STAT3 Regulates Uterine Epithelial Remodeling and Epithelial-Stromal Crosstalk During Implantation

Mice, hormone treatments, and tissue collection

All experiments involving mice were conducted in accordance with National Institutes of Health standards for the use and care of mice. The animal protocols were approved by the University of Illinois Institutional Animal Care and Use Committee. The Stat3 floxed (Stat3^f/f) mice were kindly provided by Dr Shizuo Akira of Osaka University, Osaka, Japan, and Dr Hua Yu of City of Hope, Duarte, California. Stat3^f/f mice were crossed with Wnt7a-Cre mice to generate SW^d/d mice. For mating studies, Stat3^f/f and SW^d/d female mice of C57BL/6 background were housed with wild-type C57BL/6 male mice (Charles Rivers). The presence of a vaginal plug after mating was designated as day 1 of pregnancy.

In superovulation experiments, 7- to 8-week-old female mice were injected ip with 5 IU of pregnant mare serum gonadotropin and 48 hours later with 5 IU of human chorionic gonadotropin. The mice were killed 16 to 18 hours after human chorionic gonadotropin administration, and oocytes were flushed from the oviducts and counted.

To induce delayed implantation, mice were ovariectomized in the morning (0700–0800 h) of day 4 of pregnancy and maintained by daily injections of progesterone (P) (2 mg per mouse) from days 5 to 7. To terminate delayed implantation and induce blastocyst attachment, the P-primed delayed implanted mice were given five injections of LIF (1 μg, Santa Cruz) at 2-hour intervals on the fourth day of the delay (day 8). Mice were killed at 24 hours after last LIF injection and uteri were collected.

Decidualization was experimentally induced in nonpregnant mice as described previously (24). Briefly, mice were first ovariectomized. Two weeks after ovariectomy, mice were injected with 100 ng of and estrogen (E) in 0.1 mL of sesame oil for three consecutive days. This was followed by daily injections of 1 mg P and 10 ng E for three consecutive days. Decidualization was then initiated in one horn by injection of 50 μL oil. The other horn was left unstimulated. The mice were treated with E+P for additional poststimulation and then killed to collect the uterine tissue either 20 hours after stimulus or 72 hours after stimulus. For the rescue of proliferation during decidualization, 100 μL of vehicle (Hank's balanced salt solution) or an epidermal growth factor (EGF) family cocktail, containing EGF (20 μg, Sigma-Aldrich), heparin binding-EGF (5 μg, R&D Systems), amphiregulin (AREG; 100 μg, R&D Systems), was added ip every 2 hours for 6 hours before decidual stimulation. Bromodeoxyuridine (BrdU; 100 μL) was injected 2 hours before collection and uteri were collected 20 hours poststimulus.

For analysis of gene expression during normal pregnancy, uteri were collected at different days of pregnancy for RNA isolation and immunohistochemistry (IHC). One horn of the uterus was fixed in 10% formalin before IHC, and the other horn was flash frozen in liquid nitrogen before RNA isolation.

Uterine epithelial and stromal cell isolation

Uteri were obtained from pregnant mice on day 4 or day 5. With the aid of a dissecting microscope, each uterine horn was cut transversely into two equal segments. The tissues were then incubated in 10 mL of a solution of 0.5% bovine pancreatic trypsin (Cooper Biomedical) in Ca-free and Mg-free PBS, pH 7.4. Tissues were first incubated at 4°C for 1 hour and then at 37°C for 40 minutes. After incubation, the uterine tissue was squeezed with forceps, dissociating the epithelium from the uterine pieces. The epithelial fragments in the supernatant were collected with a pipette; this procedure was repeated three or four times. The epithelial fragments were collected by centrifugation at 100 g for 5 minutes. The resulting cell pellet was subjected to RNA isolation protocol.

The remaining uterine tissue was incubated with 0.5 g/L collagenase at 37°C for 40 minutes. After incubation, the tube was shaken vigorously and the supernatant was passed through a 70-μm mesh to remove tissue debris. Stromal cells were collected by centrifugation at 1700 g for 5 minutes.

