Mice

Mice were maintained at the A*STAR Biological Resource Centre facility and maintained in accordance with the guidelines of the Institutional Animal Care and Use Committee. For the hormone replacement and LIF treatment, 8–10 wk B6C3HF1 mice were ovariectomized (ovxd) and rested for 12 days to eliminate endogenous ovarian hormones from the circulation [38]. For three consecutive days, the mice were primed daily with 100 ng of E₂ in sesame oil, rested for 3 days, and then injected with 5 mg Depo-Provera (intraperitoneally) (Pharmacia & Upjohn). After a further 3 days, they were injected with a single dose of 10 μg rLIF (intraperitoneally), a minimum dose previously determined to consistently induce implantation in B6C3HF1 mice [7], euthanized, and uteri recovered at 0, 1, 3, and 6 h after injection. Some females were also sacrificed at 30 min. For the E₂ induction, the animals were prepared in a similar manner except that the LIF injection was replaced with 50 ng E₂.

Isolation of LE

The LE were isolated from the uteri using a modified enzymatic digestion protocol [39]. Uteri were cut into 2 mm pieces and incubated with 1% trypsin (1:250 grade) in HBSS (Invitrogen), pH 7.4 (without Ca²⁺, Mg²⁺, or phenol red) at 4°C for 90 min and then room temperature for 1 h and 37°C for 10 min. Trypsin was neutralized with 0.4% soybean trypsin inhibitor with 5 mg/ml DNase1 in HBSS for 5min. The uterine pieces were washed in HBSS and the LE isolated by sliding forceps along the length of the uterine tube. The LE was recovered by gravity sedimentation, washed twice in HBSS, and stored at −70°C in TRIZOL or RLT buffer (Qiagen) prior to RNA extraction. It is estimated that this procedure results in the isolation of LE with ∼95% purity [35] because we were unable to detect vimentin mRNA in the isolates, which is a marker for stromal tissue.

Microarray Analysis

Total RNA was extracted (RNeasy; Qiagen) from LE samples. The quality and quantity of total RNA was determined with an Agilent RNA 6000 Nano Kit and Agilent Bioanalyzer 2100. Samples with RNA integrity number (RIN) above 7 were used for microarray analysis, which was performed using MouseWG-6 v2.0 Expression BeadChip (Illumina). Each experimental group consisted of six mice, and each sample (consisting of RNA from a single mouse) was run in duplicate. The samples were randomly assigned to different chips. For the microarray, 500 ng RNA was amplified into cRNA using Illumina TotalPrep-96 RNA Amplification Kit (Applied Biosystems/Ambion) according to the manufacturer's protocol. Amplified cRNA (1.5 μg) was hybridized to each array, and the beads were scanned using BeadArray Reader (Illumina) and analyzed with Genome Studio Software (Illumina).

Microarray Quality Control and Data Analysis

Data from the Genome Studio was imported into the Partek Genome Studio version 6.6. For analysis, quantile normalization was performed on background-subtracted data across all the chips followed by removal of batch effects. Principal component analysis (PCA) monitored the reproducibility of isolation with chip-to-chip and interanimal variations removed. Hierarchical cluster analysis was performed using Pearson dissimilarity correlation coefficient. Three-way analysis of variance (ANOVA) was applied between the different time points with a false discovery rate of 0.05 and a fold change of 1.2. Heat maps, volcano plots, and gene ontology (GO)-enrichment maps were generated from the gene sets obtained from the ANOVA. The gene sets were imported into Ingenuity Pathway Analysis (IPA) (Ingenuity Systems) to identify the canonical pathways. The significance of the association between the data set and the canonical pathway was measured in two ways: 1) as a ratio of the number of molecules from the data set that map to the pathway in relation to the total number of molecules that map to the canonical pathway and 2) using Fisher exact test to calculate a P value to determine the probability that the association between the genes in the data set and the canonical pathway is explained by chance alone ( www.ingenuity.com company/pdf/Citation Guidelines). For validation of candidate genes from pathways identified by IPA, only those genes in the pathways were considered that showed differences in intensity between the individual groups above 0.275, thereby excluding genes with very low expression levels.

The functional annotation tool available through the Database for Annotation, Visualization and Integrated Discovery (DAVID) ( http://david.abcc.ncifcrf.gov/) [40] was used to determine which gene sets were significantly differentially expressed between the 0 and 1 h time points. Transcription factor analysis was performed using the Genomatix MatBase TF software ( http://www.genomatix.de/solutions/genomatix-software-suite.html#1) to identify transcription factors altered by LIF at 1 h.

Quantitative Real-Time PCR

Genes identified in the microarray analysis were verified by quantitative real-time PCR (qPCR) using TaqMan (Applied Systems), with HPRT1 acting as an endogenous control with the probe sequences listed (Supplemental Table S1; Supplemental Data are available online at  www.biolreprod.org). The CT values for the samples were obtained after baseline correction. Relative fold change or relative expression was calculated by the ΔΔCt method where the expression of the gene of interest is normalized against the endogenous control. The calculated difference (ΔCt) for the experimental time points (1, 3, or 6 h) was subtracted from the control (0 h), resulting in the ΔΔCt value and expressed as relative expression ± SEM. One-way ANOVA with either the Boniferroni parametric or Games-Howell nonparametric test was used for the post hoc analysis and values obtained by comparing the ratio of the expression levels for each sample with the mean of the control samples.

Histology, Immunohistochemistry, and Immunofluorescence

For histological analysis, hematoxylin and eosin staining was performed. Images were viewed and recorded on a Zeiss Axioimager. Immunofluorescence analysis was performed on paraffin-embedded samples. Sections were deparaffinized and rehydrated through a methanol gradient. Antigen retrieval was performed in citrate buffer, pH 6 (Dako), in a 2100 Retriever (PickCell). Sections were incubated in 1% sodium borohydride for 20 min followed by a 10 min wash in PBS, pH7.4, and incubated overnight at 4°C with the primary antibodies (Supplemental Table S2). With the primary mouse monoclonal antibodies, the sections were incubated with Rodent Block M (Biocare Medical) for 20 min to prevent nonspecific binding. Nonspecific antibody binding was blocked by incubation in 10% normal goat serum or 10% donkey serum (Sigma) for 30 min and then washed in PBS. Samples were incubated with 1:500 dilution of secondary Alexa 488 antibodies (Invitrogen) for 1 h and the nuclei labeled with 4′,6-diamidino-2-phenylindole (DAPI), washed in PBS, then in distilled water, and incubated in 50 mM copper sulfate prepared in ammonium acetate buffer, pH 5, for 60–90 min to reduce autofluorescence [41]. The sections were mounted in ProLong gold antifade (Invitrogen) and imaged using a Zeiss LSM510 confocal microscope.

