Molecular Human Reproduction, Vol. 6, No. 1, 41-49,
January 2000
© 2000 European Society of Human Reproduction and Embryology
Uterus and pregnancy |
Mosaic characteristics of human endometrial epithelium in vitro: analysis of secretory markers and cell surface ultrastructure
1 Department of Obstetrics and Gynaecology, University of Glasgow, Royal Infirmary, 10 Alexandra Parade, Glasgow G31 2ER, 2 Academic Unit of Obstetrics and Gynaecology, School of Medicine, University of Manchester, St Mary's Hospital, Manchester M13 OJH, 3 Department of Ultrastructure, CRC Paterson Laboratories, Christie Hospital, Manchester M20 9BX, and 4 Department of Pathology, Withington Hospital, Manchester M20 2LR.
Abstract
Specific terminal carbohydrate structures and mucin-associated glycans increase in expression within the human endometrial epithelium during the secretory phase of the menstrual cycle but exhibit wide intercellular variation. We postulated that variation in glycosylation between cells would produce differences in the glycocalyx and result in complex mixtures of cells bearing different combinations of glycans. MUC-1 mucin, keratan sulphate and fucosylated lactosaminoglycans were examined in epithelial gland fragment cultures with antibodies (HMFG1, 5D4) and a lectin (Dolichos biflorus agglutinin). The glycocalyx was examined by transmission and high resolution scanning electron microscopy. The data were related to patterns of expression seen in vivo. The MUC-1 mucin was expressed relatively uniformly in culture, but heterogeneity was evident in mucin sialylation within the epithelial cell population. Double labelling of gland explant cultures for combinations of fucosylated lactosaminoglycans, keratan sulphate and MUC-1 demonstrated cells expressing all combinations of these markers. Ultrastructural examination confirmed remarkable intercellular variation in the glycocalyx. Though the human endometrial epithelium is relatively morphologically homogeneous, these observations reveal complex variations of cell surface glycosylation between neighbouring cells and suggest that secretory function might vary in a similar fashion.
endometrium/epithelium/glycan/human/mosaic
Introduction
Human endometrial epithelial cells are responsive to ovarian steroids and consequently change morphologically and biochemically during the course of the menstrual cycle. However, there is evidence that the glandular epithelial cell population does not respond uniformly (Graham et al., 1994
; Hey et al., 1994, 1995). In the present study, we sought to examine epithelial heterogeneity in vitro with markers of the secretory response and by ultrastructural examination and to correlate the results with observations of cells in vivo.
Heterogeneity of the epithelium first becomes apparent when the endometrium begins to regenerate after menstruation. The basal portion of the glands, which remain after shedding, form a new surface epithelium by cell migration over the denuded endometrial surface (Markee, 1940; Noyes et al., 1950; Ferenczy and Reichart, 1973; Ferenczy, 1976a,b; Ludwig, 1990). The endometrium then thickens and the glands elongate by cell proliferation for 10–11 days prior to ovulation. Epithelial proliferation in this phase of the cycle is accompanied by ciliogenesis in both the glandular and surface cells (Masterson et al., 1975). By mid-cycle the epithelial population has become a mosaic of ~20% ciliated cells, the remainder being microvillous.
After ovulation, epithelial proliferation slows and secretory differentiation commences under the influence of progesterone. The proportion of ciliated cells slowly declines. During the secretory phase, a sequence of morphological changes occurs within the glandular epithelial cells; in the surface (luminal) epithelium these are less pronounced (Noyes et al., 1950; Johannisson et al., 1982; Jansen and Johannisson, 1985; Li et al., 1988; Saleh et al., 1995). Immunohistochemical localization of secretory markers within the glandular epithelium has confirmed that the post-ovulatory morphological changes are accompanied by biochemical changes within the cells (Aplin, 1991, 1994). These markers also reveal variations in epithelial phenotype. Changes in the expression of the sialylated and sulphated lactosaminoglycan epitope recognized by monoclonal antibody (mAb) D9B1 have been detected within the first day after ovulation (Seif et al., 1989; Smith et al., 1989). Keratan sulphate, although detectable in proliferative phase epithelial cells, is greatly increased during the secretory phase (Hoadley et al., 1990; Shiozawa et al., 1991, Graham et al., 1994) but remains in a highly heterogeneous distribution in the glandular and luminal epithelium. Blood group A-related glycans containing terminal N-acetylgalactosamine and fucose residues recognized by the Dolichos biflorus agglutinin (DBA) appear in the mid secretory phase and are also heterogeneously expressed in endometrial epithelial cells (Aoki, 1989; Aplin, 1991; Jones et al., 1998). The appearance of these binding sites is inhibited by anti-progestins (Gemzell-Danielsson et al., 1994, 1996; Aplin et al., 1997; Jones et al., 1998).
