The Proteome of Prostate Cancer Bone Metastasis Reveals Heterogeneity With

Author Manuscript Published OnlineFirst on July 24, 2018; DOI: 10.1158/1078-0432.CCR-18-1229 Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited.

1 The proteome of prostate cancer bone metastasis reveals heterogeneity with

2 prognostic implications

* 5 Diego Iglesias-Gatoa,b , Elin Thysellc, Stefka Tyanovad, Sead Crnalice, Alberto Santosf,

6 Thiago S. Limaa,b, Tamar Geigerg, Jürgen Coxd, Anders Widmarkh, Anders Berghc,

7 Matthias Mannd,f, Amilcar Flores-Moralesa, † , Pernilla Wikströmc, †

8 Affiliations:

9 aDepartment of Drug Design and Pharmacology, Faculty of Health and Medical Sciences,

10 University of Copenhagen, Copenhagen, Denmark

11 bThe Danish Cancer Society. Copenhagen. Denmark.

12 cDepartment of Medical Biosciences, Pathology, Umeå University, Umeå, Sweden

13 dDepartment of Proteomics and Signal Transduction, Max Planck Institute of Biochemistry,

14 Martinsried, Germany

15 eDepartment of Surgical and Perioperative Sciences, Orthopaedics, Umeå University,

16 Umeå, Sweden.

18 fThe Novo Nordisk Foundation Centre for Protein Research, Faculty of Health Sciences,

19 University of Copenhagen, Copenhagen, Denmark.

20 gDepartment of Human Molecular Genetics and Biochemistry, Sackler School of Medicine,

21 Tel Aviv University, Tel Aviv, Israel.

22 hDepartment of Radiation Sciences, Oncology, Umeå University, Umeå, Sweden

23 †Shared senior authorship

Downloaded from clincancerres.aacrjournals.org on October 1, 2021. © 2018 American Association for Cancer Research. Author Manuscript Published OnlineFirst on July 24, 2018; DOI: 10.1158/1078-0432.CCR-18-1229 Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited.

* 24 Correspondence: Diego Iglesias-Gato ([email protected]); telephone:

25 +4535330640, fax: +4535325001.

26 Keywords: Bone metastasis; quantitative proteomics; castration resistant prostate cancer;

27 MCM3; PSA; prognosis metastatic prostate cancer.

29 Running title: The proteome of prostate cancer bone metastasis

32 Translational relevance

33 We performed a global quantification of the proteins expressed in prostate cancer bone

34 metastatic tumors and found wide variation in their proliferative and metabolic features.

35 This variability is related to disease prognosis and should be accounted for when

36 stratifying patients into different treatments.

38 Abstract

39 Background: Bone is the most predominant site of distant metastasis in prostate cancer

40 (PCa) and patients have limited therapeutic options at this stage.

41 Experimental Design: We performed a system-wide quantitative proteomic analysis of

42 bone metastatic prostate tumors from 22 patients operated to relief spinal cord

43 compression. At the time of surgery, most patients had relapsed after androgen-

44 deprivation therapy, while 5 were previously untreated. An extended cohort of prostate

45 cancer bone metastases (n=65) was used for immunohistochemical validation.

46 Results: 5067 proteins on average were identified and quantified per tumor. Compared to

47 primary tumors (n=26), bone metastases were more heterogeneous than and showed

48 increased levels of proteins involved in cell cycle progression, DNA damage response,

49 RNA processing, and fatty acid -oxidation; and reduced levels of proteins related to cell

50 adhesion and carbohydrate metabolism. Within bone metastases, we identified two

51 phenotypic subgroups: BM1, expressing higher levels of AR canonical targets, and

52 mitochondrial and Golgi apparatus resident proteins; and BM2, with increased expression

53 of proliferation and DNA repair related proteins. The two subgroups, validated by the

54 inverse correlation between MCM3 and PSA immunoreactivity, were related to disease

55 prognosis, suggesting that this molecular heterogeneity should be considered when

56 developing personalized therapies.

57 Conclusions: This work is the first system-wide quantitative characterization of the

58 proteome of PCa bone metastases and a valuable resource for understanding the etiology

59 of PCa progression.

61 Introduction

63 Mortality in prostate cancer (PCa) is strongly associated with development of bone

64 metastatic disease (1,2). PCa metastases are generally treated with androgen ablation(3).

65 This therapy is initially effective, but most patients relapse to a castration resistant stage.

66 Despite recent treatment developments for patients with castration resistant bone

67 metastases (mCRPC), their prognosis remains poor with a median survival below 2 years

68 (4). This is in contrast to patients harboring prostate confined tumors, who have a 10-year

69 cancer specific survival rate above 90%, even for patients managed by observation

70 protocols (5,6). Why metastatic tumors have a more aggressive behavior than localized

71 tumors is not fully understood. Recent genetic studies have identified only a modest

72 increase in the number and frequency of genetic aberrations in lethal, therapy-resistant

73 metastatic cancer as compared to localized tumors (e.g. TP53, PTEN, FOXA1, PI3KCA)

74 with the exception of androgen receptor (AR) amplifications, rearrangements, and

75 mutations, seen almost exclusively in CRPC (7,8). One possible explanation for these

76 findings is that many of the low frequency mutations occurring in individual tumors may

77 confer growth advantages through the regulation of a set of core signaling pathways

78 required for metastatic growth. For example, mutations in genes such as FOXA1 or MLL2,

79 with apparently unrelated functions may influence tumor growth through a common

80 signaling mechanism: the regulation of AR activity (9). Many of the alterations in protein

81 signaling associated to metastatic progression have remained undiscovered, as only few

82 studies have actually examined prostate bone metastases in depth at the proteome level

83 (10).

85 Transcriptomics or genetic analysis provides a partial view of the signaling pathways

86 driving tumor growth because of their limited capacity to predict concentration changes at

87 the protein level or inform about status of protein driven mechanisms (11,12). Quantitative

88 proteomic analysis of tumor samples may have a large potential to unveil novel signaling

89 mechanisms of clinical relevance. We have recently analyzed the proteome of primary

90 prostate tumors and demonstrated the upregulation of various proteins involved in

91 metabolic processes in malignant compared to non-malignant tissue (13). Remarkably, we

92 did not observe a general increase of proteins involved in cell cycle progression or DNA

93 synthesis, perhaps reflecting the low proliferative capacity of most localized prostate

94 tumors.

96 The objective of this first-in-class study was to characterize PCa bone metastases in depth

97 at the protein level and, by integrating these results with the previously described

98 proteomic profiles of localized prostate tumors (13), to describe the proteome evolution

99 during PCa progression. Furthermore, we aimed to identify putative targets for novel

100 therapies of metastatic CRPC and also to explore whether subgroups of PC bone

101 metastasis exist that could potentially guide therapy decisions. We used system-wide

102 quantitative mass spectrometry-based proteomic analysis to describe the proteome of PCa

103 bone metastases and we found a high degree of heterogeneity among the analyzed

104 samples with at least two phenotypically distinct subgroups of bone metastases. We

105 propose that these subgroups should be accounted for when designing novel therapeutic

106 alternatives.

107

108 Materials and Methods

109

110 Cohort description

111

112 Bone metastasis samples were obtained from men with PCa, who underwent surgery for

113 metastatic spinal cord compression at Umeå University Hospital between 2003 and 2014.

114 Metastasis biopsies were fixed in 4% buffered formalin and decalcified in 20% formic acid

115 for 1-3 days, according to clinical routines, before being embedded in paraffin. Clinical

116 characteristics of the explorative, proteomic cohort (n=22) and the extended cohort used

117 for immunohistochemical validation (n=65) are described in Table 1. For some cases of

118 the proteomic cohort (n=12-14), there were also available diagnostic, primary tumor

119 biopsies for immunohistochemical analysis. In summary, most patients were diagnosed

120 with tumors considered as high-risk due to extensive growth in the biopsies, poor

121 differentiation (high GS) and with locally advanced or metastatic disease. In patients where

122 PCa was not diagnosed until it caused neurological symptoms (patients without ADT at

123 metastasis surgery) the primary tumor was not biopsied. Most patients were directly

124 treated with ADT and only 2 patients were treated with curative intent (Table 1). At relapse

125 to castration, patients received second line treatments as indicated (Table 1). The study

126 was approved by the local ethic review board of Umeå University (Dnr 03-185, Dnr 04-

127 026M, decision 2007-08-24, Dnr 2013-372-32M) and conducted following the principles of

128 the Declaration of Helsinki. All participants gave written or verbal consent. Due to the

129 acute situation when bone metastasis surgery is performed in order to relief spine

130 symptoms and paresis, logistics do not always allow written consent and the local ethic

131 review board therefore specifically approved also verbal consent. Verbal consent is

132 documented by the physician in the patient journal.

