Stage-specific Integration of Maternal and Embryonic Peroxisome Proliferator-activated Receptor δ Signaling Is Critical to Pregnancy Success*

Successful pregnancy depends on well coordinated developmental events involving both maternal and embryonic components. Although a host of signaling pathways participate in implantation, decidualization, and placentation, whether there is a common molecular link that coordinates these processes remains unknown. By exploiting genetic, molecular, pharmacological, and physiological approaches, we show here that the nuclear transcription factor peroxisome proliferator-activated receptor (PPAR) δ plays a central role at various stages of pregnancy, whereas maternal PPARδ is critical to implantation and decidualization, and embryonic PPARδ is vital for placentation. Using trophoblast stem cells, we further elucidate that a reciprocal relationship between PPARδ-AKT and leukemia inhibitory factor-STAT3 signaling pathways serves as a cell lineage sensor to direct trophoblast cell fates during placentation. This novel finding of stage-specific integration of maternal and embryonic PPARδ signaling provides evidence that PPARδ is a molecular link that coordinates implantation, decidualization, and placentation crucial to pregnancy success. This study is clinically relevant because deferral of on time implantation leads to spontaneous pregnancy loss, and defective trophoblast invasion is one cause of preeclampsia in humans.

Implantation, decidualization, and placentation are critical steps for successful pregnancy in most mammals, including humans. Although various signaling pathways play important roles in these processes (1)(2)(3)(4), whether there is a common molecular link that coordinates them involving both uterine and embryonic components remains unknown.
Implantation is normal only if it occurs within a limited time period termed window implantation when the blastocyst is implantation-competent and the uterus is receptive (3). We have shown previously that compromised prostaglandin (PG) 6 signaling resulting from either cytosolic phospholipase 2␣ (cPLA2␣) or cyclooxygenase-2 (COX-2) deficiency causes deferral of on time implantation that creates adverse ripple effects impacting decidualization, placentation, and fetal well being (5,6). We also had circumstantial evidence that prostacyclin (PGI 2 ) derived from the cPLA2␣-COX-2 axis plays a role during early pregnancy via activation of peroxisome proliferator-activated receptor ␦ (PPAR␦, also known as PPAR␤) (7). These studies led us to speculate that PPAR␦ serves as a molecular link coordinating various critical steps during pregnancy.
Combining multiple approaches, we show here that although uterine PPAR␦ is essential for normal implantation and decidualization, embryonic PPAR␦ via coordinated interaction with AKT and leukemia inhibitory factor (LIF)-STAT3 signaling is vital for specifying trophoblast cell differentiation during placentation.
Animals-PPAR␦ null mice originally established on a C57BL/6J/Sv129 background (22) were back-crossed with CD1 mice for 10 generations. FLK1 lacZϩ/Ϫ mice were originally generated on a C57BL/6J/Sv129 background (23) and underwent backcrossings with CD1 mice for 10 generations. PPAR␦ Ϫ/Ϫ / FLK1 lacZϩ/Ϫ mice were generated by crossing PPAR␦ Ϫ/Ϫ males with FLK1 lacZϩ/Ϫ females on CD1 background. All mice used were housed in the Institutional Animal Care Facility according to National Institutes of Health and institutional guidelines for laboratory animals.
Ovulation and Fertilization-To examine ovulation and fertilization, wild-type (WT) or PPAR␦ Ϫ/Ϫ mice were bred with fertile males with same genotypes, respectively. Mice were killed on day 2 of pregnancy and oviducts were flushed with Whitten's medium to recovery eggs and embryos. Their morphology was examined under a dissecting microscope (5).
Implantation and Decidualization-Implantation sites on day 4 midnight and days 5 and 6 midmorning of pregnancy were visualized by intravenous injections of Chicago Blue dye solution (24). Uteri of mice without blue bands were flushed with Whitten's medium to recover unimplanted blastocysts. To experimentally induce artificial decidualization, pseudopregnant WT or PPAR␦ Ϫ/Ϫ mice received intraluminal infusion of sesame oil (25 l) in one uterine horn on day 4 and were killed 4 days later. Uterine weights of infused and noninfused (control) horns were recorded, and fold increases in uterine weights served as an index of decidualization (25).
Reciprocal Blastocyst Transfer-Day 4 WT or PPAR␦ Ϫ/Ϫ blastocysts were transferred into uteri of WT or PPAR␦ Ϫ/Ϫ pseudopregnant recipients on day 4 (24), and recipients were killed on the morning of days 5 and 6 to examine implantation by the blue dye method. Other recipients were sacrificed on day 10 for placental analysis.
Immunostaining and Lectin Histochemistry-Immunofluorescence staining of PPAR␦ in preimplantation embryos was performed as described by us (26). Fluorescence signals were viewed under a Zeiss LSM 510 confocal laser microscope. Immunolocalization of PPAR␦, COX-2, phospho-AKT, phospho-STAT3, cytokeratin, VCAM-1, PL-I, SOCS3, and CDX2 was performed in 10% neutral-buffered formalin or Bouin's fixed and paraffin-embedded sections of uteri and placentas using a Histostain-SP kit as described (21). To examine the maternal vasculature of the labyrinth zone, we performed BS-I Isolectin binding assay, which identifies the cell-surface carbohydrate structure of trophoblast cells lining maternal blood spaces within the labyrinth. Staining was done by the procedure as described (27). Immunohistological and lectin binding analysis were performed on at least 4 -6 different implantation sites or placentas at each developmental stage obtained from three different mice.
