PCR Analysis of Androgen-Regulated Genes in Human Lncap Prostate Cancer Cells

Home , EAF2, HEBP2, Medroxyprogesterone

Oncogene (2009) 28, 2051–2063 & 2009 Macmillan Publishers Limited All rights reserved 0950-9232/09 $32.00 www.nature.com/onc ONCOGENOMICS Microarray coupled to quantitative RT–PCR analysis of androgen-regulated genes in human LNCaP prostate cancer cells

S Ngan, EA Stronach, A Photiou, J Waxman, S Ali and L Buluwela

Department of Oncology, Imperial College London, London, UK

The androgen receptor (AR) mediates the growth- 2006). The importance of AR in male development is stimulatory effects of androgens in prostate cancer cells. shown by the androgen insensitivity syndromes char- Identification of androgen-regulated genes in prostate acterized by mutations in the AR gene (Gottlieb et al., cancer cells is therefore of considerable importance for 2004). The prostate is a prototypical androgen-depen- defining the mechanisms of prostate-cancer development dent organ (Cunha et al., 1987; Davies and Eaton, 1991) and progression. Although several studies have used and prostate cancer, which has become the most microarrays to identify AR-regulated genes in prostate commonly diagnosed cancer in males in the western cancer cell lines and in prostate tumours, we present here world and is the second leading cause of male cancer the results of gene expression microarray profiling of the death, is androgen-dependent for its growth (Carter and androgen-responsive LNCaP prostate-cancer cell line Coffey, 1990; McConnell, 1991). Therefore, treatment is treated withR1881 for theidentification of androgen- directed at inhibiting prostate cancer growth by regulated genes. We show that the expression of 319 genes suppressing the action of the endogenous androgen or is stimulated by 24 hafter R1881 addition, witha similar its production. Standard treatment involves surgical or number (300) of genes being significantly repressed. pharmacological orchidectomy or inhibition of AR Expression of the upregulated genes, as well as of 60 of activity using anti-androgens that compete with andro- the most robustly downregulated genes, was carried out gen for binding to AR. Following initial response, using quantitative RT–PCR (Q-RT–PCR) over a time- however, almost all tumours eventually progress, as course of R1881 treatment from 0 to 72 h. Q-RT–PCR reflected by the growth of androgen-independent cells was also carried out following treatment withotherAR and the development of hormone-refractory disease. agonists (dihydrotestosterone, estradiol and medroxypro- The continued involvement of the AR in resistant gesterone) and antagonists (cyproterone acetate, hydro- disease is evident from its continued expression in a xyflutamide and bicalutamide). This study provides a large proportion of resistant cases, as well as from the comprehensive analysis of androgen-regulated gene ex- detection of AR gene amplification and activating pression in the LNCaP prostate cancer cell line, and mutations in the AR gene and/or activation of AR identifies a number of androgen-regulated genes, not through crosstalk with other signalling pathways described previously, as candidates for mediating andro- (Feldman and Feldman, 2001; Taplin and Balk, 2004; gen responses in prostate cancer cells. Agoulnik and Weigel, 2006). Oncogene (2009) 28, 2051–2063; doi:10.1038/onc.2009.68; Androgen binding by the AR leads to its recruitment published online 13 April 2009 to specific gene promoters and consequent regulation of gene expression (Shang et al., 2002; Dehm and Tindall, Keywords: prostate cancer; LNCaP cells; androgens; 2006). Gene-expression microarray analysis allows the anti-androgens; real-time RT–PCR; microarrays global interrogation of the complete genome for the determination of changes in gene expression, thereby permitting the identification of gene networks that may be important in mediating AR-regulated prostate-cancer ONCOGENOMICS Introduction cell growth, thus providing a clearer understanding of the AR function in prostate cancer, and identifying new Normal male development and growth requires the prognostic markers and therapeutic targets. In order to action of androgens. These hormones function by identify androgen-responsive genes in prostate cancer activating the androgen receptor (AR), a member of cells, we used the LNCaP cell line as it expresses AR, the ligand-activated nuclear receptor superfamily of shows androgen-regulated expression of androgen- transcription factors (Brinkmann et al., 1999; Lu et al., responsive genes, such as the prostate cancer biomarker PSA, grows in an androgen-regulated manner in cell culture and forms androgen-dependent tumours in Correspondence: Drs S Ali and L Buluwela, Department of Oncology, xenograft models (Sobel and Sadar, 2005). Earlier Imperial College London, Du Cane Road, London W1 0NN, UK. microarray studies for profiling androgen-regulated E-mail: [email protected] or [email protected] Received 4 August 2008; revised 25January 2009; accepted 4 March genes in prostate cancer cell lines have been carried 2009; published online 13 April 2009 out using the synthetic androgen R1881 or dihydro- Androgen-regulated gene expression in LNCaP cells S Ngan et al 2052 testosterone (DHT), but few studies have carried out and 300 annotated genes whose expression is up- and a detailed investigation of AR-regulated gene downregulated by R1881, respectively (Supplementary expression, in response to anti-androgens (for a review Table 1). see Dehm and Tindall, 2006). In addition, AR activity can be stimulated by oestrogen (E2) and progesterone, Functional categories and pathways for genes regulated whereas the anti-androgen cyproterone acetate (CPA) by R1881 is a partial agonist for AR (Doesburg et al., 1997). As described above, 70% of the upregulated genes and Finally, some AR mutations, such as the AR-T877A 78% of the downregulated genes encode proteins with a mutation in LNCaP cells, increase the agonist activity of known or an inferred function. These gene lists were some of these weak AR agonists, as well as the anti- analysed for gene ontology (GO) within the ‘molecular androgens CPA and hydroxyflutamide (OHF) (Steketee function’ principal group (http://www.geneontology. et al., 2002). However, detailed analysis of the expres- org/). In order to maximize information from this sion of androgen-regulated genes in response to the analysis, gene ontology classifications for subcategories weak agonists and AR antagonists !been confined, by within the two largest categories; namely ‘binding’ and and large, to reporter gene studies (Brooke et al., 2008). ‘catalytic activity’, were also determined. The collective To gain new insight into the regulation of gene results from this ontology analysis show that 38 gene expression by AR, we carried out microarray profiling in ontology classifications are represented, which, by LNCaP cells treated with R1881 and Q-RT–PCR proportion, show a similar distribution in the R1881 analysis for the genes whose expression is stimulated upregulated and downregulated gene sets (Figure 1a). and for a proportion of the genes whose expression is The exception to this is the signal transducer activity inhibited by R1881, over a time course of R1881 grouping (GO: 4871), which functionally represents a treatment, and after the treatment of LNCaP cells with higher proportion of R1881 downregulated genes (4.6% other agonists and anti-androgens. representation in the upregulated gene set, compared with 9.3% in the downregulated set). The R1881 upregulated and downregulated gene lists were further combined to generate an R1881 regulated Results and discussion gene set and were analysed by functional annotation, using web-based tools provided by the Database for Identification of genes whose expression is regulated Annotation, Visualization, and Integrated Discovery by R1881 (DAVID) resource (http://david.abcc.ncifcrf.gov) (Dennis To perform gene-expression microarray analysis of et al., 2003; Huang da et al., 2007). This resulted in the prostate-cancer cells treated with the synthetic androgen identification of 64 functional annotation clusters with R1881, RNA prepared from three bio-replicate cultures enrichment scores in the range 0.01–2.43, with three of LNCaP cells treated with R1881 for periods of time functional pathway clusters having enrichment scores ranging from 4 to 72 h was evaluated for the R1881 >1 (Figure 1b). stimulation of expression of four well-characterized The most significant functional-pathway cluster iden- androgen-regulated genes (PSA, TMPRSS2, NDRG1 tified within this gene list was found to be one in which and GREB1). Expression of these genes was stimulated major determinants of the transforming growth factor within 4 h after R1881 addition and levels continued to (TGF)-b signalling pathway, including SMAD1, increase up to 24 h, after which time expression was SMAD3, SMAD6 and SMAD7, were downregulated, reduced, but remained high up to 72 h (Supplementary whereas ID3, a candidate gene for metastatic prostate Figure 1). On the basis of these results, RNA from each cancer (Burmester et al., 2004; Yuen et al., 2006), was of the three replicates, for 24-h time points, was chosen upregulated. Although it has been reported that TGF-b for hybridization to the Applied Biosystems (ABI, signalling is downregulated in prostate cancer, this has Foster City, CA, USA) human genome survey micro- largely been attributed to transcriptional regulation of array V2.0, which has probe sets for 29 098 genes. Raw TGF-b expression by AR (Qi et al., 2008), or through data were quality-assessed and filtered according to the direct interaction of AR with SMADs, leading to recommendations in the ABI1700 data analysis user inhibition of SMAD DNA-binding activity (Chipuk guide and the filtered data were ‘vsn normalized’ (Huber et al., 2002). The microarray analysis carried out here et al., 2002). Differential expression was assessed using suggests that in addition to these mechanisms, androgen linear models and empirical Bayes algorithms as treatment causes a programmed inhibition of TGF-b described (Smyth, 2004). This analysis defined 452 signalling, primarily through the downregulation of key probes whose expression was stimulated by androgen signal transduction molecules, including SMADs, and by two-fold or greater, with positive lod scores (B further highlights the importance of this pathway to values) for differential expression. We also identified 382 androgen-mediated prostate-cancer cell growth and probes whose levels were inhibited by androgens by the survival (Guo et al., 1997; Guo and Kyprianou, 1998, same margins. Of these, 133 probes from the upregu- 1999). lated set and 82 probes from the downregulated set were Fatty-acid metabolism was also highlighted as an removed from further evaluation as they represented important pathway, with 16 genes being identified in genes that were unassigned, or represented multiple this cluster. Thirteen of these genes including those for probes for the same gene. This resulted in a list of 319 the peroxisomal enzymes acetyl-CoA acyltransferase 1

