Primary and immortalized mouse epicardial cells undergo differentiation in response to TGFβ

Authors


Abstract

Cells derived from the epicardium are required for coronary vessel development. Transforming growth factor β (TGFβ) induces loss of epithelial character and smooth muscle differentiation in chick epicardial cells. Here, we show that epicardial explants from embryonic day (E) 11.5 mouse embryos incubated with TGFβ1 or TGFβ2 lose epithelial character and undergo smooth muscle differentiation. To further study TGFβ Signaling, we generated immortalized mouse epicardial cells. Cells from E10.5, 11.5, and 13.5 formed tightly packed epithelium and expressed the epicardial marker Wilm's tumor 1 (WT1). TGFβ induced the loss of zonula occludens-1 (ZO-1) and the appearance of SM22α and calponin consistent with smooth muscle differentiation. Inhibition of activin receptor-like kinase (ALK) 5 or p160 rho kinase activity prevented the effects of TGFβ while inhibition of p38 mitogen activated protein (MAP) kinase did not. These data demonstrate that TGFβ induces epicardial cell differentiation and that immortalized epicardial cells provide a suitable model for differentiation. Developmental Dynamics 237:366–376, 2008. © 2008 Wiley-Liss, Inc.

INTRODUCTION

Fifty-four percent of all cardiovascular disease in the United States effects the coronary arteries (American Heart Association, 2004). A detailed understanding of the cell populations and molecules that regulate coronary vessel development will be required to reveal novel drug targets and therapeutic strategies to direct the repair or remodeling of coronary vessels in adults. The importance of cells derived from the proepicardium (PE) and epicardium (EP) in the formation of the coronary vessels is well established (reviewed in Olivey et al.,2004; Tomanek,2005). During development these cells give rise to both the endothelial and smooth muscle components of the coronary vessels. In chick embryos, the PE arises from mesothelial cells along the caudal border of the pericardial cavity (Ho and Shimada,1978). The PE contacts the heart at the atrioventricular groove and migrates to the heart by means of an extracellular matrix bridge (Nahirney et al.,2003) between the myocardium and the PE. These cells maintain polarity while migrating as an intact epithelium with the luminal surface in contact with the myocardium. In mammals clusters of PE cells detach as vesicles that are transferred to the heart by means of the pericardial fluid (Komiyama et al.,1987; Kuhn and Liebherr,1988). In both avians and mammals, after contacting the myocardium, a fraction of cells undergo epithelial–mesenchymal transformation (EMT) and migrate into the subepicardial space. A subset of these cells continues into the compact zone of the myocardium (Mikawa and Fischman,1992; Poelmann et al.,1993). Coronary vessel formation begins as angioblasts coalesce to form a primitive vascular plexus in the subepicardial space and myocardium. These nascent endothelial tubes form larger vessels to become the coronary arteries and veins that attach to the ascending aorta and the right atrium. Once attached, these vessels recruit PE-derived cells to form the smooth muscle and fibroblast components of the vascular network. Therefore, precursor cells are delivered to the heart by the PE and form coronary vessels by a process of vasculogenesis (Munoz-Chapuli et al.,2002). In zebrafish, epicardial cells are able to reinitiate this developmental program and contribute to the genesis of new coronary vessels in injured myocardium (Lepilina et al.,2006; Poss,2007).

Studies of EMT in explanted PE (Mikawa and Gourdie,1996) and EP (Dettman et al.,1998) have identified factors that regulate EMT. For example, both vascular-derived endothelial growth factor (VEGF) and fibroblast growth factor (FGF) stimulate EMT of epicardial cells in vitro (Morabito et al.,2001). Recently, we showed that transforming growth factor β (TGFβ) induces EMT and smooth muscle differentiation in chick epicardial explants (Compton et al.,2006). TGFβ ligands are abundantly expressed in the developing heart (Akhurst et al.,1990; Pelton et al.,1991; Dickson et al.,1993; Jakowlew et al.,1994; Molin et al.,2003) and are known to play a prominent role in stimulating EMT (Kalluri and Neilson,2003) and smooth muscle differentiation (Owens et al.,2004). Three ligands, TGFβ1, TGFβ2, and TGFβ3 (Roberts and Sporn,1990; Sanford et al.,1997; Hu et al.,1998), bind four cell surface proteins. These include two transmembrane serine/threonine kinase receptors, the type I TGFβ receptor (TGFβR1) and the type II TGFβ receptor (TGFβR2; Lin et al.,1992; Ebner et al.,1993; Bassing et al.,1994). In the canonical signaling pathway (Shi and Massague,2003) ligand binding to TGFβR2 results in recruitment of the TGFβR1, activin receptor-like kinase (ALK) 5, to the complex. The constitutively active kinase of TGFβR2 phosphorylates and activates the kinase domain of ALK5, which subsequently phosphorylates and activates downstream receptor associated (R-) Smads 2 and 3 (Kretzschmar and Massague,1998). These activated R-Smads complex with Smad 4 and translocate into the nucleus to alter gene transcription. TGFβ can activate additional downstream effectors including RhoA (Bhowmick et al.,2001a,2003; Edlund et al.,2002; Masszi et al.,2003; Deaton et al.,2005), Ras (Ward et al.,2002), mitogen activated protein (MAP) kinases (Bhowmick et al.,2001b; Bakin et al.,2002; Xie et al.,2004; Deaton et al.,2005), and PI3K/AKt (Bakin et al.,2000), although the mechanisms by which TGFβ regulates these effectors is less well described. A second class of TGFβ binding proteins contains two transmembrane proteins termed the type III TGFβ receptor (TGFβR3), or betaglycan, and endoglin. Both TGFβR3 and endoglin contain a short, highly conserved intracellular domains with no apparent signaling function (Lopez-Casillas et al.,1991; Wang et al.,1991; Cheifetz et al.,1992). Mutations in the endoglin gene are linked to human hereditary hemorrhagic telangiectasia (McAllister et al.,1994), whereas targeted inactivation of the gene encoding TGFβR3 has been shown to result in embryonic death at E14.5 associated with a failure of coronary vessel development (Compton et al.,2007).

