Friend leukemia integration 1 overexpression decreases endometrial receptivity and induces embryo implantation failure by promoting PART1 transcription in the endometrial epithelial cells

Patient enrollment and sample collection

This study recruited 20 patients with RIF who had no clinically clear etiology in the RIF group. Additionally, 20 subjects who achieved clinical pregnancy through 1–2 embryo transfers were recruited in the control group. Patients in the control group underwent in vitro fertilization solely due to male factors. Patients of both groups were assisted with artificial pregnancy through in vitro fertilization/intra-cytoplasmic sperm injection-embryo transfer. In the RIF group, the characteristics of the patients were as follows: failed clinical pregnancy, underwent 2–3 transplants, cumulative number of high-scoring blastocyst embryos transferred was ≥4, or cumulative number of high-scoring blastocysts transferred was 2–3. To reduce interference and ensure the credibility of the results, patients with other medical conditions were excluded from this study. Additionally, patients who received hormone therapy or other uterine procedures in the last three months were excluded. The endometrial tissue was harvested at days 6–8 post-ovulation in the natural cycle (embryo implantation window). This study was approved by the Ethics Committee at the First Affiliated Hospital of Zhengzhou University (2020-KY-256). Written informed consent was obtained from all study patients.

Single-cell suspension preparation

The endometrial tissue of patients was washed with pre-cooled phosphate-buffered saline (PBS) and incubated with ethylenediaminetetraacetic acid (EDTA) to reduce the stability of the cell membrane. Next, the endometrial tissue was digested with protease/collagenase to prepare the single-cell suspension. The mononuclear cells were prepared using single-cell nucleus separation solution and subjected to scRNA-seq.

Library construction and RNA sequencing

Library construction and RNA sequencing were performed, following the previously reported methods (Corces et al., 2017). Briefly, 50,000 cells were washed with PBS and lysed. Next, the lysate (50 µL) was centrifuged at 800 g and 4 °C for 10 min to obtain cell nuclei. The samples were immediately subjected to the transposition reaction. The transposition reaction mix (including the reaction buffer, Nextera Tn5 transposase (the amount of enzyme was optimized), and ribozyme-free water) was prepared and incubated with the prepared cell nucleus with mixing at 37 °C for 30 min. The DNA was purified using the Qiagen purification kit and eluted to obtain 10 µL of eluate. The transposable DNA was amplified using polymerase chain reaction (PCR) (first a pre-PCR was performed to evaluate the number of PCR cycles). The amplicons were purified and subjected to quality control using Bioanalyzer. The library was quantified using the Kapa library quantification kit (observe the presence of a typical nucleosome distribution pattern). High-throughput sequencing was performed using a computer.

Generation of single-cell gene expression matrices

Each nucleus of the qualified sample was mixed with reagents and magnetic beads with barcode sequence successively on the 8-channel microfluidic chip and covered with oil drops to form gel beads in extrusion containing single nuclei and single gel beads. Each gel bead carries several sequences that are combined with Illumina connector sequence (P5), 16 bp 10x cell barcode (Barcode), and some Illumina read1 primers.

The constructed library was qualified using Qubit 2.0. The insert DNA of the library was examined using Agilent Bioanalyzer High Sensitivity DNA chip. After the expected size of the insert was obtained, the effective concentration (2 nM) of the library was accurately quantified using quantitative real-time PCR (qRT-PCR) analysis to ensure the quality of the library. Next, sequencing was performed using the Nextseq 500 platform.

Quality control, cell type clustering, and major cell type identification

To ensure high-quality data, offline data quality control measures, such as the verification of sequencing quantity, barcode carrying rate, and sequencing quality Q30 value, were employed. Additionally, the barcode and sample index sequences were subjected to Q30 quality control to ensure high resolution at both the cell and sample levels.

The nucleosome, which comprises DNA and histone, is the basic structural unit of a chromosome. Each histone octamer comprises two histone molecules with approximately 200 bp of DNA sequence coiled around the octamer core structure to form a nucleosome. In this 200-bp sequence, the 146-bp DNA sequence coils directly around the histone octamer core and is resistant to nuclease digestion, while the remaining DNA sequence links the nucleosome to the next nucleosome. Nucleosome localization refers to the precise determination of nucleosome location on the genome, which is always accompanied by gene transformation from inhibition to transcription. Nucleosome localization, which has a major role in transcriptional regulation, DNA replication, and repair, is currently a research hotspot in epigenetic research. Based on nucleosome structural characteristics and the distribution of ATAC-seq inserted fragments, fragments within 147 bp of inserted fragments are designated as nucleosome-free regions, while fragments with a size of 147–294 bp are considered nucleosome distribution regions, providing accurate nucleosome localization.