Several hormonally-modulated glycans, including keratan sulphate, sialyl Lewis X and sialyl Lewis a are associated with the mucin MUC-1 which is also up-regulated after ovulation (Hey et al., 1994; Aplin and Hey, 1995; Hey and Aplin, 1996; Aplin et al., 1998). Using a combination of in-vivo and cell culture techniques, we have investigated the extent of cellular mosaicism in the epithelium, and whether it can be observed at the level of the MUC-1 polypeptide, or results from intercellular variation in glycosylation. A panel of glycan-specifc probes has been used in combination with mAbs BC3, which recognizes an epitope in the highly O-glycosylated tandem repeat domain of the MUC-1 core protein (Xing et al., 1989) and HMFG1, which binds a site in the same region that is masked by increasing glycosylation (Burchill et al., 1987; Hey et al., 1994; Aplin et al., 1998).
Materials and methods
Endometrial specimens
Endometrial biopsies were obtained after dilatation and curettage or hysterectomy carried out for benign conditions. Tissue sampling was undertaken according to protocols agreed with the local pathology departments and approved by the local ethics committees. In cases where additional material was removed from the uterus for the purpose of research, written and informed consent of the patient was obtained. Material was not collected for the purpose of research where there was any clinical or macroscopic suspicion of neoplasia. In specimens used for culture, histopathological examination of adjoining tissue blocks revealed no signs of metaplasia, neoplasia, inflammation, or infection. The patients were aged (mean ± SD) 37 ± 8.0 years (range 25–57) and included parous and non-parous women. Tissue specimens were routinely formalin-fixed, paraffin-embedded, and sections stained with haematoxylin and eosin for evaluation (Noyes, 1950). Samples that were taken for gland isolation and cell culture, or whole mounted for confocal microscopy, were histologically normal.
Isolation of endometrial epithelial gland fragments
Tissue was collected in sterile phosphate-buffered saline (PBS; Sigma, Poole, UK) supplemented with penicillin and streptomycin (Life Technologies, Paisley, UK). The tissue was transferred to a 90 mm diameter disposable bacteriological grade Petri dish and examined with a stereo-microscope. Blood clots were removed and the tissue fragments washed with PBS, then chopped into pieces ~1–2 mm in length with a scalpel. The fragments were transferred to a 15 ml transparent conical bottomed centrifuge tube containing 5 ml of 1 mg/ml hyaluronidase (Sigma), 1 mg/ml crude collagenase (Sigma) dissolved in Optimem culture medium (Life Technologies). The solution was changed after 15–30 min and the process repeated up to three times over 45–60 min. The digestion procedure varied in length due to the fact that gland fragments were released more quickly from some specimens than others. Before each change of digestion mixture, gland and tissue pieces were sedimented by pulse centrifugation on a bench top centrifuge. The supernatant containing red blood cells, single endometrial cells and small cell clumps was then discarded. After addition of fresh medium the glands and tissue fragments were inspected by placing the conical bottomed tube horizontally on the stage of an inverted microscope. Digestion was stopped when the bulk of the glands had been liberated from the tissue and when the external surface of the glands were smooth in outline, i.e. there were no adherent cells or fragments of connective tissue. The glands and remaining tissue were then sedimented by centrifugation in a bench top centrifuge (100 g for 5 min) and selectively filtered. Undigested tissue pieces were removed by filtration through an autoclaved 400 µm polyester mesh (Locker Wire Weavers, Warrington, UK). This filter was taped to a 100 mm square piece of stainless steel into which a 60 mm diameter hole had been cut. The assembled filter was placed on a 90 mm Petri dish and the tissue digest filtered to remove large tissue pieces. The initial filtrate containing gland fragments, epithelial clumps and single cells was passed through a 30 µm mesh and washed with 15 ml PBS to remove single cells (fibroblasts and epithelial cells). The filter was inverted and the trapped glands were washed into another Petri dish with tissue culture medium. The gland suspension was examined with an inverted phase microscope. If any single cells were present the filtration process was repeated, even though this resulted in a loss of gland fragments. Careful adherence to this isolation method produced explant cultures in which no colonies of fibroblast morphology developed over a period of 2 weeks.
Epithelial cultures
Primary epithelial explant cultures were successfully established from 24 out of 30 endometrial specimens from the proliferative and secretory phases, and maintained for a period of up to 14 days. The explants were established either on plastic tissue culture flasks, glass coverslips, or on polycarbonate membranes in bicameral chambers. These surfaces were coated with either rat tail tendon collagen or Matrigel. The basal medium was either a 50:50 mixture of Dulbecco's modified Eagle's medium and Ham's F12 (Sigma) supplemented with 10% (v/v) fetal calf serum, or a serum-free medium (Optimem; Life Technologies) supplemented with 2% (w/v) albumin (Life Technologies). The medium was supplemented with oestradiol (2 ng/ml, Sigma) or oestradiol and progesterone (100 ng/ml, Sigma) or medroxyprogesterone acetate (MPA, 50 ng/ml; Sigma). Replicate cultures were established so that the effect of either the basal medium, the type of substratum used or the hormonal milieu and combinations of these conditions could be examined by lectin and antibody binding or ultrastructure. These were examined by an unbiased observer for differences in marker expression. None of the conditions tested produced an obvious qualitative difference in secretory marker expression, morphology or ultrastructure, or the mosaic patterns observed.