Oncogene Androgen-regulated gene expression in LNCaP cells S Ngan et al 2053

Enrichment Genes in Pathway Score Cluster TGFβ signalling 2.43 8 Fatty Acid Metabolism/PPAR Signaling 1.91 16 Biosynthesis of steroids 1.1 5

Keyword/ Feature Protein phosphatase 1.86 10 Endoplasmic Reticulum-golgi transport 1.61 6 Ras-association 1.6 6 Glycosyltransferase 1.51 19 Frizzled CRD region 1.47 4 Phosphatidic acid phosphatase 1.27 3 Transferase 1.25 97 Fibronectin, type III 1.0 13 Figure 1 Functional categorization of genes regulated by R1881. (a) The genes showing >2-fold regulation in LNCaP following the addition of R1881 for 24 h were analysed for gene ontology (GO) (http://www.geneontology.org/). The number of genes in each molecular function group is shown as a percentage of all genes whose expression was stimulated by R1881, whereas genes whose expression was repressed by R1881 are shown as a percentage of genes in the GO category, relative to all the repressed genes. (b) Functional analysis of all R1881-regulated genes was carried out using the database for annotation, visualization, and integrated discovery (DAVID) resource (http://david.abcc.ncifcrf.gov/). The functional annotation clusters are shown as Pathways and Keywords/Features that had enrichment scores of X1.0, together with the number of genes represented in each cluster. CRD, Cysteine rich domain; PPAR, peroxisome proliferator-activated receptors.

(ACAA1) and fatty acyl-CoA oxidase 3 (ACOX3), were The third most significant pathway involved genes upregulated by R1881, whereas the gene of acetyl-CoA required for steroid biosynthesis. Androgen regulation acyltransferase 2 (ACAA2), a mitochondrial enzyme, of steroid-biosynthesis genes is mediated through an was found to be one of the three downregulated genes in indirect mechanism involving the activation of sterol this set. This suggests that androgens preferably regulatory-element binding proteins SREBP-1 and stimulate peroxisomal over mitochondrial branched- SREBP-2 (Heemers et al., 2006). Further, kruppel-like fatty acid b-oxidation, this being consistent with earlier factor 5( KLF5) has recently been identified as an observations, which show that peroxisomal branched- androgen-regulated gene that acts as a positive regulator fatty acid b-oxidation is upregulated in prostate cancer of SREBP-1 function (Lee et al., 2009). In agreement (Zha et al., 2005), with this pathway being active in with these reports, expression of SREBP-1 and SREBP- regulating prostate cancer cell growth (Zha et al., 2003). 2 was not androgen-regulated in our microarray The importance of peroxisomal function as being a new analysis, whereas KLF5 expression was stimulated androgen-regulated target is further emphasized by our 3.1-fold by R1881. finding that 10 of the 16 genes in the fatty acid- An additional eight groupings with enrichment metabolism functional cluster are assigned to this scores X1.0 were highlighted by the DAVID analysis. organelle. These groupings feature genes with similar functions,