133

134

135 Proteomic analysis

136 The effects of formic acid-based decalcification in the proteomic analysis were evaluated

137 in a separate set of samples. No obvious quantitative or qualitative protein alterations were

138 observed (Suppl Fig. S1A). Protein extracts were obtained from approximately 20

139 microtome sections of 10 µm thick 25-50 mm2 tumor enriched areas. Mass spectrometry-

140 based proteomic analysis of the bone metastasis samples was performed as previously

141 described using Super-SILAC spike-in standards for accurate quantification ((13) and

142 extended M&M). The proteomic pipeline is summarized in Suppl. Fig.S1B. Raw data files

143 were uploaded to ProteomeXchange. Accession references for published primary tumors

144 and non-malignant adjacent tissue are PXD004159; PXD004132; PXD003636;

145 PXD003615; PXD003515; PXD003452; PXD003430 (13). Accession reference for

146 metastatic samples is PXD009868.

147

148 Bioinformatic tools and statistics

149 Unsupervised hierarchical clustering of metastasis samples was performed based on

150 Pearson correlation coefficients as distance metrics (Figure 1E). GO classification and

151 enrichment analysis was performed using DAVID (14) and results were graphically

152 represented with Cytoscape (15) (Figures 2 and 3).

153

154 Metastatic subgroups were defined by the Pearson’s correlation coefficients of the SILAC

155 ratios of each protein across the cohort with the average SILAC ratio of the

156 minichromosome maintenance complex components (MCM2-7) of each individual tumor

157 (Figure 3A). Only proteins showing 2/3 of valid SILAC ratios were used in the analysis.

158 SILAC ratios of proteins with significant correlation (p < 0.05) to the MCM complex were

159 used for hierarchical clustering. DAVID (14) was used to analyze the GO enrichment of the

160 proteins with significant, positive or negative, correlation to the MCM. Protein-protein

161 interactions of components of significantly enriched processes were plotted using

162 Cytoscape (15) according to their String score (16) (Figure 3B). Comparison between

163 gene expression and protein profiles (Figures 4A,B) was carried out using the GSEA

164 software (17).

165

166 To estimate the relative amount of subcellular components per cell (Figure 4D) the sum of

167 the label free quantification intensity (LFQ, (18)) of all proteins described as components of

168 the organelle of interest were divided by the sum of all the nucleosomal proteins quantified

169 in a given tumor sample. Only proteins quantified in all the samples were used.

170 171 Immunohistochemistry

172 Tissue sections were deparaffinized in xylene and rehydrated through graded ethanol.

173 Immunohistochemical staining was performed using MCM3 (HPA004789, Atlas

174 Antibodies, diluted 1:100) and PSA (A0562, DAKO, diluted 1:1000) antibodies, the

175 ultraView Universal DAB Detection kit (760-500) and the automatic VENTANA Benchmark

176 Ultra system, according to manufacturer´s description (Roche Diagnostics). PSA was

177 stained without tissue pre-treatment and MCM3 by using the CCL2 standard method

178 (VENTANA, Roche Diagnostics). Immunostained sections were mounted and imaged and

179 evaluated by using the Pannoramic 250 FLASH scanner and the Pannoramic viewer

180 1.15.2 software (3D HISTECH). The fraction of MCM3 immuno-positive tumor nuclei was

181 evaluated by counting on average 188 cells (117-437) cells per sample. The PSA staining

182 were quantified by scoring the intensity (0=no staining, 1=weak, 2= moderate, 3 = intense

183 staining) and the percentage of stained tumor cells (1=1-25%, 2=26-50%, 3= 51-75%, 4=

184 76-100%). A combined PSA immune-reactive score, ranging from 0-12, was calculated by

185 multiplying intensity with distribution.

186

187 Results

188

189 Proteomic profiling of PCa bone metastasis samples provides additional information to

190 transcriptomic profiling

191

192 Formalin-fixed paraffin embedded PCa bone metastasis samples, resected from 22

193 patients (Table 1), were decalcified and analyzed by quantitative LC-MS/MS using a

194 spiked-in standard derived from isotopically labeled PCa cell lines, as previously described

195 ((13); suppl. Fig.S1B and extended M&M). On average, 5067 proteins per sample were

196 identified showing valid ratios for a total of 7663 proteins (Figure 1A, suppl. Table S1).

197

198 We previously obtained global transcriptomic profiles from 20 of these bone metastases

199 (included in GEO Datasets GSE29650 and GSE101607 (19,20)). Similar to observations

200 in other cancer types (12,21), and recently reported for prostate cancer (22), we found a

201 positive but limited correlation between variations in mRNA and changes in the

202 corresponding protein expression across the samples analyzed. The mean Spearman’s

203 was 0.28 (-0.69 to 0.92), with 31% of those correlations being statistically significant (p <

204 0.05) (Figure 1B). Particularly, genes involved in mRNA translation, oxidative

205 phosphorylation and components of the ribosomes and proteasome exhibited low degree

206 of mRNA-protein correlation, with some even showing significantly negative correlations

207 (Figure 1C). These results suggest that in PCa, protein expression measurements are

208 required to fully investigate the relevance of many pathways for which analysis of

209 variations in RNA steady-state levels would lead to inaccurate conclusions. The

210 discrepancies between RNA and protein expression should be taken into consideration at

211 the time of experimental planning and data interpretation.

212

213 PCa bone metastases are more heterogeneous than tumors restricted to the prostate

214 gland

215

216 The use of the same super-SILAC spike-in standard for the analysis of bone metastases

217 as for localized tumor tissue obtained from prostatectomy specimens (of independent

218 patients) (13) allowed us to compare these protein profiles, in order to obtain a global

219 picture of PCa progression at the proteome level. Protein expression profiles of localized

220 tumors and bone metastases were significantly correlated (average Pearson’s correlation

221 of : 0.61, p<0.001; Figure 1D). However, the average correlation for protein profiles within

222 the metastatic group (: 0.64  0.077) was lower than that of the localized tumors (: 0.8 

223 0.037) and the benign prostate tissues (: 0.83  0.03), (Figure 1E), indicating a higher

224 degree of expressional heterogeneity among bone metastases than prostatectomy

225 samples. Accordingly, unsupervised hierarchical clustering based on Pearson’s correlation

226 of all the samples retrieved two main clusters; one containing all but one metastatic tumor

227 and the other, grouping the localized tumors and benign samples (Figure 1E). This

228 indicates major differences between localized and metastasized PCa at the proteome

229 level. Strikingly, protein profiles from treatment-naïve bone metastasis tumors (Samples

230 PC31-35) and one metastasis collected shortly after the beginning of the castration

231 protocol (PC36), clustered together with CRPC metastases instead of with primary tumors.

232

233 Major cellular processes altered in the transition from localized PCa to bone metastasis

234

235 In order to identify biological pathways involved in PCa progression, we performed gene

236 ontology (GO) enrichment analysis of the proteins differentially expressed in bone

237 metastases compared to primary tumors (Figure 2). Proteins functionally related to cell

238 cycle progression, DNA replication and DNA damage repair were significantly upregulated

239 in the metastatic setting (Figure 2A). Importantly, these alterations were not observed in

240 localized prostate tumors when compared to neighboring control tissue (13), suggesting

241 that increased proliferation rates are acquired at a later stage during PCa progression. On

242 the other hand, some of the changes observed already in localized tumors get

243 exacerbated in the metastases, as is the case of reduced expression of proteins involved

244 in cell adhesion and carbohydrate metabolism (Figure 2B) as well as increased expression

245 of proteins involved in RNA biogenesis and transport, lipid transport and fatty acid

246 oxidation (Figure 2A). Strikingly, these general features of bone metastases were

247 observed regardless of whether the patients were previously subjected to ADT or

248 presented as treatment naïve cases. This suggests that transition from localized to

249 metastatic spreading requires the acquisition of aggressive features that are retained or

250 further developed when relapsing after castration. It also supports the notion that

251 castration resistant metastases rely on similar mechanisms (including AR regulated

252 pathways) for growth as advanced, castration naïve tumors.

