Transforming Growth Factor β (TGFβ) Signaling via Differential Activation of Activin Receptor-like Kinases 2 and 5 during Cardiac Development

Little is known regarding factors that induce parasympathetic responsiveness during cardiac development. We demonstrated previously that in atrial cells cultured from chicks 14 days in ovo, transforming growth factor β (TGFβ) decreased parasympathetic inhibition of beat rate by the muscarinic agonist, carbamylcholine, by 5-fold and decreased expression of Gαi2. Here in atrial cells 5 days in ovo,TGFβ increased carbamylcholine inhibition of beat rate 2.5-fold and increased expression of Gαi2. TGFβ also stimulated Gαi2 mRNA expression and promoter activity at day 5 while inhibiting them at day 14 in ovo. Over the same time course expression of type I TGFβ receptors, chick activin receptor-like kinase 2 and 5 increased with a 2.3-fold higher increase in activin receptor-like kinase 2. Constitutively active activin receptor-like kinase 2 inhibited Gαi2 promoter activity, whereas constitutively active activin receptor-like kinase 5 stimulated Gαi2 promoter activity independent of embryonic age. In 5-day atrial cells, TGFβ stimulated the p3TP-lux reporter, which is downstream of activin receptor-like kinase 5 and had no effect on the activity of the pVent reporter, which is downstream of activin receptor-like kinase 2. In 14-day cells, TGFβ stimulated both pVent and p3TP-lux. Thus TGFβ exerts opposing effects on parasympathetic response and Gαi2 expression by activating different type I TGFβ receptors at distinct stages during cardiac development.

A decrease in heart rate in response to parasympathetic stimulation (negative chronotropic response) involves the binding of acetylcholine to M 2 muscarinic receptors and the dissociation of the heterotrimeric G-protein, G i2 , into ␣ i2 and ␤␥ subunits. The latter activates the inward rectifying K ϩ channel, GIRK1, increasing diastolic depolarization and decreasing heart rate (1). A decrease in the force of contraction in response to muscarinic stimulation (negative inotropic effect) involves the binding of the ␣ i2 subunit to adenylate cyclase followed by a decrease in cAMP production. Several studies support the conclusion that control of G␣ i2 expression regulates the response of the heart to parasympathetic stimulation. The development of parasympathetic responsiveness in the embryonic chick heart is associated with an increase in G␣ i2 expression (2). Furthermore, growth of chick atrial cells in the absence of lipoproteins, which has been shown to result in an increased response to parasympathetic stimulation, is associated with an increase in the expression of G␣ i2 (3,4). Finally, expression of G␣ i2 in the porcine atrioventricular node resulted in an increase in parasympathetic tone (5).
A role for TGF␤ 1 in the development of the parasympathetic response of the heart was suggested by studies in which medium conditioned by co-culture of chick heart cells and ciliary ganglia induced a negative chronotropic response to carbamylcholine in chick heart cells 3.5 days in ovo (dio). This induction of a parasympathetic response was accompanied by an increase in G␣ i2 expression (6) and was reversed by addition of a neutralizing antibody to TGF␤ 1 to the medium. 2 In contrast, we recently demonstrated that in atrial cells from hearts 14 dio, TGF␤ 1 decreased the expression of G␣ i2 and decreased the negative chronotropic response to carbamylcholine (7). These data suggest that TGF␤ exerts opposing effects on parasympathetic responsiveness at different stages of cardiac development.
