Synergistic Roles of Neuregulin-1 and Insulin-like Growth Factor-I in Activation of the Phosphatidylinositol 3-Kinase Pathway and Cardiac Chamber Morphogenesis*

Cardiac chamber morphogenesis requires the coordinated growth of both cardiac muscle and endocardial cell lineages. Paracrine growth factors may modulate the coordinated cellular specification and differentiation during cardiac chamber morphogenesis, as suggested by the essential role of endothelial-derived growth factors, neuregulin-1, and insulin-like growth factor-I. Using the whole mouse embryo culture system for delivery of diffusible factors into the cardiac chamber, neuregulin-1 was shown to promote trabeculation of the ventricular wall. Another factor, insulin-like growth factor-I, had no apparent effect by itself. Combined treatment with neuregulin-1 and insulin-like growth factor-I strongly induced DNA synthesis of cardiomyocytes and expansion of both the ventricular compact zone and the atrioventricular cushions leading to chamber growth and maturation. In cultured cardiomyocytes, combined neuregulin-1 and insulin-like growth factor-I also had a synergistic effect to promote DNA synthesis and cellular growth, which were prevented by wortmannin, an inhibitor of phosphatidylinositol 3-kinase. Adenoviral delivery of dominant negative Rac1, which acts downstream of phosphatidylinositol 3-kinase, blocked the effect of combined neuregulin-1/insulin-like growth factor-I treatment. These studies support the concept that the interaction of neuregulin-1 and insulin-like growth factor-I pathways plays an important role in coordinating cardiac chamber morphogenesis and may occur through convergent activation of phosphatidylinositol 3-kinase.

The formation, maturation, and septation of distinct cardiac chambers during heart development may require the interplay of signals between myocardial and endocardial cell lineages (1,2). These distinct cell types present in the cardiogenic mesoderm will give rise to the double-layered heart tube, with myocardial cells forming the outer layer and endocardial cells forming the inner layer (3,4). Within the looped cardiac tube, positional signals in the ventricular segment induce specific maturational steps in ventricular muscle cells, including trabeculation, expansion of the ventricular compact zone and atri-oventricular cushions, and development of the interventricular septum. Another set of cues leads to the generation of cushion mesenchyme derived from endocardial cells, which contributes to septation of the conotruncus and outflow tract and remodeling of the atrioventricular canal, which ultimately separates the atrial and ventricular chambers (1,2,5). It has been previously shown by in vitro assay systems that cushion formation depends on extracellular matrix, and signals emanating from the adjacent cardiomyocytes (5,6). However, the potential effect of endocardial-derived signals in cushion morphogenesis has not yet been explored. The maturational steps in ventricular muscle and endocardial cells might be mutually coordinated. However, the interplaying signals involved at these different steps of the developing embryonic heart are still unknown.
Recent studies from targeted gene inactivation in the mouse indicate that two likely candidates for exerting paracrine effects on ventricular chamber and cushion development are neuregulin-1 (NRG-1) 1 and insulin-like growth factor-I (IGF-I) (7,8). Both NRG-1 and IGF-I are endothelial-derived signals expressed early in embryonic development that bind and activate receptors expressed throughout the cardiac chamber (7,9,10). Targeted ablation of these growth factors or their cognate receptors severely affects murine cardiac morphogenesis. Homozygous null embryos for NRG-1 or the cognate receptors erbB2 or erbB4 display a lack of trabeculation of the ventricular wall, whereas individual knockouts for IGF-I or IGF-I receptor die perinatally with deficient myofibrillogenesis of ventricular myocytes (7,8,(11)(12)(13)(14)(15)(16).
