In Vivo Quantitative Phosphoproteomic Profiling Identifies Novel Regulators

ORIGINAL ARTICLE In vivo quantitative phosphoproteomic proﬁling identiﬁes novel regulators of castration-resistant prostate cancer growth

N Jiang1,2, K Hjorth-Jensen1, O Hekmat1, D Iglesias-Gato1, T Kruse1, C Wang2,WWei2,BKe2,BYan2, Y Niu2, JV Olsen1 and A Flores-Morales1

Prostate cancer remains a leading cause of cancer-related mortality worldwide owing to our inability to treat effectively castration- resistant tumors. To understand the signaling mechanisms sustaining castration-resistant growth, we implemented a mass spectrometry-based quantitative proteomic approach and use it to compare protein phosphorylation in orthotopic xenograft tumors grown in either intact or castrated mice. This investigation identified changes in phosphorylation of signaling proteins such as MEK, LYN, PRAS40, YAP1 and PAK2, indicating the concomitant activation of several oncogenic pathways in castration-resistant tumors, a notion that was confirmed by tumor transcriptome analysis. Further analysis demonstrated that the activation of mTORC1, PAK2 and the increased levels of YAP1 in castration-resistant tumors can be explained by the loss of androgen inhibitory actions. The analysis of clinical samples demonstrated elevated levels of PAK2 and YAP1 in castration-resistant tumors, whereas knockdown experiments in androgen-independent cells demonstrated that both YAP1 and PAK2 regulate cell colony formation and cell invasion activity. PAK2 also influenced cell proliferation and mitotic timing. Interestingly, these phenotypic changes occur in the absence of obvious alterations in the activity of AKT, MAPK or mTORC1 pathways, suggesting that PAK2 and YAP1 may represent novel targets for the treatment of castration-resistant prostate cancer. Pharmacologic inhibitors of PAK2 (PF-3758309) and YAP1 (Verteporfin) were able to inhibit the growth of androgen-independent PC3 xenografts. This work demonstrates the power of applying high-resolution mass spectrometry in the proteomic profiling of tumors grown in vivo for the identification of novel and clinically relevant regulatory proteins.

Oncogene (2015) 34, 2764–2776; doi:10.1038/onc.2014.206; published online 28 July 2014

INTRODUCTION findings have been observed using the human LNCaP xenograft 11 Most patients with castration-resistant prostate cancer (CRPCa) die model, which respond better to a combined treatment of within 2 years after diagnosis.1,2 Studies of gene copy number castration and rapamycin than to the each of the treatments 12 variations, transcriptomics and most recently exon sequencing alone. In contrast to these findings, phase II clinical trials have have revealed recurrent genetic changes associated with CRPCa.3,4 shown little effects of the rapamycin analogue RAD001 as an add- Loss of phosphatase and tensin homolog (PTEN) is the most on therapy to the AR antagonist bicalutamide for the treatment of 13 common genetic alteration observed in prostate cancer irrespec- castration-resistant prostate tumors in human patients. Likewise, tive of the tumor stage, affecting approximately 45% of all tumors; mTORC1 inhibition alone is ineffective in reducing tumor load in while the amplification of the androgen receptor (AR) gene (20%) the prostate-specific PTEN knockout model,14 indicating that or point mutations within the AR (25%) are the most frequently additional pathways are critical for the growth of castration- observed genetic alteration specific to CRPCa.5–7 Also associated resistant tumors. with advanced tumors is an increased frequency of point Phosphorylation is a reversible posttranslational modification mutations within the p53 tumor suppressor gene, loss of Rb and that can inform about the activity status of kinase-driven signaling amplification of c-myc.4 How these genetic events regulate the pathways. Mass spectrometry (MS)-based proteomic profiling in pathways that promote survival and growth in CRPCa remains combination with SILAC labeling has been used to analyze growth poorly understood. factor-driven kinase-dependent signaling pathways in tissue Recent studies using genetically modified mice have demon- cultures.15,16 This approach allows for the precise and unbiased strated that prostate tumors initiated by PTEN inactivation develop quantitation of phosphorylation changes in hundreds of peptides into castration-resistant tumors that exhibit reduced expression of in a biologic sample.17 So far, this powerful methodology has been androgen-regulated genes, as a consequence of reduced levels of mainly applied to studies in vitro. Here, we take advantage of the AR expression and activity.8,9 Moreover, PTEN-negative tumors are capacity of LNCaP cells to grow both in culture and as xenografts resistant to androgen deprivation but exhibit enhanced sensitivity to perform a quantitative proteomic profiling of LNCaP tumors to treatments with PI3 kinase and mTORC1 inhibitors, thereby grown orthotopically in mice. To gain a systemic understanding demonstrating a dependency for the mTORC1 pathway for growth of the mechanisms that contribute to the growth of CRPCa, a in conditions of low androgen circulating levels.8–10 Similar comparative analysis was performed of tumors grown in intact or