TGFβ induced EMT in both PE and EP explants from chick embryos (Compton et al.,2006; Olivey et al.,2006). Experiments using EP explants cultured with the addition of growth factor, specific small molecule inhibitors, and adenoviral gene transfer demonstrated that TGFβ-stimulated loss of epithelial character was accompanied by smooth muscle differentiation (Compton et al.,2006). These effects of TGFβ are dependent on ALK5 kinase activity, and ALK5 kinase activity is sufficient to induce EMT and invasion of a three-dimensional matrix by epicardial cells. Overexpression of Smad 3 is not sufficient to induce cell invasion. The loss of epithelial character in response to TGFβ requires the activity of the downstream effectors, p160 rho kinase and p38 MAP kinase. Induction of smooth muscle differentiation requires p160 rho kinase but not p38 MAP kinase.

To determine whether TGFβ signaling pathways play a similar role in regulating epicardial cells in mammals, we characterized mouse epicardial cell explants. We have defined culture conditions and demonstrated the ability to use adenoviral gene transfer to introduce genes of interest into mouse explants. We determined that, similar to the chick, the addition of TGFβ1 or TGFβ2 causes the loss of epithelial character and smooth muscle differentiation in mouse epicardial cells. The type I TGFβ receptor ALK5 is both required and sufficient to mediate EMT and smooth muscle differentiation in mouse epicardial cells. Immortalized epicardial cells were generated using a transgenic mouse where the large T antigen is temperature regulated (Jat et al.,1991). Immortalized epicardial cells retain the expression of the epicardial cell marker WT1 and respond to TGFβ in a manner indistinguishable from primary epicardial cells. These data demonstrate that TGFβ regulates epicardial cell differentiation in the mouse and that immortalized epicardial cells may be used as a model system to study smooth muscle differentiation.

RESULTS

TGFβ1 or TGFβ2 Causes the Loss of Epithelial Character and Smooth Muscle Differentiation

Epicardial explants from E11.5 mice were incubated with vehicle, 250 pM TGFβ1 or 250 pM TGFβ2 for 72 hr. We examined responses to both TGFβ1 and TGFβ2, because these two ligands are known to differ in binding affinity for TGFβR2 (Lin and Moustakas,1994). As soon as 24 hr after ligand addition, epicardial explants incubated with TGFβ appeared less epithelial when compared with vehicle. At 72 hr, cells in epicardial explants incubated with vehicle are compact and display a rounded, epithelial phenotype (Fig. 1A). Cells in explants incubated with TGFβ1 or TGFβ2 are elongated and separated from one another (Fig. 1B,C). Immunostaining for zonula occludens-1 (ZO-1), a tight junction protein indicative of an epithelial phenotype, demonstrates abundant staining at 72 hr in vehicle-incubated explants (Fig. 1D) and a significant decrease in explants incubated with TGFβ1 or TGFβ2 (Fig. 1E,F). Similarly, explants incubated with vehicle show a perinuclear staining pattern for cytokeratin and incubation with TGFβ1 or TGFβ2 decreased cytokeratin expression (data not shown). These data demonstrate that TGFβ causes a loss of epithelial character in epicardial cells.

Figure 1.

Transforming growth factor β (TGFβ) induces epithelial–mesenchymal transformation (EMT) in epicardial cells. Epicardial explants were incubated with vehicle, 250 pM TGFβ1 or 250 pM TGFβ2 for 72 hr before fixation. A: Cells in explants incubated with vehicle are compact and display a rounded, epithelial phenotype. B,C: Cells in explants incubated with TGFβ1 or TGFβ2 are elongated and separated from one another. D: Cells in explants incubated with vehicle express zonula occludens-1 (ZO-1) along the cell–cell borders consistent with an epithelial phenotype. E,F: Cells incubated with TGFβ1 (E) or TGFβ2 (F) are elongate and have lost cell–cell contacts and ZO-1 expression. G: Cells in explants incubated with vehicle lack distinct SM22α expression. H,I: Cells incubated with TGFβ1 (H) or TGFβ2 (I) express SM22α in organized fibers consistent with a smooth muscle phenotype. J: Cells in explants incubated with vehicle lack expression of calponin. K,L: Cells incubated with TGFβ1 (K) or TGFβ2 (L) express calponin in a pattern of organized extended fibers consistent with smooth muscle. Original magnification, ×100 in A–C, ×400 in D–L.

