TY - JOUR
AU - Van Blerkom, Jonathan
AB - Abstract Ovarian follicular granulosa cells express temporally and spatially distinct functions throughout the follicle cycle. During the entire cycle, granulosa cells exhibit an unusually broad range of activities including the secretion of steroid hormones, enzymes, growth factors and cytokines. To date, the identity(ies) of these cells (lineage/cell type) remains unknown. We demonstrate expression of the Tie, Tek, cKit, Flt-1, CD-31 and vWF proteins and the ability to rapidly internalize acetylated low density lipoprotein among mural and cumulus subpopulations of human and murine follicular granulosa cells. In addition, we provide evidence that human and murine granulosa cells can engage in tube-forming activity in vitro. To the best of our knowledge, the six phenotypic and two functional markers examined during this study, as a group, are associated only with endothelial or endothelial-like cells. In total, the findings suggest that some granulosa cells may have the potential to actively participate in the vascularization of the corpus luteum, by way of an inherent capacity which is likely to be a characteristic of their unique identity and lineage. This inherent capacity of granulosa cells to behave and respond, at least to some extent, like endothelial cells may be of possible importance in the aetiology of certain follicular pathologies. capillary-like structures, corpus luteum, endothelial cell granulosa cells, vasculogenesis Introduction Ovarian follicular granulosa cells synthesize and secrete a wide variety of growth factors and cytokines. Many of these proteins are potent effector molecules of endothelial function(s), the majority are also known to be produced by endothelial cells. Granulosa cells have also been shown to produce endotheliotropic chemoattractants (Katz et al., 1992) and mitogens (Rose and Koos, 1988), plasminogen activators (Karakji and Tsang, 1995) and their corresponding inhibitors (Piquette et al., 1993), metalloproteinases (collagenases and gelatinases) (Aston et al., 1996), anticoagulant heparin sulphate proteoglycan (Hosseini et al., 1996), angiotensin I-converting enzyme (ACE) (Daud et al., 1990) and thrombospondin (Dreyfus et al., 1992). These proteins are expressed by endothelial cells and compromised or damaged tissues and represent cellular activities that promote and facilitate angiogenesis. In addition, bovine follicular granulosa cells have been shown to rapidly internalize acetylated low density lipoprotein (AcLDL) (Spanel-Borowski and Ricken, 1997), an activity that is frequently used to identify cells as endothelial. Taken together these findings indicate that a potent inductive capacity develops within the maturing follicle, based on the endothelial-like secretions of the enclosed granulosa. It is widely held that one of the primary purposes for this capacity is the promotion of cellular and angiogenic events which are external to the granulosa cells of the follicle but which are, nonetheless, required for normal follicle growth, ovulation, and corpus luteum formation. For the preovulatory follicle, the production of angiogenic factors by granulosa cells may influence the degree of expansion of the perifollicular capillary bed and, as a consequence, increase the levels of intrafollicular oxygen. Following ovulation in healthy young women, factors produced by the granulosa cells may stimulate angiogenesis within the vessels of the theca interna and direct both their entry and participation in the formation of the corpus luteum. However, the angiogenic potential associated with the follicle, as a result of the synthetic activities of the granulosa cells, may affect the vascular behaviour(s) of cells within, as well as outside, the bounds of the follicle's basal lamina. Specifically, this angiogenic potential may, in part, elicit and support the expression of specialized vascular functions among subpopulations of the granulosa cells themselves. Some of these vascular functions may be of particular importance during the early stages of corpus luteum formation. Here, we provide evidence that human and murine follicular granulosa cells express a unique set of phenotypic (Tie, Tek, cKit, Flt-1, CD-31, vWF proteins) and functional (rapid AcLDL uptake and tube-forming ability in vitro) markers which, taken together, characterize them as specialized, endothelial-like cell populations. Materials and methods Isolation of human cumulus and mural granulosa cells, murine cumulus granulosa cells and human and murine cumulus cell-oocyte complexes Human granulosa cells were recovered from women undergoing IVF procedures after ovarian stimulation and ovulation induction (Van Blerkom et al., 1995). Cultures of human cumulus granulosa cells were obtained from cumulus cell-oocyte complexes (COC), following their dissociation from the oocyte. Cultures of human mural granulosa cells were established using cells recovered from the walls of preovulatory follicles punctured during oocyte retrieval procedures. Cell suspensions of human mural and cumulus granulosa cells were: (i) plated onto uncoated, sterile glass coverslips 1-3 days after initial recovery for immunofluorescent analyses, (ii) seeded into dishes, in some cases coated with specialized substrate material(s), for the analysis of tube-forming potential (see below), or (iii) were utilized in 35S-labelling experiments in preparation for immunoprecipitation and Western analysis (cumulus granulosa cells only). For immunofluorescent analyses, cells were routinely grown for an additional 1–4 days on glass coverslips, prior to fixation. During this culture period, cells were maintained in Dulbecco's modified Eagle's medium (DMEM) low glucose medium (1 g/l) (Gibco BRL, Grand Island, NY, USA) supplemented with 10% fetal calf serum (Research Sera, Summit Biotechnology, Fort Collins, CO, USA) and 10 μg/ml gentamicin (Gibco BRL). Human cumulus and mural granulosa cells grown on coverslips were fixed and permeabilized as previously described (Antczak and Van Blerkom, 1997b). On a few occasions, excess human COC were donated by patients undergoing gamete intra-Fallopian transfer procedures. These supernumerary complexes were fixed intact, shortly after recovery, in a phosphate-buffered saline (PBS) solution containing 3.7% formaldehyde (pH 7.3) for 1–2 h (depending on the size of the complex) at room temperature. Fixed human COC were processed further, as described below, in preparation for immunostaining. Murine cumulus granulosa cells studied as part of freshly isolated COC were derived from immature follicles and mature preovulatory and ovulated follicles. Granulosa precursor cells were analysed as part of freshly isolated intact primordial follicles derived from the ovaries of 1 week old mice. Freshly isolated murine COC, at all stages, and freshly isolated, intact primordial follicles were fixed in a PBS solution containing 3.7% formaldehyde (pH 7.3) for 1 h at room temperature. These fixed specimens were processed further, as described below, in preparation for immunostaining. In most cases, human COC and preovulatory and ovulated murine COC were exposed to a limited hyaluronidase (Sigma Biochemicals, St Louis, MO, USA) digestion, either prior to or following fixation, in order to `trim' away excessive layers of outlying cumulus granulosa cells and expose underlying follicle cells more proximal to the oocyte. No differences in immunoreactivity were detected in specimens exposed to hyaluronidase either prior to or following formaldehyde fixation. Fixed human and murine COC and the intact follicles of 1 week old mice were permeabilized in a PBS solution containing 0.1% Triton X-100 (Sigma) and 0.1% NP40 (Sigma) for 1–2 h at room temperature, prior to antibody analysis. Permeabilized complexes and intact follicles were washed in PBS containing 1% bovine serum albumin (BSA) and stored in the same solution for up to 1 week at 4°C, prior to immunofluorescent analysis. The protocols used for processing paraffin sections have been described previously (Antczak and Van Blerkom, 1997a). Immunofluorescent analysis Processed human granulosa cells grown on coverslips, freshly isolated human and murine COC, murine intact follicles and paraffin sections were examined for their reactivity towards the following primary antibodies, in accordance with previously described protocols (Antczak and Van Blerkom, 1997a,b): cKit (C-19, 4 μg/ml), Flt-1 (C-17, 2 μg/ml), Tie (C-18, 1–2 μg/ml), Tek (C-20, 1–2 μg/ml) (all from Santa Cruz Biotechnology, Inc., Santa Cruz, CA, USA) and vWF [1:200 dilution (human); 1:100 (mouse)] (Dako Corporation, Carpenteria, CA, USA). In addition, murine COC were examined for their reactivity towards an antibody directed against the murine CD-31 protein [PECAM-1 (M20, 6 μg/ml) Santa Cruz Biotechnology]. Following reaction with primary antibodies, samples were exposed to secondary antibody solutions containing a 1:100 dilution of a goat anti-rabbit fluorescein isothiocyanate (FITC) conjugate, a 1:200 dilution of rabbit anti-goat FITC conjugate or, in the case of the analysis of paraffin sections of mouse ovaries, a 1:100 dilution of a goat anti-rabbit biotin conjugate (all from Sigma), as appropriate, for 2 h at 4°C. To confirm the specificity of the cKit, Flt-1, CD-31, Tie and Tek antibodies utilized during this investigation, solutions containing these antibodies were preincubated with their corresponding immunizing peptides (the peptides used to generate the antibodies) (cKit, SC-168P; Flt-1, SC-316P; CD-31, SC-506P; Tie, SC-342P; Tek, SC-324P; Santa Cruz Biotechnology) at a concentration 20-fold (Flt-1 and CD-31 blocking peptide analysis