A Novel Pathway Involving Progesterone Receptor, 12/15-Lipoxygenase-derived Eicosanoids, and Peroxisome Proliferator-activated Receptor (cid:1) Regulates Implantation in Mice*

The 12/15-lipoxygenases (12/15-LOX) catalyze the ste-reo-specific oxygenation of arachidonic and linoleic acids into a complex series of signaling molecules, including the hydroxyeicosatetraenoic acids (HETEs) and hydroxyoctadecadienoic acids (HODEs). Our previous studies, using high density oligonucleotide microarrays, suggested a novel link between progesterone receptor (PR) signaling and 12/15-LOX-mediated fatty acid metabolism in preimplantation mouse uterus. In this paper, using PR knockout mice, we established that the transcripts encoding leukocyte-12/15-LOX (L-12/15-LOX) and epidermal-12/15-LOX (E-12/15-LOX) are indeed downstream targets of regulation by PR in the uterine surface epithelium. Maximal induction of both L- and E-12/15-LOX on the day of implantation resulted in a marked increase in the uterine levels of the eicosanoids, 12-HETE, 15-HETE, and 13-HODE. Mice with null muta-tion in L-12/15-LOX had significantly reduced uterine levels of arachidonic acid metabolites and exhibited a partial impairment

The steroid hormones progesterone (P) 1 and estrogen (E) play crucial roles during early pregnancy by coordinating a complex series of interactions between the implanting blastocyst and the uterus (1)(2)(3)(4)(5). In mice, implantation is initiated 4 days after fertilization when the blastocyst reaches the uterus (1,6). It is thought that the action of the steroid hormones during the preimplantation phase prepares the endometrium for the attachment of the blastocyst and the subsequent events leading to the establishment of pregnancy. The cellular effects of P and E are mediated through intracellular progesterone receptor (PR) and estrogen receptor, which are hormone-inducible transcription factors (7,8). Hormone-occupied PR or estrogen receptor triggers the expression of specific gene networks in different cell types within the uterus, and the products of these genes mediate the hormonal effects during implantation. In order to understand the molecular basis of this complex physiological process, it is critical to identify the steroid-regulated pathways that are induced or suppressed at the time of implantation and analyze their functional roles.
We recently used RU486, a well characterized antiprogestin, as a tool to uncover the PR signaling pathways in the pregnant uterus (9). RU486 counteracts the action of P by binding directly to PR and impairing its gene regulatory function (10,11). Administration of RU486 to pregnant rodents during the preimplantation phase effectively blocks implantation by inhibiting PR-dependent gene expression (12). By using oligonucleotide microarrays, we identified several potential PR-regulated genes whose expression was markedly altered by RU486 in the uterus at the time of implantation (9). Interestingly, two of the mRNAs repressed in response to RU486 were leukocyte-12/15lipoxygenase (L-12/15-LOX) and epidermal-12/15-lipoxygenase (E-12/15-LOX), which belong to a family of polyunsaturated fatty acid-metabolizing enzymes.
The lipoxygenases metabolize arachidonic and linoleic acids, which are major cellular substrates in the synthesis of biological mediator substances known as eicosanoids (13,14). There are three major types of mammalian lipoxygenase activity: 5-LOX, 12-LOX, and 15-LOX. The primary metabolites of arachidonic acid generated by the 5-LOX are the leukotrienes and lipoxines, whereas the 12-and 15-LOX enzymes produce hydroxyeicosatetraenoic acids (HETEs) (15, 16). Metabolism of linoleic acid by 12-and 15-LOX yields hydroxyoctadecadienoic acids (HODEs). In mice, there are three major isoforms of the 12-and 15 LOX: leukocyte (L)-type, epidermal (E)-type, and platelet (P)-type (15)(16)(17). Whereas the platelet-type predominantly produces 12-HETE from arachidonic acid, the leukocyte and epidermal types show dual specificity as they generate significant amounts of both 12-and 15-HETE metabolites and are referred to as 12/15-LOX (18). The 12/15-LOX-derived eicosanoids have been implicated in diverse inflammation-related and other physiological pathways such as lymphocyte activation and migration, thrombocyte aggregation, chemotactic stimulation of leukocytes, synaptic transmission, tumor cell metastasis, and cellular apoptosis (16 -19).
In this paper, we analyzed the roles played by the fatty acid metabolites generated by progesterone-induced 12/15-LOX enzymes in controlling uterine function during implantation. We identified peroxisome proliferator-activated receptor (PPAR␥) as a critical downstream target of these lipid mediator molecules. Most important, our studies revealed a functional link between steroid hormone action in the preimplantation uterus and the downstream events such as eicosanoid biosynthesis and activation of PPAR␥-dependent gene networks that regulate implantation.

MATERIALS AND METHODS
Reagents-Progesterone and AA-861 were purchased from Sigma. RU 38486 was a gift of Roussel-Uclaf, France. Rabbit polyclonal anti-L-12/15-LOX antibody and rosiglitazone were purchased from Cayman Chemical, Ann Arbor, MI. ELISA kits for 12-HETE; 15-HETE and 13-HODE were purchased from Assay Designs Inc., Ann Arbor, MI.
Animals and Tissue Collection-All experiments involving animals were conducted in accordance with the National Institutes of Health standards for the use and care of animals. The animal protocols were approved by the University of Illinois Institutional Animal Care and Use Committee. Female mice (CD-1 from Charles River, Wilmington, MA), in proestrus, were mated with adult males. The presence of a vaginal plug after mating was designated as day 1 of pregnancy. The animals were killed at various stages of gestation and the uteri collected. The uteri were freed of embryos by repeated flushing as described previously (9,57). The tissues were then flash-frozen and stored at Ϫ80°C. In order to examine the effects of RU486, mice on day 3 of pregnancy (4 p.m.) were injected with either vehicle (sesame oil) or RU486 (8 mg/kg body weight). The mice were killed 24 h after the injection on day 4 (4 p.m.) to collect the uteri. The PRKO and L-12/15-LOX KO mice were bred and homozygotes were confirmed by genotyping as described previously (20,29).