The TGF␤ family is composed of at least three 25-kDa homodimeric proteins, TGF␤ 1 , TGF␤ 2 , and TGF␤ 3 . TGF␤ signaling involves the binding of TGF␤ ligand to two transmembrane serine threonine kinases, the type I TGF␤ receptor I (TBRI) and the type II TGF␤ receptor (TBRII). TBRII has a constitutively active cytoplasmic kinase domain and an extracellular domain that binds TGF␤ 1 and TGF␤ 3 . TGF␤ binding results in the phosphorylation of TBRI by TBRII. TBRI then activates a signaling cascade, which may include a series of transcription factors known as Smads (8). Other TGF␤ family members such as the activins and bone morphogenic proteins (BMPs) also signal through a type I receptor by binding to specific type II receptors for activin (ActRII and ActRIIB) and BMP (BMPRII) (9). To date, seven type I receptors have been identified and designated activin receptor-like kinases (ALKs) 1-7. The ligand specificity of these ALKs has been determined by their ability to bind to a given ligand and to activate downstream signals in the presence of a specific type II receptor subtype. ALK1 and ALK5 are activated by TGF␤ via TBRII (9,10). ALK5 in association with TBRII specifically stimulates the plasminogen activator inhibitor (PAI-1) promoter. ALK5 mediates growth arrest in mink lung epithelial cells following the formation of the ALK5/TRBII complex and the phosphorylation of ALK5 (11). ALK2 interacts with TBRII as well as ActRII and BMPRII type II receptors (12). ALK2 does not mediate TGF␤ signaling in mink lung epithelial cells but has been implicated in the TGF␤-stimulated epithelial-mesenchymal transformation in the mammary gland of the mouse (13). The regulation of TGF␤ receptor signaling by selective interactions with different type I receptors is an intriguing mechanism that might help explain the pleiotropic effects of TGF␤. Here we demonstrate that TGF␤ mediates opposing effects on G␣ i2 expression and the response of the heart to parasympathetic stimulation at different stages of chick heart development and that these pleiotropic effects are due to differential activation of ALK2 and ALK5 by TGF␤.

EXPERIMENTAL PROCEDURES
Cell Culture-Embryonic chick atrial myocyte cultures were prepared by a modification of the method of DeHaan (14) as described previously (15). Eggs were staged according to the method of Hamberger and Hamilton (16). The embryos 5 dio corresponded to stage 27, and the 14-day embryo corresponded to stage 40.
RNase Protection Analysis-A G␣ i2 RNase protection probe was generated from a PstI fragment derived from the chick G␣ i2 cDNA subcloned into pBluescript and linearized with BamHI (17). Using T7 RNA polymerase (Roche Molecular Biochemicals) in the presence of [ 32 P]UTP (800 Ci/mmol, PerkinElmer Life Sciences), this template gave a 307nucleotide antisense riboprobe. The glyceraldehyde phosphate dehydrogenase (GAPDH) RNase protection probe, used as a control, was generated from a cDNA template (gift of R. Runyan), which was linearized with HindIII. Using T3 RNA polymerase, this template gave a 250nucleotide antisense riboprobe. Probes were purified by PAGE on a 6% gel, and the major band corresponding to the predicted molecular weight for the riboprobe was excised and eluted overnight. Total RNA was isolated from cultures of embryonic chick atrial cells 14 dio using guanidinium CsC1 2 centrifugation as described (18). RNase protection was carried out as described previously (15). Riboprobes were hybridized to 15 g of total RNA prepared from cells treated with either vehicle or 5 ng/ml TGF␤ 1 . The samples were treated with RNase and analyzed by PAGE on 6% gels containing urea followed by autoradiography. Radiographic exposure was 6 h for G␣ i2 and 2 h for GAPDH. The relative intensity of the bands was determined by densitometry scanning using NIH Image Pro.
Measurement of Changes in Beat Rate-Embryonic chick atrial cells from hearts of embryos 5 dio cultured on coverslips at 5 ϫ 10 5 cells/cm 2 were treated either with vehicle (4 mM HCl and 0.5 mg/ml bovine serum albumin) or with 5 ng/ml TGF␤ 1 and placed in a perfusion chamber as described (15), on the stage of a Zeiss inverted phase contrast microscope enclosed in a Lucite box maintained at 37°C. The inlet side of the chamber was connected via polyethylene tubing to two syringe pumps allowing the cells to be sequentially perfused by two different solutions. Perfusion at 0.98 ml/min did not disturb cell adhesion to the coverslip. Cells were perfused with an HEPES-buffered salt solution containing 1% fetal calf serum, 11 mM glucose, 0.6 mM HEPES, 0.6 mM CaCl 2 , 4.0 mM KCl, 140 mM NaCl, and 5 mM MgCl 2 . In this study, each cell served as its own control with the spontaneous beat rate determined before and after exposure to 0.1 mM carbamylcholine. Beating was determined by monitoring the movement of the border of a single cell with a video-motion detector and recording the output with a physiologic recorder (Hewlett-Packard Co., Palo Alto, CA) as described (3).
