ROCK1 Expression is Regulated by TGFβ3 and ALK2 During Valvuloseptal Endocardial Cushion Formation

Authors


Abstract

During early heart development at the looped heart stage, endothelial cells in the outflow tract and atrioventricular (AV) regions transform into mesenchyme to generate endocardial cushion tissue. This endocardial epithelial–mesenchymal transition (EMT) is regulated by several regulatory pathways, including the transforming growth factor-beta (TGFβ), bone morphogenetic protein (BMP), and Rho-ROCK pathways. Here, we investigated the spatiotemporal expression pattern of ROCK1 mRNA during EMT in chick and examined whether TGFβ or BMP could induce the expression of ROCK1. At the onset of EMT, ROCK1 expression was up-regulated in endothelial/mesenchymal cells. A three-dimensional collagen gel assay was used to examine the mechanisms regulating the expression of ROCK1. In AV endocardium co-cultured with associated myocardium, ROCK1 expression was inhibited by either anti-TGFβ3 antibody, anti-ALK2 antibody or noggin, but not SB431542 (ALK5 inhibitor). In cultured preactivated AV endocardium, TGFβ3 protein induced the expression of ROCK1, but BMP did not. AV endothelial cells that were cultured in medium supplemented with TGFβ3 plus anti-ALK2 antibody failed to express ROCK1. These results suggest that the expression of ROCK1 is up-regulated at the onset of EMT and that signaling mediated by TGFβ3/ALK2 together with BMP is involved in the expression of ROCK1. Anat Rec, 291:845-857, 2008. © 2008 Wiley-Liss, Inc.

During early avian and mammalian heart development, one of the most critical events is the “remodeling of the heart chambers,” during which a single heart tube is remodeled to a four-chambered heart. Endocardial cushion tissue established in the outflow tract (OT) and atrioventricular canal (AV) regions of the single tubular heart plays a major role in septation and valvulogeneis (Markwald et al., 1975). At the early looped heart stage, the heart consists of inner endocardium and outer myocardium that are separated by an expanded thick extracellular matrix called cardiac jelly (Davis, 1924). As development proceeds, some of the endocardial cells lining the OT and AV regions are activated by signaling molecules secreted by the adjacent myocardium and undergo epithelial–mesenchymal transformation (EMT, Runyan and Markwald, 1983; Krug et al., 1987). Subsequently, transformed mesenchymal cells invade the thick cardiac jelly to form mesenchymal tissue called endocardial cushion tissue, the precursor to the valves and septa of the adult heart. Impaired development of endocardial cushion tissue causes several congenital heart defects such as atrioventricular septal defect, ventricular septal defect, and conotrucal malformations (Sakabe et al., 2005). During EMT, activated endothelial cells initially show cell–cell separation, then transformation to the mesenchymal phenotype and subsequently invasion into the extracellular matrix (Markwald et al., 1984). Extensive studies using a three-dimensional collagen culture model of the AV region as well as null-mutants of mouse embryos have revealed that transforming growth factor-beta (TGFβ) and bone morphogenetic protein (BMP) and their cognate receptors play an obligate role in the regulation of EMT (Potts and Runyan, 1989; Potts et al., 1991; Nakajima et al., 1994, 1998; Sanford et al., 1997; Boyer et al., 1999; Yamagishi et al., 1999; Lai et al., 2000; Kim et al., 2001; Camenisch et al., 2002; Sugi et al., 2004; Desgrossellier et al., 2005; Wang et al., 2005; Ma et al., 2005; Rivera-Feliciano and Tabin, 2006). During chick endocardial cushion formation, TGFβ2 is expressed in the AV endocardium before the onset of EMT (Boyer et al., 1999), and TGFβ2 and -3 in activated endothelial cells as well as in invading mesenchymal cells. TGFβ2 and TGFβ3 blocking antibodies inhibit endothelial activation and transformation, respectively (Camenisch et al., 2002). During chick endocardial EMT, both TGFβ type I receptor (activin receptor-like kinase [ALK] 5) and BMP type I receptor (ALK2) are expressed in AV endocardial cells (Lai et al., 2000). Prototypical ALK5 interacts with the TGFβ type II receptor (TβRII) and mediates several TGFβ-induced responses, and ALK2 interacts with TβRII as well as BMP type II receptors in the presence of the respective ligands (Feng and Derynck, 2005). Studies in chick and mouse suggest that ALK2 plays a critical role in the regulation of EMT during the cushion tissue formation (Lai et al., 2000; Desgrossellier et al., 2005; Wang et al., 2005).

The small GTPase Rho proteins, which include RhoA, RhoB, and RhoC, act as intracellular molecular switches controlling a variety of signal transduction pathways by the cycling of GDP-bound inactive and GTP-bound active forms (Aelst and Symons, 2002; Riento and Ridley, 2003; Raftopoulou and Hall, 2004; Loirand et al., 2006). GTP-Rho activates its downstream effectors, the Rho kinases (ROCK1 and ROCK2), which have been found to regulate actin organization, thereby controlling a wide range of fundamental cell functions such as proliferation, contraction, and motility (Matsui et al., 1996; Loirand et al., 2006). We and others reported that the Rho-ROCK pathway plays an important role in the regulation of endocardial cushion EMT, and that ROCK regulates migration/invasion during EMT (Zhao and Rivkees, 2004; Sakabe et al., 2006; Tavares et al., 2006). The expression of RhoA, an upstream activator of ROCK, is regulated by TGFβ during EMT (Travers et al., 2006). However, the mechanisms responsible for the expression of ROCK are largely unknown (Loirand et al., 2006).

