Free apical surface hetero-cellular CDH1 homophilic binding is a major mediator of blastocyst–endometrium interaction in humans

Abstract

STUDY QUESTION

Is E-cadherin (CDH1) expressed on the free apical surface of endometrial epithelial cells (EECs) involved in embryo attachment?

SUMMARY ANSWER

Embryonic signals induced apical expression of endometrial CDH1, which facilitated attachment via hetero-cellular CDH1 homophilic binding.

WHAT IS KNOWN ALREADY

Understanding human blastocyst–endometrium interaction helps fertility treatment by facilitating the evaluation of endometrial receptivity and blastocyst quality. CDH1 is an adhesion molecule commonly known for the maintenance of epithelial integrity through adherens junction formation on the lateral membranes of adjacent epithelial cells. The apical region of some epithelia also expressed CDH1 of unknown function.

STUDY DESIGN, SIZE, DURATION

Laboratory experimental study.

PARTICIPANTS/MATERIALS, SETTING, METHODS

The potential apical cell plasma membrane proteins participating in adhesion between blastocysts and EEC were identified by comparing the surface proteomes of polar trophectoderm-like trophoblastic spheroids (BAP-EB) and receptive EEC. The apical surface expression of CDH1 in human blastocysts and primary EEC was confirmed by live-cell immunofluorescent staining. Antibody blocking and gene knockdown approaches were used to study the functional roles of hetero-cellular homophilic binding of CDH1. The adhesiveness of EEC and polar trophectoderm-like cells was confirmed by atomic force microscopy upon gene knockdown. The embryonic signals induced endometrial apical surface relocations of CDH1 and its stabilizer delta-1 catenin (CTNND1) were studied by treatments with conditioned media of BAP-EB and human blastocysts, as well as with HCG. The correlations of endometrial CDH1 and CTNND1 with pregnancy outcomes were confirmed by western blotting analysis of their protein expressions in EEC and immunofluorescent staining of endometrium collected from women who gave live birth, those who failed to become pregnant after IVF, and those with repeated implantation failure.

MAIN RESULTS AND THE ROLE OF CHANCE

We identified 27 pairs of potential interactive cell plasma membrane transmembrane proteins between BAP-EB and EEC, and CDH1 homophilic binding was one of them. CDH1 was localized to the apical surface of receptive EEC and the polar trophectoderm of human blastocysts. The CDH1 expression in EEC was higher during the window of implantation than those at the pre-receptive phase. With the use of atomic force microscopy, we revealed that the adhesiveness of the apical surface of polar trophectoderm-like cells and EEC decreased when CDH1 was silenced. The attachment rates of BAP-EB onto EEC were decreased when the surface CDH1 of BAP-EB and/or the EEC was blocked by an anti-CDH1 antibody, indicating that homophilic binding of trophectodermal CDH1 with endometrial CDH1 mediated the attachment of BAP-EB onto EEC. Besides cell surface adhesiveness, knockdown of CDH1 or CTNND1 in EEC also reduced expression of adhesion-related molecules and BAP-EB attachment. The conditioned media of BAP-EB and human blastocysts and HCG enhanced apical expression of CDH1 in EEC, suggesting embryonic signals, in particular, modulated the expression of CDH1 and stimulated the adhesion of EEC. The protein expression of CDH1 was significantly higher in EEC isolated from patients who gave live birth as compared to those with repeated implantation failure. Importantly, the expression of CDH1 in the apical region of the luminal epithelium of the endometrium from fertile women was also significantly higher than that from women with repeated implantation failure.

LARGE SCALE DATA

none.

LIMITATIONS, REASONS FOR CAUTION

This study only includes in vitro experiments. Due to the limited availability of human blastocysts and difficulty in long-term expandable culture of primary EEC, embryo surrogates BAP-EB and endometrial adenocarcinoma cell lines were used for most of the experimental work. The sample size of the primary EEC for protein analysis in this study was small, and the 2D culture of primary EEC in vitro might have impaired its native epithelial polarity.

WIDER IMPLICATIONS OF THE FINDINGS

Endometrial factors can lead to infertility, but the exact molecules involved in endometrial receptivity remain unclear. Human endometrial receptivity is best assessed by determining whether a human blastocyst can attach and implant onto the endometrium of interest. With the use of a human embryo surrogate named BAP-EB, we demonstrated the important roles of CDH1 and CTNND1 in human embryo attachment. In particular, we showed that homophilic binding between trophectodermal CDH1 and epithelial CDH1 was one of the major mediators of the first contact in embryo–maternal interaction. Embryo-derived factors induced apical redistribution of CTNND1 and CDH1 in the EEC. Most importantly, the expression of CDH1 on the apical region of EEC represented endometrial receptivity. The results strengthen our understanding of embryo–maternal interaction in humans, and CDH1 could potentially be used for predicting IVF outcomes, which helps clinicians to better counsel the couples who fail with ART.

STUDY FUNDING/COMPETING INTEREST(S)

This study was supported by Grant for Fertility Innovation 2016 from Merck; Health and Medical Research Fund (grant numbers: HMRF 04151546; 10212996; 11222296) from the Food and Health Bureau, Government of the Hong Kong Special Administrative Region; InnoHK initiative of the Innovation and Technology Commission of the Hong Kong Special Administrative Region Government (Health@InnoHK); Shenzhen Science and Technology Program (Grant No. KQTD20190929172749226); General Research Fund (grant number: 17212922); Collaborative Research Fund C7100-22GF from the Research Grants Council of Hong Kong; and Shenzhen Sanming Project of Medicine (SZSM202211014). The authors have no conflicts of interest to declare.

TRIAL REGISTRATION NUMBER

N/A.

Introduction

Infertility has been a major healthcare issue in recent decades. According to the World Health Organization, one in seven couples suffer from infertility. Despite advances in IVF technologies, such as time-lapse embryo monitoring and preimplantation genetic testing, the success rate remains unsatisfactory, partly due to implantation failure of unknown etiologies (Bashiri et al., 2018; Hernández-Vargas et al., 2020). It is estimated that up to two-thirds of implantation failures are related to endometrial factors (Craciunas et al., 2019). The causes of implantation failure are still poorly understood, and reliable biomarkers of endometrial receptivity are unavailable (Hernández-Vargas et al., 2020).

Implantation involves apposition, adhesion, and invasion of a blastocyst into the endometrium (Bischof and Campana, 1997; Massimiani et al., 2019). Apposition was initially thought to occur via weak interaction between L-selectin and its ligand because endometrial pinopodes express L-selectin ligand (Nejatbakhsh et al., 2012), and endometrial L-selectin ligand level is positively correlated with IVF outcome (Wang et al., 2008). However, L-selectin is absent in human blastocysts (Bloor et al., 2002), and L-selectin-deficient mice are fertile (Robinson et al., 1999). The first maternal contact with the blastocyst is believed to occur via cell adhesion molecules, such as integrins on trophectoderm (TE) and endometrial epithelial cells (EECs) (Achache and Revel, 2006). The use of integrin expression levels as a predictive marker of IVF outcome is controversial (Quenby et al., 2007).

E-cadherin (CDH1) is best known as a component of adherens junctions on the lateral membrane of epithelial cells. Like other epithelia, the endometrial epithelium expresses CDH1-containing adherens junctions. Most data on the physiological roles of CDH1 in implantation are from mice or in vitro human cell line models. Uterine infusion of anti-CDH1 antibody or knockout of CDH1 impairs embryo implantation in mice, and the effects are attributed to alterations in endometrial functions, including altered MMP-2 and -9 expression (Liu et al., 2006) and disorganized uterine epithelium development (Reardon et al., 2012), respectively. Although CDH1 is expressed in the TE and endometrial epithelium in mice at implantation, and there is a redistribution of CDH1 from the basolateral to apicolateral membrane of the EEC during progression of implantation (Tiwari et al., 2021), immunoelectron microscopy revealed that CDH1 is not responsible for the initial contact between the TE and EEC (Kadokawa et al., 1989). The current evidence does not support a direct involvement of CDH1 in the implantation of mouse blastocysts.

The involvement of CDH1 in human implantation is unclear. In fact, there are controversies on whether CDH1 expression in human endometrium is cyclically regulated, an important characteristic of implantation-related molecules. Some studies found a stable CDH1 level during the menstrual cycle (e.g. Dawood et al., 1998; Poncelet et al., 2002), while others reported elevated CDH1 levels at the secretory phase (Stern-Tal et al., 2020). On the other hand, reduced endometrial CDH1 expression in the mid-secretory phase is related to repeated implantation failure (RIF) (Yang et al., 2017), suggesting a positive relationship between CDH1 and fertility. Interestingly, transmission electron microscopy reveals the formation of adherent junctions between the apical cell surface of the TE cells and the EEC in an in vitro human blastocyst implantation model (Fukuda and Sugihara, 2012). However, whether CDH1 is involved in the formation of these adherent junctions remains unclear.

Apart from EEC, CDH1 is expressed in preimplantation embryos (Rowlands et al., 2000). Although CDH1 knockout mouse embryos fail to implant, the failure is due to a lack of TE formation resulting from dissociation of the blastomeres of the morula shortly after compaction (Larue et al., 1994; Riethmacher et al., 1995). In humans, CDH1 expression shifts from the basolateral region in the morula to the apicolateral region in blastocysts, consistent with its roles in embryo compaction and cavitation (Meistermann et al., 2021). Interestingly, CDH1 protein is also localized to the free apical membrane of TE cells (Alikani, 2005; Meistermann et al., 2021). The function of CDH1 on the apical surface of human TE cells is unknown.

Studying human implantation is difficult because of the lack of human embryos for research. To overcome the problem, we derived TE-like spheroids (BAP-EB) from human embryonic stem cells as embryo surrogates (Lee et al., 2015; Yue et al., 2020). BAP-EB at 48 h post-induction of differentiation (BAP-EB-48 h) exhibited a blastocyst-like cystic structure. The transcriptome of BAP-EB-72 h resembled that of Day 7 human polar TE (pTE) (Yue et al., 2020). Importantly, the BAP-EB-72 h attached to the receptive but not the non-receptive EEC and subsequently invaded the EEC and endometrial stromal cells (Lee et al., 2015; Yue et al., 2020). The migration of endometrial stromal cells toward BAP-EB-96 h was higher in women with live births (LBs) than those with RIF (Qiu et al., 2023). Our clinical trial further indicated that the attachment of BAP-EB-72 h onto EEC modestly predicted cumulative LBs in women aged ≥35 years (Lee et al., 2023).