To assess the role of NRG-1 and IGF-I in cardiac chamber maturation and morphogenesis in intact embryos, we have utilized the whole mouse embryo culture system for injection of these peptide growth factors through the ventricular wall of mouse embryos. The distinct advantage of the ex vivo mouse embryo culture system is that it allows functional studies of exogenous growth factors during cardiac morphogenesis in the context of the intact embryo. The present study documents that NRG-1 induces trabeculation of the ventricular wall without a significant increase in the proliferation of cardiomyocytes, in agreement with an essential role of NRG-1 to initiate trabeculae (7). Another secreted growth factor, IGF-I, which is essential for cardiac growth in neonatal heart, had no apparent effect on development. However, combined injection of NRG-1 and IGF-I induced DNA synthesis in ventricular myocytes and significant growth of the ventricular compact zone. The potentiation of NRG-1 and IGF-I on cardiomyocyte proliferation was corroborated in cultured cardiac muscle cells. To dissect downstream signals activated by NRG-1 and IGF-I, we used specific inhibitors or dominant negative mutants to block either the MAP kinase or phosphatidylinositol (PI) 3-kinase signaling pathways in cultured cardiomyocytes. Our results indicate that the synergistic effect was mediated by convergent activation of PI 3-kinase rather than the MAP kinase pathway, which was activated by each growth factor alone. Taken together, these results support a synergistic role for NRG-1 and IGF-I in coordinating ventricular chamber growth and maturation, which may occur through convergent signaling pathways at the level of PI 3-kinase activation.

MATERIALS AND METHODS
Mouse Embryo Culture-Preparation of rat serum was performed as described by Cockroft (17). Whole mouse embryos were cultured according to the method of Sturm and Tam (18) as described previously (19). Briefly, timed pregnant female C57Bl6 mice maintained and bred in colonies were sacrificed by cervical dislocation. The uterus was dissected and rinsed in phosphate-buffered saline (PBS; 137 mM NaCl, 2.7 mM KCl, 10 mM Na 2 HPO 4 , 1.8 mM KH 2 PO 4 ) to remove residual blood and then transferred to PB1 medium (137 mM NaCl, 2.7 mM KCl, 0.5 mM MgCl 2 , 8.04 mM Na 2 HPO 4 , 1.47 mM KH 2 PO 4 , 0.9 mM CaCl 2 , 0.33 mM sodium pyruvate, 1 g/liter glucose, 0.01 g of phenol red, pH 7.35, 100 ml/ml streptomycin, 100 unit/ml penicillin; all reagents were from Sigma). Embryos of 11.5 days post-coitum (E11.5) were dissected from the uterus, and the decidual mass and Riechert's membrane were removed. The embryos were released from the yolk sac and amnion, which remained attached to ensure continuity of the vessels connecting the embryo to the yolk sac or the umbilical vessels from the embryo to the placenta. They were then transferred to pre-equilibrated medium containing 50% rat serum and constant gassing (95% O 2 , 5% CO 2 ) in roller culture bottles and incubated for the following 26 h. To confirm specific embryonic stages of murine development we have used the remodeling of the forelimb buds of cultured embryos as described previously (19). The smooth distal borders of the handplate at E11.5 are indented by E12.5 (20).
Microinjection Protocol-After 1 h in culture, the embryos were placed in a Petri dish and microinjected into the left ventricle using a 6-m-diameter glass pipette. The micropipettes were pulled in a multistage pipette puller (Sutter Instruments, Co., Novato, CA) using 1-mm glass capillaries. The micropipette was attached to a MX-110-R 4 axis manual micromanipulator (Newport Instruments, Newport, CA) via electrode holders. All intracardiac injections were performed with a volume of 1 l, which proceeded at a low flow rate. All capillaries were calibrated by aspirating 3.2 l into the needle and injecting one-third of the total volume into three successive embryos. In control experiments, PBS was used. The concentration of NRG-1 solutions of the ␤ form (21) (Genentech, Inc., San Francisco, CA) or ␣ form (gift from Kuo-Fen Lee, Salk Institute, La Jolla, CA) were 50 ng/l, and IGF-I was 50 ng/l (Genentech, Inc.). These concentrations were selected based on doseresponse curves performed in cultured cardiomyocytes and adjusted in the embryo culture for NRG-1.
BrdUrd Incorporation and Immunostaining in Cultured Embryos-Embryos in culture were grown for the last 5 h in the presence of 30 g/ml of BrdUrd added to the culture medium. Embryos were fixed in 4% paraformaldehyde and embedded in paraffin at a thickness of 5 m. Immunostaining of nuclear BrdUrd was performed on four sections/ embryo treatment using antibodies against BrdUrd coupled to biotin followed by incubation with streptavidin coupled to peroxidase. The slides were then subjected to light microscopy for visualization of incorporated BrdUrd, which was detected by the peroxidase color reaction as indicated in the BrdUrd labeling kit (Zymed Laboratories Inc., CA). The number of BrdUrd positive and negative cells were determined in at least two images of 200ϫ magnified fields/section in at least four sections/embryo at a location of the ventricular free wall immediately dorsal to the apex of the chamber. The average number of BrdUrd positive cells was determined per 100 cells corresponding to each embryo treatment.