Western Blotting-Polyclonal (rabbit) antiserum raised to the carboxyl-terminal decapeptide from rat G␣ i2 was a gift of David Manning. TBRII, ALK2, and ALK5 antibodies were prepared as described (8,19). Cultured chick atrial cells 5 and 14 dio were grown for 3 days in fetal calf serum, homogenates were prepared and Western blot analysis was carried out as described (15). Equal amounts of protein were loaded as determined by a DC protein assay (Bio-Rad). Equal loading was determined by Coomassie staining.
Luciferase and Alkaline Phosphatase Assays-Embryonic chick atrial cells 5 and 14 dio were cultured in medium supplemented with fetal calf serum. On the second culture day, 1 g of G␣ i2 -Luc consisting of 1.5 kb of the 5Ј upstream region of the chick G␣ i2 promoter ligated to a luciferase reporter (7) and 0.2 g of a human placental alkaline phosphatase under the control of an SV40 promoter (pSV2Apap, a gift of L. Ercolani) were transfected into heart cells cultured on 35-mm plates by the use of FuGENE 6 transfection reagent (Roche Molecular Biochemicals) as recommended by the manufacturer. Total DNA was maintained at 2.1 g by addition of pBluescript (pBS) DNA. At 16 h prior to harvesting, cells were incubated as indicated. At 72 h after transfection, cells were washed in phosphate-buffered saline and solubilized in lysis buffer at 425 l/plate (24 mM glycyl-glycine, 15 mM MgSO 4 , 4 mM EGTA, 1% Triton X-100, and 1 mM dithiothreitol). The extract was sonicated three times for 10 s and then centrifuged at 13,000 ϫ g for 3 min at 4°C, and the supernatant was assayed for luciferase and alkaline phosphatase activity as described (20). In other experiments, cells were transfected with the pVent promoter luciferase reporter construct, Xvent2-luc, containing ϳ250 bp of Xvent2 promoter sequences, which was a gift from Christof Niehrs, or p3TP-lux containing the putative TGF␤-responsive region of the human PAI-1 (plasminogen activator inhibitor) promoter, which was a gift of Joan Massague.
Statistics-Statistical analysis was by Student's t test.

TGF␤ 1 Enhances the Negative Chronotropic Response to Muscarinic Stimulation in Atrial Cells from Hearts 5 dio-
During embryonic development, the negative chronotropic response of the chick heart to muscarinic stimulation developed between 2 and 7 dio (25). To determine the effect of TGF␤ on the development of the parasympathetic response, embryonic chick atrial cells from 5 dio hearts were incubated for 16 h with either 5 ng/ml TGF␤ 1 or vehicle, and beat rate was determined in the presence of carbamylcholine. In the absence of TGF␤ 1 , 0.1 mM carbamylcholine decreased beat rate by 30 Ϯ 1% (Ϯ S.E., n ϭ 21, p Ͻ 0.001, Fig. 1). However, after incubation with TGF␤ 1 , carbamylcholine decreased beat rate by 76 Ϯ 1% (Ϯ S.E., n ϭ 21, p Ͻ 0.01, Table I). These effects on beat rate were reversible within 5 min after reperfusion of cells with carbamylcholine-free medium. Thus TGF␤ 1 increases the chronotropic response to carbamylcholine by more than 2.5-fold in atrial myocytes from hearts 5 dio. This result is opposite to the effect of TGF␤ 1 in cells from atria of hearts 14 dio in which we demonstrated that TGF␤ 1 decreased the chronotropic response to carbamylcholine by more than 5-fold (Table I) (7).
Developmental Changes in the Expression of TGF␤ Receptors-These opposing effects in the response of chick atrial cells to TGF␤ might reflect changes in the expression of TGF␤ receptors involved in signaling at different stages of cardiac development. Western blot analysis demonstrated that embryonic chick atrial cells expressed TBRII, ALK2, and ALK5 (Fig.  4, A and C). ALK2 and ALK5 have been reported to mediate distinct responses to TGF␤ signaling (9 -11, 13). For this reason, we studied developmental changes in these two TGF␤ receptors. chALK2 and chALK5 were initially expressed at low levels at day 5 in ovo but increased markedly between days 5 and 14 in ovo (Fig. 4A). Comparison of the fold increase in ALK2 and ALK5 expression between 5 and 14 dio demonstrated that chALK2 increased 2.30 Ϯ 0.20-fold (Ϯ S.E., n ϭ 4, p Ͻ 0.01) more than chALK5 (Fig. 4B). TBRII levels increased 4.40 Ϯ 0.20-fold (Ϯ S.E., n ϭ 3) between 5 and 14 dio (Fig. 4, C and D). Thus each receptor increased between 5 and 14 dio with the largest increase in chALK2.
