Estrogen-induced Production of a Peroxisome Proliferator-activated Receptor (PPAR) Ligand in a PPARγ-expressing Tissue*

Peroxisome proliferation has been associated with carcinogenesis in the liver, and estrogen intake has been associated with increased risk of cancer in the hormone target tissues. Estrogen-induced peroxisome proliferation has been observed in an estrogen target tissue, the uropygial gland in the duck. To elucidate the molecular mechanism of this process, we previously isolated the cDNA of peroxisome proliferator-activated receptor γ1 (PPARγ1) from the duck uropygial gland and found that its expression was high exclusively in this tissue of duck. However, the nature of the ligand for PPARγ1 and how estrogen might enhance PPARγ1-regulated gene expression were not known. Here we demonstrate that estrogen treatment of animals enhanced the metabolism of arachidonic acid in the uropygial gland. Conversion of prostaglandin D2 to a metabolite was induced by estradiol treatment preceding peroxisome proliferation. High performance liquid chromatography and TLC analyses showed that the metabolite behaved chromatographically similar to prostaglandin J2 and Δ12-prostaglandin J2. Gas chromatography/mass spectrometry revealed a striking similarity of the metabolite to Δ12-prostaglandin J2, the only form among the J2 series whose natural occurrence has been detected. Furthermore, this metabolite was able to activate duck PPARγ1 to the same extent as the same concentrations of Δ12-prostaglandin J2 and 15-deoxy-Δ12,14-prostaglandin J2, whereas under the same conditions, prostaglandin D2 was not effective. The results suggest that estrogen treatment induced the formation of a prostaglandin D2 metabolite that activated duck PPARγ1, causing the induction of peroxisome proliferation in the duck uropygial gland.

series. TLC, HPLC, and GC/MS analyses provide compelling evidence suggesting that the prostaglandin D 2 metabolite is extremely similar to ⌬ 12 -prostaglandin J 2 . Our results demonstrate for the first time that in response to estradiol treatment, a potent activator of PPAR␥1 is produced in a tissue where PPAR␥1 is expressed at a high level and where peroxisome proliferation occurs in response to estrogen treatment.
The 1-2-year-old female mallard ducks (Anas platyrhynchos) obtained from Whistling Wings were kept in outdoor cages on a highenergy breeder ration and injected daily with 1 mg of 17␤-estradiol benzoate (17␤-estradiol benzoate was dissolved in 2.5% ethanol in olive oil). The control ducks were injected only with the ethanol/olive oil mixture.
Incubation and Isolation of Eicosanoids with Sep-Pak C 18 Cartridges-The frozen duck uropygial gland (200 mg) was homogenized with a Brinkmann homogenizer in 2.5 ml of 0.05 M Tris-Cl (pH 7.5). One ml of the homogenate was incubated at 37°C with constant shaking for various periods of time with 0.5 Ci of [ 3 H]arachidonic acid or with 2.0 Ci of [ 3 H]prostaglandin D 2 . For large-scale production of prostaglandin D 2 metabolites, 2-10 mg of unlabeled prostaglandin D 2 was included in the reaction mixture along with [ 3 H]prostaglandin D 2 . The reaction was terminated by adding 3 ml of ethanol, and the mixture was kept at room temperature for 5 min. Then, 16 ml of water was added to adjust the final concentration of ethanol to 15%. The mixture was centrifuged at 500 ϫ g for 10 min, and the supernatant was further acidified to pH 3.0 with 1 N HCl. Sep-Pak vac RC C 18 cartridges (Waters Associates) were used to isolate eicosanoids from the incubation mixtures as described (25). Briefly, the Sep-Pak cartridges were conditioned before loading the samples; the fractions of eicosanoids were eluted with methyl formate; and the solvent was evaporated under nitrogen. The dried material was dissolved in methyl formate or ethanol for subsequent studies.