DNA microarray analysis

Uterine epithelial cells were isolated from day 4 pregnant Stat3^f/f and SW^d/d mice and total RNA was prepared. The RNA was hybridized to Affymetrix mouse arrays (GeneChip Mouse Genome 430A 2.0 array) containing probes that represented ∼14 000 known genes. They were processed and analyzed according to the Affymetrix protocol.

Real-time PCR analysis

Either uterine tissue was homogenized or uterine epithelium was isolated and total RNA was extracted by using TRIZOL reagent, according to the manufacturer's protocol. cDNA was prepared by standard protocols. The cDNA was amplified to quantify gene expression by real-time PCR, using gene-specific primers and SYBR Green (Applied Biosystems). The expression level of Rplp0 (36B4) or Cytokeratin 18 (Ck18) was used as the internal control. For each treatment, the mean Ct (cycle threshold) and SD were calculated from individual Ct values obtained from three replicates of a sample. The normalized ΔCt in each sample was calculated as mean Ct of target gene subtracted by the mean Ct of internal control gene. ΔΔCt was then calculated as the difference between the ΔCt values of the control and treatment sample. The fold change of gene expression in each sample relative to a control was computed as 2^−ΔΔCt. The mean fold induction and SEs were calculated from three or more independent experiments.

Immunohistochemistry

Formalin-fixed uterine pieces were processed for paraffin embedding. Cross-sections (5 μm thickness) were mounted onto microscope slides (Fisher Scientific). For immunostaining, uterine sections were deparaffinized in xylene (three times for 5 minutes each), rehydrated through a graded series of treatment with ethanol (100%, 95%, 85%, and 70% for 5 minutes each), and rinsed in tap water. For all samples, antigen retrieval was performed by boiling the sections in 0.01 M sodium citrate buffer (pH 6.0) for 20 minutes, followed by incubation at room temperature for 30 minutes. A 5% solution of normal donkey serum (Jackson Immunoresearch) in PBS was used as a blocking buffer. Sections were incubated with the following primary antibodies diluted in blocking solution: (0.25% BSA, 0.3% Triton X-100, sterile PBS) overnight at 4°C: Ki67 (BD Pharmingen), STAT3-79D7 (Cell Signaling), pSTAT3-Y705 (Abcam), ERBB1 (Santa Cruz), pERBB1 (Cell Signaling), E-cadherin (Santa Cruz), β-Catenin (Santa Cruz), CLDN-1 (Santa Cruz). The sections were washed and incubated with either biotinylated secondary antibodies (Jackson Immunoresearch Laboratories Inc) or Cy3/Dylite conjugated secondary antibodies for 60 minutes. For biotinylated samples this was followed by incubation with streptavidin-conjugated horseradish peroxidase (Histostain kit; Zymed Laboratories Inc) for 45 minutes. Sections were stained with 3-amino-9-ethyl carbazole solution (Zymed Laboratories Inc) and counterstained with Mayer's hematoxylin (Sigma). Stained sections were mounted in diazabicyclo (2,2,2) octane, Tris-Cl (pH 8.5), and sterile double-distilled water. Immunofluorescence samples were mounted in Prolong GOLD with 4′,6-diamidino-2-phenylindole and cured for 24 hours before imaging. Negative controls included incubation with donkey serum and omission of the primary antibody for all samples.

Image capture and quantitation of immunostaining

The images of immunohistochemical staining were captured by using a Leica DM2500 light microscope fitted with a Qimaging Retiga 2000R camera (Qimaging). For counting, at least five to six individual ×20 fields from each sample were captured. The numbers of positively stained uterine nuclei in each field were averaged and expressed as a percentage of the total number of these cells. Images were processed on ADOBE Photoshop version 8. The standard deviation was determined for each averaged total. ANOVA single-factor analysis was conducted on the grouped means to determine statistical significance at a significance of a P value of <.01. Immunofluorescence imaging was performed on a Leica 700 confocal microscope. These images were directly documented from the scope with minimal processing.