The size distribution of ATAC-seq insertion fragments provides information on the packaging and location of nucleosomes, while the fragment length captures the periodicity of nucleosome positioning. Fragments with a size less than 147 bp indicate nucleosome-free regions, whereas those with a size of approximately 147–294 bp suggest a single nucleosome distribution region. The length and quantity of insertion fragments periodically vary due to the nucleosome structure. The fluctuation frequency is correlated with the length or number of nucleosomes.

To evaluate the number of cells captured in a barcode-labeled multi-cell mixed sample sequencing library, peak calling and insertion fragment data alignment to peak regions were performed. The number of cells successfully captured was determined by evaluating the insertion fragment count and length covering the peak region. Based on the barcode and corresponding insertion fragment count statistics, cells are classified into cellular and non-cellular categories. Cells that exceed the threshold of captured insertion sequences are classified as single cells, while those with insufficient captured insertion sequences are classified as non-cells.

The simultaneous sequencing of libraries from multiple cells was performed by 10x Genomics to obtain the ATAC data for multiple cells. The Cell Ranger ATAC package uses the Python library MOODS (https://github.com/jhkorhonen/MOODS) to scan for peak matches with group positions in each cell and annotate peaks using the JASPAR database to group cells. Two-dimensional cell clustering results are obtained using the t-distributed stochastic neighbor embedding (t-SNE) dimensionality reduction algorithm.

qRT-PCR analysis

Freshly collected endometrial tissue from the implantation stage was transferred to a culture dish, cut into pieces, and digested with collagenase at 37 °C for 30 min. The digestate was filtered using a filter to remove endometrial stromal cells and obtain primary endometrial epithelial cells. Total RNA was extracted from the samples. The isolated RNA (500 ng) was reverse-transcribed into complementary DNA using the HiScript II SuperMix (Vazyme, Beijing, China). qRT-PCR analysis was performed using the SYBR green master mix with the ABI 7500 System (Thermo Fisher Scientific, Waltham, MA, USA). The PCR conditions were as follows: 46 cycles of 94 °C for 10 min, 94 °C for 10 s, and 60 °C for 45 s. GAPDH was used as an internal reference.

Western blotting

The protein samples were subjected to sodium dodecyl sulfate-polyacrylamide gel electrophoresis using a 10% gel. The resolved proteins were transferred to a polyvinylidene difluoride membrane. The membrane was blocked with 5% skimmed milk powder solution for 1 h and incubated with the following primary antibodies overnight: anti-FLI1 (554266; 1:500, BD Biosciences, rabbit anti-vimentin (ab92547, 1:1,000, Abcam), rabbit anti-E-cadherin (ab40772, 1:1,000, Abcam), and anti- β-actin (#4970L, 1:5,000, Cell Signaling Technology) antibodies. Next, the membrane was incubated with secondary antibodies for 2 h. Immunoreactive signals were developed using enhanced chemiluminescence.

Source and analysis of long non-coding RNA (lncRNA) Data

The GSE26787 and GSE111974 datasets, which comprised gene expression data of five and 24 pairs, respectively, of endometrial tissues of pregnant women and patients with RIF, were downloaded from the Gene Expression Omnibus (GEO) database. Based on the similarity of lncRNA expression levels, a clustering heatmap was generated using the heatmap package in R software. The differentially expressed lncRNAs (DElncRNAs) between the RIF and control groups were identified based on the following criteria: —logFC—>1.5 and p < 0.01. The DElncRNAs of the two datasets were intersected to obtain the key lncRNAs that regulate the RIF process.

Evaluation of FLI1-PART1 binding

The interaction of FLI1 with PART1 in endometrial epithelial cells was examined using the chromatin immunoprecipitation (ChIP) assay. Briefly, cross-linking and cracking were performed to stabilize the complex. The nuclear components were separated with anti-FLI1 or anti-IgG antibodies. Chromatin fragmentation was performed, and protein-DNA complexes were captured and subjected to immunoprecipitation. Next, cross-linking and DNA purification were performed. The differential enrichment of PART1 between different groups was analyzed. The dual-luciferase reporter assay was used to further verify the interaction between FLI1 and PART1.

