Transforming Growth Factor-β-stimulated Endocardial Cell Transformation Is Dependent on Par6c Regulation of RhoA*

Valvular heart disease due to congenital abnormalities or pathology is a major cause of mortality and morbidity. Understanding the cellular processes and molecules that regulate valve formation and remodeling is required to develop effective therapies. In the developing heart, epithelial-mesenchymal transformation (EMT) in a subpopulation of endocardial cells in the atrioventricular cushion (AVC) is an important step in valve formation. Transforming growth factor-β (TGFβ) has been shown to be an important regulator of AVC endocardial cell EMT in vitro and mesenchymal cell differentiation in vivo. Recently Par6c (Par6) has been shown to function downstream of TGFβ to recruit Smurf1, an E3 ubiquitin ligase, which targets RhoA for degradation to control apical-basal polarity and tight junction dissolution. We tested the hypothesis that Par6 functions in a pathway that regulates endocardial cell EMT. Here we show that the Type I TGFβ receptor ALK5 is required for endocardial cell EMT. Overexpression of dominant negative Par6 inhibits EMT in AVC endocardial cells, whereas overexpression of wild-type Par6 in normally non-transforming ventricular endocardial cells results in EMT. Overexpression of Smurf1 in ventricular endocardial cells induces EMT. Decreasing RhoA activity using dominant negative RhoA or small interfering RNA in ventricular endocardial cells also increases EMT, whereas overexpression of constitutively active RhoA in AVC endothelial cells blocks EMT. Manipulation of Rac1 or Cdc42 activity is without effect. These data demonstrate a functional role for Par6/Smurf1/RhoA in regulating EMT in endocardial cells.

Congenital heart defects occur at a rate of almost 1 in 100 live births (1,2) and are responsible for the largest proportion of deaths due to birth defects in the first year of life and remain a major cause of death during childhood (3). The majority of these defects include abnormalities of valve formation. Valvular injury in adults occurs as a result of infection (4), pharmacolog-ical therapy (5), or tumorogenesis (6). Together, these factors result in over 100,000 Americans a year undergoing surgery to repair or replace heart valves (7). Valve formation is initiated in the region between the common atria and ventricle and in the distal aspect of the common outflow tract when the heart is a simple tubular structure composed of two concentric cylinders of epithelia separated by a gel-like matrix termed the cardiac jelly. A critical step in valvulogenesis is the transformation of a subpopulation of endocardial cells from the inner cell layer into mesenchymal cells that invade the cardiac jelly and later contribute to the connective tissue of the valves and septa of the adult heart (8). Transformation occurs at the atrioventricular boundary to initiate formation of the mitral and tricuspid valves and somewhat later in the outflow tract to initiate aortic and pulmonary valve formation. Endocardial cell transformation is dependent on factors derived from the myocardium that diffuse through the cardiac jelly (9). Because only a subset of endothelial cells become mesenchyme, it is important to reveal the mechanisms regulating transformation that balance the ratio of transformed to non-transformed cells. Although much of the later mechanisms of valve maturation are poorly understood, it is clear that cushion mesenchyme is critical to the process of valve formation (8,10).
Transformation in the atrioventricular cushion (AVC) 3 has been studied extensively in avian systems using an in vitro assay in which the AVC is excised and placed on a collagen gel (11). In this assay, transformation can be divided into three steps based on cellular morphology. Endocardial cells separate from the epithelial sheet and elongate in a step termed activation. Next, elongate mesenchymal cells enter the matrix, a step termed invasion. Finally, cells migrate through the gel in the migration step. These three steps, activation, invasion, and migration, constitute EMT. As in the developing heart, EMT is tightly restricted where endocardial cells in AVC explants undergo EMT, whereas endocardial cells in the ventricle do not (9). Transforming cells alter their pattern of gene expression down-regulating molecules such as the endothelial marker platelet endothelial cell adhesion molecule-1 and up-regulating smooth muscle ␣ actin and procollagen type I (12,13). EMT can be quantified by counting the number of cells in the gel, and molecules may be scored for the ability to induce loss-of-function in AVC explants or gain-of-function in ventricular explants.