Steroid hormone assay

The serum P4 and E levels were measured by RIA by the University of Virginia Center for Research in Reproduction Ligand Assay and Analysis Core.

Statistical analysis

Statistical analysis was performed by t test or ANOVA. The values were expressed as mean ± SEM and considered significant if P < .05. To test data in some experiments a Bonferroni's multiple comparison test was applied to compare multiple data points.

Biphasic expression and activation of STAT3 in the preimplantation uterus

The spatiotemporal profiles of Stat3 mRNA and the active phosphorylated form of STAT3 protein were examined in the mouse uterus during early pregnancy by real-time PCR and IHC, respectively. As shown in Figure 1A, the expression of Stat3 mRNA, which was high on day 1 of pregnancy, decreased to low levels by day 3. This was followed by a second rise in mRNA levels on day 4 at the time of embryo implantation. Consistent with the mRNA profile, we observed prominent expression of active STAT3 protein, phosphorylated at Y705, in uterine luminal epithelium on day 1 of pregnancy (Figure 1B, panel a). The level of phospho STAT3 declined on days 2 and 3 of gestation (Figure 1B, panels b and c), but reappeared in uterine luminal and glandular epithelium on day 4 at the time of implantation (Figure 1B, panel d). The level of phospho STAT3 increased further on day 5 of gestation. Interestingly, this later expression was localized to both luminal epithelium and subluminal stroma (Figure 1B, panel e). The biphasic expression and activation of STAT3 in the peri-implantation uterus showed a striking overlap with previously reported biphasic LIF expression in the uterus during early pregnancy (25, 26). This observation supports the view that LIF is a modulator of STAT3 activation in the uterus during this critical period.

Conditional ablation of Stat3 in the uterine epithelium leads to infertility

Earlier studies also localized the LIF receptor predominantly in the luminal epithelium (18), indicating that an analysis of STAT3 activation and function in this tissue would provide a critical understanding of the molecular events controlled by LIF during early pregnancy. To investigate the function of STAT3 in the uterine epithelium, we created a conditional knockout of Stat3 in the uteri of adult mice. Transgenic mice expressing Cre under the control of Wnt-7a promoter was previously used to ablate “floxed” genes selectively in uterine epithelial cells (27, 28) We therefore crossed the Wnt7a-Cre mice with mice harboring “floxed” Stat3 (Stat3^f/f) (29–31) to create SW^d/d mice. In uterine epithelial cells of SW^d/d mice, Cre recombinase deletes exon 22 of Stat3, which encodes the tyrosine residue (Y705) essential for STAT activation.

We used real-time PCR and IHC to confirm the deletion of Stat3 in the uterine epithelium of these mutant mice. As shown in Supplemental Figure 1A, published on The Endocrine Society's Journals Online web site at http://mend.endojournals.org, Stat3 mRNA was markedly down-regulated in uterine epithelial cells, but not in the stromal cells, of SW^d/d mice on day 4 of pregnancy. Similarly, STAT3 protein was robustly expressed throughout Stat3^f/f uterine epithelium on day 4 of pregnancy, whereas SW^d/d mice display STAT3 expression solely in the stromal tissue (Supplemental Figure 1B), confirming successful abrogation of Stat3 gene specifically in uterine epithelium of SW^d/d mice.

