Reprogramming of Isocitrate Dehydrogenases Expression and Activity by the Androgen

Home , IDH2, IDH3G

Author Manuscript Published OnlineFirst on May 8, 2019; DOI: 10.1158/1541-7786.MCR-19-0020 Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited.

1 Reprogramming of isocitrate dehydrogenases expression and activity by the androgen

2 receptor in prostate cancer

4 Kevin Gonthier1,2, Raghavendra Tejo Karthik Poluri1,2, Cindy Weidmann1, Maude Tadros1, and

5 Étienne Audet-Walsh1,2,3

7 1 Endocrinology - Nephrology Research Axis, Centre de recherche du CHU de Québec -

8 Université Laval, Québec City, Canada

9 2 Department of molecular medicine, Faculty of Medicine, Université Laval, Québec City,

10 Canada

11 3 Centre de recherche sur le cancer de l’Université Laval, Québec City, Canada

13 Keywords: steroid; metabolism; glioma; nuclear receptor; castration-resistant prostate cancer,

14 IDH1, oncometabolism

16 Abbreviated title: Reprogramming of IDH activity by AR in PCa

18 Corresponding Author: Étienne Audet-Walsh, Centre de recherche du CHU de Québec, 2705

19 Boulevard Laurier, room R-4714, Québec City, QC, Canada, G1V 4G2

20 Phone: 1-418-525-4444 ext. 48678 ; e-mail : [email protected]

22 Disclosure statement: the authors declare no potential conflicts of interest.

Downloaded from mcr.aacrjournals.org on September 24, 2021. © 2019 American Association for Cancer Research. Author Manuscript Published OnlineFirst on May 8, 2019; DOI: 10.1158/1541-7786.MCR-19-0020 Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited.

23 Abstract

24 Mutations of the isocitrate dehydrogenase genes IDH1 and IDH2, key enzymes involved in

25 citrate metabolism, are important oncogenic events in several cancer types, including in 1-3% of

26 all prostate cancer (PCa) cases. However, if IDH1 and other IDH isoforms are associated with

27 PCa progression, as well as the regulatory factors controlling their expression and activity,

28 remain mostly unknown. Using publicly available datasets, we showed that PCa harbors the

29 highest IDH1 expression across the human cancer spectrum and that IDH1 expression is altered

30 during PCa progression. We showed that the androgen receptor (AR), a key oncogene in PCa,

31 controls multiple IDH isoforms in both in vitro and in vivo models, predominantly positively

32 regulating IDH1. Chromatin immunoprecipitation experiments confirmed the recruitment of AR

33 at several regulatory regions of IDH1 and enzymatic assays demonstrated that AR significantly

34 induces IDH activity. Genetic blockade of IDH1 significantly impaired PCa cell proliferation,

35 consistent with IDH1 having a key function in these cancer cells. Importantly, knockdown of

36 IDH1 blocked the AR-mediated induction in IDH activity, indicating that AR promotes a

37 mitochondrial to cytoplasmic reprogramming of IDH activity. Overall, our study demonstrates

38 that IDH1 expression is associated with PCa progression, that AR signaling integrates one of the

39 first transcriptional mechanisms shown to regulate IDH1, and that AR reprograms PCa cell

40 metabolism by selectively inducing extra-mitochondrial IDH activity.

41 Implications: The discovery that AR reprograms IDH activity highlights a novel metabolic

42 reprogramming necessary for PCa growth and suggests targeting IDH activity as a new

43 therapeutic approach for PCa treatment.

45 Introduction

46 Prostate cancer (PCa) is an androgen-dependent disease that relies on the androgen receptor (AR)

47 for cancer cell proliferation and survival, and thus inhibiting AR activity is currently the

48 cornerstone therapy in clinical settings (1, 2). Despite the positive response of most cancer cases

49 to AR blockage, evolution of the disease to castration-resistant PCa (CRPC) is almost inevitable

50 (1-4), highlighting the need to develop novel therapies for PCa. In >80% of all cases, resistance

51 involves AR reactivation (5-8). Cancer cells can also evolve into AR-negative PCa, which are

52 often referred to as neuroendocrine PCa (NEPC)(9).

53 Glucose, amino acids, and fatty acids can fuel mitochondrial activity through the

54 tricarboxylic acid (TCA) cycle for ATP production and to sustain biosynthesis of essential

55 cellular components (e.g., nucleotides and lipids). The prostate is metabolically unique in human

56 in that epithelial prostate cells have a blunted TCA cycle to produce high levels of citrate (10,

57 11). One of the earliest events in PCa, which occurs in virtually all cancer cases, is the

58 reprogramming of this unique metabolic program (10, 11). Given its major oncogenic functions

59 in PCa, it is not surprising that AR was identified as a major orchestrator of energy metabolism

60 and that PCa progression was linked to specific metabolic gene signatures (12-18). Moreover, AR

61 was shown to alter citrate metabolism, most notably by inducing mitochondrial respiration and

62 dysregulating lipid metabolism, contributing to the rising characterization of AR functions in the

63 transcriptional control of energy metabolism (12, 14, 15). However, what are the key limiting

64 enzymatic steps under the AR-driven metabolic program remain unknown.

65 It is now well recognized that mutations of isocitrate dehydrogenase genes IDH1 and

66 IDH2 lead to the establishment of an oncometabolic environment that causes cancer, notably by

67 altering the epigenome (19, 20). These mutations, which induce a novel enzymatic activity

68 catalyzing the formation of 2-hydroxyglutarate, interfere with α-ketoglutarate (αKG)-dependent

69 epigenetic regulatory mechanisms and lead to the hyper-methylation of the genome (20-23).

70 Although frequent in gliomas and other cancer types, IDH1 mutations occur in only 1–3% of PCa

71 and are believed to be the underlying oncogenic drivers in these cases (24-28). On the basis that

72 the prostate harbors a distinct citrate metabolic program, we hypothesized that an alteration in the

73 regulation of IDH enzymes could be linked to PCa development and progression, even in PCa

74 cases without IDH1 mutation.

75 IDHs are encoded by five different genes leading to three possible functional IDH

76 enzymatic complexes (IDH1, IDH2, and IDH3, the latter comprising IDH3A, IDH3B, and

77 IDH3G). In the mitochondria, either the IDH2 or the IDH3 complexes catalyze the

78 decarboxylation of isocitrate, produced from citrate, into αKG using respectively NADP+ and

79 NAD+ as co-substrates. This represents a key step in the mitochondrial TCA cycle. In addition,

80 cytoplasmic IDH1 promotes the NADP+-dependent production of extra-mitochondrial αKG,

81 which is required for the proper activity of several enzymes involved in epigenetic regulation

82 (20-23). Newly synthesized NADPH can be used for anabolic processes such as lipogenesis.

83 However, what are the molecular mechanisms controlling IDH gene expression and how these

84 dynamics are altered in PCa cell metabolism are still unknown.