253

254 Proteomic subtyping of PCa bone metastases based on an inverse relation between

255 proliferation and metabolic features

256

257 High proliferation rate is a general feature of metastatic tumors, and in this study,

258 increased expression of cell cycle associated proteins was also highlighted when

259 comparing the proteomes of metastatic and localized PCa. However, given the variability

260 in protein expression among metastases we set out to analyze whether subgroups of bone

261 metastases existed based on differential expression of cell cycle regulating proteins. As a

262 proxy for proliferation rate, we analyzed expression levels of the MCM complex, a DNA

263 helicase with an essential role in licensing DNA replication origin in the early S-phase of

264 the cell cycle (23). Tumors with higher expression of MCM proteins are predicted to

265 contain more cells in an actively replicative state. The expression levels of the 6 core

266 proteins of the complex (MCM2-7, (23)) showed a high degree of intra-sample correlation

267 across the tumors tested (> 0.895; Suppl. Fig.S2). Moreover, the expression of the MCM

268 complex remained stable across the localized tumors but varied widely across the bone

269 metastases (Figure 3A and Suppl. Fig.S2), suggesting variable proliferation rates among

270 metastatic tumors. We then correlated the average expression of the MCM proteins with

271 all the proteins identified that showed at least 2/3 of valid expression ratios across all

272 metastatic samples analyzed. Proteins showing expression levels that significantly (p

273 <0.05) correlated to those of the MCM complex were grouped as positive or negatively

274 correlators, respectively (Figure 3A). As expected, the biological functions among the 240

275 proteins that positively correlated with MCM expression (suppl. Table S2) were mainly

276 related to cell cycle progression, DNA metabolism (i.e. DNA replication and DNA damage

277 repair) and RNA metabolism (including RNA splicing) (Figure 3B), but surprisingly did not

278 contain the widely used proliferation marker, Ki67 (= 0.37; p >0.05). On the other hand,

279 the 260 proteins that negatively correlated to the MCM complex were enriched for proteins

280 involved in metabolic functions such as mitochondrial respiration (e.g. COX and NDUF

281 proteins), fatty acid -oxidation and amino acid catabolism (e.g. ACAD9, ACADSB, IVD,

282 etc) and Golgi-ER organization (e.g. COGs, COPA, etc) (Figure 3B, suppl. Table S2).

283 Interestingly, this group was also enriched for proteins positively regulated via the

284 canonical androgen receptor pathway (e.g. PSA, STEAP2, AGR2, GDF15, FASN, PDIA5,

285 CORO2a, NT5DC3) (20,24,25). Our results suggest that metastatic tumors with slower

286 proliferation rates preferentially rely on aerobic metabolism for energy production, while

287 the enlarged Golgi-ER compartment of these tumors could reflect a secretory function as a

288 sign of somewhat retained prostate epithelial differentiation. Interestingly, these metabolic

289 and secretory features are characteristics of tumors localized to the prostate, which are in

290 general also characterized by a low proliferation rate (13).

291 To further support these observations, we compared these two sets of proteins (positively

292 and negatively correlated to the MCM proteins, respectively) with independent system-

293 wide transcriptomic studies of PCa using Gene Set Enrichment Analysis (GSEA, (17)). As

294 anticipated, the mRNA levels corresponding to proteins positively correlated with the MCM

295 complex were typically elevated in metastatic compared to localized prostate tumors

296 (suppl. Fig.S3A, (9,26) and this study). On the other hand, negatively correlated proteins

297 showed overall reduced expression in metastases relative to localized PCa (suppl.

298 Fig.S3B; (9,26) and this study), while being genes with increased expression at early

299 stages (i.e. in localize tumors as compared to benign prostate tissue; suppl. Fig.S3C,

300 (9,13,26)). This suggests that metastases with predominant expression of proteins

301 negatively correlated with the MCM complex retain features of localized tumors with low

302 proliferation rates and would potentially show a less aggressive phenotype. We next used

303 hierarchical clustering to analyze whether the protein signature defined through correlation

304 to the MCM complex could serve to identify subgroups of metastatic tumors. Indeed, using

305 this protein signature, tumors could be grouped in two main clusters (BM1 and BM2,

306 Figure 4A) which contained all but three of the tumors (sample PC32 was excluded from

307 the analysis due to large percentage of missing values). A direct comparative analysis of

308 the tumors included on each of the clusters showed indeed that BM1 metastases express

309 significantly higher levels of mitochondrial and ER/Golgi proteins typical of localized

310 tumors, while proteins with increased expression in BM2 tumors are enriched in

311 proliferation related processes (DNA replication, mitosis, mRNA processing) (Figure 4B).

312 Taken together, these results suggest the existence of at least two subtypes of bone

313 metastases; one (BM1) including tumors that retain features of differentiated prostate

314 epithelium, with potentially less aggressive phenotype; and another (BM2), grouping highly

315 proliferative and less differentiated metastases.

316

317 We further validated the presence of these phenotypic subgroups by immunohistochemical

318 analysis of an extended cohort of 65 PCa bone metastasis. Immunoreactivity of prostate

319 specific antigen (PSA), as secretion marker related to prostate epithelial cell differentiation

320 and representative of group BM1; and nuclear MCM3, as marker of proliferation and

321 representative of the BM2 group, was assessed (suppl. Fig.S4A). In good agreement with

322 the proteomic data (PSA/MCM3 correlation = -0.77; p < 0.0001; n = 22), the fraction of

323 MCM3 immuno-positive tumor cells was inversely correlated to the PSA immunoreactivity

324 ( = -0.58; p < 0.0001; n = 65) (Figure 4C). Metastases with MCM3 scores at or below

325 median (≤ 21) and PSA scores above the median (> 6) could be considered as BM1-like,

326 while metastases with MCM3 scores above 21, and PSA scores at or below 6 are more

327 likely BM2-like. Tumors clustered as part of one of these groups comprised 2/3 of the

328 cohort analyzed, while the rest showed a heterogeneous phenotype in relation to these

329 two markers as observed in suppl. Fig.S4A.

330

331 Increased expression of proliferation markers in primary prostate tumors is correlated with

332 poorer prognosis (27) and commercial test are available to evaluate these features (e.g.

333 Prolaris®). In order to investigate whether the observed proliferation features of BM2

334 tumors were present in the primary tumors or were instead acquired at a later stage, we

335 analyzed MCM3 and PSA expression in diagnostic biopsies corresponding to 12 and 14,

336 respectively, of our proteomic analyzed metastases. On the one hand, MCM3 levels were

337 similar in primary biopsies independently of whether the consequent metastasis showed

338 features of the BM1 or BM2 phenotypic subgroups (Fig. 4D), but importantly they were

339 increased in bone metastases compared to paired diagnostic biopsies (P = 0.010, n =12)

340 and the increase seemed more pronounced in BM2 than BM1 metastases (Fig. 4D). On

341 the other hand, levels of PSA on biopsies from tumors that developed into BM2

342 metastases were lower than those resulting in BM1 metastases (P = 0.023, n =12), and

343 PSA levels were further decreased in metastases compared to diagnostic biopsies (P =

344 0.030, n =14) (Fig. 4E).

345 Taken together, the IHC analysis confirmed the proteomic data showing that PCa bone

346 metastatic progression is associated with increased proliferation, and suggests that early

347 loss of AR canonical function drives a separate path of metastatic development upon

348 castration treatment.

349

350 Phenotypic subtypes of PCa bone metastases correlate with disease outcome and could

351 provide bases for therapeutic stratification of patients

352

353 According to the protein profiles of tumors obtained from castration naïve and castration

354 resistant patients, the bone metastasis phenotypes seemed to be independent from

355 previous treatment (Figure 4A) (suppl. Fig.S3C). Instead, the large variation in MCM3

356 positive tumor cells (3 to 94 %) and PSA immune-score (0 to 12) appeared to be related to

357 patient prognosis. Patients with BM2-like metastases (as defined above, n=22) had shorter

358 survival times after first ADT compared to the BM1-like group (n=21) (median survival 30

359 vs. 68 months, Log Rank = 4.6, p=0.032, Figure 5A). Moreover, when the survival of BM2-

360 like patients was compared with the rest of the cohort (n=43), the survival time after first

361 ADT was significantly reduced (median survival 30 vs. 46 months, Log Rank = 5.7,

362 p=0.017, Figure 5B). Importantly, the combined immuno-variable defining BM2-like

363 metastases (MCM3 fraction above median and PSA IR below median) added prognostic

364 value to either MCM3 or PSA immunoreactivity analysed as single variables (Table S3),

365 also when adjusted for age and Gleason score at diagnosis. In contrast, neither of the

366 clinical variables in Table 1 were significantly associated with survival after first ADT

367 (Supp. Table S3).

368

369 In addition to a possible difference in prognosis, the distinct phenotypic characteristics of

370 the two metastasis subtypes could be of importance for patient stratification into different

371 treatments. Despite the lack of cell line models accurately resembling all the phenotypic

372 features of each subgroup, we tested, as proof of principle, whether cell sensitivity towards

373 cellular organelle poisoning was related to the amount of organelle per cell. Thus, C4-2b

374 cells with proportionally more mitochondrial content were more sensitive to drugs targeting

375 mitochondrial function (metformin and doxycycline) (suppl. Fig. 4B). On the other hand,

376 PC3 cells, with enlarged Golgi compartment were more sensitive to brefeldin A treatment,

377 a drug that interferes with ER to Golgi transport (suppl. Fig.4B).