Oncogene Androgen-regulated gene expression in LNCaP cells S Ngan et al 2054 rather than diverse genes associated with distinct et al., 2008). In order to determine whether the different pathways, and include several enzyme groups, such AR agonists act in a similar manner to regulate the as phosphatases, transferases and proteins associated expression of androgen-responsive genes, a Q-RT–PCR with secretion (endoplasmic reticulum and Golgi was carried out for the 347 androgen-regulated genes transport). described in the section above, using total RNA prepared in triplicate from LNCaP cells following addition of DHT, E2 or MPA for 24 h. In addition, Quantitative RT–PCR for determining time-course LNCaP cells were treated for 24 h with the selective of gene regulation by R1881 androgen-receptor modulators (SARMs) CPA and Real-time Q-RT–PCR was carried out for the R1881 OHF, which show mixed agonist/antagonist activities, genes identified above, using the low-density array and with the anti-androgen bicalutamide (BIC), also micro-fluidic cards from ABI, which allow simultaneous known as casodex, which has little or no agonist activity Q-RT–PCR for up to 384 genes. Q-RT–PCR primers for AR (Brooke et al., 2008). were not available for 10 of the 319 upregulated genes, When the Q-RT–PCR data were analysed relative to hence low-density array cards were designed, containing the no-ligand control, most of the genes whose expres- 309 of the upregulated genes, as well as 62 genes sion was stimulated by R1881 were also stimulated by showing the highest significance scores for downregula- DHT, although the level of regulation by DHT was tion by R1881, together with control genes (GAPDH, generally lower than that for R1881 (Figure 3a). RPLPO). Q-RT–PCR was carried out for three replicate Interestingly, a small number of genes whose expression RNA preparations from LNCaP cells treated with was stimulated by R1881 were repressed by DHT. The R1881 for 0, 4, 8, 16, 24, 48 and 72 h. Fold expression DHT-repressed genes were also repressed by the other relative to the control (0 h) was determined for each AR activators, E2 and MPA, as well as by the anti- gene, and cluster analysis was used to determine androgens. E2, MPA, CPA and OHF also stimulated patterns of R1881 regulation. No expression was the expression of most R1881 upregulated genes. detected for 22 of the upregulated genes and 2 of the LNCaP cells express oestrogen receptor b (ER-b), but downregulated genes at the 0-h time point. These genes not ER-a (Lau et al., 2000), although other studies were therefore excluded from further analysis, as it was indicate that ER-a is also expressed in LNCaP cells not possible to make a measurement of their relative (Takahashi et al., 2007). In addition to direct binding of regulation by ligand. As seen in the heat maps for the E2 by AR, it is therefore possible that some of the AR- up and downregulated genes (Figures 2a–c), 287 regulated genes analysed here are also directly regulated up-regulated genes could be clustered into seven groups, by ER, for example GREB1 (Rae et al. 2006). An labelled U1–U7, and as three clusters in the down- alternative mechanism by which ER regulates the regulated gene set (see Supplementary Table 2 for names expression of AR-regulated genes through interaction of the genes in each cluster). Plotting the average profile with AR has also been described (Arnold et al., 2007; for each of these clusters indicates that the expression of Takahashi et al., 2007). In addition to their regulation the genes in cluster U5is rapidly stimulated by R1881, by direct E2 binding to AR, it is possible therefore that reaching a peak at 24 h, after which time expression the E2 regulation of androgen-responsive genes involves generally plateaus (Figure 2d). A similar profile is direct or indirect binding of E2–ER to promoters of obtained for clusters U3 and U6, except that these androgen-responsive genes. The progestin MPA is also clusters represent genes whose average expression able to stimulate AR activity, and MPA treatment reaches a considerably lower maximum than those resulted in the stimulation of many of the R1881- observed for U5. U2 is interesting as expression is regulated genes. As progesterone receptor (PR) and AR stimulated early, peaks at 24 h and subsequently falls to bind to similar DNA motifs, it is possible that PR near-basal levels by 48 h. Finally, clusters U1, U4 and directly regulates the expression of AR-regulated genes. U7 appear to represent genes whose expression increases Indeed, PR expression in LNCaP cells has been reported slowly over the time course; these genes could therefore (Lau et al., 2000); hence MPA regulation of our gene set represent indirectly androgen-regulated genes. There could be because of MPA stimulation of PR, as well as was little to distinguish the 60 downregulated genes, as of AR. Interestingly, the SARMs (CPA, OHF) stimu- most of the genes follow a similar time course, with lated the expression of a larger proportion of the repression being seen within 4–8 h, and near maximal upregulated genes than E2 or MPA. Expression of the repression being reached 16 h after R1881 addition R1881 downregulated genes was also repressed by the (Figure 2e). This suggests that many of these genes other ligands (Figure 3b). Cluster analysis highlighted represent direct targets for AR. five groups (labelled L1–L5; see Supplementary Table 3 for the genes in each cluster). No clear breakdown could Quantitative RT–PCR for determining androgen agonist be achieved for the R1881 downregulated genes. The and antagonist regulation of the identified genes cluster analysis divided the R1881 upregulated genes Earlier gene profiling studies in LNCaP cells have into two broad groups: those genes whose expression is focused on changes in gene expression upon the addition strongly stimulated by DHT (cluster L5), and those of R1881 or DHT (Dehm and Tindall, 2006). However, genes more which are weakly activated by DHT (L1– the AR is also activated by E2, as well as by progestins L4). Cluster L5contained the majority of the genes such as medroxyprogesterone acetate (MPA) (Brooke showing rapid stimulation by R1881, with 22/40 (55%),

Oncogene Androgen-regulated gene expression in LNCaP cells S Ngan et al 2055

Figure 2 Analysis of quantitative RT–PCR for 368 androgen-regulated genes over the time course of R1881 treatment. (a) The results of quantitative RT–PCR (Q-RT–PCR) for 287 genes whose expression was stimulated >2-fold at 24 h following R1881 addition in the microarray analysis. Q-RT–PCR was carried out using TaqMan gene expression assays for three separate RNA preparations from LNCaP cells treated with 1 nM R1881 for 0, 4, 16, 24, 48 or 72 h. Fold changes in expression were determined relative to the no-ligand (0 h R1881) control. The genes were clustered using a standard correlation algorithm (GeneSpring Software, Agilent Technologies, Stockport, UK). Upregulated genes are shown in red, downregulated genes in green, non-changing genes in black, and genes for which there was no expression after 40 cycles are shown in grey. (b) The cluster analysis for Q–RT–PCR using Taqman gene expression assays for 60 genes identiﬁed from the microarray analysis as being downregulated by R1881 by >2-fold. (c) Colour scale corresponding to fold change in gene expression. (d, e) Average fold expression for all of the genes in each of the major clusters labelled in (a) and (b) is shown in the form of line graphs.

8/13 (61%) and 41/56 (73%) of the genes clustered in Androgen and anti-androgen regulation of selected genes groups U2, U3 and U5, respectively, indicating that the The genes identiﬁed in the microarray analysis included genes showing the most potent R1881 regulation are genes identiﬁed earlier as being androgen-regulated. also activated by other agonists and by SARMs. Among These included the KLK3 or PSA gene, a commonly the genes weakly activated by DHT, genes in clusters L1 used biomarker for prostate cancer, whose expression and L3 were stimulated by SARMs but not by E2 or was maximally stimulated 12-fold at 24 h with R1881 MPA, whereas groups L2 and L4 contained genes whose (Figure 4). E2, MPA, CPA and OHF also stimulated expression was not stimulated by E2, MPA, or by CPA KLK3 expression (Figure 5). KLK3 clustered in group and OHF. U5, which contains the genes reaching the greatest

Oncogene Androgen-regulated gene expression in LNCaP cells S Ngan et al 2056

Figure 3 Regulation of genes identiﬁed through microarray analysis by androgen-regulated (AR) agonists and antagonists. (a) Cluster analysis of 287 R1881-responsive genes after Q-RT–PCR of three independent RNA samples prepared from LNCaP cells 24 h following the addition of R1881, dihydrotestosterone (DHT), cyproterone acetate (CPA), hydroxyﬂutamide (OHF), 17b-oestradiol (E2), medroxyprogesterone (MPA) and bicalutamide. Fold changes in expression were determined relative to the no- ligand (0 h R1881) control. The clusters L1–L5are also shown. ( b) The cluster analysis for Q–RT–PCR data for the RNA samples described in (a) for the 60 genes showing two-fold or greater repression at 24 h after R1881 addition in the microarray analysis. (c) Colour scale corresponding to fold change in gene expression.

Oncogene Androgen-regulated gene expression in LNCaP cells S Ngan et al 2057 14 50 50 12 40 40 10 8 30 30 6 20 20 4 2 KLK3 10 KLK2 10 NDRG1 0 0 0 0 1020304050607080 0 1020304050607080 0 1020304050607080 Time (Hours) Time (Hours) Time (Hours)

40 40 50 30 30 40 30 20 20 20 10 10 TMPRSS2 FKBP5 10 MAF 0 0 0 0 1020304050607080 0 1020304050607080 0 1020304050607080 Time (Hours) Time (Hours) Time (Hours)

10 25 25 8 20 20 6 15 15 4 10 10 2 SPDEF 5 CXCR4 5 hCAP-D3 0 0 0 0 1020304050607080 0 1020304050607080 0 1020304050607080 Time (Hours) Time (Hours) Time (Hours)

