Glycomics reveal that ST6GAL1‐mediated sialylation regulates uterine lumen closure during implantation

Abstract Objectives Implantation failure is a major cause of prenatal mortality. The uterine lumen closure contributes to embryo adhesion to the uterus, but its underlying mechanisms are largely unknown. Our previous study has reported that endometrial fold extension can lead to uterine lumen closure in pigs. The objective of this study was to reveal molecular mechanisms of the uterine lumen closure by characterizing the molecular basis of the endometrial fold extension during implantation in pigs. Materials and methods Uterine and endometrium tissues during implantation were collected in pigs. MALDI‐TOF MS was used to characterize the N‐glycomic profiles. Histochemistry, siRNA transfection, Western blotting, lectin immumoprecipitation, mass spectrometry and assays of wounding healing and cell aggregation were performed to investigate the molecular basis. Results We observed that uterine luminal epithelium (LE) migrated collectively during endometrial fold extension. For the first time, we identified a large number of N‐glycan compositions from endometrium during implantation using MALDI‐TOF MS. Notably, the α2,6‐linked sialic acid and ST6GAL1 were highly expressed in uterine LE when the endometrial folds extended greatly. Subsequently, the role of ST6GAL1‐mediated 2,6‐sialylation in collective epithelial migration was demonstrated. Finally, we found that ST6GAL1‐mediated α2,6‐sialylation of E‐cadherin may participate in collective migration of uterine LE. Conclusions The study reveals a mechanism of uterine lumen closure by identifying that ST6GAL1‐mediated α2,6‐sialylation of cell adhesion molecules contributes to endometrial fold extension through regulating collective migration of uterine LE.


| INTRODUC TI ON
Prenatal mortality caused by implantation failure is a common reproductive health concern for both humans and farm animals. It is of importance to elucidate the mechanisms of the implantation.
During early implantation, the uterine lumen closure is a critical event contributing to embryo adhesion to the uterine LE. The morphological features and mechanisms of uterine lumen closure are well characterized in the mouse models, 1,2 but they remain unclear in humans and large animals due to the difficulty in collecting uterine samples during early implantation stage.
The litter-bearing pig (Sus scrofa) is one of the most important farm animals. Besides, the pig serves as a large biomedical model to address various human health issues, including reproductive health. 3 Pig blastocysts shed the zona pellucida around gestational days [6][7] (with entire gestational length of the pig of 114 days) and continue to develop into conceptuses (embryo/foetus and related extraembryonic membranes). The conceptuses freely float in the uterine lumen until gestational days 12-13 and then start to adhere to the uterine (LE) to form an epitheliochorial placenta around gestational day 18. 4,5 Previous studies have found that (1) when conceptuses remain freefloating (before gestational day 12), the uterine lumen is in an open state and a few scattered endometrial folds occur; (2) at the stage of conceptus attachment (around gestational day 15), the endometrial folds that extend greatly to the uterine lumen interlock with each other, thus resulting in closure of the uterine lumen 5,6 ; (3) around gestational day 18, the uterine lumen reopens due to the expansion of chorioallantois. 4 Therefore, the endometrial fold extension is, at least partially, the cause of the uterine lumen closure in pigs. Although a number of genes and pathways have been reported to be involved in implantation in pigs, 7,8 the mechanisms underlying the endometrial fold extension during implantation remain largely unknown.
Cell surface glycans participate in many important biological processes. [9][10][11] The maternal-foetal interface is enriched in various glycans, [12][13][14] suggesting that glycans and glycan-modified proteins play critical roles in embryo implantation. In pigs, some glycoproteins have been isolated and characterized during implantation. 15,16 By using lectin histochemistry approach, several glycans are found to be present in the pig maternal-foetal interface throughout gestation. 17,18 However, the glycome (N-and O-glycome) of the maternalfoetal interface during implantation and the functions of glycans and glycan-modified proteins in implantation remain to be investigated in large animals and humans.
The objective of this study was to reveal the molecular basis of the endometrial fold extension during implantation in pigs by investigating the N-glycomic profiles of pig endometrium during implantation and characterizing the role of glycans in regulating the collective migration of uterine LE during implantation.