For some sections, endogenous peroxidase was eliminated by incubating the sections in 0.3% H₂O₂ for 30 min. The sections were blocked in 1% normal goat serum (Sigma) in PBS for 30 min and incubated with the rabbit anti-NFkbp65 overnight at 4°C, washed twice in PBS, and incubated in 1:500 diluted biotinylated goat anti-rabbit antibody (Dako) for 30 min at room temperature. After washing, the sections were incubated with streptavidin:biotinylated horseradish peroxidase (ABC; Invitrogen) complex for 20 min followed by PBS rinse and diaminobenzidine and H₂O₂ in PBS for 1–2 min, counterstained with hematoxylin, dehydrated through methanol grades and xylene, and mounted on DPX (Electron Microscopy Sciences).

Western Blot Analysis

Purified LE extract [42] was dissolved in complete Lysis M buffer containing proteinase inhibitor cocktail and 2% SDS (Roche). Protein supernatant (25–40 μg) was electrophoresed in a 4%–15% gradient gel (Bio-Rad) and transferred to polyvinylidene difluoride membrane (PVDF, GE Healthcare), blocked with 5% nonfat dry milk in 25 mM Tris and 0.14 M NaCl containing 0.1% Tween 20 and incubated with their respective antibodies (Supplemental Table S2) prepared in 0.5% nonfat dry milk in the same buffer overnight at 4°C. After a 30 min wash, the Western blots were incubated with 1:10 000 (rabbit) or 1:25 000 (mouse) dilutions of respective secondary antibodies (Dako) for 1 h, then washed and exposed using ECL Prime detection reagent (GE Healthcare). Cyclophilin D and cytokeratin 19 were used as loading controls. Densitometry was performed using the NIH Image J program.

Dynamic Changes in the Uterine Expression of the Estrogen Receptor-α and Progesterone Receptor Following LIF Treatment

The changes in uterine receptivity are driven by a combination of E₂ and P4, acting through their respective receptors, in regulating gene expression in the LE, GE, and stroma. Because we omitted nidatory E₂ in initiating the change in receptivity, it was necessary to first determine whether the substitution of E₂ by LIF affects the estrogen receptor-α (ESR) and progesterone receptor (PGR) distribution because they function as key transcription and cotranscription factors in the different uterine compartments.

A previous study indicated that on D4, ESR expression was high in the GE, with lower levels in the LE and subepithelial stroma [43]. By D5, increased ESR expression is detected in the GE, LE, and stroma. We compared ESR expression in the uteri at successive time points after treating the ovxd mice either with 50 ng of E₂ or 10 μg LIF (Fig. 1, A and B). In the GE, at 0 h, ESR was detected in a few glands and cells in the adjacent stroma (Fig. 1Aa–Ae). E₂ induced a strong ESR signal in the GE within 1 h after injection (Fig. 1Af). The levels increased, together with the ESR localizing to the GE nuclei at later time points (Fig. 1, Ag and Ah). In the LE, ESR expression was detectable at 1 h, although predominantly cytoplasmic and at the apical edge (Fig. 1Ab). However, by 3 h, many LE and subepithelial stromal cells showed a nuclear localization of the ESR (Fig. 1, Ac and Ag). LIF treatment resulted in a similar pattern of ESR induction, however at both a slower rate and at a lower intensity, with more intense expression being detected in the LE and GE only at 6 h after LIF treatment (Fig. 1, Bd and Bh).

FIG. 1

Estrogen (ESR) and progesterone (PGR) receptor expression and localization change in response to LIF. A) ESR (green) uterine expression after E₂ treatment (a: 0 h; b: 1 h; c: 3 h; d: 6 h; i: negative control) increased levels and nuclear localization of ESR in the LE and stroma by 3 and 6 h posttreatment. In the GE, ESR expression is activated by 1 h (e: 0 h; f: 1 h; g: 3 h; h: 6 h). Bar = 50 μm. Nuclei (blue) are stained with 4′,6-diamidino-2-phenylindole. B) ESR (green) expression after LIF injection (a: 0 h; b: 1 h; c: 3 h; d: 6 h; i: negative control) reveals nuclear localization of ESR in the LE and stroma by 6 h. ESR was not strongly induced till 6 h, and nuclear localization was not seen in the GE (e: 0 h; f: 1 h; g: 3 h; h: 6 h). Bar = 50 μm. C) PGR (green) expression after LIF injection (a: 0 h; b: 1 h; c: 3 h; d: 6 h; i: negative control) identified nuclear PGR at 0 h in the LE. By 1 h, LIF caused a reduction in PGR expression in the LE and nuclei, after which it substantially increases in the LE and stroma by 3 h. PGR expression in the stroma falls by 6 h. No change in PGR expression or localization was seen in the GE within 6 h of LIF (e: 0 h; f: 1 h; g: 3 h; h: 6 h). Bar = 50 μm.

PGR was detected primarily in the LE and subepithelial stroma, with little if any in the GE on D4. By D5, PGR expression was restricted to the subepithelial stroma [43]. In the LIF-treated mice, a similar pattern of PGR expression was noted, with strong localization of PGR to the LE at 0 h and weak expression in the subepithelial stroma (Fig. 1, Ca and Ce). At 1 h, PGR expression transiently declined in the LE (Fig. 1Cb), but by 3 h, expression had intensified in the LE and subepithelial stroma, a pattern that persisted to 6 h (Fig. 1, Cc, Cd, Cg, and Ch). Little or no PGR expression was detected in the GE at any time (Fig, 1, Ce–Ch). These observations indicate that LIF-initiated changes in LE gene expression in the first 3 h are probably independent of the ESR, with the PGR being predominately localized to the LE, albeit in a dynamic manner, due to the transient decline within the LE within the first hour after LIF, though strong PGR expression is restored in the LE and stroma by 3 h after LIF.

LIF Initiates Nuclear pSTAT3 Translocation in the LE but Not GE

In the LE, LIF stimulates the JAK-STAT transcription pathway, culminating in the activation and nuclear translocation of the transcription factor STAT3 [13]. Immunofluorescence analysis revealed that both the LIFR and gp130 were expressed in the LE and GE (Fig. 2, A and B). At 0 h, the LIFR localized to the apical and basal membranes of the LE, with lower levels of expression in the basal membranes of the GE (Fig. 2Aa). By 1 h, signal intensity had markedly increased in the LE and GE (Fig. 2Ac). Thereafter, the signal intensity declined to a minimal level in the LE by 6 h (Fig. 2Ae). Total LIFR expression confirmed the decline at 3 and 6 h (Supplemental Fig. S1A). The gp130 was detectable on the LE apical membrane at 0 h (Fig. 2Ba). The signal intensity subsequently increased with a maximal intensity at 3 h before declining to minimal levels by 6 h (Fig. 2, Bb–Be). To confirm that the LE responded to LIF, we examined STAT3 activation. Nonphosphorylated STAT3 was detected in the cytoplasm at 0 h, consistent with previous results [13] (Fig. 2Ca). By 1 h expression of pTyr-STAT3 increased (as measured by pTyr-STAT3-specific antibodies) with translocation into the majority of the LE nuclei (Figs. 2Cd and Supplemental Fig. S1B). By 3 h, pTyr-STAT3 levels declined, and STAT3 was no longer concentrated in the nuclei (Fig. 2Ce). Interestingly, the GE, which in the mouse synthesizes LIF, may not apparently respond to LIF, as indicated by the absence of nuclear pSTAT3, despite pSTAT3 protein being detected in the cytoplasm of the GE at all the time points (Fig. 2D). In the stroma, nuclear pSTAT3 was observed in a few isolated cells and the vascular endothelium (Figs. 2, C and D). In contrast, STAT5B, which is also induced by E₂ in the uterus [44], was identified in only a few stromal cells at any time (data not shown).

