Mechanically altered embryonic chicken endothelial cells change their phenotype to an epithelioid phenotype
Abstract
Monolayers of retracted endothelial cells exhibiting wounds or zones denuded of cells were obtained from aortic explants from 10- to 12-day-old chicken embryos. Using time-lapse videomicroscopy, we investigated the sequence of events that occurred both during and after closure of the monolayer wounds. Such wound closure (re-endothelialization process) occurred 4–12 hr after removing the explants, depending on wound width and presence of serum. The cells from along the wound edges appeared to move toward one another. We suggest an important role for bFGF and TGFβ-2 and -3 during this process. Twenty-five hours after removal there were still some areas of retracted cells, and many of the cells displayed a weak von Willebrand's Factor (vWf) immunoreactivity. Surprisingly, after 63–65 hr many of the endothelial cells had become epithelioid in shape and the vWf immunoreactivity appeared increased. This epithelioid phenotype is currently considered typical of cultured vascular non-muscle-like cells and intimal thickening cells. By 5–7 days, the vast majority of cells in the monolayer had acquired an epithelioid morphology, showing a cobblestone appearance. These cells were significantly smaller than polygonal cells. Most importantly, they showed strong vWf immunoreactivity. At the edge of the monolayers we found that the majority of the cells had become epithelioid. Some of them detached from their neighbors and became round in shape and acquired mesenchymal characteristics, some expressing smooth muscle α-actin (SM α-actin). These findings demonstrate not only that embryonic endothelial cells that are transiently mechanically altered may change their phenotype to an epithelioid phenotype, but also that these cells may eventually transdifferentiate into mesenchymal cells expressing SM α-actin. Since some aspects of endothelial cell behavior have been shown to be regulated by locally released growth factors such as TGFβ and FGF, we also investigated TGFβ-2 and -3 and bFGF expression. Presence of TGFβ-2 and -3 and bFGF-immunoreactive epithelioid and mesenchymal cells indicates that these growth factors may be involved in the changes described. Anat Rec Part A 270A:67–81, 2003. © 2003 Wiley-Liss, Inc.
In vitro cell culture studies oriented toward a better understanding of intimal lesions have reported that intimal smooth muscle cells (SMCs), which seem to have special properties that may contribute to vascular diseases, show a very different phenotype than “traditional” medial SMCs (Schwartz et al., 1995a, 1995b; Seidel, 1997; Shanahan and Weisberg, 1998; Gittenberger-de Groot et al., 1999). These differences have suggested diverse origins for intimal cells. In this respect, studies regarding experimental intimal thickening have indicated that an SMC subpopulation, characterized by expression of nonmuscle protein isoforms and epithelioid morphology more reminiscent of endothelial cells, can be obtained after endothelial injury (Walker et al., 1986; Majesky et al., 1992; Orlandi et al., 1994; Pauletto et al., 1994; Bochaton-Piallat et al., 1996; Holifield et al., 1996; Adams et al., 1999). For this reason, these epithelioid cells have also been described as having a nonmuscle phenotype. Moreover, the same studies propose that such epithelioid cells are derived from an SMC subset. Additionally, cells exhibiting reduced smooth muscle α-actin (SM α-actin) expression and epithelioid morphology have also been isolated from the intima of normal rat aorta (Villaschi et al., 1994). This study suggested that these cells could originate from preexisting SMCs in the intima, based on the in vivo presence of scattered SMCs at the subendothelial space. Likewise, a subpopulation of epithelioid cells expressing little to no SM α-actin and virtually no smooth muscle myosin has been isolated and characterized from normal mature bovine pulmonary arterial media (Frid et al., 1997). Interestingly, epithelioid cells have also been isolated from the intima of fetal and neonatal arteries (Gordon et al., 1986; Majesky et al., 1988; Ehler et al., 1995; Kohler et al., 1999) and recently from the human internal thoracic artery (Li et al., 2001). Especially interesting, Topouzis and Majesky (1996) reported that cells originating within the cardiac neural crest give rise to SMCs that eventually form a cobblestone-appearing monolayer and display an epithelioid phenotype.
As the monoclonal theory of atherosclerosis (Benditt and Benditt, 1973; Schwartz et al., 1995b) postulates that the initial SMC proliferative event involves the clonal expansion of an SMC subpopulation in response to a specific stimulus, the role of epithelioid cells has been more frequently established not only during the remodeling process of the media, but also in neointima or intimal thickening and in vascular diseases (Majesky et al., 1992; Ehler et al., 1995; Schwartz et al., 1995b; Gittenberger-de Groot et al., 1999). However, the origin of these epithelioid cells and the signals regulating their behavior have not yet been clearly established (Schwartz et al., 1995a; Gittenberger-de Groot et al., 1999).