associated with murine COC) or 40-fold (all other blocking peptide analyses) that were used for the primary antibody, for 2 h at 37°C. Following this preincubation period, human cumulus granulosa cells or murine COC were reacted with each of these `blocked' antibody solutions as well as standard non-`blocked' antibody solutions and examined for their immunoreactivity. Following incubation with `blocked' and non-`blocked' primary antibody solutions, samples were reacted with the appropriate secondary antibody solutions (see above) and examined as described below. During the analysis of the vWF protein, positive [human umbilical vein endothelial cells (HUVEC)] and negative [human fibroblasts (HF)] cellular control populations were utilized to confirm the specificity of the antibody, at the concentrations used. To control for non-specific antibody interactions and the levels of background immunofluorescence associated with each group of specimens examined, antibody control analyses were conducted using non-specific rabbit or goat IgG primary antibodies (SC-2027 and SC-2028; Santa Cruz Biotechnology), in accordance with previously described protocols (Antczak and Van Blerkom, 1997a,b). During each of the fluorescent analyses described above, results were determined using a standard epifluorescent microscope and/or a scanning laser confocal microscope with filter sets optimized for the detection of FITC- or Texas Red-associated fluorescence (paraffin sections). The Tie, Tek, cKit, Flt-1 and CD-31 antibodies utilized during these studies were selected because each was developed using a peptide sequence representing a cytoplasmic, C-terminal region of their respective molecules. These antibodies should not recognize secreted, soluble forms of the full-length molecules to which they are targeted if those secreted, soluble forms lack their cytoplasmic, C-terminal domains. Acetylated low density lipoprotein uptake To examine the ability of human granulosa cells to rapidly internalize acetylated low density lipoprotein (AcLDL), cultures of human cumulus and mural granulosa cells grown on coverslips were incubated in standard growth medium (see above) supplemented with 10 μg/ml DiI (1,1'-dioctadecyl-3,3,3,3′-tetramethylindo-carbocyanine perchlorate)-AcLDL (Biomedical Technologies, Inc., Stoughton, MA, USA) for a period of 4–8 h, at 37°C, in a 5% CO2 incubator. To determine the ability of human and murine cumulus granulosa cells to rapidly internalize DiI-AcLDL while still part of freshly isolated COC, these complexes were incubated in standard growth medium (see above) supplemented with 10 μg/ml DiI-AcLDL for 4 h, at 37°C, in a 5% CO2 incubator. During these analyses, positive cellular control cultures of HUVEC and murine haemangio-endothelioma cells (EOMA) and negative cellular control cultures of normal HF, all grown on glass coverslips, were exposed to the same DiI-AcLDL solutions, for the same periods of time, as those utilized during the examination of populations of human cumulus granulosa cells. At the end of the incubation period, specimens were washed and fixed in preparation for fluorescent analysis, following procedures recommended by the manufacturer (Biomedical Technologies, Inc.). Results were determined through the use of a standard epifluorescent microscope and/or a scanning laser confocal microscope using filter sets optimized for the detection of rhodamine-associated fluorescence. Induction of endothelial cell-like behaviour and analysis of resulting structures For the development of tubular tracts within cultures of human or murine cumulus granulosa grown on tissue culture plastic, cells were left undisturbed (no subculturing) and maintained at high cell density, with infrequent media changes (~1 per week), for extended periods of time (2–4 weeks). For human granulosa cells, this activity was supported in the presence of growth medium consisting of medium M-199 (Gibco BRL) supplemented with 10% fetal calf serum (FCS; Gemini Bio-products, Calabasas, CA, USA), 0.7 BSA [fraction V (cell culture tested); Sigma)], 0.2 mmol/l 2-mercaptoethanol (BME; Sigma), 5 mmol/l reduced glutathione (Sigma) and 25 μg/ml gentamicin. For murine cumulus granulosa cells, this activity was supported in DMEM (high glucose)/F-10 (50/50) (Gibco BRL) supplemented with 10% FCS, and 25 μg/ml gentamicin. The nature of the tubular tracts that formed under these conditions was more closely examined by cross-section analysis. Following formaldehyde fixation of the cell monolayer in its original dish, the sample was embedded in a solution of LR White embedding medium (London Resin Company, Basingstoke, Hampshire, UK), in accordance with the manufacturer's recommended protocols. Semi-thin sections of the embedded cell monolayer were cut on a microtome, stained in a toluidine blue solution (Sigma) then analysed and photographed under microscopic examination. The development of tubular cellular tracts in cultures of human cumulus granulosa cells grown on collagen gels was supported by a growth medium consisting of medium M-199 (Gibco BRL) supplemented with 10% FCS, 10% human follicular fluid (0.22 μm filtered), 0.7% BSA, 0.2 mmol/l BME, 5 mmol/l reduced glutathione and 25 μg/ml gentamicin. Collagen gels were prepared in accordance with a previously described protocols (Montesano et al., 1983). For the detection of nuclei within the cells forming these structures, the cultures were incubated in solution of 4,6-diamidino-2-phenylindole (DAPI) (2.5 μg/ml, final concentration), prepared in the growth medium described above, and maintained for 15 min at 37°C. After DAPI staining, cultures were washed three times in a solution of medium M-199 supplemented with 10% FCS, prior to analysis and photography during microscopic examination. Western blot analysis Immunoprecipitated antibody complexes prepared from cultures of 35S-labelled human and murine cumulus granulosa cells were examined by Western analysis for the presence of the vWF, CD-31, Flt-1, Tie and Tek proteins. For the analysis of vWF, CD-31, Flt-1, Tie and Tek proteins, human and murine cumulus granulosa cells were grown for 24–48 h in the presence of methionine-free α-minimum essential medium (αMEM) medium (Irvine Scientific Inc., Santa Ana, CA, USA) supplemented with 10% FCS, 0.0 to 2.0 U/ml human chorionic gonadotrophin (HCG), 10 μg/ml gentamicin and 100 μCi/ml [35S] methionine. Immunoprecipitation of the vWF, CD-31, Flt-1, Tie and Tek proteins was conducted using a protocol obtained from Santa Cruz Biotechnology, with several modifications. In brief, at the end of the labelling period, cell cultures were washed three times with unsupplemented α-MEM at room temperature then lysed in ~1 ml of a PBS solution containing 1% NP-40, 0.5% sodium deoxycholate, 0.1% SDS (RIPA buffer), supplemented with 100 μg/ml phenylmethylsulphonylfluoride and 150 mU/ml aprotinin (all reagents from Sigma) for 10 min at 4°C. Cell lysates were then triturated through a 26 G needle attached to a 1 ml syringe to completely disrupt all cells and shear genomic DNA. Cell lysates were centrifuged at 600 g for 10 min to pellet cellular debri. Lysates were transferred to fresh 1.5 ml microcentrifuge tubes, then precleared by the addition of 1.0 μg of either normal rabbit (vWF, Flt-1, Tie and Tek) or goat (CD-31) IgG for 1 h at 4°C, with rocking. Following incubation with normal IgG, 20 μl protein A-agarose (vWF, Flt-1, Tie and Tek) or 20 μl of protein G-agarose (CD-31) conjugate was added to each lysate solution to form complexes with the previously added non-specific immunoglobulin. Lysates were incubated with the appropriate agarose conjugates for an additional 1 h at 4°C with rocking motion. Non-specific immune complexes were removed from the cell lysates following centrifugation for 5 min at 450 g. Lysate supernatants were transferred to clean microcentrifuge tubes and centrifuged two additional times as previously described to ensure complete removal of nonspecific immune complex. Cleared lysates were then reacted with specific antibodies using either a 1:500 dilution (vWF) or a final antibody concentration of 0.2–0.5 μg/ml (CD-31, Flt-1, Tie and Tek) for 1–2 h at 4°C, with rocking. In all instances, the antibodies used for Western analyses were the same as those used during immunofluorescent analysis (see above for specific details and sourcing). Following reaction with specific antibodies, cell lysates were reacted with 20 μl of the appropriate agarose conjugate (Santa Cruz Biotechnology, Inc.) for 1–2 h at 4°C, for isolation of the specific immune complex. Specific agarose-immune complexes were recovered from cleared lysates by centrifugation for 5 min at 450 g. The agarose-immune complexes were then washed and centrifuged four times (as above) with 1.0 ml of RIPA buffer containing protease inhibitors (prepared as above) to remove extraneous proteins and debri. Washed immune complexes were resuspended in 50 μl of lysis buffer containing 125 mmol/l Tris-Cl (pH 6.8), 20% glycerol, 4% sodium dodecyl sulphate (SDS), 1% 2-BME, 100 mmol/l dithiothreitol (DTT) and 10 μg/ml bromophenol blue. Following resuspension, immune complexes were either frozen down at –80°C for later analysis or boiled for 5 min prior to loading onto 5.0% or 7.5% SDS-polyacrylamide gel electrophoresis (PAGE) gels for electrophoretic separation. During each analysis protein molecular weight standards were included to enable sizing of identified protein bands. Following electrophoresis, proteins were transferred to PVDF membranes (Bio-Rad, Hercules, CA, USA) in preparation for immunodetection. Blotted membranes were probed with specific antibody solutions directed against the vWF (1:2000 