Lectin- and immuno-cytochemistry
The distribution of DBA binding sites was examined on sections of hysterectomy specimens (n = 37) with biotinylated DBA (10 µg/ml, Vector). The lectin was detected with an avidin–biotin–peroxidase complex (Elite ABC; Vector) and a diaminobenzidine (DAB) substrate kit (Vector) according to the manufacturer's instructions. The specimens were then examined without nuclear counter-staining after aqueous mounting but using differential interference contrast microscopy. Immunolabelling with mAbs HMFG1 and BC3 was as previously documented (Hey et al., 1994). Images were grabbed and immunostaining segmented using a Kontron KS400 system mounted on a Leitz Dialux microscope as described (Jones et al., 1998).
DBA binding to cultured cells was detected by a variety of methods depending on the application. When cells had been cultured on plastic, the bottom of the flask was cut into fragments with sharpened joiner's pincers. DBA–peroxidase (10 µg/ml, Sigma) and DAB (Sigma) substrate were used to localize binding. The plastic fragment was mounted (Immumount; Shandon) on a glass microscope slide and the cells were observed with an inverted phase microscope. Concanavalin A–biotin (Vector) binding to cells cultured in the same way was carried out in a similar fashion.
Multiple fluorescent localizations were carried out on the same transwell culture by fluorescence microscopy. Small rectangular strips of culture membrane (1–2x3–4 mm) were cut with spring scissors under a dissecting microscope and handled with a pair of fine forceps. The filter fragments were then incubated in small drops (12–20 µl) of the appropriate lectin or antibody solutions placed on a microscope slide on which hydrophobic rings had been drawn with a marker pen. The slides were placed on a black sheet of matt silicone rubber inside a humidified Petri dish. The drops were illuminated tangentially with a fibre optic light source so that the filter fragments were easy to identify. Between incubations the fragments were washed in 35 mm diameter Petri dishes containing 1.5 ml PBS and rinsed three times. The filter fragments were then placed in a drop of mountant and the nuclear fluorophore was examined to assess orientation of the filter fragment. If necessary the filter was then upturned before a coverslip was placed on top.
DBA–fluorescent isothiocyanate (FITC) binding to isolated gland fragments (n = 3) was also examined by conventional or confocal microscopy. Isolated glands were fixed in the same way as whole mounted tissue although processed in centrifuge tubes and pulse centrifuged on a microcentrifuge between incubations. The pellet was resuspended in ~10 µl PBS and added to a drop of mountant on a microscope slide and a coverslip placed on top. Individual gland fragments and whole mounted tissue were then examined by conventional epi-illumination or by confocal microscopy using a x63 oil immersion lens (LSM-1; Zeiss).
Fluorescence microscopy was used for dual localization studies. Initially it was confirmed that DBA–FITC (10 µg/ml) produced the same staining pattern as DBA–biotin (10 µg/ml) used with either avidin–FITC or anti-biotin FITC-conjugated antiserum. Nuclei were fluorescently labelled with bisbenzimide (10 µg/ml, Aldrich). Keratan sulphate was localized with a monoclonal antibody (diluted 1/40, clone 5D4; ICN, UK). MUC-1 core protein was examined with monoclonal antibodies HMFG1 (Serotec, diluted 1/20 or 1/40; Burchill et al., 1987) or BC3 (diluted 1/2500; Xing et al., 1989; a gift from P.Devine).
Electron microscopy
Specimens for transmission electron microscopy were examined with or without detection of DBA binding. Those in which the lectin was examined were pre-fixed in 2.5% (v/v) glutaraldehyde in 0.1 mol/l sodium cacodylate buffer pH 7.3 and lectin binding detected prior to embedding with biotinylated DBA (10 µg/ml, Sigma) in Tris-buffered saline (50 mmol/l Tris–HCl, 0.15 mol/l NaCl, 1 mmol/l CaCl2, pH 7.6). After incubation with avidin peroxidase (5 µg/ml) in a high ionic strength buffer (0.125 mol/l Tris/HCl, 0.347 mol/l NaCl pH 7.6) at room temperature for 1 h (Jones et al., 1987) sites of lectin binding were detected with DAB–H2O2 as previously described (Jones et al., 1987, 1998). After lectin binding the cultures were post-fixed in 1% osmium tetroxide in 0.05 mol/l cacodylate buffer pH 7.3, dehydrated and embedded in epoxy resin (Taab Laboratories Equipment Ltd, Aldermaston, UK). Sections (0.5 µm) were cut and stained with 1% (w/v) Toluidine Blue in 1% (w/v) borax and examined by light microscopy; thicker sections (2.5 µm) were examined unstained to detect DBA binding. Ultra-thin (pale gold) sections of suitable areas were cut and examined without further staining on a Philips 301 transmission electron microscope.