Endothelial cells have been recognized as playing an important role not only in neointimal formation after injury (Beranek, 2001), but also during vascular development (Tuder et al., 1994; DeRuiter et al., 1997; Arciniegas et al., 2000), although their direct participation in these processes is not well understood. Recently, we showed that endothelial cells scraped from the luminal surface of the bovine pulmonary artery gave rise to enlarged and binucleated cells that formed a confluent monolayer with a cobblestone appearance and adopted an epithelioid morphology, typical of neonatal and intimal thickening cells in culture (Graterol et al., 2000).
In this study we investigated the behavior of mechanically altered embryonic endothelial cells in terms of their morphology and differentiation. These cells were obtained from 10- to 12-day-old aortic explants (stages 35–37) from chicken embryos. At these stages intimal thickening and endothelial-mesenchymal transdifferentiation seem to occur.
Since some aspects of endothelial cell behavior have been shown to be regulated by locally released growth factors such as FGF and TGF-β, we also investigated bFGF or FGF-2, TGFβ-2, and TGFβ-3 expression. We demonstrated that addition of bFGF or TGFβ-2 or -3 to monolayers of embryonic endothelial cells exhibiting zones denuded of cells promotes the re-endothelialization process. We also found that during culture, embryonic endothelial cells clearly altered their phenotype to an epithelioid phenotype. This alteration corresponded with changes in the expression of von Willebrand's Factor (vWf)-related antigen. Interestingly, some epithelioid cells detached from the edges of the monolayer differentiate into mesenchymal-like cells, some of them expressing SM α-actin. Also, the presence of TGFβ-2 and -3 and bFGF-immunoreactive epithelioid and mesenchymal cells indicated that these growth factors may be involved in the aforementioned changes.
MATERIALS AND METHODS
Embryonic Aortic Explants
Fertilized chicken eggs (Brown Highland) were obtained from a local hatchery (Granja Avícola Santa Cruz C.A., Paracotos, Edo. Miranda) and incubated at 37°C and 60% humidity for 10–12 days (stages 35–37). Embryos were staged according to Hamburger and Hamilton (1992). Aortas were dissected in Hank's balanced salt solution (Life Technologies, GIBCO BRL) at 37°C. Aortic segments, approximately 8.12 ± 0.04 mm2 in surface area, were isolated (distal to the aortic arches) and opened along the longitudinal axis. Explants were rinsed with a Ca++, Mg++-free phosphate-buffered saline (PBS) (GIBCO BRL) and placed in medium 199 with Earle's salts, with L-glutamine (GIBCO BRL) supplemented with 1% chicken serum (GIBCO BRL), insulin-transferrin-selenium (ITS) (GIBCO BRL), 100 μg/ml streptomycin, and 100 U/ml penicillin (GIBCO BRL) (this medium is referred to as a complete medium).
Cell Cultures
Cell cultures were initiated by placing the explants, with the endothelial side down, in a gelatin-coated 35-mm Petri dish surface (Nunclon Delta, IL) containing 250 μl of complete medium. They were incubated at 37°C in a humidified atmosphere consisting of 5% CO2 and 95% air for 3–4 hr in order to allow stable attachment to the surface. After this interval, complete medium 199 was gently added to each dish. Two hours later the adhered explants were removed with the aid of a thin needle, leaving behind a monolayer of retracted or altered endothelial cells that exhibited zones denuded of cells. The surfaces were then rinsed five times with medium 199 without chicken serum (serum-free medium (SFM)). Fresh complete medium was then added to each dish. Each set of experiments consisted of at least eight dishes. One of each set of experiments was videorecorded every 10 min for a 12-hr period starting immediately after the removal of the explants, whereas the rest were examined every 2 hr during the first 3 days of culture and every 12 hr thenceforth until day 7 in an inverted microscope (IX-70 Olympus, Olympus America Inc., Melville, NY) with a ×10 objective. Each experiment reported in this study was repeated at least nine times with consistent results.
Time-Lapse Phase-Contrast Videomicroscopy
Time-lapse videomicroscopy of cell cultures was performed using an inverted microscope Olympus IX-70, equipped with a CO2 incubator containing distilled water and kept at 37°C. During the first 12 hr two images were captured every 10 min; one focused on the center of the monolayer and the other on the edge of the monolayer. After this period, images were captured every 2 and 12 hr until days 2 and 7, respectively. An image editing capture and processing software program (Image Pro Plus, Media Cybernetics, Silver Spring, MD) was used.