FIG. 2

LIF stimulates the JAK-STAT pathway resulting in nuclear translocation of pSTAT3 in the LE but not GE. A) LIFR expression (green) increases in the LE and GE within 1 h of LIF injection (a: 0 h; b: 30 min, c: 1 h; d: 3 h; e: 6 h; f: negative control). Bar = 100 μm. B) The gp130 (green) increases by 1 h at the apical surface of the LE (a: 0 h; b: 30 min, c: 1 h; d: 3 h; e: 6 h; g: negative control). Expression of gp130 in the GE at 1 h is shown in f. Bar = 50 μm. C) Nonphosphorylated Stat3 is abundant in the LE cytoplasm at 0 h (a). Nuclear translocation of phospho^tyr705STAT3 (green) occurs in the LE at 1 h after LIF injection (b: 0 h; c: 30 min, d: 1 h; e: 3 h; f: 6 h; g: negative control). Bar = 50 μm. D) The GE does not respond to LIF even though STAT3 is present in the GE cytoplasm as revealed by the lack of p-STAT3 (green) localization to the nuclei, despite the presence of LIFR and gp130 in the GE (a 0 h; b: 30 min, c: 1 h; d: 3 h; e: 6 h; f: negative control). Bar = 50 μm.

LIF Initiated Changes in LE Gene Expression

To determine the changes in gene expression initiated by LIF, we performed a comparative array analysis on total RNA isolated from the LE at successive time points. PCA and ANOVA-based Partek analysis were used for the statistical assessments. PCA revealed that all the samples at each specific time point were tightly clustered; indicating a high level of reproducibility across at the different times (Supplemental Fig. S2A). Hierarchical clustering analysis further identified two distinct expression patterns: 1) between 0 and 1 h and 2) between 3 and 6 h (Supplemental Fig. S2B). The overall expression profile at 6 h was closer to that at 0 h (i.e., untreated) than to the 1 h time point.

We first analyzed whether known LIF-regulated genes underwent expected changes in expression (Supplemental Fig. S3, a–f). Based on previous reports, Coch, Igfbp3, Msx1, Irg1, and Ihh are regulated by LIF in the LE [18, 19, 21, 22, 45, 46], although Ihh is also regulated by PGR receptor [47]. Our microarray data reflected the same changes in the expression of these genes: Coch, Igfbp3, Irg1, and Ihh all increased (Supplemental Fig. S3, Aa, Ab, and Ad–Af), while Msx1 declined in response to LIF (Supplemental Fig. S3Ac). We confirmed the change in expression of Coch, Irg1, and Msx1 by qPCR (Supplemental Fig. S3, Ba–Bc). Coch levels increased by 6 h, and Irg1 increased by 3 h, after which its expression declined (Supplemental Fig. S3, Ba and Bb). In contrast, Msx1 levels transiently increased at 1 h but by 6 h had declined to a level lower than that before LIF treatment (Supplemental Fig. S3Bc). Genes such as interleukin-1 alpha (Il1α) and oncostatin M (Osm), which are expressed in the LE and stroma at implantation [48, 49], although detectable at basal levels, were not altered by LIF treatment (data not shown). Other LE-restricted genes, including cyclooxygenase 1 (Ptgs1, also known as Cox1) [50] and lipocalin 2 (Lcn2), were analyzed. Ptgs1 expression fell at both the mRNA and protein levels (Supplemental Fig. S4, A, B, Ca, and Cb). COX-1 protein was expressed predominantly at the perinuclear boundary in cells of the LE. Similarly, transcript expression of Lcn2, an acute-phase protein expressed by the LE and GE and regulated by E₂, increased at 3 and 6 h (Supplemental Fig. S4D) [51]. This was contrary to a previous report where Lcn2 declines in response to nidatory E₂ [35]. Genes expressed in the stroma, such as vimentin and desmin, remained at basal levels and were unaltered by LIF treatment (data not shown) [52]. These changes in gene expression demonstrate that the LE responds to LIF to the same extent seen in normal uteri on D4 following nidatory E₂.

The array data was then subjected to a three-way ANOVA. A 1.2-fold change allowed us to identify specific gene networks in the LE and the dynamics of their expression over the 6 h following LIF treatment. The volcano plots obtained by comparison analysis of the data revealed that within 1 h of LIF administration 489 genes exhibited altered expression (144 reduced versus 345 increased). By 3 h, this had increased to 3987 genes, but by 6 h, the number had marginally decreased to 3651 (Supplemental Fig. S2, Ca–Cc, respectively). Heat maps also illustrated the overall changes in gene expression at the different time points (Supplemental Fig. S5).

GO ( http://www.geneontology.org) annotation, DAVID bioinformatics database, and IPA were used for the categorization and deep analysis of the gene lists. Deep analysis of LIF-induced changes focused on the BP category. Within 1 h of treatment, the largest BP categories of the 487 genes identified were immune response, blood vessel/vascular development, regulation of cell proliferation, and transcriptional regulation (22.12%). By 3 and 6 h, the dominant LIF-initiated BP categories were cellular metabolic processes, cellular component and complex subunit organization, together with the synthesis of biopolymers, macromolecules, carbohydrates, and lipids (38.58% and 42.13% at 3 and 6 h, respectively). Induction of stress response and cell death were also significantly induced BPs (11.74% and 7.47% at 3 and 6 h, respectively). The genes associated with each of the BPs are listed in Supplemental Table S3.

Changes in LE Gene Expression in the First Hour after LIF Injection

We reasoned that changes occurring during the first hour were likely to be those involved in making the LE receptive to blastocysts as well as those involved at initiating stromal decidualization, that is, defining which genes maintain the refractory/prereceptive state and those associated with the onset of the receptive state in direct response to LIF-STAT3. From the list of genes, after removing all RIK, LOC, and unknown transcripts, 54 annotated genes with reduced expression were identified (Supplemental Table S4). These genes were submitted to the DAVID Bioinformatics Database for GO analysis to identify gene clusters common to biologically related themes [40]. Full analyses and lists of enriched annotation terms for the gene groups can be found in Supplemental Table S5.