300 14 7 250 12 6 10 5 200 8 4 150 6 3 100 4 2 50 HPGD 2 NR4A1 1 GADD45G 0 0 0 0 1020304050607080 0 1020304050607080 0 1020304050607080 Time (Hours) Time (Hours) Time (Hours) Figure 4 Time course of expression of selected genes in gene cluster U5upon R1881 treatment. The expression of selected genes that are grouped in gene cluster U5is shown. The line graphs show the means and standard errors for Q-RT–PCR carried out with three independent RNA samples prepared for each time point. stimulation in the R1881 time-course, and includes suppressor gene in colorectal and breast cancer, with many androgen-regulated genes described earlier, epigenetic silencing of the HPGD gene being frequently including KLK2, NDRG1, TMPRSS2, FKBP5 and the observed (Myung et al., 2006; Wolf et al., 2006). hCAP- proto-oncogene c-MAF (Nelson et al., 2002). These D3 is a non-SMC subunit of the condensing II complex genes cluster to group L5on ligand Q-RT–PCR required for mitotic chromosome assembly and segrega- analysis, with their expression being activated by tion (Ono et al., 2003). It was ranked ninth in terms of DHT, CPA, OHF, E2 and MPA. Among the genes in signiﬁcance of difference in expression by microarray cluster U5were the prostate-derived ets factor SPDEF, analysis; its expression was stimulated 19-fold by R1881 which interacts with AR and with the androgen- at 24 h and was increased 1.3–3.9 fold with other regulated homeobox gene NKX-3.1 and regulates the ligands. Similarly, expression of the CXCR4 chemokine expression of the KLK3 gene (Oettgen et al., 2000; Chen receptor, which is believed to play a role in cell motility et al., 2002). NKX-3.1 was also androgen-regulated in and metastasis, and which has been implicated in many our microarray, and by Q-RT–PCR its expression cancer types (Busillo and Benovic, 2007), including peaked at 24 h, then fell by 48 h (Figures 4 and 5). 15- prostate cancer, was stimulated 19-fold by R1881 at Hydroxy-prostaglandin dehydrogenase (HPGD) was 24 h, and was stimulated 1.9–2.9-fold by DHT, CPA, the most highly androgen-responsive gene in the OHF and MPA, but was not stimulated by E2. NR4A1 microarray analysis. Q-RT–PCR showed that HPGD is an orphan nuclear receptor that regulates apoptosis in expression was stimulated 151-fold 24 h after R1881 many cancer cell types (Ke et al., 2004; Chintharlapalli addition and 271-fold at 72 h after treatment. DHT (53- et al., 2005) and whose expression is downregulated in fold), E2 (39-fold), MPA (52-fold), CPA (63-fold) and hormone-refractory prostate cancer (Tamura et al., OHF (63-fold) stimulated HPGD expression strongly. 2007). NR4A1 expression was stimulated 12-fold by Prostaglandins are growth stimulatory for many tumour R1881, with 1.9-3-fold stimulation by CPA, OHF, E2 types, with the prostaglandin-generating enzyme COX-2 and MPA. Another gene in cluster U5that inhibits cell being aberrantly expressed in many tumours. HPGD growth and induces apoptosis, its expression being converts prostaglandin E2 to biologically inactive stimulated 5.6-fold by R1881 at 24 h, was GADD45G. 15-ketoprostaglandins, hence acting in opposition to GADD45G is a member of the DNA damage-inducible COX-2, and HPGD gene has been proposed as a tumour gene family that is frequently silenced epigenetically in

Oncogene Androgen-regulated gene expression in LNCaP cells S Ngan et al 2058 14 50 50 12 KLK3 40 KLK2 40 NDRG1 10 8 30 30 6 20 20 4 10 2 10 0 0 0 E2 E2 E2 NL NL NL BIC BIC BIC DHT DHT DHT CPA CPA CPA MPA MPA MPA OH-F OH-F OH-F BIC + BIC + BIC + R1881 R1881 R1881 R1881 R1881 R1881 Ligands Ligands Ligands 40 40 40 35 TMPRSS2 35 FKBP5 35 MAF 30 30 30 25 25 25 20 20 20 15 15 15 10 10 10 5 5 5 0 0 0 E2 E2 E2 NL NL NL BIC BIC BIC DHT DHT DHT CPA CPA CPA MPA MPA MPA OH-F OH-F OH-F BIC + BIC + BIC + R1881 R1881 R1881 R1881 R1881 R1881 Ligands Ligands Ligands 180 25 10 160 140 HPGD 20 hCAP-D3 8 SPDEF 120 100 15 6 80 10 4 60 40 5 2 20 0 0 0 E2 E2 E2 NL NL NL BIC BIC BIC DHT DHT DHT CPA CPA CPA MPA MPA MPA OH-F OH-F OH-F BIC + BIC + BIC + R1881 R1881 R1881 R1881 R1881 R1881 Ligands Ligands Ligands 25 14 7 6 20 CXCR4 12 NR4A1 GADD45G 10 5 15 8 4 10 6 3 4 2 5 2 1 0 0 0 E2 E2 E2 NL NL NL BIC BIC BIC DHT DHT DHT CPA CPA CPA MPA MPA MPA OH-F OH-F OH-F BIC + BIC + BIC + R1881 R1881 R1881 R1881 R1881 R1881 Ligands Ligands Ligands 12 12 5 10 GREB1 10 NKX3-1 4 TWIST1 8 8 3 6 6 2 4 4 2 2 1 0 0 0 E2 E2 E2 NL NL NL BIC BIC BIC DHT DHT DHT CPA CPA CPA MPA MPA MPA OH-F OH-F OH-F BIC + BIC + BIC + R1881 R1881 R1881 R1881 R1881 R1881 Ligands Ligands Ligands 12 14 4 10 ELL2 12 EAF2 3 CREB3L2 8 10 3 8 2 6 6 2 4 4 1 2 2 1 0 0 0 E2 E2 E2 NL NL NL BIC BIC BIC DHT DHT DHT CPA CPA CPA MPA MPA MPA OH-F OH-F OH-F BIC + BIC + BIC + R1881 R1881 R1881 R1881 R1881 R1881 Ligands Ligands Ligands 30 9 6 SGK 8 PRKCA 5 IGF1 25 7 20 6 4 5 3 15 4 10 3 2 2 1 5 1 0 0 0 E2 E2 E2 NL NL NL BIC BIC BIC DHT DHT DHT CPA CPA CPA MPA MPA MPA OH-F OH-F OH-F BIC + BIC + BIC + R1881 R1881 R1881 R1881 R1881 R1881 Ligands Ligands Ligands Figure 5 Ligand regulation of the expression of selected genes after 24 h treatment. Fold expression of genes in Figures 4 and 6 in response to R1881, dihydrotestosterone (DHT), cyproterone acetate (CPA), hydroxyﬂutamide (OHF), 17b-estradiol (E2), medroxyprogesterone (MPA) and bicalutamide is shown. The bar graphs show the means of Q-RT–PCR for three independent RNA samples prepared following the treatment of LNCaP cells with ligands for 24 h.

Oncogene Androgen-regulated gene expression in LNCaP cells S Ngan et al 2059 12 12 5 10 GREB1 10 NKX3-1 4 TWIST1 8 8 3 6 6 2 4 4 2 2 1 0 0 0 0 1020304050607080 0 1020304050607080 0 1020304050607080 Time (Hours) Time (Hours) Time (Hours)

12 14 3.5 10 ELL2 12 EAF2 3 CREB3L2 8 10 2.5 8 2 6 6 1.5 4 4 1 2 2 0.5 0 0 0 0 1020304050607080 0 1020304050607080 0 1020304050607080 Time (Hours) Time (Hours) Time (Hours)

30 10 6 25 SGK 8 PRKCA 5 IGF1 20 4 6 15 3 4 10 2 5 2 1 0 0 0 0 1020304050607080 0 1020304050607080 0 1020304050607080 Time (Hours) Time (Hours) Time (Hours) Figure 6 Time-course of expression of selected genes in gene cluster U2 upon R1881 treatment. The expression of selected genes that are grouped in gene cluster U2 is shown over the time course of R1881 treatment from 0 to 72 h. As for Figure 4, the line graphs show the means and standard errors for Q-RT–PCR carried out with three independent RNA samples prepared for each time point.

