15S-Hydroxyeicosatetraenoic Acid Activates Peroxisome Proliferator-activated Receptor γ and Inhibits Proliferation in PC3 Prostate Carcinoma Cells¹

Recent studies have supported the possible biological significance of alterations in AA³-metabolizing enzymes in Pca. Studies in cell lines have indicated a possible contribution of 5-LOX (1, 2) and cyclooxygenase-2 (3, 4) to Pca cell proliferation. Alterations in 5-LOX expression or 5-HETE formation have not been reported in actual Pca tissues, but recent studies have suggested possible increased expression of cyclooxygenase-2 in Pca(5). Increased 12-LOX mRNA has been reported in Pca compared with benign prostate (6), although increased 12-HETE levels or catalytic activity have not been reported in actual Pca tissues thus far.

We have previously demonstrated that the recently described 15-LOX-2(7) is uniformly expressed in the differentiated apical or secretory cells of benign prostate (8). Benign prostate tissue samples synthesize 15S-HETE as the major eicosanoid product from exogenous AA. In contrast, by immunostaining and enzyme activity assays, 15-LOX-2 and 15S-HETE formation were reduced in Pca(8). More recently, using paired snap-frozen benign and malignant prostate tissue obtained intraoperatively (9),we have shown that 15-LOX-2 mRNA is reduced in tumor compared with benign in the majority (>80%) of patients and confirmed that reduced 15S-HETE formation in tumor is indeed a common alteration (>60%) in Pca.⁴In addition, by immunostaining, reduced 15-LOX-2 correlated inversely with the degree of tumor differentiation, with retained expression in the majority of Gleason score 5 tumors compared with a statistically significant reduction in Gleason 6, 7, and 8–10 tumors(10). 15-LOX-2 was reduced in high-grade prostatic intraepithelial neoplasia compared with benign glands, indicating that this may be an early alteration in Pca development (10). The goal of the present study was to examine if loss of 15-LOX-2 expression may contribute to the malignant phenotype in Pca, by examining its effects on Pca cell proliferation. Studies examining oxidized low-density lipoprotein and foam cell formation from macrophages have shown that 15-HETE may activate transcription by the nuclear receptor PPARγ (11), and other studies have shown inhibition of Pca cell line PC3 proliferation by synthetic PPARγ agonists (12). Therefore, we compared the effects of 15S-HETE to known PPARγ agonists on PC3 cell proliferation and investigated the ability of 15S-HETE to activate PPARγ-dependent transcription in Pca cell lines.

PC-3 cells were obtained from the American Type Culture Collection(Rockville, MD) and cultured in Ham’s FK12 with 10% FCS according to recommendations. Snap-frozen human prostate tissues were obtained intraoperatively during RP as described previously(9). Six-mm punch biopsies from multiple transition zone and peripheral zone sites within the prostate yield cores ∼1–1.5 cm in length, which are immediately placed in liquid nitrogen. Two to three mm from one end of the cores were processed for RNA, and the immediately adjacent 1–2 mm was processed for histology to be representative of the portion processed for RNA. The 1- to 2-mm-thick transverse section for histology was immediately placed in 10%buffered formalin for routine processing and paraffin embedding. Five-μm cross sections were stained with H&E and assessed as benign,tumor, or mixed, with ≥85% benign glands considered benign and ≥85%tumor glands considered tumor.