85 In the current study, we observed that wild-type IDH1 is predominantly expressed in PCa

86 cells and that its expression is altered throughout PCa progression. We identified AR as a key

87 regulator of several IDH isoforms, most notably IDH1. We showed that androgen stimulation

88 reprograms IDH activity, favoring cytoplasmic over mitochondrial metabolism. Importantly,

89 knockdown of IDH1 decreased cancer cell proliferation and a blockade of the IDH metabolic

90 reprogramming, possibly highlighting IDH1 as a potential new therapeutic target for PCa.

92 Material and Methods

93 Cell Culture

94 LNCaP, LAPC4, 22rv1, PC3, VCaP, and DU145 cells were grown in RPMI

95 supplemented with 10% fetal bovine serum (FBS), penicillin, streptomycin, and sodium pyruvate

96 at 37°C and 5% CO2. All cells were kept in culture for no more than 3 months after resuscitation

97 and were re-authenticated using the ATCC cell line authentication service during summer 2016.

98 For hormonal treatments with 10nM of the synthetic androgen R1881, steroid deprivation was

99 first performed for 48h in phenol red-free RMPI 1640 with 5% charcoal-stripped serum (CSS) as

100 described previously (13).

101

102 Cellular proliferation

103 AR+ PCa cells and AR- PCa cells were seeded in 96-well plates in media with 5% CSS, with

104 fresh media changed every 48h. For cell counting, culture medium was discarded and 50ul of

105 0.5% crystal violet staining solution was added to each well before being incubated at room

106 temperature on a bench rocker. After 30 minutes, plates were rinsed 4 times with water and plates

107 were dried overnight at room temperature. 200ul of SDS 2% were then added to each well and

108 plates were incubated at room temperature for 3h. Optical density (OD) was then measured at

109 570 nm using a spectrophotometer. OD values were then converted into cell numbers using

110 established standard curves for each cell line tested.

111

112 IDH enzyme activity

113 IDH enzyme activity was measured using the IDH Activity Assay Kit (Sigma), which measures

114 NADH/NADPH synthesis during the conversion of isocitrate into α-ketoglutarate by IDH

115 enzymes. For IDH activity from whole cell lysates, PCa cells were harvested in PBS and counted

116 using an automated cell counter TC10 (Biorad). For basal activity, cells were harvested following

117 48h in complete media. For androgen stimulation, steroid deprivation was performed for 48h in

118 media with 5% CSS before 48h stimulation with R1881 in media with 5% CSS. After cell count,

119 cells were resuspended in ice-cold IDH assay buffer at a 1 million cells per 200uL concentration.

120 Cells were then diluted 1:8 in IDH assay buffer before the enzymatic activity assay. For

121 purification of cytoplasmic and mitochondrial extracts for IDH enzymatic activity, cellular

122 fractionation was performed according to the protocol of Gravel et al. (29). For siRNA studies, in

123 the hour following seeding in media with 5% CSS, LNCaP and VCaP cells were transfected with

124 a siRNA pool targeting IDH1 (Dharmacon) or a non-targeted control using the Hiperfect

125 transfection reagent according to the manufacturer's instructions (Qiagen). 48h later, fresh media

126 containing 10nM R1881 or vehicle was added for 48h before harvesting the cells for IDH activity

127 assay. All experiments were performed in biological duplicates or triplicates and performed 2-5

128 times independently. Student’s T test and ANOVA with Tukey HSD post-hoc test were used to

129 evaluate statistical significance.

130

131 RNA extraction and quantitative reverse-transcriptase PCR

132 Isolation and purification of total RNA were performed using the EZ-10 RNA extraction kit (Bio

133 Basic); first strand cDNA synthesis was performed using the LunaScript RT SuperMix Kit from

134 New England BioLabs (NEB). The Luna Universal qPCR Master Mix (NEB) was then used for

135 qPCR and quantification of specific genes, and duplicate technical replicates were performed for

136 every biological samples. Relative expression was normalized to the expression of two

137 housekeeping genes, PUM1 and TBP. Primer sequences are detailed in Supplementary Table S1.

138 Results are shown as the average of 2-4 independent experiments with at least 3 biological

139 replicates per condition and the Student’s T test was used to evaluate statistical significance. For

140 RNA-seq analysis, LNCaP, 22rv1, and PC3 cells were maintained in media with 5% CSS for 72h

141 before RNA purification using the RNeasy plus purification kit from Qiagen. RNA from three

142 biological samples per cell lines were sent to the Genomic Centre of the Centre de recherche du

143 CHU de Québec – Université Laval for sequencing. Sequence quality control analyses were

144 performed using FastQC (30) and trimming using Trimmomatic (31). Alignment and

145 quantification of the samples to the transcriptome hg38 were then performed using Kallisto (32).

146 RNA-seq data are available on the Gene Expression Omnibus (GEO: GSE128201).

147

148 ChIP-qPCR

149 Chromatin immunoprecipitation (ChIP)-qPCR was performed in LNCaP and 22rv1 cells

150 using an anti-AR antibody (sc-7305, Santa Cruz BioTechnology) as previously described (13). In

151 brief, after treatment with R1881 for 16h, cells were cross-linked in 1% formaldehyde for 10

152 minutes at room temperature. Nuclei were then purified using sequential centrifugation and

153 chromatin was sonicated. Immunoprecipitation was then performed overnight at 4ºC, and AR-

154 bound chromatin was enriched using Dynabeads® (Life TechnologiesTM). DNA was then

155 purified using the fast ChIP protocol (33). For ChIP-qPCR, cells treated with vehicle were used

156 as controls of antibody non-specific binding, and two to three negative regions were used for

157 ChIP normalization between samples. Sequences of primers used for AR DNA binding positive

158 and negative regions are detailed in Supplementary Table S2. Results are shown as the average of

159 3 independent experiments. ChIP-seq experiments were performed using the same protocol with

160 the AR antibody sc-816X from Santa Cruz Biotechnology (Audet-Walsh et al., in preparation)

161 and analyzed as previously described (15). Visualization was performed using the UCSC genome

162 browser following the merge of two independent experiments.

163

164 Clinical Data and Statistical Analyses

165 IDH gene expression data across the human cancer spectrum and in PCa specimens were

166 obtained from The Cancer Genome Atlas (TCGA) consortium (34, 35) and from the Human

167 Protein Atlas website in July 2018 (www.proteinatlas.org; (36)). Data from the Cancer Cell Line

168 Encyclopedia (CCLE), including shRNA results from the Achilles project, were obtained from

169 the webportal (https://portals.broadinstitute.org/ccle) in November 2018 (37). For the Kaplan-

170 Meier and the log-rank test, data from Taylor et al. (GSE21032) were analyzed through the

171 Project Betastasis webpage (www.betastasis.com) (8). Expression data from six benign prostate

172 tissues, seven clinically localized primary prostate tumors, and six hormone-refractory metastatic

173 prostate tumors were described by Varambally et al. 2005 (GSE3325)(38). Expression data from

174 28 benign prostate tissues, 59 localized PCa, and 35 metastatic castration-resistant PCa were

175 described by Grasso et al. (GSE35988)(39). The microarray data from Sun et al. (40) were

176 analyzed for IDH gene expression in human LuCaP35 prostate tumor explants (five mice with

177 sham surgery and five castrated; GDS4120). RNA-seq from xenograft experiments with 22rv1

178 cells and their metastatic subclones were obtained from the Tsai et al. dataset (GSE99857)(41).