dilution), CD-31 (1:300–500 dilution), Flt-1 (1–300 dilution) Tie and Tek (1:200–300 dilutions) proteins and the proteins were visualized in conjunction with an amplified alkaline phosphatase immunoblot assay kit (Bio-Rad), following the protocols provided by the manufacturer. For detection of the CD-31 protein, filters were reacted with a 1:3000 dilution of a donkey anti-goat biotinylated antibody (Chemicon International, Inc., Temecula, CA, USA) in association with the previously mentioned immunoblot assay kit. Results The findings described below were obtained from the analysis of cumulus granulosa cells derived from 38 IVF patients and the mural granulosa cells derived from 26 IVF patients; seven cultures from both these groups represented cumulus and mural granulosa cells derived from the same patients. In addition, the cumulus granulosa cells associated with 23 human and 313 murine, intact, freshly isolated COC were also examined. All human cumulus granulosa cells examined during this investigation were derived from COC similar to that shown in Figure 1A, complexes that contained minimal red blood cell contamination. During the analysis of cells in their original in-vivo context (COC), outer cells of human and murine complexes, those beyond the limits of the broken white circle shown in Figure 1A (white arrow), were removed by limited hyaluronidase treatment. This enzymatic `trimming' procedure ensured efficient penetration of antibodies or DiI-AcLDL to follicle cells most proximal to the oocyte. To prevent the presentation of redundant information in this report, murine results are cited but not shown, with the exception of those results that provide novel contributions to the analyses. Internalization of AcLDL Cultures of dispersed human cumulus (n = 16) and mural granulosa cells (n = 4) and freshly isolated, intact, human (n = 2) and murine COC (n = 7) were examined for their ability to rapidly internalize DiI-AcLDL. In each culture or freshly isolated COC examined, human and murine granulosa cells stained brightly positive, and at high frequency (60 to >80%), for this activity. The remainder of the population stained to a lesser extent. Representative images of AcLDL uptake by human cumulus granulosa cells examined as part of a freshly isolated COC and as dispersed human cumulus and mural granulosa cells grown in culture are shown in Figure 1B (1B2, white light image of 1B), 1C and 1D respectively. Positively stained human and murine follicle cells showed a cytoplasmic distribution of numerous brightly stained points of punctile fluorescence, the majority of which were clustered around the nucleus of each cell (Figure 1C, D). During these analyses, cultures of HUVEC (Figure 1E) and a murine endothelial cell line, EOMA (Obeso et al., 1990) served as positive cellular controls. Uptake of DiI-AcLDL by these positive control cultures displayed a pattern of staining similar to that observed in cumulus granulosa cells (compare results in Figure 1C and D with those shown in Figure 1E). Cultures of normal human fibroblasts (HF), which served as negative cellular controls during these analyses, failed to show rapid internalization of DiI-AcLDL [Figure 1F (1F2, white light image of 1F)], under the conditions of the assay. Zona-enclosed oocytes present in each of the human and murine COC examined failed to demonstrate rapid DiI-AcLDL internalization (Oc, Figure 1B). Tie and Tek immunodetection Freshly isolated human COC and dispersed cultures of human cumulus and mural granulosa contained cells that stained positively for the Tie [COC (n = 2), Figure 1G, H (H is an internal portion of 1G); cumulus granulosa cultures (n = 19), Figure 1I; mural granulosa cultures (n = 12), Figure 1I2] and Tek [COC (n = 2), Figure 1K; cumulus granulosa cultures (n = 18), Figure 1L; mural granulosa cultures (n = 12), Figure 1L2] proteins. Freshly isolated murine COC also contained cells which stained positively for the Tie (n = 41) and Tek (n = 44) proteins. Approximately 50–70% of the cells in dispersed cultures of human cumulus and mural granulosa stained positively for the Tie and Tek proteins, Figure 1I, I2 and L, L2, respectively, following 1–7 days of in-vitro culture. A similar percentage of positively stained cells was associated with human and murine COC examined for the presence of the Tie and Tek proteins by immunofluorescent detection. Poorly or unstained cell populations could also be identified in each of these specimens. These variably stained subpopulations are readily apparent in the images of human COC stained for the Tie and Tek proteins as shown in Figure 1G and K. In these figures, black asterisks are used to denote regions with brightly stained cells and white asterisks are used to denote regions with poorly or unstained cells. Tie and Tek immunofluorescence was associated with cells of the cumulus oophorus and corona radiata in both the human and mouse, as illustrated in Figure 1G and K, in the human. In the human COC stained for Tek shown in Figure 1K; slender, brightly stained, cellular processes can be seen extending from cells of the corona radiata into the zona pellucida (white arrowheads). During each analysis, a portion of the human and murine samples examined were assayed for levels of associated non-specific antibody interaction and background fluorescence (antibody control analyses). Representative results from these analyses are shown in Figure 1N, for human COC; Figure 1O, for human cumulus granulosa; and Figure 1O2, for human mural granulosa cells. Only low-level, non-specific antibody interactions and background fluorescence were detected in each case (Figures 1O and 1O2 depict antibody control results associated with the analysis of the Tie, Tek, cKit, Flt-1 and vWF proteins in human cumulus and mural granulosa cells, respectively). A result obtained following immunodetection of the Tek protein associated with freshly isolated follicles derived from the ovaries of mice at 1 week of age is shown in Figure 1P, this image represents an internal region of the specimen. A major portion of the follicle precursor cells associated with the primordial follicle shown in Figure 1P, stained brightly for the Tek protein (white arrows), while a relatively minor portion stained poorly (white asterisk). The level of non-specific antibody interaction and background fluorescence associated with intact follicles derived from 1 week old mice, during antibody control analyses, is shown in Figure 1T2. Only signals of low intensity were detected. Immunofluorescent analysis of the Tie and Tek proteins was also conducted on paraffin sections prepared from murine ovaries recovered 11 h post HCG, from mice undergoing an ovulation induction regimen. During these analyses, individual sections of the same preovulatory follicle were immunostained for the Tek protein and non-specific control antibodies. Results from these studies clearly showed positive Tek immunoreactivity among the cells of the cumulus oophorus and corona radiata (Figure 1Q). In Figure 1Q, white arrows are used to direct attention towards the thin cellular processes extending from cells of the corona radiata into the zona pellucida; these cellular processes also stained positively for the Tek protein. Results obtained following immunodetection of the Tie protein in paraffin sections prepared from murine ovaries 11 h post HCG are shown in Figure 1R and S. In Figure 1R, bright Tie-specific immunofluorescence is almost exclusively associated with the four individual follicles present within the specimen (white arrows). Figure 1S shows the distribution of the Tie protein among the granulosa cells of an early antral follicle as well as the distribution of the protein within the neighbouring thecal and stromal tissues along one side of the same follicle. Bright Tie-specific immunofluorescence was almost exclusively associated with the cells located within individual follicles [Figure 1R, S (black asterisk)]. In Figure 1S, it should be noted that these positively stained cells are located both within the cumulus granulosa and the mural granulosa (black asterisk). In all paraffin sections examined, although fine points of Tieand Tek-specific immuno-fluorescence were scattered throughout the extrafollicular tissues, in general, thecal and ovarian stromal tissues stained poorly for these proteins (see white asterisks, Figure 1R, S). The level of non-specific antibody interaction and background fluorescence associated with paraffin sections of murine ovaries, during antibody control analyses, is shown in Figure 1T. Virtually no signal was detected, under the conditions of the analysis. To confirm the specificity of the antibodies used for Tie and Tek immunodetection, these antibodies were preincubated with blocking peptides (the peptide sequences used to generate the antibodies) prior to the analysis of populations of human and murine granulosa cells. Blocking peptide preincubation eliminated signal detection from the Tie and Tek antibodies in dispersed cultures of human cumulus granulosa cells (Tie, 1J; Tek 1M) and in cumulus granulosa cells associated with murine COC. cKit and Flt-1 immunodetection Freshly isolated human COC and dispersed cultures of human cumulus and mural granulosa contained cells which stained positively for cKit [COC (n = 3), Figure 2A; cumulus granulosa cultures (n = 7), Figure 2B; mural granulosa cultures (n = 7), Figure 2B2] and Flt-1 [COC (n = 3), Figure 2D; cumulus granulosa cultures (n = 15), Figure 2E; mural granulosa cultures (n = 17), Figure 2E2] proteins. Freshly isolated murine COC also contained cells which stained positively for the cKit (n = 38) and Flt-1 (n = 32) proteins. In