Peroxisome Proliferator-Activated Receptor Gamma Controls Prostate Cancer Cell Growth through AR-Dependent and -Independent Mechanisms

Abstract

Background:

Prostate Cancer (PC) remains a leading cause of cancer mortality and the most successful chemopreventative and treatment strategies for PC come from targeting the androgen receptor (AR). Although AR plays a key role, it is likely that other molecular pathways also contribute to PC, making it essential to identify and develop drugs against novel targets. Recent studies have identified PPARγ, a nuclear receptor that regulates fatty acid metabolism, as a novel target in PC, and suggest that inhibitors of PPARγ could be used to treat existing disease. We hypothesized that PPARγ acts through AR-dependent and independent mechanisms to control PC development and growth and that PPARγ inhibition is a viable PC treatment strategy.

Methods:

Immunohistochemistry was used to determine expression of PPARγ in a cohort of PC patients. Standard molecular techniques were used to investigate the PPARγ signaling in PC cells as well a xenograft mouse model to test PPARγ inhibition in vivo. Kaplan-Meier curves were created using cBioportal.

Results:

We confirmed the expression of PPARγ in human PC. We then showed that small molecule inhibition of PPARγ decreases the growth of AR-positive and -negative PC cells in vitro and that T0070907, a potent PPARγ antagonist, significantly decreased the growth of human PC xenografts in nude mice. We found that PPARγ antagonists or siRNA do not affect mitochondrial activity nor do they cause apoptosis; instead, they arrest the cell cycle. In AR-positive PC cells, antagonists and siRNAs reduce AR transcript and protein levels, which could contribute to growth inhibition. AR-independent effects on growth appear to be mediated by effects on fatty acid metabolism as the specific FASN inhibitor, Fasnall, inhibited PC cell growth but did not have an additive effect when combined with PPARγ antagonists. Patients with increased PPARγ target gene expression, but not alterations in PPARγ itself, were found to have significantly worse overall survival.

Conclusions:

Having elucidated the direct cancer cell effects of PPARγ inhibition, our studies have helped to determine the role of PPARγ in PC growth, and support the hypothesis that PPARγ inhibition is an effective strategy for PC treatment.

1. Introduction:

Prostate cancer (PC) is the second most diagnosed cancer and sixth leading cause of cancer mortality in men in the world [1]. Because of both the increasing prevalence of and high costs associated with PC, there is an urgent need to develop novel strategies for PC treatment. PC is a multifaceted disease, with the greatest risk factors being age, race, inherited susceptibility, and environmental and behavioral factors such as diet. The development and growth of PC is uniquely dependent on androgens and the androgen receptor (AR) [2]. Our most effective regimens for treating metastatic PC have arisen from the pioneering experiments in which suppression of testicular testosterone production was shown to cause tumor regression [3]. Since then, our ability to inhibit androgen synthesis and AR signaling has improved, and several agents are now approved for the treatment of metastatic PC [4]. Despite great progress in developing novel treatments, once PC metastasizes, it remains incurable as patients eventually develop resistance to AR-targeted therapies. This necessitates the need for novel therapeutic targets.

Many alterations outside of the AR axis have been proposed to contribute to disease initiation and/or progression, including PTEN loss, Nkx3.1 loss, Myc amplification, FOXM1 over-expression, and PI3K/AKT activity, among others [5–9]. It is likely that various combinations of these alterations occur in different patients to cause tumorigenic transformation of cells, and that distinct alterations may dictate the course of disease progression and provide distinct therapeutic targets. We and others have recently identified the peroxisome proliferator-activated receptor gamma (PPARγ) as an oncogene that contributes to PC development and progression [10, 11]. PPARγ is a ligand-dependent transcription factor belonging to the nuclear hormone receptor superfamily [12]. PPARγ is known to play a prominent role in adipocyte differentiation, the inflammatory response, lipid metabolism, and peripheral glucose utilization, and synthetic PPARγ agonists are clinically used as insulin sensitizers in patients with type II diabetes [12]. PPARγ can be activated by a variety of different of natural ligands which include polyunsaturated fatty acids (FAs) and its associated metabolites which include 9- and 13-hydroxyoctadecadienoic acid (HODE) and Prostaglandin (PGE2) [12]. However, the only high-affinity ligands are synthetic agonists including those of the glitazone family [12].