Radioimmunoassay-The gland tissue was homogenized in ice-cold ethanol, and prostaglandins were recovered as described previously (26). The level of endogenous prostaglandin D 2 was measured with the [ 3 H]prostaglandin D 2 assay system (Amersham Pharmacia Biotech) using the competition method recommended by the manufacturer. The prostaglandin preparation was resuspended in 500 l of diluted assay buffer, and 30 l of this mixture was used in the assay. The standards or samples were incubated with tracer and antiserum at 4°C overnight, and the free prostaglandin D 2 was adsorbed onto dextran-coated charcoal. The concentrations of endogenous prostaglandin D 2 were obtained by referring to the standard curve.
TLC and HPLC Analyses-TLC was carried out on Whatman linear-K TLC plates with a solvent system of chloroform/methanol/acetic acid (90:5:2). The prostaglandins were visualized under UV light after spraying the plates with a 0.1% ethanolic solution of 2,7-dichlorofluorescein. The plates were also scanned to measure the radioactivity using a Berthold automatic TLC linear analyzer. In some cases, the silica gels representing TLC fractions were also scraped and directly assayed for radioactivity in a Beckman LS3801 scintillation counter.
HPLC was performed on a 5-m Lichrosorb RP-18 column (250 ϫ 4.6 mm; Alltech Associates, Inc.) using a Hewlett-Packard 1090 liquid chromatography system. The UV detection range was 190 -400 nm. The elution program was a 15-min linear gradient of 30% acetonitrile in water, a 5-min linear gradient to 50% acetonitrile, and a 25-min isocratic elution at that acetonitrile concentration. This was followed by a 5-min linear gradient to 73% acetonitrile and a final isocratic elution. All solvents contained 0.1% acetic acid. The flow rate was 0.5 ml/min, and the eluate was collected in 0.5-ml fractions. An aliquot from each fraction was assayed for radioactivity by scintillation counting.
GC/MS Analysis-The methyl ester and trimethylsilyl ether derivatives were prepared by treatment of the compounds with ethereal diazomethane for 10 min and further with 25 l of bis(trimethylsilyl-)trifluoroacetamide/dimethylformamide (1:1) for 10 min at room temperature. GC/MS analysis was carried out using a Hewlett-Packard 5890 gas chromatograph and a Hewlett-Packard 5972 mass spectrometer with a 30-m Hewlett-Packard 5ms column. Injections were made in the splitless mode at 70°C. After 3 min, the oven temperature was increased to 240°C at a rate of 30°C/min.
Plasmid and Transient Transfection Experiments-The mammalian expression vector of the chicken estrogen receptor (CERO) was kindly provided by Dr. Pierre Chambon (Institut de Chimie Biologique, Strasbourg, France). The pCMX-mRXR␣ vector and the mouse PPAR␣ expression vector pCMX-mPPAR␣ were kindly provided by Dr. Ronald M. Evans (Salk Institute, San Diego, CA). The EcoRI fragment of duck PPAR␥1 cDNA was subcloned to the EcoRI site of the mammalian expression vector pcDNA3 (Invitrogen). Two complementary 39-mer oligonucleotides representing the PPRE of the rat fatty acyl-CoA oxidase promoter with BglII site overhangs were synthesized, phosphorylated with T 4 polynucleotide kinase, and annealed to and subcloned to the BglII site of the pGL 2 promoter that contains an SV40 promoter upstream of the luciferase gene (Promega). The clone that contained three consecutive copies of the PPRE was identified by screening and sequencing and was used in transfection experiments as the PPREcontaining reporter vector.
NIH 3T3 cells were grown routinely in Dulbecco's modified Eagle's medium with high glucose and 10% calf serum and switched to Dulbecco's modified Eagle's medium without phenol red and with 2% dextrancoated charcoal-treated calf serum prior to transfection. On the next day, the 60 -80% confluent cells were transfected using LipofectAMINE reagent (Life Technologies, Inc.) in 35-mm plates. In each transfection reaction, 2 g of DNA was used, including 200 ng of each receptor or reporter expression plasmid and 1 g of ␤-galactosidase expression vector. pBluescript plasmid was used to make up the total amount of DNA to 2 g/reaction. After 5 h, the ligands were added and replaced on the following day. After 48 h, the cells were washed and harvested. The cell lysates were prepared by three to four cycles of freezing and thawing. Luciferase expression was standardized against an internal standard of ␤-galactosidase.