Figure 2A, corona radiata cells, positioned directly on the surface of the zona pellucida (ZP), were stained brightly (white arrows) for cKit protein, following immunofluorescent analysis of the freshly isolated human COC. The anchoring process of one of these cells is clearly visible and stained positively for cKit protein (left arrow, Figure 2A). In dispersed cultures of human cumulus and mural granulosa, ~50–70% of the cells stained positively for the cKit protein; examples of these human results are shown in Figure 2B and 2B2, respectively. A roughly equivalent percentage of the granulosa cells associated with freshly isolated human and murine COC also stained positively for cKit protein. In the human COC immunostained for the Flt-1 protein and shown in Figure 2D, follicle cells in the vicinity of the zona pellucida (ZP) and oocyte (Oc) stained brightly for the Flt-1 protein (white arrows). On the basis of their relative positions with respect to the oocyte and zona pellucida, these Flt-1 positive follicle cells are derived from both the corona radiata and cumulus oophorus (Figure 2D). In general, 50-70% of the cultured human cumulus (Figure 2E) and mural granulosa cells (Figure 2E2) and the human (Figure 2D) and murine cumulus granulosa cells associated with freshly isolated COC stained positively for Flt-1 protein. A portion of the human and murine samples allotted for examination with antibodies directed against the cKit and Flt-1 proteins were analysed for levels of associated non-specific antibody interaction and background fluorescence. Representative results from these antibody control analyses are shown in Figure 2M, for human COC; Figure 1O, for human cumulus granulosa; and Figure 1O2, for human mural granulosa cells. Only low-level, non-specific antibody interaction and background fluorescence were detected in each case. To confirm the specificity of the antibodies used for cKit and Flt-1 immunodetection in the human granulosa cells and murine COC examined, these antibodies were preincubated with blocking peptides prior to the analysis of populations of human and murine granulosa cells. Blocking peptide preincubation eliminated signal detection from the cKit and Flt-1 antibodies in cultures of human cumulus granulosa cells (cKit, Figure 2C; Flt-1, Figure 2C2) and the cumulus granulosa cells associated with murine COC. Preliminary examinations of murine COC using dual staining analysis suggested that, in general, the subpopulations of cells which stained positively for the Tek protein were also those which stained positively for Tie, Flt-1 and cKit proteins. During these analyses the non-stained cell subpopulation did not appear to be reactive to any of the endothelial cell-related antibodies examined as opposed to representing a subpopulation which was reactive to some but not all of those antibodies. vWF and CD-31 immunodetection Freshly isolated human COC (n = 7) (Figure 2F, G) and dispersed cultures of human cumulus (n = 16) (Figure 2H) and mural granulosa (n = 14) (Figure 2H2) contained cells which stained positively for the vWF protein. Freshly isolated murine COC also contained cells which stained positively for the vWF protein (n = 9) and, in addition, the CD-31 protein (n = 54) (Figure 2K, L). Representative examples of each of these results are described below. CD-31 immunodetection was not conducted with human COC or granulosa cells. In the human COC stained for the vWF protein shown in Figure 2F, numerous brightly stained cumulus granulosa cells are seen in close proximity to the zona pellucida and the oocyte. In the human COC immunostained for the vWF protein shown in Figure 2G, a complex which includes a GV-stage oocyte (GV), the brightly stained body of a corona radiata cell can be seen projecting from the surface of the zona pellucida (ZP) (white arrow A). A cellular process extending from the body of this cell into the zona pellucida also stained brightly for the vWF protein (white arrow B, Figure 2G). As shown in Figure 2G-H2, the vWF protein was located within the cytoplasm and excluded from the nucleus of positively stained granulosa cells. In general, ~50–70% of the granulosa cells in cultured human cumulus and mural granulosa cells as well as those associated with human and murine COC stained positively for the vWF protein by immunofluorescent detection. During vWF protein analysis, cultures of HUVEC and HF were also examined for the presence of the vWF protein by immunodetection and served as positive and negative cellular controls, respectively. Figure 2I and J shows representative results obtained following immunofluorescent analysis of the vWF protein in HUVEC and HF. Human endothelial cells showed a pattern of vWF protein staining similar to that detected in cultures of human cumulus and mural granulosa cells, primarily a distribution of the protein throughout the cytoplasm of the cell (Figure 2I). However, some of the vWF-specific staining observed in cultures of HUVEC was associated with rod-like structures, Weibel Palade bodies (Weibel and Palade, 1964), which were not detected in cultures of human granulosa cells, either because these structures do not exist in human granulosa cells or they were too small to be easily detected. Cultures of human fibroblasts immunostained for the vWF protein failed to show any vWF-associated immunofluorescent signal (Figure 2J). Figure 2J2 shows the corresponding white light image of the culture shown in Figure 2J. Representative images depicting the levels of non-specific antibody interaction and background fluorescence associated with the analysis of the vWF protein in human COC, human cumulus granulosa, human mural granulosa and HUVEC cells, during antibody control analyses, are shown in Figures 2M, 1O, 1O2 and 2I2 respectively. Virtually no signal was detected, under the conditions of the analysis. Results obtained following immunofluorescent analysis of the CD-31 protein among granulosa cells associated with murine COC are shown in Figure 2K and L: a portion of the granulosa cells positioned directly on the surface of the zona pellucida, cells of the corona radiata, as well as those positioned one or two cell layers beyond, cells of the cumulus oophorus, stained brightly for the CD-31 protein (black asterisks). Other similarly positioned subpopulations of these cells stained poorly for the CD-31 protein (white asterisks, Figure 2K, L). In the GV-stage (GV) murine COC shown in Figure 2L, the majority of the cells associated with one hemisphere of the specimen were physically removed just prior to sample fixation (white arrow). The broken cellular processes that remained in this region, embedded in the zona pellucida, stained positively for the CD-31 protein (white arrow, Figure 2L). Roughly 50–70% of the cumulus granulosa cells associated with murine COC examined stained positively for the CD-31 protein. A representative image showing the levels of non-specific antibody interaction and background fluorescence associated with the analysis of the CD-31 protein in murine COC (antibody control analyses) is shown in Figure 2N. Only signals of very low intensity were detected. Blocking peptide preincubation of the CD-31 antibody prior to analysis of the CD-31 protein in granulosa cells associated with murine COC virtually eliminated a detectable immunofluorescent signal, as shown in Figure 2L2. Results obtained during these blocking peptide analyses confirmed the specificity of the CD-31 antibody utilized during the study. The formation of tube- and plexus-like structures by follicle cells grown in vitro As shown in Figure 3A (black arrows) each culture of ovulated murine cumulus granulosa cells maintained on basal growth medium for 2–3 weeks, without passage, consistently formed tubular structures when grown on tissue culture plastic. Cultures of human cumulus granulosa cells maintained at high density, on basal growth medium, for several weeks without passage, occasionally developed branched, anastomosed, tubular tracts which coursed throughout the monolayer. Examples of a human cumulus granulosa cell culture engaged in this activity are shown in Figure 3B1 (black arrowheads) and 3B2 (black arrows). Some of these tubular tracts extended for several millimetres in length; the portion of the tubular tract shown in Figure 3B1 is ~450 μm long (black arrowheads). To confirm that the tubular tracts observed in these cultures were capillary-like in nature, portions of the cultures shown in Figure 3B1 and B2 were examined in cross section. Following these analyses, the tubular regions observed were seen to consist of hollow, membrane bound voids insinuated between neighbouring cells of the monolayer (black arrows, Figure 3C1 and C2). Near the walls of the well containing this high density human cumulus granulosa cell culture grown on plastic, individual cells organized to form a net-like cellular formation reminiscent of a plexus as shown in Figure 3D1 (black arrows). Similar, more mature, structures consisting of a network of interconnecting tubular tracts were positioned at other locations along the walls of the same dish, as shown in Figure 3D2 [white arrow (because of the 3-dimensional nature of this tubular network only a relatively small portion was in focus within a particular plane of examination)]. The formations in Figure 3D1 and D2 were largely suspended and free of substrate attachment with the exception of cells anchored to the walls of the dish on one side and cells firmly attached to the base of the dish on the opposing side. To further examine the ability of human granulosa cells to display tube-forming ability and to analyse these capabilities under conditions designed