tumour cell lines (Ying et al., 2005). Although its for prostate cancer-cell growth, is itself androgen- expression was stimulated by R1881, GADD45G ex- regulated (Shanmugam et al., 2007), with rapid stimula- pression was inhibited 2–5-fold by the other ligands, tion in its expression, which peaked at 24 h (25-fold) and including DHT (5). decreased to near basal levels by 48 h. Interestingly, The R1881 time-course cluster U2 (Figure 2), which DHT did not stimulate serum/glucocorticoid-induced represents genes whose expression is stimulated early protein kinase expression at 24 h, whereas its expression after R1881 treatment and then falls by 48 h, included was stimulated 2-, 5-, 1.8- and 3.9-fold by CPA, OHF, the androgen-regulated genes GREB1 (Rae et al., 2006) E2 and MPA, respectively. The protein kinase C-a and NKX3-1 described earlier (Figure 6). The NKX3-1 subunit was similarly upregulated within 8 h of R1881 homeobox gene is frequently deleted in prostate cancer addition, with its expression peaking at 24 h (7.4-fold) and knockout mice are susceptible to prostatic intrae- and decreasing to near-basal levels at 48 h. The other pithelial neoplasia (Abdulkadir et al., 2002; Kim et al., ligands also stimulated PRKCA expression 2.1–3.9-fold, 2002), indicating that NKX3-1 is a tumour suppressor relative to the no-ligand control. Insulin-like growth gene in the prostate. NKX3-1 expression was stimulated factor 1 was another signal transduction gene whose 10-fold by R1881 and 3–7-fold by the other ligands expression increased rapidly within 24 h and then fell by (Figure 5). TWIST1, a bHLH transcription factor that is 48 h with 1.4–4.3-fold stimulation by the AR ligands. upregulated in prostate cancer (Kwok et al., 2005), was Hence, the groups of genes showing rapid induction upregulated 1.9–4.3-fold at 24 h by the other ligands. with R1881 identified here following the microarray ELL2 is an RNA polymerase II elongation factor analysis and Q-RT–PCR include a number of genes with (Shilatifard et al., 1997), which interacts with the potential cellular growth-regulatory roles, several of transcription factor EAF2 (ELL associated factor 2) which have already been implicated in prostate cancer (Simone et al., 2003). Interestingly, ELL2 and EAF2 through deregulation of their expression and epigenetic genes show strong co-regulation, clustering together in silencing. The other AR agonists and SARMs also the R1881 time course and in the expression profile in stimulated expression of many of these genes. response to the other AR ligands. EAF2 has also been The anti-androgen BIC is known to inhibit the shown to induce prostate cancer-cell apoptosis in androgen-stimulated growth of LNCaP cells and act as xenograft tumours, and its expression is downregulated an AR antagonist (Veldscholte et al., 1992b; Furr and in prostate tumours (Xiao et al., 2003). CREB3L2 a Tucker, 1996). BIC has been earlier shown to promote cAMP-response element binding protein homologue AR association with cytosolic heat shock protein that is fused to the FUS gene in fibromyxoid sarcomas complexes (Veldscholte et al., 1992a). By contrast, other (Storlazzi et al., 2003), was upregulated by R1881, but studies have found that BIC allows nuclear re-localiza- was not significantly upregulated by other AR ligands tion of AR and recruitment to androgen-regulated gene (Figure 5). Serum/glucocorticoid-induced protein ki- promoters (Masiello et al., 2002; Yoon and Wong, 2006; nase-1, which stimulates AR activity and is important Hodgson et al., 2007). The latter studies have shown

Oncogene Androgen-regulated gene expression in LNCaP cells S Ngan et al 2060 Table 1 Expression proﬁles for R1881-stimulated genes whose expression is not repressed by treatment with BIC alone No R1881 R1881 R1881 R1881 R1881 R1881 R1881 BICDHT CPAFLUT E2 MPA ligand 4h 8h 16 h 24 h 48 h 72 h þ BIC

HPGD 1.00 4.73 16.13 38.89 151.59 218.64 271.18 111.60 1.09 53.02 63.28 63.82 39.21 52.86 NPPC 1.00 34.81 77.51 224.78 307.24 47.23 36.07 185.53 1.85 63.69 79.14 115.07 43.55 57.17 KCNMA1 1.00 1.05 1.50 2.28 8.52 6.96 7.59 8.80 1.02 8.80 7.61 6.15 4.14 4.61 IGSF4D 1.00 0.72 0.87 6.22 11.58 4.41 1.64 11.01 1.12 11.56 5.71 9.46 6.03 4.59 FXYD3 1.00 1.251.49 1.73 3.73 5.128.63 3.20 1.34 3.31 2.73 2.17 1.93 1.86 MUC20 1.00 1.98 3.09 4.7516.42 5.48 2.63 7.77 0.96 4.06 1.98 3.26 1.72 1.17 ANKRD37 1.00 1.01 1.70 4.67 12.53 16.24 18.08 6.60 0.99 2.73 1.81 2.81 1.95 1.51 SAT 1.00 1.16 1.73 2.94 8.35 12.96 18.95 5.54 1.01 2.97 2.18 2.28 1.86 1.78 SOCS2 1.00 1.74 2.61 3.458.76 3.72 4.26 4.49 1.20 2.23 2.11 2.42 1.252.04 UGT2B28 1.00 1.02 1.30 4.28 16.81 57.51 48.21 20.10 1.17 5.34 1.69 6.97 3.59 3.94 HEBP2 1.00 1.19 1.66 1.86 4.29 5.73 5.16 4.21 0.96 2.61 1.77 2.69 2.03 1.70 PHLDB2 1.00 4.95 9.50 31.86 58.76 43.13 55.50 40.80 3.85 10.15 6.96 15.06 11.45 13.72 TPM1 1.00 1.37 1.27 0.90 1.66 1.94 2.14 7.27 6.90 10.41 6.17 6.50 6.74 5.10 ANXA2 1.00 0.82 1.04 1.38 1.90 1.58 1.84 6.72 5.16 7.01 6.46 5.20 4.77 5.41

Abbreviations: BIC, bicalutamide; CPA, cyproterone acetate; DHT, dihydrotestosterone; E2, 17b-oestradiol; FLUT, hydroxyﬂutamide; MPA, medroxyprogesterone. The fold expression of genes relative to the no-ligand control is shown.

that although AR agonists recruit co-activators and expression of these genes was reduced in the presence of co-repressors to AR-regulated gene promoters, in the the other ligands to an extent similar to the reduction presence of BIC co-repressors are recruited but there is a seen with R1881 at 24 h, although the repression lack of co-activator recruitment. In agreement with appeared to be less marked for DHT and CPA these ﬁndings, expression of the majority of R1881- compared with OHF, E2 and MPA. Interestingly, stimulated genes was repressed by co-treatment with however, although co-treatment with BIC and R1881 BIC (see Figures 3 and 5). Exceptions were genes that gave gene repression similar to that seen for R1881 featured in the cluster L5: not only HPGD (1.1-fold alone, the expression of these genes was considerably relative to the no-ligand control) and NPPC (1.9-fold), higher on average in the presence of BIC alone, both of which show strong R1881 stimulation, but also suggesting that BIC acts to reduce the AR activity on the KCNMA1, IGSF4D, FXYD3, MUC20, ANKRD37, repressed AR genes, as well as on AR-activated genes. SAT, SOCS2, UGT2B28 and HEBP2 genes (1.0–1.9- The expression of a number of R1881 downregulated fold relative to the no-ligand control), which are only genes was not repressed, or was stimulated, by BIC moderately upregulated by R1881 (4.3–16-fold). The (Figure 7). These include many of the genes whose most potent BIC stimulation was seen for PHLDB2, expression is either stimulated by R1881 at early time TPM1, ANXA2, respectively stimulated 3.8-, 6.9- and points, before being repressed, or those genes whose 5.2-fold relative to the no-ligand control (Table 1). Two expression is reduced only at later time points. These of these genes PHLDB2 and TPM1, are found in the cell genes include C1orf165, PTPRR, a gene for a protein cytoskeleton, PHLDB2 being a phospholipid binding tyrosine phosphatase, and the cytokine receptor CXCR7 protein implicated in microtubule stabilization, whereas (aka CMKOR1). CXCR7 expression was only reduced TPM1 (tropomyosin 1-a) is an actin-binding protein. at 16 h or more after R1881 treatment and was also Annexin A2 (ANXA2) is a member of the calcium- inhibited by the other ligands, with the exception of dependent phospholipid binding protein family, which BIC, which stimulated CXCR7 1.6-fold. Interestingly, regulate cell growth. PHLDB2 was potently and rapidly C1orf165 expression was stimulated about two-fold by regulated by R1881 (5-fold upregulation in 4 h), and its R1881 within 4 h, with expression falling at 8 h. expression was also stimulated by the other ligands. C1orf165 expression was repressed by the other ligands, However, TPM1 and ANXA2 expression was increased but stimulated 1.6-fold by BIC. Expression of PTPRR only 1.7-and 1.9-fold by R1881 at 24 h. Interestingly, in was also stimulated two-fold by R1881, fell to one-fold addition to showing weak activation by R1881 and by 16 h and was reduced further at 48 h. PTPRR strong stimulation by BIC, TPM1 and ANXA2 were expression was stimulated by the other ligands at 24 h, also strongly upregulated (4.8–10.4-fold) by the other including 2.5-fold by BIC. ligands.