In the present study, we examined the spatiotemporal expression of ROCK1 mRNA during endocardial cushion EMT, and whether TGFβ3 or BMP was capable of stimulating ROCK1 expression in cultured AV endocardium. In situ hybridization and reverse transcriptase-polymerase chain reaction (RT-PCR) showed that the expression of ROCK1 was up-regulated in the activated AV endothelial and invading mesenchymal cells. In AV endocardium co-cultured with associated myocardium, anti-TGFβ3 antibody, anti-ALK2 antibody or noggin, but not SB431542 (ALK5 inhibitor) inhibited the expression of ROCK1. In cultured preactivated AV endocardium, TGFβ3, but not BMP, induced the expression of ROCK1 and this TGFβ3-inducible expression of ROCK1 was inhibited by anti-ALK2 antibody. Our findings suggest that signaling mediated by TGFβ3 and ALK2 together with BMP plays a role in the expression of ROCK1 required for mesenchymal invasion/migration during the formation of endocardial cushion tissue.

MATERIALS AND METHODS

Chick Embryos

Fertilized eggs from domestic fowl (Gallus gallus) were incubated for 2–3 days at 37.8°C and 90% humidity. Embryos were collected in ice-cooled phosphate-buffered saline (PBS) and staged according to the criteria of Hamburger and Hamilton (1992). The staged embryos were subjected to the experiments described below.

In Situ Hybridization

To prepare a ROCK1 probe, cDNA from nucleotide position 5058 to 6019 of chick ROCK1 cDNA (Genbank accession no. XM_419151) was amplified from a stage 18 chick heart library. The sequences for each primer set were 5′-TGAGCTACAGATGCAGCTGGA-3′ (forward) and 5′-AGCAATCTTAACCCTGAAGCC-3′ (reverse).

The PCR product was cloned into the pGEM-T vector (Promega) and sequenced to confirm its identity. Digoxigenin (DIG) -labeled single-strand RNA was prepared using a DIG-RNA labeling kit (Roche) according to the manufacturer's instructions. Chick ROCK1 cDNA subcloned into pGEM-T was linearized using ApaI and then transcribed using SP6 RNA polymerase to construct an antisense probe. A sense probe was also generated using SalI and T7 RNA polymerase.

Section in situ hybridization was performed as described by Yamagishi et al. (1999). Embryos were fixed using 4% paraformaldehyde in PBS and embedded in paraffin. Sections were cut at 5 μm and mounted on MAS-coated slides (Matsunami Glass). Sections were deparaffinized, hydrated, refixed with 4% paraformaldehyde in 0.1 M PB (phosphate buffer) for 10 min, rinsed with 0.1 M PB, and digested with proteinase K (Roche, 0.3 unit/ml) in 0.01 M Tris HCl for 5 min at 37°C. The sections were then acetylated with 0.25% acetic anhydride in 0.1 M triethanolamine (pH 8.0) for 10 min. The sections were dehydrated with a graded series of ethanol, air-dried, and then hybridized with a DIG-labeled probe (0.5 μg/ml in hybridization buffer) at 55°C for 14–16 hr in a moist chamber. After hybridization, sections were rinsed with 4× standard saline citrate (SSC) followed by 50% formamide/2× SSC for 30 min at 50°C, 2 × SSC for 20 min at 50°C and 0.2 ×SSC for 20 min at 50°C. Hybridization was detected using alkaline phosphatase-conjugated anti-DIG antibody (Roche, diluted 1:500 with 20% sheep serum/KTBT) and 5-bromo-4-chloro-3-indolyl phosphate/nitroblue tetrazolium/ (BCIP/NBT).

Explant Co-culture on Collagen Gel

Hydrated collagen gels (300 μl/well, 1 mg/ml type I rat-tail collagen; Becton Dickinson) were prepared in four-well dishes (Nunc) according to Bernanke and Markwald (1982). The AV region was resected from stage 12–18 chick hearts and then cut longitudinally. The resulting AV explants were placed on a collagen gel lattice saturated with complete medium (CM199; medium 199 containing 1% chick serum, 5 μg/ml insulin, 5 μg/ml transferrin, 5 ng/ml selenium (ITS™ premix; BD Bioscience), and streptomycin/penicillin; Gibco). After incubation for 3 hr, the cultures were subjected to various test conditions including CM199 (300 μl) supplemented with anti-TGFβ3 antibody (R&D), anti-chick ALK2 (Exalpha Biologicals, Lai et al., 2000), noggin (R&D), or SB431542 (ALK5 inhibitor, TOCRIS). The anti-TGFβ3 antibody has less than 10% cross-reactivity with other TGFβs (manufacturer's data), and this antibody recognized 10 ng of recombinant (r) TGFβ3, but not 10 ng of rTGFβ2 on Western blotting (only a faint band was visible when 50 ng of rTGFβ2 was subjected; not shown). After 36 hr (total incubation time), cultures were assessed for EMT under Hoffman modulation optics and subjected to immunohistochemistry or RT-PCR as described below.

Endothelial Activation Assay on Collagen Gel

Hydrated collagen gels (300 μl in volume) were prepared and AV regions (containing preactivated endocardium and myocardium) from stage 14-minus hearts were placed on the gel lattice. AV explants were removed after 3–4 hr of co-culture, leaving a preactivated AV endothelial monolayer, which will remain epithelial without the addition of EMT-stimulating preparations (Krug et al., 1987), and the resulting endothelial monolayer was subjected to various test conditions including incubation with 300 μl of cardiocyte conditioned medium (CCM, Krug et al., 1987) or CM199 with or without supplementation with recombinant TGFβ3 or BMP2 (R&D). After 36 hr (total incubation time), endothelial monolayers were assessed for EMT and subjected to immunohistochemistry or RT-PCR as described below.