Ovariectomy and Hormone Treatments-Female mice were subjected to bilateral ovariectomy and, 2 weeks later, were injected subcutaneously with P (40 mg/kg body weight) or vehicle (sesame oil) as described previously (9). The animals were killed 16 h after final injection.
Delayed Implantation and the Vascular Permeability (Blue Band) Assay-To induce and maintain delayed implantation, mice were ovariectomized on day 3 of pregnancy and injected daily with P (40 mg/kg/ BW) from days 5-8. To terminate delayed implantation and induce blastocyst activation, the P-primed delayed implanting mice were given an injection of E (1 g/kg/BW) on the 3rd day of the delay (day 8). Mice were killed at 24 h after E injection.
For the vascular permeability assay, delayed mice were given an injection of E on the 3rd day of delay, and 24 h after E administration, Chicago blue dye solution was injected intravenously as described previously (6). Distinct blue bands, which arise along the uterus due to increased endometrial vascular permeability at the sites of blastocyst apposition, were counted.
Measurement of 12/15-LOX-derived Metabolites-Uteri were collected from pregnant mice at different days of gestation. The tissues were freed of embryos by repeated flushing and homogenized in 1.0 ml of 50 mM phosphate buffer, pH 7.4, using a homogenizer. The homogenates were centrifuged at 10,000 ϫ g for 20 min. The supernatant was assayed directly for 12-and 15-HETE and 9-and 13-HODE without any further extraction using ELISA kits from Assay Designs Inc., Ann Arbor, MI.
Northern Blot Analysis-Northern blot analysis was performed as described previously (9,57). Briefly, total RNA was extracted from mouse uteri using the RNA isolation kit from Promega. These extracted mRNAs were size-fractionated on formaldehyde-denaturing 1% agarose gels and transferred to Duralon-UV membranes (Stratagene, La Jolla, CA) according to the manufacturer's instructions. The blots were prehybridized at 42°C for 2-6 h in a buffer containing 50% formamide, 5ϫ SSC, 5ϫ Denhardt's solution, 0.5% (w/v) SDS, and 100 mg/ml herring sperm DNA (Sigma). Hybridization was performed overnight at 42°C in the same buffer containing radioactive probe at a final concentration of 1 ϫ 10 6 cpm/ml. The blots were washed twice in 2ϫ SSC, 0.1% SDS at room temperature for 15 min and twice in 0.2ϫ SSC, 0.1% SDS at 55°C for 15 min. The intensity of a signal on the autoradiogram was estimated by densitometric scanning. To correct for RNA loading, the filters were stripped of the radioactive cDNA probes by washing for 5 min in 0.1% SDS at 95°C. The blots were then reprobed with a 32 Plabeled glyceraldehyde-3-phosphate dehydrogenase (GAPDH) probe.
In Situ Hybridization-Uterine tissues from normal pregnant or delayed animals were collected and frozen. The in situ hybridization was performed as described previously (9,57). Briefly, tissues were fixed in 4% paraformaldehyde at 4°C. Cryostat sections were cut at 8 m and attached to 3-aminopropyltriethylsilane (Sigma)-coated slides. In situ hybridization was performed with digoxygenin (DIG)-labeled sense or antisense RNA probes complimentary to L-and E-12/15 lipoxygenase cDNAs. DIG-labeled RNA probes were synthesized from the cDNAs using T3 or T7 RNA polymerase and DIG-labeled nucleotides according to the manufacturer's specifications (Roche Applied Science). Prehybridization was carried out in a damp chamber at 55°C for 60 min in hybridization buffer (50% formamide, 5ϫ SSC, 2% blocking reagent, 0.02% SDS, 0.1% N-laurylsarcosine). Hybridization was carried out at 55°C overnight in a damp humidified chamber. To develop the substrate, sections were sequentially washed in 2ϫ SSC, 1ϫ SSC, and 0.1ϫ SSC for 15 min in each buffer at 37°C. Sections were then incubated with anti-DIG alkaline phosphatase-conjugated antibody. Excess antibody was washed away, and the color substrate (nitro blue tetrazolium salt and 5-bromo-4-chloro-3-indoylphosphate) was added. Slides were allowed to develop in the dark, and the color was visualized under light microscopy until maximum levels of staining were achieved.
Immunohistochemistry-Paraffin-embedded endometrial tissues were sectioned at 4 m and mounted on slides. Sections were deparaffinized in xylene, rehydrated through a series of ethanol washes, and rinsed in water. Endogenous peroxidase activity was blocked by incubating sections in 0.3% hydrogen peroxide in methanol for 30 min at room temperature. The slides were washed in water, and antigen retrieval was performed by immersing the slides in 0.1 M citrate buffer solution, pH 6.0, and subjected to microwave heating for 20 min. The sections were allowed to cool, washed in PBS for 20 min, and then incubated in a blocking solution containing 10% normal goat serum for 30 min before incubation in primary antibody overnight at 4°C. Polyclonal antibodies against L-12/15-lipoxygenase and PPAR␥ were diluted 1:4000 and 1:1000, respectively, for immunohistochemistry. Immunostaining was performed using Avidin-Biotin kit for rabbit primary antibody (Vector Laboratories, Burlingame, CA) and the 3-amino-9-ethylcarbazole chromogen. Sections were counterstained with hematoxylin, mounted, and examined under bright field. Red deposits indicate the sites of immunostaining.
Statistical Analysis-Statistical evaluation of the data representing the number of implantation sites in Figs. 6 and 7 were performed using the Student's t test. All data were calculated as mean Ϯ S.D. p Ͻ 0.05 were considered statistically significant.