In Situ and Northern Hybridization-In situ and Northern hybridizations were performed as described previously (28). Antisense 35 S-labeled or 32 P-labeled cRNA probes were generated using appropriate polymerases from mouse-specific cDNAs. Northern hybridized bands were quantified using a personal densitometer (GE Healthcare).
LacZ Staining-LacZ staining in frozen sections was performed as described previously by us (29).
Trophoblast Stem (TS) Cell Culture-WT or PPAR␦ Ϫ/Ϫ TS cell lines were generated as reported previously (30). Cells were maintained in a proliferative state in media containing 70% embryonic mouse fibroblast cells conditioned medium, 30% TS cell medium, FGF4 (25 ng/ml), and heparin (1 g/ml). To induce TS cell differentiation, cells were cultured in medium free of sera, FGF4, and heparin for 2-6 days with medium changes every day.
Western Blotting-Protein extraction and Western blotting were performed as described previously (31). Antibodies to total and phospho-ERK1/2, p38, AKT, or STAT3 were used. Bands were visualized using an ECL kit and quantitated using a personal densitometer (GE Healthcare).
Statistical Analysis-Comparison of means was performed using a one-tailed Student's t test. Data are shown as means Ϯ S.E.

PPAR␦ Deficiency Limits Term Pregnancy
Success-To delineate PPAR␦ functions in pregnancy, we first analyzed pregnancy outcome in PPAR␦ Ϫ/Ϫ mice on CD1 background. Although homozygous null females give birth to viable pups, the litter size is significantly smaller compared with WT females (Fig. 1A). To search for the cause of reduced litter size, we examined early pregnancy events in PPAR␦ Ϫ/Ϫ mice.
Ovulation and Fertilization Are Normal in PPAR␦ Ϫ/Ϫ Mice-To examine ovulation and fertilization status in PPAR␦ Ϫ/Ϫ mice, we recorded the number of ovulated eggs and fertilized two-cell embryos on day 2 of pregnancy. Normal ovulation and fertilization were noted in PPAR␦ Ϫ/Ϫ females mated with null males. As shown in Fig. 1B, all null females ovulated with comparable numbers of ova as WT females. The yield of two-cell embryos among ovulated eggs determined the fertilization rate. The number of two-cell embryos was similar in both WT and null females (Fig. 1C). Because PPAR␦ is dispensable for normal ovulation or fertilization, we continued our search for the cause of reduced fertility post-fertilization in PPAR␦ Ϫ/Ϫ mice.
PPAR␦ Deficiency Defers On Time Implantation-Using immunofluorescence, we first examined PPAR␦ expression in preimplantation embryos, and we noted that PPAR␦ is localized in the nucleus at all stages of embryos spanning 1-cell zygotes to blastocysts (Fig. 1D). This observation suggests that PPAR␦ signaling could be important during early pregnancy.
Thus, we next compared implantation status in WT and mutant mice.
As shown in Fig. 1, E and F, whereas 45 and 100% of WT mice showed distinct implantation sites at 2000 and 2400 h on day 4, respectively, none of the PPAR␦ Ϫ/Ϫ females showed implantation even at 0100 h on day 5 as examined by the blue dye method. Unimplanted blastocysts with normal morphology were recovered (Fig. 1G), indicating their normal development. However, over 90% (10 of 11) of null mice showed blue bands when examined at 0800 h on day 5, suggesting deferred implantation in PPAR␦ Ϫ/Ϫ females. This result provides the first genetic evidence that PPAR␦ is critical for timely onset of implantation.
Maternal PPAR␦ Is Crucial for Normal Implantation-To ascertain the relative contribution of uterine versus embryonic PPAR␦ in implantation, we performed reciprocal blastocyst transfer experiments. Day 4 WT or PPAR␦ Ϫ/Ϫ blastocysts were transferred into PPAR␦ Ϫ/Ϫ or WT recipients on day 4 of pseudopregnancy. As shown in Table 1, WT blastocysts implanted normally after transfer into WT recipients (65 and 66% on days 5 and 6, respectively). In contrast, WT blastocysts transferred into PPAR␦ Ϫ/Ϫ recipients had considerably reduced implantation rate when examined on day 5; only 21% of transferred blastocysts showed implantation in 6 of 11 mutant recipients. But when examined on day 6, all mice showed implantation with an average of 44% implantation sites, reinforcing that implantation is deferred in the absence of maternal PPAR␦. Reduced implantation rates were not observed when PPAR␦ Ϫ/Ϫ blastocysts were transferred into WT uteri; 60% of blastocysts transferred showed implantation in all mice examined on days 5 and 6. Collectively, the results show that maternal, but not embryonic, PPAR␦ is a major contributor to on time implantation. PGE 2 via EP 2/4 Receptors Compensates for the Loss of PPAR␦ during Implantation-Observations of deferred implantation, but not its complete failure, in PPAR␦ Ϫ/Ϫ mice suggested that alternative pathways by PGI 2 or other PGs may partially compensate for the loss of PPAR␦ during implantation. To address this issue, we first measured PG levels and found that PGI 2 (measured as 6-keto-PGF 2␣ ) levels are highest followed by PGE 2 and other PGs in WT and PPAR␦ Ϫ/Ϫ uteri on day 5 ( Fig.  2A), suggesting that PGI 2 and PGE 2 are major PGs during implantation.