RESULTS
Test for Estrogen Activation of Duck PPAR␥1-We recently isolated a PPAR␥1 cDNA clone from the duck uropygial gland, and this PPAR␥1 was found to be highly expressed specifically in the duck gland tissue (18). To test whether 17␤-estradiol or any of its metabolites elicit a direct response with duck PPAR␥1, duck PPAR␥1 was expressed transiently in NIH 3T3 cells, and estrogens were tested as added ligands. The results show that at a high concentration (20 M), 17␤-estradiol consistently led to a higher response through duck PPAR␥1 than at a low concentration (0.2 M) (Fig. 1, A and B). The observed induction was usually ϳ2-fold. Other 17␤-estradiol metabolites, including estriol and estrone, were also tested, but no induction of reporter gene expression was observed (Fig. 1B). Cotransfection with retinoic acid X receptor ␣ caused a significant enhancement of the induction of the target gene caused by estradiol at all concentrations (Fig. 1A). Since estradiol caused only ϳ2-fold induction with PPAR␥1 alone, it appeared unlikely that direct estrogen activation of PPAR␥ was adequate to account for peroxisome proliferation in the uropygial gland.
Test of the Synergistic Response of Duck PPAR␥1 and the Estrogen Receptor to 17␤-Estradiol-Since the ER was found earlier to be expressed in the duck uropygial gland, 2 we tested the possibility that estradiol may be a ligand for a heterodimer of duck PPAR␥1 and the ER, which might in turn induce peroxisome proliferation. The addition of the ER to the transfection assays appeared to improve the induction of the target gene by various ligands, albeit to a small degree (Fig. 1C). The effect was greater when a high concentration (10 M) of 17␤estradiol was present. However, the presence of PPAR␥ and the ER gave a higher stimulation than that obtained with either alone, but the effect was additive rather than synergistic, suggesting that it is unlikely that the estradiol response involved heterodimers of PPAR␥ and the ER. The higher level of response to estradiol was probably due to its effect through binding to the ER, rather than conversion to other metabolites.
Stimulation of the Metabolism of Arachidonic Acid by Estro-gen in the Duck Uropygial Gland-Since it was reported that selective groups of eicosanoids were able to activate PPARs differentially (14), we examined the possibility that estrogen might induce the formation of a prostaglandin ligand for PPAR in the duck uropygial gland. Initially, we tested whether arachidonic acid metabolism in uropygial glands was enhanced by estrogen treatment of the animals. 3 H-Labeled arachidonic acid was incubated with the homogenates of glands from control animals and from animals treated for 9 days with estradiol, and the metabolic products were extracted and examined by TLC. The results ( Fig. 2) show that the homogenates of untreated uropygial glands generated polar metabolites from arachidonic acid. The extracts of the glands from estrogentreated ducks converted a much higher portion of arachidonic acid to metabolites, clearly showing that estrogen treatment significantly enhanced arachidonic acid metabolism. Effect of Estrogen Treatment on the Level of Endogenous Prostaglandin D 2 in the Uropygial Gland-Prostaglandin D 2 is the immediate precursor of the prostaglandin J 2 series, including ⌬ 12 -prostaglandin J 2 and 15-deoxy-⌬ 12,14 -prostaglandin J 2 , which have been reported to be activators of PPAR␥ (14). Therefore, we examined whether the level of endogenous prostaglandin D 2 in the duck uropygial gland changes as a result of estrogen treatment of the animals. Radioimmunoassays for prostaglandin D 2 levels after treatment of the animals with estradiol for different periods of time showed that after the first 0.1% (v/v) vehicle alone as a control or with peroxisome proliferators (4-chloro-6-(2,3-xylidino)-2-pyrimidinylthioacetic acid (Wy) and 5,8,11,14-eicosatetraynoic acid (ETYA)) or estrogens at the final concentrations as indicated. Each reaction was done at least in triplicate, and the average values of luciferase activity that have been normalized with ␤-galactosidase activity are shown.  3 H-Labeled arachidonic acid was incubated for 30 min with buffer (top) or with the homogenate of the uropygial gland from an untreated duck (middle) or from an estrogen-treated (9 days) duck (bottom). The metabolites of arachidonic acid were extracted, and the same amount of radioactivity from each sample was subjected to TLC as described under "Experimental Procedures"; the chromatograms were scanned with a Berthold TLC linear analyzer. 