TGFβ Type III and TGFβ Type II receptors have distinct activities during epithelial–mesenchymal cell transformation in the embryonic heart


  • Angelique S. Boyer,

    1. Department of Cell Biology and Anatomy, University of Arizona, Tucson, Arizona
    2. Proctor and Gamble Company, Miami Valley Laboratories, Cincinnati, Ohio
    Search for more papers by this author
  • Raymond B. Runyan

    Corresponding author
    1. Department of Cell Biology and Anatomy, University of Arizona, Tucson, Arizona
    • Department of Cell Biology and Anatomy, University of Arizona, P.O Box 245044, Tucson, AZ 85724
    Search for more papers by this author


During the early stages of heart development, progenitors for the heart valves and septa come from endothelial cells via a developmental process known as “epithelial–mesenchymal cell transformation.” This process is restricted to the atrioventricular (AV) canal and outflow tract portions of the embryonic heart. TGFβ signal transduction pathways play critical roles during epithelial–mesenchymal cell transformation in heart development. Previously, we showed that both TGFβ Type II (TβRII) and Type III (TβRIII) receptors are required to mediate epithelial mesenchymal cell transformation in chick heart. Further, distinct TGFβ2 and TGFβ3 activities correspond to separate components of the embryonic cell transformation process. Studies by others of TGFβ-mediated inhibition of cell proliferation produced a model where TβRIII functions by facilitating TGFβ2 binding to TβRII. In the present study, we provide evidence that TβRIII mediates distinct cellular responses from those of TβRII. Blocking antibody for TβRIII, but not antibody against TβRII, specifically inhibits the endothelial cell–cell separation step. Examination of developmental markers, perturbed by blocking TβRIII antibody, revealed a pattern of expression distinctively different from that of TβRII antibody treatment. These data show that a distinct TβRIII-mediated process is required for endothelial cell–cell separation during epithelial mesenchymal cell transformation. As TGFβ2 mediates endothelial cell–cell separation, the data point to a specific association of TGFβ2 and TβRIII in the cell separation step of epithelial mesenchymal cell transformation. We conclude that distinct TβRII and TβRIII signal transduction pathways mediate epithelial–mesenchymal cell transformation in the heart. © 2001 Wiley-Liss, Inc.


During stage 14–18 in chick heart development (Hamburger and Hamilton, 1951), the primitive epithelium lining the lumen of early heart is activated in the AV canal and outflow tract regions. These endothelial cells then transform into mesenchymal cells and invade the underlying extracellular matrix (ECM) (Markwald et al., 1981). This event is termed an “epithelial–mesenchymal cell transformation” (Hay, 1995). There are at least two distinguishable events that occur during epithelial–mesenchymal cell transformation, endothelial cell activation and subsequent mesenchymal cell invasion. Activation of the endothelium is characterized by a loss of cell–cell contacts, cellular hypertrophy and polarization, and an increased expression of ECM molecules (Bolender et al., 1980; Krug et al., 1985; Crossin and Hoffman, 1991). Activated endothelial cells subsequently undergo morphological changes to become mesenchymal cells and invade the underlying ECM (Boyer et al., 1999a).

Transforming growth factors βs (TGFβs) inhibit cell growth, induce cell proliferation, and induce extracellular matrix synthesis. Our previous work demonstrated a role for TGFβs during AV cushion transformation. Both TGFβ2 and TGFβ3 were found in the heart at the time of cell transformation by RNase protection assay (Potts et al., 1992). A TGFβ3 antisense oligonucleotide inhibited transformation by 80% (Potts et al., 1991). Furthermore, TGFβ2 and TGFβ3 have distinctly different expression patterns during epithelial mesenchymal cell transformation (Barnett et al., 1994; Boyer et al., 1999a). Antibodies against the two TGFβ isoforms also produced inhibitory effects at different stages of epithelial–mesenchymal cell transformation. We concluded that TGFβ2 is critical to the endothelial cell activation process while TGFβ3 is critical to the mesenchymal cell formation process.