R1881-repressed genes Conclusions As mentioned earlier, most of the 60 R1881 downregulated genes showed reduced expression by at least The proﬁling of androgen-regulated gene expression in 10% within 4 h, with greater than 50% repression by LNCaP cells with the androgen R1881 identiﬁed more 24 h and less than 10% expression relative to the than 300 genes whose expression was upregulated, and a no-ligand control at 48 h after R1881 addition. The similar number whose expression was downregulated.

Oncogene Androgen-regulated gene expression in LNCaP cells S Ngan et al 2061

No R1881 R1881 R1881 R1881 R1881 R1881 R1881 Ligand 4 hrs 8 hrs 16 hrs 24 hrs 48 hrs 72 hrs + BIC BIC DHT CPA OHF E2 MPA UGT2B15 1.00 0.91 0.55 0.15 0.12 0.15 0.12 0.10 1.91 0.32 0.21 0.23 0.25 0.16 TMEM38B 1.00 1.03 0.75 0.17 0.29 0.19 0.23 0.24 1.25 0.44 0.36 0.33 0.33 0.25 SERPINI1 1.00 1.06 0.88 0.43 0.23 0.15 0.15 0.30 1.37 0.52 0.44 0.40 0.57 0.38 CMKOR1 1.00 1.10 1.20 0.25 0.41 0.47 0.47 0.29 1.62 0.75 0.70 0.45 0.56 0.44 ADCY7 1.00 1.01 0.81 0.52 0.43 0.27 0.37 0.49 1.23 0.84 1.24 0.51 0.84 0.55 SI 1.00 0.88 0.72 0.18 0.00 0.00 0.00 0.11 1.64 0.27 0.32 0.17 0.35 0.16 C1orf165 1.00 2.37 0.79 0.28 0.00 0.00 0.03 0.00 2.17 0.27 0.40 0.00 0.59 0.00 PCDH11X;PCDH11Y 1.00 0.66 0.75 0.32 0.00 0.00 0.00 0.00 2.09 0.02 0.31 0.00 0.52 0.00 BCHE 1.00 1.30 1.08 0.42 0.11 0.00 0.07 0.07 1.43 0.32 0.74 0.09 0.61 0.00 PTPRR 1.00 2.05 2.25 1.03 1.04 0.00 0.35 0.96 2.54 2.94 6.62 3.24 2.22 2.63 Figure 7 Ligand regulation of R1881 downregulated genes. (a) The mean expression levels, relative to the no-ligand control as determined by Q-RT–PCR, are shown for the 60 genes whose expression was downregulated by R1881 in the microarray analysis. The error bars represent the standard errors of the mean. (b) Fold expression of selected R1881-repressed genes, relative to the no-ligand control. BIC, bicalutamide; DHT, dihydrotestosterone; CPA, cyproterone acetate; OHF, hydroxyﬂutamide; E2, 17b-oestradiol; MPA, medroxyprogesterone.

The expression profiles of these genes over a timecourse gen targets, whereas the potential importance of TGF-b of R1881 treatment highlight distinct sets of androgen- signalling in androgen-mediated prostate-cancer pro- responsive genes whose expression shows rapid gression is further strengthened by our finding that stimulation, as well as those genes whose expression many genes that participate in TGF-b signalling are is stimulated by R1881 over a longer time period, androgen regulated in LNCaP cells. indicating that the former set comprises genes directly regulated by AR, whereas the latter group may include genes whose expression is indirectly regulated by AR. Materials and methods A comprehensive analysis of the expression of R1881- regulated genes by DHT, but also by other AR agonists Cell culture oestrogen (E2) and progestin (MPA), as well as the well- Cell lines were maintained in RPMI-1640 medium-HEPES characterized and clinically important anti-androgens, modification, supplemented with 10% fetal calf serum CPA, flutamide and bicalutamide. Most of the andro- (Invitrogen Ltd., Paisley, UK). For experiments in which the gen-regulated genes identified earlier cluster together as effects of ligands were determined, the cells were plated in genes whose expression is strongly regulated by R1881, RPMI or DMEM supplemented with 10% dextran–charcoal- other AR agonists and by SARMs. These clusters stripped fetal calf serum for 3 days before the addition of include many other genes that have not been identified ligands. The final ligand concentrations were methyltrienolone earlier as androgen-responsive genes, and that may play (R1881; 1 nM), dihydrotestosterone (100 nm), 17b-oestradiol significant roles in prostate cancer biology. Indeed, (E2; 100 nm), medroxyprogesterone acetate (MPA; 100 nm), cyproterone acetate (CPA; 10 mM), hydroxyflutamide (OHF; genes not identified earlier showing strong androgen 10 mM), bicalutamide (BIC; 10 mM). As the ligands were regulation included the genes for nuclear receptor prepared in ethanol, an equal volume of ethanol was added NR4A1, the cytokine receptor CXCR4 and insulin-like to the no-ligand controls. Cells were processed for determining growth factor-1, which have all been implicated in cell number, RNA or protein preparation, as described below. prostate cancer progression (Akashi et al., 2006; Cheng et al., 2006; Tamura et al., 2007). The cluster analysis RNA extraction and reverse–transcriptase PCR (RT–PCR) also indicates that levels of gene regulation by E2 and Total RNA was extracted using the QIAGEN RNeasy Mini MPA are similar to those of CPA and OHF, suggesting kit (Qiagen Ltd., Crawley, UK) following DNaseI treatment, that E2 and MPA act as SARMs. according to the manufacturer’s instructions. The concentra- Finally, pathway analyses particularly highlight genes tion and purity of RNA were determined by measuring involved in fatty-acid metabolism as particular andro- spectrophotometric absorption at 260–280 nm. RNA integrity