Differential Activation of chALK2 and chALK5 by TGF␤ at Days 5 and 14 in Ovo-To determine whether TGF␤ signaling might preferentially activate chALK2 or chALK5 at different stages of cardiac development, we compared the effect of TGF␤ 1 on p3TP-lux and pVent reporter activity in atrial cells from hearts 5 and 14 dio. chALK5 specifically activates the p3TP-lux reporter (12). pVent is known to be activated by BMP, not by TGF␤, and is one of the best known reporters of ALK2 activation (26). To determine whether ALK2 might be mediating a TGF␤ response, in our system, pVent was used as a reporter of ALK2 activation. In atrial cells from hearts 5 dio, 5 ng/ml TGF␤ 1 stimulated p3TP-lux activity 5.20 Ϯ 0.30-fold (Ϯ S.E., n ϭ 5, p Ͻ 0.002, Fig. 5A), whereas in atrial cells from hearts 14 dio, TGF␤ 1 stimulated p3TP-lux by 2.5 Ϯ .01-fold (Ϯ S.E., n ϭ 6, p Ͻ 0.003, Fig. 5B). In atrial cells from hearts 5 dio, TGF␤ 1 had no effect on pVent reporter activity (Fig. 5C). However, in atrial cells 14 dio, we observed an unexpected 2.2 Ϯ 0.2-fold (Ϯ S.E., n ϭ 7, p Ͻ 003, Fig. 5D) increase in pVent reporter activity in response to TGF␤ 1 . These data demonstrate that in chick atrial cells, pVent is activated by TGF␤ and that this activation is specific for cells 14 dio. These data also suggest that in chick atrial cells 5 dio TGF␤ signals via chALK5 and not chALK2.
If chALK2 mediates the inhibition of the G␣ i2 promoter by TGF␤ signaling, then overexpression of chALK2 in atrial cells from chicks 5 dio should inhibit TGF␤-stimulated G␣ i2 pro-  moter activity. In experiments summarized in Fig. 7, TGF␤ 1 stimulated G␣ i2 promoter activity 2.10 Ϯ 0.10-fold above basal (Ϯ S.E., n ϭ 4). Cotransfection of these cells with chALK2 ϩ followed by incubation with TGF␤ 1 not only reversed TGF␤ stimulation of G␣ i2 promoter activity but also decreased G␣ i2 promoter activity by 9-fold to 0.40 Ϯ 0.06 (Ϯ S.E., n ϭ 4)-fold below basal. As expected, chALK5 ϩ alone stimulated G␣ i2 promoter activity. These data demonstrate that chALK2 inhibits G␣ i2 promoter activity, whereas chALK5 stimulates G␣ i2 promoter activity, and that these effects are independent of the developmental stage of the atrial myocytes.