P Acting via PR Regulates
Uterine Expression of L-and E-12/15-LOX mRNAs-By using DNA microarray analysis, we previously identified L-and E-12/15-LOX as potential targets of P regulation in the pregnant uterus (9). In the present study, we analyzed the regulation of uterine expression of L-and E-12/15-LOX by P and PR. First, we examined the effects of P administration to ovariectomized mice. Uteri were collected from animals 16 h after hormone injection, and total RNA was isolated from these tissues for Northern blot analysis. The blot was probed with a 32 P-labeled cDNA representing either L-12/ 15-LOX or E-12/15-LOX or GAPDH (as a control) gene. As shown in Fig. 1A, mRNA transcript of L-or E-12/15-LOX was undetectable in the uteri of ovariectomized mice (lane ϪP). Both mRNAs were, however, markedly induced after treatment with P (lane ϩP), indicating that P regulates the uterine expression of both 12/15-LOX genes.
To establish whether the P regulation of L-and E-12/15-LOX genes is mediated through PR, we analyzed the expression of these mRNAs in the uteri of wild type (WT) and PR knockout (PRKO) mice (20). Female WT and PRKO mice were subjected to ovariectomy, and 2 weeks following this procedure mice were treated with either vehicle or P. After P treatment, the uteri were collected, and uterine sections were subjected to in situ hybridization using L-or E-12/15-LOX-specific probe. No expression of either 12/15-LOX gene was observed in the uteri of ovariectomized WT or PRKO mice treated with vehicle alone (Fig. 1, B, panels a and d; C, panels a and d). A marked induction of both L-and E-12/15-LOX mRNAs was induced in the surface and glandular epithelial cells of WT uteri upon administration of P ( Fig. 1, B, panel b, and C, panel b). A low level of L-12/15-LOX mRNA was also observed in the stromal cells. In contrast, no L-or E-12/15-LOX mRNA was detectable in the uteri of ovariectomized PRKO mice upon P treatment ( Fig. 1, B, panel e and C, panel e). These results indicated that P regulation of L-and E-12/15-LOX gene expression in the uterus is mediated via PR.

Induction of L-and E-12/15-LOX Gene Expression in the Pregnant Uterus
Coincides with the Time of Implantation-We next monitored the expression profiles of L-and E-12/15-LOX mRNAs in mouse uterus during early pregnancy. For this purpose, we performed Northern blot analysis employing uterine RNAs from days 1-6 of gestation. As shown in Fig. 2A, top and middle panels, the amounts of uterine L-12/15-LOX or E-12/15-LOX transcripts were low on days 1-3 of pregnancy. A marked rise in the levels of both of these transcripts was observed on day 4, the day of implantation, which then declined again to barely detectable levels by day 5 of pregnancy. These results indicated that a transient surge of L-and E-12/15-LOX mRNA expression occurs in the uterus on day 4 of gestation, and this event precisely coincides with the time of implantation. Interestingly, no transcript of either 5-LOX or platelet 12-LOX was detected in the uterus during the peri-implantation period (data not shown).
We next examined the spatial expression profile of the L-and E-12/15-LOX in the uterus during early pregnancy. We employed immunohistochemistry to analyze the expression of L-12/15-LOX protein (Fig. 2B). Uterine sections from animals on gestation days 2, 4, and 6 were probed with an antibody specific for L-12/15-LOX. We observed only weak staining in sections obtained from animals on days 2 and 6 of pregnancy (panels a and c). In contrast, an intense staining was observed in the surface and glandular epithelial cells of the uterine sections obtained from day 4 pregnant animals (panel b). Sections of uteri (day 4) incubated with control serum did not exhibit any specific immunostaining (panel d). L-12/15-LOX was, therefore, maximally expressed on the day of implantation, and this expression was restricted to the uterine epithelium. Interestingly, our previous studies indicated a similar pattern of expression for PR in the luminal epithelium (9). Although a low level of PR was detected in the luminal epithelial cells on either day 2 or 6 of pregnancy, an intense expression of this receptor was observed in these cells on days 3-4.
We monitored E-12/15-LOX mRNA expression by in situ hybridization using an antisense RNA probe. No specific signal was detected in the uterine sections on days 2 and 6 of pregnancy (Fig. 2C, panels a and c). A strong hybridization signal was, however, observed in the glandular and surface epithelial cells of the uterine sections obtained from day 4 pregnant mice (panel b). A control uterine section (pregnant, day 4) hybridized with the corresponding sense RNA probe did not exhibit any signal (panel d). These results confirmed that E-12/15-LOX mRNA is transiently expressed in the uterine epithelial cells of FIG. 1. PR regulates L-and E-12/15-LOX mRNA expression in uterine tissue. A, mice were subjected to bilateral ovariectomy, and 15 days later were treated with vehicle or P (40 mg/kg BW) for 3 consecutive days. Uteri were collected 16 h after the last injection. Total RNA was prepared and analyzed by Northern blotting followed by hybridization with 32 P-labeled cDNA probes for L-12/15-LOX, E-12/15-LOX, and GAPDH. Lane ϪP represents RNA from uteri of animals injected with vehicle only after ovariectomy; lane ϩP represents RNA from uteri of animals injected with P. B and C, female PRKO and WT mice of same genetic background (strain 129) were ovariectomized and treated with vehicle or P as described above. Twenty four hours after final P injection, animals were killed and uteri isolated, and in situ hybridization was performed as described under "Materials and Methods." A digoxygenin-labeled antisense RNA probe specific for L-12/15-LOX or E-12/ 15-LOX was employed. Sections marked by a, b, d, and e represent WT and PRKO uteri hybridized to an antisense L-12/15-LOX (A) or E-12/ 15-LOX RNA (B) probe. a, WT uterus treated without vehicle; b, WT uterus treated with P; c, WT uterus treated with P and hybridized to a sense L-12/15-LOX (B) or E-12/15-LOX (C) RNA probe; d, PRKO uterus treated without P; e, PRKO uterus treated with P. Section marked by L indicates luminal epithelium. pregnant mice overlapping the window of implantation.
Production of 12/15-LOX-derived Metabolites Is Regulated by P at the Time of Implantation-The L-or E-12/15-LOX metabolizes arachidonic acid to generate 12-HETE and 15-HETE (13)(14)(15). These enzymes also produce 9-HODE and 13-HODE from linoleic acid. As a first step toward understanding the role of these HETE and HODE metabolites in implantation, we investigated (i) whether these metabolites are induced in the pregnant uterus during implantation and (ii) whether P regulates their production.