BRL 49653 was synthesized by Glaxo Wellcome.(PPRE)3-tk-luciferase (13) was provided by Dr. Ron Evans (The Salk Institute, San Diego, CA). The 15S-HETE was prepared using the soybean lipoxygenase, as described previously(14). The 15R-HETE was prepared from 15S-HETE methyl ester by the following sequence: (a) stirring with activated manganese dioxide (10 mg/ml) in methylene chloride at room temperature for 2 h, producing the 15-keto analogue in a 20–30%yield; (b) filtration through a silica column in 5%methanol in methylene chloride and isolation of the 15-ketoeicosatetraenoate product by normal phase HPLC (Alltech; 5-μm silica column, solvent hexane/isopropanol, 100:1, by volume);(c) reduction of the 15-keto derivative using sodium borohydride in methanol; (d) isolation of the 15R and 15S enantiomers by chiral column chromatography using a Chiralpak AD column(Chiral Technologies, Inc., Exton, PA) and a solvent of hexane/ethanol(100:2, by volume); (e) saponification in aqueous 1 m KOH/methanol (1:1) for 2 h at room temperature, followed by (f) repurification of the 15R-HETE by normal-phase HPLC (hexane/isopropanol/glacial acetic acid;100:1:0.1, by volume). The 15R- and 15S-HETEs were quantified by UV spectroscopy using a molar extinction coefficient of 23,000 at 235 nm and stored in ethanol at −20°C at a concentration of 10 or 100 mm.

Subconfluent or confluent cultured cells were examined for 15-LOX-2 mRNA by Northern blot and RT-PCR (with benign prostate as positive control). Benign and malignant prostate tissues from RP specimens and cell lines were examined for PPARγ mRNA by RT-PCR. For RNA isolation from prostate tissues, ∼50 mg were diced with fine scissors in 1 ml Tri Reagent (Molecular Research Center, Inc., Cincinnati, OH). Approximately 100 μl of autoclaved 200-μm glass beads (BioSpec Products, Bartlesville, OK) were added, and the tissue was further disrupted by two 20-s periods of agitation on a Mini-Beadbeater(BioSpec Products), placing the specimen on ice between cycles. RNA was extracted from these homogenates and from cultured cells in T75 flasks using the Triazol reagent (Molecular Research Center, Inc.), according to the manufacturer’s instructions. For Northern blots, 20 μg total RNA was used, and blots were hybridized overnight at 42°C with 1 × 10⁶ cpm/ml each of ³²P-labeled cDNA probes for 15-LOX-2 and GAPDH(prostate tissues and PC3 cells) or a-FABP and GAPDH (untreated or treated PC3 cells) in ULTRAhyb (7). The 15-LOX-2 probe was a 1088-bp cDNA amplified by PCR from a partial cDNA in pcDNA3(Stratagene, La Jolla, CA), with primers 5′-TGC-CTC-TCG-CCA-TCC-AGC-T-3′ and 5′-TGG-GAT-GTC-ATC-TGG-GCC-TGT-3′. The GAPDH probe was an 1100-bp cDNA (Clontech, Palo Alto, CA). The a-FABP probe was a 500 bp fragment released by BamHI and XhoI digestion of a cDNA cloned into pBlueScript SK(−)(Stratagene; Ref. 15). Blots were exposed to Kodak X-OMAT film at −80°C for 1–3 days or to a phosphorimaging screen (Super resolution screen; Packard Instrument Co., Inc., Meriden, CT) for 1 to 4 h at room temperature and imaged on a Cyclone Storage Phosphor System (Packard Instrument Co.).

One μg total RNA was used for conventional RT-PCR reactions with the Promega Access RT-PCR System (Promega, Madison, WI), generally according to the manufacturer’s instructions. For 15-LOX-2, the primers used were 5′-GCC-TCT-CGC-CAT-CCA-GCT-3′ (forward) and 5′-TGC-CGA-GTT-CTC-CTT-CCA-TGA-3′ (reverse), which gives a 126-bp amplified product. For PPARγ, the primers were 5′-GAG-TTC-ATG-CTT-GTG-AAG-GAT-GC-3′ (forward) and 5′-CGA-TAT-CAC-TGG-AGA-TCT-CCG-CC-3′ (reverse), which generate a 233-bp amplimer corresponding to portions of exons 2 and 3 contained in both PPARγ1 and PPARγ2 isoforms (16). Primers to amplify both isoforms were chosen because the relative expression in prostate is not definitively established (17), and both contain the same ligand-binding domain that would potentially bind 15-HETE(16). 15-LOX-2 and PPARγ primers span intron-exon boundaries (7, 16), obviating the need for inclusion of DNase treatment during RNA extraction protocols. Thirty cycles of reaction at 94°C for 30 s, 60°C for 30 s, and 72°C for 60 s, respectively, were carried out on a thermal cycler(Perkin-Perkin-Elmer Corp.).