1The Novo Nordisk Foundation Center for Protein Research, Department of Health Science, Faculty of Health and Medical Sciences, University of Copenhagen, Copenhagen, Denmark and 2Tianjin Institute of Urology, Tianjin Medical University, Tianjin, China. Correspondence: Professor Y Niu, Tianjin Institute of Urology, 23 Pingjiang Road, Hexi District, 300211 Tianjin, China or Professor A Flores-Morales, The Novo Nordisk Foundation Center for Protein Research, Department of Health Science, Faculty of Health and Medical Sciences, University of Copenhagen, Blegdamsvej 3b, Building 6.1, Copenhagen 2200, Denmark. E-mail: [email protected] or [email protected] Received 4 December 2013; revised 21 April 2014; accepted 28 May 2014; published online 28 July 2014 Castration-resistant prostate cancer phosphoproteomics N Jiang et al 2765 castrated mice. SILAC-based quantitation identified phosphoryla- antibodies (Figure 2a). We also analyzed the status of the key tion sites that changed significantly depending on the hormonal signaling proteins within the mTORC1 and MAPK pathways and status of the host, thereby allowing for the identification of novel found that both mTORC1 and its downstream target p70S6 kinase pathways sustaining CRPCa growth. exhibited enhanced phosphorylation in CR-LNCAP, indicative of increased activity.24 The phosphorylation levels of the upstream fi RESULTS negative regulator AMPK were not signi cantly changed, whereas AKT phosphorylation changes exhibited a more complex pattern; Phosphoproteomic analysis of castration-resistant LNCAP tumors the phosphorylation of AKT on Thr308 increased, whereas AKT- LNCaP xenografts mimic many of the features of human prostate Ser473 phosphorylation was reduced in CR-LNCaP tumors in tumors, which make them a valuable model to study the comparison with androgen-sensitive xenografts (Figure 2). These 11,18 mechanism of progression to CRPCa. To obtain an overview results suggest that the two distinct pathways that contribute to of these mechanisms LNCaP xenografts were grown orthotopically maximal activation of AKT through increased PDK1 (AKT Thr308) in intact and castrated mice. Analysis of tumor growth showed and mTORC2 (AKT Ser473) activity thus appear to be oppositely that castration-resistant LNCaP (CR-LNCaP) tumors proliferate at a regulated in the transition from the androgen-sensitive to the higher rate than androgen-sensitive tumors (hormone-sensitive castration-resistant state.28 The activities of additional positive LNCaP (HS-LNCaP)) at the time of the harvest, 60 days after regulators of mTORC1, the extracellular-signal-regulated kinase fi surgical orthotopic implantation (Figures 1a and b). These ndings (ERK) kinases 1/2, were also increased in CR-LNCaP tumors, mimic the situation with human CRPCa, which proliferate at a suggesting that ERKs may contribute to growth in these fi 19,20 higher rate than castration-naive, prostate-con ned tumors. conditions possibly influencing mTORC1 activation.29 Finally, In addition, CR-LNCAP tumors showed an increased number of we also confirmed the upregulation of phospho-PAK2 levels Ki67-positive cells and diminished levels of active caspase-3, (Figure 2). Immunohistochemical analysis showed that both PAK2 indicating that an enhanced proliferative rate and reduced and YAP1 exhibit significant increased expression in CR-LNCaP apoptosis contribute to the increased growth rate of tumors xenograft models, with both proteins showing a predominantly grown in castrated mice (Figures 1c and d). cytosolic expression and YAP1 occasionally staining the nucleus Next, we performed a comprehensive quantitative phosphoproteomic profiling using SILAC-labeled LNCaP-FGC cells as (Supplementary Figure S1). internal, spiked-in controls in all samples analyzed.21 To account for biologic variation, four individual tumors were analyzed for Transcript profiling reveals pathways deregulated in CR-LNCaP each experimental group. In parallel, we also performed a tumors genome-wide transcriptomic analysis. A summary of the metho- If the pathways identified as regulated in castration-resistant dology is presented in Figure 1e. cancer by phosphoproteomic analysis are functionally relevant, we A total of 2782 phosphorylated peptides from a total of 1212 would expect the transcriptional program of these tumors to proteins were quantified in at least one of the samples. Sufficient reflect changes in mTORC1-, AKT-, MAPK-, YAP1- and PAK2- data across biologic replicates were obtained for 800 of these mediated pathways. Therefore, we measured changes in mRNA phosphopeptides. Statistical analysis identified differential expres- levels for all human genes in HS- and CR-LNCAP tumors and sion for 98 peptides defined as fold changes of more than 50%, analyzed whether the profile of differentially expressed genes Po0.05 (Supplementary Table S1). Of these, 44 phosphopeptides contained signatures corresponding to the activation of these showed increased levels (Table 1), whereas 54 exhibited pathways. The transcriptomic analysis identified 514 genes whose decreased levels in castrated tumors as compared with those mRNA levels were induced and 291 genes repressed in the CRPCa grown in intact mice (Supplementary Table S1). Gene ontology as compared with control tumors (fold 42, false discovery rate analysis indicated that proteins involved in cytoskeletal organiza- o0.05) (Supplementary Table S3). Gene set enrichment analysis tion, regulation of cell morphogenesis and intracellular transport was performed where the profile of genes differentially expressed were enriched among proteins with increased phosphorylation in LNCaP-CR vs LNCaP-HS was examined for similarities against (Supplementary Table S2). Phosphorylation of ACC1, the rate- fi 22 more than 3000 different expression pro les collected within the limiting enzyme in fatty acid synthesis, was the most prominent MSigDB database30 and we found strong and statistically effect observed, followed by changes in tyrosine kinase Lyn-Ser13 significant similarities with several expression profiles (Figure 3 phosphorylation. We also identified elevated phosphorylation and Supplementary Table S5). Next, we visualized these results levels of cytoskeleton regulators: MAP1B, MAP4 and Stathmin, as using Cytoscape to highlight the network of relationships well as signaling molecules MEK (P = 0.06) and PRSA40 in CR- between these profiles.31,32 The genes upregulated in CR-LNCAP LNCaP. MEK and PRAS40 are key members of the MAPK and fi mTORC1 signaling pathway, suggesting that these pathways are was signi cantly similar to the sets of genes upregulated in basal involved in CR-LNCaP growth.23–25 Furthermore, we also observed vs luminal types of breast cancer or overexpressed in ductal an increased phosphorylation of YAP1, a transcriptional coacti- invasive breast carcinomas. This is in agreement with the fi vator that has a key role in the mammalian hippo signaling signi cant upregulation in CR-LNCaP of genes involved in pathway,26 and of PAK2, a regulator of cell motility that mediate mesenchymal differentiation related to cytosqueletal function the actions of Cdc42 and Rac small GTPases.27 On the other hand, (Supplementary Table S4 and Supplementary Figure S2). In line proteins involved in the regulation of nucleic acid metabolism, with the downregulation of AKT-S473P detected in CR-LNCAP, we including the regulation of transcription and RNA processing and also observed that genes upregulated in these tumors are nucleosome organization, were among the proteins with dimin- enriched for genes regulated by FOXO1 (Figure 3), a transcription 28 ished phosphorylation (Supplementary Table S1). factor inhibited by AKT activation. Several genes repressed by We then proceeded to validate our phosphoproteomic data and KRAS transformation of prostate tissue (KRAS PROSTATE_UP to confirm the implication of the MAPK and mTORC1 pathways V1_DN) were also downregulated in LNCaP-CR in line with the and of PAK2 and YAP1 in CR-LNCaP growth. Four randomly increased activity of ERK pathway detected in these tumors. assigned individual samples within each experimental group were Interestingly, the genes upregulated in CR-LNCaP were strongly subjected to western blot analysis. In good agreement with the enriched for the gene set denoted ‘CORDENONSI YAP CONSERVED MS data analysis, CR-LNCaP cells exhibit increased phosphoryla- SIGNATURE’ comprising genes regulated by YAP1, adding support tion of ACC, PAK2 and PRASA40 as indicated by phosphospecific for an increased activity of YAP1 in CR-LNCaP.

© 2015 Macmillan Publishers Limited Oncogene (2015) 2764 – 2776 2766 nlss assetawr sdt dniypopoetdsadcagsrltv otesie ncnrlwr quanti were control in spiked chromatography the tumor with to shown. liquid pro 1:1 relative transcript mixed are tandem changes were genome-wide and images cells and The phosphopeptides LNCaP variants. Representative enrichment identify Student labeled isotopic SILAC to peptide analyzed. phosphopeptide from used of extracts were digestion, were intensities Protein spectra mice. tumors trypsin Mass nude (CR) analysis. Balb/c prior intact castration-resistant samples or castrated or protein either the in (HS) posttransplant, grown days hormone-senstitive xenografts 60 killing either of of ( time the tumors At Three weekly. signi ( have times group. mice two each castrated ( recorded in in analyzed was grown were volume mice tumors Four tumor subcutaneous and of mice rate nude growth Balb/c into orthotopically 1. Figure noee(05 2764 (2015) Oncogene d c h rhtpcxngat eehretd6 aspstasln n nlzdb H sn nioisfrK6 n ciecaspase-3. active and Ki67 for antibodies using IHC by analyzed and posttransplant days 60 harvested were xenografts orthotopic The ) Quanti ) “Heavy reference”

Metabolic labeling cell number / field 3 (Arg Tumor volume (mm ) 100 150 200 250 300 LNCaP (Arg Castrated mice 1000 1200 1400 ’ 50 s Control mice 200 400 600 800 0 LNCaP 0 nrae rwhrt fLCPxngat rw ncsrtdmc.( mice. castrated in grown xenografts LNCaP of rate growth Increased t , Lys ﬁ ts a sdt sessaitclsigni statistical assess to used was -test 10 0 aino h H nlsso i7adatv aps- nH rC uos ( tumors. CR or HS in caspase-3 active and Ki67 of analysis IHC the of cation , Lys SC SCR HS CR HS 0 0 ) caspase 3 8 atainrssatpott acrphosphoproteomics cancer prostate Castration-resistant p<0.05 ) 22 36 24 12 Castrated mice Intact mice Mix tumorproteinextracts SILAC proteinreference – 2776 Balb/c nudemice Xenografted Days

1:1 and cell number / field 100 150 200 250 50 0 860 48 b mgso xie NA rhtpcxngat rw ncsrtdadcnrlmc sindicated. as mice control and castrated in grown xenografts orthotopic LNCAP excised of Images ) p<0.01 ki67 Digestion Xenograft Orthotopic tumor proteinextracts Trypsin ﬁ

ac (* cance Ki67 Caspase 3 Phosphopeptide Jiang N Tumor disection enrichment P Castrated o Control TiO (n=4) (n=4) 0.05). tal et 2 Hormone sensitive A ﬁ ﬁ Reverse eo Na eorfe uoswsas nlzdadcompared. and analyzed also was tumors xenografted LNCaP of le phase atyices ncmaio ihtmr rw ncnrlmice. control in grown tumors with comparison in increase cantly C 18 mRNA a LNCAP-FGC orthotopicxenografts Na eort eetaslne uctnosyand subcutaneously transplanted were xenograts LNCaP ) LTQ-Orbitrap Velos High Resolution LC-MS/MS Transcript Profiling e Affymetrix xeietlsrtg o h nlsso LNCaP of analysis the for strategy Experimental )

Intensity 05McilnPbihr Limited Publishers Macmillan 2015 © MaxQuant/Perseus SILAC pairs Castration resistant m/z BIOINFORMATICS – admms spectrometry mass tandem

Intensity ﬁ dbsdo the on based ed m/z Castration-resistant prostate cancer phosphoproteomics N Jiang et al 2767

Table 1. Peptides exhibiting statistically signiﬁcant (Po0.05) increased phosphorylation in castration-resistant compared with hormone-sensitive LNCaP xenograft tissue