179

180

181 Results

182 IDH1 is highly expressed in PCa and is altered during PCa progression

183 To better understand the relationship between IDH gene expression and PCa, we screened for

184 their mRNA levels across the human cancer spectrum using The Cancer Genome Atlas dataset

185 (TCGA)(34, 35). All five IDH genes showed ubiquitous expression across the diverse types of

186 human cancer interrogated (Fig. 1 and Fig. S1), but only IDH1 had a much higher median in PCa

187 compared to other cancer types analyzed (Fig. 1A). The median IDH1 expression level in PCa

188 samples was even about 3-fold higher than in gliomas, a tumor type in which IDH1 mutation is

189 known as a key oncogenic driver linked to disease development (42). Similarly, PCa cell models

190 had the second highest IDH1 expression levels in the Cancer Cell Line Encyclopedia (Fig. 1B).

191 Based on these results, we hypothesized that IDH1 could be an important metabolic enzyme

192 associated with PCa cell metabolism.

193 In human PCa samples, we found that low expression levels of IDH1 were associated

194 with a high recurrence rate following radical prostatectomy (p=0.0473; Fig. 2A). Although PCa

195 has high basal IDH1 expression levels, lower levels are associated with more aggressive diseases.

196 Consistent with this observation, IDH1 levels were significantly lower in metastatic castration-

197 resistant PCa in comparison to those of both primary tumors and benign normal prostate samples

198 from the Grasso et al. (39) dataset (Fig. 2B). These results were validated in a second cohort with

199 similar results (Fig. 2C). Indeed, the four microarray probes targeting IDH1 indicated

200 significantly lower levels in metastatic PCa compared to primary tumors and/or benign normal

201 prostate. These results indicate that IDH1 expression is altered during PCa progression and that

202 more aggressive diseases have lower IDH1 levels.

203 Given the major role of AR in PCa, we were intrigued about the possible relationship

204 between IDH1 expression, PCa progression, and AR status. We hypothesized that evolution of

205 AR+ to AR- (neuroendocrine) PCa tumors favored a metabolic reprogramming involving the loss

206 of IDH1. Using the Grasso et al. dataset (39), we separated CRPC samples in two groups. We

207 first established whether AR-driven CRPC tumors had increased AR expression compared to

208 primary tumors (n=29). Second, we regrouped CRPC tumors that had decreased AR expression

209 compared to primary tumors concomitantly with the increased expression of the neuroendocrine

210 markers ENO2 and CHGA (n=6). As expected, we observed a significant difference in AR

211 expression levels compared to ENO2 and CHGA expression in these two subsets of CRPC,

212 categorized respectively as AR+ and neuroendocrine CRPC (Fig. 2D). Consistent with our

213 hypothesis on IDH1 reprogramming following loss of AR, IDH1 levels were significantly lower

214 in NEPC compared to tumors driven by ARs (Fig. 2D). To further link IDH1 expression with AR

215 expression and activity, we performed correlation analyses of IDH1 expression in relationship

216 with mRNA levels of AR, ENO2, and CHGA using the TCGA cbioportal to access the TCGA,

217 the Taylor et al. (8) and the Kumar et al. clinical datasets (43). Interestingly, IDH1 levels were

218 significantly correlated with AR expression (p>0.00005) in the TCGA dataset, with small yet

219 significant inverse correlation with both NEPC markers (p<0.05; Fig. 2E). We next studied these

220 correlations in a clinically more aggressive setting with the Taylor and the Kumar et al. datasets.

221 First, in the Taylor et al. dataset (Fig. 2F), IDH1 was also found to be significantly and positively

222 correlated to AR but inversely correlated to ENO2 and CHGA (p<0.0007). IDH1 was also

223 positively correlated with AR (p<3x10-5) and inversely correlated with ENO2 and CHGA

224 (p<2x10-7) in the Kumar et al. dataset (Fig. 2G). Altogether, these results demonstrate that IDH1

225 expression is altered during PCa progression and that it is correlated with AR expression.

226

227 AR controls IDH1 expression in PCa cells

228 We then studied the relative expression of the five IDH genes in in vitro models of AR-positive

229 human PCa (LNCaP, LAPC4, and 22rv1 cells; Fig. 3A). In AR-positive PCa cells, IDH1, which

230 encodes the only cytoplasmic IDH enzyme, was the most expressed gene in PCa cells compared

231 to the mitochondrial isoforms (Fig. 3A). IDH2, IDH3A, IDH3B, and IDH3G had similar

232 expression levels, which ranged between 7- and 321-fold lower than IDH1 expression in LNCaP,

233 LAPC4, and 22rv1 cells (Fig. 3A). The only different profile was observed in VCaP cells, which

234 exhibited similar levels of IDH1 and IDH3 genes. We also performed RNA-seq analyses in

235 LNCaP and 22rv1 cells. After normalization, these data also demonstrate that IDH1 is the

236 predominant IDH isoform expressed in these AR-positive PCa cells (Fig. 3B and 3C). We

237 validated our results by reanalyzing publicly available RNA-seq experiments in 22rv1 cells,

238 which also indicated that IDH1 is the most highly expressed IDH isoform in these cancer cells in

239 vitro (Fig. 3D). Tsai and colleagues used 22rv1 cells in vivo to develop a highly metastatic PCa

240 model (41). Re-analyses of their data indicate that, as seen in vitro for 22rv1 cells (Fig. 3C-D),

241 IDH1 was also the most highly expressed IDH isoform in vivo, with expression levels ranging

242 from 4- to 13-fold higher than levels of other IDH genes (Fig. 3E, light green). Interestingly, the

243 metastatic sub-line also showed a predominance of IDH1 expression, but that was statistically

244 decreased compared to the IDH1 expression measured in the parental cell lines (Fig. 3E, dark

245 green). We then studied IDH gene expression levels in AR-negative human PCa cells (PC3 and

246 DU145). These cancer cells harbored a different IDH expression pattern compared to AR-

247 positive PCa cells (Fig. 3A), as IDH1 levels were similar to those of IDH3s (Fig. 3F). We also

248 performed RNA-seq analyses in PC3 cells. Contrary to AR-positive cells, IDH1 was not

249 expressed at a higher level compared to other IDH isoforms (Fig. 3G). Accordingly, total IDH

250 activity was significantly higher in AR-positive PCa cells, with 2-10-fold higher activity than in

251 AR-negative PCa cells (Fig. 3H). Overall, these results indicate that IDH1 is the predominant

252 IDH gene expressed in AR-positive PCa cells in vitro and in vivo.

253 We next investigated the relationship between AR activation and the regulation of IDH

254 expression. Treatment of LNCaP cells with the synthetic androgen R1881 significantly induced

255 the expression of the prostate-specific antigen (PSA) mRNA (encoded by KLK3; Fig. 4A, left

256 panel). AR activation also significantly increased the expression of IDH1 (Fig. 4A, right panel).