PPARγ and a-FABP mRNA copy numbers in untreated and treated PC3 cells were determined by real-time quantitative RT-PCR using a Lightcycler fluorescence temperature rapid-air cycler (Roche Molecular Biochemicals, Indianapolis, IN) with cDNA standard curves and the double-stranded DNA-binding fluorescent probe SYBR Green(18, 19, 20, 21). Amplifications were done in glass capillary tubes using a 20-μl reaction of 100 ng total PC3 RNA, 5 mm magnesium chloride, 11 ng/μl Taq-start antibody (Clontech Laboratories), 2.0 μl 1 × SYBR Green RNA master-mix (Roche Molecular Biochemicals), 0.4 μl reverse transcriptase enzyme (Roche Molecular Biochemicals), and 1.0μ m each primer. The cDNA template for PPARγconsisted of a 761-bp fragment inserted into PCRII (Stratagene). The primers were identical to those listed above. The a-FABP template for standard curves consisted of a 500-bp fragment, released by BamHI and XhoI digestion from a full-length (619 bp) cDNA inserted into pBlueScript SK(−) (Stratagene). The primers were 5′-TCA-GTG-TGA-ATG-GGG-ATG-TGA-3′ (forward) and 5′-TCA-ACG-TCC-CTT-GGC-TTA-TGC-3′ (reverse), which generate a 288-bp amplimer. The one-step real-time RT-PCR reactions consisted of the following steps: reverse transcription at 55°C for 15 min,denaturation at 95°C for 1 min, amplification for 45 cycles, and melting curve analysis from 95°C to 65°C at a rate of 0.1°C/s under continuous fluorescence monitoring. The amplification programs consisted of heating at 20°C/s to 95°C, cooling at 20°C/s to 53°C, annealing at 53°C for 5 s, heating at 20°C/s to 72°C, elongation at 72°C for either 12 s (for PPARγ) or 16 s (for a-FABP), and heating at 5°C/s to either 84°C(PPARγ) or 83°C (a-FABP) for fluorescence acquisition. The specificity of the amplimer in each reaction was confirmed by the melting curve analysis, with initial gel confirmation that this large peak corresponded to the expected amplimer (18, 21, 22, 23). The contribution to fluorescence signal of any nonspecific products and/or primer dimers was eliminated by increasing the temperature to 2° below the melting temperature of the specific product, which eliminated any other minor cDNAs (which have lower melting temperatures; Ref. 19). Copy numbers of mRNA were calculated from serially diluted standard curves generated from purified cDNA template (24). Serial dilutions (1:10) over a range of four to five orders of magnitude were used to generate the standard curves (10¹⁰–10⁶copies for PPARγ;10¹⁰–10⁷ copies for a-FABP). The serially diluted standards were simultaneously amplified with the unknown samples to generate a linear standard curve using the fit points method of analysis with four points. Standard curves for both PPARγ and a-FABP had correlation coefficients of 1.00. Control samples run in triplicate had a variance of ∼10%. All of the biological samples fell on the standard curves, and copy numbers of the unknown samples were calculated using the Lightcycler software (version 3).