Here, we hypothesized that the free apical surface CDH1 is involved in adhesion of the TE cells to the EEC during embryo attachment. To test the hypothesis, we compared the surface proteomes of TE cells and EEC and demonstrated that homophilic binding of apical surface CDH1 between the TE cells and EEC contributes significantly to blastocyst adhesion. Homophilic binding of CDH1 on the lateral membrane of adjacent cells of the same type as in the epithelium is well studied. To the best of our knowledge, homophilic CDH1 binding on the free apical surface of different cell types from two separate entities has not been reported, though hetero-cellular CDH1 interactions occur between epidermal keratinocytes and Langerhans cells, and melanocytes and Merkel cells within the skin tissue (Tang et al., 1993, 1994; Werling et al., 2011).

Knowledge of embryo–endometrium interaction is important for understanding the implantation process, the causes of infertility, and the development of fertility treatment. In this report, we demonstrated for the first time that free apical surface homophilic CDH1 binding contributed to hetero-cellular TE-EEC adhesion. Paracrine communication, in particular HCG from human blastocysts and blastocyst surrogate BAP-EB, modulated the expression of CDH1 and its membranous stabilizer Catenin Delta-1 (CTNND1) of EEC and enhanced the binding of the spheroids to EEC. The positive correlation of CDH1 and CTNND1 expressions in EEC and the apical CDH1 expression in luminal epithelium with successful IVF outcomes indicated their potential role as biomarkers for endometrial receptivity.

Materials and methods

Human endometrial tissue collection and primary EEC isolation

Endometrial aspirates were obtained from women who were on the waiting list for IVF at Queen Mary Hospital. Written informed consents were obtained from all recruited subjects. The study protocol was approved by the Institutional Review Boards of the University of Hong Kong/Hospital Authority Hong Kong West Cluster (IRB: UW 16-094). Primary EECs were isolated from endometrial aspirates collected either 2/3 days after LH surge (LH + 2/3) or 7/8 days after LH surge (LH + 7/8), which corresponded to pre-receptive and receptive stages, respectively. The detailed method of primary EEC isolation has been described elsewhere (Lee et al., 2015). The isolated LH + 7/8 primary EECs were cultured in medium containing 500 pM estradiol (E2, Sigma, Burlington, MA, USA) and 50 nM progesterone (P4, Sigma) for 3–4 days. Recruited subjects underwent ovarian stimulation for IVF and embryo transfer in fresh and/or frozen–thawed transfer cycles. The cumulative LB rates, i.e. from fresh or frozen transfer within 6 months of ovarian stimulation, were correlated with the protein expressions in the isolated EECs. Women aged <40 who failed to get pregnant after transferring more than four good-quality cleavage-stage embryos or two blastocysts in consecutive cycles were considered as the RIF group (Coughlan et al., 2014).

Cell lines

Human embryonic stem-cell line VAL3 was obtained from the Spanish Stem Cell Bank (Spain). Cells were cultured in Matrigel (BD Bioscience, New Jersey, USA)-coated plates with mTeSR^TM Plus medium (STEMCELL Technologies, Vancouver, Canada). Human endometrial adenocarcinoma Ishikawa cells (Sigma) were cultured in Minimum Essential Medium (MEM, Sigma) supplemented with 10% fetal bovine serum (FBS, Thermo Fisher Scientific, Waltham, MA, USA), 1% pen–strep (P/S, Thermo Fisher Scientific), and 1% L-glutamine (L/G, Thermo Fisher Scientific). Human endometrial adenocarcinoma HEC-1B cells were cultured in DMEM/F12 (Sigma) supplemented with 10% FBS, 1% P/S, and 1% L/G.

Human blastocyst and spent culture media collection

Human blastocyst spent culture media (SCM) were obtained from infertile women undergoing IVF as described above. SCM of single blastocysts (∼8 µl) were frozen and stored at −80°C individually. Human blastocysts on Day 6 post-fertilization collected in the assisted reproduction program at The University of Hong Kong—Shenzhen Hospital were used for live-cell immunocytochemical staining, and the protocol was approved by the Ethics Committee of The University of Hong Kong—Shenzhen Hospital (hkuszh2022082).

Trophoblastic spheroid formation and conditioned media collection

Trophoblastic spheroids (BAP-EB) and trophoblastic cell monolayers (BAP/2D) were derived from VAL3 according to our established protocols (Lee et al., 2015; Yue et al., 2020). Briefly, VAL3 were aggregated overnight to form embryoid bodies (EB) or seeded as monolayered cells and subjected to differentiation by BAP [10 ng/ml BMP4 (R&D Systems, Minneapolis, USA), 1 μM of A83-01 (Stemgent, Beltsville, MD, USA), and 0.1 μM of PD173074 (Stemgent)] for 24 h (BAP-EB-24 h or BAP/2D-24 h), 48 h (BAP-EB-48 h or BAP/2D-48 h), and 72 h (BAP-EB-72 h or BAP/2D-72 h). Conditioned media (CM) of BAP-EB-48 h (48 h-CM) and BAP-EB-72 h (72 h-CM) were obtained at 48 h and 72 h, respectively, by collecting CM and saving the supernatant after centrifugation at 300g for 5 min.

Membranous protein isolation

Membranous proteins of Ishikawa and HEC-1B were extracted using the Native Membrane Protein Extraction Kit (Sigma); the procedure is described in Fernando et al. (2021). Collected proteins were used for western blotting analysis.

Surface protein isolation and mass spectrometry analysis

Surface proteins were isolated using the Pierce Cell Surface Protein Isolation Kit (Thermo Fisher Scientific). Endometrial cells or BAP-EB were washed with ice-cold BupH™ Phosphate Buffered Saline (PBS), and surface proteins were biotin labelled by incubation with Sulfo-NHS-SS-Biotin for 30 min at 4°C. The reaction was stopped by aspirating Sulfo-NHS-SS-Biotin and washing with BupH™ Tris Buffered Saline. Cells were then lysed with the lysis buffer. Labelled surface proteins were captured by NeutrAvidin Agarose and eluted by SDS-PAGE. Sample buffer supplemented with 50 mM dithiothreitol. The isolated surface proteins were ready for downstream analysis. Surface proteins isolated from BAP-EB-0 h, -48 h, and -72 h were identified by LC-MS (Mass Spectrometry Proteomics Services, The University of Hong Kong) using an online reverse-phase nanoLC coupled to an Orbitrap Fusion Lumos mass spectrometer. Data analysis was performed by the SEQUEST engine and validated using the Proteome Discoverer software.

Co-immunoprecipitation assay

To demonstrate the homophilic binding of CDH1 with the apical cell surface CDH1 on BAP-EB-72 h and Ishikawa cells, recombinant His-tagged CDH1 (CDH1-His) was used as the bait in the co-immunoprecipitation (Co-IP) assay using standard procedures. Briefly, the solvent buffer of surface proteins isolated from BAP-EB-72 h and Ishikawa was changed to immunoprecipitation buffer (10 mM HEPES, pH 7.5, 2.5 mM MgCl₂, 50 mM KCl, 100 mM NaCl, 10% glycerol, 0.5% NP40, and 0.9 mM CaCl₂) supplemented with protease inhibitor cocktails (Thermo Fisher Scientific) using an Amicon Ultra-0.5 Centrifugal Filter Unit (Millipore, Massachusetts, USA) before incubation with 2.5 μg/ml of CDH1-His (R&D Systems, Minnesota, USA) at 4°C overnight. Meanwhile, protein-G conjugated Dynabeads^TM (Thermo Fisher Scientific) were incubated with a 1:25 dilution of polyclonal anti-His-Tag antibody (Cell Signaling Technology, Massachusetts, USA) at 4°C overnight. The conjugated Dynabeads^TM were then mixed and incubated with the surface protein and CDH1-His complex at 4°C overnight. The beads were then washed four times with immunoprecipitation wash buffer (10 mM HEPES, pH 7.5, 2.5 mM MgCl₂, 50 mM KCl, 100 mM NaCl, 1% NP40, and 0.9 mM CaCl₂) before protein elution with sample buffer (Takara Bio Inc., Shiga Prefecture, Japan) at 80°C for 10 min. The protein content in the resulting samples was identified by western blotting after separation by SDS-PAGE. Polyclonal goat anti-CDH1 antibody (R&D Systems) was used for western blots of CDH1.

siRNA transfection

siRNAs of CTNND1, CDH1, RAB11A, or a negative control (Thermo Fisher Scientific) were transfected into Ishikawa cells at a final concentration of 125, 250, and 500 nM using Lipofectamine^® 2000 reagent (Thermo Fisher Scientific). After 4-h of culture, the media were topped up with MEM medium and cultured overnight. The transfected cells were then maintained in fresh MEM medium for another 2 days and harvested for further analysis.

Real-time quantitative PCR, western blotting, and immunocytochemistry

To measure transcript expression of the surface or membranous proteins, total RNAs were extracted from Ishikawa and HEC-1B cells using the mirVana^TM miRNA isolation kit (Ambion, Life Technologies) following the manufacturer’s instructions. RNA was reverse transcribed (RT) using the PrimeScript™ RT reagent kit (Takara, Japan), followed by real-time quantitative PCR using the Takara Premix Ex Taq™ (Takara) on a QuantStudio 5 Real-Time PCR System (Thermo Fisher Scientific). The data were analyzed by the software provided by the manufacturer (Thermo Fisher Scientific), normalizing to endogenous 18S ribosomal RNA (internal control) using the 2^−ΔΔC_T method.

The surface or membranous proteins were subjected to SDS-PAGE followed by western blotting analysis. Primary antibodies against CTNND1 (Abcam, Bristol, UK), CDH1 (Abcam and R&D), CTNNA1 (Abcam), CTNNB1 (Cell Signalling Technology, Massachusetts, USA), Integrin αV (Abcam), Annexin-A2 (Abcam), and Rab11a (Abcam) were used in the protein analysis. Densitometry of western blotting results was performed with the ImageJ software (National Institutes of Health, USA).