Symbol Uniprot Sequence Location Probability Fold change

ACAC Q13085 FIIGSVSEDNSED 62 0.54 3.20 ACAC Q13085 SRFIIGSVSEDNS 60 0.54 3.02 LYN Q6NUK7 KGKDSLSDDGVDL 104 0.99 2.75 BAT2D1 Q9Y520 QKLPDLSPVENKE 2107 1.00 2.03 MYH9 P35579 KGAGDGSDEEVDG 1943 1.00 1.90 MAP1B P46821 KGEAEQSEEEADE 1016 1.00 1.89 MAP1B P46821 RKLGDVSPTQIDV 1501 0.50 0.85 MAP1B P46821 IERTTKSPSDSGY 1915 0.82 1.12 EIF3J O75822 AAAAGDSDSWDAD 11 0.92 1.78 STMN1 P16949 VPEFPLSPPKKKD 38 1.00 1.42 NOB1 Q9ULX3 EDRKDDSDDDGGG 211 1.00 1.37 CTAGE5 O15320 ETRAFLSPPTLLE 541 1.00 1.35 MAP4 P27816 AQAKVGSLDNVGH 1073 1.00 1.29 MAP4 P27816 ALGKDVTPPPETE 521 1.00 0.59 MAP4 P27816 GLLKDMSPLSETE 507 1.00 0.58 LIMCH1 Q9UPQ0 KVVKPKSPEPEAT 718 1.00 1.26 LIMCH1 Q9UPQ0 SPSSEKSPVTTPF 769 0.95 1.11 LIMCH1 Q9UPQ0 RSRQTPSPDVVLR 217 0.99 0.85 AKT1S1 Q96B36 TQQYAKSLPVSVP 203 1.00 1.24 ARHGAP35 Q9NRY4 TSFSVGSDDELGP 1179 0.97 1.16 NFIX B4DHW2 YPGTGRSPAAGSS 309 1.00 1.05 ESYT2 A0FGR8 PTPSIASDISLPI 779 0.77 1.04 ESYT2 A0FGR8 SIASDISLPIATQ 782 0.93 1.02 TMF1 P82094 SVSEINSDDELSG 344 0.80 0.99 DENND4C Q5VZ89 SIVKVPSGIFDVN 753 1.00 0.98 SLC9A1 P19634 SRARIGSDPLAYE 703 1.00 0.97 CHMP2B Q9UQN3 TSKATISDEEIER 199 0.99 0.95 VCL P18206 EAIDTKSLLDASE 721 1.00 0.93 NAP1L1 P55209 IDNKEQSELDQDL 10 0.97 0.93 BGL P50851 GDDDTLSSVDEKD 2064 0.65 0.91 PDCD4 Q53EL6 GLTVPTSPKGRLL 94 0.84 0.90 YAP1 B4DTY1 AQHLRQSSFEIPD 163 0.50 0.89 YAP1 B4DTY1 QHLRQSSFEIPDD 164 0.50 0.89 YAP1 B4DTY1 VRGDSETDLEALF 63 0.50 0.77 GOLGA4 Q13439 LQLRVPSVESLFR 93 1.00 0.85 LYSMD1 Q96S90 LFNGLDSEEEKDG 99 1.00 0.83 USO1 Q86TB8 EDEDDESEDPGKD 963 1.00 0.82 USO1 Q86TB8 EEDELESGDQEDE 953 1.00 0.59 HSP90AA1 P07900 ERDKEVSDDEAEE 353 1.00 0.82 GBF1 Q92538 PDAGAQSDSELPS 1298 0.48 0.80 STMN2 Q93045 PKKKDLSLEEIQK 46 1.00 0.75 KIAA1321 Q7Z417 GLERNDSWGSFDL 652 1.00 0.71 PKP3 Q9Y446 TLQRLSSGFDDID 314 0.50 0.62 PAK2 Q13177 VKQKYLSFTPPEK 141 0.50 0.59 The peptide sequence as well as the Uniprot database identiﬁer of the corresponding protein is presented. The location of the phosphorylation site mapped to the amino-acid sequence of the indicated Uniprot entry is also listed as well as the probability that the site of phosphorylation indicated is correctly located. The fold change indicated is the log2-transformed ratio of the phosphopeptide levels measured in castration-resistant compared with those measuredin hormone-sensitive tumors.

Study of the involvement of PAK2 and YAP1 in CRPCa growth reduced again by long-term androgen treatment (Figure 4b). The differential phosphorylation of signaling proteins in Moreover, inhibition of AR with MDV3100 (Enzalutamide) caused a castration-resistant vs hormone-sensitive tumors suggested that significant induction of YAP1, PAK2 and PAK2-S141P levels they may be regulated by androgens. Therefore, we studied how (Figure 4c). Overall, these experiments suggest that changes in fl signaling proteins observed in CR-LNCaP tumors can, in part, be dihydrotestosterone (DHT) (10 nM) treatment in uences the fi phosphorylation of AKT, ERK, mTORC1 and PAK2, as well as the explained by androgen de ciency. While the involvement of ERK and mTOR pathways in the levels of YAP1. A time-course study of DHT effects on LNCaP cells – growth of CRPCa has been well documented,33 37 less is known showed that androgen treatment represses the phosphorylation about the role of PAK2 and YAP1. To investigate a potential role of of the mTORC1 downstream targets p70S6 kinase as well as the these proteins, we analyzed prostatectomy samples obtained from PRAS40-S183 and AKT-Thr308 phosphorylation (Figure 4a). The patients with benign prostate hyperplasia (BPH), localized PCA and phosphorylation of ERK displayed a different pattern, with DHT CRPC for the expression of PAK2, PAK2-S141P and YAP1 by repressing ERK activation in a transient manner while no effects immunohistochemistry (IHC) (Figure 5 and Supplementary were observed in the case of PAK2, PAK2-S141P or YAP1 levels Figure S3). The expression of both PAK2 and PAK2-S141P was within the time period studied. However, an extended analysis circumscribed to the epithelial compartment, with PAK2 showing using androgen treatments up to 72 hours showed that YAP1 and a pattern of progressively, significantly increased expression PAK2 levels increased upon androgen withdrawal and were from low- to high-grade localized tumors to CRPC. In the case of

12341234

AKT-T308P AKT-T308P AKT-S473P PRAS40-S183P PRAS40 2 1.2 3.0 2 1 2.5 * AKT-S473P 1.5 1.5 0.8 2.0 * 1 0.6 1.5 1 AKT 0.4 1.0 0.5 0.5 0.2 0.5 PRAS40-S183P 0 0 0 0 HS CR HS CR HS CR HS CR

PRAS40 ERK1/2 -S217/ p70S6K- T389P mTORC1-S2448P S221P p70S6K-T389P 2.5 2 * 2.5 * * 2 2 1.5 mTORC1-S2448P 1.5 1.5 1 1 1 mTORC1 0.5 0.5 0.5 ERK1/2 0 0 0 S217/S221P HS CR HS CR HS CR PAK2 PAK2-S141P ACC-S79P AMPK-T172P ERK 1/2 p=0.08 * 2.5 2.5 * 2 2 2 PAK2-S141P 2 1.5 1.5 1.5 1.5 1 1 PAK2 1 1 0.5 0.5 0.5 0.5 ACC-S79P 0 0 0 0 HS CR HS CR HS CR HS CR

AMPK-T172P

β-actin

Figure 2. Changes in the phosphorylation levels of signaling protein in hormone-sensitive (HS) and castration-resistant (CR) LNCaP orthotopic xenografts. (a) The levels of indicated proteins and phosphoproteins were measured by western blot in prostate tissue from individual mice with orthotopically grown xenografts. (b) The signal was quantiﬁed and statistical signiﬁcance analyzed by Student’s t-test (*Po0.05).