Recently a novel signaling pathway has been described in which Par6 acts downstream of TGF␤ to control apical-basal polarity in mouse NMuMG cells (30). Par6 is colocalized to the tight junctions with ALK5 by occludin. TGF␤R2 is recruited to the tight junctions by TGF␤ addition and phosphorylates Par6 resulting in the recruitment of Smurf1 (31). Smurf1 ubiquitination of RhoA leads to degradation of RhoA promoting tight junction dissolution and EMT (32). Here we ask if the Par6 signaling pathway regulates EMT in endocardial cells. Using an in vitro model of endocardial cell transformation we demonstrate that ALK5 activity is required for EMT in AVC explants. Overexpression of wild-type Par6 induces EMT in ventricular explants, whereas dominant negative (dn)Par6 blocks EMT in AVC explants. Similarly, overexpression of Smurf1 causes EMT in ventricular endocardial cells. Consistent with a role for Smurf1 in targeting RhoA for degradation, both dnRhoA and siRNA to RhoA caused gain-of-function in ventricular explants, whereas constitutively active (ca)RhoA effectively blocked EMT in AVC explants. Similar manipulations of Rac1 and Cdc42 did not alter EMT, suggesting a selectivity for RhoA in mediating these effects. These data demonstrate that the Par6 pathway is an important mediator of endocardial cell EMT in vitro.

EXPERIMENTAL PROCEDURES
Construction of Adenoviral Constructs-Adenoviruses were generated using the pAdEasy system (33). All concentrated viruses were titered by performing serial dilutions of the concentrated virus and counting the number of GFP-expressing 293 cells after 18 -24 h. The Rho GTPase dominant negative and constitutively active recombinant adenoviruses were prepared and characterized as previously described (34). Titers of virus ranged between 10 9 to 10 14 plaque forming units/ml. Injections were adjusted to achieve infection of 20 -50% of endocardial cells.
Viral Injections and Collagen Gel Assays-Stage 10 -12 chick embryos were harvested onto Whatman paper rings and imme-diately injected with adenovirus in the heart lumen. Injections were performed using 20 -50 pulses, ϳ50 ms each, delivered by a picospritzer. Fast Green Dye (Sigma) was used to monitor injection of the virus. Immediately after injection, the embryos were placed on egg agar and incubated at 39°C. After 24 h, ventricular or AVC explants were excised from the infected hearts and placed in culture as described previously (35) (supplemental Fig. S1, A and B). After 48 h, the explants were fixed in 0.8% formaldehyde, 0.05% glutaraldehyde. GFP expression by virally infected cells was observed on an inverted fluorescent microscope. The phenotype of each GFP-expressing cell was determined and scored as epithelial, activated, or transformed as described (35) (supplemental Fig. S1, C and D).
Ligand and Inhibitor Addition-TGF␤2 was purchased from R & D Systems. Ligand addition occurred 12 h post placement of explants on collagen pads to allow for attachment of explants. SB431542 was purchased from Sigma, and Calbiochem 616541 was purchased from Calbiochem. Inhibitors were added to conditioning media and preincubated 12 h prior to placement of explants on collagen pads, following removal of media at the time of explant placement.