GO analysis categorized the genes into intermediate filament-cytoskeletal organization, which included the keratins (K5, 14, 15, 17), collagen 17A1, cofilin-1, claudin 5, kinesin C2, and Arc. Chromatin organization was another GO pathway associated with reduced expression of two genes, Cbx8 and Suv420H2, which maintain pericentric heterochromatin, as well changes in the histone clusters HIST1H4K and HIST4H4. Two previously characterized transcription factors, ERRFI1/MIG-6, required for P4 inhibition of E₂-mediated proliferation, and Cited2, which functions as a PRG-regulated trans-activator for Cbp/p300, were also reduced in the LE. Intriguingly, two genes involved in regulating circadian rhythmicity, Per1 and RGS16, showed reduced expression, with Per1 having been previously reported as a PGR-regulated gene in the uterus [53] (Supplemental Table S5).

Within 1 h, 256 genes/transcripts (excluding RIK and LOC sequences) were induced in the LE. The GO BPs that were induced included inflammatory/innate immune responses (Cxcl1, NfKBIZ, Selp, Ccl2, Cfb, C3, Sphk1, Saa3, SerpinG1, STAT3, Ccl7, Cxcl10, SerpinA3N, Nupr1, Igbp4, F2R) and vascular development/angiogenesis (Sox17, Ihh, Notch, Hey2, endothelin 1). Significantly, there was increased expression of 42 transcription factors, including Sox 7, 9, and 17, Cebp/β, HoxD10, HoxB4 and B10, Hes1, Hey1 Nfkβ, Bcl3, Kfl 6 and 9, and Rel genes (Supplemental Tables S4 and S6).

LIF Induces a Diverse Repertoire of Networks in the LE

IPA revealed that more than 30 pathways were altered by LIF treatment across all the time points. Eighteen pathways at each time point were selected (Supplemental Fig. S6). These included the JAK-STAT, IGF-1, VEGF, Sonic Hedgehog, Wnt/β-catenin, ephrin, Notch, and TGF-β pathways, metabolic pathway mTOR, cytoskeletal pathways affecting actin organization, integrin receptor β activation, innate immunity/inflammation and vasculogenesis, such as stress response/apoptosis, and Toll-like receptor (TLR) pathways. To confirm the authenticity of the microarray results, we validated the dynamics in the expression of selected proteins from 10 of these pathways of which a selection are described in detail below.

IGF1 Pathway

In the LE, IGF1 receptor (IGF1r) mediates E₂-induced cell proliferation [54, 55]. Over the 6 h period, components of the IGF1 pathway underwent an initial increase followed by a decrease in expression. These included Igf1r, insulinlike growth factor-binding proteins 3, 4, and 6 (Igfbps3, 4, and 6), glycogen synthase kinase 3 beta (Gsk3β), and cysteine-rich, angiogenic inducer, 61 (Cyr61). The expression of Igf1r and Igfbp3 was validated by qPCR, Western blot, and immunofluorescence analyses. The transcript levels of Igf1r and apical membrane expression of phospho^tyr IGF1r protein were reduced at 3 and 6 h (Fig. 3, Ac and Ad, and Supplemental Fig. S7Ac). Intriguingly, a nuclear form of this protein was apparent [56] that increased at 1 and 3 h, after which its expression declined (Fig. 3, Ab, Ac, and Ae). In line with previous reports, IGFBP3 was induced by LIF [19], with transcript and protein levels increased by 3 h (Fig. 3Bc and Supplemental Fig. S7, B and C). Although the IGF1 pathway regulates epithelial proliferation prior to implantation, we did not detect any increase in the proliferation marker Ki67 within the 6 h following LIF treatment (data not shown). However, at 1 h, cyclinD1 (a marker for proliferating LE [57]) transiently localized to the LE nuclei (Fig. 3Cb), at the same time as nuclear localization of the PGR fell (Fig. 1Cb), whereas at 0, 3, and 6 h, cyclinD1 was primarily cytoplasmic. Based on the reduction in membrane-associated IGF1R, its translocation to the nucleus along with increased IGFBP3, which sequesters the IGFs, it would appear that LIF inhibits IGF1 signaling in the LE.

FIG. 3

LIFs effects on the insulinlike growth factor-1 receptor pathway. A) Phospho^tyr1161IGF1R (green) localization following LIF is reduced at the apical LE membranes (a: 0 h; b: 1 h; c: 3 h; d: 6 h; f: negative control). There is an increase in a novel nuclear form of Igf1R in the LE nuclei at 1 and 3 h (e). Bars for a, b, c, d, f = 50 μm and for e = 20 μm. B) IGFBP3 (green) increases following LIF (a: 0 h; b: 1 h; c: 3 h; d: 6 h; e: negative control) by 3 h. Bar = 50 μm. C) Cyclin D1 (green) shows transient nuclear localization at 1 h following LIF treatment in the LE (a: 0 h; b: 1 h; c: 3 h; d: 6 h; e: negative control). By 3 h, Cyclin D1 had relocated to the cytoplasm. Bar = 50 μm.

LIF treatment reduced the levels of phospho^tyr1161IGF1R at the LE cell surface with concurrent translocation to the nucleus. Although the IGF1R lacks nuclear localization signals, nuclear translocation of this receptor has been noted in tumor cells, where IGF1R associates with the WNT-regulated transcription factor LEF1 [58, 59]. Together these factors increase the levels of cyclin D1 and AXIN2 [56]. Whether nuclear IGF1R has similar functions in the LE remains to be determined.

VEGF Pathway

The first overt indication of the start of implantation is an increase in uterine vascular permeability leading to edema. Following blastocyst invasion, the uterus undergoes extensive vascular remodeling to support stromal decidualization. Vascular remodeling is regulated by a variety of growth factors, particularly the VEGFs, FGFs, and platelet-derived growth factors(PDGFs); antibody-mediated inhibition of VEGF alone is sufficient to block implantation [60].

The VEGF family consists of five genes (VEGFA₁₆₄, B, C, and D, and PLGF) with each gene generating many different isoforms [61]. Although Vegf-a is the predominant form in the LE and is primarily responsible for stimulating angiogenesis, its expression was unaltered by LIF (data not shown). Vegf-b, which is required for blood vessel survival/maintenance, increased at the transcript level at 3 h after which levels declined (Supplemental Fig. S8A). Levels of VEGFB protein in the LE declined by 1 h and continued to do so in the ensuing 6 h (Fig. 4A and Supplemental Fig. S8, Ba and Bb), although by 1 h VEGFB stromal expression significantly increased (Fig. 4Ab). Of the VEGF receptors, expression of VEGFR1 (FLT-1), which binds VEGFA and B, was unchanged in the LE until 3 h, where it localized at the apical surface, followed by more intense expression in the subepithelial stroma at 6 h (Fig. 4Cd). VEGFR2 (FLK-1) binds VEGFs A and C–E, and its expression became more obvious in the LE and stromal blood vessels by 1 h (Fig. 4Cb). By 3 h, VEGFR2 was showing marked nuclear accumulation in the LE and in many cells in the stroma (Fig. 4Cc) [62], but by 6 h, expression had declined (Fig. 4Cd). VEGFR3 (FLT4), which regulates lymphangiogenesis, appeared to be weakly expressed in the LE apical membranes and GE throughout the 6 h time frame without any overt changes in expression (data not shown).