Previously, we demonstrated that chick epicardial cells lose epithelial character and undergo differentiation to smooth muscle in response to TGFβ (Compton et al.,2006). Therefore, we examined specific markers of smooth muscle differentiation after incubation with TGFβ (Fig. 1G–L). Explants incubated with vehicle display little expression of SM22α (Fig. 1G). Cells in explants incubated with TGFβ1 or TGFβ2 (Fig. 1H,I) demonstrate increased expression of SM22α found in organized fibers consistent with a smooth muscle phenotype. Similarly, cells in explants incubated with vehicle lack calponin expression (Fig. 1J), whereas incubation with TGFβ1 or TGFβ2 (Fig. 1K,L) increases the expression of calponin. These data demonstrate that TGFβ causes the loss of epithelial character and initiates smooth muscle differentiation of epicardial explants.

ALK5 Kinase Activity Is Required for Epicardial Cell Differentiation

To determine whether ALK5 activity, a downstream molecule of TGFβ, was required for the effects of TGFβ, we incubated the explants on collagen-coated slides with or without the ALK5 kinase inhibitor SB431542. Explants were incubated with 2.5 μM SB431542 in the presence of vehicle, 250 pM TGFβ1 or 250 pM TGFβ2 for 72 hr before fixation. Explants incubated in the presence of SB431542 and vehicle display an epithelial phenotype comparable to cells incubated with vehicle alone (compare Fig. 2A with 1A). In contrast, cells in explants incubated in the presence of SB431542 and TGFβ1 or TGFβ2 did not elongate or separate but retained an epithelial appearance (compare Fig. 2B,C with 1B,C). Consistent with this observation epicardial cells incubated with vehicle, TGFβ1, or TGFβ2 in the presence of SB431542 retain the expression of ZO-1 (Fig. 2D–F). These data demonstrate that inhibition of ALK5 kinase activity prevents loss of epithelial character in response to TGFβ. To determine whether kinase activity is required for smooth muscle differentiation, we examined the expression of the smooth muscle markers SM22α and calponin. Cells in epicardial explants incubated with vehicle and SB431542 fail to express SM22α or calponin (Fig. 2G,J). Cells incubated with TGFβ1 or TGFβ2 in the presence of SB431542 fail to express SM22α or calponin (compare Fig. 2H,I,K,L with 1H,I,K,L). These data demonstrate that ALK5 kinase activity is required for TGFβ-stimulated loss of epithelial character and smooth muscle differentiation.

Figure 2.

Activin receptor-like kinase 5 (ALK5) is required for the effects of transforming growth factor β (TGFβ) in epicardial explants. Epicardial explants were incubated with vehicle, 250 pM TGFβ1 or 250 pM TGFβ2 in the presence of 2.5 μM SB431542, an ALK5 kinase inhibitor, for 72 hr before fixation. A: Epicardial explants incubated with vehicle display an epithelial phenotype. B,C: Cells incubated with TGFβ1 (B) or TGFβ2 (C) in the presence of inhibitor remain round and compact consistent with epithelial phenotype. DF: Cells incubated with vehicle (D), TGFβ1 (E), or TGFβ2 (F) in the presence of inhibitor retain zonula occludens-1 (ZO-1) expression at cell–cell borders consistent with an epithelial phenotype. GI: Cells in explants incubated with vehicle (G), TGFβ1 (H) or TGFβ2 (I) in the presence of inhibitor failed to express the smooth muscle marker SM22α. JL: Incubation of cells with vehicle (J), TGFβ1 (K), or TGFβ2 (L) in the presence of inhibitor failed to express the smooth muscle marker calponin. ×100 in A–C; ×400 in D–L.

TGFβ Effects on Smooth Muscle Differentiation Require p160 rho Kinase but not p38 MAP Kinase Activity

RhoA is a known TGFβ effector that has also been shown to signal smooth muscle differentiation in response to platelet derived growth factor (PDGF; Lu et al.,2001). To address the potential role of RhoA in TGFβ-stimulated loss of epithelial character and smooth muscle differentiation in epicardial explants, we targeted p160 rho kinase, a downstream effector of RhoA. Epicardial explants were harvested and incubated with vehicle, TGFβ1, or TGFβ2 and analyzed as described above. The addition of 10 μg/ml Y27632, a specific p160 rho kinase inhibitor, had no apparent morphological effect on vehicle-incubated explants (Fig. 3A,D,G,J) and did not completely block the loss of epithelial character in response to TGFβ1 or TGFβ2 (Fig. 3A–F). However, Y27632 effectively blocked the expression of SM22α in response to either TGFβ1 or TGFβ2 (Fig. 3, compare G–I with J–L). MAP kinase is a known downstream mediator of TGFβ signaling and p38 MAP kinase has been implicated specifically in TGFβ-induced EMT (Bhowmick et al.,2001b; Bakin et al.,2002). The addition of 4 μM SB202190, a p38 MAP kinase inhibitor, had no discernable effect on explants incubated with vehicle or on the effects of TGFβ1 or TGFβ2 (Fig. S1). These data demonstrate a requirement for p160 rho kinase activity in mediating the expression of smooth muscle markers in response to TGFβ.