378

379 Discussion

380 In this study, we have performed an extensive characterization of the PCa bone

381 metastasis proteome, as the bone is the preferred metastasis niche for advanced PCa.

382 Surgical removal of bone metastases is seldom performed and, therefore, most studies of

383 PCa metastases so far have been based on soft tissue metastases obtained from rapid

384 autopsy programs and include patients with exhausted treatment options (28). In contrast,

385 our study cohort was mainly composed of chemotherapy naïve bone metastases and even

386 included cases obtained prior to any therapy. The analysis of those valuable metastasis

387 samples, in combination with data from our previous study of the primary PCa proteome

388 (13), allowed us to describe PCa evolution at the proteome level for the first time. This

389 revealed the existence of two phenotypically distinct subtypes of PCa bone metastasis,

390 which are related to disease prognosis and may have implications for future personalized

391 treatments. Limitations of this study include the limited number of samples analyzed and

392 the unavailability of an independent replication cohort. A separate analysis of the epithelial

393 and stromal components could provide additional information regarding the relative

394 influence of tumor microenvironment on proteome variation during prostate cancer

395 progression.

396

397 The molecular transition from localized to metastatic PCa is not yet well understood. Here

398 we found that at the proteome level, castration naïve bone metastases show a proteome

399 pattern distinct from untreated localized PCa but similar to castration-resistant bone

400 metastases. This suggests that metastases growing in the bone and thus influenced by

401 this microenvironment (29), have acquired most features of aggressive tumors, such as

402 accelerated cell cycle progression, already prior to therapy selection (Figure 2; (30)).

403 Whether acquisition of this aggressive phenotype is a requirement for metastatic

404 spreading or it is achieved upon reaching the metastatic niche remains to be elucidated.

405 The increased activity in signaling pathways related to cell cycle and DNA damage

406 response in soft tissue metastases previously reported (10), together with the observed

407 increased expression of MCM3 protein in bone metastasis compared to diagnostic

408 biopsies, suggest that these are features of metastatic prostate cells. Our data also

409 indicates that castration-resistant PCa bone metastases maintain their aggressive growth

410 behavior under the environmental circumstances of low androgen availability, by often but

411 not always restoring canonical androgen receptor signaling.

412

413 We observed that PCa bone metastases are heterogeneous in their protein repertoire and

414 propose the existence of at least 2 phenotypically distinct subgroups (Figure 5). Group

415 BM1, characterized by elevated expression of many canonical AR targets and by higher

416 mitochondrial and ER/Golgi organelle content as indicatives of aerobic metabolism and

417 secretory functions and BM2, defined by increased expression of proteins required for cell

418 cycle progression and likely driven by glycolytic metabolism. Inverse correlation between

419 AR canonical activity and tumor aggressiveness has been previously suggested for

420 localized PCa (28,31), but with limited characterization of the metabolic tumor features.

421 The features of BM2 type of metastases are not commonly seen within localized PCa, but

422 importantly, when observed, they are associated with poor patient prognosis (27). On the

423 other hand, features of BM1 tumors are often recognized in early PCa (13), suggesting this

424 group includes tumors with higher degree of epithelial cell differentiation. Whether more

425 differentiated BM1 metastases retain their molecular features throughout the treatment

426 regimen until patient death or if they change as an adaptive mechanism to later therapies,

427 deserves further investigation. The use of SILAC spike-in standard from prostate cells

428 ensures quantification accuracy but limits the identification of proteins specific of other cell

429 types within the tumor microenvironment. Therefore, the existence of additional subtypes

430 of PCa bone metastasis, for instance, those exhibiting high degree of immune cells

431 infiltration, cannot be ruled out (10). Whether tumors with high immune cell infiltration

432 constitute and independent subgroup or an additional layer of complexity for the

433 stratification of bone metastasis, deserves further investigation. However, taking

434 advantage of some of the bone metastasis being analyzed in both this and the Ylitalo et at

435 study (20) , tumors with higher degree of immunoreactivity would be of the BM2 type while,

436 as expected, AR-driven tumors are classified as BM1.

437

438 The two metastatic phenotypes identified in our study show similarities with molecular

439 subtypes recently suggested for localized prostate tumors based on transcriptomic

440 analysis (31,32). These studies proposed the existence of 3 PCa subtypes: 2 luminal-like

441 (PCS1, PCS2 and luminal A, luminal B, respectively) and one basal-like (PCS3 and Basal,

442 respectively). Our BM2 metastases show similarities with the PCS1 and luminal B tumors,

443 being highly proliferative and associated with poor patient prognosis, while our BM1

444 tumors may have similarities with the PCS2 and luminal A tumors, showing increased

445 canonical AR activity. The PCS3 and basal tumors were not obviously captured by our

446 proteomic approach, probably due to the number of samples analyzed, but may have been

447 reflected as the MCM3 low/PSA low cases by the IHC analysis. The immunohistochemical

448 analysis also reflected large inter- and intra-tumoral heterogeneity, with mosaic

449 metastases (areas positive for PSA and negative for MCM3 and vice versa, next to each

450 other) represented in both subgroups (suppl. Fig.S4A). Tumor heterogeneity may reflect

451 the range of response to secondary treatment, including enzalutamide and abiraterone,

452 observed in the clinic and supports the development of personalized treatments according

453 to the phenotypic features of the tumors (33,34).

454

455 Phenotypic features of PCa bone metastatic tumors seem related to disease prognosis.

456 Patients with high metastasis PSA immunoreactivity (assessed as marker for prostate

457 differentiation) in combination with low proliferation rate appeared to have better

458 prognosis, measured as cancer-specific survival from first ADT treatment. Since the

459 subtypes do not appear to be dependent on first castration treatment, this way of

460 measuring survival could reflect the original response to ADT, predicted to be more

461 efficient against BM1, a subgroup clearly androgen receptor driven. Interestingly, we were

462 able to detect the expression of neuroendocrine markers such as CHGA by mass

463 spectrometry in a fraction of BM2 tumors (Figure 4A), in line with the de-differentiation

464 process and poorer prognosis observed. Of those, only mPC31 and mPC51 could be

465 define as canonical neuroendocrine tumors expressing low AR/PSA levels and high levels

466 of CHGA addressed by IHC (Figure 4A).

467

468 In addition to its relation to prognosis, the existence of phenotypic subgroups might be use

469 for patient stratification for suitable therapeutic options. Thus, BM1 tumors that express

470 higher levels of proteins involved in mitochondrial respiration and secretory functions and

471 maintain canonical AR activity might be sensitive to drugs targeting these metabolic

472 functions (e.g. metformin, doxycycline or Brefeldin A) presumably in combination with AR

473 targeting drugs. This possibility is supported by our in vitro data on PCa cells and by

474 previous studies showing that the combination of Enzalutamide and Metformin treatment in

475 C4-2b cell xenografts reduces tumor volume more effectively than any of the treatments

476 given alone (35). Currently there are several clinical trials testing the effectiveness of this

477 combined therapy in patients with CRPC (NCT01677897; NCT01243385, NCT01796028) but, to

478 the best of our knowledge, no patient selection is being applied based on metabolic or

479 proteomic profiles. Our results suggest that patients with maintained AR activity linked to

480 mitochondrial respiration would benefit the most from this treatment and therefore patient

481 stratification should be considered in future trials.

482

483 In an effort to identify drugs that can be repurposed for the treatment of prostate cancer

484 alone or in combination with current treatments, we summarized all approved drugs that

485 target proteins found upregulated in PCa bone metastases compared to prostate localized

486 tumors (suppl. Fig.S5) and those particularly targeting each of the subgroups (suppl.

487 Fig.S6). We hope that this overview of current drug targets in the context of prostate

488 cancer progression may serve to motivate clinical studies for drug repurposing or the

489 utilization of novel drug combinations. Indeed, targeting of proteins like mTOR, with

490 functions both in cellular proliferation (RNA metabolism and translation) and energy

491 generation (fatty acid metabolism) is currently being tested in clinical trials in combination

492 with AR targeting therapies (www.clinicaltrial.org; clinical trial IDs: NCT02407054 and

493 NCT02091531). Note that some of these drugs have already proven inefficient for the

494 treatment of metastatic PCa (e.g.(36-38)) and it would be interesting to determine whether

495 these failures are due to lack of efficacy or due to inappropriate patient selection.

