. The DNA complexes were further soni- are available at ProteomeXchange (PXD009636). cated to an average size of 300 to 500 bp using a Bioruptor plus sonication device (Diagenode; with settings 30 seconds on, Other bioinformatic tools and statistics 30 seconds off 7 times). Anti-rabbit IgG (10 mg; Millipore) or 5 To compare gene-expression variation between mRNA and mg of anti-REST (Millipore) was added to 50 mg of DNA lysate. protein, tumors were ranked according to the expression Magnetic Dynabeads (Invitrogen) were added into the mix and levels for each gene and distributions were compared using incubated overnight at 4C. After washes, the DNA was decros- Spearman correlation. The overall correlation to different slinked at 65C for 12 hours and purified using a Minelute PCR ontological processes was performed using the 1D analysis purification kit (Qiagen). The DNA was sequenced at the National tool implemented in the Perseus software as described else- High-Throughput DNA Sequencing Centre at Copenhagen Uni- where (35). Briefly, the score compares the mean RANK of versity. Raw reads were mapped to the human genome (hg19 the correlation values of given group (e.g., Ribosome) and the assembly) using bowtie (25) and converted to bam- and bed-files mean RANK of the correlation values of the rest of the proteins using Samtools and Bedtools (26, 27). Only one read per chro- under the formula s ¼ 2 (R/1n R2) ,whereR1andR2are mosomal position per strand was allowed for downstream visu- the average ranks within the group under consideration and its alization and analysis. For peak finding, measurements of dis- complement (all remaining proteins in the experiment), tances to TSSs, as well as visualization of tracks, superimposed respectively, and n is the total number of data points. Gene tracks, and heat maps, data were imported into EaSeq (28) using ontology classification and enrichment analysis were per- default settings and an extension of each read from its 50-end to a formed using DAVID (36), and results were graphically repre- total length of 250 bases. Raw ChIP files are available at the Gene sented with Cytoscape (37). Expression Omnibus database (GSE119385). Cell-cycle analysis Quantitative proteomic profiling C4-2B and PNT2 cells were transfected with two siRNAs Whole-protein extracts were purified from frozen specimens as targeting REST or a siRNA control, pulsed with BrdUrd (Life previously described (29). Twenty-five to 40 g of protein extracts Technologies) for 30 minutes, collected by trypsinization, and was trypsin digested following the Filter-Aid Sample Preparation fixed in methanol. Cells were permeabilized in 0.2 mol/L HCl (FASP) methodology (30). The resulting peptides were fraction- containing 0.25% Triton X-100 for 30 minutes. Anti-BrdUrd ated by Strong Anion Exchange chromatography (SAX) into five antibody (Santa Cruz Biotechnology) was incubated at 1:200 fractions to reduce sample complexity and maximize depth of dilution. As secondary antibody Alexa 488-anti-mouse (Life tech- proteome coverage (30). Each fraction was then analyzed by LC- nologies) at 1:500 dilution was used. Cells were washed in PBS MS/MS using the Q-Exactive mass spectrometer (Thermo scien- and resuspended in propidium iodide solution containing tific). Peptides were separated using a 4-hour gradient of water: RNaseA for 30 minutes at 37C. Samples were analyzed on a acetonitrile, on a 30 cm C18 column. MS spectra were acquired in FACSVerse instrument (BD Biosciences) using appropriate the Orbitrap with 70,000 resolution. MS/MS spectra were settings. acquired in data-dependent mode, after Higher Energy Collision- al Dissociation fragmentation, at a resolution of 17,500 (31). The Transfections, cell viability, RT-PCR, and Western blot analysis obtained mass spectrometric raw data were analyzed in the The Neon transfection system (Invitrogen) and Lipofectamine MaxQuant environment, version 1.3.7.1 (32), with the inte- 2000 (Invitrogen) were used for transfection of siRNA into grated Andromeda searching engine and false discovery rate prostate cancer cells following the manufacturer's instructions. (FDR) cutoff for peptide identification of 0.1 (ref. 33; Fig. 1B). The preparation of cell lysates and Western blot has been Proteins were identified by searching MS/MS data against the described (38). Viability analyses were performed using the cell human proteome sequences from UniProt (UniprotKB, 2012). proliferation kit I (Roche), following the manufacture's guide- To compare NEPC PDXs and androgen-sensitive (AS) adeno- lines. For evaluation of cell death, transfected cells were incubated carcinoma PDXs, the differences between the mean protein with propidium iodide (Sigma-Aldrich) and Hoechst-33342 expressiononeachgroup,defined as label-free quantification (Invitrogen) for 15 minutes at 37C in the dark and of the number (LFQ) intensity (34), were evaluated by the Student t test of dead and live cells estimated using the Celigo Imaging followed by Benjamini–Hochberg correction for multiple test- Cytometer (Nexcelom Bioscience). RT-PCR was performed as ing. FDR values smaller than 0.05 were considered statistically previously described (39). Primers used in this study: b-actin significant. Additionally, proteins with valid LFQ values for (ACTB), 50-CTGGCTGCTGA-CCGAGG-30 and 50-GAAGGTCT- each of the NEPC tumors but 25% values in the adenocar- CAAACATGATCTGGGT-30; TP53,50-ACCTACCAGGGCAGC- cinoma group were considered upregulated in NEPC. Proteins TACGGTTTC-30 and 50-GCCGCCCATGCAGGAACTGTTACA-30; with no detected in NEPC tumors but with LFQ values in REST,50-GGAGGAGGAGGGCTGTTTAC-30 and 50-ACCGACCA- >60% of adenocarcinoma tumors were defined as downregu- GGTAATCACAGC-30; p21 (CDKN1A), 50-CCTGTCACTGTCTTG- lated in NEPC (Supplementary Table S2). TACCCT-30 and 50-GCGTTTGGAGTGGTAGAAATCT-30; MDM2, To compare expression of proteins between PDX on castrated 50-TGTTGTGAAAGAAGCAGTAGCA-30 and 50-CCTGATCCAAC- and intact hosts, the difference between the mean LFQ of each CAATCACCTG-30; FAS,50-GGGCATCTGGACCCTCCTAC-30 and group was evaluated by the Student t test (P < 0.05 was considered 50-GATAATCTAGCAACAGACGTAAGAACCA-30; CHGA,50GCG- statistically significant). Additionally, proteins with no valid LFQ GTGGAAGAGCCATCAT-30 and 50-TCTGTGGCTTCACCACTT- value on any of the tumors of one group but at least on 75% of the TTCTC-30; SYN1,50-GCACGTCCTGGCTGGGTTTCTGGG-30 and tumors of the other group were considered differentially 50-AGGCTACCCGTCAGACATCCGTCTC-30.

www.aacrjournals.org Clin Cancer Res; 2018 OF3

Flores-Morales et al.

FASP/Trypsin SAX A B digestion fractionation

pH3 Primary adenocarcinoma LN met. NEPC met. pH11 pH5 pH8 pH6

Primary 313C 313D 313H 418 412 352 370 xenografts 6,000 311 313A 313B 331 MaxQuant LC-MS/MS analysis Castration 311-X 313A-X 313B-X 331-X 5,000 (3 weeks) Castration 313BR 331R 4,000 resistant Retention time

# of proteins (LFQ) 3,000 Samples m/z C D 0.8 0.4 Mean: 0.28 0 370 352 331-X 331R 418 331 412 311-X 311 313A-X 313B-X 313A 313C 313D 313H 313BR 313B -0.4 -0.8 Spearman correlation Score FDR Tryptophan metabolism 0.45 0.008 Arg and Pro metabolism 0.43 0.001 Propanoate metabolism 0.41 0.005 Val, Leu and Iso degradation 0.36 0.002 Pyruvate metab. 0.4 0.005 Resp. elec. transport chain −0.41 1.6×10−6 Mitochondria resp. chain C-I −0.5 0.0002 Ribosome −0.37 3.8×10−6 Translation termination −0.37 1.3×10−5 0.7 0.85 1 Translation initiation −0.3 0.0001 Translation elongation −0.28 0.001 Oxidative phosphorylation −0.35 1.6×10−5 Cell cycle Lipid oxidation E Chromosome F Lipid and amino acid segregation Chromatin catabolism organization Lipid catabolism Response Resp to oxidative to ROS Chromatin modification Mitosis Branched stress AA catabolism FA oxidation Chromatin remodeling Mitotic spindle Cellular response organization to oxidative stress AA catabolism Acetyl-CoA catabolism NER Aerobic Cell cycle respiration DNA repair regulation TCA cycle DNA damage Carbohydrate response signal Alcohol catabolism Neuron DNA Neuron biosynthesis differentiation replication development Lysosome Carboxylic acid Cell projection Glucose Acetyl-CoA organization biosynthesis morphogenesis metabolism catabolism DNA repair and replication Neuron projection dev Lysosome Organic acid biosynthesis Carbohydrate Carbohydrate Neuronal biosynthesis development metab.