DISCUSSION
The data presented here provide novel insight into TGF␤ signaling and the regulation of parasympathetic responsiveness in the heart. TGF␤ stimulates the negative chronotropic response of chick atrial cells 5 dio to carbamylcholine, whereas it decreases the inhibition of beat rate by carbamylcholine in atrial cells 14 dio (7). These effects of TGF␤ correlate with alterations in G␣ i2 expression. At 5 dio, TGF␤ stimulates G␣ i2 expression, and at 14 dio, TGF␤ inhibits G␣ i2 expression. Examination of two TBRIs reported to play a role in TGF␤ signaling reveals that chALK5 increases G␣ i2 expression, whereas chALK2 decreases G␣ i2 expression independent of the embryonic age of the cells. Further, TGF␤ stimulates pVent, a reporter of ALK2 activation, in 14 dio, but not in 5 dio, atrial cells. These data, taken together with the finding that chALK2 expression increases markedly between 5 and 14 dio, suggests that at 5 dio, TGF␤ activates only chALK5, but at 14 dio, TGF␤ activates both chALK5 and chALK2. These findings offer a potential mechanism to explain the change in TGF␤ regulation of G␣ i2 expression and parasympathetic response during cardiac development.   14 (B) dio cultured in medium supplemented with fetal calf serum were co-transfected with a 2-kb fragment from the 5Ј-flanking region of G␣ i2 ligated to a promoterless luciferase reporter (G␣ i2 -Luc) and a human placental alkaline phosphatase (PAP). Following transfection, cells were incubated for 16 h with either 5 ng/ml TGF␤ or an equal volume of vehicle. Cells were harvested, and luciferase activity and PAP activity were determined as described previously. Data are normalized to the ratio of luciferase to PAP activity in cells cultured with vehicle adjusted to 1.
The induction of a parasympathetic response is a critical step in the physiological development of the mammalian heart. The regulation of the parasympathetic responsiveness of the heart not only controls the rate and force of contraction but also may play a role in the development of cardiac arrhythmias (27,28). We have demonstrated previously that during embryonic development of the chick heart, the negative chronotropic response to carbamylcholine increased markedly between 5 and 7 dio, reaching a plateau at 7 dio (25). The development of the parasympathetic response in the embryonic chick heart was associated with an increase in G␣ i2 expression (2). Regulation of G␣ i2 expression has been associated with the control of parasympathetic responsiveness in the adult heart. A recent study demonstrated that overexpression of G␣ i2 in the porcine atrioventricular node resulted in a decrease in atrioventricular conduction and a decreased response to sympathetic stimulation consistent with an increase in parasympathetic tone (5). Here we demonstrate a striking parallel between developmental changes in TGF␤ regulation of the response of the heart to parasympathetic stimulation and TGF␤ regulation of G␣ i2 ex- pression. These data emphasize the importance of the regulation of G␣ i2 expression on parasympathetic responsiveness and cardiac function.
Our data support the notion that the transition of TGF␤ signaling in atrial cells from a stimulatory effect on G␣ i2 expression and parasympathetic response to an inhibitory effect during embryonic development reflects differential activation of the TGF␤ type I receptors, chALK2 and chALK5. Complexes of ALK5 and TBRII bind TGF␤ 1 to mediate TGF␤ effects such as growth arrest in mink lung epithelial cells (11). Although ALK2 binds TGF␤ when co-expressed with TBRII, it does not mediate growth arrest in mink lung epithelial cells. A role for ALK2 has been described during the TGF␤-dependent epithelial-mesenchymal transformation of mouse mammary epithelial cells (13). A similar TGF␤-stimulated epithelial-mesenchymal transformation occurs in the atrioventricular cushion during valvulogenesis. Studies using an in vitro culture system demonstrated that anti-chALK2 antisera blocked transformation, whereas anti-chALK5 antisera was without effect (19). The finding that specific TGF␤ effects may be attributed to ALK2 or ALK5 suggested that the specificity of the downstream response to TGF␤ signaling is dependent on the identity of the TBRI activated in a given cell type. In support of this conclusion, chALK2 and chALK5 were shown to exert opposing effects on G␣ i2 promoter activity. Constitutively active chALK2 inhibited G␣ i2 promoter activity, and constitutively active chALK5 stimulated G␣ i2 promoter activity independent of the embryonic age of the cell in which they were expressed.
Hence differential activation of chALK2 and chALK5 by TGF␤ at 5 and 14 dio might result in opposing effects of TGF␤ on G␣ i2 expression during cardiac development. To test this hypothesis, we compared the effect of TGF␤ 1 on pVent and p3TPlux reporter activity in cells from atria 5 and 14 dio. The pVent reporter is activated by BMP signaling via ALK2 (22,26,29), whereas the p3TP-lux reporter is activated by ALK5 signaling (11). TGF␤ 1 stimulated p3TP-lux reporter activity in atrial cells from hearts 5 dio but had no effect on pVent reporter activity in these cells. Furthermore, although TGF␤ stimulated both p3TPlux and the pVent reporter in cells 14 dio, the stimulation of pVent was significantly higher than p3TP-lux in these cells.
