Human endometrial CD98 is essential for blastocyst adhesion

Abstract

Background: Understanding the molecular basis of embryonic implantation is of great clinical and biological relevance. Little is currently known about the adhesion receptors that determine endometrial receptivity for embryonic implantation in humans.

Methods and principal findings: Using two human endometrial cell lines characterized by low and high receptivity, we identified the membrane receptor CD98 as a novel molecule selectively and significantly associated with the receptive phenotype. In human endometrial samples, CD98 was the only molecule studied whose expression was restricted to the implantation window in human endometrial tissue. CD98 expression was restricted to the apical surface and included in tetraspanin-enriched microdomains of primary endometrial epithelial cells, as demonstrated by the biochemical association between CD98 and tetraspanin CD9. CD98 expression was induced in vitro by treatment of primary endometrial epithelial cells with human chorionic gonadotropin, 17-β-estradiol, LIF or EGF. Endometrial overexpression of CD98 or tetraspanin CD9 greatly enhanced mouse blastocyst adhesion, while their siRNA-mediated depletion reduced the blastocyst adhesion rate.

Conclusions: These results indicate that CD98, a component of tetraspanin-enriched microdomains, appears to be an important determinant of human endometrial receptivity during the implantation window.

Figure 1. Flow cytometry analysis of the expression of cell adhesion molecules in endometrial epithelial cells and cell lines.

Histogram profiles of the plasma membrane expression of CD9 (VJ1/20), ICAM-1 (HU5/3), β1 integrin (TS2/16), αvβ3 integrin (8D6), CD147 (VJ1/9) and CD98 (FG1/10) on HEC-1-A (green) and RL95-2 (blue) cell lines as well as on primary EEC cultures (pink). Negative control pX63 is shown in gray line.

Figure 2. CD98 is expressed in the implantation window in human endometrium.

A. Immunolocalization of ICAM-1 (HU5/3 mAb), CD147 (VJ1/9 mAb) and CD9 (VJ1/20 mAb) in human endometrium throughout the menstrual cycle. The micrographs show representative immunohistochemical stains of luminal epithelium in group 4 samples (day 20), corresponding to the implantation window. Images correspond to a 200× magnification. The charts plot expression based on semi-quantitative staining analysis throughout the menstrual cycle in three to five endometrial samples per group in epithelial glands (▪), luminar epithelium (♦) or stroma (▴). B. Immunolocalization of CD98 (anti-CD98 pAb) in human endometrium throughout the menstrual cycle. Micrographs depict the luminar epithelium in the 5 menstrual cycle groups in sequence: group1,  =  day 5, group 2 =  day9, group 3 =  day 15, group 4 =  day 20, and group 5 =  day 25. Images correspond to a 200× magnification. The chart plots semi-quantitative analysis of the stainings throughout the menstrual cycle in three to five endometrial samples per group as in A.

Figure 3. CD98 is inserted into tetraspanin-enriched microdomains on the apical surface of human endometrial cells.

A. Confocal analysis of adhesion molecule expression in cultures of polarized human primary endometrial epithelial cells. Vertical sections were obtained with Leica Confocal Software (LASAF). Dashes mark the substratum position, whereas arrows point to the apical surface of the monolayer. B. Confluent EEC monolayers were double stained for CD98 (FG1/10) and β1 integrin (TS2/16) or β3 integrin (P97) with CD9 (VJ1/20) and analyzed by confocal microscopy. Nuclei were labeled with Hoechst (blue). Merged images and vertical sections of the double staining are also shown. Bars: 10 µm. C. Confluent EEC monolayers were lysed in buffer containing 1% Brij 96 and lysates were immunoprecipitated with anti-β3 (P97), anti-β1 (TS2/16) integrin chains mAbs or anti-CD98 serum. Non-immune Ig was included as negative control. Western blots were probed with TS2/16 for β1, and rabbit polyclonal Abs against β3 integrin and CD98. Molecular weights (kDa) are shown. D. Confluent EEC monolayers were lysed in buffer containing 1% Brij 96 and lysates were immunoprecipitated with anti-CD9 (VJ1/20), anti-ICAM-1 (HU5/3) or anti-CD147 (VJ1/9) mAbs or anti-CD98 serum. Non-immune Ig was included as negative control. Western blots were probed for CD9 (VJ1/20) and CD98 (anti-CD98 pAb). Different exposures of the same membrane are shown for CD98 IP. Molecular weights (kDa) are shown.

Figure 4. Endometrial CD98 expression is induced by hormones in vitro.