FIG. 4

VEGF pathway alterations induced by LIF. A) VEGFB (green) (a: 0 h; b: 1 h; c: 3 h; d: 6 h; e: negative control) increases at 3 h. Bar = 50 μm. B) VEGFR1/FLT-1 (green) (a: 0 h; b: 1 h; c: 3 h; d: 6 h; e: negative control) increases in the subepithelial stromal cells at 6 h. Higher magnification images at 3 and 6 h are shown in f and g, respectively. Bar = 50 μm. C) VEGFR2/FLK-1 (green) (a: 0 h; b: 1 h; c: 3 h; d: 6 h; e: negative control) increases in the LE and stromal blood vessels at 1 h and nuclear localization in LE and stroma at 3 h. Bar = 50 μm. D) The HIF 1α transcription factor declines by 6 h (a: 0 h; b: 1 h; c: 3 h; d: 6 h; e: negative control). Bar = 50 μm.

From the array data, expression of the hypoxia inhibitory transcription factor 1α (HIF1α), which regulates VEGF expression [63], was initially high but had fallen by 6 h. HIF1α protein was localized in the LE nuclei at 0 and 1 h, with a marked increase in the stromal nuclei appearing by 1 h (Fig. 4, Da and Db). By 3–6 h, nuclear localization in the LE had declined, and by 6 h, overall levels in both the LE and stroma had declined (Fig. 4, Dc and Dd). There was some relocalization of HIF1α from the nucleus to cytoplasm at 3 h (Fig. 4Dc). Overall, the relative paucity in changes in the different components of the VEGF family of ligands/receptors in response to LIF suggests that the implanting blastocyst may be the prime inducer of the VEGF network. Nevertheless, it is intriguing that in the retina and other tissues, LIF appears to inhibit VEGF, HIF-1 expression, and microvessel formation [64].

TLR Signaling

The interleukin-1 receptor/TLR system (IL1r/TLR) is a key component of the innate immune system (IIS) and is the first line of the host defense system against infectious microorganisms. In mammals, bacterial lipopolysaccharide stimulates the IIS by binding to the receptor complex consisting of CD14, MD2, and TLR4. This receptor complex activates the MyD88 (myeloid differentiation factor 88) pathway by recruiting the myeloid differentiation primary response gene 88 (Myd88) and culminating in activation of the NFκB and MAPK transcription factors [65]. These then induce the production of proinflammatory cytokines, such as TNFα and IL-1 and -6. From the array data, genes altered in this pathway were lipopolysaccharide-binding protein (Lbp), the TLR 1, 2, and 4 (Tlr1, Tlr2, and Tlr4), CD14 (Cd14), Myd88, and ECSIT homolog (Ecsit). The data revealed significant activation of the Toll pathway by 3–6 h. Expression of TLR2 and CD14, a coreceptor for TLR4, were further validated, with transcripts for Tlr2 increasing by 3 h and then declining by 6 h (Supplemental Fig. S9A), while TLR2 protein localized to the cell membrane and cytoplasm of the LE (Fig. 5A), as previously reported [35, 66]. The expression of TLR2 protein in the LE prior to LIF stimulation was minimal, with a slight increase being noted by 3 h (Fig. 5Ab–Ad). In the stroma, robust protein expression of TLR2 and CD14 was evident by 3 h (Fig. 5, Ac and Bc), possibly due to soluble protein diffusing from the LE to the stroma (Fig. 5, Ab and Ac) because CD14 transcripts are only present in the LE [35]. Alternatively, TLR2 and CD14 stromal expression may be due to some secondary signal, induced by LIF in the LE, transiently stimulating expression of these factors. Total membrane and soluble protein levels for TLR2 fell by 6 h (Supplemental Fig. S9, Ba–Bc). CD14 protein was detected both on LE membranes and in the cytoplasm (Fig. 5B), in agreement with a previous report based on mRNA in situ analysis [35]. By 3 h, the expression increased, but by 6 h, CD14 levels had fallen (Fig. 5B and Supplemental Fig. S9, C, Da, and Db). As with TLR2, we observed increased CD14 levels at 1 and 3 h in the stroma (Fig. 5, Bb and Bc).

FIG. 5

The innate immune response and ephrin receptor pathways are modulated by LIF. A) Membrane expression of TLR2 (green) in the LE peaks at 3 h after LIF (a: 0 h; b: 1 h; c: 3 h; d: 6 h; e: negative control) as does stromal TLR2 expression. Bar = 50 μm. B) CD14 (green) increases in both the LE and stroma at 1 and 3 h (a: 0 h; b: 1 h; c: 3 h; d: 6 h; e: negative control). Bar = 20 μm. C) Ephrin A1 (green) increases in the LE. Stromal expression also increases at 1 and 3 h. (a: 0 h; b: 1 h; c: 3 h; d: 6 h; e: negative control). Bar = 50 μm. D) Membrane localization of EPHA2 (green) initially increases at 1 h, then declines, and then further increases at 6 h (a: 0 h; b: 1 h; c: 3 h; d: 6 h; e: negative control). Bar = 50 μm.

Induction of all these changes is initiated in the absence of embryos, suggesting that the presence of the embryo in the uterus is, at some level, anticipated. However, it seems unlikely that the immune response is essential to the switch in receptivity as deletion or inhibition of some key components of the immune response (e.g., Cd14, Il1r, and Coch, which has bacteriostatic properties [67]) has little or no effect on implantation rates in mice [18, 68, 69].

Stress Response/Apoptosis Signaling

In parallel with the activation of the TLR pathway, we characterized the activation of stress response/apoptosis pathways. The RelA viral oncogene, B-cell CLL/lymphoma 2, 3, and 6 (Bcl2, 3, and 6), nuclear factor of kappa light chain gene enhancer in B-cells inhibitor, alpha (Nfkbiα) or I-kappa-B-alpha (IκBα), Fas (TNF receptor superfamily, member 6), BCL2-antagonist/killer 1 (Bak1), BH3-interacting domain death agonist (Bid), lamin A (Lmna), nuclear factor of kappa light polypeptide gene enhancer in B-cell inhibitor, beta (Nfkbiβ), tumor necrosis factor receptor, member 1B (Tnfrsf1β), and tumor necrosis factor receptor, member 1A (Tnfrsf1α) were induced. Three genes—RelA, Bcl3, and Nfkbiα—were further validated. The levels of all three transcripts increased by 3 h, after which they declined (Supplemental Fig. S10, A, E, and F). Nfκbp65 transcript and protein levels were confirmed by Western blot analysis and immunofluorescence (Supplemental Fig. S10, C, Da, and Db). Cytoplasmic localization of total NFκBp65 increased at 3 h and peaked at 6 h in the LE (Supplemental Fig. S10B).