Effect of Growth Factors on Re-Endothelialization Process
To determine whether bFGF and TGFβ mediate the re-endothelialization process, some monolayers of retracted cells exhibiting zones denuded of cells were maintained in SFM containing bovine or recombinant human FGF basic (bFGF; 1–100 ng/ml) (R&D Systems, Minneapolis, MN), porcine TGFβ-2, and recombinant human TGFβ-3 (rhTGFβ-3; 0.5–10 ng/ml) (R&D Systems, Minneapolis, MN). The recombinant human bFGF was kindly provided by Dr. C. Reynolds, National Cancer Institute, Rockville, Maryland. Cultures were videorecorded every 2 hr for a 20-hr period, starting immediately after the removal of the explants.
Cell Counting and Measuring
These determinations were done in microscope fields from a monolayer that had been captured 4 hr and 7 days after the removal of the explants. The cell area, diameter, and perimeter were quantified from the images captured using the Image Pro Plus software program (Image Pro Plus, Media Cybernetics, MD). The values were expressed as mean ± S.D. using descriptive statistical methods.
Indirect Immunofluorescence
Fixed and permeabilized cells were processed for staining for vWf (rabbit anti-human vWf) (Dako, Carpinteria, CA), SM α-actin (mouse ascites monoclonal) (clone 1A4) (Sigma, St. Louis, MO), and calponin (mouse monoclonal anti-smooth muscle calponin) (clone CP93) (Sigma, MO) as previously described (Arciniegas et al., 2000). Negative controls were performed by using purified normal serum or PBS instead of primary antibody. Cell cultures were examined and photographed with an inverted microscope (IX-70 Olympus).
Immunoperoxidase
To investigate the spatial distribution of immunoreactive TGFβ-2 and -3 and bFGF at the aortic wall and in cell cultures, we used commercial neutralizing antibody anti-TGFβ-2 produced in rabbits and anti-TGFβ-3 produced in mouse (R&D Systems, Minneapolis, MN), both at a 20 μg/ml concentration. We also used anti-FGF basic commercial neutralizing antibody produced in rabbits (20 μg/ml) (R&D Systems, Minneapolis, MN) in cell cultures.
An immunoperoxidase technique using avidin-biotin-peroxidase complex (ABC) was used as described (Taylor and Cote, 1994), with some modifications. Briefly, rehydrated sections and cell cultures were rinsed in PBS and endogenous peroxidase activity and nonspecific binding were blocked with 3% H2O2 in methanol and 2.5% normal horse serum, respectively. Sections and cell cultures were incubated overnight at 4°C in a humid chamber with the primary antibodies. After several PBS washes, a biotinylated secondary antibody (Universal Quick Kit, Vector) was applied for 10 min at room temperature, followed by a streptavidin horseradish peroxidase conjugate. The sections and the dishes were developed for 5 min with 3-3′-diaminobenzidine as the chromogen (Dako and Vector Laboratories, Inc., Burlingame, CA). Both were then rinsed in tap water and counterstained with Mayer's hematoxylin. Negative controls were performed using purified normal serum instead of the primary antibody or omitting the primary antibodies in the incubation.
Cell Proliferation Assays
DNA synthesis was determined in wounded endothelial cell cultures by 5-bromo-2′-deoxyuridine (BrdU) incorporation (Cell Proliferation Kit, Amersham Biosciences, Buckingshire, UK).
After the removal of the explants, some cultures were maintained 9 hr in complete medium. During the final 2 hr, cultures were incubated with the labeling medium (BrdU) containing 5-fluoro-2′-deoxyuridine, which inhibits thymidylate synthetase and thereby maximizes BrdU incorporation. After several PBS washes, the cultures were fixed in acid-ethanol for 30 min and rehydrated by several PBS washes. The BrdU incorporated was detected immunocytochemically with an anti-BrdU monoclonal antibody-nuclease, followed by peroxidase-labeled anti-mouse immunoglobulin G and the chromogen diaminobenzidine. The culture dishes were mounted and examined in an Olympus IX-70 inverted microscope with ×10 objective. All cells in the repaired area and in the immediately adjacent regions were counted in each microscope field (one per culture dish), and the labeled nuclei were scored. Counting was performed from the images captured using the Image Pro Plus software program (Image Pro Plus, Media Cybernetics, MD). At least 700 cells were counted for each field, and the percentage of labeled cells ± S.D. was calculated.
RESULTS
When aortic explants obtained from 10- to 12-day-old chicken embryos (stages 35–37) were cultured for 5–6 hr in gelatin-coated Petri dishes with complete medium, no cells grew from the edges of the explants. However, after removing the explants, a monolayer of retracted endothelial cells exhibiting zones denuded of cells or wounds was found adhered to the surface of the dish where the endothelial face of the explants had been located (Fig. 1a). These zones were most probably a consequence of the mechanical force exerted on the attached cells upon removal of the explants.