| Ethics statement
All animal procedures were approved by the Ethics Committee of Huazhong Agricultural University (HZAUSW-2016-015).

| Sample collection
Sample collection was carried out according to Wang et al. 6 Briefly, Yorkshire gilts were mated to Yorkshire boars at the onset of the second estrus (Day 0) and again 12 h later. On gestational days 12, 15, and 18, the uterus was taken from each gilt and cut into segments (3-4 gilts/gestational day). Some of the uterine segments were randomly selected to be flushed with precooled RNase-free phosphate buffer saline (PBS). When pregnancy was confirmed by the presence of conceptuses in the uterine flushing, the uterine segments in which the conceptuses were flushed out, were used for endometrium collection. The uterine segments that were not flushed were fixed immediately in 10% neutral-buffered formalin followed by paraffin embedding.

| Histological and histochemistry
The paraffin-embedded intact uterine samples in which the conceptuses were not flushed out were cross-sectioned into 4 μm thick using a Leica microtome (RM2235, Leica, Germany) and then deparaffinized and rehydrated by passage through xylene and a grade alcohol. Sections that were stained with haematoxylin and eosin were used for histological analysis.
Immunofluorescence assays were carried out as described by Wang et al. 6 The primary antibodies used were ST6GAL1 (1:50

| Isolation of N-linked glycans from pig endometrium
The endometrial tissues obtained from 3 gilts at each gestational day were pooled in equal weight (one pool/gestational day). Each pooled tissue sample (500 mg) was homogenized in denaturing buffer (0.5 M Tris-HCl, pH 8.5, containing 7 M guanidine-HCl and 10 mM EDTA) using an Ultrasonic Processor (Sonics VCX130, Sonics & Materials Inc). The sonicated solution was then centrifuged to collect the supernatant which contains proteins. The proteins (300 μg per sample) were reduced with 10 mM DL-Dithiothreitol (DTT) at 60°C for 30 min, alkylated with 20 mM iodoacetamide (IAA) at room temperature for 60 min in the dark, dialysed against 10 mM ammonium bicarbonate (PH 8.6) for 48 h and digested with TPCK treated trypsin (T1426; Sigma) at 37°C for 18 h. After inactivation of trypsin by heat treatment (95°C for 5 min), 3 mU peptide N-glycanase F (PNGase F, P0704S; New England Biolabs) was added to release the intact Nglycans. The released N-glycans were treated using the BlotGlyco ® glycan purification kit according to the manufacturer's protocol (Sumitomo Bakelite Co.). Briefly, the released N-glycans were captured using BlotGlyco beads, followed by methylesterification of sialic acids with 3-methyl-1-p-tolytriazene (98%; Sigma-Aldrich).
Then, the captured N-glycans were labelled and released with an aminooxy-functionalized peptide reagent (aoWR). The derivatized glycans were recovered from the resin by washing with 50 μl of distilled water. Finally, excess reagent was removed using a cleanup column provided with the kit. The obtained solution containing the N-glycan derivatives was analysed via mass spectrometry (MS).

| Characterization of N-glycomic profiles using MALDI TOF/TOF mass spectrometry
Mass spectra were acquired with a MALDI TOF/TOF mass spectrometer (New ultrafleXtreme; Bruker Daltonik). For sample preparation, 2,5-Dihydroxybenzoic acid (DHB, 10 mg/ml in 30% ethanol) was used as the matrix and the DHB matrix (0.5 μl) was spotted onto a target plate (MTP 384 target plate ground steel; Bruker Daltonik) and dried.
Subsequently, an aliquot (0.5 μl) of the N-glycan solution was spotted onto the 2,5-dihydroxybenzoic acid crystal and dried. Ions were generated by signal averaging over 4,000 laser shots from a Smartbeam-II Nd:YAG laser operating at 355 nm and a repetition rate of 2 kHz. All spectra were obtained using the reflectron mode with a delayed extraction of 200 ns. The area of the isotopic peak of each glycan was normalized to an internal standard (maltoheptaose) with a known concentration. The ratio of each glycan's peak area to the sum of all glycans' peak area represents the content of a certain glycan. The glycan compositions were determined by database searching using GlycoWorkbench software. 19 Briefly, exact mass of 'labelling reagent' was 432.22, and glycans labelled with the reagent were mainly de-