Filter fragments processed for scanning electron microscopy were fixed in 2.5% (v/v) glutaraldehyde in either Sorensen's buffer, post-fixed in 1% aqueous osmium tetroxide and dehydrated through a graded series of ethanol concentrations. The cultures were then incubated in Arklone to remove the ethanol (ICI, Runcorn, UK), critical point dried from liquid CO2, and coated with a very thin layer (~1–2 nm) of chromium in a coating unit fitted with a cryo-pump and planar magnetron sputter head (Auto 301; Edwards, Crawley, UK). The surface of the cultured cells was then examined at high resolution by field emission in-lens scanning microscopy (DS 130F; Topcon-ABT-ISI, Tokyo, Japan). Effects of culture variables were again monitored blind in replicate cultures.
Results
DBA binding to endometrial epithelium in vivo
DBA binding was almost completely undetectable in five proliferative phase tissues except for very small patches of staining on the epithelium. A similar pattern was observed in early secretory phase endometrium. From the mid-secretory phase, DBA-binding glycans were localized to the glandular epithelium and secretions, but were absent from the stroma as described previously (Aoki et al., 1989; Jones et al., 1998). Heterogeneity was a very striking feature in the secretory phase glandular epithelium (Figure 1). In some cases, DBA binding was either absent or largely restricted to the apical surface, while in other cells it was found in a punctate distribution in the apical cytoplasm, or more widely distributed. Glands in close proximity to one another varied widely in the amount of staining found within both epithelial cells and lumen. Staining was generally less abundant in menstrual phase specimens although, in a small proportion of glands, intense and widely distributed staining was still observed. In the luminal epithelium, staining tended to appear rather later in the cycle than in glands, being restricted to very few cells in the early secretory phase, patches of cells in mid-secretory and essentially continuous by late secretory phase (not shown). Experiments were also carried out in which whole-mounted glands or sheets of luminal epithelium were examined by conventional and confocal fluorescence microscopy (not shown), and heterogeneous expression of DBA binding sites was again observed within both cell populations.
|
MUC-1 core protein
After staining with mAb BC3, essentially all the cells in the epithelium, both glandular and luminal, were observed to be immunopositive both in the cytoplasm and at the apical cell surface (Figure 2A,B
). In contrast, staining with HMFG1 was highly heterogeneous (Figure 2C,D). Desialylation led to increased HMFG1 binding and a reduction in heterogeneity (not shown); however, the staining remained less extensive than that observed with BC3.
|
DBA binding, keratan sulphate and MUC-1 core protein localization to isolated gland fragments, explants and cell cultures
Heterogeneity in DBA binding was a major feature of freshly isolated mid-secretory phase gland fragments, explants, and outgrowing cells (Figure 3A–C
). The distribution of staining within outgrowths reflected that found in the explanted tissue. Differences were observed between colonies in both the proportion of cells stained by DBA and the intensity of signal from different cells in the population. In some cases, entire cell islands were stained, while in others very few positive cells were present. DBA localization in dense confluent areas was usually punctate and primarily on the cell surface. Examination of well-spread cells at the edges of epithelial islands, however, revealed punctate and elongated intracellular arrays characteristic of intracellular membrane-bounded compartments (not shown). Inspection of semi-thin plastic sections of the cultures (not shown) revealed intense staining on the apical surface of some cells while in others there was no detectable staining.
|
Many cells were stained in cultures established from secretory phase specimens while very few were present in those established from proliferative phase glands, reflecting the observations in vivo (Jones et al., 1998
). Set against a background of extensive heterogeneity between replicate wells, cells from the same specimens cultured with or without progesterone or MPA did not exhibit discernible differences. One further variable, the composition of substrate, was examined using wells coated with Matrigel or collagen, or a hydrated Matrigel layer. Cells that had grown out on hydrated Matrigel tended to destroy the substrate within a few days, leaving residual rounded mounds to which the cells were adherent. These areas and the surrounding outgrowth showed similar mosaicism to that seen on collagen-coated plastic or Matrigel-coated filters (Figure 3B–C
).
The distribution of DBA-binding glycans was compared to that of keratan sulphate (recognized by mAb 5D4) and MUC-1 (mAb HMFG1), using dual immunofluorescence. When used singly, each antibody produced a wide range of fluorescence intensity present largely at the cell surface. When DBA was combined with either 5D4 (Figure 3D,E) or HMFG1 (Figure 3H–J), the binding pattern of each marker overlapped but did not coincide; some cells expressed both markers, some cells expressed only one of each pair and others expressed neither. However, relatively few cells bound both DBA and 5D4. Staining with mAb BC3 showed that essentially all cells contained the MUC-1 core protein (not shown), suggesting that, as in tissue, much of the heterogeneity observed was at the level of glycosylation. Concanavalin A, used as a positive control to confirm the presence of glycoprotein glycans, reacted with all cells in tissue (both proliferative and secretory phase), explants and outgrowing epithelial islands.