Cell lines and cell culture

The non-receptive endometrial epithelial cell line (HEC-1-A) and the receptive endometrial epithelial cell lines (Ishikawa and RL95-2) were purchased from Wuhan Punosai Life Technology Co., Ltd (Hubei, China). These cell lines have stable phenotypes and can be used to effectively simulate primary endometrial epithelial cells in vitro. BeWo cells were sourced from Wuhan Punosai Life Technology Co., Ltd (Hubei, China). Cell lines were cultured at 37 °C and 5% CO₂ in an incubator with saturated humidity. The cells were subcultured, and logarithmic phase cells were used for subsequent cell experiments.

Construction of stable FLI1-overexpressing and FLI1 knockdown cell lines

The FLI1 overexpression plasmid (OE-FLI1) and the corresponding negative quality control plasmid (technical control, OE-NC) were constructed by Shanghai Hanheng Biotechnology Co., Ltd (Shanghai, China). Meanwhile, the FLI1 knockdown plasmid (sh-FLI1) and the corresponding negative quality control plasmid (technical control, sh-NC) were constructed by Guangzhou Huijun Biotechnology Co., Ltd (Guangdong, China).

The sequences of sh-FLI1 and sh-NC are as follows:

5′-CCGGCGTCATGTTCTGGTTTGAGATCTCGAGATCTCAAACCAGAACATGACG TTTTTGAATT-3 ′ (sh-FLI1#1);

5′-CCGGCCCATGAACTACAACAGCTATCTCGAGATAGCTGTTGTAGTTCATGGG TTTTTGAATT-3 ′ (sh-FLI1#2);

5′-CCGGCCCTTCTGACATCTCCTACATCTCGAGATGTAGGAGATGTCAGAAGGG TTTTTGAATT-3 ′ (sh-FLI1#3);

5′-CCGGTCCTAAGGTTAAGTCGCCCTCGCTCGAGCGAGGGCGACTTAACCTTAGG TTTTTGAATT-3 ′ (sh-NC).

Logarithmic phase Ishikawa and HEC-1-A cells were randomly divided into the following four groups: OE-NC group, transfected with OE-NC construct; OE-FLI1 group, transfected with OE-FLI1 construct; sh-NC group, transfected with sh-NC construct; sh-FLI1 group, transfected with sh-FLI1 construct. Transfection was performed using Lipo8000™ (Beyotime Biotechnology, Shanghai, China). At 24 h post-transfection, the cell line stably expressing the target vectors was obtained.

Cell function assays

The effect of FLI1 on the proliferation of endometrial epithelial cells was examined using the cell counting kit-8 (CCK-8) kit (Beyotime Biotechnology, Shanghai, China), following the manufacturer’s instructions. Meanwhile, the effect of FLI1 on the migration of endometrial epithelial cells was evaluated using the scratch assay. Image J (National Institutes of Health, Bethesda, USA) was used to calculate the scratch area. The migration rate was calculated as follows: migration rate (%) = [(scratch area at 0 h –scratch area at 48 h)/(scratch area at 0 h) ×100%]. The effect of FLI1 on the epithelial-to-mesenchymal transition of endometrial epithelial cells was examined by evaluating the protein expression levels of vimentin and E-cadherin using western blotting. Furthermore, the effect of FLI1 on embryo adhesion was examined using the BeWo cells.

Mouse model construction and adeno-associated virus (AAV) infection

Sexually mature male and female ICR mice (aged 8 weeks) of specific pathogen-free (SPF) grade were purchased from Skbex Biotechnology Co., Ltd. (Heinan, China). The males were fertile. The animals were caged for mating in a ratio 1:2 (male:female). Pregnancy in female mice was confirmed through vaginal suppositories. Pregnant mice were selected for subsequent in vivo experiments.

Epithelial cell-specific PART1 overexpression AAV vector (AAV-PART1) and its corresponding negative control vector (AAV-NC) were constructed and packaged by Shanghai Hanheng Biotechnology Co., Ltd. (Shanghai, China). The virus titer was 1.00 × 10¹² VG/mL. AAV-NC and AAV-PART1 were injected into the mouse uterine cavity through the uterine horn. On days 0, 2, 5, and 8 of injection, endometrial epithelial cells were isolated to examine the overexpression efficiency of AAV-PART1 using qRT-PCR and western blotting.