siRNA Treatment of AVC and Ventricular Explants-AVC or ventricular explants were collected in 100 l of M199 medium (Mediatech, Inc.) at room temperature for each treatment condition. For transfection, first 4 l of siPORT NeoFX (Ambion) was incubated in a final volume of 100 l of M199 for 10 min at room temperature. Next, the appropriate final concentration (for 300 l of total volume) of siRNA derived from 21-bp RNA (target sequence) was added to a final volume of 100 l. These two tubes were mixed and incubated for 15 min at room temperature to allow siRNA complexes to form. Next, the 200-l mixture was added to the 100 l containing AVCs or ventricular explants, and this solution was incubated at 37°C, 5% CO 2 for 45 min. Explants were then placed on conditioned collagen gels and incubated under the same conditions for 48 h. After 48 h explants were fixed in 0.8% formaldehyde, 0.05% glutaraldehyde for 5 min at room temperature and washed twice with phosphate-buffered saline. Fixed explants were scored for the number of mesenchymal cells invading the gel. The siRNA target sequence for RhoA was: 5Ј-AATTATGTAGCAGATATT-GAA-3Ј. For control siRNA, a scrambled 21-oligonucleotide template containing the same number of the bases of the RhoA siRNA target that did not blast to any gene in the chicken genome, 5Ј-AGACTGTCGCGTGCTCTGTCC-3Ј was used as previously described (36). RhoA siRNA and control siRNA were a generous gift from R. Runyan (37). Two independent siRNA constructs against each chicken Par6c and Smurf1 (designated Par6c-A, Par6c-B, Smurf1-A, and Smurf1-B) were designed using Silencer Select custom siRNA (Ambion). The siRNA target sequences for Par6c-A were: sense, 5Ј-GCCUC-CAACUCAUUGCAtt-3Ј and antisense, 5Ј-UCUGCAAUGA-GUUGGAGGCga-3Ј; Par6c-B sense, 5Ј-CGCAGUACCUCA-GCAACUAtt-3Ј and antisense: 5Ј-UAGUUGCUGAGGUAC-UGCGag-3Ј; Smurf1-A sense, 5Ј-GCACAAGGGUUUUAAC-GAAtt-3Ј and antisense, 5Ј-UUCGUUAAAACCCUUCUGCag-3Ј; Smurf1-B sense, 5Ј-GGAACUUGAGCUAAUCAUAtt-3Ј and antisense, 5Ј-UAUGAUUAGCUCAAGUUCCtt-3Ј. Control siRNA was used for these experiments as described above.
All siRNA constructs were confirmed for their ability to decrease mRNA measured by real-time PCR in chick embryonic fibroblasts (supplemental Fig. S2). Additional methods are found in the supplemental materials.

ALK5 Kinase Activity Is Required for EMT in Endocardial
Cells-We have previously demonstrated that ALK2 is sufficient and required for endocardial cell EMT, whereas ALK5 is not sufficient (35). Several recent studies have shown that, even if not sufficient, ALK5 activity may be required for TGF␤ signaling (38 -40). To test whether ALK5 kinase activity is required for endocardial cell EMT, we took advantage of the availability of several specific, small molecule inhibitors of ALK5 kinase activity. Two ALK5 inhibitors with different potency and distinct structural characteristics were chosen, SB431542 and Calbiochem 616451. Each inhibitor or vehicle was added to the collagen gel 12 h prior to the placement of AVC explants isolated from Stage 14 chick embryos. Explants were incubated for 48 h and fixed, and the number of transformed cells was quantitated in each group (supplemental Fig.  S1, C and D). Transformed cells were defined as those that enter the collagen gel. Both compounds significantly inhibited the number of cells entering the gel. SB431542 at a final concentration of 2.5 M inhibited the number of cells entering the collagen gel by 71 Ϯ 4% when compared with vehicle ( Fig. 1A). Calbiochem 616451 at a final concentration of 150 nM similarly reduced the number of cells that enter the gel by 77 Ϯ 3% when compared with control (Fig. 1B). These data demonstrate that ALK5 kinase activity is required for AVC endocardial cell EMT.