Indirect Immunofluorescence Microscopy

Cultures were drained of medium, rinsed with PBS, and fixed in 4% paraformaldehyde in PBS for 30 min at room temperature. After rinsing in PBS for 15 min, cultures were blocked with 1% bovine serum albumin in PBS containing 0.1% Triton X-100 for 1 hr, then incubated with anti–ROCK1-coil1 antibody (rabbit polyclonal diluted 1:100, kindly provided Dr. K. Kaibuchi, Nagoya, Japan), anti–smooth muscle α-actin antibody (mouse IgG2a diluted 1:400, Sigma) or the antibody mixture in a moist chamber overnight at 4°C. The anti–ROCK1-coil1 antibody recognizes a single 160-kDa band in chick embryonic extracts (Lu et al., 2001). After rinsing with PBS, cultures were incubated with fluorescein isothiocyanate–conjugated goat anti-rabbit IgG (Cappel, diluted 1:100), rhodamine isothiocyanate (RITC) -conjugated goat anti-mouse IgG2a (SouthernBiotech, diluted 1:100) or the secondary antibody mixture for 2 hr. After rinsing with PBS, cultures were incubated with DAPI (4,6-diamidino-2-phenylindole dihydrochloride, diluted 1:500) for nuclear detection, then transferred to slides and cover-slipped. Samples were observed under a laser confocal microscope (Zeiss) or conventional fluorescent microscope (OLYMPUS). At least 10 explants were analyzed for ROCK1 expression in endothelial cells on the gel surface or in the total endothelial/mesenchymal cells. Endothelial cells were defined as rounded cells presented on the gel surface. Mesenchymal cells were defined as elongated cells migrated on/within the gel lattice. The percentage of ROCK1-positive cells ([number of ROCK-1–positive cells/DAPI-positive cells] × 100) was calculated for each explant and statistical analysis was performed using analysis of variance (ANOVA) with the significance level set to less than 5%.

RT-PCR

Cultured endothelial/mesenchymal cells were collected from 5–10 explants and RNA was extracted using ISOGEN (Nippon Gene). cDNA was synthesized from 0.1 μg of total RNA, and PCR was carried out in 15-μl reaction mixtures using the One-Step RT-PCR system (QIAGEN). The sequences of ROCK1-specific primers were described above and the primers for GAPDH (glyceraldehyde-3-phosphate dehydrogenase) and Nkx2.5 have been described elsewhere (Sakata et al., 2007). The reactions were cycled at 94°C (30 sec), 60°C (30 sec), and 72°C (1 min), with a final extension at 72°C (10 min). Semiquantitative RT-PCR analysis was performed according to Wei et al. (2001). For each primer set, at least three cycle numbers were tested to be certain that PCR products were accumulated within the linear range. The number of cycles used for ROCK1 was 26–30 cycles, and that for GAPDH was 26–30 cycles. RT-PCR at the appropriate cycle numbers was performed at least three times. PCR products were electrophoresed and stained with ethidium bromide, and the density of the appropriate bands was analyzed using Image Gauge (FUJIFILM). Relative expression levels normalized to the expression level of GAPDH were calculated. Statistical analysis wasperformed using ANOVA or Student's t-test with a significance level of 5%. Amplification without RT revealed that PCR products produced from genomic DNA were not detected in any of the RT-PCR reactions (not shown).

RESULTS

ROCK1 Expression Is Up-regulated at the Onset of EMT

Previous studies showed that the expression of ROCK1 is initiated in the AV and OT endocardium at the onset of EMT (as demonstrated by immunohistochemistry), and that ROCK1 plays a role in the regulation of mesenchymal invasion/migration during EMT (Zhao and Rivkees, 2004; Sakabe et al., 2006). To investigate the mechanisms stimulating the expression of ROCK1, we first examined the spatiotemporal expression patterns of ROCK1 by in situ hybridization. At stage 12, expression of ROCK1 was detected in the myocardium (m in Fig. 1A), but only a weak or background level signal was found in the endocardium (arrowheads in Fig. 1A2). At stage 14, some of the AV endothelial cells are activated to undergo EMT and showed cellular hypertrophy (Markwald et al., 1990). At this stage, ROCK1 was detectable in hypertrophied AV endothelial cells (arrows in Fig. 1B2), but some endothelial cells showed only a weak signal (arrowheads in Fig. 1B2). Myocardial cells also expressed ROCK1 (m in Fig. 1B). At stage 15–16, some of the endothelial cells in the AV region transform to acquire the mesenchymal phenotype and invade the adjacent extracellular matrix known as cardiac jelly. At this stage, AV endothelial cells with migratory appendages expressed ROCK1 extensively (arrows in Fig. 1C2). At stage 18, EMT occurs extensively and delaminated mesenchymal cells migrate into the cardiac jelly. At this stage, ROCK1 was detectable extensively in AV endothelial (arrows in Fig. 1D) and migrating mesenchymal cells (* in Fig. 1D2). As previously reported, the inner myocardial layer subjacent to the endocardial cushion tissue expressed ROCK1 extensively (m in Fig. 1D; Sakabe et al., 2006). To examine the relative expression levels of ROCK1, RT-PCR was performed in endothelial/mesenchymal cells obtained from stage 12, 14, 16, and 18 hearts. AV or ventricular endocardium was co-cultured with associated myocardium (co-culture, six explants/region/stage) on a collagen gel lattice. Ventricle was defined as a trabeculated region between the AV and outflow tract. After 3–4 hr, myocardium was removed and the obtained endothelial/mesenchymal tissue (or endocardium) was subjected to RT-PCR. Semiquantitative RT-PCR was performed as described in the Materials and Methods section (Wei et al., 2001) and showed that the PCR products of ROCK1 were increased in the endothelial/mesenchymal tissue of stage 14–18 hearts (Fig. 1E). The level of the PCR products of ROCK1 in the ventricular endocardium was low in comparison with that in the AV of stage 16–18 hearts. There was no detectable PCR product of Nkx2.5 in these samples (not shown). These results suggest that the expression of ROCK1 was up-regulated in AV endocardium at around stage 14, in which EMT is activated to take place by myocardium-derived inductive signal(s).

Figure 1.