Series of time-lapse phase-contrast videomicroscopy of the replenishment of zones denuded of cells. Time course between frames. a: Time zero. Shows a monolayer of retracted endothelial cells exhibiting zones denuded of cells or wounds after removing the explant. b: 4 hr. Cells from the wound edges, moving toward each other, extending lamellipodia into denuded area, showing a polygonal shape. In c, the asterisk marks retracted endothelial cells. Large arrows indicate the overall direction of cell movement. c: Detail of b showing polygonal cells. d: 12 hr. Following contact, the closure of the last wound takes place. e: 63 hr. Many of the endothelial cells have become epithelioid in shape. f: After 7 days in culture, the cells of the monolayer acquire an epithelioid morphology showing a cobblestone appearance. Scale bars = 40 μm, 20 μm (detail of b).
To determine whether there was a replenishment of zones denuded of endothelial cells, and if so, to examine the manner in which this process occurred, we used time-lapse phase-contrast videomicroscopy.
When the monolayer of retracted endothelial cells was examined 2–4 hr after removing the explant, we found that the cells from along the wound edges appeared to move toward each other, extending lamellipodia into the denuded area, showing a polygonal shape (Fig. 1b and c). When the cells from the two edges of the wound came into contact, the closure of the wound (re-endothelialization) was almost complete (Fig. 1d). This process was completed between 4 and 10 hr after removing the explants, depending on the width of the wound. Next, we investigated in detail the sequence of events that occurred after the closure of the wound. We found that 25 hr after removing the explants some regions of retracted endothelial cells still remained (not shown). When the monolayer was examined 63–65 hr after removal of the explant, we found that an expansion of the monolayer had taken place. Surprisingly, many of the endothelial cells had become epithelioid in shape (Fig. 1e). Importantly, this epithelioid cell morphology observed is currently considered typical of cultured vascular non-muscle-like cells and intimal thickening cells. Therefore, we decided to maintain the cultures for longer periods (5–7 days). By 5–7 days, the vast majority of the cells of the monolayer had acquired an epithelioid morphology, showing a cobblestone appearance (Fig. 1f). Interestingly, they exhibited a prominent nucleus and were significantly smaller than polygonal endothelial cells (Table 1).
| Type of cells | Area (μm2) | Diameter (μm) | Perimeter (μm) |
|---|---|---|---|
| Polygonal (n = 215) | 159.41 ± 39.11 | 13.68 ± 2.02 | 45.99 ± 7.57 |
| Epithelioid (n = 252) | 88.97 ± 33.41 | 9.736 ± 1.94 | 35.157 ± 8.96 |
- Values are mean ± S.D. (n, total number of cells).
When the edge of the monolayer was examined 3 hr after removal of the explant, we found some regions of retracted endothelial cells (Fig. 2a). By 6–10 hr, many of the cells displayed a polygonal shape. At this time, some of the cells located at the marginal edges began to migrate toward cell-free areas (Fig. 2b). After 14 hr, cells assumed an epithelioid morphology with a cobblestone appearance, whereas the migrating marginal cells became elongated (not shown). After 41 hr, the epithelioid morphology became more evident. These cells were also observed detaching from the sheet and migrating away (Fig. 2c). Surprisingly, after 64 hr in culture, the majority of the cells had become epithelioid. Some of them continued to detach from their neighbors and to become round in shape and acquire mesenchymal characteristics (Fig. 2d).
Series of time-lapse videomicroscopy of the edge of the monolayer of retracted endothelial cells. Time course between frames. a: 3 hr after removal. Regions of retracted cells may be seen (asterisk). b: 10 hr. Many of the retracted cells display a polygonal shape (arrows). Cells located at the marginal edge appear to be migrating toward cell-free areas. c: 41 hr. The epithelioid-like morphology becomes evident. Some cells appear to detach from the sheet and migrate (arrow). d: 64 hr. The majority of the cells have become epithelioid. Some of them continue to detach from their neighbors and become round in shape (arrows) and acquire mesenchymal characteristics. Scale bar = 40 μm.
As a whole, these results demonstrate that embryonic endothelial cells that are transiently mechanically altered may change their phenotype into an epithelioid phenotype and that these cells may eventually acquire mesenchymal characteristics.
Effects of Growth Factors on Re-endothelialization Process
To determine whether growth factors such as bFGF and TGFβ are involved in the closure of the wounds (re-endothelialization process), we investigated the expression of these growth factors with immunoperoxidase staining of monolayers of retracted endothelial cells that had been maintained in culture for 6–10 hr after removing the explants, in presence of complete medium.
Strong cytoplasmic bFGF immunoreactivity was found in those cells that appeared to move toward each other into the denuded area and in the cells of the monolayer (Fig. 3). When the TGFβ-2 or -3 expression was investigated in similar conditions, a nuclear immunoreactivity was also observed in some of the cells that were migrating into the denuded area (not shown).