It was originally thought that PPARγ acted as a tumor suppressor in PC and that PPARγ agonists could be used as therapeutics [13–18]. However, further analysis clearly demonstrated that PPARγ agonists were working via PPARγ-independent mechanisms to inhibit the growth of PC cells in vitro [19–23]. In addition, newer studies support the notion that PPARγ is an oncogene in PC. For instance, PPARγ expression is greater in PC tissue than in benign tissue and expression increases with stage and grade of PC [11, 24, 25]. Two recent molecular studies further support an oncogenic role for PPARγ in PC. In our previous work, we sought a molecular mechanism to explain the large retrospective studies that have shown that long-term use of warfarin reduced the risk of PC diagnosis [10, 26–29]. Although warfarin plays an important role in blood coagulation, we used a variety of approaches to demonstrate that warfarin inhibited PPARγ signaling in prostate and PC cells, and that this inhibition of PPARγ signaling also led to an inhibition of AR signaling, which contributed to the inhibition of PC cell growth. Independently, Ahmad et al. identified Pparg as a novel gene that drives prostate carcinogenesis using a Sleeping Beauty screen in prostate-specific Pten^−/− mice [30]. Mice with insertions upstream of the Pparg gene that caused increased expression of the PPARγ protein had decreased survival and increased metastases to the lungs and lymph nodes compared to littermate controls. Increased PPARγ expression was associated with increased levels of PPARγ target genes, and overexpression of PPARγ in three AR-negative PC cell lines, DU-145, PC3, and PC3M, increased cell proliferation and migration. In comparison, siRNA knockdown of PPARγ had the opposite effect in vitro, and treatment with a PPARγ antagonist decreased tumor size. Furthermore, Ahmad et al. confirmed that levels of PPARγ positively correlated with PC grade [30]. Here, we investigate the mechanism by which PPARγ controls PC cell growth and demonstrate that PPARγ inhibition affects both AR-dependent growth, primarily through regulation of AR transcript levels, and AR-independent growth, primarily through regulation of FA synthesis. Finally, we show that PPARγ signaling is associated with overall survival in human PC.

2. Materials/Methods:

Reagents

Dihydrotestosterone was purchased from Steraloids. PPAR agonists and antagonists were purchased from: GW1929 (Sigma), Pioglitazone (Santa Cruz), GW9662 (Sigma), T0070907 (Cayman Chemical).

Cell lines and culture conditions

LNCaP cells (ATCC), BPH-1 cells (gift from Ann Donjacour), and LAPC4 cells (gift from Charles Sawyers) were maintained in phenol red-free RPMI 1640 supplemented with 10% FBS and antibiotics. LNCaP cells stably expressing AR and PPARγ were generated through the transfection of fluorescent and HA-tagged AR and PPARγ (clone HsCD00455985) HEK 293 cells stably expressing fluorescent (CARY) and HA-tagged AR were generated previously, and along with PC3 and DU-145 cells (both from ATCC) were maintained in Dulbecco’s modified Eagle’s medium supplemented with 10% FBS and antibiotics [31].

Transfection and transcriptional assays

Cells were transfected using Lipofectamine Plus (Thermofisher) with PSA-luciferase [48], MMTV-luciferase or PPRE-luciferase, and pRL-SV40 (Promega) as a control. Gene specific or negative control siRNA (Qiagen) was transfected together with the plasmids when applicable. Cells were transferred to a 96-well plate 24 hours after transfection and treated with the appropriate drugs dissolved in media supplemented with charcoal-stripped serum for another 24 hours. Luciferase activity was assayed 24 hours after treatment using the dual-luciferase reporter assay system (Promega). Student’s t-test (two-sided and equal variance) was performed and association was considered significant when p < 0.05 and indicated by an asterisk.