A 6-month breeding study demonstrated that SW^d/d mutant mice are infertile (Table 1). Although Stat3^f/f mice exhibited normal litter size and pregnancy rates, the SW^d/d females failed to become pregnant when mated with wild-type males. However copulatory plugs were observed on mating, indicating normal mating behavior. To investigate the cause of infertility in SW^d/d females, we examined their ovarian functions by inducing superovulation. Prepubertal Stat3^f/f and SW^d/d mice were treated with a regimen of gonadotropins as described in Materials and Methods. We observed that, on gonadotropin stimulation, the number of eggs produced by the Stat3^f/f females was comparable to that produced by the SW^d/d females (Supplemental Figure 2A), indicating that ovulation is not affected in these mutant mice. To examine the ovulation and fertilization in these mice further under normal physiological conditions, blastocysts were recovered from uteri of Stat3^f/f and SW^d/d mice on day 4 of pregnancy before implantation. Once again, no significant difference was found in either the number or the morphology of the embryos recovered from Stat3^f/f and SW^d/d uteri (Supplemental Figure 2B and data not shown). In further support of normal ovarian activity, the serum levels of P and E were found to be comparable in Stat3^f/f and SW^d/d females on day 4 of pregnancy (Supplemental Figure 2, C and D). Collectively, these results suggested that the infertility of SW^d/d females is not due to impairment in the hypothalamic-pituitary-ovarian axis or lack of fertilization, but is likely due to defective implantation or pregnancy failure after implantation.

Deletion of Stat3 from uterine epithelium prevents embryo attachment

In mice, attachment of the embryos to the uterine wall is accompanied by increased vascular permeability at the implantation sites, which can be scored visually as distinct blue bands after an iv injection of Chicago blue dye (32). As shown in Figure 2A, Stat3^f/f mice displayed distinct implantation sites on day 5 of pregnancy. In contrast, the SW^d/d females did not show any sign of implantation. Histological analysis of uterine sections from Stat3^f/f females on day 5 of pregnancy showed, as expected, a close contact of embryonic trophectoderm with uterine luminal epithelium (Figure 2B, panel a). In contrast, in SW^d/d uteri, embryos did not attach to luminal epithelium. Instead, blastocysts remained free-floating in the lumen (Figure 2B, panel b). Taken together, these results indicated that the loss of Stat3 expression in the uterine epithelium resulted in the inability of the luminal epithelium to attach to the embryos.

STAT3 regulates junctional remodeling in uterine luminal epithelium

To gain insights into the mechanisms underlying the implantation defect observed in SW^d/d mice, we isolated luminal epithelial cells from Stat3^f/f and SW^d/d uteri on day 4 of pregnancy and performed gene expression profiling, using Affymetrix Mouse GeneChip arrays. Interestingly, our study revealed an up-regulation of transcripts corresponding to two distinct classes of junctional molecules, catenins and claudins, in SW^d/d uterine epithelium compared with Stat3^f/f control uteri. The microarray data (GEO accession no. GSE44087) were validated by real-time PCR analysis. As shown in Figure 3A, we confirmed a marked up-regulation of mRNAs corresponding to claudins 1, 3, and 4, and α- and β-catenin in the luminal epithelium of SW^d/d uteri.

An important feature of the receptive uterus is the reorganization of junctional proteins in uterine epithelial cell membrane at the time of implantation. In the nonreceptive state, E-cadherin, a Ca²⁺-dependent transmembrane adhesion molecule, connects adjacent epithelial cells by linking their cytoskeletons via catenins to help maintain cell-cell adhesion and apical-basal polarity (31, 33). Similarly, claudins, which are components of tight junctions, help maintain a paracellular barrier between the cells of an epithelium (34, 35). These junctional proteins associate with various peripheral membrane proteins, which are in turn connected to the actin cytoskeleton. As the uterus attains receptive status, the epithelial cell junctions are reorganized as evidenced by the redistribution of membrane proteins at junctional complexes, resulting in the loss of cell polarity that allows acquisition of adhesiveness for the embryo (36, 37).