To answer the first question, we determined the uterine profiles of 12-HETE, 15-HETE, 9-HODE, and 13-HODE during early pregnancy. We isolated uteri from pregnant mice on days 2-6 of gestation and determined the concentrations of these metabolites in uterine homogenates by ELISA. We observed that the level of 12-HETE, 15-HETE, and 13-HODE rose sharply and transiently on day 4, the day of implantation (Fig.  3). In contrast, the level of 9-HODE stayed relatively low (less than 50 ng per mg of protein) throughout early pregnancy, and no pronounced alteration in its level was seen during implantation (data not shown). The implantation stage-specific increase in the levels of 12-HETE, 15-HETE, and 13-HODE is consistent with the transient induction of 12/15-LOX gene expression in the uterus on day 4 of gestation ( Fig. 2). We observed that the peak level of 12-HETE was ϳ4and 10-fold higher than those of 13-HODE and 15-HETE, respectively ( We also examined whether P regulates the transient surge of 12-HETE, 15-HETE, and 13-HODE during implantation. We measured these metabolites in the uteri of day 4 pregnant mice treated with or without RU486. As shown in Fig. 4, administration of RU486, which down-regulates the expression of Land E-12/15-LOX mRNAs in the pregnant uterus, also strongly suppressed the levels of 12-and 15-HETEs and 13-HODE in this tissue. These results indicated that P acting via PR promotes the uterine synthesis of L-and E-12/15-LOX enzymes and their metabolic products in an implantation stage-specific manner. AA-861 Blocks the Generation of 12/15-LOX-derived Metabolites during Implantation-To address 12/15-LOX function in the pregnant uterus, we used AA-861, a coenzyme Q-like compound that is reported to inhibit LOX activity in certain tissues without affecting the other arachidonic acid-metabolizing enzymes COX-1 and COX-2 that produce various prostaglandins (21)(22)(23). To test the specificity of this inhibitor, we initially examined its effects on both COX and 12/15-LOX activities in the uterus at the time of implantation. Administration of AA-861 on the day 3 of gestation strongly inhibited the levels of 12-HETE, 15-HETE, and 13-HODE generated by the 12/15-LOX enzymes in the day 4 pregnant uterus (Fig. 5, left and two center panels). Treatment with AA-861, however, had no significant effect on the production of prostacyclin (prostaglandin I 2 ) by the COX pathway (Fig. 5, right panel). Our results therefore showed that AA-861 is a specific inhibitor of 12/15-LOX activity in the uterus.
Inhibition of 12/15-LOX Function by AA-861 Impairs Implantation-We next examined the effects of the LOX-specific inhibitor, AA-861, on the implantation process. Previous studies indicated that 12/15-LOX enzymes might influence ovarian function (24 -26). Interference with ovarian function during pregnancy may affect uterine receptivity and implantation. To avoid this possibility, we examined the effects of AA-861 on uterine function using mice that were subjected to delayed implantation (27,28). In these mice, the ovaries were surgically removed (ovariectomy) on the morning of day 4 of gestation.

FIG. 2. Expression profiles of L-and E-12/15-LOX mRNAs in mouse uterus during early pregnancy.
A, total RNA (20 g) isolated from uteri of animals at days 1-6 of gestation was subjected to Northern blot analysis. Top and middle panels represent the patterns of signals obtained after hybridization with 32 P-labeled L-12/15-LOX and E-12/15-LOX probes, respectively. The bottom panel shows the same blot after hybridization with a control 32 P-labeled GAPDH probe. B, uterine sections were obtained from mice at pregnancy days 2, 4, and 6, and stained with an antibody specific for L-12/15-LOX. Panels a-c represent sections of uteri at gestation days 2, 4, and 6, respectively. An intense signal was localized in the luminal and glandular epithelia of day 4 pregnant uterus (panel b). Panel d represents a day 4 uterine section incubated with control serum. L and G indicate luminal and glandular epithelium, respectively. C, uterine sections were obtained from mice at pregnancy days 2, 4, and 6 and subjected to in situ hybridization. The hybridization was performed employing a 300-bplong digoxygenin-labeled antisense RNA probe specific for E-12/15-LOX. Panels a-c represent sections of uteri at gestation days 2, 4, and 6, respectively. An intense signal was localized in the luminal and glandular epithelia of day 4 pregnant uteri (panel b). Panel d represents a day 4 uterine section hybridized with the corresponding sense RNA probe of equal length.
These animals were treated with P alone for two consecutive days so that the blastocysts remained viable, but its attachment to the uterine epithelium would not occur in the absence of E. On the 3rd day, upon administration of E along with P, the delay was terminated, and implantation was induced. The removal of the ovaries in the delayed implanting mice ensured that any impact of AA-861 on implantation would be independent of any ovarian effect it may exert.
We first ascertained that the L and E-12/15-LOX genes are expressed in the uterus under the conditions of delayed implantation. Northern blotting (Fig. 6, left panel) showed that L-12/ 15-LOX mRNA was expressed, as expected, in the P-treated delayed uteri. The level of its expression in the delayed uterus after P treatment for 24 and 48 h was comparable with that in the normal pregnant (day 4) uterus (compare lanes 1-3). The mRNA level increased modestly at 12 h following an implantation-inducing dose of E (lane 4). Interestingly, the L-12/15-LOX mRNA declined at 24 h after E treatment, thus mimicking the post-implantation decline of this mRNA on day 5 of normal pregnancy (lane 5). We observed a similar expression profile for E-12/15-LOX mRNA, although at a relatively lower level, during delayed implantation (data not shown).