PAK2-S141P, we found the average expression intensity levels to clonogenic activity in Matrigel as well as in their invasion activity increase significantly when comparing low-grade localized tumors (Figures 6b and c). In addition, proliferation of C4-2b cells was to high-grade localized tumors or CRPCa, although no significant clearly inhibited following PAK2 downregulation. In contrast, YAP1 differences were observed between the two latter categories. The downregulation failed to inhibit cell proliferation (Figure 6d and levels of YAP1 were found to be significantly increased in CRPC Supplementary Figure S4). Because PAK2 effects on cell division relative to localized PCA as well as in high-grade vs low-grade are poorly understood, we used live cell microscopy to analyze tumors. In contrast, no differences were found between localized mitotic timing in cells transfected with PAK2 targeting siRNAs. As tumor and the control tissues obtained from BPH patients, as shown in Figure 6e, cells expressing reduced levels of PAK2 show these samples exhibited strong YAP1 expression localized to the a significant delay in the duration of mitosis, which can help to basal cell layer within the epithelial compartment. In addition, we explain the observed effects in cell proliferation. In conclusion, observed diffuse cytosolic YAP1 staining in stromal compartment YAP1 and PAK2 have independent and overlapping functions in of samples from patients with BPH and low-grade localized androgen-independent prostate cancer cells that may contribute tumors, while nuclear YAP1 staining was evident in basal epithelial to the growth and spread of castration-resistant tumors. cells of BPH samples as well as in high-grade localized and castration-resistant tumors. Having demonstrated the increased expression of PAK2 and Pharmacologic inhibition of PAK2 and YAP1 inhibits CRPCa growth YAP1 in CRPCa and their regulation by androgens, we next in vivo analyzed their regulatory functions in tumor cells. Transfection of The results presented so far suggest that pharmacologic inhibition C4-2b cells, an androgen-independent cell line derived from of YAP1 and PAK2 may serve to inhibit growth of CRPCa tumors. CR-LNCaP xenografts,38 with specific small interfering RNAs To test this possibility, we used preclinical models of CRPCa (siRNAs) targeting PAK2 and YAP1 resulted in effective down- growth and previously described PAK2 kinase (IPA-3 and – regulation of the targeted proteins (Figure 6a). Measurements of PF-3758309) and YAP1 (Verteporfin) inhibitors.39 41 First, the AKT-S473P, AKT-T308P, P-ERK1/2 and p70S6 kinase phosphoryla- proliferation capacity of C4-2b, 22rv1 and PC3 androgen- tion showed no significant changes upon knockdown of PAK2 or independent cells was tested in the presence or absence of the YAP1 protein levels (Figure 6a). The downregulation of both YAP1 inhibitors (Figure 7a and Supplementary Figure S7). In all cases, a and PAK2 in C4-2b cells caused a significant reduction in their significant inhibition of cell proliferation was observed upon

Figure 3. Network of gene sets showing significant similarity with genes identified as differentially expressed in CR-LNCaP vs HS-LNCaP orthotopic xenografts. Gene set enrichment analysis (GSEA) was applied to the mRNA expression data from CR-LNCaP vs HS-LNCaP to identify similarities (false discovery rate o0.1) with data sets contained in the MSigDB from the Broad Institute (Cambridge, MA, USA). The GSEA output files were then analyzed using Cytoscape (Enrichment Map plug-in) to visualize the interconnectivity of the overlapping gene sets. The area of the circles reflects the degree of overlap between the indicated gene sets and the genes differentially expressed in CR-LNCaP vs HS-LNCaP tumors. Gene sets depicted in red or blue are enriched with genes that are increased or reduced in their expression in CR-LNCaP vs HS-LNCaP, respectively. treatment with IPA-3, PF-3758309 or Verteporfin parallel to a led to the identification of several pathways that are concomi- decreased activity of PAK2 (PAK2-T184P) and decreased levels of tantly altered in castration-resistant tumors and seem to YAP1 in cells treated with their respective inhibitors (Figure 7b). contribute to tumor growth in the context of reduced androgen We then tested whether these effects could be observed in vivo. circulation levels. Notably, we found a significant and parallel Mice harboring PC3 xenografts treated with either PF-3758309 or activation of MAPK and mTORC1/2 pathways but only partial Verteporfin showed significantly reduced tumor growth (Figures activation of AKT.28,42 In addition, we found evidences that the 7c and d). IHC analysis showed that PF-3758309-treated tumors PAK2 kinase, a downstream effector of GTPases Cdc42 and Rac, express reduced levels of PAK2-S141P in parallel to diminished cell and YAP1, a key effector of the mammalian hippo signaling proliferation and increased apoptosis, as indicated by reduced pathway, are also overexpressed in castration-resistant tumors. expression of Ki67 and increased active caspase-3 (Figure 7e and In vitro and in vivo studies demonstrated that PAK2 and YAP1 Supplementary Figure S5). Likewise, treatment of mice harboring contribute to the proliferative and invasive capacity of androgen- PC3 xenografts with Verteporfin caused a more modest but independent prostate cancer cells and the growth of castration- significant decrease in the tumor growth. These tumors expreseds resistant tumor xenografts. reduced levels of nuclear YAP1 and the Ki67 antigen proliferation Studies using mouse models have demonstrated that PI3 kinase marker, whereas expression levels of active caspase-3 increased activation resulting from the loss of tumor suppressor PTEN in (Figure 7f and Supplementary Figure S6). These results suggest prostate epithelium antagonizes AR signaling in prostate and that PAK2 and YAP1 signaling proteins may constitute valid drug sustains tumor growth in castrated mice.8,9 In these mice, dual targets for the treatment of CRPCa. inhibition of mTORC1/2 and PI3 kinase restores AR activity and enhances the expression of Her2/3. This effect is also observed in PTEN-negative LNCaP cells, where Her2/3 induction is accompa- DISCUSSION nied by activation of ERK kinase.9 Reciprocally, castration of these We have performed an unbiased analysis of the changes in mice leads to increased phosphorylation of AKT owing to the protein phosphorylation associated with the onset of CRPCa inhibition of PHLPP (pleckstrin homology domain leucine-rich in vivo, using the LNCaP orthotopic xenograft model.11,18 This has repeats protein phosphatase) activity. Based on these results, a

DHT -30″ 2h 4h 8h 24h Full 24h 48h 72h Med. DHT AKT-T308P PAK2 AKT-S473P YAP1

AKT b-actin

PRAS40-S183P

PRAS40 MDV3100 (24h) - + p70S6K-T389P PAK2-S141P

mTORC1-S2448P PAK2 ERK1/2 YAP1 S217/S221P β-actin ERK 1/2

PAK2-S141P

PAK2

YAP1

b-actin

Figure 4. Androgen regulation of signaling proteins elevated in CR-LNCaP xenograft tumors. (a) LNCaP cells were cultured in steroid depleted medium for 48 h and subsequently treated with DHT (10 nM) for the indicated time. Protein levels were analyzed by western blot. (b) LNCaP cells were cultured in steroid depleted medium for 48 h and subsequently treated with DHT (10 nM) or solvent ( − ) for the indicated time. Cells grown in full (steroid containing) media were used as control. Protein levels were analyzed by western blot. (c) The levels of YAP1 and PAK2 were measured in LNCaP cells treated with the AR antagonist MDV3100 (enzalutamide, 100 nM) for 24 h.

model has been proposed where castration-resistant growth of which promote degradation of YAP1, are downregulated in PTEN-null prostate cancer tumors (representing 70% of metas- metastatic prostate cancer.44 Given that Lats1/2 mRNA changes tases in humans) is sustained by pathways downstream of are small and have not been analyzed at the protein level, phosphoinositide 3-kinase, with both AKT1/2 and mTORC1/2 additional experiments would be needed to confirm the role of kinases contributing to these effects.8,9,12,43 Our study confirms Lats1/2 in the regulation of YAP1 levels in castration-resistant the activation of mTORC1 in CR-LNCaP tumors. However, the prostate tumors. Interestingly, we established that both YAP1 and effects on AKT appear more complex considering our findings of PAK2 levels are repressed by AR signaling, thereby providing a increased AKT1-Thr308P and reduced AKT1-Ser473P in compar- possible explanation as to why long-term androgen depravation ison with hormone-sensitive tumors. As peak activity of AKT1 leads to increased levels of PAK2 and YAP1 in castration-resistant requires the phosphorylation of both the 308 and 473 residues, tumors. A similar mechanisms may also explain why both PAK2 these results question the necessity for maximal activation of AKT and YAP1 are upregulated in high-grade localized tumors, as these for the induction of mTOR pathway observed in CR-LNCaP have previously been characterized as deficient in AR signaling.45 tumors.28 They also suggest the presence of additional mechan- The regulation of PAK2 and YAP1 by androgens could be isms driving mTORC1 activation. Indeed, when comparing mediated to some extent by transcriptional repression, as it is CR-LNCaP to HS-LNCaP tumors, we observed strong induction of has been reported that long-term hormone depletion may lead to ERK, which is known to participate in the activation of mTORC1 increased mRNA levels PAK2 and YAP1.43 However, our own through inhibitory phosphorylation of the tuberous sclerosis transcriptomic analysis in vivo failed to confirm those findings. complex complex and has been found to be activated in CRPCa Alternatively, posttranslational mechanisms may have a role, clinical samples.29,33 Furthermore, phosphorylation of PRAS40- especially in the case of YAP1 whose activity is known to be S183, which is required for maximal activation of mTORC1, regulated through controlled, ubiquitin-mediated proteolysis.26 is also increased in these castration-resistant tumors, possibly Indeed, several genes known to modulate YAP1 turnover were compensating for the partial activation of AKT.25 overexpressed in CR-LNCAP (Supplementary Table S3), including Our phosphoproteomic analysis was able to identify novel Ajuba, a tight junction protein that positively regulates YAP1 signaling events that contribute to CRPCa growth and appear to content.46 be of clinical relevance. Specifically, we demonstrated that the Members of the type I family of PAK kinases (PAK1, PAK2 and expression of PAK2 is increased in CR-LNCaP and correlates with PAK3) act downstream of the small GTPases, Rac and Cdc24, and disease progression in patients samples, showing significant have an important role in regulating cytoskeletal dynamics, upregulation in high-grade localized and castration-resistant proliferation and cell survival,27 but their role in prostate cancer tumors. We also demonstrated that YAP1, a transcriptional is poorly understood. We demonstrated that siRNA-mediated coactivator that drives the biologic response of the mammalian targeting of PAK2 in the castration-resistant C4-2b cell line delays hippo signaling pathway, is increased in castration-resistant mitosis and diminishes its proliferation rate and its invasion xenografts and human tumors.41 These findings are in line with activity. On the other hand, YAP1 inhibition has little effects on previous reports showing that mRNAs coding for kinases Lats1/2, adherent cell proliferation but inhibits clonogenic activity and cell