We next investigated whether the nonreceptive state of the uterine epithelium of SW^d/d mice is associated with an altered expression of specific junctional proteins, such as E-cadherin, β-catenin, and claudin-1 (CLDN1), in the epithelial junctions. As shown in Figure 3B, panel a, low levels of E-cadherin expression were seen only in the apical tips of Stat3^f/f uterine epithelium during the receptive phase. The lateral junctions were devoid of any E-cadherin expression. Low levels of β-catenin were present in lateral junctions of Stat3^f/f uterine epithelium (Figure 3B, panel b). In contrast, in SW^d/d uterine epithelium, prominent expression of E-cadherin was observed in both apical and lateral junctions (Figure 3B, panel d) and markedly elevated levels of β-catenin were seen in the lateral junctions (Figure 3B, panel e). Interestingly, E-cadherin and β-catenin were colocalized specifically in the lateral junctions of SW^d/d uterine epithelium, consistent with the presence of intact intercellular cadherin-catenin adherent junctional complexes (Figure 3B, panel f).

We also examined the status of the tight junctions in Stat3^f/f and SW^d/d uterine epithelia. As shown in Figure 3C, the CLDN1 protein, which was barely detectable in Stat3^f/f luminal epithelium (Figure 3C, panel b), was present in the apical and basolateral junctions of SW^d/d uterine epithelium (Figure 3C, panel e). Collectively, these results suggested that, in the absence of epithelial Stat3, proper reprogramming of luminal epithelial membrane proteins fails to occur with consequent dysregulated distribution of adherent and tight junctional complexes and lack of epithelial function required for embryo attachment.

LIF-induced STAT3 activation triggers uterine epithelial reorganization

We next examined whether LIF-induced activation of STAT3 during the transition of uterine luminal epithelium from nonreceptive to receptive phase is accompanied by altered expression and redistribution of the catenins, claudins, and E-cadherin in uterine epithelial junctional complexes. In this experiment, we used the delayed implantation model in which pregnant mice are ovariectomized to remove the source of ovarian hormones and then treated with exogenous LIF to initiate embryo attachment to uterine luminal epithelium. Our results revealed that uterine epithelial STAT3 remains unphosphorylated (at Y705), cytosolic, and inactive in the absence of exogenous LIF (Figure 3D, panels a and b). Both apical and lateral junctions of luminal epithelium contained E-cadherin (green, Figure 3D, panel c), and β-catenin (red) was present mainly in the lateral junctions. Prominent expression of CLDN1 was observed in both apical and basolateral surfaces of the uterine luminal epithelium (Figure 3D, panel d). Interestingly, administration of LIF, which initiated implantation in delayed mice, also led to phosphorylation (at Y705) and nuclear localization of phosphorylated STAT3 in uterine epithelial cells (Figure 3D, panels e and f); however, total STAT3 protein levels remain unaltered (compare Figure 3D, panels a and e). We noted concomitant down-regulation of E-cadherin at epithelial apical and lateral adherent junctions (Figure 3D, panel g). In the presence of LIF, β-catenin staining, although reduced, was still detectable in the lateral surfaces. We observed marked down-regulation of CLDN1 in the uterine luminal epithelial apical and basolateral tight junctions in response to exogenous LIF (Figure 3D, compare panels d and h). Taken together, our results are consistent with the hypothesis that LIF-mediated activation of STAT3 critically regulates the dynamics of uterine luminal epithelial adherent and tight junctional complexes containing E-cadherin/β-catenin and CLDN1, respectively. Hence, we propose that a major cause of implantation failure in mice lacking uterine epithelial Stat3 is the inability of LIF to down-regulate the expression and distribution of critical components of epithelial junctional complexes. The resulting epithelium fails to present a proper surface for attachment to the embryonic trophectoderm and remains nonreceptive and refractive to implantation.