In Situ Hybridization Analysis-Riboprobes for acetylcholinesterase cDNA were generated using either T7 or T3 polymerase (22) and radiolabeled in the presence of 35 S-UTP (23). In situ hybridization was performed on 5-m paraffin sections from mouse embryos according to the method described by Lyons et al. (24). Equal amounts (10,000 cpm/ml) of 35 S-labeled riboprobes were added to each section and washed after overnight hybridization at 60°C. The slides were exposed in D19 Kodak emulsion, counterstained with 0.2% toluidine blue solution, and mounted in Permount.
Microdissection and Scanning Electron Microscopy-To assess internal morphological characteristics, embryos were perfusion fixed, microdissected, and processed for scanning electron microscopy as described previously (25). Briefly, embryos used for scanning electron microscopy were removed from the culture bottles and perfusion fixed with 2% glutaraldehyde, 1% formaldehyde in 0.1 M cacodylate buffer. Hearts were then processed to obtain the parietal and septal views of the right ventricle using a previously described standardized procedure (26). Primary magnification of scanning electron micrographs was 160ϫ for atrioventricular (AV) cushion distances and 1000ϫ for analysis of ventricular wall morphology. Measurements of AV cushion distance were acquired from 160ϫ images at a point midway between the lateral boundaries of the superior and inferior cushion pads. Measurements of the thickness of the compact zone were acquired from the 1000ϫ images at a location of the right ventricular free wall immediately dorsal to the apex of the chamber. At least three measurements were averaged from each image for both AV cushion and compact zone distances.
Isolation, Culture, and Treatment of Cardiomyocytes-Primary cardiac myocytes were isolated from neonatal rat heart ventricular chambers and prepared by a Percoll gradient technique as described previously (28). Cardiomyocytes were densely plated in gelatin-coated 10-cm plates or in chamber slides precoated with 1% gelatin and laminin (Sigma) as described previously (28). After the first day in culture medium containing 10% serum, cardiomyocytes were washed and cultured in minimal medium (4:1 Dulbecco's modified Eagle's medium/ medium 199, 100 units/ml penicillin, 100 g/ml streptomycin, and 10 mM glutamine). NRG-1 and/or IGF-I were added to the culture medium at a concentration of 2.5 and 5 nM, respectively, in agreement with their maximal activities in dose-response curves. Cardiomyocytes were grown for 24 h in the presence of growth factors and for additional 6 h in the presence of 40 M BrdUrd to measure DNA synthesis or for 72 h to follow cardiomyocyte morphological changes. Alternatively, cardiomyocytes were treated with 1 M wortmannin or 10 M PD098059 or infected in serum-free medium for 12 h with adenovirus at a multiplicity of 50 -100 particles/cell prior to addition of growth factors and then cultured as described above.
Immunostaining-Cells were rinsed with 1ϫ PBS and fixed in 4% paraformaldehyde for 10 min. Cells were washed in PBS and permeabilized with PBS containing 1% bovine serum albumin, 0.2% Triton X-100 for 15 min. Incubation with the primary antibody myomesin (a kind gift from Dr. J.-C. Perriard, Institute of Cell Biology, ETH-Zurich, Switzerland), was performed for 1 h in PBS containing 1% bovine serum albumin. Myocytes were washed in PBS/1% bovine serum albumin then incubated with secondary antibodies coupled to rhodamine and the fluorescent dye bisbenzimide (Hoechst dye 22358) 1 g/ml for 30 min. The slides were then washed and mounted in a glycerol-based mounting medium containing Gelvatol and 2.5% 1,4-diazabicyclo[2.2.2]octane (Sigma).
Alternatively, 10 l of rabbit anti-erbB4 serum raised against recombinant protein (generous gift of Dr. Cary Lai, Scripps Research Institute, La Jolla, CA) was added to 0.3 ml of cell lysate. For control immunoprecipitates, preimmune serum was added to 0.3 ml of cell lysate. The protein precipitates were washed four times with lysis buffer, and proteins were analyzed by 6.5% SDS-polyacrylamide gel electrophoresis. Immunoblotting was performed with anti phosphotyrosine antibodies (RC20H, Transduction Laboratories), erbB2, or erbB4 antibodies.