Conventional immunocytochemistry (fixed-cell ICC) was performed by fixing the cells with 4% paraformaldehyde (PFA, diluted in PBS; Sigma). The fixed cells were permeated with 0.1% Triton before blocking with the corresponding normal serum. After overnight primary antibody incubation followed by 1 h of fluorescence-conjugated secondary antibody incubation, the cell nuclei were stained with 4′,6-diamidino-2-phenylindole (DAPI, 1:1000; Sigma). Live-cell immunocytochemistry (live-cell ICC) was based on a published protocol (Zhuang and Matsunami, 2008) with modifications. Briefly, cells were incubated with primary antibodies against CTNND1 (1:100) and CDH1 (1:100) that were diluted in corresponding culture media on ice for 1 h. Corresponding normal IgG was used as a negative control. After washing out the primary antibody by cold culture media, fluorescence-conjugated secondary antibodies (1:500) diluted in culture media were incubated with endometrial cells or BAP-EB for 1 h on ice in the dark. After washing with culture media and PBS, endometrial cells or BAP-EB were fixed by 4% PFA at room temperature for 20 min or 1 h, respectively. The nuclei were stained by DAPI. Fluorescence images were obtained by using a Carl Zeiss LSM780 or an LSM900 confocal microscope (Germany) at the Faculty Core Facility, The University of Hong Kong, and the images were processed with the ZEN software (Carl Zeiss, Germany).

Quantification of CTNND1 and CDH1 in single cells

The quantification of Ishikawa cells was achieved by seeding Ishikawa cells at a density of 9 × 10⁴ cells/cm² on a µ-Slide (ibidi, Gräfelfing, Germany). Two days later, the cells were treated with 48 h-CM, 72 h-CM, or 50 IU/l HCG (Sigma; diluted in MEM phenol red-free medium) for 30 min, 1 h, and 3 h. For human blastocyst SCM treatment, Ishikawa cells were seeded at a density of 6 × 10⁴ cells per droplet on a µ-Slide before treatment by SCM on the next day for 3 h. Conventional ICC was performed. A randomly selected region was imaged, and 12 optical z-slices were acquired for a total of 18.04 μm through a Carl Zeiss LSM 780 confocal microscope. Quantification was performed according to a published method (Tiwari et al., 2021) with modifications. To calculate the percentage of protein fluorescence intensity in the apical region (top 10% of cell height) over that in the entire cell, intensity values of each pixel along the entire height of individual cells were quantified from apical to basal side using ImageJ. All individual cells from two XZ views of each condition were quantified. BAP-CM and HCG data were collected from two independent replicates, while individual human blastocyst SCM (n = 18) were used for the cell treatment. Single-cell 3D surface plots were generated using ImageJ.

The quantification of luminal epithelium of primary endometrial tissue was performed by conducting immunohistochemistry of CDH1 and CTNND1 on the paraffin sections of endometrial tissues. Images were captured through a Carl Zeiss LSM 900 confocal microscope. Apical and basal boundaries of luminal epithelium were labeled using the Labelme annotation tool (Wada, 2016). Using Python 3.7, points from the apical and basal boundaries were uniformly sampled, and vectors were created between them. The outermost 10% of each vector was then used to generate a line that, together with the original apical boundary, outlined the luminal epithelium’s apical region. Signal intensities were measured in the apical region and the entire region of luminal epithelium. The percentage of apical CDH1 or CTNND1 was calculated as the ratio of signal intensity in the apical region to the intensity in the entire luminal epithelium.

BAP-EB attachment and spreading assays

Attachment assay was performed as described (Lee et al., 2015). Briefly, blocking of CDH1 on BAP-EB-72 h or Ishikawa cells was achieved by pretreatment with 5 μg/ml anti-CDH1 antibody or 10 μg/ml CDH1 blocking peptide (H-SWELYYPLRANL-NH2, Severn Biotech Ltd., UK) (Devemy and Blaschuk, 2009) for 2 h. After washing away the unbound antibodies or peptides with PBS, BAP-EB were cocultured with Ishikawa cells for 1 h. The unattached BAP-EB were removed by centrifugation at 10g for 10 min. The attachment rate was calculated as the percentage of attached BAP-EB out of the total number of BAP-EB seeded. Spreading assay was performed by co-culture of BAP-EB-72 h with Ishikawa cells for 24 h. The spreading area of each spheroid was measured by the ISCapture software (Tucsen, China) or Image J.

Atomic force microscopy imaging on live cells

Atomic force microscopy (AFM) images were generated on live cells using a Resolve BioAFM (Bruker Inc.). The cells were washed with PBS and refreshed with new culture medium before image capture. The whole process was maintained at 37°C, and disturbance of the laser light was prevented by fully immersing the probe into the culture medium. Imaging was performed at the Peakforce tapping mode with the Peakforce QNM-LC-A probes. The spring constant was set at 0.08–0.11 N/m with a small tip for 3-dimensional images and mechanical property mappings. Data were recorded at a scan rate of 0.1–0.3 Hz, an applied force of 100–400 pN, and a Peakforce tapping frequency of 0.5–1 khz with a resolution of 256  ×  256 pixel². Multiple scan areas varying from 2  ×  2 μm to 40  ×  40 μm were taken for each condition. To monitor the details of surface morphology, the scan size was set at 30  ×  30 μm to 40  ×  40 μm to ensure that the whole cell body was scanned properly.

Force curves generation, mechanical analysis, and functionalization of tips

The cell stiffness and CDH1 binding events were described by the Young’s modulus and rupture event force value, respectively. The force–distance curves were obtained in the Force Volume mode. The curves were recorded with a moderated applied force of 400–800 pN, a ramp size of 800–1400 nm, and a ramp frequency of 1 Hz. Each force–distance curve included the approach part and the retract part, where the tips approached to and departed from the cells, respectively. For each cell, a map of 16 × 16 curves was recorded in a 2000 × 2000 nm² area on the central apical region of the cell membrane to avoid edge effects.

For mechanical analysis, the Young’s modulus was calculated by the approach part of the curves fitting with the Sneddon model over a tip radius of 20 nm and considering a conical tip with an opening angle of 15°.

$$F=\mathrm{ }\frac{2}{\pi }\frac{E}{1-{\nu }^2}\mathrm{t}\mathrm{an}\left(\alpha \right){\delta }^2$$

where υ was the Poisson ratio of the cell, with a typical value of 0.5 for the cell. Δ was the indentation depth, which was equal to the z movement of the system subtracting the deflection of the probe. $\alpha$was the half-opening angle of the AFM tip. F was the loading force calculated by multiplication of the spring constant of the probe and the deflection of the cantilever. The adhesive forces of Ishikawa cells were obtained from the retract part of the curves, which equaled to the force difference at a contact point and adhesive force peak location (Li et al., 2021).

Functionalization of AFM tips was performed to assess adhesive forces between specific molecules. AFM DNP-10-A tips with a stiffness of 0.35 N/m and a resonance frequency in air of 65 kHz were used. The tip was first subjected to acetone washing, followed by ultrasonic cleaning with deionized (DI) water. It was then functionalized with 3% (3-aminopropyl)triethoxysilane (Sigma-Aldrich) diluted in acetone for 30 min to acquire the alkoxysilane molecules. Subsequently, the AFM tips were washed with DI water and incubated with 0.5% PEG crosslinker polyethylene glycol 2-aminoethyl ether acetic acid (Sigma-Aldrich) diluted in DI water for 30 min. After washing with DI water, the tips were then immersed in 30–60 µg/ml recombinant CDH1 proteins (R&D Systems) for 1 h at 30°C. The unreacted aldehyde groups were passivated by incubating the tips with 30% Bovine Serum Albumin (Sigma-Aldrich) for 30 min. The functionalized tip was washed with PBS before multiple indentation experiments and was prepared fresh for each experiment. The force–distance curves on the central apical surface of the cells were recorded following the steps described before. Data were analyzed offline using the Nanoscope Analysis data processing software, Matlab R2020a, and open-source software Atomic J (Hermanowicz et al., 2014).

Statistical analysis

Data were analyzed and plotted using the Prism 9 (GraphPad Software Inc., USA) or SigmaPlot software (Jandel Scientific, USA). All samples were used once. The data normality was tested by Shapiro–Wilk normality test. If data were normally distributed, a two-sided t-test was performed, otherwise Mann–Whitney U-test was used. P < 0.05 was considered as significantly different. Enriched Gene Ontology (GO) terms and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways were analyzed using the web-based DAVID 2021 tool. Significant GO terms were further analyzed using the REVIGO web platform (http://revigo.irb.hr). The published RNA-seq data were obtained from E-MTAB-3929 (Petropoulos et al., 2016). The plasma membrane proteins (PMP) and plasma transmembrane proteins (TMPs) were classified based on the UniProt database (Bateman et al., 2023). Protein binding pairs were identified in silico through the BioGRID (v4.4.220) protein interaction database (Oughtred et al., 2021).

Results

Identification of potential apical cell surface protein binding pairs on pTE-like BAP-EB and receptive EEC

To determine the potential binding partners between TE and EEC, we analyzed the cell-apical-surface proteins of BAP-EB-48 h and -72 h, receptive EEC (Ishikawa), non-receptive EEC (AN3CA), and oviductal epithelial cells (OE-E6/E7) (Lee et al., 2001). It is well established that blastocyst–endometrium communication promotes implantation (Liu et al., 2022). Therefore, we also tested if such communication changed the apical surface proteins (apicoproteome) of EEC by comparing the apicoproteome of the Ishikawa cells after treatment with CM from moderately adhesive BAP-EB-48 h (48 h-CM) and highly adhesive BAP-EB-72 h (72 h-CM).

In total, 1026 and 977 proteins were identified in all triplicates of BAP-EB-48 h and -72 h apicoproteomes, respectively. BAP-EB-72 h transcriptionally resembled pTE of human blastocyst at 7 days post-fertilization (Yue et al., 2020). Among the identified proteins, 94% of both BAP-EB-48 h (967/1026) and -72 h (918/977) apicoproteomes were reported in the pTE of preimplantation (E6 and E7) human blastocysts (Liu et al., 2022). In Ishikawa cells treated with 48 h-CM or 72 h-CM, 285 proteins were identified as apicoproteome, and the majority (90.5% or 258/285) of them were reported in Ishikawa cells that had been cocultured with human blastocysts (Liu et al., 2022), demonstrating the high reliability of our models and dataset.