Ablation of Stat3 in the uterine epithelium leads to a defect in decidualization

During implantation, embryo attachment to the uterine epithelium is followed by decidualization of underlying stromal cells (7, 38). Because embryos fail to attach to epithelium of SW^d/d uteri, we assessed the decidual response in these mutant mice by performing experimentally induced decidualization, which occurs independent of embryo attachment, as described previously (24). Ovariectomized Stat3^f/f and SW^d/d mice were treated with a well-established regimen of steroid hormones and then decidualization reaction was initiated in one uterine horn by intraluminal injection of oil, while the other horn was left unstimulated. We then examined the gross anatomy of the stimulated and unstimulated uterine horns of Stat3^f/f and SW^d/d mice. As expected, the uterine horns of Stat3^f/f mice exhibited a robust decidual response at 72 hours after receiving the artificial stimulation (Figure 4A, panel a). In contrast, the SW^d/d uteri under identical conditions showed a drastically reduced decidual response (Figure 4A, panel b). When the decidual response was assessed by measurement of uterine wet weight gain, the SW^d/d uteri exhibited a markedly reduced weight gain relative to that seen in the Stat3^f/f uteri (Figure 4B).

Uterine epithelial Stat3 regulates stromal proliferation during decidualization

During decidualization, the uterine stromal cells initially undergo proliferation for 24 to 48 hours followed by differentiation (39). To investigate the decidualization defect in mice lacking Stat3 in uterine epithelium further, we monitored the proliferation and differentiation of steroid hormone-primed stromal cells in response to a decidual stimulation, using known markers of these events. As shown in Figure 4C, panel a, uteri of Stat3^f/f mice exhibited an intense staining for Ki67 in the stromal cells within 20 hours of receiving the decidual stimulus, clearly indicative of extensive stromal cell proliferation, which is typically seen during normal pregnancy in response to embryo attachment (Figure 4C, panel c). In contrast, the uterine sections from SW^d/d mice showed a markedly altered pattern of Ki67 staining under identical conditions (Figure 4C, panels b and d). A generally reduced proliferative activity was observed in the stromal compartment of the mutant uteri. However, a few bands of proliferative cells persisted in the subepithelial stroma, indicating that proliferation of these cells is independent of epithelial STAT3. Quantitation of the Ki67 signal in areas of reduced proliferation revealed a significant reduction in the number of Ki67-positive cells in the stromal compartment of the SW^d/d uteri (Figure 4D), indicating a defect in stromal cell proliferation in these mutant mice.

We further investigated the proliferation defect in SW^d/d uterine stroma by analyzing the expression of factors that control various cell cycle checkpoints, including cyclins, cyclin-dependent kinases, and cyclin-dependent kinases inhibitors (40). Our results revealed a marked down-regulation of cyclin E1 (Ccne1) and E2 (Ccne2) in stromal cells of SW^d/d uteri on day 5 of pregnancy. In contrast, the level of p21, an inhibitor of cyclin E, was up-regulated in stromal cells of SW^d/d uteri during decidualization (Figure 4E).

We next investigated whether uterine stromal differentiation is also impaired in the absence of epithelial Stat3. Our results revealed an intense expression of alkaline phosphatase, a classical marker of decidualization, in the decidual cells of Stat3^f/f uteri at 72 hours after decidual stimulation. In contrast, no expression of this decidual marker was seen in the uteri of SW^d/d mice at a similar time point, indicating a complete blockade of stromal cell differentiation in these mice (Figure 4F, compare panels a and b). Taken together, our results indicated that Stat3 is a critical regulator of stromal cell proliferation as well as differentiation during decidualization.

Uterine epithelial STAT3 controls EGF receptor signaling in the stroma

A defect in stromal proliferation resulting from ablation of uterine epithelial Stat3 in SW^d/d mice raised the interesting possibility that action of this transcription factor in the epithelium controls stromal proliferation by a paracrine mechanism. Previous studies reported that several members of the EGF family growth factors are expressed in the uterine epithelium and the EGF receptors (ERBB) are expressed in the uterine stroma during pregnancy (10, 11). It was also reported that the expression of EGF-like factors in uterine luminal epithelium is drastically reduced in the Lif-null mice during implantation (41, 42). However, the mechanisms that regulate the production of EGF-like growth factors downstream of LIF in the preimplantation uterus, and most importantly, the precise uterine functions controlled by these factors remained unclear. We, therefore, considered the possibility that EGF ligands are potential paracrine mediators that function downstream of epithelial STAT3 to control stromal proliferation. Indeed, our studies revealed a dramatic reduction in the expression of several EGF family ligands in the epithelial cells of SW^d/d uteri compared with Stat3^f/f controls on day 5 of pregnancy (Figure 5A).