257 IDH2 and members of the IDH3 complex were much less stimulated, if at all, by R1881 (Fig. 4A,

258 right panel). As in LNCaP cells, IDH1 was significantly induced in VCaP (Fig. 4B) and 22rv1

259 (Fig. 4C) cells following R1881 treatment. No other IDH genes were significantly modulated by

260 AR in these cells, except a trend for an increase in IDH3A in VCaP cells. In LAPC4 cells, IDH2,

261 IDH3A and IDH3G were moderately induced by androgens (Fig. S2). It should be noted that

262 LAPC4 cells harbor the highest basal IDH1 expression levels across all cell lines tested (Fig. 3A).

263 In human PCa xenografts of LuCaP35, inhibition of the androgen-signaling pathway by

264 castration led to a significant decrease in several IDH genes in vivo (Fig. 4D). Indeed, several

265 microarray probes targeting IDH1, IDH2, and IDH3A showed a significant decrease after

266 castration, sustaining a positive regulation by androgens of these genes in PCa cells in vivo.

267 AR ChIP-seq analyses identified two and three AR DNA binding sites in the vicinity of

268 IDH1 and IDH2, respectively, in LNCaP cells (Fig. 4E and S3A). In the absence of androgens, no

269 peak was observed, while R1881 treatment significantly induced AR recruitment to IDH1 (Fig.

270 4E). No recruitment of AR was detected for IDH3A, IDH3B, or IDH3G in ChIP-seq analyses

271 from LNCaP cells (Supplementary Fig. S3B-D). Similarly, reanalysis of AR ChIP-seq

272 experiments from Massie et al. (16) and Singh et al. (44) also indicates a significant AR DNA

273 binding site 2kb upstream IDH1 transcriptional start site (corresponding to IDH1 #1, Fig. 4E).

274 Validation was performed using ChIP-qPCR experiments, which confirmed the recruitment of

275 AR to genomic DNA regions in proximity to the IDH1 promoter in LNCaP cells (Fig. 4E). Given

276 that AR binding to these sites was relatively low, particularly for the second site, we cannot

277 exclude that AR also controls IDH1 expression in an indirect manner. Complementary validation

278 was performed in 22rv1 cells (Fig. 4E). In accordance with the qPCR results showing that only

279 IDH1 is increased by R1881 in 22rv1 cells (Fig. 4C), AR was exclusively recruited to IDH1

280 promoter in ChIP-qPCR analysis (Fig. 4E), while no recruitment was observed at IDH2 (Fig.

281 S3A). These results indicate that AR controls the expression of IDH1, thus identifying one of the

282 first transcription factors controlling this gene expression in the context of PCa.

283

284 IDH1 modulation reprograms PCa cell metabolism

285 To understand the functional role of IDH1 in AR+ PCa cells, we performed IDH enzymatic

286 activity assays. Following AR activation in LNCaP cells, a significant induction of 3.8-fold of

287 total IDH activity was observed (Fig. 5A). Similarly, R1881 treatment led to a 5.3-fold induction

288 of total IDH activity in VCaP cells (Fig. 5B), correlating with a positive regulation of IDH1

289 expression by AR in these cancer cells (Fig. 4).

290 To further delineate the contribution of IDH1 (in the cytoplasm and peroxisomes) and

291 IDH2/3 (in the mitochondria) to the total IDH activity in PCa cells, with and without AR

292 activation, we performed siRNA-mediated knockdown of IDH1 in LNCaP cells. As expected,

293 transfection of a pool of siRNAs targeted against IDH1 significantly decreased its expression by

294 >90% (Fig. 5C). Importantly, this knockdown was selective and did not affect other IDH gene

295 expression (Fig. 5C). IDH1 knockdown, in absence of androgens, led to a 3.7-fold decrease in

296 total IDH activity in LNCaP cells, indicating that IDH1 represents the major active isoform in

297 these cells (Fig. 5D), correlating with IDH expression data (Fig. 3A). Importantly, IDH1

298 knockdown completely blocked the AR-mediated induction in total IDH activity, with the loss of

299 87% of total IDH activity in LNCaP cells (Fig. 5D). Similarly, we observed in VCaP cells that

300 IDH1 knockdown was specific and did not affect other IDH gene expression (Fig. 5E). As

301 anticipated, IDH1 knockdown completely blocked the R1881-mediated increase in IDH activity

302 in VCaP cells (Fig. 5F).

303 Loss of IDH1 in LNCaP cells significantly decreased proliferation in complete media

304 (Fig. 6A). Similarly, loss of IDH1 also decreased proliferation in absence of androgens,

305 indicating that IDH1 is required for maximal proliferation (Fig. 6B). The reanalysis of the

306 Achilles project data, which uses genome-scale RNAi and CRISPR-Cas9 to silence or knockout

307 gene expression across hundreds of cancer cell lines (45, 46), also confirmed that PCa cells are

308 one of the most sensitive cancer types to the loss of IDH1 (Supplemental Fig. S4). This seemed to

309 be mostly specific to AR-positive cells as IDH1 knockdown in PC3 and DU145 AR-negative

310 cells did not significantly or only modestly change proliferation (Supplemental Fig. S5). These

311 data suggest that an induction of cytoplasmic/mitochondrial IDH activity is required for maximal

312 PCa cell proliferation driven by AR. We thus performed cytoplasmic and mitochondrial

313 preparations known to preserve metabolic activities (29). In extracts from R1881-treated cells, we

314 observed a significant 3.24-fold induction of the cytoplasmic IDH activity, further indicating that

315 IDH1 is the major target of AR in LNCaP (Fig. 6C) cells. Importantly, no increase in

316 mitochondrial IDH activity was observed in extracts from R1881-treated cells (Fig. 6C).

317 Knockdown of IDH1 completely blocked this induction of cytoplasmic IDH activity by AR in

318 LNCaP cells (Fig. 6D), demonstrating that AR promotes a reprogramming of IDH activity

319 favoring cytoplasmic over mitochondrial metabolism.