Control samples of benign prostate were homogenized as described previously (8). For determination of possible 15-HETE formation from cell lysates, cultured cells that were 80–100%confluent in T75 flasks were removed by trypsinization, pelleted, and washed. The entire pellet was resuspended in 100 μl Dulbecco’s PBS(pH 7.4) and sonicated for 3 s. One hundred μl (equal or greater to the protein amount routinely incubated from positive prostate tissues) were transferred to a new tube and incubated with 50μ m [1-¹⁴C]AA for 45 min. Two hundred fifty μl methanol were added, followed by addition of 125μl dichloromethane, vortex-mixing, and evaporation under nitrogen. After resuspension in 50 μl methanol, the sample was collected on a C18 Bond Elute cartridge. One-fifth and subsequent four-fifth fractions were analyzed by reverse-phase HPLC, with radioactivity monitored with an on-line Radiomatic Instrument Flo-One detector as described previously, using a solvent system of methanol:water:acetic acid(80:20:0.01) at a flow rate of 1.1 ml/min and addition of cold HETE standards to monitor retention times (8). In additional experiments, formation of possible AA metabolites was assessed using intact monolayers incubated with labeled AA and calcium ionophore. Eighty to 100% confluent PC3 cells in T75 flasks were incubated in 5 ml PBS, with 2 μCi [¹⁴C]AA with or without cold AA (50 μm total) and 5 μm calcium ionophore A23187 for 15 min at 37°C.

Effects of BRL 49653 and 15-HETE on PC3 proliferation were assessed using an agar cloning technique. An underlay of 0.5% agar in Ham’s F12 containing 5% FCS was prepared by mixing equal volumes of 1% agar and 2× Ham’s F12 plus 10% FCS. Two ml of this mixture were pipetted into the wells of six-well plates and allowed to set. PC3 cells,70–100% confluent, were trypsinized, and the cells were resuspended in growth medium and counted with a hemocytometer. The cells were then diluted to a final concentration of 2000/ml in a mixture of 0.7% agar and 2× Ham’s F12. BRL 49653 and 15-HETE were added from appropriate stock solutions in 50% Etoh/50% DMSO to achieve indicated final concentrations and a final solvent concentration of 0.5%. Vehicle controls received similar volumes of solvent alone. Two ml of the cell suspension were aliquoted into each well. The agar was allowed to set,and the plates were incubated in a humidified chamber at 37°C for 14 days. Colonies were counted in a blinded manner using a 4× objective on a Zeiss inverted microscope. Data are expressed as percent of control.

PC3 cells (5 × 10⁵) were seeded into T75 flasks, and 24 h later, media was replaced with 10 ml fresh media (Ham’s F12K with 10% fetal bovine serum), with or without indicated final concentrations of BRL 49653 or 15S-HETE added in 50%Etoh/50% DMSO (final solvent concentration < 1%). Vehicle controls received equal volumes of solvent alone. After 72 h, cells were removed by trypsin-EDTA, washed, pelleted, and resuspended in 300 μl of 1× PBS with subsequent dropwise addition of 700 μl ice-cold 100% Etoh (final concentration, 70%). Fixed cells were stored at −20°C until stained with propidium iodide with RNase, using standard methods. Cell cycle analysis was performed using a Becton Dickinson FACScan and linked Modfit cell cycle analysis and Winlist software (Verity Software House, Topsham, ME).

PC3 and DU-145 cells (5.0 × 10⁵)were transfected using FUGENE 6 at a lipid:DNA ratio of 3:1. Cells were exposed to a mix containing 150 ng/ml of (PPRE)3-tk-luciferase(a gift from R. Evans, Salk Institute, La Jolla, CA), 150 ng/ml pCDNA3.1, 1.0 ng/ml pRL-SV40 in Opti-MEM (Life Technologies). The transfection mix was replaced after 5 h with 10%charcoal-stripped fetal bovine serum-containing media (Hyclone)supplemented with either 0.1% vehicle (DMSO or ethanol) or the indicated concentrations of BRL 49653 and 15S-HETE. After 24 h,cells were harvested in 1× luciferase lysis buffer. Relative light units from firefly luciferase activity were determined using a luminometer (MGM Instruments) and normalized to the relative light units from renilla luciferase using the Dual Luciferase kit (Promega).