496

497 In summary, the analysis of the PCa proteome revealed that while localized to the

498 prostate, most tumors are slowly proliferating and exhibit elevated mitochondrial activity

499 and intracellular vesicle transport, associated with secretion and somewhat retained

500 epithelial cell differentiation. Metastatic tumors are more heterogeneous showing various

501 levels of AR activity, metabolic profiles, and proliferation rates. We propose the existence

502 of at least two phenotypically different subtypes of PCa bone metastases, where patients

503 with highly proliferative, low differentiated PCa bone metastases (BM2) have worse

504 prognosis and potentially would benefit from receiving early chemotherapy-based

505 treatments targeting cell cycle progression (Figure 5C). Patients with more differentiated

506 metastases (BM1), seem to show better response to ADT and probably would benefit from

507 second line AR targeting therapies and treatments directed towards metabolic pathways

508 (Figure 5C). The proteome analysis of larger cohorts of PCa metastases is warranted in

509 order to validate and refine group features. Whether metastasis heterogeneity can be

510 predicted from analysis of the primary tumor also deserves further investigation.

511

512

513 References

514

515 1. Coleman RE. Clinical features of metastatic bone disease and risk of skeletal morbidity. Clin 516 Cancer Res 2006;12(20 Pt 2):6243s-9s doi 10.1158/1078-0432.CCR-06-0931.

517 2. Gandaglia G, Karakiewicz PI, Briganti A, Passoni NM, Schiffmann J, Trudeau V, et al. Impact 518 of the Site of Metastases on Survival in Patients with Metastatic Prostate Cancer. Eur Urol 519 2015;68(2):325-34 doi 10.1016/j.eururo.2014.07.020. 520 3. Parker C, Gillessen S, Heidenreich A, Horwich A, Committee EG. Cancer of the prostate: 521 ESMO Clinical Practice Guidelines for diagnosis, treatment and follow-up. Ann Oncol 522 2015;26 Suppl 5:v69-77 doi 10.1093/annonc/mdv222. 523 4. Nuhn P, De Bono JS, Fizazi K, Freedland SJ, Grilli M, Kantoff PW, et al. Update on Systemic 524 Prostate Cancer Therapies: Management of Metastatic Castration-resistant Prostate 525 Cancer in the Era of Precision Oncology. Eur Urol 2018 doi 10.1016/j.eururo.2018.03.028. 526 5. Barbosa PV, Thomas IC, Srinivas S, Buyyounouski MK, Chung BI, Chertow GM, et al. Overall 527 Survival in Patients with Localized Prostate Cancer in the US Veterans Health 528 Administration: Is PIVOT Generalizable? Eur Urol 2016;70(2):227-30 doi 529 10.1016/j.eururo.2016.02.037. 530 6. Wilt TJ, Brawer MK, Jones KM, Barry MJ, Aronson WJ, Fox S, et al. Radical prostatectomy 531 versus observation for localized prostate cancer. N Engl J Med 2012;367(3):203-13 doi 532 10.1056/NEJMoa1113162. 533 7. Cancer Genome Atlas Research N. The Molecular Taxonomy of Primary Prostate Cancer. 534 Cell 2015;163(4):1011-25 doi 10.1016/j.cell.2015.10.025. 535 8. Robinson D, Van Allen EM, Wu YM, Schultz N, Lonigro RJ, Mosquera JM, et al. Integrative 536 Clinical Genomics of Advanced Prostate Cancer. Cell 2015;162(2):454 doi 537 10.1016/j.cell.2015.06.053. 538 9. Grasso CS, Wu YM, Robinson DR, Cao X, Dhanasekaran SM, Khan AP, et al. The mutational 539 landscape of lethal castration-resistant prostate cancer. Nature 2012;487(7406):239-43 doi 540 10.1038/nature11125. 541 10. Drake JM, Paull EO, Graham NA, Lee JK, Smith BA, Titz B, et al. Phosphoproteome 542 Integration Reveals Patient-Specific Networks in Prostate Cancer. Cell 2016;166(4):1041-54 543 doi 10.1016/j.cell.2016.07.007. 544 11. Schwanhausser B, Busse D, Li N, Dittmar G, Schuchhardt J, Wolf J, et al. Global 545 quantification of mammalian gene expression control. Nature 2011;473(7347):337-42 doi 546 10.1038/nature10098. 547 12. Zhang B, Wang J, Wang X, Zhu J, Liu Q, Shi Z, et al. Proteogenomic characterization of 548 human colon and rectal cancer. Nature 2014;513(7518):382-7 doi 10.1038/nature13438. 549 13. Iglesias-Gato D, Wikstrom P, Tyanova S, Lavallee C, Thysell E, Carlsson J, et al. The 550 Proteome of Primary Prostate Cancer. Eur Urol 2016;69(5):942-52 doi 551 10.1016/j.eururo.2015.10.053. 552 14. Huang da W, Sherman BT, Stephens R, Baseler MW, Lane HC, Lempicki RA. DAVID gene ID 553 conversion tool. Bioinformation 2008;2(10):428-30. 554 15. Cline MS, Smoot M, Cerami E, Kuchinsky A, Landys N, Workman C, et al. Integration of 555 biological networks and gene expression data using Cytoscape. Nature protocols 556 2007;2(10):2366-82 doi 10.1038/nprot.2007.324. 557 16. Szklarczyk D, Morris JH, Cook H, Kuhn M, Wyder S, Simonovic M, et al. The STRING 558 database in 2017: quality-controlled protein-protein association networks, made broadly 559 accessible. Nucleic Acids Res 2017;45(D1):D362-D8 doi 10.1093/nar/gkw937. 560 17. Subramanian A, Tamayo P, Mootha VK, Mukherjee S, Ebert BL, Gillette MA, et al. Gene set 561 enrichment analysis: a knowledge-based approach for interpreting genome-wide

562 expression profiles. Proc Natl Acad Sci U S A 2005;102(43):15545-50 doi 563 10.1073/pnas.0506580102. 564 18. Cox J, Hein MY, Luber CA, Paron I, Nagaraj N, Mann M. Accurate proteome-wide label-free 565 quantification by delayed normalization and maximal peptide ratio extraction, termed 566 MaxLFQ. Mol Cell Proteomics 2014;13(9):2513-26 doi 10.1074/mcp.M113.031591. 567 19. Hornberg E, Ylitalo EB, Crnalic S, Antti H, Stattin P, Widmark A, et al. Expression of 568 androgen receptor splice variants in prostate cancer bone metastases is associated with 569 castration-resistance and short survival. PLoS One 2011;6(4):e19059 doi 570 10.1371/journal.pone.0019059. 571 20. Ylitalo EB, Thysell E, Jernberg E, Lundholm M, Crnalic S, Egevad L, et al. Subgroups of 572 Castration-resistant Prostate Cancer Bone Metastases Defined Through an Inverse 573 Relationship Between Androgen Receptor Activity and Immune Response. Eur Urol 574 2017;71(5):776-87 doi 10.1016/j.eururo.2016.07.033. 575 21. Chen G, Gharib TG, Huang CC, Taylor JM, Misek DE, Kardia SL, et al. Discordant protein and 576 mRNA expression in lung adenocarcinomas. Mol Cell Proteomics 2002;1(4):304-13. 577 22. Latonen L, Afyounian E, Jylha A, Nattinen J, Aapola U, Annala M, et al. Integrative 578 proteomics in prostate cancer uncovers robustness against genomic and transcriptomic 579 aberrations during disease progression. Nat Commun 2018;9(1):1176 doi 10.1038/s41467- 580 018-03573-6. 581 23. Li N, Zhai Y, Zhang Y, Li W, Yang M, Lei J, et al. Structure of the eukaryotic MCM complex at 582 3.8 A. Nature 2015;524(7564):186-91 doi 10.1038/nature14685. 583 24. Cai C, Wang H, Xu Y, Chen S, Balk SP. Reactivation of androgen receptor-regulated 584 TMPRSS2:ERG gene expression in castration-resistant prostate cancer. Cancer Res 585 2009;69(15):6027-32 doi 10.1158/0008-5472.CAN-09-0395. 586 25. Sharma NL, Massie CE, Ramos-Montoya A, Zecchini V, Scott HE, Lamb AD, et al. The 587 androgen receptor induces a distinct transcriptional program in castration-resistant 588 prostate cancer in man. Cancer Cell 2013;23(1):35-47 doi 10.1016/j.ccr.2012.11.010. 589 26. Taylor BS, Schultz N, Hieronymus H, Gopalan A, Xiao Y, Carver BS, et al. Integrative genomic 590 profiling of human prostate cancer. Cancer Cell 2010;18(1):11-22 doi 591 10.1016/j.ccr.2010.05.026. 592 27. Cuzick J, Swanson GP, Fisher G, Brothman AR, Berney DM, Reid JE, et al. Prognostic value of 593 an RNA expression signature derived from cell cycle proliferation genes in patients with 594 prostate cancer: a retrospective study. Lancet Oncol 2011;12(3):245-55 doi 595 10.1016/S1470-2045(10)70295-3. 596 28. Kumar A, Coleman I, Morrissey C, Zhang X, True LD, Gulati R, et al. Substantial 597 interindividual and limited intraindividual genomic diversity among tumors from men with 598 metastatic prostate cancer. Nat Med 2016;22(4):369-78 doi 10.1038/nm.4053. 599 29. Bergstrom SH, Rudolfsson SH, Bergh A. Rat Prostate Tumor Cells Progress in the Bone 600 Microenvironment to a Highly Aggressive Phenotype. Neoplasia 2016;18(3):152-61 doi 601 10.1016/j.neo.2016.01.007. 602 30. Tomlins SA, Mehra R, Rhodes DR, Cao X, Wang L, Dhanasekaran SM, et al. Integrative 603 molecular concept modeling of prostate cancer progression. Nat Genet 2007;39(1):41-51 604 doi 10.1038/ng1935.