GO analyses were performed on the apicoproteomes. Significant GO terms were analyzed using the REVIGO web platform (http://revigo.irb.hr). Among the 462 proteins with higher expression (fold change >1) in the apicoproteome of BAP-EB-72 h than that of BAP-EB-48 h (72 h > 48 h), ‘cell adhesion’ and ‘cadherin binding’ were among the top-ranked enriched GO terms (Fig. 1A; Supplementary Fig. S1A; Supplementary Table S1). ‘Cell adhesion’ and ‘cadherin binding’ were also among the top-ranked enriched GO terms for the top 20 highly expressed proteins in the Ishikawa cells (IshiHigh) when compared to the AN3CA and OE-E6/E7 cells (fold change > 1) (Fig. 1B;  Supplementary Fig. S1B; Supplementary Table S2). Likewise, GO terms ‘embryo implantation’, ‘cell adhesion’, and ‘cadherin binding’ were enriched in the 235 up-regulated proteins (P < 0.05 or q < 0.1) in the Ishikawa cells upon treatment with 72 h-CM when compared to those treated with 48 h-CM (Ishi-72 h-CM > 48 h-CM) (Fig. 1C;  Supplementary Fig. S1C; Supplementary Table S3). KEGG analyses revealed that the pathways of adherens junction and cell adhesion molecules were enriched in the three sets of proteins (Supplementary Tables S1, S2, and S3).

Figure 1.

Potential apical cell surface protein binding pairs on pTE-like BAP-EB and receptive EEC. (A–C) REVIGO analysis showing the biological process (BP) and molecular function (MF) of Gene Ontology (GO) terms enriched for apicoproteomes in (A) BAP-EB-72 h higher than BAP-EB-48 h (72 > 48 h), (B) Ishikawa higher than AN3CA and OE-E6/E7 (IshiHigh), and (C) Ishikawa treated with BAP-EB-72 h CM higher than BAP-EB-48 h CM (Ishi-72 h-CM > 48 h-CM); (D) Venn diagrams showing the common TMP encoded genes between TE in preimplantation embryos (Petropoulos et al., 2016) and TMP of apicoproteaomes of 72 > 48 h (left) and between Ishi-72 h-CM > 48 h-CM and IshiHigh (right). The protein-binding pairs as analyzed in silico based on BioGRID database (Oughtred et al., 2021) between the surface of BAP-EB and endometrial epithelial cells (EEC) was shown as Sankey plot (middle). BAP-EB, embryonic stem cell-derived trophoblastic spheroids; BAP-EB-48 h or -72 h, BAP-EB at 48 h or 72 h post-induction of differentiation; CM, conditioned media; TMP, transmembrane protein; TE, trophectoderm; pTE, polar trophectoderm.

The biotinylation-based surface protein pulldown assay, though sensitive, has a propensity to isolate intracellular proteins. For example, it was reported that only 29.82% of the isolated proteins were membrane proteins (Meng et al., 2023). After matching with the known PMP in the UniProt dataset (Bateman et al., 2023), we found that 38.6% of BAP-EB-48 h, 42.6% of BAP-EB-72 h, and 48.42% of Ishikawa cells treated with 48 h-CM and 72 h-CM were PMP in their top 500 enriched proteins of apicoproteomes. We further analyzed the published total lysate proteomes of human blastocyst (Zhu et al., 2025) and Ishikawa cells (Fayezi et al., 2024) and found that only 27.0% and 26.1% of the top 500 proteins were PMP, respectively (Supplementary Fig. S1D). The data suggested the successful enrichment of PMP in the apicoproteomes in this study.

We applied a filter to the apicoproteome based on the plasma membrane TMPs listed in the UniProt dataset (Bateman et al., 2023), aiming to refine the selection of surface molecules potentially involved in blastocyst–endometrium binding. Among them, 115, 9, and 63 TMPs were identified in the apicoproteomes of 72 h > 48 h, IshiHigh, and Ishi-72 h-CM > 48 h-CM protein sets, respectively (Table 1). We also identified 87 TMP from 438 TE-specific genes of human preimplantation embryos (relative to epiblast and primitive endoderm) from a published dataset (Petropoulos et al., 2016). Among them, 18 TE-specific TMPs (PLXNA1, EGFR, ENPEP, ADAM9, STX7, PRSS8, BCAM, FDFT1, SLC16A1, DSG2, MYOF, TFRC, TMPRSS2, SLC7A5, MPZL1, ITGB4, ST14, and CDH1) were shared with the 72 h > 48 h TMP list. There were six TMPs (SLC3A2, CDH1, SLC7A5, TFRC, BSG, and ATP1A1) common between the IshiHigh and the Ishi-72 h-CM > 48 h-CM protein list. After subsequent matching of BAP-EB and EECTMPs in the BioGRID protein interaction database (Oughtred et al., 2021), 26 binding pairs were identified (Fig. 1D). CDH1-CDH1 homophilic binding was one of them and was selected for further study because homophilic CDH1 binding exhibited the highest binding affinity among a set of 21 transmembrane adhesion molecules identified as binding partners of CDH1 in a previous study (Shafraz et al., 2020).

Table 1.

Lists of plasma membrane transmembrane proteins (TMP) in BAP-EB (72 > 48 h), IshiHigh and Ishi-72 h-CM > 48 h-CM.

	TMPs
72 > 48 h	ITGA5, LMAN1, ESYT1, MET, ROR2, PLXNA1, MMP15, CLDN6, NCSTN, NOTCH2, SDK2, ITGA2, LRP1, MMP14, EGFR, SLC25A4, ERBB2, ENPEP, ACSL1, SLC3A2, ATP2B4, PLXNB2, ESYT2, SLC39A10, LSR, ROR1, ITGB5, GPR107, DDOST, HMOX2, ITGB4, DCBLD2, ATP11A, PCDHB8, PLXNB1, ITGA1, SLC25A13, MPZL1, ADAM9, STX7, PCDHGB2, SIGIRR, ERBB3, SDK1, IGF1R, PCDHB16, SLC25A6, ST14, ITGA9, FLOT1, FZD6, PRSS8, NOTCH3, PCDHGB1, ATP2B1, NAALAD2, CXADR, ACSL3, CDH10, ALCAM, ERLIN2, BCAM, VTCN1, ITGB1, FDFT1, EPHB3, PCDHB13, SLC2A1, EPCAM, BSG, CACNA2D2, CDH3, TENM3, TMED10, FLRT3, STS, PTPRM, ITGA7, PVRL4, SLC16A1, EPHB4, CKAP4, PRTG, FLVCR1, NRP1, SLC44A2, ATP1B1, NRP2, DSG2, VAPB, MYOF, SDC4, FGFR2, ITGAV, PCDHB2, PTK7, ADAM10, SLC4A2, TFRC, FREM2, TMPRSS2, PVRL2, SEMA6A, PLXNA2, PLXDC2, SLC7A5, CDH1, SLC27A4, ICAM1, TMEM30A, F11R, DDR1, ATP2A2, LGR4, ANO6
IshiHigh	PODXL, SLC3A2, ATP1B3, ATP1A1, BSG, CKAP4, TFRC, SLC7A5, CDH1
Ishi-72 h-CM > 48 h-CM	ITGA5, ITGA3, CDIPT, PLXNA1, TPBG, MICA, NLGN4X, MMP14, EGFR, SLC30A1, SLC3A2, PLXNB2, LSR, PTPRJ, VAMP7, ITGB4, BCAP31, FAT2, RAET1G, NLGN2, ROBO1, MCAM, NLGN1, ST14, SLC29A2, VANGL1, CDH12, NOTCH3, NDC1, LRIG2, ATP2B1, TKT, CXADR, ATP1A1, ITGB1, LRIG1, BSG, CDH3, PTPRF, ARL6IP5, VASN, SLC7A1, WLS, PROM2, SLC44A2, EPHA2, APMAP, SLC1A5, LRIG3, ITGAV, SLC44A1, TFRC, PVRL2, PLXDC2, SLC7A5, CDH1, MICB, SLC16A3, CDCP1, ITGB3, NLGN4Y, F11R, DDR1

The apical surface CDH1 on pTE cells is adhesive

To be involved in blastocyst–endometrium adhesion, the binding molecules have to be expressed on the apical surface of the TE cells and EEC. The presence of apical cell surface CDH1 was confirmed by non-permeabilized immunocytochemical staining on live cells (live-cell ICC). Confocal imaging confirmed the apical surface localization of CDH1 on non-permeabilized BAP-EB-48 h and BAP-EB-72 h, but not on BAP-EB-24 h (Fig. 2A). An increase of CDH1 from BAP-EB-48 h to BAP-EB-72 h was indicated in the mass spectrometry results of the BAP-EB apicoproteome (Supplementary Fig. S2A). Bioinformatic analysis of a published dataset (Petropoulos et al., 2016) showed that the expression of CDH1 transcripts was significantly higher in TE than epiblast and higher in pTE than mural TE of human blastocysts on Day 7 post-fertilization (Supplementary Fig. S2B). The pTE is responsible for adhesion of human blastocysts to EEC. To determine the surface expression of CDH1 in human blastocysts, we performed live-cell ICC on four good-quality human blastocysts donated for research (Fig. 2B; Supplementary Video V1). As shown in the representative blastocyst images (Fig. 2B), more intense signals of apical surface CDH1 were observed in pTE when compared to mTE by live-cell ICC. Further fixed-cell ICC confirmed the pTE localization, as the enhanced CDH1 signal was mainly localized to the NANOG-positive inner cell mass (single and z-stacked confocal images of Blastocyst 2). We also performed fixed-cell ICC of CDH1 and found that CDH1 was evenly distributed in the cell–cell junctions within the whole blastocyst (Blastocyst 3; Fig. 2B). Furthermore, we also confirmed the expression of CTNND1, one of the CDH1 intracellular binding proteins, by fixed-cell ICC. However, live-cell ICC demonstrated the non-apical surface localization of CTNND1 (Blastocyst 3; Fig. 2B).

Figure 2.

Expression and adhesiveness of apical cell surface CDH1 on pTE of human blastocyst. (A) Confocal images of live-cell ICC of CDH1 (green) in BAP-EB-24 h (24 h), BAP-EB-48 h (48 h), and BAP-EB-72 h (72 h). Scale bar: 50 μm; (B) Confocal image (Blastocysts 1–3) and maximum intensity projection of z-stacked confocal images (top panel of Blastocyst 2) of live-cell ICC (live) followed by permeabilized-cell ICC (fixed) in human blastocysts. White arrows denotes regions with inner cell mass. Scale bar: 50 μm. Cell nuclei were stained with DAPI (blue); (C) Schematic diagram showing the setup of high-resolution atomic force microscope-based single molecules force spectroscopy (AFM-SMSF); (D and E) The adhesion (rupture) forces distribution with Gaussian fit of AFM tips functionalized with recombinant CDH1 onto the cell surface of (D) BAP/2D-24 h (24 h) and BAP/2D-72 h (72 h) (BAP/2D-24 h: n = 1217, BAP/2D-72 h: n = 1491, biological repeat n = 3, P<0.0001, Mann–Whitney U-test), and (E) BAP/2D-72 h transfected with 500 nM control (Si-C) or CDH1 (Si-CDH1) siRNA (Si-C: n = 1353, Si-CDH1: n = 1352, biological repeat n = 3, P = 0.0121, Mann–Whitney U-test). Data were presented as bar plots representing the mean ± SEM and showing individual data points; *, P < 0.05. ICC, immunocytochemistry; pTE, polar trophectoderm; BAP-EB, embryonic stem cell-derived trophoblastic spheroids; BAP-EB-24 h, -28h or -72-h, BAP-EB at 24 h, 48 h or 72 h post-induction of differentiation; BF, bright field; AFM, atomic force microscopy; BAP/2D, embryonic stem cell-derived trophoblastic cell monolayers; BAP/2D-24 h or -72h, BAP/2D at 24 h or -72h post-induction of differentiation; DAPI, 4′,6-diamidino-2-phenylindole.