The EGF family ligands exert their responses through the cell surface ERBB receptors. Stimulation of ERBBs by these growth factors leads to phosphorylation of specific tyrosine residues on the receptors, which in turn activate downstream signaling pathways (43). Although ERBB1 is the major EGF receptor expressed in the uterine stroma, ERBB2 and ERBB3 are exclusively expressed in the luminal epithelium during the implantation window and ERBB4 is expressed in the submyometrial stroma and the myometrium (10, 44). The signaling by EGF, heparin binding-EGF, and AREG is mediated by either an ERBB1 homodimer or a heterodimer of ERBB1 and ERBB2–4 (43). We, therefore, determined the expression of ERBB1 in uteri on day 5 of pregnancy. We found it to be widely expressed in the stroma and the levels were comparable in uteri of Stat3^f/f and SW^d/d mice (Figure 5B, compare panels a and b). We also monitored the expression of the active, phosphorylated form of ERBB1 in uteri of Stat3^f/f and SW^d/d mice on day 5 of pregnancy. Although a widespread expression of the phosphorylated ERBB1 was observed in the uterine stroma of Stat3^f/f mice on day 5 of pregnancy (Figure 5B, panel c), its levels were drastically reduced in the stromal cells of SW^d/d uteri (Figure 5B, panel d), indicating a marked reduction in stromal EGF signaling in the absence of epithelial Stat3. We also analyzed the expression of ERBB4 and phospho-ERBB4, which are located in the submyometrial stroma. No significant differences in their levels were observed between Stat3^f/f and SW^d/d mice on day 5 of pregnancy (data not shown).

We further examined whether LIF, acting upstream of STAT3 activation, is indeed the key factor controlling the production of the EGF family ligands in the epithelial cells. To test this possibility, we again used the delayed implantation model in which embryo attachment to the uterine epithelium is initiated in response to LIF. As shown in Figure 5C, administration of LIF to mice undergoing delayed implantation stimulated the synthesis of Egf, Hbegf, and Areg in uterine luminal epithelium. These results strongly supported our hypothesis that LIF-mediated activation of luminal epithelial STAT3 drives the production of the EGF family ligands in this tissue.

Paracrine control of EGF receptor signaling is critical for stromal proliferation during decidualization

To address whether the impaired production of the EGF family ligands in the luminal epithelium of SW^d/d mice was responsible for the observed defect in stromal cell proliferation during decidualization, we conducted an experiment in which these mutant mice were subjected to experimentally induced decidualization with or without supplementation of the EGF family ligands. Because Egf, Hbegf, and Areg were collectively down-regulated in uteri of SW^d/d mice, we administered a combination of these factors into the mutant uteri and then assessed their ability to rescue stromal proliferation. As shown in Figure 6A, the stroma of vehicle-treated SW^d/d uteri showed significantly reduced proliferative activity (BrdU immunostaining) and expression of activated pERBB1 relative to control Stat3^f/f uteri (Figure 6A, compare panels a and b, d and e). The levels of both BrdU-positive cells and pERBB1 were markedly elevated in the uterine stroma of SW^d/d mice treated with the mixture of EGF family ligands (Figure 6A, panels c and f). It is of interest to note that this treatment, which also increased the levels of pERBB1 in the luminal epithelium, did not trigger proliferation of these cells. Consistent with these observations, we found a marked up-regulation of mRNAs corresponding to cell cycle regulators Ccne1 and Ccne2 in uterine stromal cells in response to stimulation by the EGF ligands (Figure 6B). Taken together, these results uncovered a paracrine regulatory mechanism in which the EGF family ligands, produced downstream of LIF-induced activation of STAT3 in luminal epithelial cells, trigger EGF receptor signaling in the stromal cells to critically control the proliferation of these cells during decidualization.