9 days of treatment, the level of prostaglandin D 2 decreased to less than half of the original level, and then the level started to go up, approaching the starting level with extended estrogen treatment of the animals when the peroxisome level is known to have reached a maximal level (Fig. 3) (2). The measured values are net levels of prostaglandin D 2 , reflecting both the production and the catabolism or clearance of prostaglandin D 2 . Therefore, the changes in the prostaglandin D 2 level could indicate decreased production or increased metabolism of prostaglandin D 2 . The composition of the gland secretion changed due to the estradiol treatment. First, long chain fatty acid esters began to appear together with the usual short chain acid esters, and subsequently, the entire secretion was composed of long chain esters prior to the appearance of diesters, the characteristic products of the peroxisomes. The observed changes in the level of prostaglandin D 2 could be directly related to estrogen-induced peroxisome proliferation in the duck uropygial gland since the lowest level of prostaglandin D 2 was reached after 9 days of estrogen treatment, preceding the appearance of 3-hydroxy fatty acid diesters produced by the peroxisomes (Fig. 3) (2).
Induction of the Enzymatic Conversion of Prostaglandin D 2 to a Metabolite in the Uropygial Gland by Estrogen Treatment of Ducks-To determine whether the decrease in prostaglandin D 2 caused by estrogen treatment represents increased conversion of the prostaglandin to metabolites, we incubated 3 Hlabeled prostaglandin D 2 with homogenates of glands and examined the products using TLC. The results revealed that prostaglandin D 2 was converted to a single major product in 30 min by homogenates of glands from animals treated with estrogen for 9 days (Fig. 4A). Very little product was detectable with buffer alone. A small amount of product was found after incubation with extracts of control glands from untreated animals. With the homogenates of glands from animals that had been treated with estrogen for 6 days, almost half of the prostaglandin D 2 was converted to the metabolite after 30 min of incubation (Fig. 4B); with boiled gland homogenates, virtually no conversion of the prostaglandin occurred, showing that the conversion of prostaglandin D 2 to the observed product was an enzyme-catalyzed process. The homogenates from animals treated with estrogen for 9 days converted prostaglandin D 2 The composition of the secretion from the uropygial gland was examined by TLC and is indicated for each period: short, esters of short chain fatty acids; long, esters of long chain fatty acids; diester, diesters of 3-hydroxy C 8 , C 10 , and C 12 acids. The prostaglandins from the gland were prepared, and the level of prostaglandin D 2 (PGD 2 ) was measured by radioimmunoassay as described under "Experimental Procedures." FIG. 4. Radio thin-layer chromatographic analysis of metabolites produced from 3 H-labeled prostaglandin D 2 by uropygial gland extracts from estradiol-treated and control ducks. A, 3 Hlabeled prostaglandin D 2 (PGD 2 ) was incubated for 30 min with buffer alone (top) or with the homogenate of the uropygial gland from an untreated duck (middle) or from an estrogen-treated (9 days) duck (bottom). B, 3 H-labeled prostaglandin D 2 was incubated for 30 min with the homogenate of the uropygial gland from an estrogen-treated (6 days) duck (top), with the same homogenate that had been boiled (middle), or with the homogenate of the uropygial gland from a longterm estrogen-treated (18 days) duck (bottom). In all cases, the prostaglandins from the incubation mixtures were extracted, and the same amount of radioactivity from each sample was subjected to TLC. The chromatograms were scanned with a Berthold TLC linear analyzer. almost completely to the metabolite, whereas with gland homogenates from animals treated for 18 days, the production of the prostaglandin D 2 metabolite decreased to the control level (Fig. 4, A and B). Thus, the conversion of prostaglandin D 2 to its metabolite during estrogen treatment was inversely correlated with the changes in the endogenous prostaglandin D 2 level, suggesting that the decreased level of prostaglandin D 2 probably was due to the enhanced conversion of prostaglandin D 2 to its metabolite.