Immunolocalization of bFGF in a monolayer of retracted endothelial cells kept in culture for 10 hr after removing the explant in the presence of complete medium. Note the strong cytoplasmic immunoreactivity for this growth factor in the monolayer cells and in the cells moving toward each other into the denuded area (arrows). Scale bar = 20 μm. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]
To gain insight into factors that mediate the re-endothelialization process, some monolayers of retracted endothelial cells exhibiting zones denuded of cells (Fig. 4a) were cultured in SFM for 10 hr after removal of the explants. In this condition, no cells migrating from the wounded edges were observed after 10 hr (Fig. 4b). Therefore, the re-endothelialization process did not occur. However, if the medium of some cultures that had been maintained in SFM for 10 hr was switched to complete medium and maintained for an additional 10 hr, the re-endothelialization process took place (Fig. 4c).
Series of time-lapse videomicroscopy of a monolayer of retracted endothelial cells kept in SFM for 10 hr after removal of the explant (a and b) and for an additional 10 hr after switching to complete medium (c). Time course between frames. a: Time zero. Shows the monolayer exhibiting zone denuded of cells after removing the explant. b: 10 hr. No cells migrating from the wounded edges are observed. c: 20 hr (10 hr after switching). Closure of wounds takes place. Cells at the marginal edge appear to be migrating toward cell-free areas. Large arrows indicate the direction of migration. Scale bar = 40 μm.
To determine whether bFGF and TGFβ-2 or -3 were mediating the re-endothelialization process, other monolayers that had been maintained in SFM for 10–12 hr were cultured in SFM or SFM containing either bFGF (1–100 ng/ml) or TGFβ (0.5–10 ng/ml) for an additional 10 hr. We found that in SFM the re-endothelialization process did not occur (Fig. 5a and b), whereas addition of bFGF stimulated this process (Fig. 5c and d). In contrast, the addition of TGFβ (10 ng/ml) did not stimulate re-endothelialization (not shown). However, on addition of TGFβ at a 0.5 ng/ml concentration, re-endothelialization was promoted (Fig. 5e and f). Additionally, we also examined if detaching and migrating cells were present at the marginal edges of these monolayers when the time of incubation was prolonged for up to 48 hr. In the SFM condition, neither detaching nor migrating cells were observed (not shown). Under this condition the viability of the cells was found to be particularly compromised. In contrast, the presence of bFGF (1–10 ng/ml) or TGFβ (0.5–1.0 ng/ml) promoted the detachment and migration of cells located at the marginal edges (Fig. 6a–c).
Series of time-lapse videomicroscopy of the monolayers kept in SFM for 12 hr after removal of the explants (a, c, and e) and for an additional 10 hr in SFM (b), SFM containing bFGF (10 ng/ml) (d), and SFM containing TGFβ-3 (0.5 ng/ml) (f). Note that in SFM alone the re-endothelialization process does not occur (b), whereas addition of bFGF (d) and TGFβ-3 (f) stimulates this process. Scale bar = 40 μm.
Series of time-lapse videomicroscopy of the edge of a monolayer kept in SFM for 12 hr after removal of the explant (a) and for up to 24 hr after switching to SFM containing bFGF (1 ng/ml) (c). Time courses between frames: 12 hr (a), 24 hr (b), and 36 hr (c). Some cells appear to detach from the monolayer and migrate away. Scale bar = 40 μm.
Endothelial Cell Identification by Immunofluorescence
To determine whether the alterations observed in the phenotype of the embryonic endothelial cells corresponded with changes in the expression of vWf-related antigen, monolayers of the initially retracted endothelial cells that had been maintained in culture for 24 hr, 64 hr, and 5–7 days in the presence of complete medium were examined by immunofluorescence.
This technique showed weak immunoreactivity for vWf after 24 hr in the monolayer cells (Fig. 7a and b). At this time, some of the elongated cells located at the edges of the monolayer displayed immunoreactivity for vWf (Fig. 7c and d). These cells did not show any immunoreactivity for SM α-actin or calponin (not shown). By 64 hr, when many of the cells had become epithelioid, this immunoreactivity appeared to have increased (not shown). After 5–7 days, when the vast majority of the monolayer cells had acquired an epithelioid morphology with a cobblestone appearance, a strong immunoreactivity for vWf was observed in essentially all of the cells (compare Fig. 7b and f). At this time, some cells of the marginal edges of the monolayers had detached and acquired mesenchymal characteristics (Fig. 8a). To determine whether these cells expressed an SMC phenotype, double immunofluorescence was performed. Interestingly, this technique revealed that some of the epithelioid cells that appeared migrating displayed immunoreactivity for both vWf and SM α-actin, the latter in a fibrillar pattern mainly distributed delineating the cellular margins (Fig. 8b and c), whereas cells of the monolayer displayed a strong immunoreactivity only for vWf (Fig. 8b). To examine the presence of mesenchymal cells expressing both vWf and calponin, double immunofluorescence was also performed. Double immunostaining revealed the presence of some mesenchymal cells expressing calponin (Fig. 9a and b).