Transmission and scanning electron microscopy of epithelial cells in vitro
Transmission electron microscopy (TEM) of epithelium cultured on transwell filters showed that the cells varied in morphology. Some cells were highly polarized (Figure 4A) whilst others were more elongated. When viewed at higher magnification, the glycocalyx on the apical surface of polarized or elongated microvillous cells could be observed (Figure 4B). An abundant glycocalyx was present on the microvilli and inter-microvillous membrane of some cells whilst on others it was thin or undetectable (Figure 4C). Localization of DBA binding sites obscured the glycocalyx but demonstrated binding mainly at the cell surface, both on microvilli and the intervening membrane, and strong variation between adjacent cells (Figure 4D). Weaker staining was present on subapical membrane-bounded compartments within the cultured cells. This distribution corresponds to the pattern of staining observed in semi-thin section by light microscopy. The surface of ciliated cells was not observed by TEM due to their infrequent occurrence.
|
High resolution scanning electron microscopy (SEM) permitted examination of a much larger area of the cell surface. Both microvillous and cilated cells were present in the cultures (Figure 5A
). Ciliated patches were observed on some cells; this appears to be an in-vitro artefact since the whole of the surface is ciliated in this type of cell in vivo. On the microvillous cells there was a great variation in the density of surface projections which bore no apparent relationship to the cell's apical surface area (Figure 5B). No differences could be detected between progestin-treated and untreated cultures. The glycocalyx present on cilia and microvilli was observed as collections of small knobs of ~20 nm in diameter (Figure 5C and D). These were found in variable amounts on the microvilli and on the intervening membrane (Figure 5E and G) but they were absent from some cells (Figures 5F). Comparison with transmission microscopy (Figure 4) suggested that the blebs represent apical glycocalyx and are formed by the collapse of a more extended structure at the time of critical point drying.
|
Discussion
Studies of endometrial ultrastructure demonstrated heterogeneity within the epithelium based on the presence of microvillous or ciliated cells (Masterson et al., 1975; Ferenczy, 1976a,b). Despite their relative morphological homogeneity in the light microscope, the cells also exhibited significant variations in the expression of glycan markers (Campbell et al., 1988; Aplin et al., 1994; 1998; Jones et al., 1998; Aplin, 1999). The relative proportion of these cell types, and their surface topology, varied with position in relation to gland openings, as well as with time of the cycle. Several markers have now been employed to analyse heterogeneity at the molecular level in the context of the change of phenotype that occurs in the secretory phase. The data demonstrate heterogeneity based on variations in glycosylation patterns in both the glandular and luminal epithelium (Aplin et al., 1998; Aplin, 1999). Glycan structures that show variable expression can be associated with both ciliated and microvillous cell surfaces (Campbell et al., 1988) suggesting ciliogenesis and glycan heterogeneity are not directly correlated.
During the secretory phase, glycogen accumulation occurs in the apical cytoplasm on days 4 and 5 after the luteinizing hormone surge (LH+4 and +5), accompanied by Golgi development and the appearance of glycoprotein-rich post-Golgi secretory vesicles (Dockery et al., 1988; Aoki et al., 1989; Smith et al., 1989; Aplin et al., 1994). Glycoproteins and glycogen are both released into gland lumens from where they diffuse into the lumen of the uterus (Hey et al., 1995). Considerable variation is evident in the secretory activity of different glands (Seif et al., 1989; Smith et al., 1989; Graham et al., 1994; Hey and Aplin, 1996; Aplin et al., 1998). However, it is not known whether this results from temporal variation in response across the epithelial population or a qualitatively and quantitatively different response in some glands.
Large variations in probe binding to individual cells have been observed in vivo with DBA (Aoki et al., 1989; Jones et al., 1998), mAbs 5D4 (Graham et al., 1994; Aplin et al. 1998) and HMFG1 (Hey et al., 1994) and other glycan-binding mAbs (Thor et al., 1987; Smith et al., 1989; Hey and Aplin, 1996) and lectins (reviewed in Aplin, 1991). Previous studies have demonstrated that 5D4-binding glycans are associated with endometrial MUC-1, while experiments using a capture-detect enzyme-linked immunosorbent assay (ELISA) to determine whether DBA-binding glycans might also be associated with MUC-1 gave a weak positive signal, suggesting that the sites are associated with MUC-1 but also probably with other glycoproteins present in the tissue (N.A.Hey, unpublished data).
Heterogeneity is most evident in distal components of epithelial oligosaccharides that are added during late stages of processing. Thus, for example, DBA binding requires terminal 
-GalNAc in the structure GalNAc1–3[Fuc1–2]Galß1–4GlcNAcß1- in which both GalNAc and Fuc are added as late Golgi modifications; mAb 5D4 recognizes a family of sulphated lactosaminoglycans in which sulphation is a late Golgi event (Mehmet et al., 1986); mAb HMFG1, which binds to a subset of the total MUC-1 core protein present in endometrial epithelium, is inhibited from binding by sialic acid (Burchill et al., 1987; Hey et al., 1994) which is added at a late processing stage; mAbs D9B1 (Campbell et al., 1988; Smith et al., 1989; Hoadley et al., 1990; Aplin et al., 1998) and B72.3 (Thor et al., 1987; Aplin et al., 1998), both of which have been previously reported as binding in a mosaic fashion, recognize structures containing terminal sialic acid. Extensive glycan mosaicism is also apparent in cell culture, where patterns of expression of several markers within cells and at the apical surface are approximately consistent with their abundance and cellular distribution at the time of biopsy. There is also very considerable variation in the amount of glycocalyx associated with the apical cell surface, an observation that is presumably related to the observed variation in glycan abundance.