Because PGI 2 can also function via its cell surface prostaglandin I receptors (IP), we next analyzed expression IP in WT and null uteri to see whether PGI 2 signaling via IP receptors is relevant during implantation in PPAR␦ Ϫ/Ϫ mice. Similar to our previous report (7), uterine IP expression was very low during implantation in WT mice, but in null mice, its expression was detected in a subpopulation of stromal cells just underneath the circular muscle and in interstitial cells within the myometrium on days 4 and 5 (Fig. 2B). This expression pattern suggests that the role of PGI 2 via IP is less important for blastocyst attachment with the luminal epithelium even in the absence of PPAR␦.
We next compared the expression profile of prostaglandin E receptor (EP) subtypes EP 1-4 in WT and null uteri to assess the significance of PGE 2 signaling during implantation in the absence of PPAR␦. We found that the expression of these receptor subtypes is comparable in both WT and PPAR␦ Ϫ/Ϫ uteri. EP 4 was primarily expressed in the receptive luminal epithelium on day 4 and EP 2 in the luminal epithelium surrounding the implanting blastocyst on day 5 (Fig. 2C), suggesting their potential roles in implantation in null females. In contrast, EP 1 is expressed at a very low to undetectable level, and EP 3 is pri-marily expressed in mesometrial stromal cells on day 5, 7 suggesting their minimal, if any, involvement in blastocyst attachment. These results would imply that PGE 2 , via EP 2/4 receptors, partially offsets the loss of PPAR␦ during implantation. Thus, we asked if pharmacological inhibition of EP 2/4 receptors would further deteriorate implantation in PPAR␦ Ϫ/Ϫ mice. Indeed, whereas coadministration of AH-6809 (EP 1/2 antagonist) with AH-23848 (EP 4 antagonist) at 5 mg each/ kg⅐body weight on day 4 failed to inhibit implantation in WT mice, the same treatment remarkably blocked implantation in PPAR␦ Ϫ/Ϫ mice when examined on day 5 (Fig.  2D). This finding points toward a compensatory contribution of PGE 2 to implantation in the absence of PPAR␦.
Decidualization Is Compromised in PPAR␦ Ϫ/Ϫ Mice-Because PPAR␦ is highly expressed in stromal cells with the onset and progression of decidualization (7), we speculated that PPAR␦ deficiency would deter normal decidual development. Indeed, overall decidual growth, as assessed by weights of implantation sites, was substantially reduced in PPAR␦ Ϫ/Ϫ females compared with WT females on days 5-8 of pregnancy (Fig. 3A). Reduced decidual responses were also seen in experimentally induced decidualization by intraluminal oil infusion in day 4 pseudopregnant PPAR␦ Ϫ/Ϫ in the absence of embryo (Fig. 3B), providing evidence that defective decidualization is because of uterine loss of PPAR␦. The reduced decidual growth was correlated with an aberrant pattern and levels of ERK1/2 and p38 7 H. Wang and S. K. Dey, unpublished results.

FIGURE 2. Pharmacological silencing of PGE 2 signaling exacerbates the deleterious effects of loss of PPAR␦ on implantation.
A, PGI 2 and PGE 2 are the major PGs produced in day 5 implantation sites of WT and PPAR␦ Ϫ/Ϫ mice. No significant differences in uterine PG levels were noted between WT and null mice (p Ͼ 0.05, n ϭ 5-6; Student's t test). PGI 2 was measured as 6-keto-PGF1␣. Data are means Ϯ S.E. B and C, expression of IP and EP 2/4 receptors in WT and PPAR␦ Ϫ/Ϫ peri-implantation uteri. In situ hybridization analysis shows that IP is expressed in a subpopulation of stromal cells just underneath the circular muscle layer and in interstitial cells within the myometrium on days 4 and 5 in PPAR␦ Ϫ/Ϫ mice, whereas its expression remains low in WT uteri. EP 2 and EP 4 transcripts are comparable in peri-implantation uteri of WT and null mice. Arrows indicate the location of implanting blastocysts. lu, lumen; myo, myometrium; s, stroma. Bar, 200 m. D, co-treatments with EP 2 -and EP 4 -selective antagonists, AH-6809 plus AH-23848, further reduce implantation in PPAR␦ Ϫ/Ϫ mice. Numbers within the bar indicate the number of mice with blue bands/mice examined. Bl, blastocysts; Veh, vehicle.