Cell Growth Curves

For growth curves, cells were plated at a density of approximately 20,000 cells/well in 48-well plates, in quadruplicate. The following day, medium with or without drugs was added to the cells. Proliferation was determined by measuring the DNA content of the cells in each well. Cells were fixed in 2% PFA, followed by staining for 5 min at room temperature with 0.2 ng/mL DAPI in PBS solution. The cells were washed with PBS solution, then read on a fluorescence plate reader using 365/439 excitation/emission wavelengths. A Student t test was used to determine significant differences between control and drug treated populations.

Cell Toxicity assays

MTT Cell Proliferation Assay Kit purchased from Cayman Chemical. Cells were treated with drugs for 24–72 hours. MTT reagent was dissolved in 5 mL of assay buffer and added after drug treatment for 4 hours. After formazan production, crystal dissolving solution (crystal dissolving SDS was dissolved in crystal dissolving solution made of hydrochloride) was added and cells were placed in the 37 °C incubator for 12 hours. Absorbance was measured at 570 nM using a spectrophotometer.

Immunoprecipitation

LCP cells were treated with drugs for 24 hours and lysed in TBS, 0.1% Triton X-100, protease and phosphatase inhibitors (Roche). Immunoprecipitation was performed using anti-AR (AR441, Santa Cruz) or anti- PPARγ antibody (Cell Signaling Technology). Western blot was used to detect AR (PG-21, Millipore) and PPARγ (Cell Signaling and Technology).

Quantitative PCR (qPCR)

Total RNA was extracted from cells or homogenized tissue using GeneJet RNA purification kit (ThermoFisher). Reverse transcription was performed using M-MLV reverse transcriptase (ThermoFisher). Transcript levels were quantified relative to the RPL19 housekeeping gene using SYBR green (ThermoFisher) with Rox reference dye (ThermoFisher) on a StepOne Real Time PCR System (ThermoFisher). Relative gene expression was calculated by ΔΔCt. Student’s t-test was performed and association was considered significant when p < 0.05 and indicated by an asterisk.

Immunohistochemistry

Radical prostatectomy or metastatic biopsy tissue was obtained from 51 patients with approval from the City of Hope Institutional Review Board under protocol 11058. Immunohistochemistry was performed using standard protocols. Antigen retrieval was performed on paraffin-embedded sections using citrate-based antigen unmasking solution (Vector Labs, Burlingame, CA). Slides were blocked with 10% normal goal serum, and then stained with anti-PPARγ antibody (CST)diluted 1:200 in TBST or normal rabbit IgG (Santa Cruz) overnight at 4°C. Slides were then incubated in biotinylated anti-rabbit secondary antibody (Vector Labs) followed by Vectastain Elite ABC reagent (Vector Labs) and developed using DAB substrate (Vector Labs). Sections were counterstained with Harris hematoxylin (Poly Scientific, Bay Shore, NY).

Immunofluorescence

PC3 cells were treated with 500nM Mitotracker Red (Cell Signaling Technology) for 45 minutes prior to fixation. Both PC3 and LCP cells were fixed in 2% paraformaldehyde in PBS and permeablized in 0.1% triton X in PBS. Cells were blocked for 60 min at RT with 10% normal goat serum in PBS. Anti- PPARγ (Cell Signaling Technologies) was added overnight at 4C, followed by washing and treatment with an Alexafluor-488 (PC3) or Alexafluor-647 (LCP) labeled, goat-anti-rabbit secondary antibody for 60 min. Cells were washed and mounted and pictures were taken with the aid of a CCD-camera attached to a fluorescent microscope.