We have previously determined that TGF␤R3 is sufficient and required for endocardial cell EMT in vitro. Blockade of TGF␤R3 in AVC explants inhibits EMT. Overexpression of TGF␤R3 in non-transforming ventricular endocardial cells, which normally lack TGF␤R3 expression, results in EMT following the addition of TGF␤ ligand (28). To determine if TGF␤R3-mediated EMT requires ALK5 kinase activity, we overexpressed TGF␤R3 in ventricular endothelial cells and incubated explants in the presence or absence of 2.5 M SB431542 plus or minus 250 pM TGF␤2. Stage 10 -12 chick embryos were harvested and injected with adenovirus expressing either GFP alone or TGF␤R3 and GFP (see supplemental Fig. S1). Injected embryos were cultured 18 -24 h, and ventricular explants from embryos between Stages 13 and 15 were excised and placed on collagen gels preincubated with vehicle or SB431542 as above. Twelve hours after explant placement either vehicle or 250 pM TGF␤2 was added, and the incubation continued for an additional 36 h. Explants were fixed and GFPpositive cells scored as epithelial (round cells in a sheet on the surface of the gel), activated (elongated, individual cells on the surface of the gel), or transformed (elongated, individual cells in the collagen gel). Infection with GFP alone defined the basal distribution of cells as epithelial, activated, or transformed (Fig.  1C). The addition of TGF␤ did not alter this distribution. Cells infected with TGF␤R3 and GFP incubated with vehicle had a distribution comparable to cells infected with GFP alone. In contrast, the addition 250 pM TGF␤2 resulted in a significant increase in the percentage of transformed cells (557%) and a concomitant decrease in the percentage of cells scored as epithelial (57%). Incubation with SB431542 prevented the ability of TGF␤2 to stimulate EMT in TGF␤R3-expressing cells. Importantly, incubation with SB431542 did not alter the distribution of cells expressing TGF␤R3 and incubated with vehicle. These data demonstrate that ALK5 kinase activity is required for TGF␤R3-dependent EMT in endocardial cells.
Par6 Is Sufficient and Required for EMT-Par6 has been demonstrated to play a role in TGF␤-stimulated dissolution of tight junctions during EMT (30). Par6 is expressed in several epithelial cell populations in the chick embryo as well as in endocardial cells during embryonic development (data not shown). To test the hypothesis that Par6 activity regulates endocardial cell EMT, we took advantage of our assay where both gain-and loss-of-function experiments can be performed in ventricular and AVC explants, respectively. We chose to overexpress wild-type Par6 in ventricular explants to score for gain-of-function and overexpress dnPar6 in AVC explants to score for loss-of-function. Embryos were injected, and ventricular or AVC explants were harvested and scored as described above. Representative cell phenotypes are depicted (Fig. 2,  A-D). Injection of an adenoviral construct coexpressing wildtype Par6 and GFP into normally non-transforming ventricle  Table S1.
induced a statistically significant 75% increase in transformed cells with a concomitant decrease in epithelial cells (Fig. 2E). When compared with explants infected with GFP alone, infec-tion by dnPar6-GFP adenovirus in normally transforming AVCs resulted in a larger percentage of GFP-positive cells which are epithelial, with a concomitant 35% decrease in the percentage of GFP-positive cells that are transformed (Fig. 2F). These data demonstrate that Par6 is both sufficient and required for endocardial cell EMT in vitro.
As a second independent approach to decrease Par6 activity we delivered siRNA constructs against Par6 to AVC explants. Explants from Stage 14 embryos were harvested. Following incubation with siRNA as described under "Experimental Procedures" the explants were placed on a collagen gel. Explants were fixed and scored after 48 h. As compared with control siRNA, two independent siRNA constructs to Par6 decreased the number of transformed cells by 70% (Fig. 2, G and H). These data are consistent with those obtained using dnPar6 and support a role for Par6 activity in mediating endocardial cell transformation.
Smurf1 Is Sufficient and Required for EMT-Given that ALK5 and Par6 are required for EMT in endocardial cells, we sought to determine if Smurf1, which is activated downstream of Par6, is sufficient for transformation. Adenovirus coexpressing Smurf1 and GFP or GFP alone was injected into the heart of Stage 10 -12 embryos followed by the harvest of both ventricles and AVCs. Following incubation of explants for 48 h, GFPpositive cells were scored as described. Smurf1 overexpression in the ventricle led to a statistically significant gain-of-function (i.e. EMT) as measured by a 300% increase in the number of transformed cells. This was accompanied by a 23% decrease in epithelial cells. Overexpression of Smurf1 did not cause a significant increase in transformation in the AVC. These data demonstrate that Smurf1 can stimulate EMT, consistent with Smurf1 acting downstream of Par6 (Fig. 3A). To determine if Smurf1 activity is required for endocardial cell transformation, we delivered two independent siRNA constructs against Smurf1 to AVC explants as described earlier.