To determine whether the apical surface CDH1 conferred adhesiveness to TE cells, we treated monolayers of a human embryonic stem cell line, VAL3, with BAP (BAP/2D). The XZ views of live-cell ICC confirmed the apical localization of CDH1 on the BAP/2D (Supplementary Fig. S2C). There was minimal surface immunoreactive signal of CDH1 on BAP/2D at 24 h (BAP/2D-24 h) and 48 h (BAP/2D-48 h) post-BAP treatment. Prominent and intense apical surface signal was only detected at 72 h post-BAP treatment (BAP/2D-72 h), when BAP-EB-72 h was pTE-like and had acquired strong adhesiveness.

Next, a high-resolution atomic force microscope (AFM) was used to characterize the adhesiveness of the pTE-like BAP/2D cells. Compared with BAP/2D-24 h, the cells of BAP/2D-72 h were shorter in cell height (Supplementary Fig. S2D) and with increased roughness of the cell surface (Supplementary Fig. S2E), suggesting the potential increase in cell surface protein expressions. The AFM-based single-molecule force spectroscopy (AFM-SMFS) method was used to confirm a direct binding between proteins on the surface of BAP/2D and exogenous CDH1 (Fig. 2C). In this method, recombinant CDH1 protein was cross-linked to the AFM tip, and the success of functionalization was validated (Supplementary Fig. S2F). The presence of CDH1-binding protein on the cell surface of BAP/2D-24 h and BAP/2D-72 h was determined by the force–distance curves. It was found that the rupture forces between the CDH1-functionalized tip and BAP/2D-72 h (36.0 ± 0.8 pN) were significantly higher (P < 0.0001) than those between the tip and BAP/2D-24 h (30.8 ± 0.8 pN) (Fig. 2D), showing a higher expression of CDH1-binding proteins on the surface of BAP/2D-72 h. On the other hand, knockdown of CDH1 reduced the cell surface roughness (Supplementary Fig. S2G) and significantly decreased (P = 0.0121) the rupture forces between the CDH1-functionalized tip and BAP/2D-72 h (31.37 ± 0.5 pN) when compared to the siRNA control (33.40 ± 0.5 pN) (Fig. 2E), confirming the adhesive properties of apical surface CDH1 on BAP-EB-72 h.

The apical surface CDH1 on endometrial epithelial cells is adhesive and contributes to CDH1 homophilic binding between BAP-EB and endometrium

Apicoproteome analyses showed that treatment with 72 h-CM upregulated the expression of CDH1 in the receptive Ishikawa cells when compared with those treated with 48 h-CM (Supplementary Fig. S3A). Western blotting analysis confirmed significant induction (P = 0.0264) of membranous CDH1 expression in the Ishikawa cells upon 72 h-CM treatment (Fig. 3A), and that the Ishikawa cells exhibited significantly higher expression of CDH1 transcripts (P = 0.0015) (Supplementary Fig. S3B) and CDH1 in the membranous fraction (P = 0.0033) (Fig. 3B) than the non-receptive endometrial HEC-1B cells. The XZ view of fixed-cell ICC samples localized the CDH1 signal to the apical region of the Ishikawa cells (Fig. 3C) and primary EEC from the mid-secretory phase (LH + 7) of the menstrual cycle (Fig. 3D). Additionally, live-cell ICC confirmed the apical cell surface location of CDH1 in the Ishikawa cells and primary EEC (Fig. 3C and D). CDH1 immunoreactivities were absent in the HEC-1B cells (Supplementary Fig. S3C).

Figure 3.

Spatial expression of CDH1 in EEC. (A) Representative image of western blotting and the relative surface protein expression of CDH1 in Ishikawa cells after 3-h of treatment with 48 h-CM or 72 h-CM (n=3, P = 0.026, t-test); (B) western blotting image and the relative membranous protein expression of CDH1 in Ishikawa and HEC-1B cells (n = 4, P = 0.003, t-test); (C and D) Z-stack images from apical to basal region and XZ view of fixed-cell ICC and live-cell ICC of CDH1 (green) in (C) Ishikawa cells and (D) cultured primary EEC, scale bar: 20 μm; (E) representative images of immunohistochemistry of CDH1 (red) in endometrial biopsies obtained on LH + 2 or LH + 7 day, cell nuclei were stained by DAPI (blue), magnification: 200×; (F) western blotting analysis showing the relative protein expressions of CDH1 in EEC isolated from patients at their LH + 2 (n = 9) and LH + 7 (n = 28) days, scatter dot plot with median was shown (P = 0.0020, Mann–Whitney U-test); (G) the relative surface protein expression of CDH1 in Ishikawa cells after 3-h treatment with 500 pM β-estradiol and 50 nM progesterone (E2 + P4) (n = 5, P=0.0079, Mann–Whitney U-test). Bar plots represent the mean ± SEM. The P-value is displayed on each plot. EEC, endometrial epithelial cells; 48 h- or 72 h-CM, conditioned media of embryonic stem cell-derived trophoblastic spheroids at 48 h or 72 h post-induction of differentiation; ICC, immunocytochemistry; LH + 2 or +7, Day 2 or 7 post LH surge; DAPI, 4′,6-diamidino-2-phenylindole.

Histologically, CDH1 was expressed in the luminal and glandular epithelium but not the stroma of human endometrial tissues (Fig. 3E). We then compared CDH1 expression in primary EEC from women in their early secretory phase (LH + 2) and mid-secretory phase (LH + 7). Significantly higher levels of CDH1 proteins were found in the receptive (LH + 7) than the pre-receptive (LH + 2) phase (P = 0.0020) (Fig. 3F). A steroid environment mimicking the mid-secretory phase (E2+P4) also induced the expression of membranous CDH1 protein in the Ishikawa cells (P = 0.0079) (Fig. 3G). These results were consistent with a potential role of endometrial apical CDH1 in blastocyst adhesion.

AFM with non-functionalized tips was then used to characterize the general adhesiveness of Ishikawa cells. Individual Ishikawa cells were around 20–40 μm in width and 2–3 μm in height, with the height apex localized to the nucleus area and decreased toward the lamellipodia (Supplementary Fig. S3D). The stiffness of the cells increased from the central part to the edge of the cells from 1 to 20 kPa (Supplementary Fig. S3E), but the adhesive force of the cell membrane surface, which ranged from nearly 0 to a maximum of 100 pN, showed an opposite trend with stronger adhesive force on the central part of the cell than the cell edge (Supplementary Fig. S3F). To study the roles of CDH1 on adhesiveness, we knocked down CDH1 in the Ishikawa cells by siRNA transfection as demonstrated by western blotting (Supplementary Fig. S3G) and fixed-cell ICC (Supplementary Fig. S3H). The AFM 3-dimensional measurement showed that the knockdown of CDH1 reduced cell surface roughness (Fig. 4A; Supplementary Fig. S3I) and significantly reduced the Young’s modulus [Si-control (12.0 ± 0.1 kPa) vs si-CDH1 (10.0 ± 0.1 kPa), P < 0.0001] and adhesive forces [Si-control (26.0 ± 0.4 pN) vs si-CDH1 (17.1 ± 0.3 pN), P < 0.0001] (Fig. 4B) of the cells. The results indicated that the apical CDH1 contributed to the adhesiveness of the EEC.

Figure 4.

Contribution of apical surface CDH1 to homophilic binding between BAP-EB and EEC. (A) AFM topology showing cellular morphology 3 days after control (Si-C) or CDH1 (Si-CDH1) siRNA transfection. Bottom scale bar: 10 μm; right bar: height scale. Arrows point to some cell surface area. Biological repeats, n = 3; (B) the Young’s modulus (upper panel) and adhesive (rupture) forces (lower panel) distribution with Gaussian fit in Si-C and Si-CDH1 transfected Ishikawa cells (biological repeats, n = 3, P < 0.0001, Mann–Whitney U-test); (C) the BAP-EB-72 h attachment rate (%) after 1-h coculture with si-C or Si-CDH1 transfected Ishikawa cells (n = 6, P = 0.0146, t-test); (D and E) the BAP-EB-72 h spreading area (μm²) after 24-h coculture with si-C or Si-CDH1 knockdown Ishikawa cells where (D) unattached BAP-EB-72 h were not removed after 1-h coculture (Si-C: n = 60, Si-CDH1: n = 59, P = 0.0003, Mann–Whitney U-test); (E) unattached BAP-EB-72 h were removed after 1-h coculture; (F and G) the BAP-EB attachment rate (%) after 1-h coculture with Ishikawa cells with either or both BAP-EB and Ishikawa cells pretreated with (F) 5 μg/ml mouse IgG (Mo-IgG) or anti-CDH1 antibody and (G) 10 μg/ml CDH1 blocking peptide; (H and I) Co-immunoprecipitation results of recombinant human CDH1 protein (CDH1-His) and CDH1 protein in surface protein (SP) isolated from (H) BAP-EB-72h and (I) Ishikawa cells. Bar plots represent the mean ± SEM. The P-values were displayed on each plot. BAP-EB, embryonic stem cell-derived trophoblastic spheroids; BAP-EB-72 h, BAP-EB at 72 h post-induction of differentiation.

The functional roles of CDH1 on BAP-EB-endometrium adhesion were then tested by coculture of BAP-EB-72 h with EEC. CDH1 knockdown in the Ishikawa cells significantly reduced the attachment rates of BAP-EB-72 h (P = 0.0146) after 1 h of coculture (Fig. 4C). In addition, the knockdown decreased the spreading area of BAP-EB-72 h on Ishikawa cells (P = 0.0003) (Fig. 4D; Supplementary Fig. S3J) after 24-h coculture, which measured the migration of BAP-EB. On the contrary, when the unattached BAP-EB-72 h were removed after the first hour of coculture, the spreading area of the attached BAP-EB were comparable between control and CDH1-silenced Ishikawa cells (Fig. 4E). This result indicated that BAP-EB attachment might be delayed by CDH1-deficient EECs, causing the reduced migration area of the trophoblast cells.