605 31. You S, Knudsen BS, Erho N, Alshalalfa M, Takhar M, Al-Deen Ashab H, et al. Integrated 606 Classification of Prostate Cancer Reveals a Novel Luminal Subtype with Poor Outcome. 607 Cancer Res 2016;76(17):4948-58 doi 10.1158/0008-5472.CAN-16-0902. 608 32. Zhao SG, Chang SL, Erho N, Yu M, Lehrer J, Alshalalfa M, et al. Associations of Luminal and 609 Basal Subtyping of Prostate Cancer With Prognosis and Response to Androgen Deprivation 610 Therapy. JAMA Oncol 2017 doi 10.1001/jamaoncol.2017.0751. 611 33. Beer TM, Armstrong AJ, Rathkopf D, Loriot Y, Sternberg CN, Higano CS, et al. Enzalutamide 612 in Men with Chemotherapy-naive Metastatic Castration-resistant Prostate Cancer: 613 Extended Analysis of the Phase 3 PREVAIL Study. Eur Urol 2017;71(2):151-4 doi 614 10.1016/j.eururo.2016.07.032. 615 34. Ryan CJ, Smith MR, de Bono JS, Molina A, Logothetis CJ, de Souza P, et al. Abiraterone in 616 metastatic prostate cancer without previous chemotherapy. N Engl J Med 617 2013;368(2):138-48 doi 10.1056/NEJMoa1209096. 618 35. Nguyen HG, Yang JC, Kung HJ, Shi XB, Tilki D, Lara PN, Jr., et al. Targeting autophagy 619 overcomes Enzalutamide resistance in castration-resistant prostate cancer cells and 620 improves therapeutic response in a xenograft model. Oncogene 2014;33(36):4521-30 doi 621 10.1038/onc.2014.25. 622 36. Araujo JC, Trudel GC, Saad F, Armstrong AJ, Yu EY, Bellmunt J, et al. Docetaxel and 623 dasatinib or placebo in men with metastatic castration-resistant prostate cancer (READY): a 624 randomised, double-blind phase 3 trial. Lancet Oncol 2013;14(13):1307-16 doi 625 10.1016/S1470-2045(13)70479-0. 626 37. Michaelson MD, Oudard S, Ou YC, Sengelov L, Saad F, Houede N, et al. Randomized, 627 placebo-controlled, phase III trial of sunitinib plus prednisone versus prednisone alone in 628 progressive, metastatic, castration-resistant prostate cancer. J Clin Oncol 2014;32(2):76-82 629 doi 10.1200/JCO.2012.48.5268. 630 38. Smith M, De Bono J, Sternberg C, Le Moulec S, Oudard S, De Giorgi U, et al. Phase III Study 631 of Cabozantinib in Previously Treated Metastatic Castration-Resistant Prostate Cancer: 632 COMET-1. J Clin Oncol 2016;34(25):3005-13 doi 10.1200/JCO.2015.65.5597. 633

634 Acknowledgements.

635

636 The authors are grateful to Pernilla Andersson, Susanne Gidlund and Igor Paron for

637 excellent technical assistance and acknowledge the Uppsala-Umeå Comprehensive

638 Cancer Consortium for giving access to biobank samples. This work was supported by

639 grants from the Danish Research Council (DFF 4004-00450, Flores-Morales), the

640 Movember Foundation (Flores-Morales) and the Danish Cancer Society (R90-A6060-14-

641 S2, Flores-Morales); Swedish Research Council (K2013-64X-20407-04-3, Wikström), the

642 Swedish Cancer Society (CAN 2016/824, CAN 2013/1324, Wikström and 130293,

643 Bergh), Swedish Foundation for Strategic Research (RB13-0119, Wikström); Swedish

644 Cancer Society (CAN 2015/732 Widmark). The Novo Nordisk Foundation Center for

645 Protein Research, University of Copenhagen is supported financially by the Novo Nordisk

646 Foundation (Grant agreement NNF14CC0001).

647

648 Author contributions:

649 D.I-G. contributed to Study concept and design, Acquisition, analysis and interpretation of data,

650 Drafting of the manuscript and Statistical analysis; E.T. contributed to Acquisition, analysis and

651 interpretation of data and Statistical analysis; S.T. contributed to Acquisition, analysis and

652 interpretation of data and Statistical analysis; S.C. contributed to Technical or material support;

653 A.S. contributed to Acquisition, analysis and interpretation of data; T.S.L. contributed to

654 Acquisition, analysis and interpretation of data; T.G. contributed to Acquisition, analysis and

655 interpretation of data; J.C. contributed to Acquisition, analysis and interpretation of data and

656 Statistical analysis; A.W. contributed to Technical or material support; A.B. contributed to Critical

657 revision of the manuscript for important intellectual content and Technical or material support;

658 M.M. contributed to Critical revision of the manuscript for important intellectual content and

659 Technical or material support; A.F-M. contributed to Study concept and design, Acquisition,

660 analysis and interpretation of data, Drafting of the manuscript and Technical or material support;

661 P.W contributed to Study concept and design, Acquisition, analysis and interpretation of data,

662 Drafting of the manuscript and Statistical analysis.

663

664 Competing interests:

665 The authors declare no conflict of interest

666

667 Figure legends

668

669 Figure 1. Proteomic profile of bone metastasis PCa. (A) Number of valid SILAC ratios

670 obtained per PC bone metastatic tumor. (B) Spearman´s correlation between protein and

671 mRNA expression variation pairs obtained from the same tumor sample. Black bars

672 indicate correlation with p< 0.05 (C) Biological processes with generally high and low

673 correlation between protein and mRNA expression based on the variation of their

674 components (see M&M section). (D) Scatter plot of the average protein expression in

675 localized prostate tumors (T) compared to PCa bone metastases (Met). Color represents

676 density. (E) Hierarchical cluster of localized prostate tumors (Blue), neighboring benign

677 prostate tissue (Green) and PCa bone metastases (Black, castration resistant tumors;

678 Brown, hormone naïve tumors; Grey, tumor collected shortly after the beginning of the

679 androgen deprivation therapy).

680

681 Figure 2. Network representation of gene ontology terms enriched among proteins

682 differentially regulated between localized and metastatic PCa tumors. Functional

683 categories overrepresented among the proteins with elevated (A, red) or reduced (B, blue)

684 expression in the bone metastasis compared to localized tumors. Categorical enrichment

685 was calculated using DAVID and enrichment results were plotted using Cytoscape. Line

686 thickness represents number of shared proteins between categories. On each node, the

687 size of the red inner circle and the thickness of the red ring relates to the number of

688 proteins up-regulated and down-regulated, respectively.

689

690 Figure 3. Set of proteins that positively or negatively correlate with the expression

691 of the MCM complex. (A) The protein expression profile across metastatic tumors of

692 MCM complex protein components were set as reference (black lines). Expression profiles

693 of proteins that positively and negatively correlated with the average expression of the

694 MCM proteins are presented in red and blue, respectively. Profile shows the expression as

695 Log2 of the SILAC intensity ratio for each protein after application of Z-transformation.

696 Tumors: brown (mPC31-35), hormone naïve tumors; black (mPC37-52), castration

697 resistant tumors; grey (mPC36), tumor collected shortly after the beginning of the

698 androgen deprivation therapy. (B) Functional categories overrepresented among the

699 proteins that positively (red) and negatively (blue) correlate with the average expression of

700 the MCM complex components. Network representation of the proteins of each category

701 was displayed according to the String database score and plotted using Cytoscape

702 (www.string-db.org (16))

703

704 Figure 4. Phenotypic subgroups of PCa bone metastasis. (A) Hierarchical clustering of

705 PCa bone metastatic tumors (mPC31-52) according to the expression of the selected

706 proteins that significantly correlate (Positively or negatively) with the expression of the

707 MCM complex, resulting in two main clusters; BM1 and BM2. AR and CHGA expression is

708 indicated. Grey represents protein not detected in the sample; Blue, expression lower than

709 the mean in localized tumors minus 2 standard deviations; Red; expression higher than

710 the mean in localized tumors plus 2 standard deviations; Black, similar levels as in

711 localized tumors. For CHGA, red indicates protein being detected and quantified. *

712 Indicates tumors with canonical neuroendocrine features of low AR and high CHGA

713 expression as addressed by IHC. (B) Gene ontology of the biological processes and

714 cellular compartments enriched among the proteins significantly upregulated (p <0.05; fold

715 change 1.5) in BM1 compared to BM2 metastases (blue) and vice versa (red). Y axis

716 indicates the –log10(FDR of the enrichment). (C) Correlation between the immunoreactivity

717 of MCM3 and PSA in each tumor of the validation cohort. (D,E) MCM3 (D) and PSA (E)

718 immunoreactivity was evaluated on diagnostic biopsies and paired PCa bone metastasis.