TABLE 1 Implantation of blastocysts transferred into pseudopregnant WT or PPAR␦ ؊/؊ mice
Day 4 WT or PPAR␦ Ϫ/Ϫ blastocysts were transferred into PPAR␦ Ϫ/Ϫ or WT recipients on day 4 of pseudopregnancy. Recipients were sacrificed on days 5 and 6 to examine the number of implantation sites (IS) by the blue dye method. Uteri without IS were flushed with Whitten's medium to recover any unimplanted blastocysts. NA indicates not applicable.

PPAR␦ in Implantation, Decidualization, and Placentation
MAPK phosphorylation in the stroma during blastocyst-and oil-induced decidualization (Fig. 4, C-F). These molecules are known to participate in decidualization (31). These findings again suggest that maternal PPAR␦ is critical for normal implantation and decidualization. PPAR␦ Deficiency Confers Abnormal Angiogenesis in the Decidua Basalis-By day 7 of pregnancy, the antimesometrialmesometrial oriented deciduum slows down growth but undergoes tissue remodeling. The antimesometrial decidua capsularis progressively degenerates creating room for the growing embryo, whereas the mesometrial decidua basalis undergoes angiogenesis forming a vascularized zone that brings maternal and fetal blood vessels in close proximity to form a functional placenta (1). This timely decidual remodeling is well correlated with shifting of PPAR␦ expression from the antimesometrial decidua to the mesometrial decidua basalis and to the trophoblast sprouting from the ectoplacental cone (supplemental Fig.  S1).
Using Flk1 lacZϩ/Ϫ and PPAR␦ Ϫ/Ϫ /Flk1 lacZϩ/Ϫ mice that express ␤-galactosidase as a read out for Flk-1 promoter activity in endothelial cells (23), we observed an abnormal pattern of angiogenesis in the decidua basalis of PPAR␦ Ϫ/Ϫ females. As shown in Fig. 4A, the density of lacZ-stained blood vessels was comparable in mesometrial decidua in both WT and null mice on day 8, but the vessel density progressively decreased in WT mesometrial decidua as opposed to sustained higher density in PPAR␦ Ϫ/Ϫ decidua on day 10. This aberrant angiogenesis in the absence of PPAR␦ was associated with altered expressions of COX-2, vascular endothelial growth factor (VEGF), and angiopoietins (ANG) in decidual basalis. For example, COX-2 transcripts and proteins were highly expressed in WT mesometrial decidual cells underneath the degenerating luminal epithelium on day 8, but the expression was dramatically down-regulated with the onset of placentation on day 10 ( Fig. 4B and supplemental Fig. S1). This reduced decidual COX-2 expression is associated with reduced mRNA expression of VEGF 164 (Fig. 4C), an active and abundant isoform in mice (32), and matrix metallopeptidase 2 (MMP2) (Fig. 4D), a key regulator of decidual remodeling (33). Interestingly, COX-2 expression was sustained at higher levels in the decidua basalis in the absence of PPAR␦ (Fig. 4B) and so was the expression of VEGF 164 and MMP2 (Fig. 4, C and  D). However, the expression of ANG-1 and ANG-2 was comparable in WT and null females, with low levels of ANG-1 in the developing metrial gland and heightened ANG-2 expression in the decidua basalis (Fig. 4, E and F), indicating that their expression is independent of PPAR␦ signaling during this period. Collectively, the results suggest that PPAR␦ is required for appropriate expressions of COX-2 and VEGF for normal angiogenesis during decidual basalis development.
PPAR␦ Deficiency Disrupts Normal Placentation-Because normal implantation and decidualization is important for subsequent embryonic growth and placentation (3), observations of defective implantation and decidual growth in PPAR␦ Ϫ/Ϫ mice led us to explore whether these early defects affected the process of placentation in null mice. As expected, impaired fetoplacental development was noted in PPAR␦ Ϫ/Ϫ mice with significant anomalies in ectoplacental trophoblast development and chorioallantoic fusion on day 10 of pregnancy. For example, cytokeratin immunolocalization showed shallow trophoblast invasion into the decidua basalis (Fig. 5A). Although the fusion of the chorion and allantois occurred regardless of the presence of PPAR␦ with normal expression of VCAM-1 required for chorioallantoic attachment (34,35), the process was developmentally behind for about 24 h (Fig. 5A). In fact, chorioallantoic villous branching in day 11 null placentas was comparable with that seen in day 10 WT placentas. Because the ectoplacental cone, but not the chorion and allantois, expresses PPAR␦ on day 8 (supplemental Fig. S1), this delay in chorioallantoic attachment and villous branching in null placentas C-F, phosphorylation status of ERK1/2 and p38 MAPK in blastocyst-induced decidua in pregnant mice and in oil-induced deciduoma in pseudopregnant females. Western blot analysis demonstrates an enhanced activation of ERK, but not p38, in the stroma particularly on day 6 of pregnancy in the absence of PPAR␦. This altered ERK phosphorylation is perhaps because of the deferral of embryo implantation, which slows down stromal decidualization, making the developmental stage equivalent to day 5. In oil-induced decidualization model that precludes embryonic influence and defects associated with deferral of attachment reaction, the levels of total ERK in PPAR␦ Ϫ/Ϫ deciduoma are much higher. With respect to p38, its phosphorylation activity declines much earlier with the loss of PPAR␦. Uterine samples were pooled from 3 to 4 mice, and immunoblotting experiments were repeated twice with similar results. Quantitative analysis of ERK and p38 activation is presented as the ratio of the band intensity of phosphorylated proteins relative to their total proteins. cont, noninfused uterine horns.
is perhaps due a secondary effect imposed by deferred implantation and defective decidual growth.