Immunolocalization of vWf after removing the explants. Phase-contrast of the monolayer of retracted cells after 24 hr in culture in the presence of complete medium (a). Weak immunoreactivity for vWf (red) in a punctated and granular pattern typical of endothelial cells is observed in the monolayer cells after 24 hr in culture (b). At this moment, some of the elongated cells located at the edges of the monolayer (c) display immunoreactivity for vWf (d). After 7 days, when the majority of the monolayer cells acquire an epithelioid morphology with a cobblestone appearance (e), a strong vWf immunoreactivty is observed (f). Scale bars = 15 μm (a–c), 10 μm (d–f).
Double immunofluorescence of the cells located at the marginal edges of a monolayer after 6 days in culture. Note that some of the epithelioid cells that appear migrating (a) display immunoreactivity for both vWf (orange), in a punctate and granular pattern, and SM α-actin (green) (b, arrow), in a fibrillar pattern delineating the cellular margins (c, arrow). Scale bars = 15 μm (a), 10 μm (b and c).
Immunolocalization of vWf and calponin at the marginal edge of a monolayer after 6 days in culture. Some of the mesenchymal cells display immunoreactivity for calponin (green), whereas other display immunoreactivity for vWf (orange). Scale bar = 20 μm.
In Vitro TGFβ-2, TGFβ-3, and bFGF Immunolocalization
Because some aspects of endothelial cell behavior have been shown to be regulated by locally released growth factors such as TGFβ and FGF, we next investigated TGFβ-2 and -3 and bFGF expression by immunoperoxidase staining in monolayers of retracted endothelial cells that had been maintained in culture in the presence of complete medium over 48 hr.
When these cultures were examined, strong cytoplasmic immunoreactivity for TGFβ-2 and -3 was observed in patches of epithelioid cells (Fig. 10a and c). When the edges of the same monolayers were examined, TGFβ-2 and -3 expression was limited to the detaching and migrating cells (Fig. 10b and d). Interestingly, the cells that had acquired mesenchymal characteristics exhibited strong immunoreactivity for both growth factors. Like TGFβ-2 and -3, bFGF immunoreactivity was detected in some detaching and migrating cells and those cells that had acquired mesenchymal characteristics (Fig. 11).
Immunolocalization of TGFβ-2, and -3 in a monolayer kept in culture for 48 hr in the presence of complete medium. Immunoreactivity for TGFβ-2 (a and b) and for TGFβ-3 (c and d) is observed in patches of epithelioid cells (a and c) and some detaching and migrating cells located at the edge of the same monolayer (b and d). Scale bar = 20 μm. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]
Immunolocalization of bFGF in a monolayer kept for 48 hr in the presence of complete medium. Immunoreactivity is observed in some detaching and migrating cells. Scale bar = 20 μm. [Color figure can be viewed in the online issue, which is available at www. interscience.wiley.com.]
In Vivo TGFβ-2, TGFβ-3, and bFGF Immunolocalization
To determine if TGFβ-2 and -3 and bFGF are expressed in vivo during day 11 (stage 36) of development, serial frozen sections from the aorta were examined by immunoperoxidase staining.
At this stage, when intimal thickening is detected, strong immunoreactivity for these growth factors was observed in most endothelial cells and in some of the radially oriented mesenchymal cells (Fig. 12a and c).
Distribution of immunoreactive TGFβ-2 (a), TGFβ-3 (b), and bFGF (c) in the aortic wall at day 11. Strong immunoreactivity for TGFβ-2 and -3 and bFGF is observed in most endothelial cells (e) and some radially oriented mesenchymal cells (arrows) that form an intimal thickening. Scale bars = 10 μm (a and b), 20 μm (c).
Cell Proliferation Assays
Endothelial cell spreading, migration, and proliferation have been considered important events during re-endothelialization. In this study, cell proliferation was assessed by BrdU incorporation.