Figure 3.

Inhibition of p160 rho kinase blocks transforming growth factor β (TGFβ) -stimulated SM22α expression in epicardial cells. Epicardial explants were incubated with vehicle, 250 pM TGFβ1 or 250 pM TGFβ2 for 72 hr before fixation. AC: Cells incubated with vehicle (A), 250 pM TGFβ1 (B), or 250 pM TGFβ2 (C). TGFβ1 and TGFβ2 cause the loss of zonula occludens-1 (ZO-1) expression at cell–cell borders. DF: Cells were incubated with vehicle (D), 250 pM TGFβ1 (E), or 250 pM TGFβ2 (F) in the presence of 10 μg/ml Y27632, a p160 rho kinase inhibitor. Cells elongate and have diminished cell–cell contacts and ZO-1 expression. GI: Cells incubated with vehicle (G), 250 pM TGFβ1 (H), or 250 pM TGFβ2 (I). Cells incubated with vehicle did not express SM22α (G). Cells incubated with TGFβ1 or TGFβ2 are elongate and display SM22α expression (H,I). JL: Cells preincubated with 10 μg/ml Y27632 before the addition of vehicle (J), TGFβ1 (K), or TGFβ2 (L) lacked SM22α expression. ×400 in A–L.

Immortalized Epicardial Cells Respond to TGFβ

Immortalized epicardial cells were examined for responsiveness to TGFβ-stimulated loss of epithelial character and smooth muscle differentiation. Cells were isolated as described and had been in culture for at least 3 months. Before assay cells were switched from immortomedium to standard growth medium for 24 hr, vehicle or ligand added, and the incubation continued for an additional 72 hr. Cells from E11.5 embryos retain expression of the epicardial marker WT1 (Fig. S2) and form tightly packed epithelia (Fig. 4A,D). The addition of either TGFβ1 or TGFβ2 resulted in elongation and separation of cells and the loss of ZO-1 expression (Fig. 4B,C,E,F). As in primary epicardial explants, the addition of TGFβ1 or TGFβ2 increased the expression of both SM22α and calponin (Fig. 4G–L). To determine whether ALK5 kinase activity is required for the effects of TGFβ, these same measures were performed in cells incubated with 2.5 μM SB431542. The results obtained were comparable to those obtained in primary epicardial explants. The ALK5 kinase inhibitor effectively blocked TGFβ-stimulated loss of epithelial character and the expression of both SM22α and calponin (Fig. 5). Cells incubated in the presence of SB431542 and vehicle display an epithelial phenotype comparable to cells incubated with vehicle alone (compare Figs. 5A and 4A). Similar results were obtained with cells immortalized from E10.5 and E13.5 embryos. Therefore, both cells in primary epicardial explants and immortalized cells respond similarly to TGFβ, and this response requires ALK5 kinase activity.

Figure 4.

Transforming growth factor β (TGFβ) induces epithelial–mesenchymal transformation (EMT) in immortalized epicardial cells. Cells were incubated with vehicle, 250 pM TGFβ1 or 250 pM TGFβ2 for 72 hr before fixation. AC: Cells incubated with vehicle form a compact monolayer and display a rounded, epithelial phenotype. A: Cells incubated with TGFβ1 (B) or TGFβ2 (C) are elongated and separated from one another. DF: Cells incubated with vehicle express zonula occludens-1 (ZO-1) along the cell–cell borders consistent with an epithelial phenotype (D). Cells incubated with TGFβ1 (E) or TGFβ2 (F) are elongate and have decreased cell–cell contacts and ZO-1 expression. GI: Cells incubated with vehicle lack SM22α expression (G). Cells incubated with TGFβ1 (H) or TGFβ2 (I) express SM22α in organized fibers consistent with a smooth muscle phenotype. JL: Cells incubated with vehicle lack expression of calponin (J). Cells incubated with TGFβ1 (K) or TGFβ2 (L) express calponin in a pattern of organized extended fibers in the cytoskeleton. ×100 in A–C; ×400 in D–L.

Figure 5.