Homophilic CDH1 binding between EEC and BAP-EB-72 h was first tested by antibody blocking and peptide blocking experiments. Pre-treatment of either BAP-EB-72 h or Ishikawa cells, or both cells with specific antibody or blocking peptide (Devemy and Blaschuk, 2009) against the ectodomains of CDH1 significantly reduced the attachment rates (Fig. 4F and G). The three treatments reduced attachment to a comparable extent, consistent with participation of homophilic binding between the CDH1 on BAP-EB and those on the Ishikawa cells. Co-IP assay confirmed that C-terminal His-tagged recombinant human CDH1 protein (∼90 kDa) could bind to the CDH1 (∼120 kDa) isolated from the cell surface of BAP-EB-72 h (Fig. 4H) and Ishikawa cells (Fig. 4I), supporting homophilic binding of CDH1 on the apical surfaces of BAP-EB-72 h and Ishikawa cells. The results indicated that the homophilic binding of CDH1 was a major molecular interaction responsible for the BAP-EB-EEC adhesion, as blocking of the binding reduced the attachment by ∼50%.

BAP-EB conditioned media induce expression of CTNND1 in EECs

Apart from CDH1, apicoproteome analyses showed that the 72 h-CM treatment upregulated the expression of the molecules associated with CDH1 in the adherens junction of Ishikawa cells, namely CTNND1, CTNNB1, and CTNNA1, when compared with Ishikawa cells treated with 48 h-CM (Supplementary Table S3). CTNND1 was selected for further study because it is a molecular stabilizer of CDH1 in epithelial cells (Venhuizen et al., 2020), but its role in endometrial receptivity is not known. Western blotting analysis confirmed significant upregulation of membranous CTNND1 proteins in the Ishikawa cells upon treatment with 72 h-CM (Supplementary Fig. S4A). The XZ view of fixed-cell ICC localized the CTNND1 signal to the apical region but beneath that of the CDH1 signal in both the Ishikawa cells (Fig. 5A) and primary EEC collected on Day LH + 7 (Fig. 5B). CTNND1 immunoreactivities were absent in the non-receptive endometrial HEC-1B cells (Supplementary Fig. S4B). In contrast to CDH1, the CTNND1 immunoreactivity was not detected on the non-permeabilized live-cell ICC (Fig. 5A and B), indicating association of CTNND1 with the cytoplasmic face of the plasma membrane. In histological sections of human endometrial tissues, CTNND1 immunoreactivities were localized to the luminal and glandular epithelium but not the stroma (Fig. 5C). Similar to CDH1, the expression was significantly higher at LH + 7 than LH + 2 (P = 0.017) (Fig. 5D), and E2+P4 induced the expression of membranous CTNND1 proteins in the Ishikawa cells (P = 0.0079) (Fig. 5E).

Figure 5.

Spatial expression and mediation in adhesion of CTNND1 in EEC. (A and B) Z-stack images from apical to basal region and XZ view of fixed-cell and live-cell ICC of CTNND1 (red) and CDH1 (green) in (A) Ishikawa cells and (B) primary EEC, scale bar: 20 μm; (C) representative images of immunohistochemistry of CTNND1 (green) in endometrial biopsies obtained on LH + 2 or LH + 7 day, cell nuclei were stained with DAPI (blue), magnification: 200×; (D) western blotting analysis showing the relative protein expressions of CTNND1 in EEC isolated from patients at their LH + 2 (n = 9) and LH + 7 (n = 28) days, scatter dot plot with median is shown (P=0.017, Mann–Whitney U-test); (E) western blotting analysis showing the relative surface protein expression of CTNND1 in Ishikawa cells after 3-h treatment with 500 pM β-estradiol and 50 nM progesterone (E2 + P4) (n = 5, P = 0.0079, Mann–Whitney U-test); (F and G) Relative protein expressions of CDH1, CTNND1, CTNNB1, CTNNA1, Integrin αV, ANXA2, and β-actin 3 days after control siRNA (Si-C) and (F) CDH1 (Si-CDH1) or (G) CTNND1 (Si-D1) siRNA transfection (n = 3). Protein expression levels were normalized to the cells treated by same amount of lipofectamine 2000 (lipo). Ishikawa cells treated with transfection basal medium (Opti-MEM, control) was included as control; (H) AFM topology showing the cellular morphology 3 days after Si-C or Si-D1 transfection. Scale bar: 10 μm; right bar: height scale. Arrows point to some cell surface area. Biological repeat n = 3; the (I) Young’s modulus and (J) adhesive (rupture) forces distribution with Gaussian fit in Si-C and Si-D1 transfected Ishikawa cells (biological repeat n = 3, P < 0.0001, Mann–Whitney U-test); (K) the BAP-EB-72 h attachment rate (%) after 1-h coculture with si-C or Si-D1 transfected Ishikawa cells (n = 7, P = 0.0040, t-test); and (L) the BAP-EB-72 h spreading area (μm²) after 24-h coculture with si-C or Si-D1 (Si-C: n = 71, Si-D1: n = 49, P < 0.0001, Mann–Whitney U-test) transfected Ishikawa cells. Bar plots represent the mean ± SEM. The P-value was displayed on each plot. EEC, endometrial epithelial cells; ICC, immunocytochemistry; LH + 2 or +7, Day 2 or 7 post LH surge; BAP-EB-72 h, embryonic stem cell-derived trophoblastic spheroids at 72 h post-induction of differentiation; DAPI, 4′,6-diamidino-2-phenylindole.

Knockdown experiments were performed to determine the relationship between CTNND1 and CDH1 in the Ishikawa cells. While CDH1 knockdown had no effect on CTNND1 expression (Fig. 5F), CTNND1 knockdown led to abolishment of CDH1 expression (P = 0.0059) (Fig. 5G; Supplementary Fig. S4C). We further analyzed the effects on other adhesion-related proteins. Knockdown of CDH1 significantly reduced the expression of CTNNB1 (P = 0.0262), but not CTNND1, CTNNA1, integrin αV, and annexin-A2 in the Ishikawa cells (Fig. 5F). However, CTNND1 knockdown significantly reduced the expression of CTNNA1 (P = 0.0379), CTNNB1 (P = 0.0068), and integrin αV (P = 0.0464), but not annexin-A2 (Fig. 5G), and decreased the roughness of the cell surface (Fig. 5H; Supplementary Fig. S4D).

Knockdown of CTNND1 significantly reduced the Young’s modulus [si-control (14.2 ± 0.1 kPa) vs si-CTNND1 (9.6 ± 0.1 kPa), P < 0.0001] (Fig. 5I) and adhesive forces [si-control (22.4 ± 0.2 pN) vs si-CTNND1 (10.0 ± 0.1 pN), P < 0.0001] (Fig. 5J). The attachment rates (P = 0.0040) (Fig. 5K) and spreading areas (P < 0.0001) (Fig. 5L; Supplementary Fig. S4E) of BAP-EB-72 h onto the CTNND1 knockdown Ishikawa cells were significantly reduced. These results suggested that CTNND1 worked with CDH1, contributing to the adhesiveness of the EEC.

Conditioned media of human blastocysts and HCG induce apical redistribution of endometrial CDH1 and CTNND1

A short period of BAP-EB CM treatment (3 h) induced surface expression of CDH1 on the Ishikawa cells (Fig. 3A). To further study the cellular localization of CDH1 and CTNND1 in response to embryonic signals, Ishikawa cells were treated with BAP-EB-CM (48 or 72 h) for different durations (Fig. 6A) or SCM of individual human blastocysts (n = 18) for 3 h (Fig. 6B), followed by fixed-cell ICC. Enhanced CDH1 and CTNND1 expressions in BAP-EB-CM-treated Ishikawa cells, but not HEC-1B cells, were observed (Supplementary Fig. S5A).

Figure 6.

Effects of embryonic signals on the spatial expression of CDH1 and CTNND1 in EEC. (A and B) Representative figures showing XZ views of CDH1 (green) and CTNND1 (red) in randomly selected regions in Ishikawa cells after treatment for (A) 0 min (0’), 30 min (30’), 60 min (60’), and 180 min (180’) with 48 h-CM or 72 h-CM or (B) 0 min (control) and 3 h with human blastocyst spent culture medium. Cell nuclei were stained with DAPI (blue). (C and D) Representative 3D surface plots (upper panel: scale on the right represents the measure of thermal gradients) and bar plots (lower panel) showing percentages of apical CDH1 and CTNND1 expressions in single Ishikawa cells after treatment for (C) 0 min (0’), 30 min (30’), 60 min (60’), and 180 min (180’) with 48 h-CM or 72-CM or (D) 3-h human blastocyst SCM. Bar plots represented the mean ± SD upon treatment of individual human blastocyst SCM; (E) western blotting analysis showing the relative surface protein expression levels of CTNND1 and CDH1 in Ishikawa cells after 3-h treatment with HCG at concentrations of 5 IU/l, 50 IU/l, or 500 IU/l and compared to the cells without treatment (control) (n = 5); (F) representative figures showing XZ views of CDH1 (green) and CTNND1 (red) in randomly selected regions in Ishikawa cells after treatment for 0 min (0’), 30 min (30’), 60 min (60’), and 180 min (180’) with 50 IU/l; (G) representative 3D surface plots (upper panel) and bar plots (lower panel) showing percent of apical CDH1 and CTNND1 expressions in single Ishikawa cells after treatment for 0 min (0’), 30 min (30’), 60 min (60’), and 180 min (180’) with 50 IU/l HCG; (H) representative figures showing XZ views (top panel, scale bar: 20μm) of CDH1 (green) and CTNND1 (red) in randomly selected regions, representative 3D surface plots (middle panel), and bar plots (bottom panel) showing percent of apical CDH1 and CTNND1 expressions in single Ishikawa cells 3 days after control (Si-C) or Rab11A (si-RAB11A) siRNA transfection and treated by HCG for 30 min; and (I) the BAP-EB-72 h attachment rate (%) after 1-h coculture with si-C or Si-RAB11A transfected Ishikawa cells (n = 5, P = 0.0435, t-test). Scale bar: 20 μm. Bar plots (C, E, G, H, and I) represented the mean ± SEM. The P-values were displayed on each plot. EEC, endometrial epithelial cells; BAP-EB-72 h, embryonic stem cell-derived trophoblastic spheroids at 72 h post-induction of differentiation; 48 h- or 72 h-CM, conditioned media of embryonic stem cell-derived trophoblastic spheroids at 48 h or 72 h post-induction of differentiation; DAPI, 4′,6-diamidino-2-phenylindole.