AAV injection

Twelve pregnant mice were randomly divided into the following two groups (6 mice/group): AAV-NC and AAV-PART1 groups. This study was approved by the Ethics Committee at Zhengzhou University (No. ZZU-LAC20200828[08]). All mice were housed in an SPF animal room under the following conditions: temperature, 24–26 °C; humidity, 50%–60%; circadian cycle, 12-h light/dark cycle (lights on from 6:00 AM to 6:00 PM); feed, irradiated feed; water, sterile water; environmental enrichment, each cage contained nesting material and chew stick. At day 1.5 of pregnancy, pregnant mice were anesthetized with isoflurane inhalation and placed prone on a small animal operating table insulated at 37 °C. The skin on the mouse back was fully exposed. A small incision was introduced on the back to fully expose the uterine horn. An insulin syringe was used to inject 20 µL of AAV into the uterine horn (mice in the AAV-NC and AAV-PART1 groups were injected with AAV-NC and AAV-PART1 constructs, respectively). The wound was sutured, and the animal was placed back in the animal room for separate feeding after it regained consciousness. On day 9 of pregnancy, blue dye was injected into the tail vein, and the mice were euthanized using a CO₂ euthanasia chamber. The chamber was initially filled with a mixture of CO₂ and O₂ in a ratio 6:4. The CO₂ concentration was gradually increased to 100% after the animals lost consciousness. The mice were maintained in the chamber for a minimum of 10 min to confirm death during which they were completely unconscious. None of the animals survived at the end of the study. The uterus was removed to observe the number of implanted embryos.

Statistical analysis

All data are represented as mean ± standard deviation. GraphPad Prism 6.0 was used for statistical analysis and plotting. Means between two groups were compared using the Student’s t-test or Fisher’s exact test, while those between multiple groups were compared using one-way analysis of variance, followed by Dunnett’s test. Differences were considered significant at p < 0.05.

Single-cell clustering analysis and identification of cells in the endometrial tissue

Single-cell sequencing data were subjected to cluster analysis and cell gene labeling to identify the following eight types of cell populations in the endometrial tissue: stromal fibroblast cells, endometrial cells, epithelial cells, T cells, B cells, dendritic cells (DCs), smooth muscle cells, and neural progenitor cells (Fig. 1A). The cell populations of the RIF and control group were subjected to t-SNE nonlinear dimensionality reduction analysis. Seven cell types were identified in both RIF and control groups with stromal fibroblast cells accounting for the highest proportion (Fig. 1B). Furthermore, heatmaps were used to display the top 10 differentially expressed genes of seven cell types (Fig. 1C). The key genes in each cell group were analyzed, and these genes can serve as biomarkers for identifying cell types. CDH11, SPINK4, DMD, CD247, CD70, CD83, ITGA8, and UBE2C were biomarkers for stromal fibroblasts, endometrial cells, epithelial cells, T cells, B cells, DCs, smooth muscle cells, and neural progenitor cells, respectively (Figs. 1D–1K).

Endometrial epithelial cell numbers are downregulated and FLI1 expression is upregulated in patients with RIF

The frequency and proportion of different cell populations in the RIF and NC groups were statistically analyzed. Compared with that in the NC group, the proportion of interstitial fibroblasts and epithelial cells was significantly downregulated and the proportion of endothelial cells, T cells, and B cells was significantly upregulated in the RIF group. The difference in the proportion of epithelial cells between the RIF and NC groups was the highest (0.17% vs 20.94%, Fig. 2A).

ATAC-seq analysis of endometrial epithelial cells revealed that the chromatin accessibility was significantly different between the RIF and control groups. Most peaks were located at the enhancer position (enhancer was defined as the region that was >2 kb away from the transcription start site). The active enhancer histone-modified marker H3K27Ac was analyzed using ChIP-seq. Most of the upregulated peaks were enriched with H3K27Ac, suggesting that these peaks are mainly located in the active enhancer region, which may be related to gene transcription activation. These peaks were subjected to transcription factor (TF) motif analysis (MEME software). The most significantly enriched motif was FLI1, which was 65% enriched among all differentially upregulated peaks. Figure 2B shows the peak distribution of FLI1. Further analysis revealed that the chromatin accessibility of FLI1 in the endometrial epithelial cells of the RIF group was significantly higher than that in the endometrial epithelial cells of the control group (log FC = 0.27, p = 0.009). These findings suggest that FLI1 regulates embryo implantation.