TGFβs propagate their signal through three major classes of receptors. Two receptors of TGFβs, TGFβ Type II receptor (TβRII) and TGFβ Type III receptor (TβRIII), are expressed in the AV canal. We showed that TβRII propagates a signal from TGFβ to mediate endothelial cell transformation and mesenchymal cell migration (Brown et al., 1996). Subsequently, we found that TβRIII is required for epithelial mesenchymal cell transformation and is distributed where epithelial–mesenchymal cell transformation takes place in vivo (Brown et al., 1999). Importantly, directed expression of TβRIII enables transformation of endothelia from ventricular explants in the presence of TGFβ2.

In this study, we explore two questions. First, which TGFβ receptor is involved in the initial endothelial cell activation process? Second, does TβRIII mediate a signal transduction pathway distinct from TβRII-mediated signal transduction? Our results demonstrate that TβRIII, but not TβRII, mediates endothelial cell activation, as measured by cell–cell separation. Further, perturbation of TβRIII produces distinguishable cellular responses from perturbation of TβRII. Together, these data suggest that a standard model of TβRIII facilitation of TβRII function (Lopez-Casillas et al., 1993) does not apply to the process of epithelial–mesenchymal cell transformation in the heart.

Abbreviations used:

AV, atrioventricular; ECM, extracellular matrix; GalTase, β1,4-galactosyltransferase; PFA, paraformaldehyde; TGFβ, transforming growth factor beta; TβRII, TGFβ Type II receptor; TβRIII, TGFβ Type III receptor.


TGFβ Type III Receptor Antibody Blocks Endothelial Cell Separation

We showed that TβRIII is required for epithelial–mesenchymal cell transformation during endocardial cushion formation (Brown et al., 1999). TβRIII can induce ventricular endocardial cells to transform in the presence of TGFβ2. In addition, TβRIII blocking antibody inhibits transformed mesenchymal cell migration. Because TGFβ2 is the critical factor for AV endothelial cell–cell separation process (Boyer et al., 1999a), we hypothesized that TβRIII mediates the AV endothelial cell–cell separation process. To test our hypothesis, we treated stage 14 AV explants with both TβRII and TβRIII blocking antibodies. At stage 14 (the endothelial activation stage), TβRII antibody has no effect on the endothelial cell–cell separation process (Fig. 1A,B). Such separation is the first morphological step in embryonic cell transformation (Boyer et al., 1999a; Romano and Runyan, 1999). In contrast, treatment with TβRIII receptor blocking antibody completely inhibited endothelial cell–cell separation (Fig. 1C,D). Perturbation of cell–cell separation was assessed by counting endothelial cells within a defined area. Measurement of endothelial cell density revealed that TβRIII antibody-treated cultures have an 80% higher endothelial cell density compared to control cultures. Endothelial cell density in TβRII antibody-treated cultures was the same as its preimmnue control (Fig. 2).

Figure 1.

Effect of TGFβ Type III receptor antibody on endothelial cell separation. Stage 14 explants were harvested and placed onto collagen gels and treated as described in Experimental Procedures. Representative cultures are shown here after 21 hr of incubation. A: Endothelial cells treated with preimmune IgG corresponding to the TβRII antibody. This culture shows activated and separated endothelial cells. B: Surface of culture treated with TβRII antibody showing activated endothelial cells. C: Surface view of culture treated with preimmune IgG for the TβRIII antibody showing activated endothelial cells. D: Surface view of culture treated with TβRIII receptor antibody showing nonactivated, flattened endothelial cells.

Figure 2.

Effect of TβRIII antibody on endothelial cell density. Stage-14 chick AV canal cultures were prepared and treated with anti-TβRII (Antiβ), preimmune IgG (PIβ), anti-TβRIII (antiK) or its preimmune IgG (PIK). The number of endothelial cells within a given frame (240 × 340 μm) were counted and photographed. These data demonstrate a 55% greater cell density in the presence of TβRIII antibody that reflects the relative lack of cell–cell separation (P < 0.01).