Localization of ROCK1 mRNA during the formation of atrioventricular canal (AV) endocardial cushion tissue. A: Before the onset of endocardial cushion epithelial–mesenchymal transformation (EMT) at stage 12, in situ hybridization showed that ROCK1 mRNA was expressed in the myocardium (m), but a weak or background level signal was found in AV endocardium (arrowheads). B,C: At stage 14, activated and hypertrophied AV endothelial cells expressed ROCK1 mRNA (arrows in B2), but some endothelial cells showed a weak signal (arrowheads in B2). At stage 16, endothelial cells with migratory appendages expressed ROCK1 (arrows in C2). Myocardial cells expressed ROCK1 (m). D: At stage 18, ROCK1 mRNA was highly expressed in both AV endothelial (arrows in D2) and cushion mesenchymal cells (* in D2). E: Semiquantitative reverse transcriptase-polymerase chain reaction (RT-PCR) was performed as described in the Materials and Methods section, and the results showed that ROCK1 transcripts were increased in AV endothelial/mesenchymal tissue from stage 14 onward (Student's t-test). The relative expression level of ROCK1 was lower in ventricular endocardium. There was no detectable PCR product of Nkx2.5 in these samples. ‡Note that RNA was obtained from endothelial/mesenchymal cells of six explants and three RT-PCR reactions were performed. e, endocardium; m, myocardium; St, stage. Scale bar = 100 μm in A1–D1; 50 μm in A2–D2.

Myocardial Signal Up-regulate ROCK1 in Cultured AV Endocardium

The above observations suggested that the expression of ROCK1 mRNA was stimulated in AV endothelial cells at the onset of EMT, and therefore the expression of ROCK1 appears to be up-regulated by inductive signal(s) regulating EMT. It has been reported that endocardial EMT is initiated by myocardially derived inductive molecules (Krug et al., 1985, 1987; Mjaatuedt et al., 1991). To examine whether the myocardial inductive signal was capable of inducing ROCK1 expression in AV endothelial cells, stage 14-minus preactivated AV endothelial monolayers were cultured on three-dimensional collagen gel lattice with associated myocardium (co-culture) or CCM (cardiocyte conditioned medium that contains myocardially secreted inductive molecules that can initiate EMT in culture and induce the expression of TGFβ in preactivated endothelial cells; Krug et al., 1987; Nakajima et al., 1994). After 36 hr, AV endothelial cells that were co-cultured with associated myocardium or treated with CCM seeded mesenchymal cells into the gel lattice, while endothelial cells cultured in control medium (CM199) maintained the epithelial phenotype and did not seed mesenchymal cells into the gel lattice (Fig. 2A1, B1). The resulting cultures were subjected to immunostaining or RT-PCR to examine the expression of ROCK1. Endothelial cells cultured with CM199 did not show any apparent anti-ROCK1 immunoreactivity (Fig. 2A2). In AV endothelial cells treated with CCM (Fig. 2B) or co-cultured with associated myocardium (not shown), some endothelial cells and migrating mesenchymal cells showed anti-ROCK1 immunoreactivity (Fig. 2B2). Endothelial/mesenchymal cells that expressed ROCK1 were also labeled with EMT marker, anti-smooth muscle α-actin antibody (Fig. 2B2-4). Therefore, ROCK1 is likely to be expressed in activated endothelial cells as well as in migrating mesenchymal cells. Semiquantitative analysis (percentage of ROCK1-positive cells) showed that more than 10% of endothelial/mesenchymal cells expressed ROCK1 when the cells were treated with CCM or co-cultured with associated myocardium (co-culture; Fig. 2C). The percentage of ROCK1-positive endothelial cells on the gel surface was also increased in cultures treated with CCM or in co-cultures (Fig. 2D). On the other hand, when endothelial cells were cultured in CM199 alone, approximately 1% of the cells showed anti-ROCK1 immunoreactivity (Fig. 2C). Semiquantitative RT-PCR showed that the amount of PCR-product for ROCK1 was much higher in endothelial cells treated with CCM or associated myocardium (co-culture) than in those cultured in CM199 (P < 0.01, Fig. 2E). The results indicate that the expression of ROCK1 in cultured AV endocardium is up-regulated by CCM or associated myocardium.

Figure 2.

ROCK1 expression was induced by cardiocyte conditioned medium (CCM). A: Stage 14-minus preactivated atrioventricular canal (AV) endothelial cells were cultured with or without CCM. After 36 hr, AV endothelial cells cultured in control medium (CM199) maintained an epithelial phenotype (A1), and did not show any apparent staining of ROCK1 (A2) or the epithelial–mesenchymal transformation (EMT) marker smooth muscle α-actin (A3). B: When explants were treated with CCM, endothelial cells seeded mesenchymal cells into the gel lattice (B1). Smooth muscle α-actin (SMA) -positive activated endothelial (arrowheads) and migrating mesenchymal cells (arrows) expressed ROCK1 (B2, B3). Blue staining (DAPI, 4,6-diamidino-2-phenylindole dihydrochloride, diluted 1:500) indicates nuclei (A4, B4). C: The percentage of ROCK1-positive endothelial/mesenchymal cells was significantly increased in AV endothelial cells treated with CCM or co-cultured with associated myocardium (P < 0.01, analysis of variance [ANOVA]). D: The percentage of ROCK1-positive endothelial cells was also increased in AV endothelial cells treated with CCM or co-cultured with myocardium (P < 0.01, ANOVA). E: Semiquantitative reverse transcriptase-polymerase chain reaction (RT-PCR) was performed as described in the Materials and Methods section, and the results showed that the amount of ROCK1 mRNA was significantly increased in endothelial cells treated with CCM or co-cultured with associated myocardium (P < 0.01, ANOVA). ‡Note that RNA was obtained from six to eight explants and three RT-PCR reactions were performed. n, number of explants examined; SMA, smooth muscle α-actin. Scale bar = 50 μm in A1,B1; 25 μm in A2–4,B2–4.