The mRNA and protein expression levels of FLI1 in primary epithelial cells obtained from the endometrial tissues of the RIF and control groups were examined using qRT-PCR and western blotting. Compared with those in the control group, the endometrial cell mRNA and protein expression levels of FLI1 were markedly upregulated in the RIF group (Figs. 2C–2D). These data indicate that the chromatin accessibility of the TF FLI1 in endometrial epithelial cells of patients with RIF is enhanced. As the expression level of FLI1 was significantly upregulated, it may be involved in inducing decreased endometrial receptivity in RIF.

Effect of FLI1 on cell function in vitro

Non-receptive endometrial epithelial cell line (HEC-1-A cells) and receptive endometrial epithelial cell lines (both Ishikawa and RL95-2) were selected to perform the in vitro assays. qRT-PCR analysis revealed that the mRNA expression level of FLI1 in HEC-1-A cells (1.02 ±0.12) was significantly higher than that in Ishikawa (0.81 ±0.15) and RL95-2 cells (−3.91 ±0.50) (p < 0.0001, Fig. 3A). Thus, FLI1 expression is upregulated in non-receptive endometrial epithelial cells. Stable transfection lines were established by transfecting FLI1 overexpression and knockdown plasmids into Ishikawa and HEC-1-A cells, respectively (Figs. 3B–3C). FLI1 knockdown significantly increased the proliferation of HEC-1-A cells (p < 0.0001). Meanwhile, FLI1 overexpression significantly inhibited the proliferation of Ishikawa cells (p < 0.0001, Fig. 3D). These findings indicate that the proliferation of endometrial epithelial cells is downregulated upon FLI1 overexpression and upregulated upon FLI1 knockdown. The results of the scratch test revealed that FLI1 knockdown significantly increased the migration of HEC-1-A cells (Fig. 3E). Meanwhile, FLI1 overexpression significantly inhibited the migration of Ishikawa cells (Fig. 3F). Thus, the migration of endometrial cells is downregulated upon FLI1 overexpression and upregulated upon FLI1 knockdown. FLI1 knockdown upregulated vimentin protein expression and downregulated E-cadherin protein expression in HEC-1-A cells (Figs. 3G–3H). Meanwhile, FLI1 overexpression significantly downregulated vimentin protein expression and upregulated E-cadherin protein expression in Ishikawa cells (Figs. 3G–3H). These findings indicate that the epithelial-to-mesenchymal transition of endometrial epithelial cells is downregulated upon FLI1 overexpression and upregulated upon FLI1 knockdown. The ability of FLI1-overexpressing Ishikawa cells to adhere to BeWo cells was significantly downregulated (Fig. 3I). In contrast, the ability of FLI1 knockdown HEC-1-A cells to adhere to BeWo cells was significantly upregulated (Fig. 3I). These findings suggest that the ability of endometrial cells to adhere to BeWo cells was downregulated upon FLI1 overexpression and upregulated upon FLI1 knockdown.

Effect of FLI1 on mouse embryo implantation

To determine the effect of FLI1 on mouse embryo implantation, AAV-OE-NC and AAV-OE-FLI1 were injected into the uterus of female mice through the uterine horn. The FLI1 expression levels in the endometrial epithelial cells of each group of mice on days 0, 2, 5, and 8 post-injection were examined using qRT-PCR and western blotting. qRT-PCR analysis revealed that the FLI1 mRNA expression levels were significantly upregulated in female mouse endometrial epithelial cells from day 2 post-AAV-OE-FLI1 injection (Fig. 4A). Consistent with the qRT-PCR analysis results, western blotting revealed that the expression level of FLI1 protein in female mouse endometrial epithelial cells was significantly upregulated from day 2 post-AAV-OE-FLI1 injection (Fig. 4B). AAV-OE-FLI1 injection into the uterine horn promoted FLI1 overexpression in mouse endometrial epithelial cells. Furthermore, the number of implanted embryos in the AAV-OE-FLI1-injected group was significantly lower than that in the AAV-OE-NC-injected group (Fig. 4C). These data indicate that epithelial cell-specific overexpression of FLI1 can inhibit mouse embryo implantation.