Bioinformatics:

To determine putative binding sites for PPARγ in the AR gene promoter we used the motif searching algorithm STORM [32]. PPARγ motifs taken from the transfac motif database were searched for in a 5kb window centered on the transcriptional start site of the AR gene using the hg19 reference genome for human and the mm9 reference genome for mouse. A stringency threshold of 1e-3 p-value was used to filter out low confidence matches. We interrogated the Memorial Sloan Kettering Cancer Center “Prostate Adenocarcinoma” dataset using the cBioportal platform [33–35]. Our analysis included mutations, putative copy-number alterations, and mRNA expression in PPARγ, ACC (also known as ACACA), ACLY, FASN, and LIPE where we compared Kaplan Meier Survival Curves of patients with and without genomic alterations.

T007 Treatment of Xenograft Mouse Model:

All animal experiments were conducted with approval from the institutional Animal Care and Use Committee of City of Hope under Protocol 11010. Male nude mice, aged 8 weeks, were e NCI breeding program. 500,000 LCP cells in matrigel were subcutaneously injected into the flank of 8 week old male nude mice. Once a palpable tumor formed, mice were treated with 15 mg/kg T007 (in DMSO/PEG/PBS) or vehicle for four weeks. Tumor size was measured using calipers every week. Mice were euthanized after 4 weeks and tissues collected for analysis.

3. Results:

3.1. PPARγ expression in PC

To determine the direct cellular effects of PPARγ antagonists on the growth of PC cells, we examined the expression of PPARγ in PC cell lines by Western blot (Figure 1A) and found that AR-negative PC3 and DU145, but not the AR-positive LAPC4 or LNCaP cell lines expressed PPARγ. We created a LNCaP cell line that stably expresses AR and PPARγ (LCP) so that we would have an AR-positive model with which to work. PPARγ was activated by the PPARγ-selective agonist pioglitazone (pio) and inhibited by the PPARγ-selective antagonist T0070907 (T007) in each PPARγ-expressing cell line, as determined by a PPAR response element (PPRE) luciferase activity (Figure 1B). We next determined the expression of PPARγ in a cohort of localized and metastatic PC samples (Figure 1C), and found that PPARγ was expressed in a majority of cancers (36/43 localized, 5/8 metastatic). While we found that PPARγ was expressed predominantly in cancers as has been previously reported, we saw that it was also occasionally expressed in benign glands (Figure 1C) [24, 25].

Figure 1:

Expression and activity of PPARγ in PC cells and tissues.

(A) Western blot for PPARγ in the indicated PC cell lines.

(B) PC3, DU-145 and LCP (LNCaP constitutively expressing PPARγ and AR) cells were transfected with PPRE-firefly luciferase and control renilla luciferase plasmids, and treated with 1 μM pioglitazone (pio) or pio + 100nM T0070907 (T007) overnight. The following day, the luciferase activity of renilla normalized firefly was quantified (* p<.01).

(C) Examples of PPARγ staining in localized and metastatic PC, and in benign tissue sample.

3.2. PPARγ controls the growth of AR-negative and AR-positive PC cells

We next determined if PPARγ antagonists had a direct effect on the growth of PC cells. We found that growth of both PPARγ-positive cell lines (PC3 and LCP), but not PPARγ-negative LAPC4 cells, was inhibited by T007 (Figure 2A). We also found that transfection with siRNAs targeting PPARγ significantly inhibited the growth of PC3 and LCP cells in culture (Figure 2B). We then wanted to determine if PPARγ antagonists controlled the growth of PC cells in vivo. Ahmad et al previously demonstrated that the PPARγ antagonist GW9662 inhibited the growth of PC3 xenografts in immunocompromised mice [11]. We decided to test the effects of T007 on the growth of our LCP cells in a similar study. We first determined that 15mg/kg once daily IP was an effective concentration of T007 by measuring the effect of treatment on PPARγ target gene expression in fat tissue (Supplementary Figure 1). Immunocompromised mice were injected subQ with LCP cells and once tumors were established, mice were treated once daily with 15mg/kg T007 for 4 weeks. We found that T007 treatment significantly inhibited the growth of LCP cells in this model (Figure 2C). In fact, four out of seven tumors completely disappeared during the course of the treatment. These data together suggest that PPARγ inhibition can control the growth of both AR-positive and AR-negative cells.