In principle, intercellular variation in glycoprotein glycan could arise as a result of variations either in core protein abundance or post-translational events up to and including the final processing stages of glycosylation or sulphation. MUC-1 mRNA is, however, present at highest relative abundance in the secretory phase (Hey et al., 1994). mAb BC3 is much less sensitive than HMFG1 to variation of glycosylation. While we cannot exclude some quantitative intercellular variation in core protein abundance, relatively homogeneous binding of BC3 in endometrial epithelium in vitro and in vivo suggests that glycan heterogeneity is likely to arise at least in part by post-translational mechanisms.
The mosaic characteristic of the epithelial cell population has added degrees of complexity, since we have observed that at least four qualitatively different phenotypic combinations are present, as demonstrated by dual labelling with DBA and mAb 5D4. This is presumably a result of combinatorial expression (or action) of a repertoire of steroidally regulated glycosyl transferases. 5D4 sites appear earlier in the secretory phase than DBA sites (Graham et al., 1994; Jones et al., 1998), accounting for the relative scarcity of cells bearing both markers in culture. Both however,, are progesterone-dependent Aplin et al., 1997; Jones et al., 1998; unpublished data).
The lack of up-regulation by progestins in culture of glycans that are progesterone-dependent in vivo (Gemzell-Danielsson et al., 1994; 1996; Aplin et al., 1997; Jones et al., 1998) agrees with findings from several laboratories that commonly employed culture protocols do not conserve the hormonal responsiveness of endometrial epithelium (Satyswaroop et al., 1979; Kirk and Irwin, 1980; reviewed in Aplin, 1989). The use of matrix-coated porous filters can conserve steroid receptors (Classen-Linke et al., 1997), but mucin expression has not been reported in this model. An additional factor in monolayer culture is the loss of tissue architecture and paracrine signals from adjacent mesenchyme and vasculature (Koji et al., 1994). Thus the mechanisms giving rise to mosaicism of hormonally responsive components have not yet been elucidated. Nonetheless, glycan heterogeneity is conserved in vitro to an extent that should allow future examination of its biosynthetic basis and control.
Our results show intercellular variations sufficient to generate a readily observable mosaic pattern within endometrial glands. It is possible that this reflects intercellular variability in cellular functions, and has a role in tissue physiology. For example, by stimulating different cells in the population at different rates, the tissue may maintain the progesterone-dependent secretory response over a longer time window. Alternatively, different cells may contribute secretions of different composition. Glycoprotein- and glycogen-rich secretions contribute to the uterine luminal fluid (Hey et al., 1995) which provides the pre-implantation environment of the embryo. After implantation, maternal glands are invaded and both the cells and their secretions are phagocytosed by the expanding trophoblast (Hertig et al., 1956). MUC-1 has been shown to be anti-adhesive and so its co-expression at the cell surface with adhesion molecules is intriguing in relation to the regulation of embryo–epithelial interactions (Aplin et al., 1996; Aplin, 1997). Further studies are in progress to examine the relevance of mucin heterogeneity at the cell surface in interactions with the embryo (Aplin, 1997).
Acknowledgments
The authors are grateful to Wellbeing for financial support and to clinical staff who assisted with collection of tissue. We thank P.Devine for the gift of antibody BC3, F.Maclachlan and M.D.Jefferies for histopathological evaluation, and C.Daley for help with confocal microscopy. S.C. thanks M.Elstein for his support during the early stages of this work.
Notes
5 To whom correspondence should be addressed at Research Floor, St Mary's Hospital, Manchester M13 0JH, UK 
References
Aoki, D., Kawakami, H., Nozawa, S. et al. (1989) Differences in lectin binding patterns of normal human endometrium between proliferative and secretory phases. Histochemistry, 92, 177–184.[Web of Science][Medline]
Aplin, J.D. (1989) Cellular biochemistry of the endometrium. In Wynn, R.M. and Jollie W.P. (eds), Biology of the Uterus. Plenum Press, New York, USA, pp. 89–129.
Aplin, J.D. (1991) Glycans as biochemical markers of human endometrial secretory differentiation. J. Reprod. Fertil., 91, 525–541.
Aplin, J.D. (1994) Products of endometrial differentiation. In Chard, T. and Grudzinskas, J.G. (eds), The Uterus. Cambridge University Press, Cambridge, UK, pp. 125–147.