719 Color represent the metastatic subgroup of each patient.

720

721 Figure 5. Phenotypic features of PCa bone metastases show correlation with

722 disease prognosis after first ADT.

723 (A,B) Kaplan-Meier analysis of survival after first ADT of patients bearing PCa bone

724 metastases with different expression of MCM3 and PSA, defined as MCM3 high/low and

725 PSA high/low using median levels (21 and 6, respectively) as cut-offs. (A) Patients were

726 distributed as their tumors expressed: low MCM3 and high PSA (BM1-like); high MCM3

727 and low PSA (BM2-like); low MCM3 and low PSA; and high MCM3 and high PSA. (B)

728 Patients were dichotomized as high MCM3 and low PSA (BM2-like) compared to all other

729 tumors. (C) Schematic representation of PCa progression model.

730

731

Table I. Patient characteristics at prostate cancer diagnosis and at time for orthopedic surgery due to bone metastasis complications.

Proteomic cohort (n=22) Validation cohort (n=65) Age at diagnosis (yrs) 72 (63; 77) 69 (63; 75) Age at metastasis surgery (yrs) 73 (65; 81) 71 (67; 78) PSA at diagnosis (ng/ml) 110 (46; 990) 100 (42; 750) PSA at metastasis surgery (ng/ml) 241 (41; 1100) 260 (61; 830) Follow up after diagnosis (mo.) 38 (26; 69) 43 (25; 73) Follow up after first ADT (mo.) 38 (26; 64) 42 (25; 71) Follow up after metastasis surgery (mo.) 12 (2.4; 32) 10 (3.0; 24) Gleason score at diagnosis 7 4 17 8 6 17 9 5 13 Not available 7 18 Radical prostatectomy Yes 0 2 No 22 63 Previous ADT: None 5 15 Short-term 1 Long-term 16 50 Treatment for CRPC: Bicalutamide 8 27 Chemotherapy 1 8 Ra223 1 6 Bisphosphonate 2 6 Radiation towards operation site 1 8 Median (25th; 75th percentiles)

Figure 1 Spearman´s correlation A C 0.8 Biological Process Score FDR 0 7000 -0.8 6000 Arginine and proline metab. 0.45 7.1x10-4 5000 Hexose Biosythetic process 0.58 1.4x10-5 4000 Propanoate metab 0.43 0.004 Cell. amino acid catab 0.41 1.6x10-7

# Valid Ratios Valid # 3000 Cell. carbohydrate biosynth. 0.4 6.8x10-8 BM B Glucolysis/Gluconeogenesis 0.36 0.0024 Mean 0.277 Mitochondrial matrix 0.33 1.1x10-8 600 Mitch. Respiratory chain C- I -0.48 0.003 400 Ribosome -0.36 5.9x10-6 Translation initiation -0.35 3.5x10-6 200 Counts Proteasome -0.33 0.007 0 Oxidative phosphorylation -0.26 0.004 .4 0 8 Translation elongation -0.36 1.5x10-5 -0.8 -0 0. 0.4 0. Spearman´s correlation D E mPC32 mPC50 mPC41 mPC34 mPC45 mPC52 mPC35 mPC37 mPC47 mPC48 mPC39 mPC44 mPC46 mPC31 mPC36 mPC33 mPC38 mPC43 mPC51 mPC40 mPC49 PC56 PC63 PC11 PC7 mPC42 PC17 PC9 PC23 PC55 PC53 PC64 PC60 PC54 PC62 PC58 PC61 PC57 PC10 PC24 PC1 PC27 PC28 PC6 PC3 PC16 PC13 PC26 PC5 PC4 PC12 PC25 PC8 PC19 PC21 PC20 PC18 PC2

10 ρ: 0.61 mPC32 mPC50 mPC41 mPC34 6 mPC45 mPC52 mPC35 2 mPC37 mPC47 0 mPC48 -2 mPC39 mPC44

Av_Log2 (T) Av_Log2 mPC46 -6 mPC31 mPC36 mPC33 -6 -2 0 2 6 10 mPC38 mPC43 mPC51 Av_Log2(Met) mPC40 mPC49 PC56 PC63 PC11 PC7 Correlation mPC42 PC17 Benign_ρ: 0.83 ± 0.03 PC9 PC23 PC55 Tumor_ρ: 0.80 ± 0.037 PC53 PC64 Metastasis_ρ: 0.64 ± 0.077 PC60 PC54 PC62 B vs T_ρ: 0.79 ± 0.04 PC58 PC61 B vs Met_ρ: 0.58 ± 0.072 PC57 PC10 PC24 T vs Met_ρ: 0.61 ± 0.077 PC1 PC27 PC28 PC6 PC3 PC16 PC13 PC26 PC5 PC4 PC12 PC25 PC8 PC19 PC21 PC20 PC18 PC2 Downloaded from clincancerres.aacrjournals.org0.35 0.45 0.55 on0 .65October 0.751, 2021. ©0.85 2018 American Association for Cancer Research. Author Manuscript Published OnlineFirst on July 24, 2018; DOI: 10.1158/1078-0432.CCR-18-1229 Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited.

Figure 2

Collagen A Biosynthetic DNA Damage & replication Lipid Transport Process Fatty Acid Reg Lipid degradation DNA Duplex Transport Rho Protein Unwinding DNA Signal Replication Plasma Transduction Initiation Lipoprotein Particle Mitochondrion Acute Assembly Organization Inflammatory DNA Response Geometric Change Cellular Protein−lipidPhospholipid Macromolecular Complex Efflux Assembly Chromatin Organization DNA−dependent Complex "DNA Damage DNA DNA Assembly Chromatin Response, Replication Replication Modification Reverse Signal Cholesterol Transduction" DNA Repair Transport DNA Cellular Metabolic Regulation Of Macromolecular Chromatin Regulation Of Process Helicase Cellular Complex Chromatin Organization Cell Cycle Activity Protein Subunit Assembly Or Complex Organization Disassembly Chromosome Assembly Lipoprotein Organization Particle DNA Damage Nucleotide−excision Clearance Checkpoint Repair Positive Negative Reg Cellular Somatic Cell Regulation Of Response To Helicase Cellular Cell Cycle DNA Activity Component Checkpoint Stress Recombination Organization DNA Packaging Protein Chromosome Protein Amino RNA Localization Biosynthetic In Organelle SegregationNuclear Acid N−Linked Protein Division Glycosylation Process Cell Cycle Via Asparagine Targeting Organelle Protein Mitosis Fission Import M Phase RNA Ribonucleoprotein Localization Complex Mitotic Cell Cell Division Cell Cycle Nuclear Biogenesis Cycle Process Peptidyl−asparagine Nucleocytoplasmic Transport Transport ncRNA Modification Processing Nuclear rRNA M Phase Of Mitotic "Transcription, Export Mitotic Spindle Metabolic Cell Cycle Microtubule−based DNA−dependent" Nucleic Acid Process Organization Process RNA Transport ncRNA Establishment Metabolic Spindle Transport Process rRNA Microtubule RNA Processing Organization Cytoskeleton mRNA Localization Organization Transport Ribosome Nucleic Acid RNA Biogenesis Transport" tRNA Processing Metabolic Process Cell Cycle RNA Processing B Actin Organization Actin Carbohydrate Membrane Filament−based Process Secondary Actin Cell−substrate Metabolic Junction Cell Junction catabolism Lipids Catabolism Filament Process Assembly Organization Assembly Golgi Vesicle Hexose (Lysosome) Transport Glucose Metabolic Metabolic Process Actin Process Cytoskeleton Cell−substrate Organization Adhesion Fatty Acid Cellular Lipid Cofactor Metabolic Catabolic Metabolic Process Process Cell junction Carbohydrate Process Catabolic Process (no mitochondrial) Glycolipid Tetrahydrobiopterin Cellular Catabolic Vesicle−mediated Process Peptide Transport Biosynthetic Carbohydrate Metabolic Process Catabolic Monosaccharide Polysaccharide Metabolic Glycosphingolipid Process Process Metabolic Catabolic Process Process Membrane Process Lipid Lysosome Catabolic Sulfur Organization Process Metabolic Regulation Of Process Vesicle−mediated Sphingolipid Transport Cellular Catabolic Tetrahydrobiopterin Carbohydrate Process Metabolic Biosynthetic Downloaded from clincancerres.aacrjournals.orgProcess on October 1,Process 2021. © 2018 American Association for Cancer Vesicle Transport Research. Author Manuscript Published OnlineFirst on July 24, 2018; DOI: 10.1158/1078-0432.CCR-18-1229 Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited.