Activin receptor-like kinase 5 (ALK5) is required for the effects of transforming growth factor β (TGFβ) in immortalized epicardial cells. Cells were incubated with vehicle, 250 pM TGFβ1 or 250 pM TGFβ2 in the presence of 2.5 μM SB431542, an ALK5 kinase inhibitor, for 72 hr before fixation. AC: Cells incubated with vehicle form a compact monolayer and display a rounded, epithelial phenotype (A). Cells incubated with TGFβ1 (B) or TGFβ2 (C) remain round and compact, consistent with an epithelial phenotype. DF: Cells incubated with vehicle express zonula occludens-1 (ZO-1) along the cell–cell borders (D). Cells incubated with TGFβ1 (E) or TGFβ2 (F) retain ZO-1 expression. GI: Cells incubated with vehicle did not express SM22α, a smooth muscle marker (G). Cells incubated with TGFβ1 (H) or TGFβ2 (I) fail to express SM22α in response to TGFβ. JL: Cells incubated with vehicle (J), TGFβ1 (K), or TGFβ2 (L) fail to express calponin. Original magnification, ×100 in A–C, ×400 in D–L.

To compare the potency of TGFβ1 and TGFβ2 in inducing smooth muscle gene expression, E10.5 Sm22α-lacZ::Immorto epicardial cells were incubated with TGFβ1 or TGFβ2 and monitored for lacZ activity (Fig. 6A,B). Incubation with concentrations from 125 to 1250 pM TGFβ1 or TGFβ2 demonstrated a dose-dependent increase in lacZ activity with a similar maximal response and effective concentration for 50% maximal response (Fig. 6C). These data demonstrate similar potency for TGFβ1 or TGFβ2 in inducing SM22α gene expression.

Figure 6.

Transforming growth factor (TGF) induces differentiation and Sm22α-lacZ expression in immortalized epicardial cells. Sm22α-lacZ::Immorto epicardial cells were isolated from embryonic day (E) 10.5 embryos and assayed for lacZ activity. A,B: Cells incubated with 250 pM TGFβ1 (A) or 250 pM TGFβ2 (B) for 72 hr and stained for lacZ activity. Cells underwent transformation and displayed a flattened smooth muscle phenotype. Vehicle-incubated cells displayed very low levels of single cell staining (not shown). C: To determine the dose dependence of Sm22α-lacZ, expression cells were incubated with increasing concentrations of TGFβ1 (triangles) or TGFβ2 (squares) for 72 hr in triplicate. Sm22α-lacZ activity was assayed with Galacto-light Plus assay. Luminescence readings were normalized by dividing the relative light unit readings for each sample by the mean value for the corresponding mock incubated cells. Data points presented are mean fold induction of triplicate samples ± SEM. Vehicle-incubation readings were no different from mock incubation.

ALK5 Activity Is Sufficient to Induce Loss of Epithelial Character and Smooth Muscle Differentiation

To determine whether ALK5 activity is sufficient to induce differentiation of epicardial cells, we infected immortalized epicardial cells from E11.5 embryos with adenovirus encoding either constitutively active (ca) ALK5 and green fluorescent protein (GFP) or GFP alone. The titer of the adenovirus was adjusted so as to infect only a fraction of the epicardial cells to allow for the scoring of individual GFP-positive cells. Overexpression of caALK5 resulted in loss of ZO-1 from the membrane and a concomitant detachment of the cell from the epithelial sheet (Fig. 7). Overexpression of caALK5 also induced SM22α expression (data not shown). Quantitation of the percent of cells that undergo transformation after infection with adenovirus coexpressing caALK5 and GFP or GFP alone in immortalized cells revealed that caALK5 expression was sufficient to cause loss of epithelial character. Only 20% of cells expressing GFP only were scored as transformed, whereas 80% of cells expressing caALK5 were transformed (Fig. 7). These results demonstrate that ALK5 kinase activity is sufficient to induce both loss of epithelial character and smooth muscle differentiation in epicardial cells.

Figure 7.

Activin receptor-like kinase 5 (ALK5) is sufficient to induce the effects of transforming growth factor β (TGFβ) in immortalized cells. Cells were infected with adenovirus expressing green fluorescent protein (GFP) or GFP and constitutively active (ca) ALK5. A,B: Photomicrographs of infected cells. GFP-positive cells express zonula occludens-1 (ZO-1) at cell–cell borders (A). Cells overexpressing caALK5 show absent or discontinuous ZO-1 expression (B). C: Quantitation of the percent of cells that undergo transformation after infection with adenovirus. Data are plotted as the mean percent ± SEM of three independent experiments (P < 0.05). Original magnification, ×400 in A,B.

Smooth Muscle Differentiation Requires p160 rho Kinase

Experiments next addressed the role of p160 rho kinase and p38 MAPK in mediating the effects of TGFβ in immortalized epicardial cells. The addition of the specific p160 rho kinase inhibitor Y27632 to immortalized epicardial cells had actions comparable to that seen in primary explants. Y27632 did not completely block the loss of epithelial character in response to TGFβ1 or TGFβ2 (Fig. S3A–F). However, as in primary explants, Y27632 effectively blocked the expression of SM22α in response to either TGFβ1 or TGFβ2 (Fig. S3G–L). The addition of 4 μM SB202190, a p38 MAP kinase inhibitor, had no discernable effect on explants incubated with vehicle or on the effects of TGFβ1 or TGFβ2 (Fig. S4). These data demonstrate comparable requirements for p160 rho kinase activity in mediating the expression of smooth muscle markers in response to TGFβ in both primary and immortalized epicardial cells.