| Lectin-histochemical analysis
The lectin-histochemical study was performed on pig uterine cross sections. Slides were incubated with either fluorescein isothiocy-

| Cytochemistry analysis
Ishikawa cells were cultured in DMEM (C11995500BT; Gibco) with 10% foetal bovine serum (FBS, SE200-ES; VisTech™) and 1% antibiotics (15240062; Gibco). Cells were seeded at a density of 1 × 10 5 cells on 12-well glass bottom plates and fixed with 4% paraformaldehyde for 10 min, washed three times with cold PBS and permeabilized with 0.1% Triton X-100 in PBS for 10 min. After being washed with cold PBS for 3 times, cells were blocked with PBST buffer containing 1% BSA and 10% goat serum for 2 h at room temperature. Olympus Corporation).

| Lectin inhibition assay and siRNA transfection
To investigate the effect of α2,6-sialylation on cell migra- Q uantitative RT-PCR (qRT-PCR) and Western blot were performed after cells were transfected with si-ST6GAL1 or si-NC for 72 h. Takara Biomedical Technology). GAPDH was used as the internal reference. All primer sequences are shown in Table S1.

| Western blotting
Protein extraction from tissues or cells was performed using RIPA

| Wounding healing assay
Ishikawa cells were seeded in a 6-well plate ( Olympus Corporation). The cells were counted from three fields, and three independent experiments were performed.

| Lectin immumoprecipitation
A total of 500 mg pig endometrium from gestational days 12, 15 and The immunoprecipitated proteins that were probed with SNA was used as the loading control. In addition, the expression level of Ecadherin in proteins that were not immunoprecipitated were detected in parallel and β-actin (1:2500, AF5003; Beyotime) was used as the loading control. Proteins were detected using the ECL Western blot kit (170-5,060; Bio-red) and analysed by a chemiluminescent imaging system (Tanon-5200; Tanon Science and Technology).

| Identification of α2,6-sialylated proteins using mass spectrometry analysis
The proteins immunoprecipitated from pig endometrium of ges-

| Statistical analysis
Statistical analyses were carried out using GraphPad Prism Software version 5 (GraphPad Software). The statistical significance of differences was analysed using nonparametric Mann-Whitney one tail/ two tail test. The data were expressed as means ±standard error.
p < 0.05 was considered statistically significant. All graphs were generated using GraphPad Prism software.

| Uterine LE migrates collectively during endometrial fold extension
The histological analysis of pig uterine cross section revealed that  Figure 1F). In addition, Western blotting analysis revealed that WAVE1 was expressed in pig endometrium on gestational days 12 and 15 ( Figure 1G). Taken together, these results showed that E-cadherin, β-catenin and RhoA were consistently expressed in uterine LE, while Rac1 and WAVE1 were expressed on gestational day 15 in uterine LE and endometrium, respectively, indicating that the uterine LE might migrate collectively at which the endometrial folds extend greatly.