320

321

322

323 Discussion

324 In the current study, IDH1 was shown to be the predominant IDH gene expressed in PCa and to

325 be tightly linked with AR expression and activity. Accordingly, IDH1 was shown to be sensitive

326 to AR activation in both in vitro and in vivo PCa models. We further identified AR DNA binding

327 sites close to the promoters of IDH1 that were validated by ChIP-qPCR. Using siRNAs, we

328 demonstrated that loss of IDH1 significantly impairs proliferation, but also blocks the

329 reprogramming by AR of cytoplasmic to mitochondrial IDH activity. Consistent with a

330 reprogramming of cell metabolism during carcinogenesis, IDH1 gene expression was

331 significantly altered during PCa progression. Overall, our study highlights the importance of the

332 wild-type IDH1 gene in PCa.

333 In gliomas, glioblastomas, and certain types of leukemia, mutations in IDH1 are

334 frequently identified as a major oncogenic driver (19, 20). IDH1 mutations can occur in PCa but

335 are generally rare (24-28). Given the predominance of IDH1 mutations in other cancer types, it

336 was surprising to observe such a high expression in PCa samples. Rather than an overexpression

337 in PCa primary tumors, IDH1 expression is not different compared to peri-tumor normal tissues

338 (Fig. 2), suggesting that IDH1 levels are normally high in the prostate. This might be related to

339 the citrate-secretory phenotype of prostate epithelial cells that require a unique metabolic

340 program to sustain citrate production but also to control intracellular citrate levels and usage.

341 Indeed, prostate epithelial cells exhibit a truncated mitochondrial citric acid cycle to allow the

342 massive production of citrate, which is subsequently secreted (11).

343 This citrate-secretory phenotype is loss early in the course of PCa, at a stage were the

344 tumor is still localized. The decrease of IDH1, observed in metastatic CRPC samples, probably

345 reflects a subsequent change in citrate metabolism not directly linked to the loss of the prostate

346 citrate-secretory phenotype that occurs earlier. It is not completely clear why IDH1 is decreased

347 in metastatic CRPC samples compared to primary tumors. One explanation is that these tumors

348 comprise 15-20% of NEPC samples, induced upon sequential anti-androgen treatments, which

349 are characterized by low AR and IDH1 expression (Fig. 2). This is probably why low IDH1

350 expression is linked to faster recurrence (Fig. 2A), possibly indicating more aggressive PCa. In

351 addition, it has been shown that cancer cells further undergo metabolic reprogramming during

352 establishment at distal metastatic sites, notably in the breast cancer setting (47). This

353 phenomenon most probably also occurs during PCa metastasis establishment, which is supported

354 by the 22rv1 metastatic model (Fig. 3E). Nevertheless, IDH1 is still highly expressed in AR-

355 positive PCa cells derived from CRPC patients, such as the cellular models used in the present

356 study, as well as responsive to androgens.

357 To better understand the role of IDH1 in PCa, we performed our studies in AR-positive

358 PCa cells, which better represent clinically localized and locally advanced PCa as they are

359 characterized by a dependency on AR signaling. Accordingly, in VCaP, LNCaP, and 22rv1 cells,

360 IDH1 was significantly induced following treatment with R1881 while other IDHs were not

361 consistently modulated upon AR activation. This regulation by androgens was observed in

362 several PCa models and correlation studies in human PCa samples also support the concept that

363 AR controls IDH1 expression. Mostly studied for its enzymatic activity, not much is known

364 about IDH1 gene regulation, and thus AR appears as one of the first transcriptional regulatory

365 mechanisms controlling its expression. Indeed, AR activation induced its expression by up to 4-

366 fold and, more importantly, also induced its activity up to 6-fold in PCa cells. It is also possible

367 that AR controls IDH1 expression and activity in other contexts, such as in cancers with IDH1

368 mutation, and it is tempting to speculate that we could use anti-androgens in these cases to

369 decrease the expression of mutant IDH1 oncogenic enzymes.

370 By specifically inducing IDH1 activity, AR promotes an unexpected metabolic rewiring

371 in PCa cells, favoring cytoplasmic/mitochondrial IDH ratios, which was associated with PCa cell

372 proliferation status. Given that IDH1 is an important pathway for the replenishment of NADPH,

373 its increase could favor biosynthetic pathways that rely on this cofactor for proper activity, such

374 as lipid synthesis that is also activated by AR in PCa cells (12, 15). In addition, IDH1 promotes

375 the synthesis of αKG, which is an important cofactor for several proteins associated with the

376 epigenome modulation. How modulation of αKG levels after modulation of IDH1 will affect the

377 epigenome regulation has yet to be determined and mechanistic studies are required to fully

378 capture the implication of IDH1 activity in PCa cell genomic activity.

379 In PCa samples, IDH1 expression is correlated to AR expression and decreased during the

380 transition of PCa to AR-negative NEPC. IDH1 expression was significantly lower in CRPC

381 human samples compared to primary tumors in two independent datasets. Interestingly, a similar

382 pattern was observed in 22rv1 xenograft experiments, in which the highly metastatic sub-clone

383 also had significantly lower levels of IDH1 compared to the parental cell line. The loss of IDH1

384 expression in AR-negative PCa cells could be a consequence of the absence of a functional AR

385 pathway in these cells. Given that IDH1 expression is required for maximal proliferation in AR+

386 PCa cells and that it is associated with a significant metabolic reprogramming, these results

387 suggest that loss of IDH1 is an important step for the evolution to metastatic or NEPC cancer

388 cells. Further studies are required to fully understand the implication of IDH1 in the metastatic

389 potential of PCa cells.

390 In summary, wild-type IDH1 is an important metabolic gene expressed in the prostate and

391 for which expression is altered during PCa progression. Importantly, the androgen-signaling

392 pathway was identified as one of the first transcriptional regulatory mechanisms that govern

393 IDH1 expression and activity. Indeed, AR was shown to positively regulate IDH1 expression in

394 several PCa models, analogous with a positive correlation between AR and IDH1 expression in

395 human PCa samples. Consequently, AR activation led to a reprogramming of total IDH activity,

396 favoring a cytoplasmic/mitochondrial IDH ratio to sustain PCa cell proliferation. Overall, our

397 study highlights the key role played by IDH1 in AR-positive PCa cell metabolism and suggests it

398 could be used as a novel therapeutic target for the treatment of this disease before evolution to

399 NEPC.

400

401 Acknowledgments. This work was supported by funding to EAW from the Canadian Institutes

402 for Health Research (CIHR; PJT159530) and the Fondation du CHU de Québec (3061). The

403 authors thank Camille Lafront for critical reading of the manuscript.