To confirm that administration of AA-861 inhibits 12/15-LOX activity in the uterus under the conditions of delayed implantation, we measured uterine 12-HETE levels in the uteri of treated animals (Fig. 6, middle panel). Ovariectomized pregnant animals were divided into three groups. One group was injected with P (for 3 days) plus E on the 3rd day but received no AA-861. A second group received the same regimen of P and E and also received AA-861 at 0.5 mg per mouse per day of the delay. A third group received P and E and a higher dose of AA-861 at 2 mg per mouse per day of the delay. No appreciable decline in uterine 12-HETE level was observed when AA-861 was used at 0.5 mg/day/mouse. We, however, found a sharp decline in the level of this metabolite in response to AA-861 at 2 mg/day/mouse, indicating that the drug is as effective under delayed implantation conditions as in normal pregnancy.
We then examined the effects of AA-861 on implantation. The number of implantation sites in each treatment group was assessed at 24 h after E administration by counting the blue bands along the uterus following injection of Chicago blue B dye solution as described under "Materials and Methods." It is well established that in the mice, one of the earliest (day 5) macroscopic evidence for the initiation of implantation is an increased endometrial vascular permeability at the sites of blastocyst apposition (6). This increased vascular permeability can be visualized as discrete blue bands along the uterus immediately after an intravenous injection of Chicago blue B dye solution. As shown in Fig. 6, right panel, when mice were injected with a low dose (0.5 mg/day/mouse) of AA-861, which did not appreciably inhibit 12/15-LOX activity, we observed no significant decline in the number of implantation sites compared with animals that have received only P plus E. Interestingly, when mice were administered with the higher dose (2 mg/day/mouse) of the inhibitor, which efficiently blocked 12/15-LOX activity, we observed a dramatic decline in the number of implantation sites. The mice that were injected with the higher dose of AA-861 harbored only 2 Ϯ 0.8 implantation sites compared with the 11 Ϯ 2 sites that were normally generated in mice without administration of the drug (Fig. 6, right panel). This reduction was found to be statistically significant (p Ͻ 0.05). Taken together, these results clearly indicated that inhibition of uterine L-and E-12/15-LOX activities during early pregnancy compromises uterine functions during implantation.
Partial Reduction of Implantation Sites in the L-12/15-LOX Null Mice-We also examined the role of 12/15-LOX enzymes in implantation by using an L-12/15-LOX knockout (KO) mouse model (29). L-12/15-LOX KO mice and WT mice of the same genetic background (strain C57/BL6) were subjected to delayed implantation as described above. The metabolite levels and the number of implantation sites in the delayed uteri of WT and L-12/15-LOX KO were compared. As shown in Fig. 7, the uterine level of 12-HETE in the L-12/15-LOX KO mice was ϳ50% less compared with that in the WT C57/BL6 mice (top panel, lanes WT versus KO). The numbers of implantation sites in L-12/15-LOX KO mice were also significantly reduced (ϳ40%) compared with WT mice (Fig. 7, bottom panel, lanes WT versus  KO). Our results thus confirmed a link between the level of endogenous 12/15-LOX-derived metabolites and the number of implantation sites. That the L-12/15-LOX KO mice displayed a partial but not a total impairment in implantation is likely due to the residual 12/15-LOX activity contributed by E-12/15-LOX in the pregnant uterus. As shown in Fig. 1, the E isoform is expressed in uterine epithelial cells during implantation, and it generates the same 12-or 15-HETE and 13-HODE metabolites, although the overall level of the metabolites is substantially lower than that in the WT uteri.
We also tested whether the suppression of the residual 12/ 15-LOX activity in pregnant L-12/15-LOX KO mice by AA-861 further impairs delayed implantation. We observed that administration of only 0.5 mg of AA-861 effectively blocked the generation of 12-HETE in the uteri of L-12/15-LOX KO mice (Fig. 7, top panel, lanes WTϩAA versus KOϩAA). Consistent with our observation in CD-1 mice (Fig. 6), when WT mice of C57/BL6 strain were administered with 0.5 mg of AA-861, we observed only a marginal decline in the number of implantation sites (Fig. 6, bottom panel,

lanes WTϩAA versus KOϩAA).
Interestingly, when the same low dose of AA-861 (0.5 mg) was given to the L-12/15-LOX KO mice, we observed a dramatic decline (ϳ90%) in the number of implantation sites. It is likely that a 0.5-mg dose of AA-861, which is insufficient to inhibit the activity of both L-and E-12/15-LOX enzymes in the WT animals, is able to adequately block the E-12/15-LOX activity in the L-12/15-LOX KO mouse. Collectively, these results support our hypothesis that the L-and E-12/15-LOX enzymes are critical regulators of implantation.

Identification of Genes Regulated by the 12/15-LOX Pathway in the Pregnant Uterus Using Oligonucleotide Microarrays-
How do the 12/15-LOX-derived metabolites regulate implantation? We postulated that the metabolites generated by the 12/15-LOX enzymes in the uterus during early pregnancy act as activators of a downstream transcriptional factor, which triggers the expression of specific gene networks in various uterine compartments. Therefore, we examined changes in uterine mRNA profiles in response to AA-861 by using high density oligonucleotide arrays (GeneChip, Affymetrix). Mice were ovariectomized on day 4 of pregnancy and divided onto two groups, and each group was subjected to delayed implantation by sequential treatment with P and E as described above (Fig. 6). Whereas one group was also treated with AA-861 (2 mg/day/mouse), the other served as an untreated control group. Mice were killed 12 h after E treatment. Total RNA was isolated from uteri collected from mice treated with or without AA-861, and the microarray analysis was performed by using a set of arrays that contained oligonucleotides corresponding to ϳ6000 known mouse genes and ϳ6000 unnamed expressed sequence tags (ESTs) as described previously (9). We applied a threshold of a 3-fold change in expression level between AA-861-treated samples and untreated controls for identifying putative 12/15-LOX-regulated mRNAs. Applying this stringent cut-off, we identified 27 known genes plus several unnamed EST tags whose expression was inhibited in the uterus at the time of implantation in response to the 12/15-LOX inhibitor (Table I). A total of 16 genes were up-regulated Ͼ3-fold in response to this drug (data not shown). These included small proline-rich protein 2E and several EST tags.