It has been reported previously that during mating season, the composition of the lipids secreted by the female duck uropygial gland undergoes specific changes (2). The monoesters of short chain fatty acids are replaced by long chain fatty acid monoesters, followed by the appearance of unique diesters of 3-hydroxy fatty acids that are thought to be pheromones (2). We found that with gland tissue, which was producing short chain fatty acid monoesters, overnight incubation of the homogenate with prostaglandin D 2 did not result in detectable conversion to its metabolite (Fig. 5A). We tested the ability of the gland at each stage to produce the prostaglandin D 2 metabolite by incubating gland homogenates with 250 g of prostaglandin D 2 . With homogenates of glands producing short chain monoesters, very little metabolite was formed (Fig. 5A). With homogenates of glands that were at the stage preceding the appearance of long chain fatty acid monoesters, specific production of the prostaglandin D 2 metabolite was observed with overnight incubation (Fig. 5B). With the extracts of glands that were at the stage preceding diester formation, overnight incubation resulted in an increased level of conversion of prostaglandin D 2 (Fig. 5C), although these long incubations were not done for actual rate measurements. With the high level of prostaglandin D 2 (250 g), complete conversion was not seen in any case with overnight incubation. Compared with short incubation times, more minor products were obtained with the overnight incubation.
Identification of the Prostaglandin D 2 Metabolite-When the prostaglandin D 2 metabolite extracted from the duck uropygial glands was compared with various prostaglandin standards by TLC, the metabolite showed the same R F value as prostaglandin J 2 and ⌬ 12 -prostaglandin J 2 , but the R F was clearly different from those of the members of the F series (Table I). HPLC analysis showed that the duck metabolite was eluted at ϳ39 min, whereas prostaglandin J 2 and ⌬ 12 -prostaglandin J 2 were eluted between 39 and 40 min (Table I). To date, there are two known pathways for prostaglandin D 2 conversion to other prostaglandins. One is the in vitro spontaneous dehydration of prostaglandin D 2 into prostaglandin J 2 (27,28) and the further conversion of prostaglandin J 2 to ⌬ 12 -prostaglandin J 2 and 15-deoxy-⌬ 12,14 -prostaglandin J 2 catalyzed by albumin or human plasma (29,30). The other is the in vivo stereospecific reduction of prostaglandin D 2 to 9␣,11␤-prostaglandin F 2 catalyzed by NADPH-dependent 11-keto-reductase and the further conversion to 15-keto-9␣,11␤-prostaglandin F 2 and 13,14dihydro-15-keto-9␣,11␤-prostaglandin F 2 in the presence of NAD ϩ (31-34). The duck metabolite appeared to be chromatographically more similar to prostaglandin J 2 and ⌬ 12 -prostaglandin J 2 than to any other known metabolite.