These data indicate that, at least at stage 14, TβRIII is critical for endothelial cell–cell separation while TβRII does not play a significant role in this process. These data fit a previous observation that treatment of stage 14 cultures with TβRIII blocking antibody produced a stronger inhibition of transformation compared to TβRII blocking antibody-treated cultures (Brown et al., 1996, 1999). Comparison of endothelial cell morphology in TβRII and TβRIII antibody treated cultures to TGFβ2 and TGFβ3 antibody treated cultures revealed a correlation between specific ligands and receptors. Both TβRII and TGFβ3 blocking antibodies blocked epithelial–mesenchymal cell invasion but failed to block cell–cell separation (Brown et al., 1996; Boyer et al., 1999a). In contrast, AV explant cultures treated with TGFβ2 antibody showed inhibition of endothelial cell–cell separation as seen here with TβRIII antibody treated cultures (Boyer et al., 1999a). These data indicate that, at the endothelial cell–cell separation step, the ligand required is TGFβ2 and that TGFβ2 propagates a signal through TβRIII. These data fit our previous observation that TβRIII mediated epithelial–mesenchymal cell transformation is TGFβ2 dependent (Brown et al., 1999). As TβRII blocking antibody failed to block cell–cell separation, the paradigm of TβRIII facilitation of TβRII signaling seen previously (Lopez-Casillas et al., 1993) does not appear to be functional in this specific developmental process.

Recent data concerning transcriptional regulation of cell transformation provides additional evidence that TβRII and TβRIII propagate different cell responses to the TGFβ ligands. The zinc finger transcription factor, Slug, mediates endothelial cell–cell separation during epithelial mesenchymal cell transformation (Romano and Runyan, 1999). Slug expression is regulated by TGFβ2 but not TGFβ3 (Romano and Runyan, 2000). In contrast, the homeobox transcription factor, Mox1, is regulated by TRBII, not TβRIII (below). This is consistent with recent experiments showing that TGFβ3, and not TGFβ2, regulates Mox1 (Wendler and Runyan, unpublished data).

TGFB Type III Receptor Blocking Antibody Inhibits Protein Expression Differently From TGFB Type II Receptor Blocking Antibody

During epithelial–mesenchymal transformation, both endothelial cells and mesenchymal cells begin to express a variety of new products including extracellular matrix molecules, extracellular matrix receptors, and new cytoskeletal components (Brown et al., 1996, 1999; Boyer et al., 1999b; Wunsch et al., 1994; Nakajima et al., 1999; Sinning et al., 1994). Immunofluorescent staining of protein markers revealed that pertussis toxin and TβRII affected protein expression differentially in both endothelial and mesenchymal cells (Boyer et al., 1999b). This method of analysis reveals differences in cellular responses that would be inaccessible by RT-PCR or Western blot analysis. Here we examined the perturbation of protein expression in stage-16 AV explants and measured separately changes in endothelial cells and mesenchymal cells after TβRIII antibody treatment. The expression pattern and function of these markers were discussed in Boyer et al. (1999b). Briefly, Mox-1 is a transcription factor; Fibrillin 2, procollagen I, and Tenascin are extracellular matrix molecules; and cell surface β1, 4 galactosyltransferase (GalTase), Integrin a6, and Integrin β1 are extracellular matrix receptors. Figure 3 shows representative changes in protein expression in TβRIII antibody-treated mesenchymal cells. There are three general types of responses to TβRIII antibody treatment: (1) Protein expression is not inhibited by TβRIII treatment (Mox1, Fig. 3A,B). (2) Protein expression is partially inhibited by TβRIII antibody treatment (integrin β1, Fig. 3C,D). (3) Protein expression is completely inhibited by type III receptor treatment (procollagen type I, Fig. 3E,F). A fourth response, an increase in expression, is not shown here but was also seen after TβRII treatment and published in Boyer et al. (1999b).

Figure 3.