Actin Cytoskeleton Organization

The morphology of the LE changes during implantation in rodents, particularly in organization of the subapical actin network, lateral membrane tight junctions, and persistence of pinopod/microvilli [70, 71]. LIF induces many changes in the LE apical actin cytoskeletal network (Supplemental Fig. S11A–E). Prominent genes affected in this network included beta actin (Actnβ), breast cancer anti-estrogen resistance 1 (CAS) (Bcar1), p21 protein (Cdc42/Rac)-activated kinase 3 (Pak3), profilin 1 (Pfn1), profilin 2 (Pfn2), PTK2 protein tyrosine kinase 2 (Ptk2), moesin (Msn), and actin-related protein 2/3 complex, subunit 1B (Arpc1β). Furthermore, two genes, Shroom3 and Nuak2, that induce stress fiber formation and apical constrictions during epithelial invagination increased by 1 h. We validated the changes in expression of Actn1, Actnβ, and Bcar1 by qPCR. Actn1 transcripts increased at 3 h and then declined by 6 h (Supplemental Fig. S11A), while expression of Actnb and Bcar1 transcripts increased at 1 h (Supplemental Fig. S11, B and C) and then fell. β-Actin localized to the apical surface of the epithelium, but this localization declined by 3 h (Fig. 6, Ab and Ac). Protein levels of β-actin showed a steady decline by 3 and 6 h (Supplemental Fig. S11, Da and Db).

FIG. 6

LIF induced changes in actin cytoskeleton organization, integrins, and extracellular matrix components in the LE. A) β-Actin immunostaining (green) (a: 0 h; b: 1 h; c: 3 h; d: 6 h; i: negative control) suggests a reduction in β-actin levels underlying the apical LE membrane at 1 and 3 h. The subpanels e, f, g, and h are magnified images of the boxed areas shown in subpanels a, b, c, d. Bar = 50 μm. B) ITGA3 (green) expression declines at the LE apical membrane (a: 0 h; b: 1 h; c: 3 h; d: 6 h; e: negative control). Bar = 50 μm. C) Fibronectin 1 (green) expression declines in the LE and stroma by 3 and 6 h (a: 0 h; b: 1 h; c: 3 h; d: 6 h; e: negative control) . Bar = 100 μm. D) Laminin C1 (green) declines but then transiently increase in the sub-LE stroma at 3 h (a: 0 h; b: 1 h; c: 3 h; d: 6 h; e: negative control). Bar = 50 μm. E) Osteopontin (SPP1) (green) expression increases in the LE at 3 h (a: 0 h; b: 1 h; c: 3 h; d: 6 h; e: negative control). Bar = 100 μm. F) MUC1 (green) expression in the LE progressively increases by 6 h (a: 0 h; b: 1 h; c: 3 h; d: 6 h; e: negative control). Bar = 50 μm.

The cell surface adhesion molecule, E-Cadherin (CDH1) is essential for formation and development of the uterine epithelium. Deletion of E-Cadherin in the neonatal uterus leads to infertility due to a failure in endometrial gland formation [72]. Although E-Cadherin was not identified as a LIF-regulated gene in the array analysis, it did undergo significant redistribution in the LE, becoming largely restricted to the apical surface, with lower levels in the basal and lateral membranes, in agreement with previous observations (Fig. 8D) [73].

Eph/Ephrin Receptor Pathway

Few studies have reported on the potential role of the Ephrin pathway in regulating uterine function/blastocyst implantation. This is somewhat surprising because Eph/ephrin signaling is important in regulating cell-cell interactions, including cell repulsion, attraction, and migration, during embryonic segmentation and angiogenesis [74]. One study indicated decreased expression of EPH-A2 in the murine LE during implantation [75], whereas in the human endometrium, EPH-A2 appeared to be abundant during the proliferative phase compared to the secretory phase of the cycle [76] and Eph-A1-null mice have abnormal uterine development [77]. Our analysis indicated a general elevation in expression of components of the Eph pathway at 3 h followed by reduced levels by 6 h. Genes whose expression changed were: ephrin A1 (Efna1), ephrin A5 (Efna5), ephrin receptor A2 (EphA2), ephrin receptor B2 (Ephb2), v-Ki-ras2 Kirsten rat sarcoma viral oncogene homolog (Kras), syndecan binding protein (syntenin) (Sdcbp), and LIM domain kinase 1 (Limk1). Validation of mRNA expression levels of the ligand Efna1, the receptor EphA2, and the intermediate gene, Kras, was performed. The proangiogenic ephrin Efna1 transcript levels increased over 6 h (Supplemental Fig. S11A). At 0 h, EFNA1 protein localized to the basal cell membrane of the LE (Fig. 5Ca) with expression subsequently increasing. By 1 h, it was also transiently present in the stroma (Fig. 5, Cb and Cc) with stromal expression declining and LE expression increasing by 6 h (Fig. 5Cd).

Ephrin receptor A2 (Epha2), the receptor for EFNA1, underwent dynamic changes in its subcellular localization. At 0 h, EPHA2 concentrated intensely at the perinuclear envelope (Fig. 5Da). By 1 h, signal intensity increased and the receptor redistributed to the cell membrane (Fig. 5Db). At 3 h, the expression decreased at both the membrane and perinuclear regions followed by a slight increase at 6 h at the basal and apical LE membranes (Fig. 5, Dc and Dd). Transcript levels either did not change or declined slightly (Supplemental Fig. S11C). Kras transcripts had increased at 1 h and subsequently fell (Supplemental Fig. S11D).

Ephrin A1 activation also results in STAT3 phosphorylation in myotubes [78], and there is evidence that in gastric cancer STAT3 binds to the Tlr2 promoter to drive cell survival, proliferation, and suppression of apoptosis [79]. The increased LE levels of ephrinA1 and STAT3 within 1 h of LIF administration suggest an association between the two events. Similar to the human endometrium, the ephrin A1 system might be involved in reduction of the cell-to-cell integrity in the LE [75]. Ephrin A1 can also modulate integrin function by dephosphorylating focal adhesion kinase (FAK) and paxillin [80]. BCAR1/ p130^cas and DOCK1/DOCK180, both LIF-regulated proteins, are associated with focal adhesions [81, 82].

Integrins and the Extracellular Matrix Proteins Laminin, Fibronectin, Osteopontin, and Muc1

Integrin receptors are made up of alpha and beta subunits, and signaling through this complex is implicated in mediating cell-cell interactions during embryo implantation [83, 84]. This pathway was upregulated at 3 and 6 h. The alpha integrin subunits affected by LIF included integrins A3 (Itga3), A5 (Itga5), and A6 (Itga6), while ITGB5 was the beta integrin that changed. Other responsive genes associated with the integrin-signaling pathway included: dedicator of cytokinesis 1 (Dock1), Wiskott-Aldrich syndrome-like (Wasl), ras homolog family member B (Rhob), ras homolog family member D (Rhod), and v-ral simian leukemia viral oncogene homolog A (ras related) (Rala). We confirmed the alteration in expression of Itga3 by qPCR analysis. The mRNA for this integrin peaked by 3 h and then declined (Supplemental Fig. S12E).