To facilitate structural analysis of the prostaglandin D 2 metabolite produced by the gland extract, we used a larger amount of unlabeled prostaglandin D 2 along with a trace amount of 3 H-labeled prostaglandin D 2 in the overnight incubation study. HPLC fractionation of the products showed that the prostaglandin J 2 -like metabolite appeared to be the predominant product detected by UV absorbance and the presence of 3 H. The fractions showing the highest level of radioactivity were combined, and the UV spectrum was obtained. The UV spectrum showed a single absorbance peak at 225 nm, which was slightly shorter than the max (248 nm) of ⌬ 12 -prostaglandin J 2 , but much shorter than the max (307 nm) of 15deoxy-⌬ 12, 14 -prostaglandin J 2 (Fig. 6). When the prostaglandin FIG. 5. Radio HPLC analysis of the metabolites produced by the extracts that produce the different types of secretory lipids in the duck uropygial gland. Prostaglandin D 2 ((PGD 2 ; 250 g) with a trace amount of 3 H-labeled prostaglandin D 2 was incubated with the duck uropygial gland homogenate overnight. The prostaglandin extracts were further fractionated by HPLC, and the fractions were assayed for radioactivity. The glands were taken from ducks of different biological stages as indicated. D 2 metabolite was subjected to GC/MS as the trimethylsilyl ether derivative of the methyl ester, a major peak emerged with a retention time of ϳ19 min under conditions in which the standard ⌬ 12 -prostaglandin J 2 derivative had the same retention time, and prostaglandin J 2 showed a retention time of 17 min. The endogenous compound from the duck gland could be co-eluted with the product generated from exogenous prostaglandin D 2 during HPLC and GC. The mass spectrum of the prostaglandin D 2 metabolite isolated from the gland extract showed a striking similarity to that of standard ⌬ 12 -prostaglandin J 2 (Fig. 7 A and B). The standard ⌬ 12 -prostaglandin J 2 derivative had all the expected ions, except the molecular ion at m/e 420, under our experimental conditions. The prostaglandin D 2 metabolite also did not have an ion at m/e 420, but showed all of the fragment ions present in the spectrum of ⌬ 12 -prostaglandin J 2 , including ions at m/e 405 (M Ϫ 15, loss of ⅐CH 3 ), 349 (M Ϫ 71, loss of C 5 H 11 ), 320 (M Ϫ100, loss of C 5 H 11 CHO), and 173 ((CH 3 ) 3 SiϭO ϩ -CH 2 -C 5 H 11 ). However, the metabolite showed a molecular ion at m/e 492 that cannot be explained by the empirical formula of the ⌬ 12 -prostaglandin J 2 derivative. One possible explanation is that under our experimental conditions, the keto group of ⌬ 12 -prostaglandin J 2 was enolyzed and formed a trimethylsilyl derivative, producing the molecular ion at m/e 492. The trimethylsilylated methyl ester of dehydrated prostaglandin D 2 could also give an ion at m/e 492. The possibility that this ion is from an impurity present in the isolated metabolite cannot be ruled out. Based on all of our evidence, it is likely that the prostaglandin D 2 metabolite formed in the gland extract was very similar to ⌬ 12 -prostaglandin J 2 .
Activation of Duck PPAR␥1 by the Duck Prostaglandin D 2 Metabolite-To test whether the duck prostaglandin D 2 metabolite was able to activate PPAR␥1, which was highly expressed in the duck uropygial gland, we transfected NIH 3T3 cells with duck PPAR␥1 expression vector along with the luciferase re-porter plasmid that contains three copies of the well tested PPRE from the rat fatty acyl-CoA oxidase promoter, and the prostaglandin D 2 metabolite was added as ligand. The concentration of the added ligand was estimated based on its UV absorbance and the extinction coefficient of prostaglandin J 2 . We found that the metabolite at 10 M induced cell death (data not shown), as observed with the prostaglandin A 2 and J 2 series (35). At 2.8 M, this compound activated duck PPAR␥1 by 5-6-fold, and under the same conditions, similar activation was caused by the prostaglandin J 2 series (Fig. 8). Prostaglandin D 2 did not activate duck PPAR␥1 at 3 M. With our system, higher induction by 15-deoxy-⌬ 12,14 -prostaglandin J 2 was not observed, probably due to the toxicity of the compound to the cells or to its extremely short half-life in the aqueous environment. DISCUSSION We have investigated the molecular mechanism involved in the estrogen induction of peroxisome proliferation in the duck uropygial gland. Initially, we explored the possibility that estrogen can directly activate PPAR␥, which we found was uniquely and highly expressed in the gland, or can activate a heterodimer of PPAR and the ER. However, estrogen did not effectively function as a ligand for PPAR␥1 from the duck uropygial gland, indicating that estrogen induction of peroxisomes probably involves more than direct binding of PPAR or its heterodimer with the estrogen receptor. These findings are consistent with other observations that 17␤-estradiol did not bind directly to PPAR␥ to any extent (25) and that PPAR and the ER did not form a heterodimer in vitro (36). Thus, the induction of peroxisome proliferation by estrogen probably also involves a more indirect mechanism.