OF4 Clin Cancer Res; 2018 Clinical Cancer Research

The Proteome of Neuroendocrine Prostate Cancer

Results These findings demonstrate that label-free quantitative proteomic profiling of PDXs can precisely segregate tumor samples Quantitative proteomic data can cluster PDXs by origin and based on patient/genetic origin. In addition, it can also provide hormone status unbiased information on effects of treatment and tumor progres- We analyzed the proteome of 17 prostate cancer PDX tumors, sion that cannot be obtained through genetic profiling. of which 4 tumors were obtained 3 weeks after castration of the host and in 2 cases tumors relapsed after castration and were Proteomic profiling provides additional information to collected (Fig. 1A; ref. 20). On average, 4,881 proteins were gene-expression studies in prostate cancer identified in each of the tumors (Fig. 1B). Unsupervised hierar- Transcriptomic profiling is widely used to estimate protein chical cluster analysis revealed a high positive correlation between expression in biological samples under the presumption that all samples (0.85 0.05; Fig. 1C). Untreated tumors derived from changes in mRNA levels would translate into changes in protein the same patient such as PDXs in the LTL-313 series (0.93 0.02) abundance. To test this hypothesis, specifically in prostate tissue, or PDXs LTL-352 and LTL-370 (0.96) are grouped together, and to identify concordant and discordant mRNA/protein pairs, reflecting a common genetic origin. On the other hand, the we compared how changes in mRNA levels correlate to protein response to castration therapy shows a larger degree of variability. expression changes for gene product across all the samples ana- PDX LTL-311 shows little alteration of proteome expression lyzed. We found a positive but limited correlation (average after therapy (LTL-311-X), while PDX LTL-331 shows a more of 0.28) between mRNA and protein variation across samples pronounced response. The proteome of PDX LTL-331 after 3 (Fig. 1D). The correlation between protein/mRNA pairs variation weeks of castration (LTL-331-X) shows higher similarity to tumors measured across the different experiments does not seem to be a with NEPC PDXs (LTL-370 and LTL-352), than to the parental function of abundance of the mRNA or the protein, indicating that untreated PDX (LTL-331), suggesting that a castration-induced the calculated correlation coefficients are not likely to be an transdifferentiation process occurred in this case (Fig. 1C). artifact of the measurement methodologies (Supplementary Indeed, once castration resistant, the LTL-331R PDX shows a Fig. S1A). Overall, genes involved in intermediary metabolism neuroendocrine phenotype characterized by low expression of of lipids and carbohydrates show positive mRNA/protein corre- AR-regulated genes and elevated expression of neuronal-specific lation, while genes related to ribosomal biogenesis, translational genes (Supplementary Tables S1 and S2; ref. 20). In contrast, control, as well as mitochondrial proteins show close to random castration-resistant LTL-313BR PDX exhibits a proteome profile correlation (Fig. 1D). Therefore, transcriptomic studies involving similar to the parental LTL-313B PDX, reflecting both the genetic the latest functional categories must by cautiously interpreted. similarity between LTL-313B and LTL-313BR and the AR-driven phenotype (Supplementary Table S2; ref. 20). Protein-driven mechanisms of NE transdifferentiation Interestingly, despite the efficacy of androgen ablation therapy Androgen deprivation therapy (ADT) can induce regression of in limiting tumor growth of advanced prostate cancer, limited most advanced prostate cancer, at least temporarily, before CRPC proteome alterations were observed only in tumors collected arises. Primary or therapy-induced NEPCs are highly aggressive, 3 weeks after castration of the host compared with tumors but their proteome is poorly characterized, beyond the fact that propagated in intact mice (Supplementary Table S3). Tumors on they express high levels of neurosecretory proteins and have castrated hosts showed signs of reduced androgen signaling, as reduced expression of AR-regulated proteins (14). Therefore, we suggested by the lower expression of androgen-regulated proteins compared the protein expression profiles of NEPCs to those from such as SARG, DHCR24, and FASN and of proteins involved in AR-dependent adenocarcinomas. We identified 861 proteins fatty acid biosynthesis (FASN, PCCB, and ACSM3). Castration (FDR < 0.05; Supplementary Table S2) differentially expressed also resulted in reduced expression of proteins involved in rRNA between these tumor types. Among the proteins with differential processing and DNA metabolism (DDX56, HEATR1, NOB1, expression, we found some previously validated by IHC (e.g., AR, RPS27A, PCNA, and ATR), while leading to elevated expression PSA), as well as some of bona fide markers of NED (e.g., SYN1, SYP, of proapoptotic proteins such as BCL2L1. Moreover, proteins and CHGA), thereby supporting the validity of our proteomic related to axon guidance (DPYSL2, MATN2, SPTAN1, and data (20). Gene ontology enrichment analysis of the biological SPTBN1) increased their expression, although the canonical NE processes connected to differentially expressed proteins showed markers did not consistently show upregulation at this point after that NEPCs have increased expression of proteins that are castration (Supplementary Table S3).

Figure 1. Proteome characterization of prostate cancer PDXs. A, Schematic representation of the grafting process. Tumors were grafted from primary adenocarcinomas obtained after radical prostatectomy procedures (LTL-311, LTL-418, and LTL-331); biopsies (LTL-313 series); lymph node metastasis (LTL-412); and distant metastases (LTL-352 and LTL-370). Tumors were established and propagated through various generations of mice. "X" indicates that tumors were collected 3 weeks after castrating the host (LTL-311-X; LTL-313A-X; LTL-313B-X; LTL-331-X). LTL-313BR and LTL-331R tumors relapsed after initial response to castration (castration-resistant). B, Layout of proteomic methodology. Proteins from whole tumor extracts were digested with trypsin and peptides fractionated using Strong Anionic Exchange (SAX). Each fraction was analyzed on an Exactive-Q mass spectrometer (Thermo). The number of proteins identified and quantified by label-free quantification (LFQ) using the MaxQuant software is shown. C, Unsupervised hierarchical cluster of all tumors based on Pearson correlation coefficient. D, Limited correlation between mRNA and protein expression in prostate cancer. Tumors were ranked according to the expression levels for each gene and distributions compared using Spearman correlation. The overall correlation to different ontological processes was performed as described in the Materials and Methods section. E and F, Network representation of gene ontology terms enriched among proteins differentially regulated between NEPC and adenocarcinoma PDX tumors. Functional categories overrepresented among the proteins with elevated (D, red) or reduced (E, blue) expression in the neuroendocrine tumors compared with adenocarcinomas. On each node, the size of the red inner circle and the thickness of the red ring relates to the number of proteins upregulated and downregulated, respectively. Edges connect categories with shared proteins.

www.aacrjournals.org Clin Cancer Res; 2018 OF5

Flores-Morales et al.