to more closely parallel those experienced by these cells in vivo, human cumulus and mural granulosa cells were grown on collagen gels or Matrigel in the presence of 10% follicular fluid. Under these conditions, cumulus and mural granulosa cells associated to form tubular structures of greater girth than those formed by monolayer cultures grown on plastic and they did so more quickly (~14 days). Representative examples of these tubular structures, formed by human cumulus granulosa cells grown on the surface of collagen gels (white asterisk), are shown in Figure 3E (arrows, white and black). The nuclei of the individual cells engaged in the formation of these tubular structures were visualized by DAPI staining and subsequent examination under epifluorescent illumination (Figure 3F, white arrows). Media supplementation with 10% follicular fluid was required to support tube formation in low-density human granulosa cell cultures. Western analysis of vWF, CD-31, Flt-1, Tie and Tek proteins in human and murine cumulus granulosa cells Representative results obtained following the analysis of the vWF protein in FCS (lane A1), 35S-labelled cell lysates of human cumulus granulosa cells (lane A3) and the conditioned medium from those cells (lane A5) are shown in Figure 4A. More specifically, an aliquot of standard growth medium supplemented with 10% FCS, prior to use in cell culture, was run in lane 4A1 without immunoprecipitation. A single protein species consistent with the 220–225 kDa secreted form of the bovine vWF protein (Lynch et al., 1983a) was visualized following immunodetection. Autoradiographic analysis of the results from lane 4A1 detected no radioactive bands, as expected (lane 4A2). Results obtained following Western analysis of vWF-specific protein immunoprecipitates derived from lysates prepared from 35S-labelled human cumulus granulosa cells are shown in lane 4A3. A comparison of the results obtained in lane 4A3 and that obtained following autoradiographic analysis of those results, lane 4A4, identified three 35S-labelled protein bands. The largest of these vWF-specific bands in lane 4A3 is projected to be ~275 kDa and is consistent with the size expected for the pre-propeptide form of the vWF protein (top arrow, lane 4A3) (Bonthron et al., 1986). The broad band immediately beneath the 275 kDa protein (4A3) is likely to represent two specific proteins which have failed to adequately resolve on the gel, a protein of ~240 kDa (middle arrow, lane 4A3) and a smaller protein of ~220–225 kDa (bottom arrow, lane 4A3). The 240 kDa protein, putatively identified as the propeptide form of the vWF protein (Lynch et al., 1983b), at the upper edge of the broad band in lane 4A3, was found to be radioactive following autoradiographic analysis (bottom arrow, lane 4A4). The 220–225 kDa protein associated with the broad band in lane 4A3 (bottom arrow) is likely to be derived, at least in part, from the secreted form of the vWF protein present in the FCS used to supplement the growth medium. This 225 kDa protein was not synthesized by the cumulus granulosa cells during their incubation in the [35S]methionine-containing medium (no corresponding band in lane 4A4). The smallest of the radioactive bands (asterisk, lane 4A4) identified in cumulus granulosa cell lysates has a molecular weight of ~190 kDa and is probably a proteolytic peptide originating from the higher molecular weight forms of the protein [asterisk, lane 4A3 (see below)]. Results obtained following the analysis of immunoprecipitates of vWF-specific proteins derived from the conditioned medium of the cells whose lysates were analysed in lane 4A3 are shown in lane 4A5. Weak Western bands corresponding to the 260–275 kDa pre-pro (upper arrow, lane 4A5), 240 kDa pro (middle arrow, lane 4A5) as well as a more prominent band of ~225 kDa, consistent with the secreted form of the human vWF protein (bottom arrow, lane 4A5) (Bonthron et al., 1986), were detected. Each of these proteins was radioactively labelled, following autoradiographic analysis (arrows, lane 4A6). Proteolytic fragments of the vWF protein have been detected in the plasma of normal individuals of approximate sizes 189, 176 and 140 kDa (Zimmerman et al., 1986). Levels of protease inhibitors in the immunoprecipitation buffers used during the analyses may have been inadequate to completely arrest activities of proteolytic proteins released from the cumulus granulosa cells and/or present in the whole cell lysates. These proteolytic proteins may be responsible for the ~190 kDa protein appearing in lanes 4A3, 4A4 and 4A5 (asterisks) and the ~170 kDa protein appearing in lanes 4A3 and 4A5 (closed circles). All vWF results shown were derived from individual samples run on the same SDS–PAGE gel or subsequent autoradiographic analysis of the individual filter strips derived from that gel. Representative results obtained following immunoprecipitation and Western analysis of radiolabelled human and murine granulosa cell lysates for the CD-31, Flt-1, Tek and Tie proteins are shown in Figure 4, lanes B1-E2, respectively. Specific results obtained from these analyses are as follows: (i) Figure 4, lane B1, human and murine cumulus granulosa cell lysates examined for the CD-31 protein, murine results shown; (ii) Figure 4, lane, C1, human cumulus granulosa cell lysates examined for the Flt-1 protein; (iii) Figure 4, lane D1 murine cumulus granulosa cell lysates examined for the Tek protein; (iv) Figure 4, lane E1, murine cumulus granulosa cell lysates examined for Tie protein. Bands of appropriately sized proteins were obtained in each case and those bands were identified as radioactive, following autoradiographic analysis, as shown for: (i) CD-31, ~130 kDa, (Xie and Muller, 1993), arrow, lane 4B1; radioactive band, arrow, lane 4B2; (ii) Flt-1, ~ 180 kDa (Seetharam et al., 1995), arrow, lane 4C1; radioactive band, arrow, lane 4C2; (iii) Tek, ~140 kDa (Ziegler et al., 1993), arrow, lane 4D1; radioactive band, arrow, lane 4D2; (iv) Tie, ~130 kDa (Hashiyama et al., 1996), arrow, lane 4E1; radioactive band, arrow, lane 4E2. Discussion The endothelial-like character of follicle cells This study was designed to test the hypothesis that follicular granulosa cells might represent a specialized endothelial-like cell population. Results derived from immunofluorescent analysis and the examination of rapid AcLDL uptake in these cells demonstrated the following points. (i) High numbers of human and murine cumulus and mural granulosa cells (~50 to >80%, combined total) stained positively for each of the markers examined (Tie, Tek, cKit, Flt-1, CD-31, vWF and rapid AcLDL uptake). (ii) In human and murine COC, cells of the cumulus oophorus and corona radiata stained positively for each of the markers examined. In most cases, thin processes extending from cells of the corona radiata were stained positively and embedded in the surface of the zona pellucida. This particular relationship, displayed by immunopositive cells among the innermost layer of follicular granulosa cells, is a unique and defining characteristic of the cells of the corona radiata (Van Blerkom and Motta, 1979). It is not the behaviour of an extrafollicular (outside the basal lamina of the follicle) cell contaminant. (iii) In paraffin sections of murine ovary, the granulosa cells (mural and cumulus granulosa) within individual, intact follicles stained brightly for the Tie and Tek proteins; in general, thecal and ovarian stromal tissues stained poorly for both proteins. (iv) Granulosa cells cultured for limited periods of time in vitro or those examined as part of freshly isolated COC yielded similar results, demonstrating that results obtained following the analysis of granulosa cells grown in vitro were not artefacts of cell culture. The veracity of each of these immunofluorescent results was confirmed by the following. (i) The antibodies used recognized only the C-terminal, cytoplasmic domains of the receptor (Tie, Tek, cKit, Flt-1) or adhesion molecules examined (CD-31), virtually eliminating the possibility that the results obtained were a consequence of the detection of soluble molecular variants. (ii) Blocking peptide analyses (Tie, Tek, cKit, Flt-1 and CD-31) or the analysis of positive and negative cellular control populations (rapid AcLDL uptake and vWF) were used to verify the specificity of each of the antibodies or activities investigated. (iii) Western analyses were conducted using 35S-labelled cell lysates [Tie, Tek, Flt-1, CD-31, vWF (cell synthesis of cKit, previously demonstrated by Tanikawa et al., 1998, at the protein and RNA levels)], which demonstrated both the presence and synthesis of the proteins assayed. (iv) The examination of non-specific antibody interactions and the levels of background fluorescence associated with each antibody and group of samples provided additional support for the authenticity of each result. During immunofluorescent analyses, a portion of the granulosa cells associated with each sample failed to stain positively for each of the proteins examined. Two possible explanations for these results, representing opposite ends of the spectrum, are: (i) the nonstained cells had not yet differentiated to express endothelial-like features, or (ii) the non-immunoreactive cells had differentiated towards another, as yet unknown, lineage/cell identity. We are currently attempting to discriminate between these and other potential explanations for the results obtained. While one or several of the markers examined during this investigation might be associated with populations of cells other than endothelial cells, to the best of our knowledge, only endothelial cells or