To determine the cellular location of the two proteins upon 72 h-CM or SCM treatments, we analyzed the distribution of CDH1 and CTNND1 proteins in single cells (Fig. 6A and B). In this analysis, the proportion of signal in the top 10% of the height over the total intensity of a cell was defined as the percentage of apical protein expression. Before the 72 h-CM or SCM treatments (time 0), the expressions of CDH1 and CTNND1 were mainly localized to the lateral region and minimally to the apical and basal regions of the cells. Upon treatments, the percentages of apical expressions of the two molecules were significantly increased (Fig. 6C and D). Specifically, the 72 h-CM induced a significantly higher percentage of apical signal than the 48 h-CM at each time point studied (30 min: CDH1, P = 0.0004; CTNND1, P = 0.0004; 60 min: CDH1, P = 0.0375; CTNND1, P = 0.0427; 180 min: CDH1, P = 0.0019; CTNND1, P = 0.0009) (Fig. 6C). However, human blastocyst SCM treatment only increased the percentages of apical CDH1 (P < 0.0001), but not CTNND1 expression in the treated cells (Fig. 6D).

Human blastocysts secrete HCG. Consistently, BAP-EB-72 h, but not BAP-EB-48 h, secreted HCG into the culture medium (Supplementary Fig. S5B). The 72 h-CM contained HCG (7.6 ± 3.0 IU/l) at a concentration comparable to that produced by one human blastocyst on Day 5 post-fertilization in 25 µl of culture medium (Xiao-Yan et al., 2013). We studied the action of HCG on the expression of CDH1 and CTNND1 in EEC. Western blotting analyses showed that Ishikawa cells treated with HCG at the serum level of HCG in the peri-implantation period (50 IU/l) increased the apical membranous expressions of both CDH1 and CTNND1 (Fig. 6E). The HCG treatment also increased the percentage of apical (Fig. 6F and G; Supplementary Fig. S5C) CDH1 (P < 0.0001) and CTNND1 (P < 0.01) expressions in the Ishikawa cells, supporting that HCG was an embryonic factor inducing apical CDH1 and CTNND1 expression for embryo adhesion.

Rab11A is a GTPase that regulates intracellular trafficking and is a CDH1 transporter to the cell membrane (Kakar-Bhanot et al., 2019). Its role in CDH1 trafficking to the apical cell surface induced by embryonic signals was studied. Here, we knocked down Rab11A in Ishikawa cells by siRNA (Supplementary Fig. S5C) and treated the cells with HCG for 30 min. The apical expression of HCG-induced CDH1, but not CTNND1, was significantly reduced in the si-RAB11A-treated cells (P = 0.0092) (Fig. 6H). Consistently, the BAP-EB attachment rate was reduced (P = 0.0435) on the RAB11A knockdown Ishikawa cells (Fig. 6I).

Expressions of CDH1 and CTNND1 correlate with successful IVF outcomes

The involvement of CDH1 and CTNND1 in attachment processes was also suggested by their differential expressions in primary EEC of women with or without a LB after IVF treatment. Western blotting of primary EEC isolated from endometrial tissues indicated that significantly higher levels of CDH1 (P = 0.0378) and CTNND1 (P = 0.032) were detected in the LH + 7 EEC from women with a LB (n = 29) than those without a pregnancy (NP, n = 30) following IVF. Significantly lower levels of CDH1 (P = 0.005), but not CTNND1, were detected in the LH + 7 EECs from women with RIF (n = 10) when compared to those from the LB group (Fig. 7A). We further specifically analyzed the expression of CDH1 and CTNND1 in the apical region of luminal epithelial cells by immunofluorescence staining of the paraffin sections (Fig. 7B). It was revealed that the percentage of apical CDH1 in the luminal epithelium was significantly higher (P = 0.0278) in the LB group (n = 12) when compared to that of the RIF group (n = 7) (Fig. 7C).

Figure 7.

Expression of CDH1 and CTNND1 in primary EEC. (A) The relative protein levels of CDH1 and CTNND1 in receptive EECs (LH + 7 day) from women who gave live birth (LB, n = 29), were not pregnant (NP, n = 30) and with recurrent implantation failure (RIF, n = 10), Mann–Whitney U-test; (B) representative images of immunohistochemistry of CTNND1 (red) and CDH1 (green) in endometrial tissues obtained on LH+7 day from women with LB (upper panel) and women with RIF (bottom panel). Cell nuclei were stained with DAPI (blue), scale bar: 20 μm; and (C) the normalized percentages of apical CDH1 and CTNND1 in receptive EECs from women with LB (n = 12) and RIF (n = 7), t test. Data were presented as scatter dot plots showing the median. P-value was indicated on the graphs. EEC, endometrial epithelial cells; LH + 7 day, Day 7 post LH surge; DAPI, 4′,6-diamidino-2-phenylindole.

Discussion

This is the first report demonstrating the contribution of hetero-cellular homophilic binding between trophectodermal CDH1 and endometrial CDH1 in the adhesion process during embryo implantation (schematic Fig. 8). Specifically, embryonic signals, including HCG, induced expression of apical cell surface CDH1 on the receptive EEC, suggesting a role in the enhancement of embryo adhesion. The conclusions are based on the following: (i) CDH1 was localized to the free apical surface of pTE of human blastocysts, pTE-like BAP-EB-72 h, and receptive EEC; (ii) AFM and CDH1 knockdown experiments confirmed the adhesive nature of the apical-surface CDH1 on BAP-EB and EEC; and (iii) functional, antibody/peptide blocking, and co-IP experiments showed that the homophilic binding of CDH1 was a major factor of blastocyst–endometrium adhesion, contributing to ∼50% of the adhesion between BAP-EB and EEC. The study further showed that the expression of CDH1 and its stabilizer CTNND1 in EEC and apical CDH1 expression in luminal epithelium are clinically meaningful, as their expression correlated with successful IVF outcomes.

Figure 8.

Hetero-cellular homophilic binding between embryonic CDH1 and endometrial CDH1 contributing to human embryo adhesion. Schematic diagram showing the differential effects of blastocyst-secreted HCG on the induced expression and relocation of CDH1 and CTNND1 in EEC between normal receptive and RIF patients. EEC, endometrial epithelial cells; RIF, repeated implantation failure.

Embryo–endometrium crosstalk is critical for a successful implantation (Tabibzadeh and Babaknia, 1995; Simon et al., 2001). Consistently, CM from BAP-EB-72 h and human blastocysts induced redistribution of CDH1 and CTNND1 in receptive but not non-receptive EEC after 3 h of treatment. The rapid induction is physiologically reasonable for an incoming blastocyst to strengthen its adhesion by elevating the apical expression of CDH1 on EEC prior to its physical contact with the endometrium. As human blastocysts secrete HCG, and HCG maintains maternal tolerance of the implanting embryo ( D’Hauterive et al., 2022), we tested if HCG mediated the induction at a concentration that correlated with the physiological levels secreted by good-quality blastocysts (Chen et al., 2020). Consistently, the highly adhesive BAP-EB-72 h produced more HCG than the moderately adhesive BAP-EB-48 h (Lee et al., 2015; Yue et al., 2020). Due to the limited volume of SCM (∼8 µl) available from each blastocyst in our IVF program, the concentration of HCG in the SCM could not be determined in the study. The level of HCG in the 72 h-CM was similar to the reported range in the SCM of human blastocysts, which ranged from 0.3 to 2.3 IU/l (Xiao-Yan et al., 2013). The level of HCG acting on EEC at the implantation site during blastocyst apposition in vivo is unknown. Our data showed that HCG at level of 50 IU/l (corresponding to the serum level in early pregnancy) induced apical expression of CDH1 and CTNND1. A meta-analysis on 15 randomized controlled trials concluded that intrauterine infusion of HCG before embryo transfer increased LB rate significantly (Gao et al., 2019). It is reported that the HCG receptor, LHCGR, is expressed in Ishikawa cells (Viswanath et al., 2007), and thus we postulated that the action of HCG in EEC was mediated through the LHCGR. Yet, HCG can function through alternative mechanisms beyond its canonical receptor-binding pathway. As HCG is highly glycosylated, it can initiate ‘lectin-like’ effects by interacting with lectin-binding sites on other molecules. In addition, available evidence also suggested the internalization and trafficking of the free β-subunit of HCG to the nucleus via an LHCGR-independent pathway (Butler et al., 1999).

The luminal epithelium is receptive to blastocysts only during the window of implantation. Under the action of progesterone in the receptive phase, the luminal epithelium exhibits plasma membrane transformation (Whitby et al., 2020), loss of cell polarity with downregulation of polarity markers (Whitby et al., 2018), and reduction in glycocalyx thickness and cell surface charges (Thie and Denker, 2002) for reducing repulsion of pTE during implantation. The tight junctions become ‘tighter’, but the adherens junctions are lost with reduced CDH1 expression along the lateral membrane of the luminal epithelium in mice (Murphy, 2000) and in infertile women (Matsuzaki et al., 2010; Kakar-Bhanot et al., 2019). At the same time, the EEC gains adhesion competence for embryo implantation (Thie and Denker, 2002) via unclear mechanisms possibly involving Scribble (Whitby et al., 2018) and Rab11A (Kakar-Bhanot et al., 2019). Here, we showed that in human endometrial tissues at the receptive stage, expression of CDH1 on the apical membrane of luminal epithelium positively correlated with pregnancy outcomes, which suggested CDH1 contributes to the gain of adhesion competence of EEC. The importance of CDH1 in attachment is supported by the observation that forced expression of CDH1 in the non-receptive AN3CA EEC significantly increases the attachment of choriocarcinoma trophoblastic spheroids onto the EEC (Rahnama et al., 2009). The role of embryonic CDH1 in implantation has been implicated in the literature. The expression of CDH1 shifts from a basolateral location at the morula stage to the apical membrane of TE at the blastocyst stage (Alikani, 2005). In light of the high expression of CDH1 mRNA and protein in human blastocysts on Days 6–8 after fertilization (Petropoulos et al., 2016), we tested the possibility that CDH1 on BAP-EB-72 h was a binding partner of endometrial CDH1. Our combined results of anti-CDH1 antibody and peptide blocking further support the critical roles of apical surface CDH1 on embryonic and endometrial cells for embryo attachment. Furthermore, knockdown of Rab11A in EEC significantly reduced the apical CDH1 expression and BAP-EB attachment rate when treated by HCG, suggesting Rab11A mediated the redistribution of CDH1 to the apical cell surface of EEC for embryo attachment. The results find support from a recent study that the expression of endometrial Rab11A and CDH1 at the receptive phase is significantly lower in women with unexplained infertility than the fertile women, and knockdown of Rab11A in EEC reduces the number of surface αVβ3 expressing cells, total and lateral membrane expression of CDH1, and attachment of choriocarcinoma spheroids (Kakar-Bhanot et al., 2019).