To characterize placental defects in the absence of PPAR␦, we examined the expression of heart and neural crest derivatives expressed transcript 1 (HAND1) and achaete-scute complex homolog-like 2 (Ascl2, also known as MASH2), known to have stimulatory and repressive roles in specifying giant and spongiotrophoblast cell fates (36 -39). As shown in Fig. 5B, HAND1 and MASH2 were normally expressed on day 10 and day 11 PPAR␦ Ϫ/Ϫ placentas compared with that in day 10 WT placentas, indicating that PPAR␦ signaling does not influence HAND1 and MASH2 expression in differentiating trophoblast cells during placentation.
To better define which subtypes of trophoblast cells abnormally developed from diploid trophoblast progenitor cells in ectoplacental cones of PPAR␦ Ϫ/Ϫ placentas, we examined the expression of PL-1, a giant cell marker, and trophoblast-specific protein ␣ (TPBP, also known as 4311), a spongio-trophoblast cell marker (40). PL-I was expressed in a deferred manner in days 10 and 11 mutant placentas, but TPBP was expressed aber-rantly at lower levels (Fig. 5C), indicating an impaired spongiotrophoblast development in the absence of PPAR␦.
We then performed reciprocal embryo transfer experiments to further sort out the maternal versus embryonic contribution to observed spongy trophoblast defects in PPAR␦ Ϫ/Ϫ mice. Although WT trophoblast giant and spongy cells developed normally with correct expression of PL-I and TPBP in WT recipients on day 10, PPAR␦ Ϫ/Ϫ blastocysts developing in WT recipients showed impaired ectoplacental development again with reduced TPBP-positive spongiotrophoblasts (Fig. 5D). This impairment was not observed for WT embryos in mutant mothers, although less PL-I-and TPBP-expressing cells were noted (Fig. 5D), perhaps because of the deferred implantation in mutant recipients.
We next examined the consequences of compromised ectoplacental development from PPAR␦ deficiency on midgestational embryo-uterine development. As shown in supplemental Fig. S2, about 42% of implantation sites showed signs of resorption with intraluminal hemorrhage and retarded embryonic growth as examined on day 14 of pregnancy. To delineate causes for these defects, we analyzed trophoblast development in those surviving mutant placentas with viable embryos. We found that attenuated spongiotrophoblast development in the absence of PPAR␦ continued through day 14 with a marked increase in secondary trophoblast giant cells. For example, whereas WT placentas showed a discontinuous layer of giant cells, multiple layers of giant cells were present in day 14 null placentas (Fig. 6A). Increased numbers of trophoblast giant cells perhaps occurred at the expense of spongiotrophoblast development, because the TPBP-expressing spongy junctional zone was considerably reduced in PPAR␦ Ϫ/Ϫ placentas (Fig.  6B). It was striking to see that some giant cells in mutant placentas retained TPBP expression (Fig. 6B).
Furthermore, consistent with defective trophoblast cell invasion in day 10 null placentas, a shallow invasion of trophoblast cells into the decidual basalis was observed on day 14. For example, TPBP-and CDX2-positive glycogen trophoblast cells in WT placentas migrated into the interstitium of the decidua basalis even beyond the boundaries of cytokeratin-positive endovascular giant cell invasion but not in null placentas (Fig. 6, A-C). In fact, the number of glycogen trophoblast cells was reduced in the absence of PPAR␦. In addition, endovascular giant cells, which expressed high levels of COX-2 in WT placentas, were poorly developed, leaving an intact endothelial layer overlying maternal blood vessels in PPAR␦ Ϫ/Ϫ placentas (Fig. 6D). This defective invasion of both endovascular giant and glycogen spongy cells in null mutant females appears to be associated with failure of adaptation and close apposition of the decidual blood space with fetoplacental components. On the other hand, the maternal-fetal vasculature within the labyrinth zone in WT and mutant placentas showed normal development with a similar pattern of Isolectin B4 binding (Fig. 7A), a marker for trophoblast cells lining maternal blood spaces (27). Using Flk1 lacZϩ/Ϫ and PPAR␦ Ϫ/Ϫ /Flk1 lacZϩ/Ϫ mice, we also observed a comparable density of lacZ-stained fetal blood vessels (Fig.  7B), suggesting normal labyrinth vascular network in the absence of PPAR␦. Collectively, the results support our argument that the loss of PPAR␦ shifts trophoblast cell differentia-  DECEMBER 28, 2007 • VOLUME 282 • NUMBER 52 tion in favor of giant cell fate at the expense of spongiotrophoblasts, but it attenuates trophoblast cell invasion into the decidua basalis.