404

405

406

407 References 408 1. Watson PA, Arora VK, Sawyers CL. Emerging mechanisms of resistance to androgen 409 receptor inhibitors in prostate cancer. Nat Rev Cancer. 2015;15(12):701-11. 410 2. Wyatt AW, Gleave ME. Targeting the adaptive molecular landscape of castration- 411 resistant prostate cancer. EMBO Mol Med. 2015;7(7):878-94. 412 3. Mottet N, Bellmunt J, Bolla M, Briers E, Cumberbatch MG, De Santis M, et al. EAU- 413 ESTRO-SIOG Guidelines on Prostate Cancer. Part 1: Screening, Diagnosis, and Local Treatment 414 with Curative Intent. Eur Urol. 2017;71(4):618-29. 415 4. Cornford P, Bellmunt J, Bolla M, Briers E, De Santis M, Gross T, et al. EAU-ESTRO- 416 SIOG Guidelines on Prostate Cancer. Part II: Treatment of Relapsing, Metastatic, and Castration- 417 Resistant Prostate Cancer. Eur Urol. 2017;71(4):630-42. 418 5. Quigley DA, Dang HX, Zhao SG, Lloyd P, Aggarwal R, Alumkal JJ, et al. Genomic 419 Hallmarks and Structural Variation in Metastatic Prostate Cancer. Cell. 2018;174(3):758-69 e9. 420 6. Viswanathan SR, Ha G, Hoff AM, Wala JA, Carrot-Zhang J, Whelan CW, et al. Structural 421 Alterations Driving Castration-Resistant Prostate Cancer Revealed by Linked-Read Genome 422 Sequencing. Cell. 2018;174(2):433-47 e19. 423 7. Robinson D, Van Allen EM, Wu YM, Schultz N, Lonigro RJ, Mosquera JM, et al. 424 Integrative clinical genomics of advanced prostate cancer. Cell. 2015;161(5):1215-28. 425 8. Taylor BS, Schultz N, Hieronymus H, Gopalan A, Xiao Y, Carver BS, et al. Integrative 426 genomic profiling of human prostate cancer. Cancer Cell. 2010;18(1):11-22. 427 9. Beltran H, Prandi D, Mosquera JM, Benelli M, Puca L, Cyrta J, et al. Divergent clonal 428 evolution of castration-resistant neuroendocrine prostate cancer. Nat Med. 2016;22(3):298-305. 429 10. Massie CE, Mills IG, Lynch AG. The importance of DNA methylation in prostate cancer 430 development. J Steroid Biochem Mol Biol. 2017;166:1-15. 431 11. Costello LC, Franklin RB. A comprehensive review of the role of zinc in normal prostate 432 function and metabolism; and its implications in prostate cancer. Arch Biochem Biophys. 433 2016;611:100-12. 434 12. Audet-Walsh E, Vernier M, Yee T, Laflamme CE, Li S, Chen Y, et al. SREBF1 Activity 435 is Regulated by an AR/mTOR Nuclear Axis in Prostate Cancer. Mol Cancer Res. 2018; 436 16(9):1396-1405.

437 13. Audet-Walsh E, Yee T, Tam IS, Giguere V. Inverse Regulation of DHT Synthesis 438 Enzymes 5alpha-Reductase Types 1 and 2 by the Androgen Receptor in Prostate Cancer. 439 Endocrinology. 2017;158(4):1015-21. 440 14. Audet-Walsh E, Yee T, McGuirk S, Vernier M, Ouellet C, St-Pierre J, et al. Androgen- 441 Dependent Repression of ERRgamma Reprograms Metabolism in Prostate Cancer. Cancer Res. 442 2017;77(2):378-89. 443 15. Audet-Walsh E, Dufour CR, Yee T, Zouanat FZ, Yan M, Kalloghlian G, et al. Nuclear 444 mTOR acts as a transcriptional integrator of the androgen signaling pathway in prostate cancer. 445 Genes Dev. 2017;31(12):1228-42. 446 16. Massie CE, Lynch A, Ramos-Montoya A, Boren J, Stark R, Fazli L, et al. The androgen 447 receptor fuels prostate cancer by regulating central metabolism and biosynthesis. EMBO J. 448 2011;30(13):2719-33. 449 17. Tennakoon JB, Shi Y, Han JJ, Tsouko E, White MA, Burns AR, et al. Androgens regulate 450 prostate cancer cell growth via an AMPK-PGC-1alpha-mediated metabolic switch. Oncogene. 451 2014;33(45):5251-61. 452 18. Popovics P, Frigo DE, Schally AV, Rick FG. Targeting the 5'-AMP-activated protein 453 kinase and related metabolic pathways for the treatment of prostate cancer. Expert Opin Ther 454 Targets. 2015;19(5):617-32. 455 19. Noushmehr H, Weisenberger DJ, Diefes K, Phillips HS, Pujara K, Berman BP, et al. 456 Identification of a CpG island methylator phenotype that defines a distinct subgroup of glioma. 457 Cancer Cell. 2010;17(5):510-22. 458 20. Figueroa ME, Abdel-Wahab O, Lu C, Ward PS, Patel J, Shih A, et al. Leukemic IDH1 459 and IDH2 mutations result in a hypermethylation phenotype, disrupt TET2 function, and impair 460 hematopoietic differentiation. Cancer Cell. 2010;18(6):553-67. 461 21. Abdel-Wahab O, Levine RL. Mutations in epigenetic modifiers in the pathogenesis and 462 therapy of acute myeloid leukemia. Blood. 2013;121(18):3563-72. 463 22. Yen K, Travins J, Wang F, David MD, Artin E, Straley K, et al. AG-221, a First-in-Class 464 Therapy Targeting Acute Myeloid Leukemia Harboring Oncogenic IDH2 Mutations. Cancer 465 Discov. 2017;7(5):478-93.