μ 100 μM 100 μM 100 M (Gleason 5) (Gleason Localized PCA Localized

100 μM μ 100 μM 100 M (Gleason 8) (Gleason Localized PCA Localized

100 μM μ 100 M 100 μM CRPCa

μ 100 μM 100 M 100 μM

PAK2 PAK2-S141P YAP1

p<0.01 p<0.01 p<0.05 p<0.01 p<0.01 p=0.05 p<0.05 p<0.01 p<0.05 p<0.05 p<0.01 3 p<0.05 3 3 2 2 2 1 1 1 IHC Score IHC Score 0 0 IHC Score 0

≤6) ≤6) ≤6) BPH BPH BPH CRPC CRPC CRPC Loc(G Loc(G>6) Loc(G Loc(G>6) Loc(G Loc(G>6) Figure 5. PAK2 and YAP1 are upregulated in CRPCa. Clinical specimens of BPH, localized (Loc) prostate cancer with different Gleason scores and CRPC were analyzed by IHC for PAK2, PAK2-S141P and YAP1 expression. The IHC was scored according to the strength of the expression and Analysis of variance statistical analysis was performed to determine significance. Only significant P-values (Po0.05) are indicated. invasion. An interesting observation is that the effects observed may operate independently of these signaling pathways on the after PAK2 and YAP1 knockdown seems to be independent of the regulation of prostate tumor growth. Therefore, PAK2 and YAP1 regulation of the additional signaling pathways we found altered could constitute novel therapeutic targets for the treatment of in CRPCA, as the analysis of the activity of key molecules in the castration-resistant tumors. Indeed, pharmacologic inhibition of ERK, mTORC1 and AKT pathway does not change in response to PAK2 with IPA-3 and PF-3758309 and of YAP1 using Verteporfin PAK2 or YAP1 downregulation. This suggests that PAK2 and YAP1 effectively reduces the growth rate of several CRPCa cell lines

1000 250 +------800 CONTROL 200 -- - PAK2 -#1#2#3 600

siRNA 150 YAP1 - ---#1 #2 #3 100 400

PAK2 No of cells through matrigel colony number 50 200 YAP1 0 0 AKT-T308P CONTROL + -- CONTROL + -- #2 #3 PAK2 - #2 #3

PAK2 - siRNA

AKT-S473P siRNA

p70S6K-P 250 1000

ERK-P 200 800

β-actin 150 600

100 400 No of cells colony number 30 50 through matrigel 200 25 0 0 20 CONTROL + -- CONTROL + - - YAP1 - #2 #3 YAP1 - #2 #3 siRNA

15 siRNA 10 5 siRNA-Luciferase

Doubling time (Hours) 0 200 CONTROL +- -- PAK2 -#1#2#3 DIC siRNA -180 244860 150 25

20 Histone 100 15 siRNA-PAK2 #3

10 Time (minutes)

50 DIC 5 -180 306684

Doubling time (Hours) 0 CONTROL +--- 0

YAP1 -#1#2#3 + --Histone siRNA CONTROL - #2 #3 siRNA PAK2 Figure 6. YAP1 and PAK2 regulate proliferation and invasion of CRPCa cells. (a) Castration-resistant C4-2b cells were transfected with control siRNA or siRNAs (1, 2 and 3) directed against PAK2 or YAP1. PAK2 and YAP1 knockdown and the potential impact on the phosphorylation and/ or levels of signaling proteins were analyzed by western blot. (b and c) Effects of PAK2 and YAP1 knockdown on C4-2 cell clonogenic activity (b) and their capacity to invade through a thick layer of Matrigel (c). (d) Proliferation was monitored in real time using RTCA technology (Roche) and doubling times were determined on this basis. (e) Time in mitosis for cells transfected with PAK2 targeting siRNAs as indicated. The right panel illustrates the phenotype of C4-2 cells during mitosis. Statistical signiﬁcance was analyzed by Student’s t-test (*Po0.05; **Po0.01).

in vitro and the growth of androgen-independent PC3-xeno- growth of castration-resistant prostate tumors. In the light of grafted tumors in vivo. These results are an encouraging first step recent improvements in sample preparation methods, large-scale toward a possible use of PAK2 and YAP1 inhibitors in the quantitative phosphoproteomic analysis of human tumors is now treatment of CRPCa; however, they should be interpreted with feasible, as demonstrated here and elsewhere.34,48 During the caution given the possibility that these compounds may have preparation of this manuscript, Drake et al.34 described the additional cellular targets that are also relevant for their biologic MS-based phosphoproteomic profiling of tumors obtained by activity. rapid autopsy of CRPCa patients. Although this pioneering study Quantitative proteomic profiling has emerged as a powerful focused on tyrosine phosphorylation, which constitute approxi- tool to interrogate changes in the level of proteins and their mately 1% of protein phosphorylation (most of which occurs at PTM in a global and unbiased manner.47 Most of the global Ser and Thr residues), it served to demonstrate the interpatient phosphoproteome studies are performed in cell lines. Here we heterogeneity in tumor tyrosine phosphorylation patterns and developed a novel strategy using SILAC-labeled cultured cells to thus highlights the possibility that individualized treatment can analyze orthotopically xenograted tumors originated from the arise from phosphoproteomic profiling of clinical samples. Indeed, same cell line, allowing for precise quantification of protein our finding that pharmacologic inhibition of PAK2 and YAP1 can expression changes in vivo. This approach has allowed us to suppress castration-resistant tumor growth in vivo demonstrates identify several pathways that concomitantly contribute to the that quantitative phosphoproteome analysis can provide the

Oncogene (2015) 2764 – 2776 © 2015 Macmillan Publishers Limited Castration-resistant prostate cancer phosphoproteomics N Jiang et al 2773 a 140 PC3 140 22rv1 PC3 140 120 120 C4-2b 22rv1 120 100 100 C4-2b 100 80 80 80 60 60 60 40 40 40 20 20 20 Normalized Cell Index Normalized Cell Index 0 0

0 Normalized Cell Index Vehicle 25 uM Vehicle 10 uM Vehicle 100 nM 1 uM 10 uM -20 -20 100 nM 1 uM -20 IPA3 PF-3758309 Verteporfin

bcC4-2b 22rv1 PC3 PC3 xenografts Vehicle 50 PF-3758309 PF-3758309 IPA-3 control 40 PAK2-T184P

PAK2 30

β-actin 20 tumor volume (mm3) C4-2b 22rv1 PC3 10 Vehicle Verteporfin 01234567 days YAP1 400 PC3 xenografts β -actin Verteporfin 300 control d Vehicle Vehicle 200