Each construct decreased the number of transformed cells by at least 70% when compared with control siRNA (Fig. 3, B and C). These data demonstrate that Smurf1 is both sufficient and required for endocardial cell EMT in vitro.
RhoA Activity Regulates EMT-The activation of Smurf1 downstream of Par6 results in the targeting of RhoA for degradation (30). Therefore, we used both dn-and caRhoA constructs to determine if the levels of RhoA activity regulate EMT. We chose to overexpress dnRhoA in ventricular explants to score for gain-of-function and overexpress caRhoA in AVC explants to score for loss-of-function. Embryos were injected, and ventricular or AVC explants were harvested and scored as described above. Injection of an adenoviral construct coexpressing dnRhoA and GFP into normally non-transforming ventricle induced a statistically significant 220% increase in transformed cells with a concomitant decrease in epithelial cells (Fig. 4A). Expression of dnRhoA in the AVC had no effect (Fig. 4A). When compared with explants infected with GFP alone, infection by caRhoA-GFP adenovirus in normally trans-forming AVCs resulted in a larger percentage of GFP-positive cells, which are epithelial, with a concomitant 77% decrease in the percentage of GFP-positive cells that are transformed (Fig.  4A). Expression of caRhoA in ventricular explants also decreased the number of transformed cells, in this case by 90% (Fig. 4A). These data demonstrate that a decrease in RhoA activity is sufficient to induce EMT, whereas increased RhoA activity blocks EMT.
As a second independent approach to decrease RhoA activity we delivered siRNA constructs against RhoA to ventricular and AVC explants as described earlier. As compared with control siRNA, addition of siRNA to RhoA results in gain-of-function in ventricular explants as measured by a 690% increase in the number of transformed cells (Fig. 4B). In the AVC, addition of siRNA to RhoA had no effect (Fig. 4C). These data are consistent with those obtained using dnRhoA and support a role for a decrease in RhoA activity in mediating endocardial cell transformation.
These data are not consistent with several studies that target an indirect measure of RhoA activity by targeting the Rho kinase (ROCK1/2) downstream of RhoA using a small molecule inhibitor, Y27632 (41,42). This apparent discrepancy may be due to the use of high concentrations of ROCK inhibitor. The IC 50 for the target is 0.22 M (43), whereas typical studies employ 5-150 M. To address this issue directly we generated a dose-response curve of AVC endothelial cell EMT to Y27632 (Fig. 5A). At a concentration of 0.3 M, just above the reported IC 50 , we noted a statistically insignificant decrease in EMT. Concentrations of 3 M and higher decreased EMT with an EC 50 of ϳ50 M. Because the EC 50 required to significantly inhibit EMT was Ͼ200-fold higher than the reported IC 50 we chose to examine the effects of a low (0.5 M) and high (30 M) concentration on ventricular explants (Fig. 5B). The lower concentration of Y27632 significantly enhanced EMT, consistent with a loss of RhoA being required for EMT. However, at 30 M, Y27632 did not enhance EMT consistent with nonspecific effects that act to inhibit EMT.
Neither Cdc42 Nor Rac1 Are Sufficient or Required for EMT-To address the specificity of the effect on endocardial cell transformation seen by altering RhoA activity, we examined the related small GTPases, Rac1 and Cdc42. Although neither Rac1 nor Cdc42 are substrates for Smurf1 (44) these small GTPases have been shown to coordinately regulate signaling with RhoA (45)(46)(47)(48). Adenovirus coexpressing GFP and ca and dn forms of Cdc42 and Rac1 were injected into the heart lumen, and explants were harvested and scored as described. In ventricular or AVC explants, neither caRac1 nor dnRac1 caused significant differences in the percentage of cells that were epithelial, activated, or transformed (Fig. 6A). In ventricular explants dnCdc42 caused a small increase in the number of transformed cells with no change seen in AVC explants (Fig. 6B). In both ventricular and AVC explants caCdc42 was without effect. Taken together, these data suggest that neither Rac1 nor Cdc42 activity plays a significant role in regulating endocardial cell transformation.