466 23. Xu W, Yang H, Liu Y, Yang Y, Wang P, Kim SH, et al. Oncometabolite 2- 467 hydroxyglutarate is a competitive inhibitor of alpha-ketoglutarate-dependent dioxygenases. 468 Cancer Cell. 2011;19(1):17-30. 469 24. Cancer Genome Atlas Research N. The Molecular Taxonomy of Primary Prostate Cancer. 470 Cell. 2015;163(4):1011-25. 471 25. Kang MR, Kim MS, Oh JE, Kim YR, Song SY, Seo SI, et al. Mutational analysis of IDH1 472 codon 132 in glioblastomas and other common cancers. Int J Cancer. 2009;125(2):353-5. 473 26. Mauzo SH, Lee M, Petros J, Hunter S, Chang CM, Shu HK, et al. Immunohistochemical 474 demonstration of isocitrate dehydrogenase 1 (IDH1) mutation in a small subset of prostatic 475 carcinomas. Appl Immunohistochem Mol Morphol. 2014;22(4):284-7. 476 27. Ghiam AF, Cairns RA, Thoms J, Dal Pra A, Ahmed O, Meng A, et al. IDH mutation 477 status in prostate cancer. Oncogene. 2012;31(33):3826. 478 28. Hinsch A, Brolund M, Hube-Magg C, Kluth M, Simon R, Moller-Koop C, et al. 479 Immunohistochemically detected IDH1(R132H) mutation is rare and mostly heterogeneous in 480 prostate cancer. World J Urol. 2018;36(6):877-82. 481 29. Gravel SP, Andrzejewski S, Avizonis D, St-Pierre J. Stable isotope tracer analysis in 482 isolated mitochondria from mammalian systems. Metabolites. 2014;4(2):166-83. 483 30. Andrews S. FastQC A Quality Control tool for High Throughput Sequence Data. 2010 484 [Available from: https://www.bioinformatics.babraham.ac.uk/projects/fastqc/. 485 31. Bolger AM, Lohse M, Usadel B. Trimmomatic: a flexible trimmer for Illumina sequence 486 data. Bioinformatics. 2014;30(15):2114-20. 487 32. Bray NL, Pimentel H, Melsted P, Pachter L. Near-optimal probabilistic RNA-seq 488 quantification. Nat Biotechnol. 2016;34(5):525-7. 489 33. Nelson JD, Denisenko O, Bomsztyk K. Protocol for the fast chromatin 490 immunoprecipitation (ChIP) method. Nat Protoc. 2006;1(1):179-85. 491 34. Gao J, Aksoy BA, Dogrusoz U, Dresdner G, Gross B, Sumer SO, et al. Integrative 492 analysis of complex cancer genomics and clinical profiles using the cBioPortal. Sci Signal. 493 2013;6(269):pl1. 494 35. Cerami E, Gao J, Dogrusoz U, Gross BE, Sumer SO, Aksoy BA, et al. The cBio cancer 495 genomics portal: an open platform for exploring multidimensional cancer genomics data. Cancer 496 Discov. 2012;2(5):401-4.

497 36. Uhlen M, Zhang C, Lee S, Sjostedt E, Fagerberg L, Bidkhori G, et al. A pathology atlas of 498 the human cancer transcriptome. Science. 2017;357(6352). 499 37. Barretina J, Caponigro G, Stransky N, Venkatesan K, Margolin AA, Kim S, et al. The 500 Cancer Cell Line Encyclopedia enables predictive modelling of anticancer drug sensitivity. 501 Nature. 2012;483(7391):603-7. 502 38. Varambally S, Yu J, Laxman B, Rhodes DR, Mehra R, Tomlins SA, et al. Integrative 503 genomic and proteomic analysis of prostate cancer reveals signatures of metastatic progression. 504 Cancer Cell. 2005;8(5):393-406. 505 39. Grasso CS, Wu YM, Robinson DR, Cao X, Dhanasekaran SM, Khan AP, et al. The 506 mutational landscape of lethal castration-resistant prostate cancer. Nature. 2012;487(7406):239- 507 43. 508 40. Sun Y, Wang BE, Leong KG, Yue P, Li L, Jhunjhunwala S, et al. Androgen deprivation 509 causes epithelial-mesenchymal transition in the prostate: implications for androgen-deprivation 510 therapy. Cancer Res. 2012;72(2):527-36. 511 41. Tsai CH, Tzeng SF, Chao TK, Tsai CY, Yang YC, Lee MT, et al. Metastatic Progression 512 of Prostate Cancer Is Mediated by Autonomous Binding of Galectin-4-O-Glycan to Cancer Cells. 513 Cancer Res. 2016;76(19):5756-67. 514 42. Tateishi K, Wakimoto H, Cahill DP. IDH1 Mutation and World Health Organization 2016 515 Diagnostic Criteria for Adult Diffuse Gliomas: Advances in Surgical Strategy. Neurosurgery. 516 2017;64(CN_suppl_1):134-8. 517 43. Kumar A, Coleman I, Morrissey C, Zhang X, True LD, Gulati R, et al. Substantial 518 interindividual and limited intraindividual genomic diversity among tumors from men with 519 metastatic prostate cancer. Nat Med. 2016;22(4):369-78. 520 44. Singh AA, Schuurman K, Nevedomskaya E, Stelloo S, Linder S, Droog M, et al. 521 Optimized ChIP-seq method facilitates transcription factor profiling in human tumors. Life Sci 522 Alliance. 2019;2(1):e201800115. 523 45. Aguirre AJ, Meyers RM, Weir BA, Vazquez F, Zhang CZ, Ben-David U, et al. Genomic 524 Copy Number Dictates a Gene-Independent Cell Response to CRISPR/Cas9 Targeting. Cancer 525 Discov. 2016;6(8):914-29. 526 46. Tsherniak A, Vazquez F, Montgomery PG, Weir BA, Kryukov G, Cowley GS, et al. 527 Defining a Cancer Dependency Map. Cell. 2017;170(3):564-76 e16.

528 47. Dupuy F, Tabaries S, Andrzejewski S, Dong Z, Blagih J, Annis MG, et al. PDK1- 529 Dependent Metabolic Reprogramming Dictates Metastatic Potential in Breast Cancer. Cell 530 Metab. 2015;22(4):577-89. 531 48. Kach J, Long TM, Selman P, Tonsing-Carter EY, Bacalao MA, Lastra RR, et al. Selective 532 Glucocorticoid Receptor Modulators (SGRMs) Delay Castrate-Resistant Prostate Cancer Growth. 533 Mol Cancer Ther. 2017;16(8):1680-92. 534 535

536 Figure Legends

537

538 Figure 1. IDH1 is highly expressed in PCa. A) IDH1 expression from RNA-seq data is shown in

539 17 human cancer types from The Cancer Genome Atlas datasets. B) IDH1 expression from the

540 Cancer Cell Line Encyclopedia (CCLE). Box plots showing RNA-seq mRNA expression data

541 from the CCLE are shown, and the dashed lines within a box represents the mean. Cell lines from

542 the same area or system of the body were grouped together and lineages are indicated at the

543 bottom of the graph, with the number of cell lines in parenthesis. The “prostate” group includes

544 the PCa cell lines NCI-H660, VCaP, MDA PCa 2B, DU145, LNCaP, 22rv1, and PC3, as well as

545 the immortalized prostate epithelial cell line PRECLH.

546

547 Figure 2. IDH1 expression is inversely correlated with PCa progression. A) Kaplan-Meier curves

548 of PCa patients from the Taylor et al. dataset (8). Patients were grouped based on their IDH1

549 expression, defined as low (lowest 25%) or high (highest 75%) expression levels. The log-rank

550 test p value is shown. B) IDH1 expression in benign prostate tissues (n=28), localized PCa

551 (n=59), and metastatic castration resistant PCa (n=35) from the Grasso et al. dataset (39). C)

552 IDH1 expression in benign prostate tissues (n=6), localized PCa (n=7), and metastatic PCa (n=6)

553 from the Varambally et al. dataset (38). Note that several probes were quantified for IDH1. D)

554 CRPC tumors from the Grasso dataset were separated based on increased AR expression in

555 CRPC compared to primary tumors (n=29) or decreased AR expression with increased NEPC

556 marker expression (ENO2 and CHGA, n=6). * p < 0.05; ** p < 0.01; *** p < 0.001. Correlation

557 analysis of IDH1 mRNA with AR, ENO2, and CHGA mRNAs in the TCGA (E), the Taylor (F),

558 and the Kumar (G) clinical datasets (8, 43). Both Pearson (P) and Spearman (S) correlation

559 coefficients are indicated in insets with their respective p values in insets.