Two observations showed that CTNND1 was an upstream regulator of CDH1 in EEC. First, CTNND1 knockdown reduced CDH1 expression in the Ishikawa cells. Second, CDH1 knockdown had no effect on CTNND1 expression. In kidney cells, CTNND1 interacts with the juxtamembrane cytoplasmic domain of CDH1 to prevent internalization and degradation of CDH1 by ubiquitination (Hartsock and Nelson, 2012) and to regulate trans binding of CDH1 by dimerization at the plasma membrane (Vu et al., 2021). The knockdown of CTNND1 also downregulated the expression of other adhesion molecules, including integrin-αV, CTNNA1, and CTNNB1, whereas CDH1 knockdown only reduced the CTNNB1 expression. AFM demonstrated a prominent reduction of surface protein regions of the EEC upon CTNND1 knockdown, consistent with the reduced surface expression of CDH1 and integrin αV. In this study, we also demonstrated the presence of CTNND1 in the human blastocyst, suggesting similar regulation of CTNND1 on embryonic CDH1 during blastocyst adhesion. CTNND1 was one of the 241 genes that were significantly up-regulated in Day 7 human blastocyst TE cocultured with human primary endometrial cells compared to the un-cocultured TE (Lv et al., 2019), indicating endometrial signals could upregulate embryonic CTNND1.

The role of CTNND1 in endometrial receptivity has not been reported before, though it has been related to recurrent pregnancy loss (Craciunas et al., 2021). Here, we demonstrated that CTNND1 knockdown in EEC reduced the rupture force between the AFM probe and the cell membrane, which positively reflected cell adhesiveness (Fang et al., 2000). Consequently, the attachment of BAP-EB-72 h on the CTNND1 knockdown cells was reduced, likely resulting from loss of apical CDH1 required for BAP-EB adhesion. This possibility is supported by the reduction of cell adhesiveness and BAP-EB attachment on the CDH1 knockdown EEC. The fact that more prominent reductions in adhesion molecule expression, cell adhesiveness, and stiffness after CTNND1-knockdown, rather than CDH1-knockdown, suggested that CTNND1 is a master regulator of cell surface protein expression and cytoskeleton structure. Indeed, conditional knockout of CTNND1 in mouse epidermal cells increases expression of actin stress fibers and reduces integrin turnover (Epifano et al., 2014).

In the endometrium, CDH1 is well known for its role in maintaining the integrity of the epithelium and in epithelial-to-mesenchymal transition at implantation (Oghbaei et al., 2022). The role of CDH1 in TE-EEC adhesion of humans is unclear. We found knockdown of CDH1 in Ishikawa led to decreased attachment rate of BAP-EB and downregulation of CTNNB1. The observations are similar to conditional knockout of CDH1 in the developing uterus of mice, which results in the loss of epithelial cell–cell interactions, including loss of adherens junction molecules CTNNB1 and CTNNA1 and tight junction molecules in the neonatal uterus (Reardon et al., 2012). CDH1 has been shown to be relocated from the cytoplasm to the cell membrane in the Ishikawa cells upon exposure to estrogen and progesterone without change of mRNA level (Tiwari et al., 2021), but it is not known whether the CDH1 is translocated to the apical and/or lateral membrane of the cells. With the use of live-cell ICC, which recognized only cell-surface molecules (Zhuang and Matsunami, 2008), we confirmed the surface expression of CDH1 in the Ishikawa cells and primary EECs. We further provided the first demonstration of an increase in the apical surface expression of CDH1 on EEC upon exposure to CM from BAP-EB-72 h and human blastocysts and HCG. The available reports on human endometrial CDH1 expression are inconsistent; some reported a stable CDH1 expression throughout the menstrual cycle (van der Linden et al., 1995; Poncelet et al., 2002), while others found higher expressions in the secretory phase (Fujimoto et al., 1996; Matsuzaki et al., 2010). Here, we showed significantly higher CDH1 protein levels in EEC at the receptive phase than the pre-receptive phase, consistent with its role in implantation. The discrepancies among studies in endometrial CDH1 expression likely stem from differing cycle phase classification, patient cohorts, and precise quantification methodologies. For example, our study uniquely focused on CDH1 protein expression within purified EEC, in contrast to prior research that primarily analyzed gene expression or examining whole endometrial tissues.

We found lower CDH1 levels in EECs from non-pregnant women and women with RIF when compared to those with a LB. Similar observations were reported in endometrial biopsies from women with RIF (Yang et al., 2017; Stern-Tal et al., 2020) and those with primary infertility (Bellati et al., 2019). The expression of CDH1 in infertile women with intramural fibroids (Makker et al., 2017) and hydrosalpinges (Poncelet et al., 2010) is significantly lower than in fertile women, consistent with impairment of endometrial receptivity with the loss of CDH1. CTNND1 is the upstream regulator of CDH1 and works closely with CDH1 at the plasma membrane (Ireton et al., 2002). Therefore, it is reasonable to find positive correlations of CTNND1/CDH1 expression in EECs with successful IVF outcomes.

The limited availability of human preimplantation embryos for research makes it difficult to identify the embryonic surface proteins involved in implantation. This is solved in this study by using BAP-EB, which can be produced on a large scale. BAP-EB-72 h is used because its transcriptome is similar to that of pTE of human Day 7 blastocysts and selectively attaches to the receptive EEC (Lee et al., 2015; Yue et al., 2020). Bioinformatics comparison of the surface proteins between the highly adhesive BAP-EB-72 h and the moderately adhesive BAP-EB-48 h showed an enrichment of genes related to ‘cadherin binding involved in cell-cell adhesion’. In silico analysis identified the potential homophilic CDH1 binding between BAP-EB and EEC. Consistently, among the five reported transmembrane adhesion molecules that were identified as binding partners of CDH1 (Shafraz et al., 2020) and highly expressed in human blastocysts (Petropoulos et al., 2016), CDH1 was ranked as the highest potential binding partner, further supporting the hypothesis that CDH1 on BAP-EB might be a binding partner of endometrial CDH1.

The present study has several limitations. First, although BAP-EB mimicked human TE and was useful for studying endometrial receptivity according to our publications (Lee et al., 2015; Yue et al., 2020; Chen et al., 2023; Lee et al., 2023), it lacked the inner cell mass and primitive endoderm cells normally present in a blastocyst. Future validation with the use of more advanced embryo models like human blastoids (Yu et al., 2021; Kagawa et al., 2022) would provide stronger evidence for the roles of CDH1 during early embryo implantation. Second, endometrial carcinoma cell lines were used for most of the in-vitro experiments due to the limited availability of primary EEC. The recently published endometrial organoid-derived open-faced endometrial cell layer (Kagawa et al., 2022) could be an alternative for studying the embryo–maternal interaction. Third, to overcome the potential loss of native polarity in 2D-cultured primary EECs, a more physiologically relevant 3-dimensionally reconstructed endometrium should be developed. Fourth, CDH1 knockdown intrinsically disrupts overall epithelial integrity, which can lead to changes in apical membrane morphology as assessed by AFM. The possibility that secondary effects on AFM measurements arise from the global disruption of adherens junctions, rather than solely from decreased apical CDH1 expression, could not be ruled out.

Here, we provide the first experimental demonstration that homophilic CDH1 binding mediates the attachment of embryo surrogates onto EEC. Five observations support this conclusion: (i) the apical surface of the pTE of human blastocysts expressed higher levels of CDH1 protein than that of the mural TE; (ii) the expression of apical surface CDH1 was most prominent and intense on BAP-EB-72 h with high attachment potential onto EEC; (iii) AFM-SMFS showed higher CDH1-binding ability on BAP-72 h than BAP-24 h; (iv) pre-treatment of either the Ishikawa cells, BAP-EB-72 h, or both with anti-CDH1 blocking antibody/peptide significantly reduced the attachment rates to a similar extent; and (v) the expression percentage of CDH1 in the apical region of human endometrial luminal epithelium was higher in women with LBs than RIF patients. Our results showed that an antibody against the ectodomain of CDH1 reduced the BAP-EB-72 h by ∼50%, indicating that the homophilic CDH1 binding is a major mediator of the attachment process. A similar experiment using an antibody against integrin avβ3 reduced the number of attached trophoblast spheroids on Ishikawa cells by ∼30% (Schmitz et al., 2014).

Data availability

The data underlying this article are available in the article and in its online supplementary material.

Authors’ roles

Conceptualization: Y.L.L., W.S.B.Y., E.H.Y.N., N.X., and Y.X. Methodology: Y.L.L., W.S.B.Y., N.X., H.R., C.L.L. Investigation: H.R., A.C.H.C., Y.X., K.C.L., S.W.F., R.H., J.H., Y.F., Q.Q., Y.T. Recruitment of samples: R.H.W.L., E.H.Y.N. Supervision: Y.L.L., W.S.B.Y., E.H.Y.N., R.H.W.L., N.X., K.-F.L., T.W. Writing—original draft: H.R., Y.L.L., W.S.B.Y. Writing—review & editing: Y.L.L., W.S.B.Y.

Funding

InnoHK initiative of the Innovation and Technology Commission of the Hong Kong Special Administrative Region Government (Health@InnoHK). Health and Medical Research Fund (grant numbers: HMRF 04151546; 10212996; 11222296) from the Food and Health Bureau, Government of the Hong Kong Special Administrative Region. Grant for Fertility Innovation 2016 from Merck. Shenzhen Science and Technology Program (grant number: KQTD20190929172749226). General Research Fund (grant number: 17212922). Collaborative Research Fund C7100-22GF from the Research Grants Council of Hong Kong. Shenzhen Sanming Project of Medicine (SZSM202211014).

Conflict of interest

The funding bodies have no role in design and conduct of the study, collection and analysis of the data, and preparation of the manuscript. The final manuscript was sent to Merck before its submission to the journals. We declared no conflict of interest.