normally associated with neuronal differentiation and of proteins neuronal development related categories show an increased pro- involved in mitotic cell-cycle and cell division associated with portion in REST binding compared with the general distribution chromatin remodeling processes (Fig. 1E), in line with the highly (P< 0.007, Fig. 3D). proliferative phenotype often reported for NEPC (14). Addition- In order to study the functional significance of the REST- ally, we found a significant upregulation of proteins involved in bound regions, we analyzed how the distance between REST- DNA-repair mechanisms that have not been previously recog- occupied regions was related to the expression of genes in their nized to play a major role in NEPC (Fig. 1E). This includes vicinity. All genes significantly regulated between NEPC and AS proteins involved in homologous recombination (RAD51, adenocarcinoma tumors were sorted according to their distance PARP1, XRCC1, RECQL, and RPA2), nucleotide-excision repair to the nearest REST peak. We found that having a REST- (PCNA, POLD1, POLE, and RFC5) and mismatch repair (MSH2 occupied region within 10 kb of a gene significantly (Fisher and ERCC1) among others (Supplementary Table S2). Enzymes P < 0.0001) increases the chances of that gene being upregu- involved in nucleotide metabolism such as the thymidylate lated in NEPC tumors compared with adenocarcinomas synthetase (TYMS) were also found upregulated in NEPCs, pre- (Fig. 3A). Thus, of the genes within the vicinity of the REST- sumably to support DNA synthesis. On the other hand, NEPCs occupied region, there were 4 times more genes upregulated expressed overall lower levels of mitochondrial proteins, includ- than were downregulated (217 vs. 51) in the NEPC tumors ing those with functions in the catabolism of fatty acids, amino relative to AS tumors (Fig. 3B). Gene ontology enrichment acids, and in the tricarboxylic acid cycle (Fig. 1F) and also of analysis for the biological functions of these genes showed lysosomal proteins and of proteins involved in cellular response that genes overexpressed in NE located close to a REST-occu- to oxidative stress. In good agreement with the protein/mRNA pied region relate to neuronal differentiation and function, expression correlation, enrichment in GO terms associated with while genes related to proliferation, cell-cycle progression, and metabolic pathways involving mitochondrial and lysosomal pro- metabolism were mostly absent (Fig. 3C and D). These findings teins was only obvious in proteomic analysis and not through suggest that loss of REST during NED in prostate cancer is a key transcriptomics (Supplementary Figs. S1 and S2), suggesting a driver of the expression of neuron-specific genes but seems to higher degree of posttranscriptional regulation on the expression have limited direct impact on other processes relevant to NEPC of these genes. tumorigenesis, such as cell-cycle progression, DNA repair, cell Overall, these data suggest that rapidly growing NEPCs restrict motion, or changes in intermediary metabolism. their cellular activities to cell division. Reduced expression of mitochondria proteins suggests a greater reliance on glycolysis to Loss of REST negatively regulates cell-cycle progression fulfill their energetic needs (Fig. 1E and F). through the activation of the p53 pathway In order to functionally evaluate the effect of REST in cell REST primarily binds to chromatin in the proximity of neuron- proliferation, castration-resistant, AR-expressing C4-2B cells specific genes in prostate cancer were transfected with two independent siRNAs targeting REST Expression of neuron-specific proteins is a hallmark of NEPCs (Fig. 4A), and cell viability was measured by MTT assays and (Fig. 1). In previous studies, we identified the transcriptional compared with cells transfected with control siRNA. Surprisingly, repressor REST as an important regulator of NE transdifferentia- REST knockdown resulted in reduced C4-2B cell viability (Fig. 4B) tion in response to castration and other stimuli in prostate cancer and increased the number of dead cells over time (Fig. 4C). Cell- cells (18). In order to investigate whether REST function extends cycle profile analysis revealed that cell cultures with reduced levels to the regulation of proliferation and/or metabolism-related of REST showed an increased number of cells in G1 phase and genes, we performed an unbiased analysis of REST location on fewer cells in S phase (Fig. 4E). These results suggest a function of chromatin of prostate cancer cells using chromatin immunopre- REST in sustaining cell-cycle progression, which contrasts with the cipitation followed by DNA sequencing (ChIP-Seq), as an attempt highly proliferative phenotype of low-REST-expressing NEPC to identify gene directly repressed by REST. tumors. We then asked whether REST depletion would alter the REST-bound chromatin regions were isolated from AR- and expression of checkpoint regulators such as p53 and Rb1, thereby REST-expressing LNCaP cells and identified by next-generation resulting in cell-cycle arrest. REST knockdown in C4-2B cells did sequencing in two independent experiments. The analysis not affect total levels of Rb1 and had limited direct influence in identified 2,615 DNA regions significantly increased in REST- p53 total levels (Supplementary Fig. S3C). However, the expres- immunoprecipitated chromatin,commontobothbiological sion of the p53 downstream effectors such as p21 (CDKN1A), replicates, as compared with nonspecificIgG(Fig.2A).The FAS, and MDM2 was clearly induced (Fig. 4A, D, and E), suggest- validity of these data was confirmed by the significant enrich- ing the p53 pathway becomes activated upon REST depletion. ment of the RE-1 DNA motif within REST-occupied regions Importantly, simultaneous downregulation of REST and p53 was (P < 10 307), the bona fide recognition site for REST binding on able to rescue the cell-cycle inhibition caused by REST knockdown DNA. REST binding to the promoter regions of known target alone (Fig. 4C, D and F), suggesting that checkpoint inactivation is genes such as CHGA and SYP (Fig. 2B) as well as to BDNF, essential for cell division in the context of reduced REST expres- SYN1, SCGN,andCHGB (Supplementary Fig. S3) was also sion. Indeed, low levels of REST and mutation of TP53 are observed, further validating our experimental approach. commonly observed in NEPC tumors (7, 40, 41). Further evi- We then identified which of the genes with increased expression dence of the interplay between REST and p53 was obtained by in the NE compared with adenocarcinoma PDX tumors (20) had knocking down REST in the PNT2 prostate cell line. PNT2 cells REST binding regions in their vicinity (within 100 Kb), which were immortalized by transfecting normal prostatic epithelial occurred in 26.82% of the cases (Fig. 2C; Supplementary Table cells with the SV40 large T and small t antigens (42). These viral S4). Gene ontology enrichment analysis of the biological process proteins inhibit, among others, p53 function (43). Thus, REST of these groups of genes revealed that only genes involved in knockdown in PNT2 cells resulted in increased expression of NE

OF6 Clin Cancer Res; 2018 Clinical Cancer Research

The Proteome of Neuroendocrine Prostate Cancer

A 50 50 2615 Count IgG_1

REST_1 0 0 0 50 50 2615 Count IgG_2

REST_2 0 0 0 −10,000 bp | 10,000 bp B 30 30 IgG_1 REST_1 0 0 30 30 IgG_2 REST_2 0 0 chr14 | 93388000 | 93395000 | 93402000 chr14 | 93388000 | 93395000 | 93402000 CHGA CHGA CHGA CHGA

12,779 bp 12,779 bp

30 REST_1 0 30

REST_2 0 chrX | 49045000 | 49050000 | 49055000 chrX | 49045000 | 49050000 | 49055000 SYP SYP

7,262 bp 7,262 bp

C D % NEPC upregulated χ2 NEPC % NEPC up P value REST binding (100 kb) up & REST binding

2,500 80 DNA replication 2.3 1.8 0.53 2,000 60 1,500 DNA repair 2.5 2.3 0.88 40 1,000 Transcrip. RNA pol II 2.9 3.6 0.42 500 20 Reg. cell cycle 0.5 1.0 0.22 − 0 0 G1 S transition 1.5 1.5 0.99 NEPC upregulated

DNA repair −S transition Reg. cell1 cycle DNA replication G

Nervous system dev.Transcrip. RNA pol II

Figure 2. REST binding to chromatin is enriched in areas near to neuron-related genes. ChIP experiments are reproducible and REST binding is enriched in the promoter region of known target genes. A, Track of the duplicated experimental replicates (REST_1 and REST_2) superimposed signal (intensity on the Y-axis) at 2,616 regions from the REST- bound regions. The X-axis represents the 10,000 bp surrounding each region and was segmented into 400 bins and smoothed for 1 bin. The Y-axis reflects signal intensity. B, REST peaks are enriched in the promoter region of known target genes such as CHGA and SYP, while absent in control IgG ChIP experiments. C, Number of genes upregulated in NEPC tumors distributed by biological function containing or not REST chromatin binding region in the proximity (100 kb) as defined in ChIP analysis from LNCaP prostate cancer cells. D, Significant enrichment of genes involved in nervous system development among those upregulated in neuroendocrine tumors as compared with adenocarcinomas with REST binding regions within 100 kb. *P < 0.05 was considered statistically significant for enrichment of genes with REST binding.

www.aacrjournals.org Clin Cancer Res; 2018 OF7

Flores-Morales et al.