Figure 2:

PPARγ antagonists and siRNA control the growth of PC cells in culture and in vivo.

(A) LCP, PC3, and LAPC4 cells were treated with vehicle or the indicated concentration of T007. Growth was measured by DAPI content on day 0 and day 5.

(B) LCP or PC3 cells were transfected with control or PPARγ-targeted siRNAs. Growth was measured by DAPI content on day 0 and day 5. RT-qPCR analyses of PPARγ transcripts from identically treated cells are shown on the right.

(C) Nude mice were subcutaneously injected with 1×10⁷ LCP cells. After tumors became measurable, treatment was given once daily with vehicle or 15mg/kg T007 via IP injection (n=7). Tumor volume was measured weekly by caliper.

3.3. PPARγ inhibition induces cell cycle arrest

We next set out to determine how PPARγ inhibition was slowing the growth of PC cells. No decrease in mitochondrial activity via MTT assay was observed with either T007 treatment or PPARγ siRNA in PC3 or LCP cells (Figure 3A). Furthermore, PPARγ did not co-localize with mitochondria in PC cells and was expressed solely in the nucleus of these cells, as shown by immunofluorescence (IF, Figure 3B), suggesting that PPARγ was not acting in its reported role as a mediator of mitochondrial function [36] to inhibit the growth of PC cells. Effective concentrations of T007 and GW9662 had no effect on levels of cleaved caspases (Figure 3C) suggesting that they do not cause apoptosis. However, we found that T007 treatment arrested LCP cells in G1/G2, as shown by the increase in these populations by flow cytometry (Figure 3D). This corroborates previous studies in which suppression of key PPARγ-target genes, including FASN and ACC, induced G1 cell cycle arrest [37].

Figure 3:

PPARγ regulates PC cell growth by cell cycle arrest.

(A) LCP or PC3 cells were treated with T007 or PPARγ siRNA for 24 hrs, at which point an MTT assay was performed. No treatment resulted in a significant decrease in MTT activity.

(B) PC3 cells were fixed and stained for expression of PPARγ (green), mitochondria (Mitotracker Deep Red) and DNA (DAPI).

(C) Western Blot of caspases 3 and 9, and control p84 protein in LCP cells treated with the indicated concentrations of T007 and GW9662 for 24 hours.

(D) LCP cells were treated with 100nM T007 or cyclohexamide as a control for 24 hrs and flow cytometry was used to quantify the fraction of cells in each stage of the cell cycle.

3.4. PPARγ regulates AR transcript levels and activity

We have previously shown that PPARγ antagonists inhibit AR transcriptional activity using luciferase reporters and RT-qPCR [10]. Here we find that PPARγ siRNA has the same effect on AR activity (Figure 4A). Previous reports had shown that PPARγ agonists decreased AR activity, but this was found to be PPARγ independent [19–23]. We found that PPARγ antagonist treatment only inhibited AR activity in PPARγ-expressing cells (Supplementary Figure 2), suggesting that the effects of PPARγ antagonism on AR signaling are PPARγ dependent. Because AR activity is essential to AR-positive PC cell growth, inhibition of AR signaling would be expected to slow the growth of AR-positive PC cells. To understand how PPARγ regulates AR activity, we first determined if the two proteins interact. We found that, as expected, AR co-localizes with PPARγ in the nucleus of LCP cells treated with DHT (Figure 4B); however, we were unable to detect an interaction between AR and PPARγ by co-IP (Figure 4C). AR is known to cooperate with many other transcription factors to coordinate transcription of target genes [38]. If AR and PPARγ worked together in such a way, one would expect to find PPAR response elements near AR binding sites. However, a search of publicly-available AR ChIP-seq datasets, including multiple human cell lines and human and mouse prostate tissue, demonstrated that AR peaks are never associated with known PPAR motifs [39–53], suggesting that these two proteins do not cooperate at the DNA to control transcription. We next investigated if PPARγ regulated AR levels via its role as a transcription factor. We found that PPARγ antagonists and siRNA decreased AR transcript (Figure 4D,E) and protein levels (Figure 4F). This decrease is similar to the decrease in AR activity observed with these treatments in luciferase assays [10] suggesting that PPARγ controls AR activity by transcriptionally regulating its expression. A search for putative PPAR response elements identified a plethora of consensus or near-consensus binding sites near the transcription start site of the AR gene in both humans and mice (Supplementary Figure 3A). Furthermore, analysis of publicly available PPARγ ChIP-seq data, including data from mouse and human adipocytes [54, 55], demonstrated a PPARγ peak in the first intron of AR, which is a common position for nuclear receptors to bind and regulate the transcription of genes (Supplementary Figure 3B).