560

561 Figure 3. IDH isoforms expression pattern in PCa models. A) Relative mRNA expression of IDH

562 genes in several human PCa cells. IDH1 expression levels in LNCaP cells have been set at 1.

563 Results are shown as the mean ± SEM of three to six samples per group. B) Transcripts per

564 million (TPM) values for all 5 IDH isoforms in RNA-seq data from LNCaP cells. Results are

565 shown as the mean ± SEM (n=3). C) TPM values for all 5 IDH isoforms in RNA-seq data from

566 22rv1 cells (n=3). Results are shown as the mean ± SEM (n=3). D) Relative mRNA expression

567 levels of IDH genes in RNA-seq data from 22rv1 cells retrieved from the study by Kach et al.

568 (48). E) Relative mRNA expression levels of IDH genes in xenograft experiments with 22rv1

569 parental cells and with a 22rv1-derived metastatic cell line (n=3 per group). RNA-seq data were

570 retrieved from the study by Tsai et al. (41). * p < 0.05, Student’s T test. F) Relative mRNA

571 expression of IDH genes in PC3 and DU145 cells, normalized at 1 compared to IDH1 expression

572 levels in LNCaP cells from Fig. 3A. Results are shown as the mean ± SEM (n=6). G) TPM values

573 for all 5 IDH isoforms in RNA-seq data from PC3 cells. Results are shown as the mean ± SEM

574 (n=3). H) Total IDH activity in AR-positive and AR-negative PCa cells. Results are shown as

575 the mean ± SEM of 5 independent experiments. ANOVA p value < 0.0001; letters (a, b, c, and d)

576 indicate significantly different groups based on Tukey HSD post-hoc test with p<0.05.

577

578 Figure 4. IDH gene expression regulation by AR in PCa cells. A) Relative mRNA expression of

579 IDH genes in LNCaP cells following 24h of treatment with 10 nM R1881 (right panel). KLK3

580 expression, which encodes the prostate-specific antigen (PSA), is shown as a positive control (left

581 panel). B) Relative mRNA expression of IDH genes in VCaP cells following 24h of treatment

582 with 10 nM R1881 (right panel). KLK3 expression is shown as a positive control (left panel).

583 IDH2 was barely detectable and is not shown. C) Relative mRNA expression of IDH genes in

584 22rv1 cells following 24h of treatment with 10 nM R1881 (right panel). FASN expression is

585 shown as a positive control (left panel). For panels A, B, and C, results are shown as the mean ±

586 SEM of 2 to 4 independent experiments performed with triplicate biological replicates. D) IDH

587 gene expression following castration in human PCa xenografts of LuCaP35 (n=5/group). Log-

588 transformed data are shown normalized to controls (sham surgery). E) Analysis of AR

589 recruitment to IDH1 following treatment with R1881. The UCSC genome browser view is shown

590 for each gene as well as AR ChIP-seq signal intensity for control and R1881-treated LNCaP

591 cells. The AR fold-enrichment in LNCaP and 22rv1 cells in ChIP-qPCR validation experiments

592 are indicated below the genome browser views (the averages of three independent experiments

593 are shown). Insets indicate ChIP-qPCR validation of AR recruitment at the -12.5kb enhancer of

594 the KLK3 gene. * p < 0.05; ** p < 0.01; *** p < 0.001, Student’s T test.

595

596 Figure 5. AR activation promotes IDH1 activity in PCa cells. A) Total IDH activity was

597 measured in LNCaP cells following 48h treatment with R1881. Fold changes are indicated.

598 Results are shown as the mean ± SEM of 3 independent experiments performed in duplicate or

599 triplicate biological replicates (n=8 per group). B) Total IDH activity was measured in VCaP

600 cells following 48h treatment with R1881. Fold changes are indicated. Results are shown as the

601 mean ± SEM of 2 independent experiments performed in triplicate biological replicates (n=6 per

602 group). C) mRNA expression of IDH genes in LNCaP cells following siRNA-mediated

603 knockdown of IDH1. Cells were first transfected with a siRNA pool targeting IDH1. 48h later,

604 fresh media with and without R1881 was added and cells were harvested after 48h. Total IDH

605 activity in LNCaP (D) following knockdown of IDH1 and treatment with R1881 as described in

606 (C). E) mRNA expression of IDH genes in VCaP cells following siRNA-mediated knockdown of

607 IDH1. Cells were first transfected with a siRNA pool targeting IDH1. 48h later, fresh media with

608 and without R1881 was added and cells were harvested after 48h. F) Total IDH activity in VCaP

609 following knockdown of IDH1 and treatment with R1881 as described in (E). Results from C-F

610 are shown as the mean ± SEM of two independent experiments performed in triplicate biological

611 replicates (n=6/group). * p < 0.05; ** p < 0.01; *** p < 0.001, Student’s T test.

612

613 Figure 6. AR promotes a metabolic reprogramming favoring a cytoplasmic/mitochondrial IDH

614 activity. A) Relative cell number of LNCaP cells transfected with siC or siIDH1 following 72

615 hours in complete media (with androgens). B) Relative cell number of LNCaP cells transfected

616 with siC or siIDH1 following 72 hours without androgens. C) Cytoplasmic and mitochondrial

617 IDH activity was measured in LNCaP cells following 48h treatment with R1881. Results are

618 shown as the mean ± SEM of two independent experiments performed in triplicate biological

619 replicates (n=6/condition). D) Cytoplasmic and mitochondrial IDH activity in LNCaP cells

620 following knockdown of IDH1 and 48h treatment with R1881. Results are shown as the mean ±

621 SEM of two independent experiments performed in triplicate biological replicates per experiment

622 for enzymatic studies and with 8 biological replicates per experiment for proliferation studies. * p

623 < 0.05; ** p < 0.01; *** p < 0.001, Student’s T test.

624

Reprogramming of isocitrate dehydrogenases expression and activity by the androgen receptor in prostate cancer

Kevin Gonthier, Raghavendra Tejo Karthik Poluri, Cindy Weidmann, et al.

Mol Cancer Res Published OnlineFirst May 8, 2019.

Updated version Access the most recent version of this article at: doi:10.1158/1541-7786.MCR-19-0020

Supplementary Access the most recent supplemental material at: Material http://mcr.aacrjournals.org/content/suppl/2021/03/03/1541-7786.MCR-19-0020.DC1

Author Author manuscripts have been peer reviewed and accepted for publication but have not yet been Manuscript 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://mcr.aacrjournals.org/content/early/2019/05/08/1541-7786.MCR-19-0020. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.