A REST_1 REST_2 IgG_1 IgG_2 = 6,721) n NEPC vs. adeno signiﬁcant (

1,000,000 bp 1,000,000 bp 1,000,000 bp 1,000,000 bp Genes sorted according to distance nearest REST peak 50 50 50 50 -3 3 Log2 (FC) B 30 C REST (n = 268) Gene ontology REST Synaptic signaling (30) 20 Nervous system dev. (49) Reg. of memb. potential (18)

Count 10 Secretion by cell (23) Peptide hormone secretion (6) Calcium ion transport (6)

REST binding 10,000 bp 0

-10 -8-6-4-20246810 -log10(FDR) 0 5 10 15 NEPC vs. AS [log2 (fold change)]

D Synaptic signaling Nervous system development

SEZ6 CACNA1E OLFM1 CA10 SHOX2 OLFM3 NRXN1 SKOR1 SHC3 UNC5A CACNA1B FGF14 TRIM9 DAPK3 SEZ6 CELSR3 CPNE9 NPTX1 SYT5 CHRNB2 STMN3 SLC12A5 GAD1 SYT4 ACTL6B NPTX2 UNC13A SLC12A5 ELAVL3 SYNJ1 KCNB1 SNAP25 ADCYAP1 TNR PCLO SNAP25 SHC3 UNC13A DTX1 BSN RIC3 CDK5R2 MYT1 PCSK2 GRIK2 SYT4 NFASC CHRNB2 BSN POU3F1 TNR NRSN1 PCLO ADCYAP1 CRTAC1 CELF4 SCRT1 HCN1 GABBR2 GRM2 GRIN2C SRRM4 SCN3B SYNJ1 KCNC1 POU4F3 NRXN1 VGF HTR1A HPCA GNAO1 SPAG9 FGF14 DPYSL5 MYT1L GPRIN1 NFASC SLITRK1 NPTX1 MYT1L

Regulation of membrane potential Secretion by cell Peptide hormone secretion SEZ6 TRPM4 CARTPT HCN3 POU2F2 REST CELF4 CHRNB2 SYNJ1 FGF14 CDK5R2 SCN3B PTPRN CHGA VGF KCNC2 KCNH7 ATP1A3 TRIM9 CPLX3 GHRHR RAB3C PCLO KCNH6 KCNB1 KCNB1 ADCYAP1 CHRNB2 SNAP25 CACNA1B UNC13A Calcium ion CACNA1B GRIK2 CACNA1E NRXN1 GAD1 SYT5 transmembrane transport PKD2L1 GRIN3A GRIN2C HCN2 GRM2 CACNA1E TRPM8 SYT4 HTR1A HCN1 NRXN1 ADCYAP1 TRPM4 GRIN1 JPH3

Figure 3. Genes with nearby REST-occupied regions show higher expression levels in NEPC compared with adenocarcinoma PDXs. A, Heat map representing the nearest REST-occupied chromatin region relative to a significantly regulated gene in NEPC versus adenocarcinoma tumors (left) and nonsignificantly regulated genes (right). Fold change expression in NEPC tumors compared with adenocarcinomas is color coded from blue (downregulated to red, upregulated). B, Histogram display of significantly regulated genes between NEPC versus adenocarcinoma tumors with REST-binding region within 10,000 base pairs (bp). C, Gene ontology enrichment analysis of the genes in B. D, Network representation of genes in B based on their protein–protein interaction score according to the String database (http://string-db.org) and segregated by ontological categories.

OF8 Clin Cancer Res; 2018 Clinical Cancer Research

The Proteome of Neuroendocrine Prostate Cancer

A C4-2B B C 500 48 h 96 h REST 100 *** 400 *** PHB 80 *** *** 300 *** p53 60 ** 200

p21 40 Cell death

(% of control) 100 CHGA 20 % of cell viability β-Actin 0 0 + − − + − − + − − siCtrl + − − siCtrl siCtrl − + − siREST_1 − + − − + − siREST_1 − + − siREST_1 − − + − − + − − + siREST_2 − − + siREST_2 siREST_2 C4-2B D 250 p21p53 MDM2 FAS REST REST * 200 ** *** p53 ** ** 150 **** p21 100 SCGN * *** % of control 50 ** ** * *** ** β-Actin *** 0 − − − siCtrl + − − − − − siCtrl + −−− sip53 − +++−− sip53 + − − siREST_1 − − + − + − siREST_1 − + − − siREST_2 − − − + − + siREST_2 − + )h84( B2-4C )h84( B2-4C )h69( ***** *** •••***••• *** ** *** ••• E 100 ### ### ## 100 ### ### 80 C4-2B 80 C4-2B 60 G2 60 G2 S S 40 40 G1 G1 Cell count (%) 20 Cell count (%) 20 0 0 siCtrl + − − − − − siCtrl + − − − − − sip53 − + − − + + sip53 − + − − + + siREST_1 − − + − + − siREST_1 − − + − + − siREST_2 − − − + − + siREST_2 − − − + − +

120 ** C4-2B (48 h) ••• 120 C4-2B (96 h) ••• F ## •••*** ## 100 100 ### *** 80 *** *** ### 80 ### ### *** ### *** 60 60 ### 40 40 % of cell viability

20 % of cell viability 20 0 0 siCtrl + − − − − − siCtrl + − − − − − sip53 − + − − + + sip53 − + − − + + siREST_1 − − + − + − siREST_1 − − + − + − siREST_2 − − − + − + siREST_2 − − − + − +

G H 800 CHGA I 100 ns ns *** 80 PNT2 PNT2 600 *** 60 G2 400 SYN1 S REST *** *** 40 200 G β-Actin 20 1 Cell count (%) % of control 0 − − − − − − 0 siCtrl + siCtrl + + siCtrl + − − siREST_1 − + − siREST_1 − + − − + − siREST_1 − + − − − − − − − siREST_2 + siREST_2 + + siREST_2 − − +

Figure 4. REST depletion in prostate cancer cells inhibits cell-cycle progression through activation of the p53 pathway. A, Western blot analysis of the effects of REST knockdown on REST, PHB, p53, p21, CHGA, and Beta-Actin protein expression in C4-2B cells. B, REST knockdown reduces C4-2B cell viability and (C) increases cell death. D, Western blot analysis of the effects of REST and p53 knockdown on REST, p53, p21, and SCGN protein expression in C4-2B cells (left) and relative expression of p53, p21, MDM2, FAS, and REST in cell transfected with siRNAs targeting REST or p53. E, Quantification of the percentage of cells in the different phases of the cell cycle upon REST and p53 knockdown in C4-2B cells. F, Effects of REST and p53 in the viability of C4-2B cells. G, Western blot analysis of the effects of REST knockdown in PNT2 cells. H, Relative mRNA expression of CHGA and SYN1 in the same as in G. I, Quantification of the percentage of cells in the different phases of the cell cycle upon REST knockdown in PNT2 cells. Comparing knockdown samples with control, P < 0.05, P < 0.01, P < 0.001; comparing samples with p53, #P < 0.05, ##P < 0.01, ###P < 0.001; comparing REST knockdown to combination of REST and p53 knockdown samples, *P < 0.05, **P < 0.01, ***P < 0.001. ns, nonsignificant. www.aacrjournals.org Clin Cancer Res; 2018 OF9