In our efforts to elucidate the mechanisms involved in the induction of peroxisome proliferation by estrogen, we examined PPARs, the nuclear receptors involved in the transcriptional enhancement of genes involved in peroxisome proliferation. We found that PPAR␥ is uniquely and highly expressed in the duck uropygial gland. The finding that prostaglandins can serve as ligands for PPAR␥ (14,15) raised the possibility that estrogen could indirectly induce peroxisome proliferation by affecting prostaglandin metabolism to generate a PPAR␥ ligand that could be involved in the transcriptional activation of genes involved in peroxisome formation. Estrogen is known to regulate prostaglandin production in target tissues such as the uterus (19 -24). Estradiol was reported to increase the release of arachidonate, the precursor of prostaglandins, via its effect on phospholipase A 2 (37,38). Steroids are known to regulate prostaglandin synthesis at the prostaglandin G/H synthase level (19,39) and at steps subsequent to the formation of prostaglandin G and H (21,22). Thus, it appeared possible that in the uropygial glands, estrogen might regulate prostaglandin metabolism, resulting in the increased production of a PPAR␥ ligand. In fact, we found that estrogen administration enhanced arachidonic acid metabolism in the uropygial gland, as indicated by increased conversion of labeled arachidonic acid to polar metabolites in gland homogenates. We did not attempt to resolve the variety of compounds generated from arachidonic acid. Instead, we tested for the effect of estradiol treatment on the fate of prostaglandin D 2 , a possible precursor of a PPAR␥ ligand.
Estradiol treatment of ducks caused a decrease in prostaglandin D 2 levels. This decrease most probably reflects its increased conversion to a metabolite. Estrogen induction of prostaglandin D 2 conversion to a metabolite was readily demonstrated with uropygial gland homogenates. The estrogen-induced metabolite of prostaglandin D 2 appears to be extremely similar to ⌬ 12 -prostaglandin J 2 . Structural analysis FIG. 6. UV spectrum of the duck prostaglandin D 2 metabolite. The metabolite was produced by overnight incubation with the gland extract from estradiol-treated ducks as described under "Experimental Procedures" and purified by HPLC. An ethanolic solution of the metabolite was used to record the UV spectrum. revealed that the metabolite behaved chromatographically similar to ⌬ 12 -prostaglandin J 2 in both TLC and HPLC analyses. GC/MS analysis revealed all the fragment ions that are also present in the spectrum of authentic ⌬ 12 -prostaglandin J 2 . Most important, the metabolite was able to activate PPAR␥1 as efficiently as ⌬ 12 -prostaglandin J 2 , whereas prostaglandin D 2 could not activate PPAR␥1. Thus, our results show for the first time that a natural ligand for PPAR␥ was produced locally in a PPAR␥-expressing tissue. Whether the natural ligand for PPAR␥ is ⌬ 12 -prostaglandin J 2 in other tissues instead of 15deoxy-⌬ 12,14 -prostaglandin J 2 remains to be elucidated. Other possibilities also remain open since antidiabetic thiazolidinediones are even better ligands for PPAR␥ than 15-deoxy-⌬ 12,14 -prostaglandin J 2 in terms of receptor binding as well as activation, and they are structurally different (14, 40 -42).