Figure 4:

PPARγ regulation of AR activity.

(A) LCP cells were transfected with control or PPARγ siRNAs along with PSA-firefly luciferase and control renilla luciferase plasmids. The cells were then treated with vehicle or 1nM DHT overnight. The following day, the renilla normalized firefly luciferase activity was quantified (* p<.01).

(B) LCP cells that were treated with DHT were fixed and stained for expression of PPARγ (red), AR (yellow), and DNA (DAPI – blue).

(C) Co-immunoprecipitation of AR and PPARγ shows no interaction between AR and PPARγ.

(D & E) RT-qPCR analysis of AR transcript levels in LCP cells transfected with PPARγ siRNA or treated with T007 compared to controls.

(F) Western blot for expression of AR in lysates of LCP cells treated with vehicle, T007, or GW9662 for 24 hrs.

3.5. PPARγ regulation of FA synthesis contributes to PC cell growth

As PPARγ inhibition or loss also results in decreased growth of AR-negative cells, AR independent mechanisms must also be involved in control of cell growth. One of the major functions of PPARγ is to control fatty acid (FA) synthesis which impacts overall cellular lipogenesis. Although benign cells get required FAs from dietary intake, cancer tissues have been shown to generate their own FAs and phospholipids through de novo lipogenesis to meet the needs for increased growth as well as providing material for key structural components of cell membranes [56]. As PPARγ plays a key role in FA metabolism through its transcriptional regulation of lipogenic enzymes [30, 57], inhibiting PPARγ could disrupt the lipid synthesis pathways which would combat tumor cell growth, proliferation, and survival. As expected, PPARγ inhibition or knockdown in PC cells reduces the transcript levels of the key lipogenesis genes ACC, ACYL, FASN and LIPE in vitro (Figure 5A–B). FASN itself is of particular interest as it is upregulated in a variety of different cancers, including PC, where it correlates with poor prognosis [58–60], and FASN expression increases proliferation and growth in soft agar of both benign and cancerous prostate cells [61]. Conversely, FASN blockade inhibits the proliferation of LNCaP cells in vitro and in vivo [62, 63].Treatment of PC3 and LCP cells with the selective FASN inhibitor Fasnall [64] demonstrated that FASN inhibition decreased the growth of both cell lines (Figure 5C–D). Importantly, the combination of T007 and Fasnall did not have an additive or greater effect. This suggested that T007 and Fasnall were both working in the same pathway, and that PPARγ works through FASN to mediate its effect on PC growth and survival. Supplementing cell culture media with palmitate, which is the primary end product of the FASN enzyme activity [65], rescued the growth of PC cells treated with T007 or FASN (Figure 5C–D). This supports a dominant role for the PPARγ→FASN pathway in mediating the ability of PPARγ antagonists to inhibit PC cell growth.

Figure 5:

PPARγ inhibition regulates PC cell growth via FA metabolism.

PC3 (A) and LCP (B) cells were treated with control or PPARγ siRNAs or treated with the indicated drugs respectively, and PPARγ target transcript levels were quantified by RT-qPCR. PC3 (C) and LCP (D) cells were treated with the indicated drugs and growth was quantified on day 5. (T: T007, F: Fasnall, P: palmitate).