Loss of PPAR␦ Favors Giant Cell Differentiation by Turning
Off the AKT Pathway-To better understand the basis of aberrant trophoblast cell fate decision with the loss of PPAR␦, we analyzed the placental phosphorylation status of AKT (phospho-AKT), a downstream pathway of PPAR␦ signaling during cell proliferation and survival (16,41). By Western blotting, we observed an overall reduction of AKT phosphorylation in day 14 PPAR␦ Ϫ/Ϫ placentas (Fig. 8A). To trace the cell-specific contribution of this defect, we employed immunohistochemistry and observed that although phospho-AKT was predominantly detected in diploid spongiotrophoblasts with much lower levels in glycogen and giant and labyrinth trophoblast cells, it was barely present in spongiotrophoblasts in null placentas on day 14 (Fig.  8B). These results suggest that PPAR␦-induced AKT activation is important for normal spongy trophoblast functions.
Because PPAR␦ deficiency leads to accelerated secondary trophoblast giant cell formation, and because LIF-STAT3 signaling is critical for giant cell function (42)(43)(44)(45), we next explored the phosphorylation status of STAT3 in placentas. Our Western blotting results show comparable levels of STAT3 phosphorylation in both WT and null placentas on day 14 (Fig.  8A). Immunohistochemical analysis further revealed that phospho-STAT3 was primarily detected in diploid spongiotrophoblasts in placentas of either genotype (Fig. 8C). Interestingly, in the absence of PPAR␦, suppressor of cytokine signaling 3, a natural repressor of LIFR/gp130 signaling (46), was mostly expressed in trophoblast giant cells with reduced levels in the spongy layer as opposed to its higher expression in giant and diploid spongiotrophoblast cells in WT placentas (Fig. 8D). These results suggest that STAT3 signaling continuously promotes trophoblast giant cell formation from diploid spongiotrophoblasts in the absence of PPAR␦.
To test that a balance between PPAR␦-AKT and LIF-STAT3 signaling directs normal spongiotrophoblast differentiation into glycogen and giant cells during placentation, we utilized TS cells in culture. We first analyzed PPAR␦ expression in differentiating WT TS cells by Northern blotting. As shown in Fig. 9, A and B, although PPAR␦ expression was low in proliferating TS cells, its expression was up-regulated if cultured under differentiation conditions for 2-4 days. However, PPAR␦ expression declined in cultured TS cells on day 6 with terminal differentiation (Fig. 9, A and B). Interestingly, GW501516, a selective PPAR␦ agonist (10), but not LIF, up-regulated PPAR␦ expression in differentiating TS cells (Fig. 9, A  and B). This auto-induction of PPAR␦ expression by its ligand indicates a potential role for this signaling axis in influencing trophoblast cell fate during differentiation.
We next examined HAND1 and MASH2 expression in differentiating WT and null TS cells in culture. Consistent with our in vivo findings, the loss of PPAR␦ had little impact on HAND1 and MASH2 expression in TS cells undergoing default differentiation (Fig. 9, C and D), suggesting that PPAR␦ is not essential for normal HAND1 and MASH2 expression in trophoblasts during placental development.
To further explore cell-specific defects from the loss of PPAR␦, we analyzed PL-1 and TPBP expression in cultured TS cells. The aberrant elevation of PL-1 and TPBP expression in PPAR␦ Ϫ/Ϫ TS cells during default differentiation (Fig. 10, A-C) provides evidence that PPAR␦ signaling is required for normal trophoblast cell differentiation. Regarding higher levels of TPBP expression in differentiating null TS cells, it is possible that PPAR␦ deficiency facilitates trophoblast progenitor cell differentiation into giant cells via a TPBP-positive stage as observed in vivo (Fig. 6B). However, upon activation of PPAR␦ by the selective agonist GW501516, we observed that this agonist substantially elevated the expression of both TPBP and PL-I in WT but not in null cells (Fig.  10, A-C), suggesting that activation of PPAR␦ also promotes TS cell differentiation. This paradoxical observation of increased trophoblast cell differentiation either by silencing or activating PPAR␦ supports the concept that an appropri- ately regulated PPAR␦ signaling is critical to normal trophoblast cell development.
To search for the cause of increased giant cell population at the expense of spongiotrophoblasts in the absence of PPAR␦ in vivo, we examined the effect of LIF on TS cell differentiation in culture. We noted that LIF significantly promoted WT TS cell differentiation to giant cells with similar high expression levels of PL-I as in PPAR␦ null cells. The most striking observation was the significant reduction in TPBP expression in PPAR␦ Ϫ/Ϫ TS cells when treated with LIF during differentiation (Fig. 10, A-C). This extensive giant cell formation from PPAR␦ Ϫ/Ϫ TS cells at the expense of TPBP-expressing spongy cells in response to LIF recapitulates in vivo defects in PPAR␦ Ϫ/Ϫ placentas, suggesting that a balance between PPAR␦ and LIF signaling is important for normal trophoblast cell development.