These data support the conclusion that TGF␤ signaling at 5 dio occurs via chALK5 and that signaling at 14 dio occurs via both chALK2 and chALK5, with chALK2 predominating. Although it is not possible to directly compare the level of expression of chALK2 and chALK5 at 5 or 14 dio, we noted a larger increase in ALK2 expression than ALK5 expression, consistent with the conclusion that the increase in ALK2 signaling at 14 dio was due at least in part to an increase in expression levels. Taken together with the data which demonstrate that ALK5 stimulates G␣ i2 promoter activity and ALK2 inhibits G␣ i2 promoter activity, the finding of differential activation of ALK2 and ALK5 would account for the opposing effects of TGF␤ on G␣ i2 expression at 5 and 14 dio.
The unexpected observation that TGF␤ stimulates pVent expression in chick atrial cells 14 dio is the first report of activation of a BMP-like signal by TGF␤. TGF␤ signaling via ALK5 has been shown to involve Smads 2/3 (30). We demonstrated that constitutively active chALK5 did not stimulate pVent promoter activity, which indicates that Smads 2/3 cannot activate pVent in these cells. Furthermore, studies of pVent have demonstrated stimulation by the BMP-specific Smads 1/5/8 (26). This would suggest that TGF␤ stimulation of pVent might be mediated by a BMP-specific pathway in these cells.
The significance of these developmental changes in TGF␤ signaling may be related to a dual role of TGF␤ signaling in cardiac physiology and development. In an in vitro model for parasympathetic innervation of the heart, we have demonstrated that induction of a negative chronotropic response to carbamylcholine and the expression of G␣i2 were dependent on the release of a soluble factor (6) whose effect was inhibited by a neutralizing antibody to TGF␤. 2 These findings implicate TGF␤ in the development of the parasympathetic response. Studies in explanted, intact chick heart have previously demonstrated a marked increase in the response of the heart to parasympathetic stimulation between days 2 and 7 in ovo (31). Here TGF␤ stimulates a significant increase in both G␣i2 expression and parasympathetic response in atrial cells 5 dio. These data support the conclusion that TGF␤ plays a role in the development of a parasympathetic response in the heart. At 14 dio, functional parasympathetic innervation of the chick heart is complete (31). The significance of TGF␤ inhibition of G␣i2 expression and parasympathetic responsiveness at this developmental stage is unclear. However, TGF␤ has been shown to play a role in a number of processes important to cardiac function such as angiogenesis, cardiac hypertrophy, inflammation, and the response of the heart to myocardial infarction (32,33). The relationship between TGF␤ inhibition of parasympathetic responsiveness and G␣i2 expression to these processes remains to be determined. FIG. 6. Effect of constitutively active TBRIs (chALK2 ؉ and chALK5 ؉ ) on G␣ i2 promoter activity. A, effect of chALK5 ϩ . Cells 14 dio were transfected with G␣ i2 -Luc and cotransfected with either pBS or chALK2 ϩ and human PAP as described. Following transfection, cells were incubated for 16 h with either vehicle or 5 ng/ml TGF␤ 1 , and luciferase activity was determined. B, effect of chALK2 ϩ . Experiments were carried out as described in panel A except that cells were cotransfected with either pBS or chALK2 ϩ . FIG. 7. Effect of chALK2 ؉ on TGF␤ 1 stimulation of G␣ i2 promoter activity in atrial cells from hearts 5 dio. Cells from hearts 5 dio were co-transfected with G␣ i2 -Luc and either pBS, chALK2 ϩ , or chALK5 ϩ . Transfected cells were incubated with either vehicle or 5 ng/ml TGF␤ 1 for 16 h, and luciferase activity was determined.
These data suggest that TGF␤ is an important regulator of parasympathetic responsiveness during cardiac development and may regulate the parasympathetic response at least in part by modulating G␣ i2 expression. Further, we suggest that TGF␤ signaling may involve the activation of both ALK5 and ALK2 in atrial cells and that the relative contribution of each of these receptors determines the level of G␣ i2 expression and parasympathetic responsiveness. Our observations suggesting differential activation of two different type I receptors are an attractive mechanism to explain the pleiotropic effects of TGF␤.