Finally, we interrogated the Memorial Sloan Kettering Cancer Center “Prostate Adenocarcinoma” dataset using the cBioportal platform [33–35]. We found that there was not a statistical difference in overall survival (OS) between cases that had alterations in the PPARG gene compared to cases without alteration that gene (Figure 6A). However, Kaplan-Meier curves demonstrated significantly worse OS in cases that had increased expression of the PPARγ target genes ACC, ACLY, FASN, and LIPE, compared to those that didn’t (Figure 6B). This suggests that increased PPARγ signaling portends worse prognosis in PC patients.

Figure 6:

Increased expression of PPARγ target genes is associated with worse survival.

Kaplan-Meier curves were generated using data from the MSKCC cohort for (A) patients with or without alterations in PPARG or (B) patients with alterations in a panel of PPARγ target genes (FASN, ACC, ACLY, LIPE).

4. Discussion:

Our IHC data corroborates previous findings of increased PPARγ expression in cancerous prostate cells [11, 24, 25], which again supports the notion that PPARγ acts as an oncogene in PC. Interestingly, we observed strong expression of PPARγ in select benign glands (Figure 1C) which, to our knowledge, hasn’t been previously reported. This could implicate the importance of PPARγ in the development of PC, and may suggest that PPARγ antagonists could be used for PC prevention as every prostatectomy sample examined had at least one positive benign gland, suggesting that this is a widespread phenomenon. The location of the PPARγ-positive cells appears to be between the basal and fully differentiated luminal epithelial cells, suggesting that these may be “intermediate” epithelial cells [66]. Interestingly, analysis of the single cell prostate atlas [67] from the Strand lab website shows that PPARγ transcripts are enriched in the stem-like “club” and “hillock” cells. PPARγ transcripts are also enriched in the transition zone of the prostate, where those cells reside, and where organogenesis begins. The nature of these cells warrants further investigation, but if they do represent a cell type with pluripotent potential, it might suggest that PPARγ has a role in prostate epithelial cell development and possibly in the early stages of tumorigenesis, not just in PC cell growth. Indeed, we identified PPARγ as a potential target through our investigation into the mechanism by which warfarin reduces the risk of PC [68]. As we demonstrated that warfarin acts primarily as an inhibitor of PPARγ signaling, it is possible that PPARγ antagonists might also be able to decrease PC risk. While warfarin itself is a poor choice for a chemopreventative due to difficult dosing and potential bleeding issues, it is possible that PPARγ antagonists could mimic its protective effects in PC.

While significant additional studies need to be completed to test the chemopreventative potential of PPARγ antagonists, our data demonstrates a role for these agents in treating both AR-positive and AR-negative PCs. Although the inhibition of LCP xenografts in our study appeared to be greater than that of the PC3 xenografts in the Ahmad et al study, this could be due to differences in the model, choice of inhibitor, or biology of the cells. Additional studies that include PPARγ-positive xenograft lines are warranted. However, the fact that PPARγ antagonists are effective against both AR-positive and AR-negative cancers has important clinical implications because recent evidence suggests that truly AR-negative metastatic PCs, which were once thought to be exceedingly rare, are on the rise with the use of advanced AR targeting agents [69]. No effective treatments exist for this type of PC, so it is possible that PPARγ antagonists could be useful in this setting. PPARγ activity, as measured by the expression of its target genes, could provide a possible biomarker for patients who would best respond to PPARγ targeted therapies.