Flores-Morales et al.

markers (Fig. 4G and H) but had no inhibitory effects in cell-cycle are being implemented to manage these patients. In order to test progression (Fig. 4I). whether SCGN could serve as a prognostic biomarker for these Overall, these results indicate that during NED in prostate, patients diagnosed with "low-risk" tumors, we studied the loss of REST induces the expression of neuronal genes, while association of SCGN expression to biochemical relapse in uni- negatively regulating cell-cycle progression through activation of variate and multivariate Cox regression analyses on patients with p53. Thus, only tumors that have overcome p53 activation asso- GS 3þ4 tumors. Strikingly, both analyses retrieved statistically ciated with REST (and AR; refs. 9 and 13) loss would be able to significant association of SCGN expression with BF (Table 1B). grow as NEPCs. Moreover, the 10-year cumulative incidence of BF was 27.8% (95% CI, 19.9–35.7) in the SCGN-low group compared Expression of secretagogin (SCGN) correlates with increased with 45.2% (95% CI, 35.2–55.2) in the SCGN-high group risk of biochemical recurrence after prostatectomy (P ¼ 0.0007; Fig. 5D). Prostate cancer tumors are heterogeneous and often exhibit These results suggest that in patients managed with conserva- mixed phenotypes, including the expression of neuronal proteins tive regimens, elevated SCGN expression might be considered as a (9). Expression of neuroendocrine proteins has been associated sign of aggressiveness, suggesting the earlier application of cura- with mutation-driven cell lineage switching (44) as well as tive treatments such as RP or radiation. reduced AR activity. Because both the processes are associated with increased tumor aggressiveness, we hypothesized that Discussion expression of REST-regulated proteins in primary tumors could have prognostic value. SCGN was the most highly upregulated Patients with prostate carcinomas with NED have limited protein in NEPC tumors (fold change > 150; P < 0.003) in our therapeutic options and subsequently poor prognosis. In order proteomic profile among the ones with REST binding in the to better understand the processes involved in NED in prostate promoter region (Supplementary Fig. S3A). We first tested the cancer, we have, for the first time, characterized the proteome prognostic value of SCGN gene expression using data from a differences between NEPC and prostate adenocarcinoma PDXs publicly available cohort of radical prostatectomies (n ¼ 131; using quantitative mass spectrometry–based proteomics. NEPCs ref. 24). As expected, only a handful of tumors exhibited high exhibit enhanced levels of proteins involved in the regulation of SCGN expression, but most of them were associated with BF after cell proliferation and response to DNA-damage stress and express RP (HR log-rank: 4.82; Supplementary Fig. S3D). This prompted lower levels of mitochondrial and lysosomal proteins, indicating us to validate these results using a Copenhagen Prostate Cancer the use of alternative pathways for energy production. Expression cohort composed of 336 consecutive prostatectomies operated at of neuron-specific proteins in NEPCs was directly linked to the Righshospitalet (Copenhagen, Denmark) between 2002 and reduced expression of the transcription repressor REST. Expres- 2005 (Fig. 5B, Supplementary Table S5). We analyzed the expression of one of these proteins, SCGN, correlated with increased risk sion of SCGN by IHC (Fig. 5A). Tumors were classified according of BF after RP. to the expression of SCGN as "high" and "low" (see Materials and We recently characterized the proteome of primary prostate Methods). High SCGN protein expression was detected in 46% tumors and detected increased mitochondrial activity and of the cases (n ¼ 147), and its expression correlated with RP enhanced expression of proteins involved in oxidative phosphor- Gleason score (GS; P ¼ 0.02; Supplementary Table S5). Analysis ylation and TCA cycle when compared with neighboring benign of the 10-year cumulative incidence of BF was 36.2% (95% CI, prostate tissue (29, 46). Here, we observe that NEPCs exhibit 28.6–43.7) in the SCGN-low group compared with 50.1% (95% reduced levels of proteins involved in these metabolic processes CI, 41.8–58.4) in the SCGN-high group (P ¼ 0.0007; Fig. 5C). compared with prostate adenocarcinomas (Fig. 1). These changes Moreover, univariate Cox regression analysis found a statistical occur shortly after castration, before other NE features (increased association of SCGN expression with the risk of suffering BF after expression of neuronal and proliferation related proteins) were RP (HR, 1.7; P ¼ 0.002; Table 1A). A trend toward this association revealed, suggesting this metabolic transition to be related to being true was maintained after multivariate analysis (HR: 1.4; reduced AR activity. This hypothesis is further supported by the P ¼ 0.07; Table 1A). No association was found, however, with limited chromatin association of REST to metabolic gene pro- other clinical endpoints such as development of CRPC or prostate moters (Fig. 3) and the absence of changes in mitochondrial cancer–specific death (Supplementary Table S6). We further val- content observed in REST-depleted, AR-expressing cells (Fig. 4A). idated these results using an independent cohort of RP collected at The opposite regulation of mitochondrial proteins in NEPC the Malmo€ University Hospital (18, 21). The incidence of tumors compared with prostate adenocarcinoma indicates the existence with high SCGN expression was lower in this cohort (19%). Given of fundamental differences on the pathways utilized for energy the foci pattern of SCGN expression in primary prostate tumors, metabolism, and suggests that NEPCs would have relatively this lower incidence is most likely related to the fewer cores per higher glycolytic activity. This conclusion is supported by studies RP sampled when building the TMA: two in the Malmo€ cohort using fluorodeoxyglucose (FDG) positron emission tomography compared with 4 to 12 cores in the Copenhagen cohort. In this (PET) in patients with suspected neuroendocrine tumors. Visceral cohort, SCGN expression also tended to correlate with increased metastases, the preferred site for NEPCs, show higher uptake of risk of suffering from BF after RP (HR ¼ 2.8; P ¼ 0.06; Supple- FDG as when compared to bone metastases, the preferred site for mentary Fig. S3D). AR-expressing castration-resistant adenocarcinomas (47). Clinical trials over the years have shown that prostate cancer Interestingly, we also detect a significant downregulation of patients bearing tumors with GS 3þ4 have reduced risk of multiple proteins within several cytosolic compartments in suffering biochemical relapse and of dying with prostate cancer NEPCs, including the lysosome, ER, and Golgi. The general after RP than those with higher histologic grades (45). Conse- reduction of these compartments would be consistent with the quently, more conservative regimens such as "active surveillance" reduced cytosol volume that is characteristic of small-cell

OF10 Clin Cancer Res; 2018 Clinical Cancer Research

The Proteome of Neuroendocrine Prostate Cancer

A B 0:RI 1:RI RP in the study period n = 336

Missing malignant tissue. n = 17

50 μm SCGN Low 250 μm 250 μm TMA cohort n = 319 IR:2 IR:3 IR:4

Missing PSA follow-up. n = 2

Analysis of BF

SCGN High n = 317 250 μm 250 μm 052 μm C Biochemical failure Biochemical failure

BF Death before BF SCGN low Gray test: P = 0.0007 SCGN high Cumulative incidence of BF Cumulative Cumulative incidence of BF Cumulative

Time from RP (years) 03691215 Time from RP (years) 0 3 691215 No. at risk 317 222 162 129 60 1 No. at risk 170 129 93 72 39 1 147 93 69 57 21 0 D Biochemical failure (GS ≤ 3 + 4) Biochemical failure (GS ≤ 3 + 4) BF SCGN low Gray test: P = 0.0007 Death before BF SCGN high Cumulative incidence of BF Cumulative Cumulative incidence of BF Cumulative

Time from RP (years) 03691215 Time from RP (years) 03691215 Not at risk 238 182 135 107 54 1 Not at risk 135 111 83 65 36 1 103 71 52 42 18 0

AR Fatty acid metabolism E CR-Neuroendocrine REST Neuronal gene expression E2F1 Cell-cycle progression (AURKB, MYCN)

ADT

Fatty acid metabolism

AR Cell-cycle progression REST AR CR-Adenocarcinoma REST p53

Neuronal gene expression

Figure 5. SCGN expression correlates with increased risk of biochemical failure after radical prostatectomy. A, Representative images of SCGN immunoreactivity (IR). IR of 0 and 1 was accounted as "low" SCGN expression and 2, 3, and 4 were considered as "high." Arrows indicate positive staining of scattered prostate neuroendocrine cells. B, Schematic description of the cohort. C, The cumulative incidence of BF for the entire cohort and in relation to SCGN expression. Death without BF was computed as competing event. D, The same as B but for patients carrying tumors with Gleason score of 3 þ 4 or below. E, Schematic representation of the proposed model of prostate cancer progression after ADT. Prostate cancer relapse after ADT can occur in several ways. Typically, prostate cancer will relapse in the form of adenocarcinoma showing metabolic features of primary tumors and with expression of AR and REST and limited expression of neuronal proteins. Alternatively, castration-resistant NE tumor with reduced AR and REST expression will express neuronal proteins and increased activity of proteins involved in cell-cycle progression, likely on a background of reduced activity of the p53 pathway.

www.aacrjournals.org Clin Cancer Res; 2018 OF11

Flores-Morales et al.