We verified the results of microarray analysis by performing Northern blot analysis using cDNAs corresponding to a number of genes such as kallikrein 1, kallikrein 5, epidermal growth factor-binding protein A, and glutathione peroxidase (listed in Table I) as probes. For all of these genes, mRNA signals of marked intensity were observed in the delayed implanting uterus in the absence of AA-861, but these signals were undetectable in the presence of AA-861 (Fig. 11, top and   middle panels, lanes 1-3, data not shown). These results provided initial validation that the genes identified by the microarray method are indeed regulated by the 1215-LOX inhibitor.
The 12/15-LOX-derived Metabolites Function as Activating Ligands of PPAR␥-Certain 12/15-LOX-derived metabolites, such as 9-HODE and 13-HODE, have been reported previously (30 -32) to bind to PPAR␥ and activate its gene regulatory function in vitro. As a first step to test the possibility that gene expression by the 12/15-LOX-derived eicosanoids in the pregnant uterus might be mediated by PPAR␥, we examined the ability of various metabolites of 12/15-LOX to induce transcriptional activity of PPAR␥ in cultured cells using a transient transfection protocol. CV1 or Ishikawa endometrial cells were transfected with a reporter construct in which a PPAR-response element is linked to a luciferase reporter gene. Because PPARs are known to bind to DNA as stable heterodimers with Middle panel, the ovariectomized pregnant (day 4) mice were divided into three groups. One group received P for 2 days and then P plus E on the 3rd day. The second group received the same regimen of P and E plus AA-861 (0.5 mg/mouse/day). The third group received the same regimen of P and E plus AA-861 (2 mg/mouse/day). The 12-HETE levels were measured in delayed mouse uteri obtained from different treatment groups. The means of three independent determinations Ϯ S.D. are shown. Right panel, the number of implantation sites in different treatment groups was determined by monitoring increased endometrial vascular permeability at the implantation sites at 24 h after E administration. This was visualized as discrete blue bands along the uterus immediately after an intravenous injection of a Chicago blue B dye solution as described previously (6). The means of three independent determinations Ϯ S.D. are shown.
the retinoid X receptor (RXR), we transfected the cells with both PPAR␥ and RXR␣ expression vectors.
We found that the transfected receptors did not significantly activate reporter gene expression in CV1 cells in the absence of a PPAR␥ ligand (Fig. 8, 1st lane). Addition of ciglitazone, a known PPAR␥ agonist, led to a remarkable ϳ50-fold induction of luciferase activity (2nd lane). Interestingly, addition of 12-or 15-HETE to these cells led to an ϳ20-fold induction of PPAR␥dependent gene activation. 9-and 13-HODE also activated transcription by PPAR␥ but to a lesser extent (ϳ15-fold). Essentially similar results were obtained with the Ishikawa cells (data not shown). These results demonstrated that the 12/15-LOX-derived fatty acid metabolites function as potent activators of PPAR␥ in cells expressing this receptor.
Expression of PPAR␥ in the Preimplantation Mouse Uterus-To explore whether PPAR␥ could be a target of 12/15-LOXderived eicosanoids in the uterine tissue, we analyzed its expression in mouse uterus during early pregnancy. We first monitored PPAR␥ mRNA in non-pregnant as well as pregnant uteri on days 2, 4, and 6 of gestation by semi-quantitative reverse transcriptase-PCR. The PPAR␥ transcripts were undetectable in non-pregnant uterus (Fig. 9, left, lane NP). A marked induction in these transcripts was observed on day 2, and this expression continued on days 4 and 6 of pregnancy (lanes D2, D4, and D6). The induction of PPAR␥, therefore, temporally coincided with the peri-implantation phase of pregnancy.
We also investigated the site(s) of expression of PPAR␥ protein in the pregnant mouse uterus by employing immunocytochemistry. Uterine sections from days 1, 4, 5, and 6 pregnant animals were stained with an antibody specific for PPAR␥. Distinct nuclear staining of PPAR␥ protein was localized in the luminal epithelial cells as well as in the stromal cells on day 1 (Fig. 9, right, panel A). The stromal staining intensified on day 4 (panel B). The level of PPAR␥ protein declined in both cell types, especially in the epithelium, following implantation (panels C and D). The expression of PPAR␥ in the epithelium and stroma of pregnant uterus therefore overlapped with the window of implantation.
Treatment with a PPAR␥ Agonist Reverses Inhibition of Implantation by AA-861-If PPAR␥ mediates the functional effects of 12/15-LOX-derived metabolites in the preimplantation uterus, then administration of a synthetic agonist of this receptor may by-pass the AA-861-mediated inhibition and may allow reversal of the impairment in implantation. To test this possibility, rosiglitazone, a potent agonist of PPAR␥, was administered to AA-861-treated delayed implanting mice, and the effect on steroid-induced vascular permeability was assessed as described above (Fig. 6). Strikingly, treatment with rosiglitazone efficiently reversed the impairment of implantation in the AA-861-treated mice (Fig. 10). These results strongly suggested that the AA-861-mediated inhibition of implantation is indeed due to a block in the generation of endogenous ligands of PPAR␥, and it can be by-passed by supplementing with a synthetic agonist of this receptor.
Target Genes of 12/15-LOX-derived Metabolites Are Also Regulated by a PPAR␥ Agonist-If one or more 12/15-LOX-derived metabolites function as endogenous ligands of PPAR␥, then one would expect that at least some of the putative 12/15-LOX target genes identified by the microarray analysis (Table I) may also respond to a synthetic PPAR␥ agonist. Therefore, we tested the effects of rosiglitazone on gene expression in pregnant mice treated with AA-861. As shown in Fig. 11 These results provided strong support to our hypothesis that at least certain of the biological actions of the 12/15-LOX-derived eicosanoids in pregnant uterus are mediated by activation of PPAR␥-regulated gene networks by these ligands. FIG. 7. Impaired implantation in L-12/15-LOX KO mice. Ovariectomy was performed in pregnant L-12/15-LOX KO and WT C57/BL6 mice. Delayed implantation was induced in these mice as described in the legends to Fig. 6. Mice were treated with or without 0.5 mg of AA-861/mouse/day starting from the day of ovariectomy and steroid replacement. Top panel, the 12-HETE levels were measured in delayed mouse uteri obtained from WT, WTϩAA-861, KO, KOϩAA-861 treatment groups. The means of three independent determinations ϮS.D. are shown. Bottom panel, the number of implantation sites in different treatment groups was determined by monitoring increased endometrial vascular permeability at the implantation sites at 24 h after E administration. The means of three independent determinations Ϯ S.D. are shown.