PC-3 cells were selected for the majority of studies on the basis of previous demonstration of constitutive expression of PPARγ mRNA and protein and the susceptibility to growth inhibition by known PPARγagonists (12). Because 15-LOX-2 mRNA and enzyme activity are reduced in most Pcas compared with benign tissue, we investigated whether this is a property of PC3 cells and if these cells are a valid model for 15-LOX-2-negative prostate tumors. Compared with the usual benign prostate tissue, 15-LOX-2 mRNA was not detected in PC3 cells by either Northern blots or RT-PCR (Fig. 1). Similarly, compared with the usual presence of 15-LOX-2 catalytic activity in benign prostate (8), 15-HETE formation from exogenous AA was not demonstrated in similar incubations of PC3 cells(Fig. 1). DU-145 cells were also negative for 15-LOX-2 mRNA by Northern blot and RT-PCR. LNCaP cells were negative for 15-LOX-2 mRNA by Northern blot and showed only a faint band by RT-PCR compared with benign prostate. Neither DU-145 nor LNCaP cells synthesized detectable 15-HETE from exogenous AA (not shown).

RNA from histologically documented pure benign (n = 18) and pure tumor (n = 9) specimens snap-frozen intraoperatively during RP (including six pairs of benign and tumor from the same patients) were examined for the presence of PPARγ by RT-PCR. PPARγ mRNA was detected in all of the benign and malignant prostate tissues examined, as well as in PC3 carcinoma cells(Fig. 2). PPARγ was also detected in DU-145 and LNCaP cells (not shown). PPARα and PPARβ(δ) were also detected by RT-PCR in all of the benign and malignant prostate tissues. PPARβ(δ) was also detected in PC-3, DU-145, and LNCaP cells, whereas PPARα mRNA was detected in LNCaP but not in PC-3 or DU-145 cells (not shown).

Similar to results previously reported by others (12), the synthetic PPARγ agonist BRL 49653 caused inhibition of PC-3 cell proliferation in a soft agar colony-forming assay. A dose-dependent effect was achieved over the 10 nm to 10 μmrange tested, with an IC₅₀ of ∼3μ m (Fig. 3,A). Similarly, exogenous 15S-HETE caused a dose-dependent inhibition of PC3 colony formation, with an IC₅₀of ∼30 μm (Fig. 3,A). These levels are similar to those that have achieved PPARγ-mediated effects in other cell types (25) and were clearly not toxic to cells,based on flow cytometric and ultrastructural analyses (not shown). Inhibition was clearly evident at 10 μm15S-HETE and was essentially complete at 100μ m. In addition to fewer colonies, formed colonies were discernibly smaller at the higher concentration ranges(not shown). Possible stereospecificity of the 15-HETE effect was examined by comparing exogenous 15S- versus 15R-HETE. In each of four different experiments, the inhibition of colony formation was greater for 15S- than for 15R-HETE (Fig. 3 B).

Possible effects of exogenously added BRL 49653 and 15S-HETE on the cell cycle in PC3 cells were examined by flow cytometry. At 3 days of treatment, 3 μm BRL 49653 caused a slight increase in the G₀-G₁ fraction, with a similar reduction in the S-phase fraction (Fig. 4). Thirty μm 15S-HETE (which causes a similar degree of growth inhibition in the soft agar assay) produced a similar slight increase in the G₀-G₁fraction (Fig. 4), whereas 3 and 10 μm were without discernible effect (not shown).

Although the inhibition of PC3 proliferation by BRL 49653 and other known PPARγ agonists has been reported (12, 17), the ability of these agents to actually activate PPARγ-dependent transcription in these cells, which supports mechanism of action, has not been reported. The functional status of endogenous PPARγin PC3 cells was determined using luciferase reporter assays. Concentrations of BRL 49653 inhibiting PC3 proliferation caused PPARγ-dependent transcription in PC3 cells transiently transfected with a PPRE-luciferase reporter construct (Fig. 5,A). Similarly, concentrations of 15S-HETE inhibiting PC3 proliferation also caused PPARγ-dependent transactivation (Fig. 5 A). BRL 49653 (1 μm) and 15S-HETE(10 and 30 μm) caused an approximately threefold and a greater than twofold increases in luciferase activity,respectively (n = 3). Similarly, BRL 49653 caused PPARγ-dependent transactivation in DU-145 cells. Tenμ m 15S-HETE caused a >2.5-fold increase in luciferase activity in similarly transfected DU-145 cells(n = 3, not shown).