Nine hours after removing the explant, BrdU incorporation was generally detected in a small percent of the cells located both at the repaired area and in the immediately adjacent regions (Table 2; Fig. 13a and b), suggesting that cell spreading and migration, but not cell proliferation, were the major mechanisms responsible for closure of the wound (re-endothelialization).
| Total number of cells | Labeled nuclei | % of labeled cells |
|---|---|---|
| 1635 | 30 | 1.83 |
| 1249 | 23 | 1.84 |
| 927 | 11 | 1.18 |
| 759 | 10 | 1.31 |
| 873 | 12 | 1.37 |
| 1092 | 21 | 1.90 |
| 1089 | 7 | 0.60 |
| Mean: 1103.429 | 16.286 | 1.433 |
- % of labeled nuclei (Mean % ± S.D.): 1.433 ± 0.4687. % of total number of cells (Mean % ± S.D.): 98.567 ± 0.4687.
Determination of DNA synthesis by incorporation of BrdU. a: Nine hours after removal of the explant, BrdU incorporation is detected in a small percent of the cells located both at the repaired area (arrows) and in regions of retracted cells. b: Phase-contrast shows the total number of cells. Scale bar = 20 μm.
DISCUSSION
In the present study we investigated in detail the sequence of events leading to wound closure of a monolayer of retracted endothelial cells. Retracted endothelial and epithelial cells, as well as zones denuded of cells or wounds, have been described when monolayers of endothelial (Wong and Gotlieb, 1984, 1988) and epithelial (Lotz et al., 2000) cells were wounded.
By using time-lapse videomicroscopy, we showed that the closure of wounds occurred between 4 and 12 hr after removing the explants, depending on wound width and presence of serum. During this process, cells along wound edges appeared to move toward one another, extending lamellipodia, whereas the retracted cells were observed moving together. These observations are consistent with the possibility that both cell spreading and migration play an important role in in vitro re-endothelialization. Previous studies have emphasized the importance during this process of not only endothelial cell spreading and migration, but also cell proliferation (Clowes et al., 1986; Lindner et al., 1989; Ettenson and Gotlieb, 1994). However, the cell proliferation assays performed in this study showed BrdU incorporation in a small percent of cells during the closure of the wounds, suggesting that endothelial cell spreading and migration, but not cell division, were the major mechanisms accounting for wound closure. This low proliferation could be associated with reorganization of the cytoskeleton, which in turn would be influenced by the presence of certain growth factors released after transient or permanent endothelial injury (Sato and Rifkin, 1988; Akong and Gotlieb, 1999). In this respect, several studies have indicated that a number of growth factors, including bFGF or FGF-2, TGFβ, and platelet-derived growth factor (PDGF), play roles in the re-endothelialization process. bFGF, known to stimulate the production of plasminogen activator and endothelial cell migration (Montesano et al., 1986; Tsuboi et al., 1990), has recently been shown to induce changes in shape and cytoskeletal alterations in vascular cells (Cavallaro et al., 2001a, b). Additionally, immunolocalization studies have shown that this growth factor is located in the nucleus and cytoplasm of endothelial cells in culture, suggesting that such locations are associated with the activity of these cells (Renko et al., 1990; Dell'Era et al., 1991; Muthukrishnan et al., 1991; Acevedo et al., 1993; Yu et al., 1993). Consequently, the strong cytoplasmic immunoreactivity observed for bFGF when monolayers of retracted endothelial cells were maintained in culture for 6–10 hr in the presence of complete medium may be associated with the migration of the cells into the denuded area. We assessed the addition of bFGF to monolayers maintained in SFM and found that bFGF promoted the re-endothelialization process, considering that in SFM alone usually no migrating cells were observed.
TGFβ has been demonstrated to transiently inhibit proliferation and migration of endothelial cells during the re-endothelialization process (Heimark et al., 1986). It has also been shown to inhibit the in vitro effects of bFGF on endothelial cells (Pepper et al., 1990). In this study, as for bFGF, TGFβ immunoreactivity was found in migrating cells. Since endothelial cell migration, proliferation, and differentiation probably involve the complex interaction of both stimulatory and inhibitory factors, the similarities we detected in immunoreactivity and localization for bFGF and TGFβ may reflect a dynamic balance between these growth factors during the re-endothelialization process.