DISCUSSION

Here, we demonstrate that TGFβ1 and TGFβ2 induce loss of epithelial character and the appearance of smooth muscle markers in mouse epicardial cells. To facilitate the study of epicardial cell differentiation, we generated immortalized epicardial cells and determined that immortalized cells respond to TGFβ1 and TGFβ2 in a manner indistinguishable from primary cells. ALK5 kinase activity is both required and sufficient for the effects of TGFβ. Inhibition of p160 rho kinase activity blocks the effects of TGFβ on smooth muscle differentiation, whereas the inhibition of p38 MAPK activity is without effect. These data demonstrate that the effects of TGFβ on mouse epicardial cells are similar to those we described in chick epicardial cells (Compton et al.,2006). Our observations support a role for TGFβ in the regulation of epicardial cell differentiation during coronary vessel development.

These data are consistent with the well-described actions of TGFβ in mediating EMT in several experimental systems and cell types (Kalluri and Neilson,2003). However, Morabito et al. (2001) noted that addition of TGFβ2 or TGFβ3 to epicardial explants did not support invasion of cells into a collagen matrix, whereas TGFβ1 weakly stimulated invasion. All TGFβ isoforms inhibited FGF2 and heart conditioned media-stimulated EMT. Our experiments in chick, and now mouse, demonstrate that TGFβ1 and TGFβ2 cause the loss of epithelial character as measured by altered morphology and ZO1 expression. These effects are blocked by inhibition of ALK5 kinase activity, while overexpression of caALK5, which activates the canonical TGFβ signaling pathway, results in the loss of epithelial character. Together, these data suggest an important role for TGFβ in supporting epicardial cell EMT.

In both primary and immortalized cells, the addition of TGFβ induces loss of epithelial character and smooth muscle differentiation in the majority of the cells examined, suggesting that most, if not all, epicardial cells have the capacity to assume a smooth muscle cell phenotype. In vivo, many epicardial cells remain epithelial and only a fraction of cells undergo EMT, invade the subepicardial matrix and the myocardium, and differentiate into smooth muscle cells. This finding suggests that the restriction of TGFβ ligand availability to the epicardium may be an important mechanism to regulate EMT and smooth muscle cell differentiation. Although TGFβ ligands are abundantly expressed in the myocardium as well as the epicardium in vivo (Akhurst et al.,1990; Pelton et al.,1991; Dickson et al.,1993; Jakowlew et al.,1994; Molin et al.,2003) the ligands are made as inactive precursors that can be stored in the extracellular matrix and later activated (Annes et al.,2003). Therefore, despite the observation that mRNA expression of the ligands is discretely localized, the protein is found throughout the embryo in the extracellular matrix (Ghosh and Brauer,1996) consistent with the localized activation of ligand being an important regulatory event.

The effects of TGFβ in both epicardial explants and immortalized cells require the activity of both ALK5 and p160 rho kinase. RhoA has been implicated as a downstream mediator of TGFβ in multiple systems. Both pharmacological and genetic approaches in nontransformed murine mammary epithelial cells reveal a requirement for RhoA/p160 rho kinase in mediating TGFβ-stimulated EMT (Bhowmick et al.,2001b). In LLC-PK1 cells, a proximal tubule epithelial porcine cell line, RhoA signaling downstream of TGFβ stimulates EMT accompanied by up-regulation of smooth muscle α-actin (SMαA) gene expression (Masszi et al.,2003). Although EMT was not measured directly, both long- and short-term actin reorganization is stimulated by TGFβ in a RhoA-dependent manner in human prostate carcinoma cells (Edlund et al.,2002). Our data in mouse epicardial cells also support a role for RhoA in signaling loss of epithelial character downstream of TGFβ.

TGFβ is important in recruiting undifferentiated mesenchyme into the smooth muscle cell lineage during blood vessel assembly and remodeling (Grainger et al.,1998; Darland and D'Amore,2001; Owens et al.,2004). TGFβ or conditioned medium from endothelial cells up-regulates smooth muscle myosin, SM22α, and calponin in embryonic 10T1/2 cells (Hirschi et al.,1998). Up-regulation of smooth muscle proteins by endothelial cell conditioned medium was blocked by preincubation with neutralizing antisera to TGFβ. Subsequent studies showed TGFβ-mediated induction of the late phenotypic marker smooth muscle γ-actin by means of serum response factor (Hirschi et al.,2002). TGFβ also induces smooth muscle cell differentiation in both embryonic stem cells (Sinha et al.,2004) and neural crest stem cells (Chen and Lechleider,2004). Using explanted quail hearts as an in vitro model of coronary vasculogenesis, TGFβ was found to inhibit endothelial cell tube formation, a result consistent with TGFβ inducing smooth muscle cell differentiation at the expense of tube formation (Holifield et al.,2004). Although targeted deletion of the gene encoding TGFβR3 results in a failure of coronary vessel development, smooth muscle recruitment to the nascent vessels that do form appears to occur normally (Compton et al.,2007). Because the null mice die during the time of smooth muscle recruitment, a role for TGFβR3 during later stages of coronary vascular smooth muscle recruitment or differentiation cannot be excluded. However, at least the initial stages of coronary smooth muscle differentiation and recruitment may be independent of TGFβR3.