In addition to transactivation of the reporter construct, the ability of these agents to up-regulate expression of native genes regulated by PPARγ was investigated. The gene for a-FABP contains PPRE elements in its promoter and is up-regulated by PPARγ agonists in adipocytes (26). Although a-FABP mRNA was not detected by Northern blots in untreated or vehicle-treated control cells at 3 days,concentrations of BRL 49653- and 15S-HETE-inhibiting proliferation of PC3 cells caused up-regulation of a-FABP mRNA expression at 3 days(Fig. 5 B).

Treatment of the breast cancer cell line 21PT with synthetic PPARγagonists increased expression of PPARγ mRNA (27), and the PPARγ agonist 15-deoxy-Δ^12,14-PGJ₂increased PPARγ2 but not PPARγ1 protein in Pca cell lines(17). Whether changes in a-FABP in BRL 49653- and 15S-HETE-treated cells may be attributable in part to increases in PPARγ was investigated. PPARγ mRNA was not detected by Northern analysis in control or treated cells (not shown). By quantitative real-time RT-PCR, we observed a reduction in PPARγ mRNA in PC3 cells treated for 3 days with either BRL 49653 or 15S-HETE, using primers detecting both isoforms. PPARγ mRNA copy numbers were 3.2- and 2.7-fold greater in vehicle control than in 3 μm BRL 49653- and 30 μm 15S-HETE-treated PC3 cells,respectively. In contrast, this was accompanied by a marked up-regulation in a-FABP mRNA copy number, which was 19.5- and 2.7-fold higher in 3 μm BRL 49653- and 30 μm15S-HETE-treated cells, respectively, compared with vehicle controls. These results further support that a-FABP up-regulation was attributable to ligand-dependent activation of PPARγ-mediated transcription, rather than attributable to increases in PPARγ itself.

15-LOX-2 is uniformly expressed in differentiated secretory cells of benign prostate, and 15S-HETE is formed by benign prostate tissues from exogenous AA as the major eicosanoid detected (8). In contrast, 15-LOX-2 expression and 15S-HETE formation are reduced in the majority of Pcas (8, 10). In the present study, we have shown that PPARγ mRNA is uniformly present in benign and malignant prostate and that 15S-HETE, similar to synthetic PPARγ agonists,activates PPARγ-dependent transcription and inhibits proliferation of PC3 Pca cells. 15S-HETE also activated PPARγ-dependent transcription in DU-145 cells, and others have recently reported inhibition of DU-145 proliferation by multiple PPARγ agonists (17, 28). These results support the hypothesis that reduced 15-LOX-2 may etiologically contribute to Pca development or progression by reduced expression of PPARγ-regulated genes.

PPARγ agonists have been shown to inhibit proliferation and potentially induce differentiation in carcinoma cell lines from multiple organs, including breast (27), colon(13), and bladder (15), in addition to their ability to promote adipocytic differentiation in benign adipocytes(26) and liposarcomas (29). Although these studies suggest PPARγ agonists as potential therapeutic agents in these malignancies, possible endogenous ligands in many, if not all, of these other organs have not been identified. The uniform expression of 15-LOX-2 and formation of 15S-HETE in benign prostate and the ability of 15S-HETE to activate PPARγ-dependent transcription in Pca cells indicates that 15S-HETE may represent a true endogenous ligand for PPARγ in prostate. Although concentrations of 15S-HETE added exogenously in the present study and in a previous study showing PPARγ activation (25) represent essentially pharmacological doses, the actual concentrations reaching the nucleus under these conditions or formed intracellularly in vivo are unknown.