Effect of TβRIII blocking antibody protein expression in mesenchymal cells. Stage-16 explants were harvested and placed onto collagen gels. Explants were treated with preimmune IgG for TβRIII antibody (control) (A, C, E) or TβRIII blocking antibody (B, D, F). The explants then were fixed and immunostained for the expression of Mox-1 (A, B), Integrin β1 (C, D), and Procollagen I (E, F). A: Mox-1 expression in control explant. B: Mox-1 expression in mesenchymal cells is unaffected in the presence of blocking TβRIII antibody. C: Integrin β1 expression in control mesenchymal cells. D: Reduced Integrin β1 expression in mesenchymal cells treated with TβRIII antibody. E: Procollagen type I expression in control mesenchymal cells. F: Procollagen type I expression in mesenchymal cells is strongly inhibited by treatment with TβRIII antibody. Bar = 20 μm.

Comparison of protein expression in TβRII antibody-treated cultures (Boyer et al., 1999b) and TβRIII antibody-treated cultures reveals distinct profiles. Data for TβRII receptor inhibition (Boyer et al., 1999b) is presented here in comparison with data obtained using TβRIII antibody (Fig. 4). Several significant differences were seen. For example, Mox1 and fibrillin 2 expression are completely inhibited by TβRII antibody (Boyer et al., 1999b) while TβRIII antibody treatment has no effect on these markers in either endothelial (Fig. 4A) or mesenchymal cells (Fig. 4B). Conversely, TβRIII inhibits Integrin β1 expression while TβRII has no effect on this protein. As seen previously in the comparison of pertussis toxin and TβRII antibody–treated cultures (Boyer et al., 1999b), there are differences in response of endothelial cells (Fig. 4A) and mesenchymal cells (Fig. 4B) when treated with the two receptor blocking antibodies. This is most notable with integrin subunit expression patterns. We note that Tenascin, an extracellular matrix molecule, was dramatically up-regulated during epithelial–mesenchymal cell transformation in the heart (Crossin and Hoffman, 1991) but is largely unaffected by either TβRII or TβRIII perturbation. As this molecule was also unaffected by pertussis toxin treatment (Boyer et al., 1999b), it demonstrates the existence of at least one additional unreported signal transduction pathway during epithelial–mesenchymal transformation in the heart.

Figure 4.

Quantitation of marker protein labeling in the presence of TβRII and TβRIII antibodies. Fluorescent images of immunostained explants were obtained using a confocal microscope as in Figure 3. Through-focus endothelial and mesenchymal cell images were collected at 1-μm intervals and stacked electronically. Immunostaining intensity from each cell was quantified using NIH Image software. The data represent the average immunostaining intensities from a population of cells after treatment. Each cell population constitutes 20 to 40 cells. A measurement of 100% would reflect an unchanged level of marker expression between preimmune and blocking antibody treated cultures. A: Endothelial cells remaining on the surface of the collagen gel show differing levels of inhibition or enhancement of marker expression. Note that no marker responds identically to both blocking antibodies. B: Comparison of mesenchymal cell marker expression after treatment. Two markers, Tenascin and Integrin α6, are unaffected by either blocking antibody. These data show some differences in response between cell types and strong differences in response to TβRII or TβRIII blockade. Data showing TβRII antibody effects on staining intensity in endothelial and mesenchymal cells (all but Integrin β1) were previously published in a different form (Boyer et al., 1999b, fig. 6)

These studies clearly demonstrate distinct roles for TβRII and TβRIII receptors in the heart; but they do not identify potential receptor partners. The interaction of TβRII with a TGFβ Type I receptor (TβRI) is well described (for a review of TGFβ signaling, see Massague and Chen, 2000). We establish here that TβRIII can act independently of TβRII but we cannot formally prove whether TβRIII is interacting with a TβRI or mediates a heretofore-undescribed signal transduction pathway. Taken together, these data demonstrate separate functions of two TGFβ isoforms and two corresponding receptors during epithelial–mesenchymal cell transformation. TGFβ2 is required for endothelial cell separation and its signal is propagated through TβRIII. In contrast, TGFβ3 is required for invasion of mesenchymal cells and its signal is propagated through TβRII. Further studies in this developmental system have the potential to reveal novel aspects of ligand specificity and signal transduction in the TGFβ family of growth factors.


