Kreyberg's stain showed the presence of mucin within the lumina of the glands, indicating that they .... Kreyberg's stain for keratin and mucin (4,21) or antibodies ...
In VitroCell. Dev. Biol.26:636-642, June 1990 9 1990TissueCulture Association 0883-8364/90 $01.50+0.00
HUMAN ENDOMETRIAL C E L L S G R O W N O N AN E X T R A C E L L U L A R MATRIX FORM SIMPLE COLUMNAR EPITHELIA AND GLANDS TACEY E. K. WHITE, P. ANTHONY DI SANT'AGNESE, ANDRICHARD K. MILLER' Environmental Health Sciences Center, Departments of Obstetrics and Gynecology (T. E. K., R. K. M.), and Surgical Pathology (,4. di S'AL University of Rochester, Rochester, New York 14642
{Received 19 February. 1990; accepted 5 March 1990) SUMMARY Normal human endometrial cells were grown on an extracellular matrix containing type IV collagen, lain{Bin, heparan sulfate proteoglycan, and entactin (natrigel). On the extraceilular matrix, dispersed endometrial cells remained rounded, and aggregated to form mounds of cells, which continued to grow in this arrangement. At 10 d, light microscopy demonstrated that these mounds were comprised of an eosinophilic substance, containing individual fusiform stromal cells. About 50% of the mounds were covered with a single layer of polarized cuboidal to columnar cells with basal nuclei, whereas 60% contained columnar cells forming glandular structures with open lumina. These polarized cuboidal and columnar cells were epithelial, based on their positive staining for cytokeratins and the possession of microvilli, tonofilaments, abundant glycogen, ribosomes, and primitive junctional complexes. Kreyberg's stain showed the presence of mucin within the lumina of the glands, indicating that they were functional. Thus, human endometrial cells grown on an extracellular matrix form a simple cuboidal to columnar epithelium, a stromal component, and glandular structures, thereby mimicking the in vivo morphology of the endometrium. Key words: endometrium; in vitro; in vivo morphology; extracellular matrix; Matrigel.
INTRODUCTION
populations and maintenance of cultures for extended periods of time. Epithelial cells are better able to achieve a differentiated morphology and function when grown on an extracellular matrix iECM~ (1,7,8,17,18,35). Mouse mammary gland cells grown on floating collagen gels acquire a columnar morphology and secrete casein in response to hormones (1,7,8). Sengupta and coworkers {28) grew mouse endometrial cells on floating collagen gels where they showed columnar epithelial growth over a stromal core of cells; however, no gland formation was observed. Because m e s e n c h y m a l elements are necessary to induce epithelial gland formation in vitro i15), perhaps other E C M components, beside collagen, are necessary for the formation of endometrial glands and a fully differentiated state in vitro. In the present study, human endometrial cells were grown on an E C M containing laminin, type IV collagen, heparan sulfate proteogiycan, and entactin. In this environment, a simple columnar epithelium was formed, which covered a matrix containing individual fusiform cells resembling stromal endometrial cells and columnar cells forming glandular structures. Therefore, human endometrial cells grown on an E C M mimic the in vivo morphology of the human endometrium.
The functioning of the uterus in different phases of the menstrual cycle, during the peri-implantation period, and during pregnancy is not yet fully understood. This is particularly true in the human, where the opportunity for in vivo study is limited. Valuable information has been gained about processes such as growth and hormonal responsiveness of cell populations in the uterus through the use of primary cell culture systems for endometrium from the human (3,6,10,20,25,27,30,31,33) and from other mammalian species (9,11,19,24). However, the majority of these studies utilize a monolayer culture, which does not maintain an in vivo-type morphology for endometrial cells and therefore has only a limited application for studying higher level functions of the uterus, such as decidualization and implantation. Organ cultures of endometrium have the advantage of preserving the original cytoarchitecture of the cells, as well as their differentiated state i22,23). However, the contribution of different cell populations to a particular process cannot be discerned by such systems, and these cultures are limited by the amount of time that they can be successfully maintained in culture (usually 14 d or less) (7). An ideal culture system for the endometrium would provide for the production of an in vivo-like morphology, while still allowing for manipulation of different cell
MATERIALS AND METHODS ' To whom correspondence should be addressed at Department of Obstetrics and Gynecology, Box 668, University of Rochester Medical Center, 601 Elmwood Avenue, Rochester, NY 14642.