The mechanisms whereby 15S-HETE activation of PPARγdependent transcription leads to growth inhibition remain to be characterized. PPARγ-mediated terminal differentiation in adipocytes leads to cell cycle arrest (30). In the present studies, BRL 49653 and 15S-HETE caused a slight delay in cell cycle progression in liquid cultures. The degree of these changes may be sufficient to account for the prominent effect observed in soft agar colony-forming assays. In previous studies, inhibition of PC3 proliferation by synthetic PPARγagonists was not accompanied by detectable effects on the cell cycle at 4 days in liquid culture (12), whereas Butler et al. recently reported that 15-deoxy-Δ^12,14-PGJ₂induced cell death in Pca cell lines with accumulation in S phase at 48 h (17). Whether this result reflects an agonist-specific effect or some other mechanism besides PPARγactivation for 15-deoxy-Δ^12,14-PGJ₂ is not clear. The activation of PPARγ by PGJ₂ in Pca cell lines has not been demonstrated in prior studies (12, 17), and a more direct effect at the plasma membrane of added exogenous oxidized lipids cannot be excluded. PGJ₂ is not a natural PG product, such that any demonstrated effects are more pharmacological that potentially physiological in vivo. This is in contrast to 15-HETE,because there is a high level of 15-LOX-2 catalytic activity in the benign prostate (8, 31). In contrast, inhibition of colon cancer cell lines in soft agar was accompanied by a more prominent increase in the G₀-G₁fraction than observed herein (13). Inhibition of bladder cancer cell lines by troglitazone in mitogenic assays was accompanied by possible G₁ arrest, as indicated by increased expression of cyclin-dependent kinase inhibitors, p21^WAF1/CIP1 and p16^INK4, and decreased expression of cyclin D1(15). The differential effects of PPARγ agonists on cell cycle and possible induction of differentiation are likely dependent on the specific tumor cell lines studied.

The function of 15-HETE formed from the previously characterized 15-LOX-1 has been uncertain. 15-LOX-1 is able to oxygenate polyunsaturated fatty acids esterified to phospholipids in cell membranes and to catalyze oxygenation at other positions, with formation of 12-hydroperoxyeicosatetraenoic acid and other lipid hydroperoxides. This has led to the speculation of its function in degradation of cell organelles by initiating lipid peroxidation as in maturation of reticulocytes and in oxidation of low-density lipoprotein during atherogenesis (32). However, this latter function may also depend at least in part on activation of PPARγ in macrophages (11, 25). We believe that the highly specific formation of 15S-HETE from phospholipase A₂released AA by 15-LOX-2 is in keeping with a ligand function dependent on structure and stereoconfiguration(7). Previous stereospecificity has been demonstrated for 8S- versus 8R-HETE activation of PPARα (33). In the present study, 10 μm 15S-HETE demonstrated a modestly greater ability to inhibit PC3 proliferation than 10 μm 15R-HETE. However, the pharmacological concentrations used in the present study are the final concentrations added to the media. Concentrations actually achieved at the nucleus, where PPARs are located, are unknown, such that stereospecificity of PPARγ activation may be greater than indicated in the present studies. This remains to be demonstrated in cell-free systems. There are inherent limitations of adding exogenous lipophilic agents to cells in culture. 15S-HETE made by benign prostate epithelial cells (8, 31) could be secreted and function in a paracrine manner. However, because 15-LOX-2, similar to other lipoxygenases, seems to undergo translocation to cell membranes on activation (31), it is unlikely that adding exogenous 15S-HETE completely mimics the intracellular sites of formation under more physiological conditions. 15S-HETE made at the nuclear membrane could function as a PPARγ ligand within the same cell in which it is formed. We have recently identified several cell strains established from prostatectomy specimens, which in contrast to commercially available tumor cell lines, express high levels of 15-LOX-2 mRNA and form 15S-HETE from exogenous AA.⁵Future studies selectively eliminating expression of 15-LOX-2 and PPARγ in 15-LOX-2-positive cells may shed more insight on other aspects of this signaling pathway in prostate.