Both PDGF -BB- and TGFβ-stimulated smooth muscle differentiation requires the activity of rhoA and p160 rhokinase (Lu et al.,2001; Compton et al.,2006). RhoA and p160 rho kinase regulate the expression of two major smooth muscle marker genes, SM22α and smooth muscle α actin (SMαA), in rat thoracic aorta smooth muscle cells (Mack et al.,2001). In rat pulmonary artery smooth muscle cells, RhoA signals by means of both p160 rho kinase dependent and independent pathways to activate smooth muscle gene expression (Deaton et al.,2005). A chemical inhibitor of p160 rho kinase was incubated with quail PE before isochronic transplant into chick embryos to directly address the role of RhoA in the developing coronary vasculature (Lu et al.,2001). Although some quail cells entered the subepicardial matrix, none differentiated into smooth muscle cells or contributed to the coronary vasculature. Our data in both chick (Compton et al.,2007) and mouse are consistent with a RhoA and p160 rho kinase dependent pathway downstream of TGFβ in epicardial cells that regulates smooth muscle gene expression. p38 MAP kinase has also been implicated in TGFβ-stimulated EMT (Bhowmick et al.,2001b; Bakin et al.,2002; Deaton et al.,2005). TGFβ-stimulated EMT in nontransformed murine mammary epithelial cells is dependent on both RhoA and p38 MAP kinase activity leading to the suggestion that p38 MAP kinase might function downstream of RhoA (Bhowmick et al.,2001a; Bakin et al.,2002; Deaton et al.,2005). However, experiments using a dominant negative RhoA or a p160 rho kinase inhibitor showed that p38 MAP kinase activity is independent of RhoA activity (Bakin et al.,2002). Studies of TGFβ regulation of actin cytoskeleton mobilization in human prostate carcinoma cells showed that p38 MAPK functions in a pathway parallel to RhoA and downstream of Cdc42 (Edlund et al.,2002). Therefore, several lines of evidence support a model where RhoA and p38 MAP kinase are in separate pathways downstream of TGFβ. Our data, although implicating RhoA in smooth muscle differentiation, fails to identify a role for p38 MAPK in either epicardial cell EMT or smooth muscle differentiation.

Our data demonstrate that TGFβ signaling by means of the canonical type I receptor induces loss of epithelial character and smooth muscle cell differentiation in both primary and immortalized epicardial cells. ALK5 activity is both required and sufficient for the effects of TGFβ. Smooth muscle differentiation in epicardial cells requires p160rho kinase activity but not p38 MAP kinase activity. These observations suggest that TGFβ plays an important role in the recruitment and differentiation of epicardial cells into coronary smooth muscle cells. Our characterization of immortalized epicardial cells with properties comparable to primary cells suggests that these cells may be an important experimental tool to probe epicardial cell differentiation. The generation of immortalized epicardial cells from transgenic mice with epicardial or coronary vessel defects will provide a powerful approach to explore the role of specific genes in epicardial cell behavior.

EXPERIMENTAL PROCEDURES

Epicardial Explant Culture

Mouse hearts (E10.5, E11.5, and E13.5) were harvested in Hanks buffered salt solution (HBSS) and cultured by modification of the method of Compton (Compton et al.,2006). Embryonic hearts were placed dorsal side down on collagen culture slides (BD Bioscience, Bedford, MA) covered with explant medium (M199, 5% heat inactivated fetal bovine serum (FBS) and 1:400 antibiotic/antimycotic) and cultured at 37°C, 5% CO2. After 12–15 hr, the hearts were removed to reveal epicardial explants as monolayers attached to the collagen coated surface. Explants were washed twice with phosphate buffered saline (PBS), covered with explant medium, and cultured for up to 72 hr. Vehicle (TGFβ - 0.1% Bovine Serum Albumen in 4 mM HCl; inhibitors - DMSO) or ligand was added to the explants immediately after removal of the heart where indicated.

Immortalized Epicardial Explant Culture

To generate inducible immortalized epicardial cell line wild type or Sm22α-lacZ mice were crossed with the ImmortoMouse line (Jat et al.,1991). ImmortoMouse contains an interferon inducible temperature sensitive large T antigen that renders derived cells conditionally immortalized at 33°C in the presence of interferon gamma. Mouse hearts at E10.5, E11.5, and E13.5 were placed in culture and epicardial cells derived as above. Cells were propagated at 33°C in DMEM 10% heat inactivated FBS, insulin-transerrin-selenium (Biosource) and 10 units/ml mouse gamma interferon (Peprotech). For differentiation cells were transferred to standard M199 media without interferon as described and cultured at 37°C for 24 hr before the 72-hr differentiation protocol.