To provide further insights into the mechanism that differentially couples PPAR␦ with LIF signaling, we examined the effects of GW501516 and LIF on AKT and STAT3 phosphorylation during TS cell differentiation. AS shown in Fig. 11, A-D, whereas GW501516 activated AKT signaling in WT differentiating TS cells, LIF induced STAT3 phosphorylation in both WT and PPAR␦ Ϫ/Ϫ TS cells. Collectively, these results provide evidence that a coordinated interaction between the PPAR␦-AKT and LIF-STAT3 signaling pathways is important for normal spongiotrophoblast and giant cell differentiation.

DISCUSSION
Successful implantation and the establishment of chorioallantoic placenta are gateways to successful pregnancy. Recent evidence suggests that a short deferral of implantation timing adversely affects subsequent developmental processes in the face of compromised PG signaling resulting from either cPLA2␣, COX-2, or LPA3 null mutation (5,6,25,47). However, it remained elusive regarding the underlying mechanism and the molecular link between the critical steps. Our present investigation provides evidence that PPAR␦ serves as a molecular link that coordinates multiple signaling pathways in a stagespecific manner for the success of pregnancy. The genetic loss of PPAR␦ signaling does not influence ovulation, fertilization, and preimplantation development. In contrast, although maternal PPAR␦ signaling is essential for normal implantation and decidualization, embryonic PPAR␦ is required for normal placentation.  PPAR␦ Is Essential for On Time Implantation-PGs are involved in a variety of physiological and pathological processes, including multiple steps during pregnancy. Although PGE 2 via its membrane EP receptors is critical for normal ovulation and fertilization (25,48,49), PGI 2 plays an important role in implantation and decidualization (7,50). In the absence of appreciable levels of uterine IP receptor and adverse reproductive phenotypes in IP Ϫ/Ϫ mutant mice during pregnancy (7, 51), we speculated and later provided pharmacological evidence that PGI 2 impacts implantation via its nuclear target PPAR␦ (7). Our present observation of deferral of the implantation window in the presence of normal ovulation, fertilization, and preimplantation embryo development in CD1 PPAR␦ Ϫ/Ϫ mice provides the first genetic evidence that PGI 2 -PPAR␦ signaling is physiologically relevant for early pregnancy events. More importantly, we found that PGE 2 , via its EP 2/4 receptors plays a complementary and important role in implantation because pharmacological silencing of EP 2/4 receptors exacerbates the incidence of implantation failure in PPAR␦ Ϫ/Ϫ females. This observation is consistent with our earlier finding that PGE 2 improves implantation in COX-2-deficient mice when coadministered with cPGI, a stable analog of PGI 2 (7). The physiological function of uterine PPAR␦ in implantation is unique among the PPAR family, because PPAR␣ and PPAR␥ are expressed at very low levels in the uterus during implantation (7), and female mice lacking PPAR␣ have no apparent reproductive defects (52).
PPAR␦ Ensures Normal Angiogenesis during Decidua Basalis Development-It is not clearly understood how PPAR␦ deficiency leads to abnormal tissue remodeling and angiogenesis during the development of decidual basalis prior to the onset of fetoplacentation. One may argue that aberrant uterine angiogenesis is a consequence, not a cause, of compromised decidual growth in PPAR␦ Ϫ/Ϫ mice. VEGF and angiopoietins participate in ovarian and uterine angiogenesis in a yin-yang manner, whereas ANG-2 stimulates vessel sprouting by blocking the sta- bilizing signaling of ANG-1 in the presence of VEGF, it antagonizes ANG-1 signaling in the absence of VEGF, contributing to vessel regression (50,53). Because ANG-1 expression is low to undetectable in the decidual basalis, it is conceivable that declining VEGF expression in the face of normal ANG-2 expression from days 8 to 10 leads to progressive vessel regression in mesometrial decidua during normal pregnancy. In the absence of PPAR␦, an aberrantly sustained VEGF expression with unaltered ANG-2 expression causes abnormal angiogenesis with increased vessel leakage because of the lack of vessel stabilization normally driven by ANG-1. Potential roles of PGE 2 in decidual angiogenesis may also not be excluded because EP 3 receptors are expressed in mesometrial decidua (54). The ability of PGE 2 to up-regulate COX-2 (48) suggests that PGE 2 generated by this auto-induction loop maintains VEGF expression in the absence of PPAR␦, contributing to abnormal angiogenesis in the decidua basalis.
Coordinated PPAR␦-AKT and LIF-STAT3 Signaling Cascades Determine Normal Trophoblast Cell Fate during Placentation-The results of marker gene expression and reciprocal embryo transfer experiments provide novel physiological and genetic evidence that intrinsic embryonic PPAR␦ signaling is vital for specifying normal trophoblast cell fates, the loss of which increases trophoblast giant cell populations at the expense of spongiotrophoblasts. The execution of PPAR␦ function by coordinating AKT and LIF-STAT3 signaling shows diversified roles of this transcription factor at various stages of development.