Tissue preparation. Within 3 h of surgery samples of human uterine tissue were obtained from nonpregnant
636
ENDOMETRIAL CELLS CULTURED ON ECM
hysterectomy specimens, from regions that showed grossly normal endometrium. These specimens were made available through the departments of Obstetrics and Gynecology and Surgical Pathology of Strong Memorial Hospital (University of Rochester Medical Center) and from Highland Hospital. All patients were less than 50 yr of age at the time of hysterectomy. Use of human hysterectomy specimens was approved by the Human Investigations Committee of the University of Rochester Medical Center and Highland Hospital, and patient permission was granted through a signed consent form. Only tissues from the menstrual or proliferative phases of the menstrual cycle were used in the present study. Tissue samples were placed endometrial side down in dispase (Collaborative Research, Bedford, MA) [10 caseinolytic U / m l in Hanks' balanced salt solution, (GIBCO, Grand Island, NY)], and incubated either for 1 h at 37 ~ C or overnight at 4 ~ C. Pieces of the softened endometrium were then pulled away from the underlying myometrium, incubated for 10 to 15 min in a phosphate buffered, 0.25% trypsin (GIBCO) solution (100 mg/ml E D T A , Ca ++- and Mg§247 at 37 ~ C, and triturated with Pasteur pipettes (with decreasing bore sizes) to achieve a suspension that contained single cells pins cell and tissue clumps. To obtain a nearly single cell suspension, the cells were filtered through a 200-#m stainless steel mesh, and 105- and 37- /~m nylon meshes. This filtering resulted in a suspension that was comprised of single cells and small cell clumps (usually less than 10 cells). Endometrial cell cultures. Matrigel (~0.15 ml) (Collaborative Research), an E C M material containing type IV collagen, laminin, heparan sulfate proteoglycan, and entactin, was layered, undiluted, onto the transparent Biopore membrane (0.4-~m pore size) of a 12-mmdiameter Millicell-CM culture dish insert (Millipore Corp., Bedford, MA). The inserts were then placed into 24-well culture dishes (Costar, Cambridge, MA), and incubated at 37 ~ C, in an humid environment, for 30 min. Endometrial ceils (1.0 X 10~), which were a mixed population of stromal cells, lining, and glandular epithelial cells, and bone marrow derived cells were seeded onto the Matrigel and maintained in a 37 ~ C humidified incubator (5% CO~:95% air) in R P M I 1640 medium (GIBCO), supplemented with 10% fetal bovine serum, 1% glutamine, 1% penicillin-streptomycin, 1% sodium bicarbonate, and ITS-Premix supplement tinsulin, transferrin, selenium, Collaborative Research). The culture medium was changed every 2 to 3 d, and the cultures were usually maintained for 10 to 30 d. All cultures were viewed using a Nikon E L W D 0.3 inverted, phase contrast microscope, and photographed using a Nikon-UFX camera attachment. Histology. To examine the cells in situ under the light microscope, cultures were fixed with a 4% formaldehyde (or paraformaldehyde), 1% glutaraldehyde solution, pH 7.3, for a t least 1 h. The membranes (supporting Matrigel and cells) were cut out with a scalpel, dehydrated in a series of ethanol and xylene, and embedded in paraffin.
637
Four-micrometer sections were cut, placed on slides, heated in a 60 ~ C oven overnight, and rehydrated in a series of xylene, ethanol, and water. Every fourth slide was stained with hematoxylin and eosin, and additional selected slides were stained with a modification of Kreyberg's stain for keratin and mucin (4,21) or antibodies against cytokeratins. Slides were viewed under a Leitz Wetzlar microscope and photographed using a Leitz camera attachment. Immunohistochemistry. F o r immunohistochemistry, slides were prepared as described for light microscopy. Mter rehydration, all endogenous peroxidase activity was blocked by incubating slides in 3% H~O2 for 5 to 10 min. Antigenic sites were unmasked by digesting sections at 37 ~ C in a 0.1% pepsin (Sigma, St. Louis, MO) solution ~in 0.01 N HCI) for 15 to 20 min. A monoclonal antibody against all acidic and basic cytokeratins (AEI:AE3, Boehringer-Mannheim, Indianapolis, IN) was applied to the slides (1:1000 dilution), which were then incubated in a moist environment at 37 ~ C for 30 rain. A streptavidinbiotin secondary antibody system (Zymed, San Francisco, CA), with aminoethylcarbazole as the chromagen, was employed to visualize the antigenic sites.
FIG. 1. Endometrial cells in culture as viewed under the inverted microscope, a, In monolayer culture, cells grow flat, and epithelial cells acquire either a pancake or a whorled morphology ~ 1 4 d in culture); b, on extracellular matrix IECMI, cells remain rounded, aggregate, and grow in mounds (26 d in culture). )< 104.
638
WHITE ET AL.