We sought to determine the direct cellular effects of PPARγ antagonists on the growth of PC cells to better understand how inhibition of PPARγ decreases PC growth, which has not been previously characterized. We found that PPARγ acts in the nucleus, not the mitochondria, to inhibit PC growth, and that PPARγ antagonists induce G1/G2 cell cycle arrest in vitro. This growth arrest was powerful enough to cause regression of some tumors in vivo. In AR-positive cells, this growth arrest is mediated at least in part by inhibition of AR signaling. Our data indicate that PPARγ regulates the transcription of AR, likely by binding in the first intron of the AR gene. We attempted PPARγ ChIP to confirm that binding location in our cell lines, but despite being proficient at the technique, we were unable to identify an antibody and conditions conducive to efficient immunoprecipitation. Thus, we were forced to rely on bioinformatics analyses and publicly available PPARγ ChIP-seq data from other cell lines that clearly demonstrate PPARγ binding in AR intron 1. It should also be noted that although others have reported effects of AR inhibition of PPARγ activity [70], we observed no such effect in our cells. This could be due to differences in cell lines used and assay design.

We found that PPARγ works through FASN and other key lipogenic genes (ACC, ACLY, and LIPE) to mediate its effect on PC growth and survival. Specifically, we found that in all PC cells, both AR-positive and AR-negative, growth arrest is also mediated at least in part by inhibition of FASN and FA synthesis. This was especially clear in PC3 cells, where palmitate was able to completely rescue the growth of cells treated with T007 or Fasnall. The rescue was only partial in LCP cells however. It is possible that PC3 cells might be more dependent on palmitate for FAS and/or PC3 cells have increased FASN expression. Migita et al. showed that overexpression of FASN increased PC cell proliferation only in the presence of AR. FASN knockdown induced apoptosis, which demonstrates that FASN, in the presence of AR, has activities beyond that of increasing FASN concentration [71]. In addition, palmitate has also been shown to induce apoptosis by increasing the level of reactive oxidative species [72].

The importance of PPARγ signaling to human PC growth is also supported by analysis of publicly available datasets. Increased expression of the key PPARγ target genes ACC, ACLY, FASN, and LIPE was significantly associated with decreased OS. These studies should be repeated in other cohorts with longer follow-up to confirm the findings. Expression level or mutations in PPARγ itself were not associated with these clinical variables, suggesting that mechanisms other than protein over-expression or mutation are responsible for increased PPARγ signaling activity, which also warrants further investigation. It is important to mention that AR itself can regulate some aspects of FA metabolism, and so in AR-positive PC cells, there might be cross-regulation of some of the lipogenesis genes. However, PC3 cells are AR-negative, which allowed us to demonstrate that PPARγ can directly control the lipogenesis genes to control PC cell growth. Finally, in addition to FA metabolism, PPARγ has been found to regulate other pathways that could play a role in PC development and progression, including inflammation and regulation of tumor-infiltrating immune cells, which warrants further investigation as well [55].

5. Conclusions:

In summary, our study supports PPARγ as an important new target in PC in both established metastatic cancers and possibly the development of PC. However, current inhibitors lack suitable drug-like properties (poor solubility, high toxicity, irreversible binding, etc.), which suggests that novel PPARγ inhibitors that possess better drug-like properties are needed so that we can test such agents in advanced PC models and eventually see them progress to clinical testing.

Supplementary Material

Supp figS1

Supplemental Figure 1: T007 has on-target activity at limit of solubility.C57/Bl6 mice were treated with 15 mg/kg of T007 once per day for one week. RNA was then extracted from the abdominal fat pad and RT-qPCR was performed to measure the transcript levels of the indicated PPARγ target genes.

Supp figS2

Supplemental Figure 2: PPARγ controls AR transcriptional activity.LNCaPs with (A) and without (B) ectopic expression of PPARγ were transfected with PPRE-firefly luciferase and control renilla luciferase plasmids, and treated with the indicated concentration of drugs overnight. The following day, the renilla normalized firefly luciferase activity was quantified which demonstrated PPARγ activity in each of these cell lines (* p<.01).

Supp figS3

Supplemental Figure 3: PPARγ binding sites in mouse and human AR gene.(A) Putative PPARγ binding sites in mouse and human AR gene using STORM analysis.(B) PPARγ binding sites in mouse 3T3L1 and human hMMADS adipocytes using data from Cistrome Data Browser.