In the absence of PPAR␦, giant cell differentiation is heightened at the expense of spongiotrophoblasts during placental development. This aberrant trophoblast cell differentiation is depicted by reduced expression of TPBP, but not of HAND1 and MASH2, which stimulate and repress, respectively, giant and spongiotrophoblast cell fates. Interestingly, our observations of persistent TPBP expression in trophoblast giant cells in mutant placentas provide evidence that the transformation of secondary TGCs from progenitor cells transits through a TPBP-positive stage as reported previously in other contexts (55,56).  In addition to reduced spongiotrophoblast formation, we also observed a remarkably decreased volume of glycogen trophoblast cells in placentas of PPAR␦ null females. Although the origin of glycogen cells remains unknown, it is thought that these cells are derived from spongiotrophoblasts, because they express the spongiotrophoblast-specific gene TPBP and appear to arise within the spongy layer after diploid spongiotrophoblasts (57,58). Recent evidence suggests that AKT signaling is essential for glycogen trophoblast cell transformation, because AKT1-mutant placentas exhibit nearly complete loss of glycogen-containing cells in the spongy layer (59). Interestingly, we observed that although AKT phosphorylation predominantly occurs in the diploid spongiotrophoblast but not in glycogen cells in wild-type placentas, phospho-Akt is largely reduced in the absence of PPAR␦. This observation suggests that an appropriate PPAR␦-AKT signaling facilitates diploid spongiotrophoblast cell proliferation and further transformation into glycogen trophoblast cells, the total cell number of which increases by 80-fold during midgestational development (57). These findings together with our observation of accelerated giant cell formation with the loss of PPAR␦ led us to suggest that PPAR␦ functions as a cell lineage sensor regulating normal trophoblast differentiation into spongy cells but not giant cells. If this is so, it still needs to be addressed whether the spongiotrophoblastto-giant cell transformation in the absence of PPAR␦ is a default pathway or influenced by other signaling mechanisms.
In this regard, we show that LIF-STAT3 signaling is functionally intact in diploid spongiotrophoblasts regardless of the PPAR␦ status. Moreover, we revealed that genetic ablation of PPAR␦ down-regulates the expression of SOCS3, a natural repressor of LIFR/gp130 signaling, in spongy trophoblasts, which in turn favors LIF-STAT3-driven giant cell differentiation. Similar observation of increased giant cell formation at the expense of spongiotrophoblasts in SOCS3 mutant mice (42) further supports our contention that tightly regulated coordination between the PPAR␦-AKT and LIF-STAT3 signaling pathways appropriately allocates spongiotrophoblast cell differentiation into glycogen trophoblast and giant cells.
Peters et al. (60) in 2000 reported that PPAR␦ mutant mice do not exhibit much embryonic lethality, suggesting that PPAR␦ does not play any important role in embryo implantation and placentation. However, it is worthy of note that the PPAR␦ gene in this mutant mouse line was truncated at the C-terminal 60 amino acids, leaving its entire DNA-binding domain and most of its ligand-binding domain intact, resulting in a hypomorph allele. Thus it is possible that the residual PPAR␦ protein retains a certain degree of bioactivity, contributing to embryonic survival in this mutant mouse line. In 2002, Evans and co-workers (61) first reported placental dysfunction leading to embryonic lethality of PPAR␦ Ϫ/Ϫ homozygotes on Sv129/Jae or C57BL/6J background, but they did not offer any explanation for the underlying mechanism of this defect. During the course of our investigation in characterizing maternal versus embryonic roles of PPAR␦ in implantation, decidualization, and placentation, Desvergne and co-worker (22) also reported placental defects with impaired trophoblast giant cell development in homozygous conceptuses from cross-breedings of PPAR␦ heterozygous mice on C57BL/6J/Sv129 mixed genetic background. This study also reported that the TPBP-expressing spongy cell population was significantly reduced in homozygous placentas arising from crossings of homozygous mutants (22). In this study, however, no explanation was provided regarding the discrepancy of different results seen in PPAR␦ null placentas from heterozygous versus homozygous crossings. Moreover, this study provided no clues as to the contribution of maternal PPAR␦ during various steps of pregnancy. However, their observations in PPAR␦ Ϫ/Ϫ placentas arising from homozygous mutant females are consistent with our present findings of reduced spongiotrophoblast cell population in PPAR␦ Ϫ/Ϫ placentas. Although the exact reasons for observed discrepancies for PPAR␦ null placental phenotypes remain elusive, contributions of the genetic background to the observed differences cannot be excluded. In fact, the extent of PPAR␦ homozygous embryonic mortality resulting from placental deficiency following heterozygous matings is dependent upon the genetic makeup of FIGURE 11. PPAR␦ via AKT signaling regulates TS cell differentiation in culture. AKT and STAT3 phosphorylation in WT and PPAR␦ Ϫ/Ϫ TS cells exposed to GW501516 (A and B) or LIF (C and D) was analyzed by Western blotting. GW501516 (GW) induces AKT phosphorylation only in WT TS cells in culture, but LIF stimulates STAT3 phosphorylation in both WT and PPAR␦ Ϫ/Ϫ differentiating TS cells in culture. Quantitative analysis of AKT and STAT3 activation is presented as the ratio of the band intensity of phosphorylated proteins relative to their total proteins. mice (62). Similar observations are noted for other gene mutant mouse lines (63,64). Nonetheless, this study presents a comprehensive and cohesive story for how this PPAR␦ signaling coordinates various pathways for guiding maternal and embryonic interactions for the success of pregnancy.