endothelial-like cells are associated with each of the eight markers analysed. The `purity' of preovulatory granulosa cell populations We recognize that the human and murine granulosa cells analysed during this investigation, and in particular the cultures of human mural granulosa cells examined, may have been contaminated to a low extent with cells of extrafollicular origin. For example, in 1991, Beckmann et al. determined that 15–17% of the cells associated with freshly isolated preparations of human ovarian granulosa cells were derived from white cell contaminants introduced as a result of follicle puncture during oocyte retrieval. These contaminants were found to consist of <2% polymorphonuclear leukocytes, ~6% macrophage/monocytes and ~10% lymphocytes (Beckmann et al., 1991). Similar results were obtained during the analyses of macrophages associated with freshly isolated human granulosa cells (Machelon et al., 1995), or white cell contributions to the total number of cells recovered from the follicular fluids of IVF patients (Loukides et al., 1990; Baranao et al., 1995). However, it is implausible that the majority of the data generated during this study was the result of this contamination for the following reasons. (i) 50% to >80% of the cells associated with each of the specimens examined stained positively for the markers analysed. (ii) White cells do not stain positively for all the markers analysed, in particular the vWF protein [only megakaryocytes and endothelial cells are known to express the vWF protein (Ruggeri and Ware, 1993)], nor are they known to be capable of forming tubular structures in vitro. (iii) If freshly isolated human and murine complexes were contaminated with white cell or other extrafollicular cell populations as a result of follicle puncture, those putative contaminants should have been located only on the periphery of the complexes at the time of recovery and fixation. If present initially, these contaminants should have been removed from the many freshly isolated human and murine complexes enzymatically `trimmed' to expose granulosa cells more proximal to the oocyte. On the contrary, positive fluorescent signals for each of the markers examined were associated with cells at all locations in multilayered COC, and fluorescent signal was not eliminated following the enzymatic `trimming' of specimens. (iv) In all cases where cells of the corona radiata were visible, positive immunofluorescent signals were associated with these cells; frequently, their positively stained, extended cellular processes were seen embedded in the zona pellucida. (v) Results obtained during the analysis of the Tie and Tek proteins in paraffin sections of murine ovary demonstrated that the majority of the immunofluorescent signal associated with these proteins was located among the granulosa cells of individual, intact follicles. In comparison, thecal and ovarian stromal tissues stained poorly for these proteins. (vi) The Tek protein was found associated with pregranulosa cells in murine primordial follicles. (vii) Adequate amounts of the Tie, Tek, Flt-1, CD-31 and vWF proteins were associated with cultures of human and murine granulosa cells to enable visualization of these proteins by Western analysis in the absence of immunoprecipitation (data not shown). (viii) During examination of human granulosa cells for expression of the VCAM protein by immunofluorescent analysis, <5% of the cells in cultures of mural granulosa and none of the cells in cultures of cumulus granulosa were found to stain positively (unpublished observation). VCAM, a marker of cytokine-stimulated endothelial cells (Haraldsen et al., 1996) and a protein identified on human ovarian microvascular endothelial cells, including those derived from thecal tissues (Ratcliffe et al., 1999), was analysed to address possible thecal endothelial cell contamination of human granulosa cell cultures. Indications that follicle cells may actively participate in the vascularization of the developing corpus luteum Not only were human and murine cumulus granulosa cells capable of engaging in tube-forming activity, when maintained at high cell density, they were able to do so in the absence of exogenous factor addition, with the exception of those factors present in the FCS (10%). Structures similar to those shown here can be formed by high density cultures of bona fide human endothelial cells [compare Figures 3B1 and B2, C1 and 2, and D1 and D2 in this report, with Figures 1a-d, 4a and 3a in Folkman and Haudenschild (1980) respectively], in roughly the same period of time (~3 weeks). However, these bona fide endothelial cells required substantial growth factor supplementation, including 5.4 mg/ml endothelial cell growth supplement (ECGS) (as described by Folkman and Haudenschild, 1980), to support this activity. This is an important distinction. It demonstrates not only that granulosa cells are capable of engaging in the formation of tubular tracts or structures under the appropriate vascular influences in vitro, but also that granulosa cells are capable of providing those putative vascular influences under the conditions of culture described here. Results obtained during this investigation demonstrated that a substantial percentage of follicle cells possess both steroidogenic and endothelial cell-like characteristics/potentials. These results included those derived during the analysis of: (i) rapid AcLDL uptake (a marker for steroidogenic activity (Chen and Menon, 1993) and endothelial character (Voyta et al., 1984), (ii) expression of the Tie, Tek and Flt-1 (Tie and Tek, Dumont et al., 1993; Flt-1, Peters et al., 1993) proteins, and (iii) tube-forming ability in vitro. Based on all of these results, and in particular the ability of follicle cells to form tube-like structures in vitro, we propose that a portion of the steroidogenic cells of the follicle might be capable of some form of de-novo vessel formation in vivo. Following the manifestation of both of these capabilities, vasculogenic/steroidogenic granulosa cells would then be positioned to secrete the products of their steroidogenic activities directly into the lumens of the very same vessels to which they contribute. The products of this granulosa cell-based vasculogenic activity might then integrate with incoming vessels from the thecal tissues, in the vicinity of the follicular rent provided by ovulation. The intermingling and association of granulosa and thecal endothelial cells during early formation of the corpus luteum may partially explain the finding that the corpus luteum is composed of distinct subpopulations of endothelial cells (Spanel-Borowski and van der Bosch, 1990). The integration of granulosa cell vasculogenic activity and thecal endothelial cell angiogenic activity may establish the conduit required for the systemic release of the essential hormones produced by the follicle cells of the corpus luteum, hormones designed to support the events of early pregnancy. Presence of the Tek protein among pregranulosa cells of primordial follicles Currently, we do not know when the cells of the follicular granulosa first acquire each of the characteristics/properties attributed to them during the present investigation. We do know, however, that the Tie, Tek, cKit, Flt-1, CD-31 and vWF proteins and the ability to rapidly internalize AcLDL were associated with granulosa cells derived from both growing (containing GV-stage oocytes) and mature follicles. During the analysis of intact follicles derived from the ovaries of 1 week old mice, the Tek protein was identified among the pregranulosa cells of primordial follicles. This particular time interval is relatively close to the developmental period in which ovarian follicles are formed in the mouse. In this regard it is important to note that the Tek protein has been postulated to mark embryonic progenitors of mature endothelial cells (Dumont et al., 1992). Additional studies are required to determine precisely when expression of the Tie, Tek, cKit, Flt-1, CD-31 and vWF proteins first begins among the granulosa cells of the follicle or their predecessors. Potential clinical relevance of the endothelial cell-like character of granulosa cells The endothelial-like character of follicle cells may underlie specific forms of follicular pathology, among them diseases involving the failure to develop a fully functional corpus luteum. Some forms of follicular pathology may result from an inability of granulosa cells to produce or respond to specific vascular elements (effector and/or receptor molecules) at particular times or in particular locations during development of the follicle. Over-expression of overt endothelial cell-like behaviour(s) by the granulosa may result in other forms of follicular disease, e.g. ovarian hyperstimulation syndrome. If abnormalities in the endothelial-like character of granulosa cells are the underlying cause in certain forms of follicular disease, additional studies may suggest particular strategies for intervention and successful remediation. Careful consideration of the endothelial-like behaviours of follicular granulosa cells may provide fundamental insights into the mechanics and operational strategies that underlie various ovarian follicular functions, at all points along its development. Additional efforts are required to further support and substantiate the possibility that follicle cells may directly participate in corpus luteum formation by way of their expression of an inherent vasculogenic potential. For example, efforts to determine whether human cumulus granulosa cells and bona fide human endothelial cells can successfully interact to form hybrid tubular