TGFβ3 Induces the Expression of ROCK1 in Cultured AV Endocardium

It has been reported that TGFβ initiates the phenotypic changes of EMT and that anti-TGFβ3 inhibits the corresponding phenomenon in culture (Potts and Runyan, 1989 , 1991; Nakajima et al., 1998). We therefore examined whether TGFβ3 was capable of stimulating the expression of ROCK1 in cultured AV endocardium. Stage 14-minus AV explants (endocardium + myocardium, co-culture) were cultured on collagen gel and incubated with or without anti-TGFβ3 antibody. After 36 hr, endothelial cells cultured in control medium (CM199 containing normal IgG) seeded many mesenchymal cells into the gel lattice and the transforming endothelial/mesenchymal cells expressed ROCK1 (Fig. 3A,C). In contrast, AV explants cultured in medium supplemented with 0.5–25 μg/ml of anti-TGFβ3 antibody (final concentration) showed suppressed seeding of mesenchymal cells and the percentage of ROCK1-positive cells was decreased (P < 0.01, Fig. 3B,C). The percentage of ROCK1-positive endothelial cells on the gel surface was also significantly decreased in cultures treated with anti-TGFβ3 antibody (P < 0.01, Fig. 3D). Therefore, it is suggested that the reduction of ROCK1-positive cells by anti-TGFβ3 treatment may not be caused by the reduced number of transformed cells. Semiquantitative RT-PCR showed that the amount of PCR product of ROCK1 was significantly reduced in endothelial/mesenchymal cells treated with anti-TGFβ3 antibody at concentration of 5 or 25 μg/ml (P < 0.05, Fig. 3E). These results suggest that TGFβ3 was necessary for induction of ROCK1 expression in the AV endocardium co-cultured with associated myocardium at the onset of EMT. We next attempted a gain-of-function experiment using preactivated AV endocardium that was cultured in medium with recombinant TGFβ3. Stage 14-minus preactivated AV endothelial cells were prepared on collagen gel and cultured in medium with or without recombinant TGFβ3 for 36 hr. Endothelial cells cultured in CM199 did not undergo transformation and approximately 1% of cells showed anti-ROCK1 immunoreactivity (Fig. 4A,C). Endothelial cells treated with recombinant TGFβ3 underwent phenotypic changes (Nakajima et al., 1998) and the percentage incidence of cells expressing ROCK1 was increased in a dose-dependent manner (Fig. 4B,C). The percentage of ROCK1-positive endothelial cells on the gel surface was also increased in cultures treated with TGFβ3 (Fig. 4D). Semiquantitative RT-PCR showed that the PCR product for ROCK1 was significantly increased in cells treated with 5 or 25 ng/ml of recombinant TGFβ3 (P < 0.01, Fig. 4E).

Figure 3.

Anti–transforming growth factor-beta3 (TGFβ3) antibody inhibited the expression of ROCK1 in preactivated atrioventricular canal (AV) endocardium co-cultured with associated myocardium. A: Control AV explants obtained from stage 14-minus hearts were cultured in CM199 containing normal IgG seeded mesenchymal cells into collagen lattice (A1), and ROCK1 expression was detected in surface endothelial cells (arrowheads in A2) and cells with mesenchymal phenotype (arrows in A2). B: When the AV explants were treated with neutralizing anti-TGFβ3 antibody, the number of mesenchymal cells invading the gel lattice was reduced (B1) and the number of ROCK1-positive cells was decreased (arrows in B2). C: The percentage of ROCK1-positive endothelial/mesenchymal cells was decreased in cultures treated with anti-TGFβ3 antibody (P < 0.01, analysis of variance [ANOVA]). D: The percentage of endothelial cells expressing ROCK1 was also decreased in cultures treated with anti-TGFβ3 antibody (P < 0.01, ANOVA). E: Semiquantitative reverse transcriptase-polymerase chain reaction (RT-PCR) was performed as described in the Materials and Methods section and the results showed that the relative amount of ROCK1 mRNA was significantly decreased in cultures treated with anti-TGFβ3 antibody (P < 0.05, ANOVA). ‡Note that RNA was obtained from six to eight explants and three RT-PCR reactions were performed. ex, explant; n, number of explants examined. Scale bar = 50 μm.

Figure 4.

Transforming growth factor-beta3 (TGFβ3) induced the expression of ROCK1 in cultured preactivated endocardium. A,B: Stage 14-minus preactivated atrioventricular canal (AV) endocardium was cultured with or without recombinant TGFβ3. After 36 hr, endothelial cells cultured in control medium (CM199) maintained endothelial phenotype (A1) and did not show any apparent staining for ROCK1 (A2). In contrast, endothelial cells cultured with recombinant TGFβ3 showed phenotypic epithelial–mesenchymal transformation (EMT) changes (B1), and some endothelial cells (arrowheads in B2) and migrating mesenchymal-like cells (arrows in B2) showed anti-ROCK1 immunoreactivity. C: The percentage of ROCK1-positive endothelial/mesenchymal cells was significantly increased in cells treated with recombinant TGFβ3 (P < 0.01, analysis of variance [ANOVA]). D: The percentage of ROCK1-positive endothelial cells was also increased in cultures treated with recombinant TGFβ3 (P < 0.01, ANOVA). E: Semiquantitative reverse transcriptase-polymerase chain reaction (RT-PCR) was performed as described in the Materials and Methods section, and the results showed that PCR products for ROCK1 were increased in cultures treated with recombinant TGFβ3 in a dose-dependent manner (P < 0.01, ANOVA). Note that RNA was obtained from six to eight explants and four RT-PCR reactions were performed. n, number of explants examined. Scale bar = 50 μm.