DISCUSSION
In this paper, we identified the L-and E-12/15-LOX genes as novel downstream targets of the PR pathway in the uterus. The 12/15-LOX enzymes metabolize arachidonic and linoleic acids into HETEs and HODEs, which are important biological mediator substances (13). These metabolites have been implicated in the regulation of several important inflammatory conditions including airway and glomerular inflammation as well as atherosclerosis (13)(14)(15)(16). Although the HETEs and HODEs oppose inflammatory responses in certain tissues, they are also associated with proinflammatory effects such as neutrophilic and eosiniphilic chemotaxis in other tissues (13)(14)(15)(16). Arachidonic and linoleic acids are also metabolized by the cyclooxygenase family to produce prostaglandin derivatives, which are well known mediators of inflammation, coagulation, and vasodila-tion (25). It is well documented that prostaglandins play essential roles in female reproduction. Most notably, COX-2 null mice exhibit impaired ovulation and implantation (33). In contrast, relatively little is known about the function of 12/15-LOX-derived eicosanoids in female reproduction. Although a role of the 12/15-LOX enzymes in controlling ovarian function during the ovulatory process has long been speculated (23,24,26), little is known about the hormonal regulation and function of LOX-derived eicosanoids in the female reproductive tract.
The expression of L-and E-12/15-LOX in pregnant mice temporally coincided with the preparatory events leading to implantation, suggesting that these genes are potential regulators of this process. We used AA-861, a selective pharmacological inhibitor of the LOX enzymes, to analyze the function of 12/15-LOX during implantation. Although the treatment of  pregnant uteri with AA-861 led to a marked suppression of the generation of 12/15-LOX-derived metabolites 12-and 15-HETEs and 13-HODE, this drug did not affect the production of prostaglandins by the COX enzymes (Fig. 5). AA-861 is also known to inhibit 5-LOX activity in certain tissues. No expres-sion of 5-LOX mRNA was, however, detected in pregnant uterus by Northern blotting (data not shown). These results, which indicated that AA-861 specifically blocks the generation of 12/15-LOX-derived metabolites in the pregnant mouse uterus, allowed us to examine the impact of this inhibitor on uterine functions during implantation.
To avoid the possibility that the effects of AA-861 on the ovary may influence its action in the pregnant uterus, we used a delayed implantation mouse model in which ovaries are removed during early pregnancy, and implantation is induced by exogenous administration of E and P (27,28). In this ovariectomized model, AA-861 was administered after the embryos have developed to the blastocyst stage. Our studies revealed that AA-861 treatment leads to a complete block in implantation in delayed mice. These results clearly indicated that 12/ 15-LOX enzymes are critical modulators of uterine function during implantation.
We also used L-12/15-LOX KO mice to study the role of the 12/15-LOX-derived eicosanoids in implantation. Previous re- The ovariectomized pregnant (day 4) mice were divided into three groups. One group received P for 2 days and then P plus E on the 3rd day (lane None). The second group was treated with the same regimen of P and E and also received AA-861 (2 mg/mouse/day) on each day (lane AA). The third group received the same regimen of P, E, and AA-861 (2 mg/mouse/day) and also received rosiglitazone (0.35 mg/mouse/day) on each day (lane AAϩR). Rosiglitazone was administered by gavage. The number of implantation sites in different treatment groups was determined by monitoring increased endometrial vascular permeability at the implantation sites at 24 h after E administration. The means of three independent determinations Ϯ S.D. are shown.
FIG. 11. Rosiglitazone modulates the expression of 12/15-LOX target genes. The ovariectomized pregnant (day 4) mice were divided into three groups. One group received P for 2 days and then P plus E on the 3rd day. The second group was treated with the same regimen of P and E and also received AA-861 (2 mg/mouse/day) on each day (lane AA). The third group received the same regimen of P, E, and AA-861 (2 mg/mouse/day) and also received rosiglitazone (0.35 mg/mouse/day) on each day (lane AAϩR). Rosiglitazone was administered by gavage. Total RNA was isolated from uteri collected from each group and subjected to Northern blotting employing 32 P-labeled probes containing kallikrein 1 (top panel) and glutathione peroxidase (middle panel) cDNA sequences, respectively. The bottom panel shows the same blot after hybridization with a control 32 P-labeled probe for ribosomal protein 36B4. ports indicated that the L-12/15-LOX gene-disrupted mice develop normally and exhibit no major defects at the gross anatomical level (29). Under delayed implantation conditions, we found that L-12/15-LOX-KO mice showed a reduced number of implantation sites concomitant with a reduced level of 12/15-LOX-derived metabolites in the uterus. The fact that we observed only a partial reduction in implantation sites in the L-12/15-LOX-KO mice is presumably due to a compensatory effect of the E-12/15-LOX activity within the pregnant uterus. As shown in Fig. 2, PR in uterine epithelial cells induces E-12/15-LOX at the time of implantation, and this enzyme is known to produce the same HETE and HODE metabolites as the L-12/15-LOX. This hypothesis is further supported by the fact that treatment of L-12/15-LOX-KO mice with a low dose of AA-861, which effectively blocks the residual activity of E-12/ 15-LOX in the preimplantation uterus, dramatically reduced the number of implantation sites in the L-12/15-LOX-KO mice.