A. Flow cytometry analysis of the membrane expression of CD98 (FG1/10 mAb), ICAM-1 (HU5/3 mAb), CD147 (VJ1/9 mAb) and CD9 (VJ1/20 mAb) in EEC cells after treatment for 48 h with hCG (100 U/ml), 17-β-estradiol (30 nM), progesterone (1 µM) or 17-β-estradiol plus progesterone. Data are the mean ± s.e.m. of the Mean Fluorescence Intensity (MFI) normalized to untreated cells in six independent primary EEC cultures. * p<0.02 ** p<0.00001 versus untreated cells in a Student T test. B. Western blot analysis of CD98 expression in total lysates of primary EEC after treatment for 48 h with 17-β-estradiol (30 nM), progesterone (1 µM) or 17-β-estradiol plus progesterone. GAPDH expression is shown as loading control. Numbers correspond to the densitometric analysis of CD98 protein content corrected by GAPDH loading and normalized to untreated cells. Molecular weights (kDa) are shown. C. Western blot of CD98 expression in total cell lysates of EEC after exposure to hCG (100 U/ml) for 48 h. α-tubulin expression is shown as loading control. Numbers correspond to the densitometric analysis of CD98 protein content corrected by tubulin loading and normalized to untreated cells. Molecular weights (kDa) are shown. D. Flow cytometry analysis of the membrane expression of CD98 in EEC cells after treatment for 48 h with IL1β (20 ng/ml), LIF (1000 U/ml), TGFβ (10 ng/ml), NGF (10 ng/ml), EGF (100 ng/ml) or IGF (1 µg/ml). Data are the mean ± s.e.m. of the MFI normalized to untreated cells in four independent experiments. * p<0.05 ** p<0.02 versus untreated cells in a Student T test. E. Mouse blastocysts were allowed to adhere to confluent HEC-1-A monolayers, and samples were fixed and stained for human CD98 (FG1/10, shown in green), and analyzed by confocal microscopy. Nuclei were labeled with Hoechst (blue). A pseudocolor image of CD98 staining intensity and a phase contrast image are shown. Mouse blastocyst is outlined on the pseudocolor image. Bar: 50 µm.

Figure 5. Overexpression of CD9 and CD98 enhances the receptivity of HEC-1-A cells, while their silencing impairs adhesion.

A. HEC-1-A cells were transiently transfected with plasmid encoding the GFP-tagged versions of human ICAM-1 or CD9. 24 h after transfection cells were trypsinized and labeled with the mAb specific for the transfected receptor (HU5/3 anti-ICAM-1 or VJ1/20 anti-CD9). Dot plots of GFP versus mAb staining confirm that the GFP-expressing cells overexpress the corresponding receptor at the plasma membrane. For silencing CD9 expression, HEC-1-A cells were transfected with siRNA oligonucleotide (filled histogram), or negative control oligonucleotide (white histogram) and stained for CD9 (VJ1/20) 48 h after transfection. B. HEC-1-A cells were transiently transfected with GFP-tagged versions of CD4, ICAM-1 and CD9, or with CD9-targetted RNAi oligonucleotides, in independent experiments. Adhesion of mouse blastocysts to transfected cells was analyzed after co-culture for 24 or 48 h. Data are mean ± s.e.m. of three independent experiments, expressed relative to the rate of adhesion on untransfected cells in each experiment. The total number of blastocysts assessed in each condition were, CD4 n = 91, ICAM-1 n = 106, CD9 n = 169, CD9 siRNA n = 154. * p<0.05 (Mann-Whitney t test) compared with the corresponding control experiment with untransfected cells. Negative controls of adhesion ranged 50–60% for overexpression experiments analyzed at 24 h, and 70% for silencing experiments analyzed at 48 h. C. Flow cytometry analysis of CD98 (FG1/10) membrane expression after infection of HEC-1-A cells with lentivirus encoding human CD98 or CD98 shRNA. Expression was analyzed 72 h after infection. Expression levels in cells infected with GFP control lentivirus are shown in the white histograms. D. HEC-1-A cells were infected with lentivirus encoding GFP, CD98 plus GFP, or CD98 shRNA plus GFP. Three days after infection HEC-1-A monolayers were cocultured with mouse blastocysts and adhesion was quantified after 24 or 48 h. Data are mean ± s.e.m. Total number of blastocysts assessed were, non-infected n = 50, GFP n = 61, CD98 n = 56, GFP n = 39, CD98 shRNA n = 64. * p<0.05 (Mann-Whitney t test) versus GFP.