Anti-ALK2 Antibody Inhibits ROCK1 Expression in Culture

The above experiments suggest that TGFβ3 plays a role in the expression of ROCK1 during EMT. Cells in the chick endocardium express both ALK2 and ALK5, which bind to TGFβ in the presence of type II TGFβ receptor (Lai et al., 2000; Desgrosellier et al., 2005). Therefore, we next examined which type I receptor was required for the expression of ROCK1. Stage 14-minus AV explants were cultured in medium supplemented with normal goat IgG, anti-ALK2 antibody or SB431542 (ALK5 inhibitor), and the resulting cultures were assessed for the expression of ROCK1 (Fig. 5). In AV explants treated with 150 nM SB431542, the number of seeded mesenchymal cells was slightly but not significantly reduced in comparison with control cultures (P = 0.054, Fig. 5A1, C), and the percentage of ROCK1-positive endothelial/mesenchymal cells was significantly reduced (P = 0.02, Fig. 5A2, D). In contrast, there was no significant difference in the percentage of ROCK1-positive endothelial cells on the gel surface between controls and cultures treated with 150 nM SB431542 (P = 0.27, arrowheads in Fig. 5A2, E). In AV explants treated with 15 or 25 μg/ml of anti-ALK2 antibody, the number of mesenchymal cells invading the gel lattice was significantly reduced (P < 0.01, Fig. 5B1, C). The percentage of ROCK1-positive endothelial/mesenchymal cells as well as the percentage of ROCK1-positive endothelial cells was reduced significantly in cultures treated with 15 or 25 μg/ml of anti-ALK2 antibody (P < 0.01, Fig. 5B2, D,E). Therefore, reduction of the number of ROCK1-positive cells in cultures treated with 150 nM SB431542 was likely to be caused by the reduced number of mesenchymal cells expressing ROCK1. We next examined whether anti-ALK2 antibody suppressed TGFβ3-inducible ROCK1 expression in cultured preactivated AV endothelial cells. Stage 14-minus AV endothelial monolayers were cultured in medium with the combined addition of TGFβ3 plus anti-AKL2 antibody for 36 hr. As previously shown, ROCK1 expression was induced in cells treated with recombinant TGFβ3 (25 ng/ml) plus normal IgG (25 μg/ml), while ROCK1 expression was significantly down-regulated in cells treated with both TGFβ3 and anti-ALK2 antibody at the mRNA and protein levels (Fig. 6). The percentage of ROCK1-positive endothelial cells was also reduced in cultures treated with recombinant TGFβ3 and anti-ALK2 antibody (Fig. 6).

Figure 5.

Anti–activin receptor-like kinase-2 (ALK2) antibody inhibited the expression of ROCK1 in cultured atrioventricular canal (AV) endothelial cells. Stage 14-minus AV explants were cultured in CM199 supplemented with normal IgG, SB431542, or anti-ALK2 antibody. In explants treated with SB431542, epithelial–mesenchymal transformation (EMT) phenotypic changes occurred and the number of mesenchymal cells seeded into the gel lattice was similar (but slightly reduced at the concentration of 150 nM, P=0.054, analysis of variance [ANOVA]) to control cultures (arrows in A1, C). In cultures treated with SB431542, some endothelial cells (arrowheads) and migrating mesenchymal cells (arrows) showed anti-ROCK1 immunoreactivity, but the percentage of ROCK1-positive cells was decreased when cultures were treated with 150 nM SB431542 (P = 0.02, ANOVA, D). There was no significant difference in the number of ROCK1-positive endothelial cells on gel surfaces between control (normal IgG) and SB431542-treated explants (arrowheads in A2, P = 0.27, E). In cultures treated with anti-ALK2, the number of mesenchymal cells that invaded the gel lattice was significantly decreased (P < 0.01, ANOVA, B1, C) and the number of ROCK1-positive cells was decreased (arrow in B2). The percentage of ROCK1-positive endothelial/mesenchymal cells and the percentage of ROCK1-positive endothelial cells were significantly reduced in cultures treated with anti-ALK2 antibody at concentrations of 15–25 μg/ml (P < 0.01, ANOVA, D,E). ex, explant. n, number of cells (number of explants) examined. Scale bar = 50 μm.

Figure 6.

Transforming growth factor-beta3 (TGFβ3) -inducible ROCK1 expression was blocked by Anti–activin receptor-like kinase-2 (ALK2) antibody. Stage 14-minus preactivated atrioventricular canal (AV) endothelial cells were cultured in medium with TGFβ3 plus anti-ALK2 antibody. After 36 hr, cultures were assessed for the expression of ROCK1 at the protein and mRNA levels. Both the percentage of ROCK1-positive cells and the percentage of ROCK1-positive endothelial cells were significantly decreased in cultures treated with TGFβ3 protein plus anti-ALK2 antibody (P < 0.01, analysis of variance [ANOVA]). Semiquantitative reverse transcriptase-polymerase chain reaction (RT-PCR) also showed that PCR products for ROCK1 mRNA were reduced in cultures treated with TGFβ3 protein plus anti-ALK2 antibody (P < 0.05 and P < 0.01, ANOVA). Note that RNA was obtained from six to eight explants and three RT-PCR reactions were performed. n, number of cells (number of explants) examined.

Noggin Inhibits the Expression of ROCK1 in Culture

BMP2 is thought to be one of the myocardially derived inductive signals involved in initiating EMT (Yamagishi et al., 1999; Nakajima et al., 2000; Sugi et al., 2004). Genetic inactivation of BMP showed that BMP2 is the major BMP from the myocardium involved in BMP signaling during EMT (Jiao et al., 2003; Ma et al., 2005; Rivera-Feliciano and Tabin, 2006). Therefore, we next examined whether BMP2 was responsible for the expression of ROCK1 during EMT. Stage 14-minus AV explants (endocardium + myocardium, co-culture) were cultured in medium with or without supplementation with noggin (a specific inhibitor of BMP2, -4, -7; Balemans and Van Hul, 2002). After 36 hr, cultures were assessed for the occurrence of EMT and expression of ROCK1. As previously reported in mouse AV explant cultures (Sugi et al., 2004), the number of mesenchymal cells that invaded the gel lattice was significantly reduced in cultures treated with recombinant noggin (not shown). Immunohistochemistry showed that when the endothelial cells were co-cultured with associated myocardium, more than 13% of cells expressed ROCK1, and the percentage of endothelial/mesenchymal cells expressing ROCK1 was significantly reduced in cells treated with noggin. The percentage of ROCK1-positive endothelial cells was also reduced in cultures treated with noggin (P < 0.01, Fig. 7A). Semiquantitative RT-PCR showed that the amount of PCR product for ROCK1 mRNA in endothelial/mesenchymal cells treated with noggin was significantly reduced in comparison to control cultures (P < 0.01, Fig. 7A). Next, we examined whether BMP2 was capable of inducing the expression of ROCK1 in cultured AV endothelial cells. Stage 14-minus AV endothelial monolayers were cultured in medium with or without recombinant BMP2 for 36 hr. As previously reported, stage 14-minus AV endothelial cells treated with BMP2 failed to undergo EMT (Yamagishi et al., 1999). Immunohistochemistry showed that ROCK1 was not up-regulated in cells treated with BMP2. There was no significant difference in the percentage of ROCK1-positive cells between controls and cultures treated with BMP2 (Fig. 7B). RT-PCR showed that there was no significant difference in the amount of PCR products between controls and cells treated with BMP2 (Fig. 7B). The extent of the increase in the PCR product of ROCK1 in cultures treated with BMP2 (Fig. 7B) was small in comparison with that in cultures treated with TGFβ3 (Fig. 4E). In addition, neither BMP4 nor BMP5 induced ROCK1 expression in cultured preactivated AV endothelial monolayers (not shown).