So far, only limited information is available on the metabolism of the prostaglandin J 2 series in vivo. The enzymatic reactions involved in the metabolism of the prostaglandin J 2 series are virtually unknown. Prostaglandin J 2 was reported to be produced by incubation of prostaglandin D 2 in buffer for days (27,28), and additional metabolites were shown to be produced from prostaglandin D 2 by human plasma or serum albumin (29,30). With antibody against chicken albumin as a probe, we detected albumin-like protein (ϳ69 kDa) in the duck uropygial gland; however, the level of this protein appeared not to be influenced by estrogen treatment (data not shown). Since the enzymatic conversion of prostaglandin to the metabolite was induced by estrogen treatment, it is unlikely that the albumin was the catalyst responsible for this conversion, unless estrogen caused some activation of the protein and consequently enhanced its ability to catalyze the reaction. The enzymology of the conversion of prostaglandin D 2 to ⌬ 12prostaglandin J 2 remains to be elucidated.
To date, studies on the regulation of prostanoid metabolism in the target tissues was mostly concerned with prostaglandins E 2 and F 2␣ because of their involvement in the biology of female reproductive systems (20 -24). In fact, the level of prostaglandin D 2 in rat uterus has been shown to decrease when the estrogen level is high (23). Since little was known about the metabolism of prostaglandin D 2 , this result was simply interpreted as the decreased synthesis and release of prostaglandin Our results raise the possibility that the decreased level of prostaglandin D 2 could be due to the increased metabolism of prostaglandin D 2 instead. Therefore, it would be interesting to reexamine the metabolism of prostaglandin D 2 in the female reproductive tissues and to determine whether the production of the prostaglandin J 2 series is involved in signal transduction in those tissues.
PPAR␥s have been shown to be involved in adipogenesis (43)(44)(45)(46)(47), plasma triglyceride metabolism (48), and inflammatory responses (8 -10), suggesting that they are of central importance in lipid metabolism and other diverse cellular signaling pathways. In our investigation of the molecular basis for estrogen-induced peroxisome proliferation, we found that two isoforms of PPAR␥ (␥1 and ␥2) are expressed in the duck uropygial gland, whereas only PPAR␥2 is expressed in the goose uropygial gland, where estrogen does not induce peroxisome proliferation. The unique high-level expression of PPAR␥1 in the duck uropygial gland, where estrogen does induce peroxisome proliferation, suggests that PPAR␥1 may be involved in the estrogen-induced peroxisome proliferation in this tissue, whereas PPAR␥2 is involved in accumulating lipids. As found in the promoters of adipose cell-specific genes, the DNA response elements for PPAR in the promoters of peroxisomal genes also consist of direct repeats with one-nucleotide spacing (5,44). Therefore, our result that production of the activator of PPAR␥ preceded peroxisome proliferation further raises the possibility that PPAR␥1 is responsible for estrogeninduced peroxisome proliferation in target tissues. The expression of human PPAR␥ in the placenta and ovaries has been reported (33). It is possible that PPARs, especially the ␥ isoforms, are also expressed in other tissues such as the uterus or breast. In fact, PPAR␥ expression was recently found in breast cancer cells (49,50). With the presence of both the ligand and receptor, it is possible that peroxisome proliferation might be induced by estrogen in such tissues. Further investigation of the expression of PPAR␥1 and its ligand in the female reproductive system might reveal possible estrogen-regulated peroxisome proliferation in such tissues.
The biological function of estrogen induction of peroxisome proliferation in the uropygial gland is to produce the 3-hydroxy fatty acid esters that serve as female pheromones during the mating season. Even though the oxidants produced by the peroxisomes in the tissue may cause DNA damage, cells in this holocrine gland produce secretory lipids, lyse, and empty their contents into the secretory lumen of the gland, making the DNA damage inconsequential. Thus, the holocrine gland can use estrogen-induced peroxisome proliferation for a biosynthetic purpose needed for a short time without any lasting adverse consequences. On the other hand, if peroxisome proliferation is induced by estrogen in the mammalian target tissues, such as the breast and reproductive organs, the resulting oxidants may cause DNA damage, leading to carcinogenesis, as postulated for tumorigenesis by liver peroxisome proliferators (1). If such is the case, the peroxisome proliferation may be only a transient event in the process of estrogen-induced carcinogenesis and may not be easily detected.