| Sialylated glycans are highly expressed in endometrium on gestational days 12, 15 and 18 by MALDI-TOF MS
The N-glycan profiles of pig endometrium on gestational days 12, 15 and 18 were characterized by MALDI-TOF MS respectively ( Figure 2A). Each peak in the MS spectra was assigned to the corresponding glycan composition (Table S2). To simplify the comparison of the profiles of the released glycans on gestational days 12, 15 and 18, the area of the isotopic peaks of each glycan (peak 2-37) in the MS spectra was normalized to an internal standard with a known concentration (peak 1) and represented as a histogram ( Figure 2B,  residues respectively. 25 The MAL-II stain signal was observed in subepithelial fibroblasts, stroma and blood vessel walls, but not in uterine LE and glandular epithelium ( Figure 3B, Figure S1). The SNA stain signal was detected in glandular lumen and blood vessel. In uterine LE, SNA stain signal was weak on gestational day 12 and strong on day 15, but undetectable on day 18 ( Figure 3B, Figure S1).
Collectively, the α2,6-linked sialic acid glycan is expressed in uterine LE in a stage-specific manner during implantation.
Further, we investigated the expression pattern of two βgalactoside α2,6-sialyltransferases, ST6GALI and ST6GALII, which are responsible for transferring sialic acid to Galβ1,4GlcNAc (Nacetyllactosamine) through α2,6-linkage. 26 The qRT-PCR results showed that the expression level of ST6GAL1 in pig endometrium was significantly higher on gestational days 12 and 15 than on day 18 ( Figure 3C, two-sample nonparametric Mann-Whitney one tail test). Further immunofluorescence assay revealed that the expression level of ST6GAL1 in uterine LE was low on day 12 and was high on day 15, but it was undetectable on day 18 ( Figure 3D). In addition, ST6GAL1 was detectable in glandular lumen and blood vessel during these 3 gestational days ( Figure S2). In contrast to ST6GAL1, neither ST6GAL2 mRNA nor ST6GAL2 protein was detectable in pig endometrium during these 3 gestational days ( Figure 3C, 3D and Figure S2). These results suggest that the expression of α2,6-linked sialic acid in pig endometrium is mediated by ST6GAL1. In addition, the expression pattern of α2,6-linked sialic acid was consistent with that of ST6GAL1 in uterine LE, and their expression pattern was stage-specific.  Table S1. (B) Histogram of the relative area of each peak. The area of peak 1, which is the internal standard (maltoheptaose) with a known concentration, was defined as 100. GD, gestational day

| α2,6-sialylation promotes cell migration and cell-cell adhesion
Fluorescent staining with SNA and MAL-II revealed that α2,6-linked sialic acid rather than α2,3-linked sialic acid was expressed in Ishikawa cells. In addition, staining signals of ST6GAL1 and E-cadherin were also observed in Ishikawa cells ( Figure S3). To investigate the effect of α2,6-sialylation on cell migration and cell-cell adhesion, the α2,6linked sialic acids were blocked through cellular SNA treatment.
As shown in Figure S4, the α2,6-linked sialic acid expression levels were decreased in Ishikawa cells after SNA treatment. Woundhealing assay results indicated that blocking of cellular α2,6-linked sialic acids by SNA treatment resulted in a decrease in the ability of wound healing at 24 h, 48 h and 72 h post-treatment ( Figure 4A, B, two-sample nonparametric Mann-Whitney test). In addition, cell aggregation assay showed that only ~26% of the aggregates containing more than 10 cells formed in the cells whose α2,6-linked sialic acids were blocked, but ~69% of the aggregates formed in Ishikawa cells whose α2,6-linked sialic acids were not blocked ( Figure 4C, D). The results suggest that the presence of α2,6-sialylation promotes cell migration and cell-cell adhesion.

| ST6GAL1-mediated α2,6-sialylation of Ecadherin contributes to collective cell migration
The small interfering RNA (siRNA) technique was used to knock  Figure 6A). As shown in Figure 6A and B, the level of E-cadherin extracted from Ishikawa cells transfected with si-ST6GAL1 was consistent with that from Ishikawa cells transfected with si-NC. However, the level of E-cadherin in the SNA immunoprecipitates from the cells transfected with si-ST6GAL1 was lower than that from the cells transfected with si-NC ( Figure 6B).
The above results indicated that the knockdown of ST6GAL1 by siRNA resulted in decreased α2,6-sialylation of E-cadherin. Taken together, the above findings suggest that ST6GAL1-mediated α2,6sialylation of E-cadherin facilitates collective cell migration.