Transmission electron microscopy. A "pop-off" technique (5) was employed for transmission electron microscopy ITEM), which allowed visualization of thick sections under light microscopy, with subsequent thin sectioning of areas of interest for ultrastructural observation. Briefly, 1 to 2.5-/am sections of cells on E C M embedded in Spurr epoxy resin were placed on slides, stained with fuchsin-methylene blue, and observed under the light microscope. Areas of interest were "popped off" using a size 3 Beem capsule, thin-sectioned, stained with acetate lead citrate, and examined using an Hitachi HS-8 electron microscope. R ESULTS
Cell growth. When human endometrial cells were grown in plastic dishes, they flattened out and grew as characteristic monolayers iFig. 1 a). At no time did monolayers form three-dimensional arrangements or differentiated structures. On the other hand, when human endometrial cells were grown on the E C M they did not flatten out, but rather the cells remained rounded and aggregated to form mounds of cells. The sizes of the aggregated mounds within a culture were inversely proportional to the number of mounds formed. This varied among cultures, but there were usually between 100 and 300 mounds per 12-mm culture insert, ranging in size from 100 to 1000 /am in length, and ranging in shape from round to long and thin. As growth of the cells continued, the mounds enlarged and appeared as nodules comprised of rounded cells, as observed under the inverted phase contrast microscope (Fig. 1 b). Frequently, cellular projections were observed beneath and extending away from the nodules. These projections were similar to stromal cell projections observed in mixed monolayer cultures of endometrial cells. The cells grown on E C M could be kept in culture for up to 30 d. However, after 2 to 3 wk in culture, rounded cells or entire nodules seemed to slough off the E C M or, in some cases, monolayer outgrowth from nodules was observed. Therefore, light and electron microscopy were performed when cells had been grown in E C M culture for about 10 d. Histology. Cross sections of endometrial cells grown on E C M revealed the presence of eosinophilic mounds of material on top of the E C M (Fig. 2). The mounds seemed to be composed of a matrixlike substance, which was quite different in appearance to the original ECM, being an amorphous arrangement of material rather than the regular parallel fibers of the Matrigel. These mounds contained cross sections of individual fusiform cells, which had the morphology of endometrial stromal cells. Mounds were never evidenced on acellular E C M and always contained cross-sections of stromal-type cells. Occasionally, cross sections of individual fusiform ceils were seen within the E C M layer itself, suggesting that those cells had the ability to migrate into the ECM. About 52% of the mounds (range: 25 to 80%) per dish were covered with a single layer of cuboidal to columnar cells, with central or basal nuclei, resembling epithelial cells from the endometrium lFig. 2). Within 60% of the mounds lrange: 28 to 83%) per dish, columnar cells
formed circular arrangements with open lumina, resembling endometrial glands (Fig. 4). In some cultures, especially those showing a small number of very long mounds, longer glandular structures were observed, which appeared to be branched and convoluted. In these cultures, glands could be up to 500 /am in length, as judged by their appearance in serial sections. Immunohistochemistry. Immunohistochemical staining with antibodies against cytokeratins (Fig. 3) uniformly showed the presence of these intermediate filaments in all columnar cells in both the epithelial layer and the glandular structures, again confirming the epithelial nature of these cells. Cells within mounds possessing a stromal morphology did not routinely stain positively for the presence of cytokeratins. However, some individual cells within mounds stained very weakly with the antibody against cytokeratins. Kreyberg's stain. To investigate the functional status of the cells grown on E C M , sections of endometrial ceI1-ECM cultures were stained with Kreyberg's stain, which stains mucin or mucopolysaccharides blue. Mounds of eosinophilic material and cell bodies did not stain blue for the presence of mucin; however, the material within the lumina of the glandular structures did (Fig. 4), indicating the presence of mucin, and suggesting that the glands were indeed functional. Transmission electron microscopy. A portion of surface columnar epithelium which had formed an involuted structure was popped off and selected for T E M (Fig. 5 a). These surface cells appeared to be well polarized columnar cells, with prominent microvilli on their apical aspect. Their cytoplasm contained tonofilaments, abundant amounts of glycogen and polyribosome rosettes, and moderate amounts of golgi and mitochondria. Between cells, on their lateral edge, primitive junctional complexes were observed which were comprised of a zonula occludens ltight junction) and a trailing zonula adherens. Also, the cells were arranged in what appeared to be a glandular orientation. These observations are consistent
FIG. 2. Cross sections of endometrial cells growing on ECM, showing eosinophilic mounds containing fusilorm cells (arrowhead), and covered by a simple cuboidal to columnar epithelium (arrow). Hematoxylin-eosin stain; )