Aplin, J.D. (1997) Adhesion molecules in implantation. Rev. Reprod., 2, 84–93.[Abstract]
Aplin, J.D. (1999) MUC-1 glycosylation in endometrium is hormonally regulated and defines key differences between the luminal and glandular epithelium. Hum. Reprod., 14 (Suppl. 2), 17–25.
Aplin, J.D., Seif, M.W., Graham, R.A. et al. (1994) The endometrial cell surface and implantation: expression of the polymorphic mucin MUC-1 and adhesion molecules during the endometrial cycle. Ann. N.Y. Acad. Sci., 734, 103–121.[Web of Science][Medline]
Aplin, J.D. and Hey, N.A. (1995) MUC1 and embryo-endometrial interactions. Biochem. Soc. Trans., 23, 826–831.[Web of Science][Medline]
Aplin, J.D., Spanswick, C., Behzad, F. et al. (1996) Integrins ß5, ß3 and v in human and mouse endometrium: expression in stromal and glandular cells. Mol. Hum. Reprod., 2, 527–534.
Aplin, J.D., Jones, C.J.P., McGinlay, P.B. et al. (1997) Progesterone regulates glycosylation in endometrium. Biochem. Soc. Trans., 25, 1184–1187.[Web of Science][Medline]
Aplin, J.D., Hey, N.A. and Graham, R.A. (1998) Human endometrial MUC1 carries keratan sulphate: characteristic glycoforms in the luminal epithelium at receptivity. Glycobiology, 8, 269–276.
Burchell, J., Gendler, S., Taylor-Papadimitriou, J. et al. (1987) Development and characterisation of breast cancer reactive monoclonal antibodies directed to the core protein of the human milk mucin. Cancer Res., 47, 5476–5482.
Campbell, S., Seif, M.W., Aplin, J.D. et al. (1988) Expression of a secretory product by microvillous and ciliated cells of the human endometrial epithelium in vitro. Hum. Reprod., 3, 927–934.
Classen-Linke, I., Kusche, M., Knauthe, R. et al. (1997) Establishment of a human endometrial cell culture system and characterization of its polarized hormone responsive epithelial cells. Cell Tissue Res., 287, 171–185.[Web of Science][Medline]
Dockery, P., Li, T.C., Rogers, A.W. et al. (1988) The ultrastructure of the glandular epithelium in the timed endometrial biopsy. Hum. Reprod., 3, 826–834.
Ferenczy, A. and Riechart, R.M. (1973) Scanning and transmission electron microscopy of the human endometrial surface epithelium. J. Clin. Endocrinol. Metab., 36, 999–1008.
Ferenczy, A. (1976a) Studies on the cytodynamics of human endometrial regeneration I. Scanning electron microscopy. Am. J. Obs. Gynecol., 124, 64–74.[Web of Science][Medline]
Ferenczy, A. (1976b) Studies on the cytodynamics of human endometrial regeneration II. Transmission electron microscopy and histochemistry. Am. J. Obs. Gynecol., 124, 582–595.[Web of Science][Medline]
Gemzell-Danielsson, K., Svalander, P., Swahn, M.L. et al. (1994) Effects of a single post-ovulatory dose of RU486 on endometrial maturation in the implantation phase. Hum. Reprod., 9, 2398–2404.
Gemzell-Danielsson, K., Westlund, P., Johannisson, E. et al. (1996) Effect of low weekly doses of mifepristone on ovarian function and endometrial development. Hum. Reprod., 11, 256–264.
Graham,R.A., Li,T.C., Cooke,I.D. et al. (1994) Keratan sulphate as a secretory product of human endometrium: cyclic expression in normal women. Hum. Reprod., 9, 926–930.
Hertig, A.T., Rock, J. and Adams, J. (1956) A description of 34 human ova within the first 17 days of development. Am. J. Anat., 98, 435–494.
Hey, N.A., Graham, R.A., Seif, M.W. et al. (1994) The polymorphic epithelial mucin MUC1 in human endometrium is regulated with maximal expression in the implantation phase. J. Clin. Endocrin. Metab., 78, 337–342.[Abstract]
Hey, N.A., Li, T.C., Devine, P. et al. (1995) MUC 1 in secretory phase endometrium: expression in precisely dated biopsies and flushings from normal and recurrent miscarriage patients. Hum. Reprod., 10, 2655–2662.
Hey, N.A. and Aplin, J.D. (1996) Sialyl Lewis X and Sialyl Lewis a are expressed by human endometrial MUC1. Glycoconjugate J., 13, 769–779.[Web of Science][Medline]
Hoadley, M.E., Seif, M.W. and Aplin, J.D. (1990) Menstrual cycle-dependent expression of keratan sulphate in human endometrium. Biochem. J., 266, 757–763.[Web of Science][Medline]
Jansen, R.P.S. and Johannisson, E. (1985) Endocrine response in the female. In Shearman, R.P. (ed.), Clinical Reproductive Endocrinology. Churchill-Livingstone, Edinburgh, UK, pp. 109–164.