Table 1. Univariate and multivariate cause-specific Cox proportional hazard of BF A Univariate analysis Multivariate analysis HR (95% CI) P HR (95% CI) P SCGN Low Ref Ref High 1.7 (1.2–2.3) 0.002 1.4 (0.98–1.9) 0.07 Age at RP For 5-year differences 1.1 (0.9–1.2) 0.4 0.9 (0.8–1.1) 0.2 PSA For 2-fold difference 1.5 (1.3–1.8) <0.0001 1.3 (1.1–1.5) 0.007 Pathologic T stage pT2a/b/c Ref Ref pT3a/b 3.3 (2.4–4.6) <0.0001 2.0 (1.4–2.9) 0.0001 N stage N0/x Ref Ref N1 2.8 (1.0–7.5) 0.04 1.3 (0.5–3.8) 0.6 RP Gleason score 6 Ref Ref 3 þ 4 2.8 (1.8–4.4) <0.0001 1.8 (1.2–2.9) 0.008 4 þ 3 4.1 (2.5–6.6) <0.0001 2.8 (1.7–4.6) 0.0001 8–10 5.0 (2.9–8.8) <0.0001 2.6 (1.4–4.9) 0.002 Margin status R Ref Ref Rþ 2.3 (1.6–3.3) <0.0001 1.5 (1.0–2.2) 0.04 B Univariate analysis Multivariate analysis HR (95% CI) P HR (95% CI) P SCGN Low Ref Ref High 1.9 (1.3–2.9) 0.002 1.9 (1.2–2.9) 0.004 Age at RP For 5-year differences 1.0 (0.8–1.2) 0.9 0.8 (0.7–1.0) 0.08 PSA For 2-fold difference 1.5 (1.2–1.9) 0.0001 1.4 (1.1–1.7) 0.004 Pathologic T stage pT2a/b/c Ref Ref pT3a/b 4.0 (2.6–6.0) <0.0001 3.0 (1.9–4.9) <0.0001 N-stage N0/x Ref Ref N1 1.8 (0.2–12.6) 0.6 2.8 (0.4–21.4) 0.3 RP Gleason score 6 Ref Ref 3 þ 4 2.9 (1.9–4.5) <0.0001 1.7 (1.0–2.7) 0.03 Margin status R Ref Ref Rþ 2.2 (1.4–3.4) 0.0006 1.2 (0.7–1.9) 0.5 Abbreviations: CI, confidence interval; HR, hazard ratio; PSA, prostate-specific antigen; REF, reference; RP, radical prostatectomy.

carcinomas (48). The reason for this phenotype is not clear. We REST binds the chromatin regions in the vicinity of the SRRM4 detect reduction in enzymes involved in lipid biosynthesis, which gene (Supplementary Fig. S3B), suggesting a feed-forward regu- could influence the size of these membrane-rich organelles. The latory loop that would result in promoting the expression of normal prostate luminal epithelium has an important secretory REST-repressed genes. To which extent the function of neuron- function that is regulated by androgen signaling and, to some specific proteins is critical in the development of NEPCs aggressive extent, maintained in primary tumors (49). Therefore, reduced phenotype remains to be defined. levels of AR, resulting in reduced expression of AR-regulated genes Similar to REST (18), the expression of SCGN, a REST- in NEPCs (Supplementary Fig. S4A and S4B), may have wide- repressed gene, in prostatectomy specimens correlated with BF spread effects in secretory pathways, in addition to the regulation after surgery, while no correlation with other clinical endpoints, of mitochondria content, a hypothesis that deserves further such as disease-specific survival, was found (51). Strikingly, the exploration. expression of SCGN in primary adenocarcinomas also occurs in The transcriptional regulator REST plays a key role in repressing AR-positive cells (Supplementary Fig. S3E), suggesting that the the expression of neuronal genes in nonneuron-related tissues. expression of this neuroendocrine marker may occur indepen- Loss of REST and the expression of REST splice variants unable to dently or precede the loss of AR that is characteristic of NEPC. exert this repressor function have been implicated in the NED in Whether or not SCGN and AR coexpressing tumors have prostate cancer cells (9, 18, 50). Thus, the expression of SRRM4, a increased chances to develop into full-blown NEPC with splicing factor involved in the alternative splicing of REST, is small-cell characteristics after RP or castration therapy remains commonly upregulated in NEPCs. Interestingly, we found that to be elucidated.