By immunofluorescence analysis, ITGA3 localized to the apical, basal, and lateral membranes of the LE. By 1 h, expression had declined and by 6 h was restricted to the basal and a lesser extent apical membranes (Fig. 6B). The decrease in ITGA3 was confirmed by Western blot analysis at 3 and 6 h (Supplemental Fig. S12Fa–Fc). The three alpha subunits specifically interact with β1, although α6 can interact with the β4 integrin. Although β1 localized to the basal membrane of the LE, its expression was unaltered by LIF (data not shown). The only subunit that was LIF regulated in the microarray was β5, which undergoes a basal to apical membrane transition in the rat uterine LE at the time of embryo attachment [85]. Whether this integrin subunit forms a unique interaction with the alpha subunits in the untreated LE remains to be determined.

The transcript levels of the two other genes, Dock1 and Wasl, transiently increased by 3 h and then fell by 6 h (Supplemental Fig. S12, G and H). The expression of three major components of the extracellular matrix, laminin, (Laminin C [y] 1), fibronectin (Fn), and osteopontin (Spp1)—all Arg-Gly-Asp peptide-containing proteins that bind integrins—was also analyzed. Prior to LIF treatment, the three proteins showed strong expression in the LE with laminin and osteopontin localizing to the apical and basal LE membranes. Fn predominantly localized to the LE cytoplasm, and in contrast to the other two proteins, Fn expression in the LE declined (Fig. 6C). Following LIF treatment, laminin C1 levels in the LE declined but then transiently increased in the sub-LE stroma at 3 h (Fig. 6D). Osteopontin localization to the LE membranes persisted throughout the 6 h period with a marked peak in expression at 3 h (Fig. 6E).

Another extracellular matrix protein that has been extensively investigated regarding its role in regulating implantation is the mucin, Muc1 [86]. MUC1 expression in the LE is repressed by P4 and is weakly expressed by D4 in the LE of the mouse and rat. In the LIF-treated mice, Muc1 transcripts and protein levels were low at 0 and 1 h (Fig. 6, Fa and Fb, and Supplemental Fig. S12I) and by 3 h increased MUC1 localized to the apical LE (Fig. 6, Fc and Fd, and Supplemental Fig. S12I). The significance of these changes in SPP1 expression remains to be seen because Spp1- and Muc1-null mice are fertile as are mice doubly deficient for both Spp1 and vitronectin [87, 88].

FGF Pathway

Another prominent pathway that was induced was that of FGFs. FGFs have a multitude of regulatory functions in cell proliferation, angiogenesis, wound healing, and development of various tissue/organ systems. Genes whose expression altered following LIF treatment included FGF 1 (Fgf1), FGF receptor 2 (Fgfr2), Grb2-associated binding protein 1 (Gab1), phosphatidylinositol-4,5-bisphosphate 3-kinase, catalytic subunit gamma (Pik3cg), heme oxygenase (decycling) 1 (Hmox1), and FGFR-like 1 (Fgfrl1). FGFs, in particular FGFs 1, 9, 10, 18, and 21, were previously identified in the stroma where they influence LE cell proliferation [46]. Loss of the transcription factors MSX1/2 or HAND2 in the stroma, enhances FGF expression, either by the MSX-regulated WNT pathway and/or HAND2 driven by P4, blocking stromal FGF expression [89]. Stromal FGFs act via their FGFRs in the LE to stimulate the ERK1/2 kinase pathway in both the LE and GE. Consequently, cell proliferation is maintained, with the uterus failing to attain a receptive state [46, 89]. We observed exclusive expression of FGFR2 in the LE, as previously reported (Supplemental Fig. S13C). Its expression, as well as that of its ligand FGF1, declined following LIF treatment (Supplemental Fig. S13A–D). GAB1, a protein phosphorylated by FGF signaling and required to activate PI3K-AKT pathway [90], also fell (Supplemental Fig. S13E). The reduction in FGF signaling during embryo attachment indicates that LIF contributes to the inhibition of this pathway, paralleling the inhibition of the IGF1 pathway.

Notch2 Pathway

Array analysis revealed an increase in Notch1 expression by 1 h together with increased expression of the Notch-regulated transcription factors HEY2 and HES1. NOTCH1 is implicated at inducing stromal cell decidualization in the mouse uterus [36]; however, we did not detect any changes in expression Notch1 either by qPCR or by immunofluorescence. Because NOTCH2, another Notch receptor, is expressed in human LE [91], we analyzed its expression by qPCR and found Notch2 transcript levels fell following LIF treatment (Supplemental Fig. S14A). At the protein level, the functional intracellular domain (ICD) of NOTCH2 localized to the apical and basal membranes and cytoplasm by 3 h (Fig. 7Ac), and by 6 h, the ICD was detected in the LE nuclei (Fig. 7Ad). Total protein levels of the Notch2 ICD (cNotch2) had risen by 6 h, (Supplemental Fig. S14B). In parallel, stromal levels of NOTCH2 also increased (Fig. 7Ad).

FIG. 7

Alterations in the Notch 2 signaling pathway. A) The intracellular domain (ICD) of NOTCH2 (green) localizes to the LE membranes at 1 h and then increases in expression at 3 and 6 h. NOTCH2 ICD then localizes to the nucleus at 6 h (a: 0 h; b: 1 h; c: 3 h; d: 6 h; e: negative control). White arrows indicate localization of NOTCH2 ICD at the LE membrane; white arrowheads indicate expression in nucleus while red arrows indicate loss of expression at the cell membrane. Bar = 50 μm. B) The notch-regulated transcription factor HES1 (green) nuclear localization increases in the LE and stroma by 6 h (a: 0 h; b: 1 h; c: 3 h; d: 6 h; e: negative control). Bar = 50 μm.

FIG. 8

LIF activates the canonical Wnt pathway. A) WNT7A (green) protein increases in the LE at 1 and 3 h and then declines at 6 h. Bar = 50 μm. B) FZD6 (green) protein decreases at the apical LE membranes. Bar = 50 μm. C) The transcriptionally active dephosphorylated form of β-catenin (in green) shows increased nuclear localization in the LE and stroma 3 and 6 h. Bar = 20 μm. Subpanels indicate expression changes (a: 0 h; b: 1 h; c: 3 h; d: 6 h; e: negative control). D) E-Cadherin (green) expression decreases at the basal membrane of the LE at 6 h. Bar = 100 μm.

The only Notch ligand showing a change in expression was Jagged 1 (Jag1). Its transcript and protein levels increased at 3 and 6 h (data not shown). Other genes altered in this pathway include Deltex4, Numb, and radical fringe (Rfng). Notch signaling results in the downstream activation of the hairy and enhancer of split-1 (HES) and hairy/enhancer-of-split related with YRPW motif (HEY) group of transcription factors [92]. We observed increased HES1, but not HEY1 expression, with transcripts and nuclear levels of HES1 increasing in the LE by 6 h (Fig. 7B and Supplemental Fig. S14, C and D). HES1 stromal expression also increased (Fig. 7Bd). Whether such localization in the stroma depends on signaling from the LE remains to be determined.