An important and novel finding in our study was that the endothelial cells from monolayers that were mechanically altered during the explant removal acquired a small epithelioid morphology with a cobblestone appearance when the cultures were maintained for longer periods (5–7 days). This epithelioid phenotype is similar to that exhibited by SMCs isolated from the intima and neonatal animal arteries (Gordon et al., 1986; Majesky et al., 1988; Ehler et al., 1995; Kohler et al., 1999), intimal thickening (Majesky et al., 1992; Ehler et al., 1995; Schwartz et al., 1995b; Gittenberger-de Groot et al., 1999), the media of normal mature bovine pulmonary artery (Frid et al., 1997), and recently the human internal thoracic artery (Li et al., 2001). From these studies, different origins for these epithelioid cells, including from medial SMCs or from normally resident smooth muscle “stem cells” within the media (Seidel, 1997), have been proposed. Interestingly, epithelioid cells have been suggested to be involved in the repair process after endothelial injury (Majesky et al., 1992). Of particular interest, and in support of the results presented here, we recently demonstrated that endothelial cells scraped from the luminal surface of normal mature bovine pulmonary artery give rise to epithelioid cells (Graterol et al., 2000). Moreover, we showed that during this process endothelial cells, which are characterized by vWf expression, appeared to gradually lose their immunoreactivity. From these observations, we suggested that epithelioid cells or nonmuscle cells located at the subendothelial space in vivo mainly derive from the endothelium. Evidence of the presence of morphologically altered endothelial cells has also been provided by others when endothelial cells were transiently wounded by mechanical forces using in vitro methods that mimic or closely reflect the effect of mechanical forces acting to wound cells in vivo (McNeil et al., 1989; Acevedo et al., 1993; Villaschi and Nicosia, 1993; McNeil and Steinhardt, 1997).
In this study, we also investigated whether the change observed corresponded with alterations in the expression of vWf-related antigen. We found that after 24 hr the monolayer cells displayed a weak immunoreactivity for vWf and that this immunoreactivity increased considerably when they had acquired an epithelioid morphology with a cobblestone appearance. It is possible that these changes in vWf immunoreactivity, apart from being related to the alterations in shape that the embryonic endothelial cells underwent, could be associated with the partial disassembly and reassembly of the adherent junctions that mediate cell-cell adhesion (Herman et al., 1987; Lampugnani and Dejana, 1997).
In this study, we also found that some epithelioid cells detached from the monolayer became round shaped and acquired mesenchymal characteristics. Interestingly, changes in the shape of mesenchymal cells from round to elongated have been reported by Yang et al. (1999) during smooth muscle differentiation. These changes in cell shape were followed by the synthesis of smooth muscle-specific proteins.
Since our previous in vivo findings emphasized that embryonic aortic endothelial cells transdifferentiate into mesenchymal cells, some of which express SM α-actin (Arciniegas et al., 1989, 2000), we decided to investigate if the epithelioid cells observed in the present study gave rise to mesenchymal cells expressing SM α-actin. We found that some epithelioid cells jointly lose vWf and gain SM α-actin expression (Fig. 8), suggesting conversion to SMCs, whereas the rest of the monolayer cells maintained their vWF immunoreactivity. Thus, our findings demonstrate not only that embryonic endothelial cells may alter their phenotype to an epithelioid phenotype, but also that these epithelioid cells may transdifferentiate into mesenchymal cells expressing SM α-actin.
Several recent studies have indicated that certain mechanical forces such as shear stress and pressure stress, to which parts of the vasculature are exposed apart from modulating endothelial cell morphology, induce the local production and activation of different growth factors that include aFGF, bFGF, PDGF, VEGF, and TGFβ superfamily members (Malek et al., 1993; Reinhart, 1994; Ohno et al., 1995; Cucina et al., 1998; Topper and Gimbrone, 1999; Rhoads et al., 2000). Importantly, some of these growth factors have been implicated in the transdifferentiation process (Arciniegas et al., 1992; Nakajima et al., 1997; Paranya et al., 2001). For example, TGFβ-2 and -3, and bone morphogenetic protein-2 (BMP-2) have been suggested to act synergistically in the transformation of endocardium into mesenchyme during chicken heart early development by regulating cell behavior and the SM α-actin expression (Ramsdell and Markwald, 1997; Boyer et al., 1999a, b; Nakajima et al., 2000). Of particular interest, recent studies have indicated that TGFβ-2 and -3 are sequentially and separately involved in the process of endocardial-mesenchymal cell transformation, TGFβ-2 mediating endothelial cell-cell detachment and TGFβ-3 cell migration and morphological changes (Boyer et al., 1999a). Other studies have indicated that FGF family members also appear to regulate epithelial mesenchymal cell transformation of other cell types (Davidson et al., 2001; Morabito et al., 2001). Thus, the presence of TGFβ and FGF-immunoreactive epithelioid and migrating mesenchymal cells indicates that these growth factors may be involved in the change to an epithelioid phenotype and in the endothelial-mesenchymal transformation. Consistent with this, we also show the in vivo presence of TGFβ-2 and -3 and bFGF-immunoreactive endothelial and migrating mesenchymal cells in the intimal thickening.
Based on the data presented in this paper and in previous studies, we suggest that the origin of the epithelioid cells could be explained in terms of the transformation that some endothelial cells undergo as a response to endogenous production of growth factors like TGFβ that may occur due to culture conditions.
Acknowledgements
The authors gratefully acknowledge María C. Castillo for valuable discussion and María Eugenia Gallinoto for help in preparation of the manuscript.