Oncogene Androgen-regulated gene expression in LNCaP cells S Ngan et al 2062 was checked by electrophoresis on 1.5% agarose gels. cDNA 7900HT fast real-time PCR system, using TaqMan gene synthesis was carried out using 2 mg of total RNA. RNA was expression assays from ABI, according to the manufacturer’s reverse transcribed to cDNA in a volume of 20 ml using instructions. Q-RT–PCR was carried out using total RNA RevertAid M-MuLV reverse transcriptase (Helena Biosciences prepared from three replicate cultures following the addition Europe, Gateshead, UK), according to the manufacturer’s of 1 nM R1881 for varying times or the different AR ligands for protocols. PCR reactions were carried out using ReddyMix Taq 24 h. TaqMan assay details are available on request. polymerase (Abgene, Epsom, UK). PCR products were resolved in 1.5% agarose gels. Conflict of interest RNA for microarray and low-density array card analysis R1881 (1 nM) was added 72 h after the seeding of 3.5 Â 106 The authors declare no conflict of interest. LNCaP cells, and total RNA was prepared as described above. Three replicates from the RNA prepared from cells treated with R1881 for 24 h were used for hybridization to the ABI Acknowledgements Human Genome Survey Microarray V2.0. Raw data were quality assessed and filtered according to the recommendations We are grateful to the members of the group for advice and supplied by the ABI1700 data analysis user guide and the support. Our particular thanks go to Drs Greg Brooke and filtered data were ‘vsn normalized’ (Huber et al., 2002). Charlotte Bevan for discussions, and to Dr Lev Soinov for Differential expression was assessed using linear models and aiding the analysis of the microarray data. This work was empirical Bayes algorithms as described (Smyth, 2004). carried out through the support of the Joron trust, Hammer- Q-RT–PCR analysis of 384 genes using TaqMan gene smith Hospital trustees, Prostate Cancer Charity and Cancer expression analysis primer sets was carried out using an ABI Research UK.

References

Abdulkadir SA, Magee JA, Peters TJ, Kaleem Z, Naughton CK, growth factor-beta signaling through interaction with Smad3. Humphrey PA et al. (2002). Conditional loss of Nkx3.1 in adult J Biol Chem 277: 1240–1248. mice induces prostatic intraepithelial neoplasia. Mol Cell Biol 22: Cunha GR, Donjacour AA, Cooke PS, Mee S, Bigsby RM, Higgins SJ 1495–1503. et al. (1987). The endocrinology and developmental biology of the Agoulnik IU, Weigel NL. (2006). Androgen receptor action in prostate. Endocr Rev 8: 338–362. hormone-dependent and recurrent prostate cancer. J Cell Biochem Davies P, Eaton CL. (1991). Regulation of prostate growth. 99: 362–372. J Endocrinol 131: 5–17. Akashi T, Koizumi K, Nagakawa O, Fuse H, Saiki I. (2006). Dehm SM, Tindall DJ. (2006). Molecular regulation of androgen Androgen receptor negatively inﬂuences the expression of chemo- action in prostate cancer. J Cell Biochem 99: 333–344. kine receptors (CXCR4, CCR1) and ligand-mediated migration in Dennis Jr G, Sherman BT, Hosack DA, Yang J, Gao W, Lane HC prostate cancer DU-145. Oncol Rep 16: 831–836. et al. (2003). DAVID: Database for Annotation, Visualization, and Arnold JT, Liu X, Allen JD, Le H, McFann KK, Blackman MR. Integrated Discovery. Genome Biol 4: P3. (2007). Androgen receptor or estrogen receptor-beta blockade alters Doesburg P, Kuil CW, Berrevoets CA, Steketee K, Faber PW, Mulder DHEA-, DHT-, and E(2)-induced proliferation and PSA produc- E et al. (1997). Functional in vivo interaction between the amino- tion in human prostate cancer cells. Prostate 67: 1152–1162. terminal, transactivation domain and the ligand binding domain of Brinkmann AO, Blok LJ, de Ruiter PE, Doesburg P, Steketee K, the androgen receptor. Biochemistry 36: 1052–1064. Berrevoets CA et al. (1999). Mechanisms of androgen receptor Feldman BJ, Feldman D. (2001). The development of androgen- activation and function. J Steroid Biochem Mol Biol 69: 307–313. independent prostate cancer. Nat Rev Cancer 1: 34–45. Brooke GN, Parker MG, Bevan CL. (2008). Mechanisms of androgen Furr BJ, Tucker H. (1996). The preclinical development of bicaluta- receptor activation in advanced prostate cancer: differential co- mide: pharmacodynamics and mechanism of action. Urology 47: activator recruitment and gene expression. Oncogene 27: 2941–2950. 13–25discussion 29–32. Burmester JK, Suarez BK, Lin JH, Jin CH, Miller RD, Zhang KQ Gottlieb B, Beitel LK, Wu JH, Triﬁro M. (2004). The androgen et al. (2004). Analysis of candidate genes for prostate cancer. Hum receptor gene mutations database (ARDB): 2004 update. Hum Hered 57: 172–178. Mutat 23: 527–533. Busillo JM, Benovic JL. (2007). Regulation of CXCR4 signaling. Guo Y, Jacobs SC, Kyprianou N. (1997). Down-regulation of Biochim Biophys Acta 1768: 952–963. protein and mRNA expression for transforming growth factor-beta Carter HB, Coffey DS. (1990). The prostate: an increasing medical (TGF-beta1) type I and type II receptors in human prostate cancer. problem. Prostate 16: 39–48. Int J Cancer 71: 573–579. Chen H, Nandi AK, Li X, Bieberich CJ. (2002). NKX-3.1 interacts Guo Y, Kyprianou N. (1998). Overexpression of transforming growth with prostate-derived Ets factor and regulates the activity of the factor (TGF) beta1 type II receptor restores TGF-beta1 sensitivity PSA promoter. Cancer Res 62: 338–340. and signaling in human prostate cancer cells. Cell Growth Differ 9: Cheng I, Stram DO, Penney KL, Pike M, Le Marchand L, 185–193. Kolonel LN et al. (2006). Common genetic variation in IGF1 and Guo Y, Kyprianou N. (1999). Restoration of transforming growth prostate cancer risk in the Multiethnic Cohort. J Natl Cancer Inst factor beta signaling pathway in human prostate cancer cells 98: 123–134. suppresses tumorigenicity via induction of caspase-1-mediated Chintharlapalli S, Burghardt R, Papineni S, Ramaiah S, Yoon K, Safe apoptosis. Cancer Res 59: 1366–1371. S. (2005). Activation of Nur77 by selected 1,1-bis(30-indolyl)-1-(p- Heemers HV, Verhoeven G, Swinnen JV. (2006). Androgen activation substituted phenyl)methanes induces apoptosis through nuclear of the sterol regulatory element-binding protein pathway: current pathways. J Biol Chem 280: 24903–24914. insights. Mol Endocrinol 20: 2265–2277. Chipuk JE, Cornelius SC, Pultz NJ, Jorgensen JS, Bonham MJ, Hodgson MC, Astapova I, Hollenberg AN, Balk SP. (2007). Activity Kim SJ et al. (2002). The androgen receptor represses transforming of androgen receptor antagonist bicalutamide in prostate cancer