To uncover the network of genes that underlie the actions of the 12/15-LOX-derived metabolites in the uterus, we employed gene-profiling experiments using oligonucleotide microarrays. We identified 27 known genes, whose expression is repressed greater than 3-fold or more in the uterus at the time of implantation in response to AA-861 (Table I). Prominent among these are a number of genes that are known components of inflammatory reactions. These genes include several kallikreins, glutathione peroxidase, ceruloplasmin, transforming growth factor-␤3, ␤-chemokine receptor D6, osteopontin-like protein (minopontin), RANK (tumor necrosis factor receptor family member), and preprotachykinin/neurokinin B (33)(34)(35)(36)(37)(38)(39)(40)(41)(42)(43)(44). Certain of these genes contribute to inflammation, whereas others appear to protect the tissue against the stress of acute inflammatory reactions (45). This finding is consistent with the known inflammation-regulatory roles of the 12/15-LOX metabolites in other cell systems and tissues (13)(14)(15)(16). In the uterus, previous studies (1-5) have described extensively the localized changes in cell proliferation and differentiation, reorganization of extracellular matrix, recruitment of leukocytes, and migration of immune cells that are associated with implantation and decidualization. All these events are similar in many ways to those in inflammatory reactions and wound healing.
We also explored the mechanisms by which the 12/15-LOXderived eicosanoids control their target genes. Recent studies (30 -32, 46 -48) indicated that various eicosanoid ligands interact with and activate the PPARs. The PPARs are members of nuclear hormone receptor superfamily (48). In the mouse, three PPARs have been identified: PPAR␣, PPAR␥, and PPAR␦ (49). Each PPAR subtype plays a critical role in fatty acid homeostasis (50). These receptors are transcription factors, which are activated in response to a wide spectrum of ligands, including arachidonic and linoleic acids, and their natural and synthetic eicosanoid derivatives (46 -48, 50). Evans and co-workers (30) have reported that the 12/15-LOX-derived metabolites, 9-HODE and 13-HODE, activate PPAR␥-dependent gene expression in monocytes, presumably by directly binding to the receptor. Glass and co-workers (31) have shown that PPAR␥ and 12/15-LOX are coordinately induced by interleukin-4 in macrophages. The 12/15-LOX-derived metabolites then act as endogenous ligands of PPAR␥ to mediate transcriptional induction of the CD36 gene in these cells. In another study, 9-HODE, 13-HODE, and 15-HETE activated PPAR␥ in primary human fibroblasts (32). Collectively, these results indicated that endogenous 12/15-LOX-derived eicosanoids are able to function via PPAR␥ in a variety of tissues, and it is plausible that a similar mechanism drives 12/15-LOX-regulated gene expression in the pregnant uterus.
Very little is known about the functional role of PPARs in the female reproductive tract. In the peri-implantation mouse uterus, expression of PPAR␣ is not detectable. Consistent with this observation, the PPAR␣ KO female mice are fertile and do not show any reproductive abnormality (51). In contrast, both PPAR␦ and PPAR␥ are expressed in the uterus during early pregnancy (52, Fig. 9). Studies by Dey and co-workers (52) indicated that COX-2-derived prostacyclin activates PPAR␦ in pregnant mouse uterus during implantation. The role of PPAR␦ in implantation, however, remains unclear. Although a large number of PPAR␦ null fetuses die in utero primarily due to developmental defects, a few null mice survive to adulthood. The adult PPAR␦ KO female mice are reported to be fertile (53,54). The PPAR␥ null mutation, on the other hand, is embryonic lethal due to placental malformation and, therefore, does not permit an analysis of the role of this receptor in adult female reproductive tract function (55,56).
Here we present several lines of evidence obtained using in vitro and in vivo approaches, which indicate that PPAR␥ is a downstream effector of the 12/15-LOX-derived metabolites in the preimplantation uterus. First, using in vitro cell-based assays, we demonstrated that 12-HETE, 15-HETE, and 13-HODE induced gene activation by PPAR␥-RXR␣ heterodimers. The levels of these eicosanoids transiently increased in the uterus at the time of implantation, giving rise to the possibility that one or more of these might act as an activating ligand for PPAR␥ in this tissue. Consistent with this scenario, we observed a marked rise in the level of PPAR␥ mRNA and protein in the uterine epithelium and stroma within the implantation window. Most importantly, treatment with rosiglitazone, a PPAR␥-selective agonist, efficiently reversed the impairment in implantation induced by an inhibitor of 12/15-LOX activity, providing a clear functional link between the 12/15-LOX pathway and PPAR␥ in vivo.
If the 12/15-LOX-derived metabolites function as endogenous agonists of PPAR␥, one would predict that the 12/15-LOX target genes identified by microarray analysis (Table I) would also be regulated by other PPAR␥-modulatory ligands. Indeed, we found that two of these genes, kallikrein 1 and glutathione peroxidase, are positively regulated by rosiglitazone. Kallikrein 1 is a member of the kallikrein family, which consists of well known modulators of inflammation acting through the kinin pathway (34,35). Glutathione peroxidase modulates host response during inflammatory conditions induced by endotoxemic drugs. It regulates vascular permeability, production of cytokines, and leukocyte migration (36). The expression of both of these genes in the pregnant uterus was described previously, although their precise functions in this tissue remain unknown.
In summary, based on our findings in this paper, we propose a chain of events (Fig. 12) that is likely to mediate signaling by the eicosanoids generated by the 12/15-LOX pathway in the pregnant uterus. During early pregnancy, a rise in the level of P due to the newly formed corpora lutea induces PR-mediated expression of 12/15-LOX genes in uterine epithelium and stroma at the time of implantation. The resulting rise in the 12/15-LOX enzymes locally increases the production of a specific class of eicosanoids including 12-HETE, 15-HETE, and 13-HODE. These metabolites then act in an intracrine or paracrine manner to activate PPAR␥ in the uterine epithelial or stromal cells. The activated nuclear receptor triggers the expression of downstream gene networks in various uterine compartments, leading to extensive cell differentiation and remodeling that regulate implantation. Future studies will investigate the precise biological events that are controlled by the PPAR␥-mediated pathways in response to the 12/15-LOXderived metabolites in the pregnant uterus.