OF12 Clin Cancer Res; 2018 Clinical Cancer Research

The Proteome of Neuroendocrine Prostate Cancer

The tumor suppression function of REST suggested by its sensitivity to drugs targeting glycolytic pathways that could prognostic value for tumor aggressiveness (18) is challenged by be exploited as therapeutic alternative. Finally, expression of our experimental results. Depletion of REST in C4-2B cells neuron-specific proteins, characteristic of NE tumors, is mostly resulted in decreased viability, increased cell death and cell-cycle a consequence of low REST expression, which, as judged by our arrest. This phenotype can be rescued by simultaneous p53 data, has limited direct effect in promoting the highly proliferative depletion (Fig. 4). This suggests that a defective p53 pathway is phenotype of these tumors. Whether or not the expression of required for the survival of tumors with low REST expression. neuronal proteins contributes to survival of prostate cancer cells Because combined p53 and REST genetic inactivation in mice under conditions of low androgen availability requires further results in more frequent brain neuroepithelial tumors than caused investigation. by the inactivation of p53 alone (52), we cannot entirely exclude that REST inactivation may also contribute to prostate neuroen- Disclosure of Potential Conflicts of Interest docrine tumor growth in vivo, through mechanisms not yet No potential conflicts of interest were disclosed. fully explored here, such as the paracrine regulation of the tumor microenvironment. The significance of p53 for NE prostate Authors' Contributions malignancies is further supported by the upregulation of the Conception and design: A. Flores-Morales, Y. Wang, D. Iglesias-Gato PEG10 protein in NEPCs, as the inhibition of p53 is also required Development of methodology: D. Lin, A. Bartels, J.M. Moreira Acquisition of data (provided animals, acquired and managed patients, for the cell-cycle promoting actions of PEG10 in NEPC (53). The provided facilities, etc.): A. Flores-Morales, T.B. Bergmann, C. Lavallee, fact that tumor suppressor p53 is often mutated in NE prostate T.S. Batth, D. Lin, S. Friis, A. Bartels, A. Krzyzanowska, H. Xue, J.M. Moreira, tumors (13) supports our hypothesis that depletion of REST A. Bjartell, Y. Wang, J.V. Olsen, C.C. Collins, D. Iglesias-Gato expression (as a result of ADTs) would be deleterious for the Analysis and interpretation of data (e.g., statistical analysis, biostatistics, prostate cancer cells if combined with an active p53 signaling computational analysis): A. Flores-Morales, T.B. Bergmann, C. Lavallee, pathway (Fig. 4C). T.S. Batth, M. Lerdrup, G. Kristensen, K.H. Hansen, M.A. Røder, K. Brasso, J.M. Moreira, D. Iglesias-Gato Expression of proteins that regulate cell-cycle progression is a Writing, review, and/or revision of the manuscript: A. Flores-Morales, S. Friis, hallmark of NEPC. Many of these proteins are controlled G. Kristensen, M.A. Røder, K. Brasso, J.M. Moreira, A. Bjartell, Y. Wang, through the activation of the E2F1 transcription factor (54). D. Iglesias-Gato We could validate this in the NEPC by analyzing the E2F1 Administrative, technical, or material support (i.e., reporting or organiz- chromatin binding regions in LNCaP cells (55) in relation to ing data, constructing databases): A. Flores-Morales, G. Kristensen, our profile of differentially regulated genes in NEPC compared A. Krzyzanowska, K. Brasso, Y. Wang Study supervision: A. Flores-Morales with prostate adenocarcinoma (Supplementary Fig. S4C and Other (pathology): L. Fazli S4D). Importantly, E2F1 also binds chromatin in the proximity of two suggested driver genes of NE prostate cancer: N-MYCN Acknowledgments and AURKB (ref. 17; Supplementary Fig. S4E). The mechanisms This work was supported by grants from the Danish Research Council (DFF leading to E2F1 activation during NED in prostate cancer 4004-00450, Flores-Morales), the Movember Foundation (Flores-Morales, remain unknown; however, these would likely be related to Collins), and the Danish Cancer Society (R90-A6060-14-S2, Flores-Morales). deletion of the Rb1 gene, an E2F1 inhibitor, commonly found This work was also supported by The Canadian Institutes of Health Research mutated in NEPC (7). (Wang), The Terry Fox Research Institute (Wang, Collins), The Lundbeck Foundation, and The Danish Cancer Research Foundation (Friis). The Novo After integrating proteomic, transcriptomic, and ChIP-seq pro- Nordisk Foundation Center for Protein Research, University of Copenhagen is filing, our data suggest that relapse from ADT as NEPC requires a financially supported by the Novo Nordisk Foundation (grant agreement series of concurrent events including loss of AR and REST protein NNF14CC0001). expression, likely in a context of p53 inactivation and inhibition of RB1 function leading to E2F activation and cell-cycle progres- The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in sion (Fig. 5E). Under these conditions, cells would adapt from accordance with 18 U.S.C. Section 1734 solely to indicate this fact. a predominant oxidative phosphorylation-based metabolism, typical of prostate adenocarcinomas, to a more glycolytic one. Received March 4, 2018; revised August 4, 2018; accepted September 25, Reduction in mitochondrial content would suggest enhanced 2018; published first October 1, 2018.

References 1. Grasso CS, Wu YM, Robinson DR, Cao X, Dhanasekaran SM, Khan AP, et al. 6. Hansel DE, Nakayama M, Luo J, Abukhdeir AM, Park BH, Bieberich The mutational landscape of lethal castration-resistant prostate cancer. CJ, et al. Shared TP53 gene mutation in morphologically and phe- Nature 2012;487:239–43. notypically distinct concurrent primary small cell neuroendocrine 2. Gundem G, Van Loo P, Kremeyer B, Alexandrov LB, Tubio JMC, Papaem- carcinoma and adenocarcinoma of the prostate. Prostate 2009;69: manuil E, et al. The evolutionary history of lethal metastatic prostate cancer. 603–9. Nature 2015;520:353–7. 7. Tan HL, Sood A, Rahimi HA, Wang W, Gupta N, Hicks J, et al. Rb loss is 3. Robinson D, Van Allen EM, Wu YM, Schultz N, Lonigro RJ, Mosquera JM, characteristic of prostatic small cell neuroendocrine carcinoma. Clin Can- et al. Integrative clinical genomics of advanced prostate cancer. Cell cer Res 2014;20:890–903. 2015;162:454. 8. Terry S, Beltran H. The many faces of neuroendocrine differentiation in 4. Dai C, Heemers H, Shariﬁ N. Androgen signaling in prostate cancer. Cold prostate cancer progression. Front Oncol 2014;4:60. Spring Harb Perspect Med 2017;7(9). 9. Zhang X, Coleman IM, Brown LG, True LD, Kollath L, Lucas JM, et al. 5. Chen H, Sun Y, Wu C, Magyar CE, Li X, Cheng L, et al. Pathogenesis of SRRM4 expression and the loss of REST activity may promote the emer- prostatic small cell carcinoma involves the inactivation of the P53 pathway. gence of the neuroendocrine phenotype in castration-resistant prostate Endocr Relat Cancer 2012;19:321–31. cancer. Clin Cancer Res 2015;21:4698–708.

www.aacrjournals.org Clin Cancer Res; 2018 OF13

Flores-Morales et al.