Figure 3

A mPC31mPC32mPC33mPC34mPC35mPC36mPC37mPC38mPC39mPC40mPC41mPC42mPC43mPC44mPC45mPC46mPC47mPC48mPC49mPC50mPC51mPC52 4

-2 Log2 (Ratio L/H)_Z score L/H)_Z (Ratio Log2 -4 B RNA metabolism Chromatin remodeling PPP1R8 GPKOW GATAD2B RBM6 MTA2 POP1 ESRP1 RBBP4 MBD3 NCOR2 CENPV HTATSF1 DHX15 DHX16 CDKN2A MORF4L2 SMARCA5 RBM8A MAGOHB RBM15 ANP32B RBM4B TRMT6 SMARCE1 RBM39 DDX39A EIF4A3 PLRG11 ACTL6A PPIH DNA metabolism PSIP1 MAGOH PRPF19 NASP ANP32E SRSF7SRSF PTMA GATAD2B SRSF11 NCBP2 LUC7L2 FAM98B PUF60 U2AF1 SRSF3SRSF PRPF4 PRPF3 DEK EXOSC6 DNAJC8 HNRNPA1HNRNPA Cell Cycle MBD3 TRA2B HNRNPH1 HNRNPA1L2 LSM1 NBN NPLOC4 PAPOLA DDX6 ARHGEF2 SRSF6 EXOSC6 MTA2 RPA2 MSH6 FIP1L1 PCBP2 PATL1 ATM NFIB ADAR TIA1 MDC1 LLGL2 DEK STMN1 TSN RBBP4 RAVER1 ELAC2 DNAJA1 RPA1 SMARCA5 CPSF1 PTBP3 ELAVL1 RPA3 MSH2 NDRG1 SSRP1 SARS MORF4L2 ALKBH5 CDK1 TDP2 NASP CDKN2A SETDB1 SUPT16H ATM USP7 FEN1 UCHL5 ACTL6A HMGB22 CDK2 APPL1 TOP2A CDKN2A RBBP4 PRPF19 UBE2I PDS5A RRM1 CDK4 NEDD1 SUMO2CDKN2AA POLD1 TOP2A MSH6 PCNA CACYBP CDK2 R 1 NUDC PCNA RFC1 25 PLRG1 PDS5A USP7 RFC2 MSH2 CKS2 PARP1 RFC3 FEN1 RFC5 SAE1 RFC4 TUBB KIF2A RCC2 HNRNPA1 CDK1 ATMM 20 UBE2I TERF2IP RRM1M1 RPA3 PRPF19 RPA2 PPP5C KPNA2 NBN NAE1 MAPRE2 MDC1 UBA3 CENPV NASP 15 POGZ MAPKAPK2 SPIN1 SCRIB SEPT3 PPP5C 10

Amino Acid & enrichment -log10(GO FDR) Golgi/ ER Fatty Acid Metab 0 TIRAP3 HIBADH AKR1A1 ERGIC1 STEAP2 IVD A repair QARS A splicingCell cycle ALDH6A1 DN TMED7 PCCB A processingRN GSTP1 ABAT EPS15 TMED2 DNA metabolismRN SCYL1 Cellular respiration BCKDHA COPB1 TMED10 ALDH4A1PCCA Fatty acid catabolism TMED9 Chromatin remodeling GolgiTricarboxylic vesicle transport acid cycle COPB2 GOPC PYCR1 HIBCH COPA HADHA AUH Carboxylic acid catabolism ETFDH ACADSB SEC23IP ARCN1 ACY1 Cellular amino acid catabolism SEC31A COPG1 GLUD1 ACAD8 SEC24C COPZ1 HADHB COG3 ASS1 ACLY COG7 SEC23B ARFGEF2 AKT1 CRYL1 GBF1 ACOX3 Oxidative Phosphorylation COG6 SCFD1 GCC1 IRS2 ACSL1 GOSR2 AASS CPT2 NDUFA4 ETFDH CRAT ADK SEC16A COG5 COG4 PRKAR2B AADAT ZADH2 UQCRQ GOLGA5 SMS DCXR COX7A2L UQCRFS11 DOPEY2 LYPLA1 ATP5S GOSR1 CYCS COX6C MACROD1 ARHGEF7 SCAMP2 NDUFB3 UQCRC1 NAPA TCA ATP5J ATP5F1 ATP5C1 IDH3A COX4I1 ATP5H IDH3A

COX6B1 MDH2 IDH3B SUCLA2 COX5A ATP5B ATP5O ACPP COASY ACO2 SOD2 SUCLA2 JUNB IDH3B MDH2 ACO2Downloaded from clincancerres.aacrjournals.orgNUDT4 on October 1, 2021. © 2018 American Association for Cancer OXA1L HK1Research. Author Manuscript Published OnlineFirst on July 24, 2018; DOI: 10.1158/1078-0432.CCR-18-1229 Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited.

Figure 4 A B

BM1 BM2 BM1 BM2 20

(FDR) 10 10 * *

mPC48 mPC41 mPC39 mPC47 mPC44 mPC37 mPC45 mPC52 mPC42 mPC35 mPC34 mPC40 mPC50 mPC43 mPC38 mPC51 mPC49 mPC33 mPC46 mPC36 mPC31 5 -log

AR 0 CHGA repair splicingA replication DN A DN mRNA ER-GolgiGolgi transport organization mit. electron transportCOPII vesicle coatingnucleosome assembly fatty acid beta-oxidation G1/S transition of cell cycle

C Bone metastasis (n=65) Positivecorrelation r = - 0.58 p < 0.0001 100

MCM3(%) 40

20 Negativecorrelation 0

0 2 4 6 8 10 12 PSA IR -3 -2 -1 0 21 3

p=0.01; n=12 p=0.023; n=14 D E 12 100 BM1 BM1 10 BM2 80 BM2 Other 8 Other 60 6

40 PSAIR 4 MCM3(%) 20 2 0 0 Bone Bone PrimaryDownloaded from clincancerres.aacrjournals.org on PrimaryOctober 1, 2021. © 2018 American Association for Cancer Metastasis Research. Metastasis Author Manuscript Published OnlineFirst on July 24, 2018; DOI: 10.1158/1078-0432.CCR-18-1229 Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited.

Figure 5 A B BM1-like (n=21) BM1-like + rest (n=43)

1,0 BM2-like (n=22) 1,0 BM2-like (n=22)

0,8 Low MCM3 0,8 & low PSA (n=12) 0,6 High MCM3 0,6 & high PSA (n=10) 0,4 0,4 CumSurvival CumSurvival

0,2 0,2

0,0 0,0

100806040200 120 140 160 06040200 1008 120 140 160 Follow up after first ADT Follow up after first ADT BM2-like vs BM1-like LogRank= 5.7 (p=0.017) LogRank= 4.6 (p=0.031) HR= 2.1 (1.143 - 3.879) HR= 211 (1.07 - 4.086) C Benign Prostate Localiced PCa Bone Metastasis

Metabolic targeting

Mitochondrial respiration Amino acid catabolism BM1 Golgi/ ER ADT

BM2 Proliferation rate

Chemotherapy Downloaded from clincancerres.aacrjournals.org on October 1, 2021. © 2018 American Association for Cancer Research. Author Manuscript Published OnlineFirst on July 24, 2018; DOI: 10.1158/1078-0432.CCR-18-1229 Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited.

The proteome of prostate cancer bone metastasis reveals heterogeneity with prognostic implications

Diego Iglesias-Gato, Elin Thysell, Stefka Tyanova, et al.

Clin Cancer Res Published OnlineFirst July 24, 2018.

Updated version Access the most recent version of this article at: doi:10.1158/1078-0432.CCR-18-1229

Supplementary Access the most recent supplemental material at: Material http://clincancerres.aacrjournals.org/content/suppl/2018/07/24/1078-0432.CCR-18-1229.DC1 http://clincancerres.aacrjournals.org/content/suppl/2019/04/03/1078-0432.CCR-18-1229.DC2

Author Author manuscripts have been peer reviewed and accepted for publication but have not yet Manuscript been edited.

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/2018/07/24/1078-0432.CCR-18-1229. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.