Canonical Wnt/β Catenin Signaling

The canonical Wnt/β-catenin pathway has been implicated in regulating uterine cell proliferation, blastocyst invasion, and decidualization [93]. Overall, LIF treatment transiently enhanced expression of many genes in this pathway between 1 and 6 h. The only canonical ligand identified was WNT7A. Transcripts of Wnt7a increased over the first 3 h and then declined (Supplemental Fig. S15A). WNT7A expression in the LE peaked at 3 h with some expression in the stroma (Fig. 8Ac). Transcript levels of the Wnt receptor, Frizzled homolog 6 (Fzd6), fell over the 6 h (Supplemental Fig. S15B), with reduced protein levels apparent only by 6 h (Fig. 8B and Supplemental Fig. S15, B and C). Other pathway genes identified were Sry (sex-determining region Y)-boxes 7 (Sox7, 9, and 17), transducinlike enhancer of split 3 (Esp1) homolog, Drosophila) (Tle3), secreted frizzled-related protein 1 (Sfrp1), and gap junction protein, alpha 1, 43kDa (Gja1). We observed a significant increase in Lgr5, a cofactor for WNT receptors and stem cell marker, in the LE at 3 h, although none of its four coreceptors showed altered expression (data not shown). Beta-catenin, the major cofactor for the TCF/LEF transcription factors, showed reduced transcript levels following treatment (Supplemental Fig. S15F); however, total protein levels remained unaltered (data not shown). Nuclear localization of the transcriptionally active form of β-catenin was detected in the LE and stroma within 1 h of LIF administration (Fig. 8, Cb and Cc). By 6 h, nuclear localization had declined, and β-catenin reverted back to the cell membrane in LE and was reduced in the stroma (Fig. 8Cd). Wnt signaling is thought to be dependent on the presence of implanting blastocysts [94], but we predict that an active Wnt/β-catenin pathway, independent of embryos, is induced in the LE before implantation because we noted strong nuclear localization of β-catenin in the LE at 1 h (Fig. 8C).

The adherens junction protein E-Cadherin, which interacts with β-catenin, showed a decline in overall transcript and protein levels by 6 h (Fig. 8D and Supplemental Fig. S15, D and E). Immunolocalization revealed E-Cadherin was expressed exclusively in the LE apical, basal, and lateral membranes. By 6 h, the basal localization of E-cadherin expression had declined, whereas apical expression persisted (Fig. 8Cd). Such apical localization persists to D5 of pregnancy (data not shown).

Tgfβ Superfamily Pathway

The transforming growth factor beta (TGFβ) signaling pathway is upregulated at 3 and 6 h. In this pathway, the genes showing changes in expression were the activin receptors type 1 (Acvr1), type 2B (Acvr2b), TGF-β receptor 1 (Tbfbr1), and the TGF-β receptor 2 (Tgfbr2). The other genes induced were TGF-β2 (Tgfb2), the TGFβ-induced factor homeobox 1 (Tgif1), related RAS viral (r-ras) oncogene homolog 2 (Rras2). Bmp2, which mediates stromal decidualization, was not detected. We validated the transcript expression of Acvr1, Tgfb2, and Rras2 by qPCR. Tgfβ2 transcript levels increased over 6 hours (Supplemental Fig. S16A), whereas expression of both Acvr1 and Rras2 maximized by 3 h but declined by 6 h (Supplemental Fig. S16, B and C).

ACVR1 protein was expressed predominantly in the LE, GE, and subepithelial stroma at 0 h (Fig. 9Aa). ACVR1 expression then transiently declined in the stroma, GE, and LE (Fig. 9b), but by 3 and 6 h, expression had increased in the LE and slightly in the subepithelial stroma (Fig. 9, Ac and Ad). Similarly TGFβR1 was maximally expressed in the LE, GE, and subepithelial stroma at 0 h (Fig. 9Ba), and then also transiently declined in the LE at 1 h. TGFβR1 in the subepithelial stroma was not detected (Fig. 9Bb). However, by 3 and 6 h, TGFβR1 expression had increased in the LE, with some increase in the subepithelial stroma (Fig. 9, Bc and Bd).

FIG. 9

The TGFβ superfamily pathways are regulated by LIF. A) Expression of the receptor ACVR1 protein (in green) predominantly in the LE and marginally in the subepithelial stroma at 0 h. ACVR1 expression in the stroma and LE transiently declines by 1 h, but by 3 h, ACVR1 returns to the LE and slightly in the subepithelial stroma. Subpanels indicate expression (a: 0 h; b: 1 h; c: 3 h; d: 6 h; e: negative control). Bar = 100 μm. B) TGFβR1 (in green) expression transiently declines in the LE and subepithelial stroma by 1 h. Expression of TGFβR1 is restored to the LE and subepithelial stroma by 3 h. Subpanels indicate expression changes (a: 0 h; b: 1 h; c: 3 h; d: 6 h; e: negative control). Bar = 100 μm. C) Nuclear translocation of pSMAD3 (green) is induced by LIF at 1 h in LE and stroma and then falls. Subpanels indicate expression changes (a: 0 h; b: 1 h; c: 3 h; d: 6 h; e: negative control). Bar = 50 μm. D) Nuclear translocation of pSMAD5 (green) is induced at 3 and 6 h in the stroma but levels fall in the LE at the same time points. Subpanels indicate expression changes (a: 0 h; b: 1 h; c: 3 h; d: 6 h; e: negative control). Bar = 50 μm.

TGFβ pathway mediates transcription by phosphorylation and nuclear translocation of SMAD2/SMAD3 proteins. Although the array analysis did not show any changes in the expression of either of these proteins, alterations in SMAD3 phosphorylation and localization were detected. At 0 h, pSMAD3 appeared to localize to the apical submembranous cytoplasm and remained unchanged throughout the subsequent 6 h (Fig. 9Ca). However within 1 h, nuclear localization of pSMAD3 was observed throughout the LE and had also increased in the stroma (Fig. 9Cb). By 6 h, pSMAD3 expression had reverted back to the levels and distribution seen at 0 h (Fig. 9Cd). Stromal expression however declined by 3 h and was absent by 6 h (Fig. 9, Cc and Cd). Array analysis indicated that SMAD5 was the only SMAD protein increased by 3 h. Analysis revealed a fall in pSMAD5 nuclear localization in the LE at 3 h, followed by an increase at 6 h (Fig. 9, Dc and Dd). Cytoplasmic expression increased at 3 h but subsequently declined by 6 h (Fig. 9, Dc and Dd). In the stroma, increased expression of nuclear pSMAD5 was noted at 3 and 6 h (Fig. 9, Dc and Dd).