100 PF-3758309 Verteporfin tumor volume (mm3)

0 01234567 days

e PAK2-S141P Caspase 3 Ki67 f YAP1 Caspase 3 Ki67 1 1 1 1 1 1 0.8 0.8 0.8 0.8 0.8 0.8 0.6 0.6 0.6 0.6 0.6 0.6 0.4 0.4 0.4 0.4 0.4 0.4 0.2 0.2 0.2 0.2 0.2 0.2 0 0 0 0 0 0 Fraction of positive cells Fraction of positive cells Fraction of positive cells Fraction of positive cells Fraction of positive cells Fraction of positive cells

vehicle Vehicle vehicle vehicle Vehicle vehicle Verteporfin Verteporfin Verteporfin PF-3758309 PF-3758309 PF-3758309 Figure 7. Inhibitors of PAK2 and YAP1 modulate CRPCa growth. (a) The proliferation of androgen-independent C4-2b, PC3 and 22rv1 cells was monitored upon treatment with IPA-3, PF-3758309, Verteporfin or vehicle using RTCA technology. The bars represent normalized cell index at 72 h after drug treatment. (b) PAK2 phosphorylation was analyzed in androgen-independent cells treated with IPA-3 (25 μM) or PF-3758309 (10 μM), whereas YAP1 levels were monitored in cells treated with Verteporfin (10 μM). (c) Balb/c nude mice harboring PC3 tumors grafted subcutaneously were treated with PF-3758309, Verteporfin or vehicle and tumor growth was monitored daily. Growth curves are shown. (d) PC3 xenograft tumors collected at the end of the drug treatment and used for further analyses. (e) In PF-3758309- and vehicle-treated tumors, levels of PAK2-S141P, active caspase-3 and nuclear proliferation antigen Ki67 were monitored by IHC analysis. (f) In Verteprofin- and vehicle-treated tumors, levels of YAP1, active caspase-3 and nuclear proliferation antigen Ki67 were monitored by IHC analysis. Statistical significance differences in tumor size were analyzed by Student’s t-test (*Po0.05; **Po0.01). means for the identification of targets amenable for pharmaceu- including increased sensitivity, precision and speed of mass tical intervention. spectrometers, simplification of sample preparation protocols, The methodology for MS-based quantitative proteomic profiling expansion of methods for analysis of common and rare is advancing at a rapid pace with improvements at all levels, posttranslational modifications and the implementation of

© 2015 Macmillan Publishers Limited Oncogene (2015) 2764 – 2776 Castration-resistant prostate cancer phosphoproteomics N Jiang et al 2774 sophisticated bioinformatics tools for the analysis and interpreta- Cambridge Isotope Laboratories Inc (Andover, MA, USA). Proteins were – tion of MS data.49 51 As it stands today, MS-based proteomics extracted with a dounce homogenizer using a lysis buffer containing stands to challenge the gene centric approach to individualized 50 mM Tris-HCl (pH 7.5), 150 mM NaCl, 1 mm EDTA, 1% NP40, 1 mM Na3VO4, medicine. Its application to the study of CRPCa would unveil a 5mM NaF, 5 mM β-glycerolphosphate, 1 mM PMSF and protease inhibitor wealth of novel information that will hopefully help to design cocktail 1 × (Roche, Basel, Switzerland). Insoluble material was removed by intervention strategies for this lethal condition. centrifugation at 20 000 g for 20 min. One microgram of the protein extracted from individual mice prostates were mixed with equal amount of SILAC-labeled LNCaP-FGC extracts before the samples were digested with MATERIALS AND METHODS trypsin and peptides purified for liquid chromatography–tandem mass spectrometry analysis. Clinical samples Tissue samples were obtained from patients who underwent radical prostatectomy or transurethral resection of the prostate at the Tianjin Phosphopeptide enrichment and liquid chromatography–tandem Medical University Hospital (Tianjin, China) and inspected by a certified mass spectrometry analysis pathologist for Gleason grading. The study was approved by the Ethics Phosphopeptides were enriched using Titansphere chromatography Committee of the Tianjin Medical University, and the Declaration of essentially as described.52 All liquid chromatography–tandem mass Helsinki of Human Rights was strictly adhered to. spectrometry experiments were performed on an EASY-nLC system (Proxeon Biosystems, Odense, Denmark) interfaced with a hybrid Cell culture LTQ-Orbitrap Velos mass spectrometer (Thermo Scientific, Bremen, Germany) through a nanoelectrospray ion source as described.53,54 The PCa cell line LNCaP-FGC was obtained from the American Tissue Culture Collection (Manassas, VA, USA) and the C4-2b cells were obtained from the MD Anderson Cancer Center, The University of Texas (Austin, TX, Data analysis USA). Cells were maintained in RPMI 1640 medium supplemented with Raw MS files from the LTQ-Orbitrap and LTQ-Orbitrap Velos were analyzed 10% fetal bovine serum, 1% penicillin-streptomycin and 1% glutamine. by MaxQuant18 (version 1.0.13.12 and version 1.0.14.11). Tandem mass DHT was obtained from Amersham (Braunschweig, Germany). spectrometry spectra were searched against the decoy International Protein Index-human database version 3.62 containing both forward and siRNA reverse protein sequences, by the Mascot search engine (version 2.2.04; Validated siRNAs were from Ambion/Invitrogen (Carsbad, CA, USA). Matrix Science, Torrance, CA, USA). Parent mass and fragment ions were Transfections were performed using the Neon transfection system (Life searched with maximal initial mass deviation of 7 p.p.m. and 0.5 Th, Technologies, Carlsbad, CA, USA) according to the manufacturer’s respectively. The search included variable modifications of methionine instructions. oxidation and N-terminal acetylation, and fixed modification of cysteine carbamidomethylation. Peptides of minimum six amino acids and maximum of two missed cleavages were allowed for the analysis. For LNCaP orthotopic xenograft models peptide and protein identification, false discovery rate was set to 0.01. For Ten male athymic Balb/c nude mice (HFK Bio-Technology Co. Ltd, Beijing, fi fi 6 identi cation of peptide modi cations, the search included phospho(STY) China) were injected subcutaneouslly with 2 × 10 LNCaP cells suspended variable modification. Full protein lists are supplied as Supplementary in 0.1 ml of Matrigel (BD Biosciences, San Jose, CA, USA). Half of the mice Table S1. Raw MS data files and peptide tables are freely available at were castrated when the tumors reached 10 mm length and the tumors Tranche (https://proteomecommons.org/tranche/; search data set ‘super- were allowed to growth back and later transplanted orthotopically into the SILAC mix’). prostate ventral lobe of intact or castrated Balb/c mice. The later procedure was repeated at least two times before tumors were dissected and snap- frozen for further analysis. The growth rate of the orthotopically grown RNA extraction and gene array analysis tumors was estimated in vivo using subcutaneous xenografts. Tumor RNA was extracted using TRIzol reagent (Invitrogen). Gene expression volume was measured two times weekly. The growth rate was estimated profiling was performed using the Affymetrix platform. The data and based on the daily change of tumor volume and used to calculate experimental details have been submitted to the Gene Expression difference in growth rate. All procedures for animal studies were Omnibus public depository with accession number: GSE46218. conducted in compliance with the policies and regulations of Tianjin Medical University Institutional Animal Care and Use Committee (Tianjin, China). Western blotting Proteins (40 μg) were resolved by sodium dodecyl sulfate–polyacrylamide PC3 xenograft model. PC3 cells in exponential growth stage were gel electrophoresis and probed by immunoblotting using the following trypsinized, washed two times with serum-free RPMI 1640 medium and antibodies: AKT-T308P (sc-16646-R), AKT-S473P (sc-7985-R) and ERK (sc-94) 6 suspended in PBS. The cells (2 × 10 /0.2 ml) were mixed with 0.1 ml of from Santa Cruz Biotechnology, Santa Cruz, CA, USA; AKT (no. 9272), fl Matrigel (BD Biosciences) and injected subcutaneously on the right ank of PRAS40-S183P (no. 5936), PRAS40 (no. 2691), p70S6K-T389P (no. 9206), – male Balb/c nude mice, 5 6 weeks old. The growth of subcutaneous mTORC1-S2448P (no. 5536), mTORC1 (no. 2983), ERK1/2 S217/S221P (no. tumors was monitored and measured with a digital caliper. The tumor 2 4370), PAK2-S141P (no. 2606), PAK2 (no. 2615), ACC-S79P (no. 3661) and volumes were calculated based on the formula L × W /2, where L is the AMPK-T172P (no. 2535) from Cell Signaling Technology, Danvers, MA, USA; length and W is the width of the tumor. When tumors reached 5 mm in β-actin (A1978) from Sigma-Aldrich (St Louis, MO, USA). Proteins were length, five of the tumor harboring mice were treated with 25 mg/kg visualized using horse radish peroxidase-conjugated secondary antibodies PF-3758309 two times daily by oral gavage for 7 days when mice were fi killed and tumor tissue was collected. In parallel experiments, five of the and signals were quanti ed by densitometry analysis using the Quantity tumor harboring mice were treated with 100 mg/kg Verteporfin daily by One software (Bio-Rad, Hercules, CA, USA) intraperitoneal injection for 10 days after which the mice were killed. All procedures for animal studies were conducted in compliance with the Cell proliferation and invasion assay policies and regulations of Tianjin Medical University Institutional Animal C4-2 cells were cultured in RPMI 1640 containing 10% fetal bovine serum Care and Use Committee. PF-3758309 and Verteporfin was purchased from with 1% l-glutamine and 1% penicillin-streptomycin. The cells were seeded Shanghai Alis Chemicals Co. Ltd (Shanghai, China). with eight replicates per sample at a concentration of 10 000 cells in 200 μl per well of an E-Plate 96 (Roche). Cell proliferation was then monitored SILAC labeling and preparation of SILAC mix using an RTCA SP station with readings performed every 15 min for the LNCaP-FGC were SILAC labeled by culturing them in RPMI in which the first 24 h and once per hour subsequently. Cell index was determined natural lysine and arginine were replaced by heavy isotope-labeled amino using the RTCA software (Roche) and normalized to the 24 h time point. acids, L-13C615N4-arginine and l-13C615 N2-lysine supplemented with 10% From this, growth curves were generated and doubling times determined. (vol/vol) dialyzed serum. Labeled amino acids were purchased from Cell invasion assays were performed essentially as described.55