structures are currently in progress. These efforts, when accomplished, may further substantiate and provide an additional basis and rationale for the endothelial-like nature of follicle cells, as presented in this report. Figure 1. View largeDownload slide Representative photomicrographs showing: observations on human cumulus-oocyte complexes (COC) (A), analysis of DiI-acetylated low density lipoprotein uptake (B, B2, C-F2), and immunofluorescent analysis of the proteins Tie (G-J, R, S) and Tek (K-M,P, Q) and controls (N, O, O2, T, T2). Analyses were performed in cultures of human cumulus granulosa cells (C, I, J, L, M, O), cultures of human mural granulosa cells (D, I2, L2, O2), cultures of positive control human umbilical cord endothelial cells (E), cultures of negative control human fibroblast cells (F, F2), freshly isolated human COC (A-B2, G, H, K, N), intact primordial follicles derived from the ovaries of mice at one week of age (P, T2), and sections of murine ovaries recovered 11 h post human chorionic gonadotrophin in a superovulation cycle (Q, R, S, T). Images were created by scanning laser confocal microscopy (G, H, K, N, P, Q, T2), conventional white light microscopy (A, B2, F2) or standard epifluorescent microscopic images (B-F, I-J, J, L, M, O, O2, R-T). Loss of specific protein recognition is shown for the Tie (J) and Tek (M) antibodies following preincubation with blocking peptides. Figures N, O, O2, T and T2 show representative images of antibody control analyses in human COC, cultured human cumulus granulosa cells, cultured human mural granulosa cells, paraffin sections of mouse ovaries and intact murine primordial follicles, respectively. Oc = oocyte; black asterisk = brightly stained granulosa cells; white asterisk = poorly stained cells. Figure 1. View largeDownload slide Representative photomicrographs showing: observations on human cumulus-oocyte complexes (COC) (A), analysis of DiI-acetylated low density lipoprotein uptake (B, B2, C-F2), and immunofluorescent analysis of the proteins Tie (G-J, R, S) and Tek (K-M,P, Q) and controls (N, O, O2, T, T2). Analyses were performed in cultures of human cumulus granulosa cells (C, I, J, L, M, O), cultures of human mural granulosa cells (D, I2, L2, O2), cultures of positive control human umbilical cord endothelial cells (E), cultures of negative control human fibroblast cells (F, F2), freshly isolated human COC (A-B2, G, H, K, N), intact primordial follicles derived from the ovaries of mice at one week of age (P, T2), and sections of murine ovaries recovered 11 h post human chorionic gonadotrophin in a superovulation cycle (Q, R, S, T). Images were created by scanning laser confocal microscopy (G, H, K, N, P, Q, T2), conventional white light microscopy (A, B2, F2) or standard epifluorescent microscopic images (B-F, I-J, J, L, M, O, O2, R-T). Loss of specific protein recognition is shown for the Tie (J) and Tek (M) antibodies following preincubation with blocking peptides. Figures N, O, O2, T and T2 show representative images of antibody control analyses in human COC, cultured human cumulus granulosa cells, cultured human mural granulosa cells, paraffin sections of mouse ovaries and intact murine primordial follicles, respectively. Oc = oocyte; black asterisk = brightly stained granulosa cells; white asterisk = poorly stained cells. Figure 2. View largeDownload slide Representative photomicrographs showing: analysis of cKit (A-C), Flt-1 (C2, D-E2), von Willebrand factor (F-J2) and CD-31(K-L2) proteins and controls (M, N) in freshly isolated human cumulus-oocyte complexes (COC) (A, D, F, G, M), cultures of human cumulus granulosa cells (B, C-C2, E), cultures of human mural granulosa (B2, E2, H2), cultures of positive control human umbilical cord endothelial cells (HUVEC) (I-I2), cultures of negative control human fibroblast cells (J-J2) and freshly isolated murine COC (K-L2, N), following respective analysis. Images were created by scanning laser confocal microscopy (A, D, E, F-H, K-N), and conventional white light microscopy (J2) or standard epifluorescent microscopy (B-C2, E2, H2-J). Loss of specific protein recognition is shown for the cKit (C), Flt-1 (C2) and CD-31 (L2) antibodies following preincubation with blocking peptides. I2, M, N show representative images obtained during antibody control analyses in HUVEC cells, human COC and murine COC, respectively. Oc = oocyte; ZP = zona pellucida; GV = germinal vesicle; black asterisk = brightly stained cells, white asterisk = poorly stained cells. (In G, arrow A points to brightly stained body of a corona cell seen projecting from the surface of the zona pellucida, and arrow B points to a cellular process from this cell extending into the zona pellucida.) Figure 2. View largeDownload slide Representative photomicrographs showing: analysis of cKit (A-C), Flt-1 (C2, D-E2), von Willebrand factor (F-J2) and CD-31(K-L2) proteins and controls (M, N) in freshly isolated human cumulus-oocyte complexes (COC) (A, D, F, G, M), cultures of human cumulus granulosa cells (B, C-C2, E), cultures of human mural granulosa (B2, E2, H2), cultures of positive control human umbilical cord endothelial cells (HUVEC) (I-I2), cultures of negative control human fibroblast cells (J-J2) and freshly isolated murine COC (K-L2, N), following respective analysis. Images were created by scanning laser confocal microscopy (A, D, E, F-H, K-N), and conventional white light microscopy (J2) or standard epifluorescent microscopy (B-C2, E2, H2-J). Loss of specific protein recognition is shown for the cKit (C), Flt-1 (C2) and CD-31 (L2) antibodies following preincubation with blocking peptides. I2, M, N show representative images obtained during antibody control analyses in HUVEC cells, human COC and murine COC, respectively. Oc = oocyte; ZP = zona pellucida; GV = germinal vesicle; black asterisk = brightly stained cells, white asterisk = poorly stained cells. (In G, arrow A points to brightly stained body of a corona cell seen projecting from the surface of the zona pellucida, and arrow B points to a cellular process from this cell extending into the zona pellucida.) Figure 3. View largeDownload slide Cultures of murine (A) and human cumulus granulosa cells (B1, B2, D-F) maintained under different growth conditions and engaged in the formation of tube or capillary-like structures. Images were created by white light microscopy (A-E) and conventional epifluorescent microscopy (F). Semi-thin, cross-section analysis of the structures shown in B1 and B2 determined the tubular regions (black arrowheads, B1; black arrows, B2) to be hollow, membrane-bound and positioned between the cells of the monolayer, C1, C2 (black arrows). See text for additional details. Figure 3. View largeDownload slide Cultures of murine (A) and human cumulus granulosa cells (B1, B2, D-F) maintained under different growth conditions and engaged in the formation of tube or capillary-like structures. Images were created by white light microscopy (A-E) and conventional epifluorescent microscopy (F). Semi-thin, cross-section analysis of the structures shown in B1 and B2 determined the tubular regions (black arrowheads, B1; black arrows, B2) to be hollow, membrane-bound and positioned between the cells of the monolayer, C1, C2 (black arrows). See text for additional details. Figure 4. View largeDownload slide Results of Western analyses for the von Willebrand factor (A), CD-31 (B), Flt-1 (C), Tek (D) and Tie (E) proteins from samples derived from fetal calf serum (A1), immunoprecipitates of cell lysates prepared from 35S-labelled human cumulus granulosa (A3, C1) and murine cumulus granulosa cells (B1, D1, E1) and conditioned medium derived from the human cumulus granulosa cells analysed in A3 (A5). In each case, specific results are paired with filter strips displaying Western analysis results shown on the left (A1, A3, A5, B1, C1, D1, E1) and corresponding results obtained following autoradiographic analysis of those filter strips shown on the right (A2, A4, A6, B2, C2, D2, E2). For each paired result, arrows delineate specific protein bands identified following immunodetection (left) and the position(s) of prominent radioactive proteins (right). Putative proteolytic peptides are identified by either an asterisk (*) or a closed circle (•). The position of molecular weight markers of the approximate sizes indicated (kDa) are shown to the left of each set of results. Electrophoresis was allowed to continue until all molecular weight markers except the 200 kDa marker had run off the gel (A and C) or until the only molecular weight markers remaining in the gel were those of 200 kDa and 97.4 kDa in size (B, D, E). This method gave optimal separation and resolution of the proteins under examination. Figure 4. View largeDownload slide Results of Western analyses for the von Willebrand factor (A), CD-31 (B), Flt-1 (C), Tek (D) and Tie (E) proteins from samples derived from fetal calf serum (A1), immunoprecipitates of cell lysates prepared from 35S-labelled human cumulus granulosa (A3, C1) and murine cumulus granulosa cells (B1, D1, E1) and conditioned medium derived from the human cumulus granulosa cells analysed in A3 (A5). In each case, specific results are paired with filter strips displaying Western analysis results shown on the left (A1, A3, A5, B1, C1, D1, E1) and corresponding results obtained following autoradiographic analysis of those filter strips shown on the right (A2, A4, A6, B2, C2, D2, E2). For each paired result, arrows delineate specific protein bands identified following immunodetection (left) and the position(s) of prominent radioactive proteins (right). Putative proteolytic peptides are identified by either an asterisk (*) or a closed circle (•). The position of molecular weight markers of the approximate sizes indicated (kDa) are shown to the left of each set of results. Electrophoresis was allowed to continue until all molecular weight markers except the 200 kDa marker had run off the gel (A and C) or until the only molecular weight markers remaining in the gel were those of 200 kDa and 97.4 kDa in size (B, D, E). This method gave optimal separation and resolution of the proteins under examination. 