Oncogene Androgen-regulated gene expression in LNCaP cells S Ngan et al 2063 cells is independent of NCoR and SMRT corepressors. Cancer Res of RNA polymerase II elongation factors. Proc Natl Acad Sci USA 67: 8388–8395. 94: 3639–3643. Huang da W, Sherman BT, Tan Q, Collins JR, Alvord WG, Roayaei J Simone F, Luo RT, Polak PE, Kaberlein JJ, Thirman MJ. (2003). et al. (2007). The DAVID Gene Functional Classification Tool: a ELL-associated factor 2 (EAF2), a functional homolog of EAF1 novel biological module-centric algorithm to functionally analyze with alternative ELL binding properties. Blood 101: 2355–2362. large gene lists. Genome Biol 8: R183. Smyth GK. (2004). Linear models and empirical bayes methods for Huber W, von Heydebreck A, Sultmann H, Poustka A, Vingron M. assessing differential expression in microarray experiments. Stat (2002). Variance stabilization applied to microarray data calibration Appl Genet Mol Biol 3: Article 3. and to the quantification of differential expression. Bioinformatics Sobel RE, Sadar MD. (2005). Cell lines used in prostate cancer 18: S96–S104. research: a compendium of old and new lines–part 1. J Urol 173: Ke N, Claassen G, Yu DH, Albers A, Fan W, Tan P et al. (2004). 342–359. Nuclear hormone receptor NR4A2 is involved in cell transforma- Steketee K, Timmerman L, Ziel-van der Made AC, Doesburg P, tion and apoptosis. Cancer Res 64: 8208–8212. Brinkmann AO, Trapman J. (2002). Broadened ligand responsive- Kim MJ, Bhatia-Gaur R, Banach-Petrosky WA, Desai N, Wang Y, ness of androgen receptor mutants obtained by random amino acid Hayward SW et al. (2002). Nkx3.1 mutant mice recapitulate early substitution of H874 and mutation hot spot T877 in prostate cancer. stages of prostate carcinogenesis. Cancer Res 62: 2999–3004. Int J Cancer 100: 309–317. Kwok WK, Ling MT, Lee TW, Lau TC, Zhou C, Zhang X et al. Storlazzi CT, Mertens F, Nascimento A, Isaksson M, Wejde J, Brosjo (2005). Up-regulation of TWIST in prostate cancer and its O et al. (2003). Fusion of the FUS and BBF2H7 genes in low grade implication as a therapeutic target. Cancer Res 65: 5153–5162. fibromyxoid sarcoma. Hum Mol Genet 12: 2349–2358. Lau KM, LaSpina M, Long J, Ho SM. (2000). Expression of estrogen Takahashi Y, Perkins SN, Hursting SD, Wang TT. (2007). 17beta- receptor (ER)-alpha and ER-beta in normal and malignant Estradiol differentially regulates androgen-responsive genes through prostatic epithelial cells: regulation by methylation and involvement estrogen receptor-beta- and extracellular-signal regulated kinase- in growth regulation. Cancer Res 60: 3175–3182. dependent pathways in LNCaP human prostate cancer cells. Mol Lee MY, Moon JS, Park SW, Koh YK, Ahn YH, Kim KS. (2009). Carcinog 46: 117–129. KLF5enhances SREBP-1 action in androgen-dependent induction Tamura K, Furihata M, Tsunoda T, Ashida S, Takata R, Obara W of fatty acid synthase in prostate cancer cells. Biochem J 417: et al. (2007). Molecular features of hormone-refractory prostate 313–322. cancer cells by genome-wide gene expression profiles. Cancer Res 67: Lu NZ, Wardell SE, Burnstein KL, Defranco D, Fuller PJ, Giguere V 5117–5125. et al. (2006). International Union of Pharmacology. LXV. The Taplin ME, Balk SP. (2004). Androgen receptor: a key molecule in the pharmacology and classification of the nuclear receptor super- progression of prostate cancer to hormone independence. J Cell family: glucocorticoid, mineralocorticoid, progesterone, and andro- Biochem 91: 483–490. gen receptors. Pharmacol Rev 58: 782–797. Veldscholte J, Berrevoets CA, Brinkmann AO, Grootegoed JA, Masiello D, Cheng S, Bubley GJ, Lu ML, Balk SP. (2002). Mulder E. (1992a). Anti-androgens and the mutated androgen Bicalutamide functions as an androgen receptor antagonist by receptor of LNCaP cells: differential effects on binding affinity, assembly of a transcriptionally inactive receptor. J Biol Chem 277: heat-shock protein interaction, and transcription activation. 26321–26326. Biochemistry 31: 2393–2399. McConnell JD. (1991). Physiologic basis of endocrine therapy for Veldscholte J, Berrevoets CA, Ris-Stalpers C, Kuiper GG, Jenster G, prostatic cancer. Urol Clin North Am 18: 1–13. Trapman J et al. (1992b). The androgen receptor in LNCaP cells Myung SJ, Rerko RM, Yan M, Platzer P, Guda K, Dotson A et al. contains a mutation in the ligand binding domain which affects (2006). 15-Hydroxyprostaglandin dehydrogenase is an in vivo steroid binding characteristics and response to antiandrogens. suppressor of colon tumorigenesis. Proc Natl Acad Sci USA 103: J Steroid Biochem Mol Biol 41: 665–669. 12098–12102. Wolf I, O0Kelly J, Rubinek T, Tong M, Nguyen A, Lin BT et al. Nelson PS, Clegg N, Arnold H, Ferguson C, Bonham M, White J et al. (2006). 15-hydroxyprostaglandin dehydrogenase is a tumor sup- (2002). The program of androgen-responsive genes in neoplastic pressor of human breast cancer. Cancer Res 66: 7818–7823. prostate epithelium. Proc Natl Acad Sci USA 99: 11890–11895. Xiao W, Zhang Q, Jiang F, Pins M, Kozlowski JM, Wang Z. (2003). Oettgen P, Finger E, Sun Z, Akbarali Y, Thamrongsak U, Boltax J Suppression of prostate tumor growth by U19, a novel testosterone- et al. (2000). PDEF, a novel prostate epithelium-specific ets regulated apoptosis inducer. Cancer Res 63: 4698–4704. transcription factor, interacts with the androgen receptor and Ying J, Srivastava G, Hsieh WS, Gao Z, Murray P, Liao SK et al. activates prostate-specific antigen gene expression. J Biol Chem 275: (2005). The stress-responsive gene GADD45G is a functional tumor 1216–1225. suppressor, with its response to environmental stresses frequently Ono T, Losada A, Hirano M, Myers MP, Neuwald AF, Hirano T. disrupted epigenetically in multiple tumors. Clin Cancer Res 11: (2003). Differential contributions of condensin I and condensin II 6442–6449. to mitotic chromosome architecture in vertebrate cells. Cell 115: Yoon HG, Wong J. (2006). The corepressors silencing mediator of 109–121. retinoid and thyroid hormone receptor and nuclear receptor Qi W, Gao S, Wang Z. (2008). Transcriptional regulation of the TGF- corepressor are involved in agonist- and antagonist-regulated beta1 promoter by androgen receptor. Biochem J 416: 453–462. transcription by androgen receptor. Mol Endocrinol 20: 1048–1060. Rae JM, Johnson MD, Cordero KE, Scheys JO, Larios JM, Gottardis Yuen HF, Chua CW, Chan YP, Wong YC, Wang X, Chan KW. MM et al. (2006). GREB1 is a novel androgen-regulated gene (2006). Id proteins expression in prostate cancer: high-level required for prostate cancer growth. Prostate 66: 886–894. expression of Id-4 in primary prostate cancer is associated with Shang Y, Myers M, Brown M. (2002). Formation of the androgen development of metastases. Mod Pathol 19: 931–941. receptor transcription complex. Mol Cell 9: 601–610. Zha S, Ferdinandusse S, Denis S, Wanders RJ, Ewing CM, Luo J et al. Shanmugam I, Cheng G, Terranova PF, Thrasher JB, Thomas CP, Li (2003). Alpha-methylacyl-CoA racemase as an androgen- B. (2007). Serum/glucocorticoid-induced protein kinase-1 facilitates independent growth modifier in prostate cancer. Cancer Res 63: androgen receptor-dependent cell survival. Cell Death Differ 14: 7365–7376. 2085–2094. Zha S, Ferdinandusse S, Hicks JL, Denis S, Dunn TA, Wanders RJ Shilatifard A, Duan DR, Haque D, Florence C, Schubach WH, et al. (2005). Peroxisomal branched chain fatty acid beta-oxidation Conaway JW et al. (1997). ELL2, a new member of an ELL family pathway is upregulated in prostate cancer. Prostate 63: 316–323.

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

Oncogene