10. Li J, Wang Z. The pathology of unusual subtypes of prostate cancer. Chin J 34. Cox J, Hein MY, Luber CA, Paron I, Nagaraj N, Mann M. Accurate proteome- Cancer Res 2016;28:130–43. wide label-free quantification by delayed normalization and maximal 11. Deorah S, Rao MB, Raman R, Gaitonde K, Donovan JF. Survival of patients peptide ratio extraction, termed MaxLFQ. Mol Cell Proteomics 2014;13: with small cell carcinoma of the prostate during 1973–2003: a population- 2513–26. based study. BJU Int 2012;109:824–30. 35. Cox J, Mann M. 1D and 2D annotation enrichment: a statistical method 12. Taplin ME, George DJ, Halabi S, Sanford B, Febbo PG, Hennessy KT, et al. integrating quantitative proteomics with complementary high-throughput Prognostic significance of plasma chromogranin a levels in patients with data. BMC Bioinformatics 2012;13:S12. hormone-refractory prostate cancer treated in Cancer and Leukemia Group 36. Huang da W, Sherman BT, Stephens R, Baseler MW, Lane HC, Lempicki RA. B 9480 study. Urology 2005;66:386–91. DAVID gene ID conversion tool. Bioinformation 2008;2:428–30. 13. Beltran H, Prandi D, Mosquera JM, Benelli M, Puca L, Cyrta J, et al. 37. Cline MS, Smoot M, Cerami E, Kuchinsky A, Landys N, Workman C, et al. Divergent clonal evolution of castration-resistant neuroendocrine prostate Integration of biological networks and gene expression data using cancer. Nat Med 2016;22:298–305. Cytoscape. Nat Protoc 2007;2:2366–82. 14. Beltran H, Rickman DS, Park K, Chae SS, Sboner A, MacDonald TY, et al. 38. Chuan YC, Pang ST, Cedazo-Minguez A, Norstedt G, Pousette A, Flores- Molecular characterization of neuroendocrine prostate cancer and identi- Morales A. Androgen induction of prostate cancer cell invasion is fication of new drug targets. Cancer Discov 2011;1:487–95. mediated by ezrin. J Biol Chem 2006;281:29938–48. 15. Beltran H, Tagawa ST, Park K, MacDonald T, Milowsky MI, Mosquera JM, 39. Iglesias-Gato D, Chuan YC, Wikstrom P, Augsten S, Jiang N, Niu Y, et al. et al. Challenges in recognizing treatment-related neuroendocrine prostate SOCS2 mediates the cross talk between androgen and growth hormone cancer. J Clin Oncol 2012;30:e386–9. signaling in prostate cancer. Carcinogenesis 2014;35:24–33. 16. Hirano D, Okada Y, Minei S, Takimoto Y, Nemoto N. Neuroendocrine 40. LapukAV,WuC,WyattAW,McPhersonA,McConeghyBJ,Brahmbhatt differentiation in hormone refractory prostate cancer following androgen S, et al. From sequence to molecular pathology, and a mechanism deprivation therapy. Eur Urol 2004;45:586–92; discussion 92. driving the neuroendocrine phenotype in prostate cancer. J Pathol 17. Mosquera JM, Beltran H, Park K, MacDonald TY, Robinson BD, Tagawa ST, 2012;227:286–97. et al. Concurrent AURKA and MYCN gene amplifications are harbingers of 41. Rickman DS, Beltran H, Demichelis F, Rubin MA. Biology and evolution of lethal treatment-related neuroendocrine prostate cancer. Neoplasia 2013; poorly differentiated neuroendocrine tumors. Nat Med 2017;23:1–10. 15:1–10. 42. Berthon P, Cussenot O, Hopwood L, Leduc A, Maitland N. Functional 18. Svensson C, Ceder J, Iglesias-Gato D, Chuan YC, Pang ST, Bjartell A, et al. expression of sv40 in normal human prostatic epithelial and fibroblastic REST mediates androgen receptor actions on gene repression and predicts cells - differentiation pattern of nontumorigenic cell-lines. Int J Oncol early recurrence of prostate cancer. Nucleic Acids Res 2014;42:999–1015. 1995;6:333–43. 19. Chang YT, Lin TP, Campbell M, Pan CC, Lee SH, Lee HC, et al. REST is a 43. Ahuja D, Saenz-Robles MT, Pipas JM. SV40 large T antigen targets multiple crucial regulator for acquiring EMT-like and stemness phenotypes in cellular pathways to elicit cellular transformation. Oncogene 2005;24: hormone-refractory prostate cancer. Sci Rep 2017;7:42795. 7729–45. 20. Lin D, Wyatt AW, Xue H, Wang Y, Dong X, Haegert A, et al. High fidelity 44. Mu P, Zhang Z, Benelli M, Karthaus WR, Hoover E, Chen CC, et al. SOX2 patient-derived xenografts for accelerating prostate cancer discovery and promotes lineage plasticity and antiandrogen resistance in TP53- and RB1- drug development. Cancer Res 2014;74:1272–83. deficient prostate cancer. Science 2017;355:84–8. 21. Ahlqvist K, Saamarthy K, Syed Khaja AS, Bjartell A, Massoumi R. Expression 45. Stark JR, Perner S, Stampfer MJ, Sinnott JA, Finn S, Eisenstein AS, et al. of Id proteins is regulated by the Bcl-3 proto-oncogene in prostate cancer. Gleason score and lethal prostate cancer: does 3 þ 4 ¼ 4 þ 3? J Clin Oncol Oncogene 2013;32:1601–8. 2009;27:3459–64. 22. Schemper M, Smith TL. A note on quantifying follow-up in studies of 46. Iglesias-Gato D, Thysell E, Tyanova S, Crnalic S, Santos A, Lima TS, et al. The failure time. Control Clin Trials 1996;17:343–6. proteome of prostate cancer bone metastasis reveals heterogeneity with 23. Gray RJ. A Class of K-Sample tests for comparing the cumulative incidence prognostic implications. Clin Cancer Res 2018; doi: 10.1158/1078-0432. of a competing risk. Ann Statist 1988;16:1141–54. CCR-18-1229. [Epub ahead of print]. 24. Taylor BS, Schultz N, Hieronymus H, Gopalan A, Xiao Y, Carver BS, et al. 47. Spratt DE, Gavane S, Tarlinton L, Fareedy SB, Doran MG, Zelefsky MJ, et al. Integrative genomic profiling of human prostate cancer. Cancer Cell 2010; Utility of FDG-PET in clinical neuroendocrine prostate cancer. Prostate 18:11–22. 2014;74:1153–9. 25. Langmead B, Trapnell C, Pop M, Salzberg SL. Ultrafast and memory- 48. Wang W, Epstein JI. Small cell carcinoma of the prostate. A morphologic efficient alignment of short DNA sequences to the human genome. and immunohistochemical study of 95 cases. Am J Surg Pathol 2008; Genome Biol 2009;10:R25. 32:65–71. 26. Li H, Handsaker B, Wysoker A, Fennell T, Ruan J, Homer N, et al. The 49. Castellon E, Venegas K, Saenz L, Contreras H, Huidobro C. Secretion of Sequence Alignment/Map format and SAMtools. Bioinformatics 2009;25: prostatic specific antigen, proliferative activity and androgen response in 2078–9. epithelial-stromal co-cultures from human prostate carcinoma. Int J 27. Quinlan AR, Hall IM. BEDTools: a flexible suite of utilities for comparing Androl 2005;28:39–46. genomic features. Bioinformatics 2010;26:841–2. 50. Li Y, Donmez N, Sahinalp C, Xie N, Wang Y, Xue H, et al. SRRM4 Drives 28. Lerdrup M, Johansen JV, Agrawal-Singh S, Hansen K. An interactive neuroendocrine transdifferentiation of prostate adenocarcinoma under environment for agile analysis and visualization of ChIP-sequencing data. androgen receptor pathway inhibition. Eur Urol 2017;71:68–78. Nat Struct Mol Biol 2016;23:349–57. 51. Adolf K, Wagner L, Bergh A, Stattin P, Ottosen P, Borre M, et al. Secretagogin 29. Iglesias-Gato D, Wikstrom P, Tyanova S, Lavallee C, Thysell E, Carlsson J, is a new neuroendocrine marker in the human prostate. Prostate et al. The proteome of primary prostate cancer. Eur Urol 2016;69:942–52. 2007;67:472–84. 30. Ostasiewicz P, Zielinska DF, Mann M, Wisniewski JR. Proteome, phos- 52. Nechiporuk T, McGann J, Mullendorff K, Hsieh J, Wurst W, Floss T, et al. phoproteome, and N-glycoproteome are quantitatively preserved in for- The REST remodeling complex protects genomic integrity during embry- malin-fixed paraffin-embedded tissue and analyzable by high-resolution onic neurogenesis. Elife 2016;5:e09584. mass spectrometry. J Proteome Res 2010;9:3688–700. 53. Akamatsu S, Wyatt AW, Lin D, Lysakowski S, Zhang F, Kim S, et al. The 31. Olsen JV, Schwartz JC, Griep-Raming J, Nielsen ML, Damoc E, Denisov E, Placental Gene PEG10 promotes progression of neuroendocrine prostate et al. A dual pressure linear ion trap Orbitrap instrument with very high cancer. Cell Rep 2015;12:922–36. sequencing speed. Mol Cell Proteomics 2009;8:2759–69. 54. Bertoli C, Skotheim JM, de Bruin RA. Control of cell cycle transcription

32. Cox J, Mann M. MaxQuant enables high peptide identiﬁcation rates, during G1 and S phases. Nat Rev Mol Cell Biol 2013;14:518–28. individualized p.p.b.-range mass accuracies and proteome-wide protein 55. Ramos-Montoya A, Lamb AD, Russell R, Carroll T, Jurmeister S, quantiﬁcation. Nat Biotechnol 2008;26:1367–72. Galeano-Dalmau N, et al. HES6 drives a critical AR transcriptional 33. Cox J, Neuhauser N, Michalski A, Scheltema RA, Olsen JV, Mann M. programme to induce castration-resistant prostate cancer through acti- Andromeda: a peptide search engine integrated into the MaxQuant envi- vation of an E2F1-mediated cell cycle network. EMBO Mol Med ronment. J Proteome Res 2011;10:1794–805. 2014;6:651–61.

OF14 Clin Cancer Res; 2018 Clinical Cancer Research

Proteogenomic Characterization of Patient-Derived Xenografts Highlights the Role of REST in Neuroendocrine Differentiation of Castration-Resistant Prostate Cancer

Amilcar Flores-Morales, Tobias B. Bergmann, Charlotte Lavallee, et al.

Clin Cancer Res Published OnlineFirst October 1, 2018.

Updated version Access the most recent version of this article at: doi:10.1158/1078-0432.CCR-18-0729

Supplementary Access the most recent supplemental material at: Material http://clincancerres.aacrjournals.org/content/suppl/2018/09/29/1078-0432.CCR-18-0729.DC1

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://clincancerres.aacrjournals.org/content/early/2018/11/18/1078-0432.CCR-18-0729. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.