3 To whom correspondence should be addressed at: Department of Molecular, Cellular and Developmental Biology, University of Colorado, Boulder, Colorado, 80309, USA.E-mail: antczak@colorado.edu This work was supported by a grant from the National Institutes of Health (HD31907). Note added at proof Preliminary RT-PCR analysis of human and/or murine granulosa cells confirmed the presence of mRNA transcripts for the vWF, CD-31, Tie and Tek proteins in these cells. References Antczak, M. and Van Blerkom, J. ( 1997a) Oocyte influences on early development: the regulatory proteins leptin and STAT3 are polarized in mouse and human oocytes and differentially distributed within the cells of the preimplantation stage embryo. Mol. Hum. Reprod. , 3, 1067–1086. Google Scholar Antczak, M. and Van Blerkom, J. ( 1997b) A novel mechanism of VEGF, leptin, and TFB2 sequestration in a subpopulation of human ovarian follicle cells. Hum. Reprod. , 12, 2226–2234. Google Scholar Aston, K.E., Stamouli, A., Thomas, E.J. et al. ( 1996) Effect of gonadotrophin on cell and matrix retention and expression of metalloproteinases and their inhibitor in cultured granulosa cells modelling corpus luteum formation. Mol. Hum. Reprod. , 2, 26–30. Google Scholar Baranao, R.I., Dain, L., Palak de Fried, E. and Rumi, L.S. ( 1995) Human granulosa cells express HLA-DR antigen and are capable of synthesizing interleukin-1. Horm. Metab. Res. , 27, 495–498. Google Scholar Beckmann, M.W., Polacek, D., Seung, L. and Schreiber, J.R. ( 1991) Human ovarian granulosa cell culture: determination of blood cell contamination and evaluation of possible culture purification steps. Fertil. Steril. , 56, 881–887. Google Scholar Bonthron, D.T., Handin, R.I., Kaufman, R.J. et al. ( 1986) Structure of pre-pro-von Willebrand factor and its expression in heterologous cells. Nature , 324, 270–273. Google Scholar Chen, Z. and Menon, K. ( 1993) Identification of a scavenger receptor in rat luteal cells recognizes chemically modified lipoproteins and media uptake of cholesterol for steroidogenesis. Biochem. Biophys. Acta , 1150, 79–88. Google Scholar Daud, A.I., Bumpus, F.M. and Husain, A. ( 1990) Characterization of angiotensin I-converting enzyme (ACE)-containing follicles in the rat ovary during the estrous cycle and effects of ACE inhibitor on ovulation. Endocrinology , 126, 2927–2935. Google Scholar Dreyfus, M., Dardik, R., Suh, B.S. et al. ( 1992) Differentiation-controlled synthesis and binding of thrombospondin to granulosa cells. Endocrinology , 130, 2565–2570. Google Scholar Dumont, D.J., Yamaguchi, T.P., Conlon, R.A. et al. ( 1992) tek, a novel tyrosine kinase gene located on mouse chromosome 4, is expressed in endothelial cells and their presumptive precursors. Oncogene , 7, 1471–1480. Google Scholar Dumont, D.J., Gradwohl, G., Fong, G.H. et al. ( 1993) The endothelial-specific receptor tyrosine kinase, tek, is a member of a new subfamily of receptors. Oncogene , 8, 1293–1301. Google Scholar Folkman, J. and Huadenschild, C. ( 1980) Angiogenesis in vitro. Nature , 288, 551–556. Google Scholar Haraldsen, G., Kvale, D., Lien, B. et al. ( 1996) Cytokine-regulated expression of E-selectin, intercellular adhesion molecule-1 and vascular cell adhesion molecule-1 in human intestinal microvascular endothelial cells. J. Immunol. , 156, 2558–2565. Google Scholar Hashiyama, M., Iwama, A., Ohshiro, K. et al. ( 1996) Predominant expression of a receptor tyrosine kinase, TIE, in hematopoietic stem cells and B cells. Blood , 87, 93–101. Google Scholar Hosseini, G., Liu, J. and deAgostini, A.I. ( 1996) Characterization and hormonal modulation of anticoagulant heparin sulfate proteoglycans synthesized by rat ovarian granulosa cells. J. Biol. Chem. , 271, 22090–22099. Google Scholar Karakji, E.G. and Tsang, B.K. ( 1995) Follicular stage-dependent regulation of rat granulosa cell plasminogen activator system by transforming growth factor-alpha in vitro. Biol. Reprod. , 52, 411–418. Google Scholar Katz, E., Goodman, A.L., Garcia, J. and Rone, J.D. ( 1992) Evidence for the production of a nonsteroidal endotheliotropic factor by human granulosa cells in culture. Fertil. Steril. , 57, 107–112. Google Scholar Loukides, J.A., Randall, A.L., Edwards, R. et al. ( 1990) Human follicular fluid contains tissue macrophages. J. Clin. Endocrinol. Metab. , 71, 1363–1367. Google Scholar Lynch, D.C., Williams, R., Zimmerman, T.S. et al. ( 1983a) Biosynthesis of the subunits of factor VIIIR by bovine aortic endothelial cells. Proc. Natl. Acad. Sci. USA , 80, 2738–2742. Google Scholar Lynch, D.C., Zimmerman, T.S., Kirby, E.P. and Livingston, D.M. ( 1983b) Subunit composition of oligomeric human von Willebrand factor. J. Biol. Chem. , 258, 12757–12760. Google Scholar Machelon, V., Nome, F., Durand-Gasselin, I. and Emilie, D. ( 1995) Macrophage and granulosa interleukin-1B mRNA in human ovulatory follicles. Hum. Reprod. , 10, 2198–2203. Google Scholar Montesano, R., Orci, L. and Vassalli, P. ( 1983) In vitro rapid organization of endothelial cells into capillary-like networks is promoted by collagen matrices. J. Cell Biol. , 97, 1648–1652. Google Scholar Obeso, J., Weber, J. and Auerbach, R. ( 1990) A hemangioendothelioma-derived cell line: Its use as a model for the study of endothelial cell biology. Lab. Invest. , 63, 259–269. Google Scholar Peters, K.G., DeVries, C. and Williams, L.T. ( 1993) Vascular endothelial growth factor receptor expression during embryogenesis and tissue repair suggests a role in endothelial differentiation and blood vessel growth. Proc. Natl Acad. Sci. USA , 90, 8915–8919. Google Scholar Piquette, G.N., Crabtree, M.E., El-Danasouri, I. et al. ( 1993) Regulation of plasminogen activator inhibitor-1 and -2 messenger ribonucleic acid levels in human cumulus and granulosa-luteal cells. J. Clin. Endocrinol. Metab. , 76, 518–523. Google Scholar Ratcliffe, K.E., Anthony, F.W., Richardson, M.C. and Stones, R.W. ( 1999) Morphology and functional characteristics of human ovarian microvascular endothelium. Hum. Reprod. , 14, 1549–1554. Google Scholar Rose, B.I. and Koos, R.D. ( 1988) The effect of human follicular fluid on endothelial cells: proliferation and DNA synthesis. Biol. Reprod. , 39, 88–95. Google Scholar Ruggeri, Z.M. and Ware, J. ( 1993) von Willebrand factor. FASEB J. , 7, 308–316. Google Scholar Seetharam, L., Gotoh, N., Maru, Y. et al. ( 1995) A unique signal transduction from FLT tyrosine kinase, a receptor for vascular endothelial cell growth factor VEGF. Oncogene , 10, 135–147. Google Scholar Spanel-Borowski, K. and van der Bosch, J. ( 1990) Different phenotypes of cultured microvessel endothelial cells obtained from bovine corpus luteum. Cell Tissue Res. , 261, 35–47. Google Scholar Spanel-Borowski, K. and Ricken, A.M. (1997) Varying morphology of bovine granulosa cell cultures. In Motta, P.M. (ed.), Microscopy of Reproduction and Development: A Dynamic Approach. Antonio Delfino Editore, Rome, pp. 91–100. Google Scholar Tanikawa, M., Harada, T., Mitsunari, M. et al. ( 1998) Expression of c-kit messenger ribonucleic acid in human oocyte and presence of soluble c-kit in follicular fluid. J. Clin. Endocrinol. Metab. , 83, 1239–1242. Google Scholar Van Blerkom, J. and Motta, P. (1979) The Cellular Basis of Mammalian Reproduction. Urban and Schwarzenberg, Inc., Baltimore, pp. 5–51. Google Scholar Van Blerkom, J., Davis, P., Merriam, J. and Sinclair, J. ( 1995) Nuclear and cytoplasmic dynamics of sperm penetration, pronuclear formation and microtubule organization during fertilization and early preimplantation development in the human. Hum. Reprod. Update , 1, 429–461. Google Scholar Voyta, J.C., Via, D.P., Butterfield, C.E. and Zetter, B.R. ( 1984) Identification and isolation of endothelial cells based on their increased uptake of acetylated-low density lipoprotein. J. Cell Biol. , 99, 2034–2040. Google Scholar Weibel, E.R. and Palade, G.E. ( 1964) New cytoplasmic components in aterial endothelia. J. Cell Biol. , 23, 101–112. Google Scholar Xie, Y. and Muller, W.A. ( 1993) Molecular cloning and adhesive properties of murine platelet/endothelial cell adhesion molecule 1. Proc. Natl. Acad. Sci. USA , 90, 5569–5573. Google Scholar Ziegler, S.F., Bird, T.A., Schneringer, J.A. et al. ( 1993) Molecular cloning and characterization of a novel receptor protein tyrosine kinase from human placenta. Oncogene , 8, 663–670. Google Scholar Zimmerman, T.S., Dent, J.A., Ruggeri, Z.M. and Nannini, L.H. ( 1986) Subunit composition of plasma von Willebrand factor: Cleavage is present in normal individuals, increased in IIA and IIB von Willebrand disease, but minimal in variants with aberrant structure of individual oligomers (types IIC, IID and IIE). J. Clin. Invest. , 77, 947–951. Google Scholar © European Society of Human Reproduction and Embryology
TI - The vascular character of ovarian follicular granulosa cells: phenotypic and functional evidence for an endothelial-like cell population
JF - Human Reproduction
DO - 10.1093/humrep/15.11.2306
DA - 2000-11-01
UR - https://www.deepdyve.com/lp/oxford-university-press/the-vascular-character-of-ovarian-follicular-granulosa-cells-ZdJYm0LFSw
SP - 2306
EP - 2318
VL - 15
IS - 11
DP - DeepDyve
ER -