DISCUSSION
Endocardial cell EMT is a necessary step in valvulogenesis in the heart. Using small molecule inhibitors we demonstrate that ALK5 kinase activity is required for TGF␤R3-mediated endocardial cell transformation. We also show that the recently described role of Par6 in regulating EMT in NMuMG cells is operative in endocardial cells. Overexpression of dnPar6 inhibits EMT in AVC explants, whereas wild-type Par6 induces EMT in ventricular endocardial cells. Consistent with Par6 regulating Smurf1 activity, overexpression of Smurf1 induces EMT in ventricular cells. In NMuMG cells, Smurf1 is proposed to target RhoA for degradation and thereby allow dissolution of tight junctions and cell migration (30). Targeting of RhoA activity either by dnRhoA or siRNA supports EMT of ventricular endocardial cells, whereas caRhoA inhibits EMT. Manipulation of Rac1 or Cdc42 activity does not support a role for these molecules. Together, these data support the notion that the Par6/Smurf1/RhoA pathway regulates endocardial cell transformation.
In the canonical TGF␤ signaling pathway, ligand binding activates ALK5 kinase activity followed by the phosphorylation and subsequent nuclear translocation of Smads (49). Our studies reveal a requirement for ALK5 activity. In a variety of systems both Smad-dependent and Smad-independent pathways have been described (50). The deletion of Smad4 clearly identifies TGF␤ responses independent of the canonical Smad signaling pathway (51). TGF␤ can signal via several pathways independent of Smads activation, including RhoA (16 -20), Ras (21), MAPKs (17,19,22,23), and phosphatidylinositol 3kinase/AKt (24), although cross-talk with Smad pathways occur (50). It has previously been shown that dissolution of tight junctions in response to TGF␤ in NMuMG cells occurs via the Par6/Smurf1/RhoA pathway and is disassociated from Smad activation (30). We tested whether this pathway regulates EMT in endocardial cells. NMuMG cells contain prototypic tight junctions, whereas endothelial cells have a modified junctional complex (52). Most notably, desmosomes are absent in the endothelium while gap junctions are present (52). Our results reveal that, despite these differences, the modified complex found in endothelial cells is subject to the same regulatory pathways as those in cells containing prototypic tight junctions. This regulation of endothelial cell junctional complexes is important in cell movement not just in embryonic development but also in angiogenesis and leukocyte extravasation in adults (53,54). The ubiquitination of RhoA by Smurf1 plays a general role in regulating cell motility, and the formation of cell protrusions which promote neurite outgrowth (55) and cell shape change, motility, and invasion in tumor cells (56).
Our data demonstrate that loss of RhoA activity leads to EMT in the ventricle. Some investigators have suggested that  . Cdc42 and Rac1 have no effect on endocardial cell EMT. Average percentage of total GFP-expressing cells scored as epithelial, activated, or transformed. Means are derived from three separate experiments. GFP control adenovirus is used to define basal levels of transformation in both the ventricle and AVC. A, effects of ca-and dnRac1 on EMT. Neither dn-or nor caRac1 induced statistically significant changes in transformed cells. B, effects of ca-and dnCdc42 on EMT. In ventricular explants dnCdc42 induced a small but statistically significant increase in transformed cells. For actual counts and statistical analysis refer to supplemental Table S1.