Johannisson, E., Parker, R.A. and Landgren, B.-E. (1982) Morphometric analysis of the human endometrium in relation to peripheral hormone levels. Fertil. Steril., 38, 564–571.[Web of Science][Medline]
Jones, C.J.P., Mosley, S.M., Jeffrey, I.J.M. et al. (1987) Elimination of the non-specific binding of avidin to tissue sections. Histochem. J., 19, 264–268.[Web of Science][Medline]
Jones, C.J.P., Fazleabas,A.T., McGinlay,P.B. et al. (1998) Cyclic modulation of epithelial glycosylation in human and baboon (Papio anubis) endometrium demonstrated by the binding of the agglutinin from Dolichos biflorus (DBA). Biol Reprod., 58, 20–27.
Kirk, D. and Irwin, J.C. (1980) Normal human endometrium in cell culture. Methods Cell Biol., 21B, 51–77.
Koji, T,. Chedid, M., Rubin, J.S. et al. (1994) Progesterone-dependent expression of keratinocyte growth factor mRNA in stromal cells of the primate endometrium: keratinocyte growth factor as a progestomedin. J. Cell Biol., 125, 393–401.
Li, T.C., Rogers, A.W., Dockery, P. et al. (1988) A new method of histological dating of human endometrium in the luteal phase. Fertil. Steril., 50, 52–60.[Web of Science][Medline]
Ludwig, H., Metzger, H. and Frauli, M. (1990) Endometrium: tissue remodelling and regeneration. In D'Arcangues, C., Fraser, I.S., Newton, J.R. and Odlind, V. (eds), Contraception and Mechanisms of Endometrial Bleeding. Cambridge University Press, Cambridge, UK, pp 441–466.
Masterson, R., Armstrong, E.M., and Moore, I.A.R. (1975) The cyclic variation in the percentage of ciliated cells in the normal human endometrium. J. Reprod. Fertil., 42, 537–540.
Markee, J.E. (1940) Menstruation in intraocular endometrial transplants in the Rhesus monkey. Contrib. Embryol. Carnegie Inst. Wash., 28, 219–308.
Mehmet, H., Scudder, P., Tang, P.W. et al. (1986) The antigenic determinants recognized by three monoclonal antibodies to keratan sulphate involve sulphated hepta- or larger oligosaccharides of the poly(N-acetyllactosamine) series. Eur. J. Biochem., 157, 385–391.[Web of Science][Medline]
Noyes, R.W., Hertig, A.T. and Rock, J. (1950) Dating the endometrial biopsy. Fertil. Steril., 1, 3–25.
Satyswaroop, R.G., Bresler, R.S., de la Pena, M.M. et al. (1979) Isolation and culture of human endometrial glands. J. Clin. Endocrinol. Metab., 48, 639–641.
Saleh, M.I., Warren, M.A., Li, T.C. et al. (1995) A light microscopical morphometric study of the luminal epithelium of human endometrium during the pre-implantation period. Hum. Reprod , 10, 1828–1832.
Seif, M.W., Aplin, J.D., Foden, L.J. et al. (1989) A novel approach for monitoring the endometrial cycle and detecting ovulation. Am. J. Obstet. Gynecol., 160, 357–362.[Web of Science][Medline]
Shiozawa, T., Tsukahara, Y., Nakayama, J. et al. (1991) Immunohistochemical reactivity of antikeratan sulphate monoclonal antibody 5D4 to various conditions of human endometrial tissues and its application as a useful marker for identifying endometrial epithelia. Gynecol. Obstet. Invest., 32, 229–242.
Smith, R.A., Seif, M.W., Rogers, A.W. et al. (1989) The endometrial cycle: the expression of a secretory component correlated with the luteinizing hormone peak. Hum. Reprod., 4, 236–242.
Thor, A., Vigliane, M.J., Muraro, R. et al. (1987) Monoclonal antibody B72.3 Reactivity with human endometrium: a study of normal and malignant tissues. Int. J. Gynaecol. Pathol., 6, 235–247.[Web of Science][Medline]
Xing, P.-X., Tjandra, J.J., Stacker, S.A. et al. (1989) Monoclonal antibodies reactive with mucin expressed in breast cancer. Immunol. Cell. Biol., 67, 183–195.
Submitted on June 2, 1999; accepted on October 13, 1999.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  M. G. Wahrenbrock and A. Varki Multiple Hepatic Receptors Cooperate to Eliminate Secretory Mucins Aberrantly Entering the Bloodstream: Are Circulating Cancer Mucins the "Tip of the Iceberg"? Cancer Res., February 15, 2006; 66(4): 2433 - 2441. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 E.C.I. Veerman, P.A.M. van den Keijbus, K. Nazmi, W. Vos, J.E. van der Wal, E. Bloemena, J.G.M. Bolscher, and A.V. N. Amerongen Distinct localization of MUC5B glycoforms in the human salivary glands Glycobiology, May 1, 2003; 13(5): 363 - 366. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||