Oncogene (2015) 2764 – 2776 © 2015 Macmillan Publishers Limited Castration-resistant prostate cancer phosphoproteomics N Jiang et al 2775 Colony formation assay 6 Sato K, Qian J, Slezak JM, Lieber MM, Bostwick DG, Bergstralh EJ et al. Single-cell (C4-2b) suspensions containing 1000 cells in 60 μl of medium Clinical significance of alterations of chromosome 8 in high-grade, advanced, were mixed with 60 μl of cold Matrigel and the mixture was placed in nonmetastatic prostate carcinoma. J Natl Cancer Inst 1999; 91: 1574–1580. 24-well plates. The culture plates were placed in a 37 °C incubator for 7 Saramaki O, Visakorpi T. Chromosomal aberrations in prostate cancer. Front Biosci 12 – 10 min to let the mixture solidify, and 500 μl of RPMI 1640 medium 2007; : 3287 3301. supplemented with 10% fetal bovine serum, 1% penicillin-streptomycin 8 Mulholland DJ, Tran LM, Li Y, Cai H, Morim A, Wang S et al. Cell autonomous role and 1% glutamine was added into the well. Colony numbers were counted of PTEN in regulating castration-resistant prostate cancer growth. Cancer Cell 19 – after 7–14 days under an Olympus light microscope (Olympus Corporation, 2011; :792 804. 9 Carver BS, Chapinski C, Wongvipat J, Hieronymus H, Chen Y, Chandarlapaty S et al. Tokyo, Japan), and size differences were also examined. All experiments Reciprocal feedback regulation of PI3K and androgen receptor signaling in were performed at least in triplicate. PTEN-deficient prostate cancer. Cancer Cell 2011; 19: 575–586. 10 Blando J, Portis M, Benavides F, Alexander A, Mills G, Dave B et al. PTEN deficiency Time-lapse microscopy is fully penetrant for prostate adenocarcinoma in C57BL/6 mice via mTOR- The experiments were performed as described previously,54 using a dependent growth. Am J Pathol 2009; 174: 1869–1879. Deltavision Elite microscope equipped with a × 40, 1.35 NA, WD 0.10 11 Horoszewicz JS, Leong SS, Kawinski E, Karr JP, Rosenthal H, Chu TM et al. LNCaP 43 – objective (GE Healthcare, Issaquah, WA, USA). To monitor mitotic effects of model of human prostatic carcinoma. Cancer Res 1983; : 1809 1818. PAK2 depletion, cells were co-transfected with PAK2 siRNAs and a 12 Schayowitz A, Sabnis G, Goloubeva O, Njar VC, Brodie AM. Prolonging hormone construct expressing the chromatin marker CFP-histone H3. Subsequently, sensitivity in prostate cancer xenografts through dual inhibition of AR and mTOR. Br J Cancer 2010; 103: 1001–1007. cells were synchronized by a 2 mM thymidine treatment for 18 h before 13 Nakabayashi M, Werner L, Courtney KD, Buckle G, Oh WK, Bubley GJ et al. Phase II release into the L15 filming medium (Invitrogen) containing 10% fetal calf trial of RAD001 and bicalutamide for castration-resistant prostate cancer. BJU Int serum. Mitotic timing was determined by scoring time from nuclear 2012; 110: 1729–1735. envelope breakdown to the onset of anaphase. All data analysis was 14 Zhang W, Zhu J, Efferson CL, Ware C, Tammam J, Angagaw M et al. Inhibition performed using the Softworx software (GE Healthcare). of tumor growth progression by antiandrogens and mTOR inhibitor in a Pten-deficient mouse model of prostate cancer. Cancer Res 2009; 69: Immunohistochemistry 7466–7472. Tissue sections were dewaxed in xylene and rehydrated in graded alcohol. 15 Olsen JV, Blagoev B, Gnad F, Macek B, Kumar C, Mortensen P et al. Global, in vivo, fi 127 Antigen retrieval was carried out under pressure for 5 min in citrate buffer and site-speci c phosphorylation dynamics in signaling networks. Cell 2006; : – (pH adjusted to 6.0). Endogenous peroxidase was blocked in 0.3% 635 648. 16 Pan C, Olsen JV, Daub H, Mann M. Global effects of kinase inhibitors on signaling hydrogen peroxide for 10 min and blocked using 1.5% horse serum. networks revealed by quantitative phosphoproteomics. Mol Cell Proteom 2009; 8: Incubation with primary antibody was carried out in humidified chamber 2796–2808. overnight at 4 °C (anti-PAK2-S141P, 1:100, from Cell Signaling Technology; 17 Olsen JV, Vermeulen M, Santamaria A, Kumar C, Miller ML, Jensen LJ et al. anti-PAK2 (2247-S), 1:100, from Epitomics (Burlingame, CA, USA); Ki67, Quantitative phosphoproteomics reveals widespread full phosphorylation site 1:100, from Abgent (Surrey, UK); cleaved caspase-3, 1:100, from Abgent occupancy during mitosis. Sci Signal 2010: 3ra3. and anti-yap, 1:150 from Santa Cruz Biotechnology). After applying poly- 18 Thalmann GN, Anezinis PE, Chang SM, Zhau HE, Kim EE, Hopwood VL horse radish peroxidase anti-rabbit immunoglobulin G (30 min), secondary et al. Androgen-independent cancer progression and bone metastasis antibody detection was performed using the Ultraview DAB Detection Kit in the LNCaP model of human prostate cancer. Cancer Res 1994; 54: (Zhongshan Co., Zhongshan, China). All immunostained sections were 2577–2581. evaluated (×200) under Zeiss microscope (Carl Zeiss AG, Oberkochen, 19 Visakorpi T, Kylmala T, Tainio H, Koivula T, Tammela T, Isola J. High cell pro- Germany). At least 10 high-power fields around the malignant glands were liferation activity determined by DNA flow cytometry predicts poor prognosis evaluated and scored. In the case of YAP1 and Ki67, only nuclear staining after relapse in prostate cancer. Eur J Cancer 1994; 30A:129–130. was considered as positive and was scored. In the case of PAK2 and 20 Niu Y, Altuwaijri S, Lai KP, Wu CT, Ricke WA, Messing EM et al. Androgen receptor p-PAK2, both cytoplasm and nuclear staining were considered. For is a tumor suppressor and proliferator in prostate cancer. Proc Natl Acad Sci USA caspase-3, cytoplasm staining was considered as positive and was scored. 2008; 105: 12182–12187. 21 Geiger T, Cox J, Ostasiewicz P, Wisniewski JR, Mann M. Super-SILAC mix for quantitative proteomics of human tumor tissue. Nat Methods 2010; 7: CONFLICT OF INTEREST 383–385. 22 Lopez-Casillas F, Bai DH, Luo XC, Kong IS, Hermodson MA, Kim KH. Structure of The authors declare no conflict of interest. the coding sequence and primary amino acid sequence of acetyl-coenzyme A carboxylase. Proc Natl Acad Sci USA 1988; 85: 5784–5788. ACKNOWLEDGEMENTS 23 Roberts PJ, Der CJ. Targeting the Raf-MEK-ERK mitogen-activated protein kinase cascade for the treatment of cancer. Oncogene 2007; 26: 3291–3310. This work was supported by grants to AF-M from the Novo Nordisk Foundation, 24 Zoncu R, Efeyan A, Sabatini DM. mTOR: from growth signal integration to cancer, Movember and the Danish Council for Independent Research. JN and NY are diabetes and ageing. Nat Rev Mol Cell Biol 2011; 12:21–35. supported by the Science Foundation of Tianjin (No.: 11JCZDJC19700) and 25 Sancak Y, Thoreen CC, Peterson TR, Lindquist RA, Kang SA, Spooner E et al. 09ZCZDSF04300 and the National Natural Science Foundation of China Grant PRAS40 is an insulin-regulated inhibitor of the mTORC1 protein kinase. Mol Cell numbers: 2012CB518304 and 2012DFG32220. 2007; 25:903–915. 26 Harvey KF, Zhang X, Thomas DM. The Hippo pathway and human cancer. Nat Rev Cancer 2013; 13:246–257. REFERENCES 27 Ye DZ, Field J. PAK signaling in cancer. Cell Logist 2012; 2:105–116. 1 Schroder FH, Hugosson J, Roobol MJ, Tammela TL, Ciatto S, Nelen V et al. 28 Manning BD, Cantley LC. AKT/PKB signaling: navigating downstream. Cell 2007; Screening and prostate-cancer mortality in a randomized European study. N Engl 129: 1261–1274. JMed2009; 360: 1320–1328. 29 Ma L, Teruya-Feldstein J, Bonner P, Bernardi R, Franz DN, Witte D et al. Identifi- 2 Crawford ED. Understanding the epidemiology, natural history, and key pathways cation of S664 TSC2 phosphorylation as a marker for extracellular signal-regulated involved in prostate cancer. Urology 2009; 73(Suppl): S4–S10. kinase mediated mTOR activation in tuberous sclerosis and human cancer. Cancer 3 Taylor BS, Schultz N, Hieronymus H, Gopalan A, Xiao Y, Carver BS et al. Integrative Res 2007; 67: 7106–7112. genomic profiling of human prostate cancer. Cancer Cell 2010; 18:11–22. 30 Subramanian A, Tamayo P, Mootha VK, Mukherjee S, Ebert BL, Gillette MA et al. 4 Grasso CS, Wu YM, Robinson DR, Cao X, Dhanasekaran SM, Khan AP et al. The Gene set enrichment analysis: a knowledge-based approach for interpreting mutational landscape of lethal castration-resistant prostate cancer. Nature 2012; genome-wide expression profiles. Proc Natl Acad Sci USA 2005; 102: 487: 239–243. 15545–15550. 5 Visakorpi T, Hyytinen E, Koivisto P, Tanner M, Keinanen R, Palmberg C et al. In vivo 31 Shannon P, Markiel A, Ozier O, Baliga NS, Wang JT, Ramage D et al. Cytoscape: amplification of the androgen receptor gene and progression of human a software environment for integrated models of biomolecular interaction prostate cancer. Nat Genet 1995; 9:401–406. networks. Genome Res 2003; 13: 2498–2504.