Figure 7.

Noggin inhibited the expression of ROCK1, but bone morphogenetic protein-2 (BMP2) did not induce the expression of ROCK1. A: Stage 14-minus atrioventricular canal (AV) explants were cultured in medium with or without noggin. After 36 hr, endothelial/mesenchymal cells were assessed for the expression of ROCK1. The percentage of ROCK1-positive endothelial/mesenchymal cells was significantly decreased in cultures treated with noggin (P < 0.01, analysis of variance [ANOVA]). The percentage of ROCK1-positive endothelial cells was also decreased in cultures treated with noggin. Semiquantitative reverse transcriptase-polymerase chain reaction (RT-PCR) showed that the PCR product of ROCK1 mRNA was reduced in endothelial/mesenchymal cells treated with noggin (P < 0.01, ANOVA). B: Stage 14-minus AV endocardium was prepared on collagen gel and cultured in medium with recombinant BMP2. After 36 hr, cells were assessed for the expression of ROCK1. Endothelial cells cultured in control medium did not show any apparent staining for ROCK1. Endothelial cells treated with BMP2 were not induced to express ROCK1. There was no significant difference in the percentage of ROCK1-positive cells between control cultures and cultures treated with BMP2 (ANOVA). Semiquantitative PCR showed that the level of PCR product for ROCK1 was not significantly increased when endothelial cells were treated with BMP2 (ANOVA). NS, no significant difference. Note that RNA was obtained from six to eight explants and three RT-PCR reactions were performed. n, number of explants examined.

DISCUSSION

ROCK1 Expression Is Up-regulated at the Onset of EMT

During early cardiac development at the looped heart stage, cardiac segments including the outflow tract (OT), right ventricle (RV), left ventricle (LV), atrioventricular (AV) canal, atrium and sinus venosus are established anteroposteriorly (Sakabe et al., 2005). As development proceeds (at stage 14 in chick and 9.5 ED in mouse embryos), some of the endothelial cells in the OT and AV regions lose their epithelial phenotype, transform into migratory mesenchymal cells, and invade the adjacent extracellular matrix, resulting in the formation of valvuloseptal endocardial cushion tissue (Markwald et al., 1975). This endocardial cushion EMT is controlled by several signaling pathways, including TGFβ and BMP (Armstrong and Bischoff, 2004). We showed that ROCK1 expression was up-regulated in the AV endocardium at the onset of or during endocardial EMT at the mRNA and protein levels (Fig. 1; Sakabe et al., 2006). In the mouse embryonic heart, ROCK1 is transcribed at 9.5 ED and ROCK2 at 10.5ED (Zhao and Rivkees, 2004). It was reported that RhoA GTPase, an upstream activator of ROCK, is highly up-regulated in AV endothelial/mesenchymal cells at stage 17–18, at which time endocardial EMT is extensively under way (Tavares et al., 2006). In addition to its high level in cushion tissue, ROCK1 expression is enriched in the primitive streak, cardiac mesoderm, developing myocardium, neural plate, and somites (Wei et al., 2001; Sakata et al., 2007). In the adult mouse, ROCK1 mRNA is expressed ubiquitously except in brain and muscle, whereas ROCK2 is expressed abundantly in the brain, muscle, heart, lung, and placenta (Nakagawa et al., 1996). These observations suggest that the expression of ROCK is regulated in a spatiotemporally restricted manner during organogenesis, including the genesis of the heart.