RhoA up-regulation (37,41,42,57) or RhoA activation and not degradation enhances EMT in vitro (18,37,41,42,57). Several of these studies use an indirect measure of RhoA activity by targeting the Rho kinase (ROCK1/2) downstream of RhoA using a small molecule inhibitor, Y27632. These conflicting data may be due to the use of high concentrations of ROCK inhibitor as the IC 50 for the target is 0.22 M (43), while typical studies employ 5-150 M. Indeed we independently generated a dose-response curve of endothelial cell EMT to Y27632 (Fig.  5A) and determined that at concentrations Ͼ0.5 M there is a linear decrease in cell motility suggestive of a nonspecific effect on cell viability. Further, at a concentration of 0.5 M, Y27632 enhanced EMT in ventricular explants, consistent with inhibition of RhoA activity being required for EMT. However, a concentration of 30 M did not enhance EMT, suggesting that offtarget effects at high concentrations of Y27632 act to inhibit EMT. These data highlight the importance of choosing appropriate concentrations of small molecule inhibitors.
Additionally, contradictory to published reports, it has recently been reported that basal ROCK activity is present at endothelial cellular junctions both in vitro and in situ (58). Inhibition of basal ROCK activity or depletion of ROCK was shown to disrupt the integrity of endothelial barriers (58), suggesting that ROCK inhibition may be necessary for breakdown of junctions for protrusive behavior leading to EMT. Our data are consistent with this report, because loss of ROCK activity is a consequence of inhibition of RhoA activity.
Given the conflicting data concerning RhoA activity in mediating EMT, in addition to the use of ca and dn constructs, we used siRNA to target RhoA in endocardial cells. We obtained a chick-specific siRNA construct reported to reduce AVC transformation in vitro (37). We were unable to demonstrate lossof-function in the AVC but noted a significant gain-of-function in the normally non-transforming ventricle consistent with our results with dnRhoA. Our data provide internally consistent results in gain-and loss-of-function in ventricle and AVC, respectively, concerning RhoA activity and EMT. Overall these data demonstrate that loss of RhoA activity is sufficient for endocardial cell EMT and suggest a role for the Par6/Smurf1/ RhoA pathway in endothelial cells.
Members of the Rho family of GTPases share guanine nucleotide exchange factors, and signaling is often coordinated among these proteins (reviewed in Refs. 59 and 60). For example, RhoA can interact with activated Cdc42 or Rac GTPases via a Cdc42-Rac interaction binding motif (61). There is a large body of evidence suggesting that RhoA and Rac1 play antagonizing roles in cell signaling (45)(46)(47)(48). In the regulation of endothelial cell barriers, RhoA has been shown to increase actomyosin contractility to facilitate the breakdown of cellular junctions, whereas Rac1 stabilizes these junctions (59). Neither the overexpression of ca-nor of dnRac1 had an effect on endocardial cell EMT in vitro, suggesting that Rac1 does not play a role in endocardial cell EMT. We also addressed the role of Cdc42. The addition of dnCdc42 to ventricular explants caused a small but statistically significant increase in the number of transformed cells with no change seen in AVC transformation. However, we did not see any opposing effects, defined as a lossof-function, with the addition of caCdc42 in the AVC. We hypothesize that the effect on transformation seen with dnCdc42 in the ventricle is due to the known role of RhoA binding Cdc42, thus effectively reducing the amount of available RhoA. Taken together these data suggest that Cdc42 does not play a role in endocardial cell EMT and are consistent with a report demonstrating Cdc42 does not regulate endothelial cell permeability (62). Further, neither Rac1 nor Cdc42 are substrates for Smurf1 (44). Because EMT is altered only with the manipulation of RhoA activity, and not Rac1 or Cdc42 activity, these data argue that other Rho GTPases cannot compensate for RhoA in mediating endocardial cell EMT.
Our data confirm our previous findings that the Type III TGF␤ receptor, TGF␤R3, is sufficient and required for endocardial cell EMT in vitro (28). Little is known concerning the signaling pathways downstream of TGF␤R3. We demonstrate that ALK5 kinase activity is required for TGF␤R3-mediated EMT leading us to hypothesize that TGF␤R3 may be upstream of the Par6/Smurf1/RhoA pathway in endocardial cells. The demonstration of a TGF␤R3-linked Par6/Smurf1/RhoA pathway that regulates endocardial cell behavior provides potential novel therapeutic targets for modulating endothelial cell transformation or angiogenesis.