© 2015 Macmillan Publishers Limited Oncogene (2015) 2764 – 2776 Castration-resistant prostate cancer phosphoproteomics N Jiang et al 2776 32 Merico D, Isserlin R, Stueker O, Emili A, Bader GD. Enrichment map: a network- 43 D'Antonio JM, Ma C, Monzon FA, Pflug BR. Longitudinal analysis of based method for gene-set enrichment visualization and interpretation. PLoS One androgen deprivation of prostate cancer cells identifies pathways to androgen 2010; 5: e13984. independence. Prostate 2008; 68:698–714. 33 Mukherjee R, McGuinness DH, McCall P, Underwood MA, Seywright M, 44 Zhao B, Li L, Wang L, Wang CY, Yu J, Guan KL. Cell detachment activates the Orange C et al. Upregulation of MAPK pathway is associated with Hippo pathway via cytoskeleton reorganization to induce anoikis. Genes Dev 2012; survival in castrate-resistant prostate cancer. Br J Cancer 2011; 104: 26:54–68. 1920–1928. 45 Tomlins SA, Mehra R, Rhodes DR, Cao X, Wang L, Dhanasekaran SM et al. 34 Drake JM, Graham NA, Lee JK, Stoyanova T, Faltermeier CM, Sud S et al. Metastatic Integrative molecular concept modeling of prostate cancer progression. Nat castration-resistant prostate cancer reveals intrapatient similarity and interpatient Genet 2007; 39:41–51. heterogeneity of therapeutic kinase targets. Proc Natl Acad Sci USA 2013; 110: 46 Das Thakur M, Feng Y, Jagannathan R, Seppa MJ, Skeath JB, Longmore GD. Ajuba E4762–E4769. LIM proteins are negative regulators of the Hippo signaling pathway. Curr Biol 35 Schaeffer EM, Marchionni L, Huang Z, Simons B, Blackman A, Yu W et al. 2010; 20: 657–662. Androgen-induced programs for prostate epithelial growth and invasion arise 47 Mann M, Kulak NA, Nagaraj N, Cox J. The coming age of complete, accurate, and 49 – in embryogenesis and are reactivated in cancer. Oncogene 2008; 27: ubiquitous proteomes. Mol Cell 2013; :583 590. 7180–7191. 48 Lundby A, Secher A, Lage K, Nordsborg NB, Dmytriyev A, Lundby C et al. 36 Nardella C, Chen Z, Salmena L, Carracedo A, Alimonti A, Egia A et al. Aberrant Quantitative maps of protein phosphorylation sites across 14 different rat organs 3 Rheb-mediated mTORC1 activation and Pten haploinsufficiency are cooperative and tissues. Nat Commun 2012; : 876. fi oncogenic events. Genes Dev 2008; 22: 2172–2177. 49 Cox J, Mann M.. MaxQuant enables high peptide identi cation rates, individua- fi 37 Balakumaran BS, Porrello A, Hsu DS, Glover W, Foye A, Leung JY et al. MYC activity lized p.p.b.-range mass accuracies and proteome-wide protein quanti cation. Nat Biotechnol 2008; 26: 1367–1372. mitigates response to rapamycin in prostate cancer through eukaryotic initiation 50 Cox J, Neuhauser N, Michalski A, Scheltema RA, Olsen JV, Mann M. Andromeda: a factor 4E-binding protein 1-mediated inhibition of autophagy. Cancer Res 2009; peptide search engine integrated into the MaxQuant environment. J Proteome Res 69: 7803–7810. 2011; 10: 1794–1805. 38 Lin DL, Tarnowski CP, Zhang J, Dai J, Rohn E, Patel AH et al. Bone metastatic 51 Olsen JV, Mann M et al. Status of large-scale analysis of post-translational mod- LNCaP-derivative C4-2B prostate cancer cell line mineralizes in vitro. Prostate 2001; ifications by mass spectrometry. Mol Cell Proteomics 2013; 12: 3444–3452. 47: 212–221. 52 Hekmat O, Munk S, Fogh L, Yadav R, Francavilla C, Horn H et al. TIMP-1 increases 39 Murray BW, Guo C, Piraino J, Westwick JK, Zhang C, Lamerdin J et al. Small- expression and phosphorylation of proteins associated with drug resistance in molecule p21-activated kinase inhibitor PF-3758309 is a potent inhibitor of breast cancer cells. J Proteome Res 2013; 12: 4136–4151. oncogenic signaling and tumor growth. Proc Natl Acad Sci USA 2010; 107: 53 Olsen JV, Schwartz JC, Griep-Raming J, Nielsen ML, Damoc E, Denisov E et al. 9446–9451. A dual pressure linear ion trap Orbitrap instrument with very high 40 Deacon SW, Beeser A, Fukui JA, Rennefahrt UE, Myers C, Chernoff J et al. An sequencing speed. Mol Cell Proteom 2009; 8:2759–2769. isoform-selective, small-molecule inhibitor targets the autoregulatory mechanism 54 Zhang G, Kelstrup CD, Hu XW, Kaas Hansen MJ, Singleton MR, Olsen JV et al. The 15 – of p21-activated kinase. Chem Biol 2008; :322 331. Ndc80 internal loop is required for recruitment of the Ska complex to establish 41 Liu-Chittenden Y, Huang B, Shim JS, Chen Q, Lee SJ, Anders RA et al. Genetic and end-on microtubule attachment to kinetochores. J Cell Sci 2012; 125(Part 13): pharmacological disruption of the TEAD-YAP complex suppresses the oncogenic 3243–3253. 26 – activity of YAP. Genes Dev 2012; : 1300 1305. 55 Chuan YC, Iglesias-Gato D, Fernandez-Perez L, Cedazo-Minguez A, Pang ST, 42 Laplante M, Sabatini DM.. mTOR signaling in growth control and disease. Cell Norstedt G et al. Ezrin mediates c-Myc actions in prostate cancer cell invasion. 149 – 2012; : 274 293. Oncogene 2009; 29: 1531–1542.

Supplementary Information accompanies this paper on the Oncogene website (http://www.nature.com/onc)