ROCK1 Expression Is Regulated by TGFβ Together With BMP at the Onset of EMT

Several investigators have reported that TGFβ plays a part in the regulation of endocardial cushion EMT (Potts and Runyan, 1989; Potts et al., 1991; Nakajima et al., 1998; Boyer et al., 1999; Camenisch et al., 2002). During mouse endocardial cushion EMT, only TGFβ2 is expressed in activated AV endothelial/invading mesenchymal cells as well as in AV myocardium. TGFβ3 is expressed in AV cushion tissue at later stages (Camenisch et al., 2002). In chick endocardial cushion formation, both TGFβ2 and TGFβ3 are present in the activated endothelial/invading mesenchymal cells as well as in AV myocardium, and TGFβ2 and TGFβ3 are sequentially and separately involved in the process of EMT: TGFβ2 mediates the AV endothelial activation and cell–cell separation, while TGFβ3 is required for the mesenchymal cell formation and invasion (Boyer et al., 1999; Camenisch et al., 2002). In addition to TGFβs, BMP2 is required and acts synergistically with TGFβ to initiate EMT (Yamagishi et al., 1999). Inactivation of Bmp2 activity in the mouse AV canal revealed that Bmp2 mutant embryos failed to form AV cushions (Ma et al., 2005; Rivera-Feliciano and Tabin, 2006). We showed that anti-TGFβ3 antibody, anti-ALK2 antibody, or noggin inhibited the expression of ROCK1 in AV endothelial cells co-cultured with associated myocardium, and that CCM (which contains myocardially secreted inductive molecules and up-regulates the expression of TGFβ in endocardium) or recombinant TGFβ3, but not BMP, induced the expression of ROCK1 in preactivated AV endothelial cells in culture. Therefore, it is likely that ROCK1 is expressed in activated endothelial cells which eventually undergo EMT, and that the expression of ROCK is triggered by signals regulating EMT such as TGFβ or TGFβ together with BMP. We showed that, although the BMP antagonist noggin inhibited the expression of ROCK1 in cultured AV endocardium, recombinant BMP2, -4, and -5 did not induce ROCK1 expression effectively in preactivated AV endocardium. In the AV canal region, several BMP isoforms are expressed by the myocardium (Yamagishi et al., 2001; Somi et al., 2004). BMP2 signaling by means of ALK3 has been shown to up-regulate mesenchymal expression of TGFβ2 in mice (Gaussin et al., 2002). It was also reported that BMP2 induces the autocrine induction of TGFβ in mouse AV canal endocardium in culture (Sugi et al., 2004). In our preliminary experiments, BMP2 failed to induce the expression of TGFβ2 and TGFβ3 in cultured chick preactivated AV endocardium. Like conditional null mutants of Bmp2 in the AV canal (Ma et al., 2005; Rivera-Feliciano and Tabin, 2006), double mutant embryos for Bmp6 and Bmp7 show defective valve morphogenesis and chamber septation (Kim et al., 2001). We do not know the reason why BMP was not able to induce the expression of TGFβ in chick preactivated AV endocardium. One possibility is that heterodimerization between BMP2 and BMP6 or -7 might strongly induce the expression of TGFβ during cushion tissue formation. Another possibility is that BMP signaling renders the AV endocardial cells permissive for TGFβ signaling by an unknown mechanism (Wang et al., 2005). To date, little is known about the mechanism that regulates the expression of ROCK (Loirand et al., 2006). Hiroki et al. (2004) reported that inflammatory stimuli such as angiotensin II or IL-1b are capable of inducing the expression of ROCK in human coronary vascular smooth muscle cells. Taken together, our results and those of others noted above suggest that TGFβ or TGFβ together with BMP may play a role in the expression of ROCK at the onset of EMT and that the Rho-ROCK pathway appears to be regulated, at least in part, at the transcriptional level during development.

ROCK1 Expression Is Regulated by TGFβ and ALK2 in AV Endocardium

In the early chick endocardium, both ALK2 and ALK5 are expressed and anti-ALK2 antibody inhibits mesenchymal formation, while anti-ALK5 antibody does not (Lai et al., 2000). Ectopic expression of constitutively active (ca) ALK2, but not caALK5, in non–transforming ventricular endocardium is capable of inducing EMT (Desgrosellier et al., 2005). Therefore, signaling through ALK2 is sufficient to initiate EMT in the chick heart (Desgrosellier et al., 2005). In support of these experiments, endothelial-specific Alk2-deficiency mutant mice show defects in AV septation and valves that result from a failure of endothelial cells to transform into mesenchyme in the AV canal (Wang et al., 2005). Other experiments showed that perturbation of ALK5 or endoglin expression by siRNA causes impaired endothelial proliferation and EMT in stage 15–16 AV endocardium, which has already received EMT-activation signal(s), suggesting that ALK5 is also involved in EMT after initiation (Mercado-Pimentel et al., 2007). It is widely accepted that ALK5 (TβRI) is the major type I TGFβ receptor that interacts with TβRII and mediates TGFβ-induced responses. In addition to the well-characterized TβRII-TβRI (ALK5) complex, TβRII forms a functional complex with ALK2, and signals from TβRII-AKL2 are different from those involving TβRI (ALK5; Feng and Derynck, 2005). In mouse mammary epithelial cells (in which high levels of ALK2 are expressed but there is no expression of ALK5), TGFβ induces EMT, suggesting that ALK2 is capable of mediating TGFβ signaling (Miettinen et al., 1994). ALK2 is also known to transmit BMP signals in association with ActRII or BMPRII (Yamashita et al., 1996; Macias-Silva et al., 1998; Feng and Derynck, 2005). In AV endocardium co-cultured with associated myocardium, anti-TGFβ3 antibody, anti-ALK2 antibody, and noggin but not SB431542, effectively inhibited the expression of ROCK1 as well as EMT. In addition, TGFβ3 but not BMP was capable of inducing the expression of ROCK1 in cultured preactivated AV endocardium and the TGFβ-inducible expression of ROCK1 was inhibited by anti-ALK2 antibody. Furthermore, our preliminary experiment showed that noggin did not inhibit nuclear transport of phospho-Smad1/5/8 in cultured AV endocardium. These results suggested that signaling mediated by TGFβ/AKL2 together with BMP may play a role in the expression of ROCK1 during EMT. In mouse null mutants for Tgfβr2, AV endocardial cushion tissue was generated normally in vivo, but perturbed in cultured AV explants (Jiao et al., 2006), suggesting that signaling mediated by BMP/ALK2 can compensate for the reduction of TGFβ-TβRII-ALK2 signaling in vivo. In mouse embryos, endocardial depletion of Alk3, a specific BMP type I receptor, severely impairs EMT in the atrioventricular canal, indicating that ALK3-mediated BMP signaling also plays a role in cushion tissue formation (Song et al., 2007). Wang et al. (2005) showed that both PSmad1/5/8 and PSmad2 were present in the nuclei of cushion endothelial/mesenchymal cell and the nuclear transport of both Smad1/5/8 and Smad2 was impaired in endocardial cells depleted of Alk2. However, whether there is cross-talk between the signaling pathways mediated by TGFβ/ALK2, TGFβ/ALK5, and BMP/ALK2 during EMT remains uncertain. Further experiments will be necessary to elucidate the scheme regulating the interaction(s) between the receptor complexes for TGFβ/BMP as well as the intracellular signaling pathways regulating EMT.

Acknowledgements

The authors thank Dr. K. Kaibuchi for providing anti-ROCK1 antibody.

Ancillary