The limited tissue distribution of 15-LOX-2 strongly suggests that its biology is crucial to normal prostate function and that its reduced expression is critical to Pca development. The mechanisms whereby 15-LOX-2 and PPARγ contribute to prostate cell function and how reductions in these pathways contribute to the malignant phenotype need clarification. 15S-HETE increased expression of the PPARγ-regulated a-FABP, an effect that was not previously observed in PPARγ-agonist-treated breast cancer or PGJ₂-treated Pca cell lines (17, 27). The activation of PPARγ-dependent transcription was not demonstrated in these latter Pca cell studies (17). In contrast, we readily demonstrated up-regulation of a-FABP by both a synthetic PPARγ-agonist and 15S-HETE in PC3 Pca cells by Northern analysis and real-time RT-PCR. Although a-FABP expression is associated with adipogenesis, it is clearly not limited in its expression to adipocytes. Its expression includes epithelial cells such as urothelium(15, 34). The function of this gene in prostate remains to be established, but it may contribute to secretory cell differentiation. As FABP overexpression inhibits proliferation in breast cancer cell lines (35), it is tempting to speculate that reduced expression in prostate because of reduced 15-LOX-2/15S-HETE activation of PPARγ could directly contribute to altered tumor cell proliferation or differentiation. Whether this or changes in expression of other PPARγ-regulated genes are more crucial in Pca development and progression remains to be established. In the present studies, increase in a-FABP expression was not attributed to increase in PPARγ, because BRL 49653 and 15S-HETE caused down-regulation of PPARγ under our experimental conditions. Concordant with these observations and further supporting a connection between these pathways in vivo, in a subset of eight snap-frozen benign and tumor pairs in which 15-LOX-2 mRNA was reduced in tumor versus benign, PPARγ mRNA quantitated by real-time RT-PCR was higher in tumor versus benign in seven of eight.⁴ These results indicate that 15S-HETE activation of PPARγ in benign prostate may result in down-regulation of PPARγ(or alternatively, that loss of the endogenous ligand results in“compensatory” up-regulation of this nuclear receptor in at least some tumors).

PPARγ is expressed in other tissues that express 15-LOX-2, further supporting a role for 15S-HETE formation in modulating expression of PPARγ-regulated genes. In addition to prostate, 15-LOX-2 mRNA is detected in skin, lung, and cornea (7, 31). 15-LOX-2 was cloned originally from human hair rootlets (7), and in the skin, 15-LOX-2 is expressed strongly in sebaceous glands.⁴Interestingly, PPARγ is also expressed in skin sebaceous glands, and synthetic PPARγ agonists are additive with androgens in inducing sebocyte differentiation (36). Hence, a 15-LOX-2-PPARγpathway may contribute to differentiation in other tissues besides prostate. That 15-LOX-2 is reduced in Pca and 15S-HETE may be an endogenous ligand for PPARγ in prostate, as demonstrated herein,further supports the rationale for evaluation of PPARγ-agonists in the treatment and/or prevention of Pca. During the preparation of this manuscript, Mueller et al. reported encouraging results from a Phase II clinical study in which use of the synthetic PPARγ-agonist troglitazone resulted in serum prostate-specific antigen decreases in 33% of patients with androgen-dependent Pca and 14% of patients with androgen-independent Pca (28). Also,understanding the relationship between androgens, 15-LOX-2, and PPARγin the prostate may help identify patient subsets that are particularly likely to benefit from such novel therapies.

We thank Dr. You-fei Guan (Vanderbilt University,Nashville, TN) for providing PPARγ and a-FABP probes and Jean McClure and Brent Weedman for help with figures.