Immunohistochemistry

For SM22α (Abcam) and calponin (Sigma) staining, explants were fixed with 2% paraformaldehyde (PFA) for 30 min at room temperature and permeabilized with PBS and 0.1% Triton X-100 for 5 min. Explants were fixed in 70% methanol before staining for ZO-1. Adenovirus infected epicardial explants for ZO-1 staining were fixed in 2% PFA and permeabilized with 0.2% Triton X-100 for 5 min. Explants for calponin and ZO-1 staining were blocked with 2% bovine serum albumin in PBS for 1 hr and incubated with dilute primary antibody (calponin, 1:400; ZO-1, 2 μg/ml) overnight at 4°C. ZO-1 staining for adenoviral-infected explants was incubated with the primary antibody for 5–6 hr. Explants for SM22α staining were blocked with 5% horse serum, and incubated with primary antibody (SM22α, 1:200) overnight at 4°C. Primary antibody detection was with goat anti-mouse cy3 (calponin), goat anti-rabbit cy3 (ZO-1), or donkey anti-goat cy3 (SM22α) secondary antibody (1:800; Jackson ImmunoResearch). Nuclei were stained with 4′,6-diamidino-2-phenylinodole (DAPI; Sigma). Photomicrographs were captured with Nikon Eclipse TE2000-E microscope and QED imaging software or Nikon microscope and camera with 160T film (Kodak).

Growth Factor or Inhibitor Addition

Growth factors (TGFβ1, 250 pM; TGFβ2, 250 pM) or small molecule inhibitors (SB431542, 2.5 μM; Y27632, 10 μg/ml; SB202190, 4 μM) were added to the medium immediately after removal of the hearts. Explants incubated with inhibitor and growth factor on collagen-coated chamber slides were preincubated with medium containing inhibitor alone for 1 hr at 37°C. Afterward, fresh medium and inhibitor were added to explants and incubation continued. Reagents were obtained from the following sources: TGFβ1 & TGFβ2 from R&D Systems, SB202190 & Y27632 from Calbiochem, and SB431542 from Sigma-Aldrich.

Adenoviral Infection of Explants

Mouse hearts (E11.5) were harvested, rinsed in HBSS, and incubated at 37°C for 30 min with approximately 107 PFU adenovirus (GFP alone or caALK5 and GFP; Desgrosellier et al.,2005). Hearts were then placed on collagen culture slides and cultured as above. GFP-positive cells on collagen culture slides were digitally photographed using the Nikon epifluoresence microscope and QED Imaging software.

Scoring of Cell Transformation

Immortalized epicardial cells were plated on collagen-coated chamberslides at a density of 150,000 cells/well in medium M199 with 5% fetal calf serum (FCS). Adenovirus expressing GFP or coexpressing caALK5 and GFP at a titer of 107 PFU was added and the incubation continued at 37°C for 72 hr. Cells were fixed in 70% methanol and immunostained for ZO-1 as described. GFP-expressing cells in random fields were scored as either expressing ZO-1 in a continuous pattern at cell margins, rounded, and in the monolayer (epithelial) or expressing ZO-1 in a discontinuous pattern at cell margins, elongated, and detached from the cell monolayer (transformed). A minimum of 100 cells in both the GFP and GFP/caALK5 groups were scored and the percentage of epithelial and transformed cells determined (Compton et al.,2006). The experiment was performed three times, the mean percentages determined, and analyzed by students t-test.

Measurement of lacZ Staining and Activity

Immortalized E10.5 Sm22α-lacZ::Immorto epicardial cells were plated in uncoated 24-well tissue culture dishes at a density of 150,000 cells/well in medium M199 with 5% FCS. After 12 hr, growth factor (TGFβ1, TGFβ2, at 125, 250, 625, 1,250 pM) was added and cells incubated for 72 hr. One well was fixed with 2% PFA for 10 min and stained for beta galactosidase visualization with X-gal. Cells were incubated overnight at 37°C in X-gal stain solution (2 mM MgCl2, 5 mM potassium ferricyanide (Sigma P-3667), 5 mM potassium hexacyanoferrate(II)trihydrate (Sigma P-9387), 0.01% Np-40, 0.1% sodium deoxycholate (Fisher BP349), 0.1% X-gal(5-bromo-4-chloro-3-indolyl-b-D galactopyranoside; RPI B718001) in Dulbecco's phosphate buffered saline (pH 8.0). Additional wells in triplicate for each concentration were assayed for beta-galactosidase activity with the Galacto-Light Plus system (Applied Biosystems) according to company protocol. Luminescence was quantitated on a Turner 20/20 Luminometer. The quantitation experiment was repeated twice with comparable results.

Acknowledgements

The authors thank members of the Barnett laboratory for helpful discussions and comments. J.V.B. acknowledges the support of the Vanderbilt-Ingram Cancer Center. J.V.B., A.F.A., and L.A.C. were funded by the NIH. J.V.B. was funded by the American Heart Association.

Ancillary