Molecular mechanism underlying the inhibitory effects of FLI1 on embryo implantation

In the GSE26787 dataset, 22 differentially expressed lncRNAs (DElncRNAs) were identified between the paired endometrial tissues of pregnant women and patients with RIF (13 upregulated lncRNAs and nine downregulated lncRNAs) (Fig. 5A). Meanwhile, in the GSE111974 dataset, 72 DElncRNAs were identified between the paired endometrial tissues of pregnant women and patients with RIF (55 upregulated lncRNAs and 17 downregulated lncRNAs) (Fig. 5B). LOC100505912 and PART1 were the only common DElncRNAs between GSE26787 and GSE111974 datasets. The expression levels of LOC100505912 and PART1 were significantly upregulated in the endometrial tissue of patients with RIF (Fig. 5C). Therefore, we hypothesized that the dysregulated LOC100505912 and PART1 in the endometrial tissue are the key lncRNAs regulating RIF.

qRT-PCR analysis revealed that the expression of PART1 was significantly upregulated in FLI1-overexpressing endometrial epithelial cells and significantly downregulated in FLI1 knockdown endometrial epithelial cells (Fig. 6A), indicating that FLI1 regulates PART1 expression. However, FLI1 overexpression or knockdown in endometrial epithelial cells did not affect the expression level of LOC100505912 (Fig. 6B). Meanwhile, analysis with the JASPAR database (https://jaspar.genereg.net/) revealed that FLI1 had binding sites in the PART1 promoter region with a score of 0.86 (Fig. 6C) but not in the LOC100505912 promoter.

The binding of FLI1 to PART1 was examined using the ChIP and dual-luciferase reporter assays. The ChIP assay results revealed that the enrichment level of PART1 in the anti-FLI1 group was significantly higher than that in the anti-IgG group (Fig. 6D), indicating the interaction between FLI1 and PART1 promoter. The results of the dual-luciferase reporter assay demonstrated that the wild-type PART1 promoter, but not the mutant PART1 promoter, enhanced the luciferase reporter activity of FLI1 (Fig. 6E). These data indicated that FLI1 directly promotes PART1 transcription by binding to its promoter.

To determine the effect of PART1 on mouse embryo implantation, AAV-OE-NC and AAV-OE-PART1 were injected into the uterus of pregnant mice through the uterine horn. The PART1 expression levels in the endometrial epithelial cells of different groups were examined using qRT-PCR analysis on day 8 post-injection. Injection with AAV-OE-PART1 significantly upregulated the expression level of PART1 mRNA in female mouse endometrial epithelial cells (Fig. 7A). Furthermore, the number of implanted embryos in the AAV-OE-PART1-treated group was significantly lower than that in the AAV-OE-NC-treated group (Fig. 7A). Thus, epithelial cell-specific overexpression of PART1 can inhibit mouse embryo implantation.

The results of the cell function experiments revealed that the proliferation of Ishikawa cells was significantly downregulated upon FLI1 overexpression and significantly upregulated upon PART1 knockdown (Fig. 7B). The scratch test results revealed that the migration of Ishikawa cells was significantly downregulated upon FLI1 overexpression and significantly upregulated upon FLI1 knockdown (Fig. 7C). Western blotting demonstrated that FLI1 overexpression downregulated the vimentin levels and upregulated the E-cadherin levels in Ishikawa cells, whereas PART1 knockdown exerted the opposite effects (Fig. 7D). FLI1 overexpression inhibited the epithelial-to-mesenchymal transition of endometrial epithelial cells by upregulating PART1 expression. The ability of Ishikawa cells to adhere to BeWo cells was significantly downregulated upon FLI1 overexpression and significantly upregulated upon PART1 knockdown (Fig. 7E).

These results indicate that the binding of FLI1 to the PART1 promoter promotes the transcription and expression of PART1. Epithelial cell-specific PART1 overexpression inhibited mouse embryo implantation and endometrial cell proliferation, invasion, epithelial-to-mesenchymal transition, and adherence to BeWo cells. Thus, PART1 contributes to decreased endometrial receptivity and embryo implantation failure.