Collagen Gel Assay

AV cushion explants from stage-14 chick embryos (Hamburger and Hamilton, 1951) were placed onto collagen gels according to previously published procedures (Potts et al., 1992; Boyer et al., 1999). After 6 hr of incubation, the explants were treated with preimmune IgG for TβRII antibody (10 μg/ml), affinity-purified TGFβ TβRII antibody (10 μg/ml), preimmune IgG for TβRIII antibody (10 μg/ml), or affinity purified TβRIII antibody (10 μg/ml). Explants then were incubated for an additional 21 hr and fixed in 4% paraformaldehyde (PFA) for 30 min.

Endothelial Cell Density Measurement

Explant cultures prepared as above were visualized with Hoffman Modulation Optics. Micrographs of endothelial cells were taken with a Dage CCD camera and a Scion frame grabber on a Macintosh 7500 computer using NIH Image software. A frame of 240 × 340 μm was placed over each micrograph centered between the edge of the endothelial cells and the edge of myocardium of the explant. The number of endothelial cells within the frame was counted. Three frames were measured per explant and micrographs from 46 explants were counted for each treatment. Statistical analysis was performed using the Student's t-test.


AV canal explants from stage-16 chick embryos were placed onto collagen gels. The explants were treated with preimmune IgG (10 μg/ml) and TβRIII antibody (10 μg/ml) after 6 hr of incubation. Explants were incubated for an additional 21 hr, and fixed in 1% PFA for 30 min. Collagen gels were rinsed extensively in phosphate-buffered saline (PBS) before staining. Characterized primary antibodies used were Mox-1 (mouse monoclonal, 1:200 dilution, gift of Dr. C. Wright), β1, 4 galactosyltransferase (GalTase) (rabbit polyclonal, 1:50 dilution, gift of Dr. B. Shur), fibrillin 2 (mouse monoclonal, 1:200 dilution, gift of Dr. C. Little), procollagen type I (mouse monoclonal, dilution 1:1, from the Developmental Studies Hybridoma Bank, University of Iowa), integrin a6 (mouse monoclonal, dilution 1:1, from the Developmental Studies Hybridoma Bank, University of Iowa), and integrin β1 (mouse monoclonal, dilution 1:1, from the Developmental Studies Hybridoma Bank, University of Iowa). Primary antibody incubation was carried out overnight at 4°C. Cy-5 conjugated anti-rabbit or anti-mouse antibodies (1:200 dilution) were used as secondary antibodies. Stained explants were viewed on a Leica confocal microscope. For each experimental group (preimmune vs. TβRIII antibody), the immunostaining procedures and the settings on the confocal microscope were identical.

Fluorescent micrographs of cells within immunostained explants cultures were obtained using a Leica confocal microscope. Through-focus images were collected at 1-μm intervals and stacked electronically. The intensity of immunoflourescent staining was determined using NIH Image. A threshold pixel value was selected to define the cell borders. A mean intensity for each defined cell was obtained for each treatment. Multiple (20–40) cells were analyzed to obtain the average fluorescent intensity for each treatment.


The authors are grateful to the following labs for providing the antibodies for this study: Drs. Joey Barnett and Christopher Brown for the Type II and Type III receptor antibodies, Dr. Christopher Wright for the Mox-1 antibody, Drs. Berry Shur and Helen Hathaway for the GalTase antibody, and Drs. Charles Little and Brenda Rongish for Fibrillin 2 antibody. The Integrin antibodies were obtained from the Developmental Studies Hybridoma Bank at the University of Iowa (supported by the NICHD). This study is supported by a fellowship from the American Heart Association, Desert/Mountain Affiliate (A.S.B), and NIH grants HL 54986 and HL 63926 (R.B.R).