| E-cadherin is highly α2,6-sialylated in uterine LE during the endometrial fold extension
To identify the α2,6-sialylated proteins in pig endometrium, the α2,6-sialylated proteins were pulled down with SNA lectin from pig endometrium on gestational days 12, 15 and 18. The precipitated α2,6-sialylated proteins from gestational day 15 were determined by LC-MS/MS (n = 3 gilts). A total of 796 α2,6-sialylated proteins were identified to be mainly enriched in functional terms related to cell adhesion ( Figure 7A, Table S3). It is worth noting that E-cadherin, one of the collective cell migration markers, is in the list of these identified α2,6-sialylated proteins. We further examined the change in the α2,6-sialylation level of E-cadherin in pig endometrium on gestational days 12, 15 and 18. The Western blotting analyses showed that the expression level of E-cadherin was similar in endometrium during these 3 gestational days, but it was much higher in SNA immunoprecipitates from endometrium lysates on gestational day 15, the day when the endometrial folds extend greatly ( Figure 7B).

| DISCUSS ION
In The sialylation plays essential roles in various biological process. 37,38 Although α2,6-sialylation alteration has been reported to have impact on endometriosis development in humans, 39,40 little is known about the role of α2,6-sialylation in implantation. In this study, we found that α2,6-linked sialic acid was expressed in uterine LE and that the α2,6-sialylation was me- reports have demonstrated that the modification of E-cadherin with high-mannose, hybrid, sialylated and complex N-glycans can affect the stability of the adhesion junction, and such an effect is dependent on specific N-glycan structures linked to E-cadherin. 46 However, the sialylation of E-cadherin was only detected in a canine mammary carcinoma cell line, and its sialylation was positively correlated with malignant phenotype. 47 In the present study, we demonstrated that E-cadherin was α2,6-sialylated in pig uterine LE and Ishikawa cells. In addition, our in vivo and in vitro data revealed that α2,6-sialylation level of E-cadherin was increased with the enhanced collective cell migration mediated by Rac1. Taken together, our results suggest that collective cell migration of pig uterine LE may be modulated by α2,6-sialylation of E-cadherin ( Figure 8).

F I G U R E 7
Identification of the changes in α2,6-sialylation levels of proteins from pig endometrium. (A) GO analysis of the α2,6-sialylated proteins identified from pig endometrium on gestational day 15. (B) The α2,6-sialylated proteins from pig endometrium were pulled down with SNA lectin, and then, Western blotting was performed to evaluate the α2,6-sialylation levels of E-cadherin. The immunoprecipitated proteins that were probed with SNA were used as loading controls (left panel). Pig endometrium lysates were probed with an antibody against E-cadherin to show the amounts of E-cadherin, and β-actin was used as a loading control (right panel). n = 3 gilts/gestational day. IP, immunoprecipitation. WB, Western blot. GD, gestational day

| CON CLUS ION
Our study indicates that pig uterine luminal epithelial cells may migrate collectively during endometrial fold extension which is a key process leading to uterine lumen closure during implantation. By using MALDI-TOF MS, we identified a large number of N-glycan compositions from endometrium during implantation. Furthermore, our study provides evidence to support that ST6GAL1-mediated α2,6-sialylation of cell adhesion molecules, such as E-cadherin, facilitates uterine lumen closure by regulating collective migration of uterine LE during endometrial fold extension. Our finding can provide an insight into embryo implantation in pigs as well as other large animals and humans whose endometrial samples are largely inaccessible during implantation.

F I G U R E 8
Effect of α2,6-sialylation mediated by ST6GAL1 on regulation of pig endometrial fold extension during implantation. The increase in α2,6-sialylation of the cell adhesion molecule E-cadherin occurs coincident with the activation of Rac1/WAVE1 signal pathway, suggesting that α2,6-sialylation of E-cadherin mediated by ST6GAL1 may have a role in regulating endometrial fold extension by promoting the collective migration of uterine LE during implantation

ACK N OWLED G EM ENTS
We thank Dr. Dingxiao Zhang for helpful advice of cell woundhealing assay and Dr. Kan Ding for helping with N-glycome analysis. We also thank Ms Jing Xu for technical support and Miss Tara Mahmood for English language editing.

CO N FLI C T O F I NTE R E S T
The authors have declared that no competing interests exist.

DATA AVA I L A B I L I T Y S TAT E M E N T
The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE partner repository with the data set identifier PXD027571. Data are available via ProteomeXchange.