Analysis of NRG-1 Function on the Developing Cardiac Chamber in Cultured Whole Mouse Embryos-We have chosen
to analyze the effects of injected growth factors on cardiac development from E11.5 to E12.5 (Fig. 1, A and B), when active trabeculation of the ventricular wall and expansion of the cushions is occurring (30). The diffusion of peptide growth factors throughout the heart following direct cardiac administration was analyzed by injection of the fluorescent markers, DiI (Molecular Probes, Eugene, OR) or fluorescein isothiocyanate-dextran. The hydrophobicity of DiI allows for its cellular uptake, whereas the molecular weight of fluorescein isothiocyanatedextran is in the range of NRG-1 or IGF-I peptides. Following intracardiac injection, DiI or fluorescein isothiocyanate-dextran diffused in both the right and left ventricular chambers and to the atria and conotruncus, suggesting that the injection protocol will effectively deliver growth factors to the entire heart. The distribution of DiI was stable after 24 h in culture ( Fig. 1, C and D).
The biologically active recombinant peptide containing the epidermal growth factor-like domain of ␤ isoform of NRG-1 was injected into the ventricular chamber of wild type embryos at E11.5 and allowed to develop in culture to E12.5. As trabeculation of the ventricular wall may depend on NRG-1, expression of acetylcholinesterase (AchE), whose enzymatic activity in the embryonic rat heart is restricted to cardiomyocytes of the trabeculae of the ventricular wall, was selected to follow induced trabecular growth (31,32). Accordingly, we observed AchE mRNA in a restricted expression pattern coincident with the trabeculae and in continuity with the conotruncus at early stages of in utero mouse embryo development, E10 -E13 (data not shown). To monitor expansion of trabeculae in NRG-1 or control PBS-injected embryos in culture, the expression of AchE was determined by in situ hybridization in comparative histological sections of the ventricular free wall. A representation of the corresponding ventricular area used for the analysis is shown in Fig. 2A. As observed by the number of projections of cardiomyocytes forming trabeculae marked by AchE expression, NRG-1-injected embryos display an increased density of trabeculae ( Fig. 2C) compared with control injected embryos (Fig. 2B). Ventricular trabeculae growth in NRG-1-injected embryos was accompanied by a 1.5-fold increase in the number of cardiomyocytes (Table I). Similar effects were observed with NRG-1 recombinant peptides of either ␣ or ␤ isoforms (data not shown). Intracardiac injection of IGF-I, a peptide growth factor not related to NRG-1, had no significant effect on trabeculation. The AchE expression pattern in IGF-I-injected embryos was not different from control PBS-injected embryos (Fig. 2D).
These results indicate that exogenous NRG-1 induced the differentiation of the highly proliferative cardiomyocytes in the ventricular compact zone into trabeculae, in agreement with previous work on the requirement of NRG-1 for the formation of trabeculae (7,33,34).
NRG-and IGF-I-induced Expansion of Atrioventricular Cushion Tissue-The NRG-1 expression pattern indicates a potential role in cushion mesenchyme development. Therefore, the activity of NRG-1 during expansion of AV cushion tissue after intracardiac injection of NRG-1 was analyzed by scanning electron microscopy. The three-dimensional morphology of the developing atrioventricular cushions is shown in Fig. 3. We used the physical distance between the superior and inferior AV cushions as an indicator of AV mesenchymal expansion. The atria have been removed for ease of visualization of cushion tissue. Accordingly, representative dorsal views of the atrioventricular cushions in E11.5 mouse embryos after 24 h in culture are shown in Fig. 3 and summarized in Table I. As shown in Table I, NRG-1 induced a consistent growth of cushion mesenchyme, although it was not statistically significant. The distance between the two AV cushions was reduced in response to NRG-1 to 33 Ϯ 13 m (Fig. 3B) compared with 48 Ϯ 12 m in control PBS-injected embryos (Fig. 3A). This slight cushion growth in response to NRG-1 suggested that additional interacting signals were required for cushion development. Accordingly, we examined whether another growth factor, IGF-I, which is expressed in cushion mesenchyme (10), may potentiate the NRG-1 effect on cushion morphogenesis because in vitro work by others has demonstrated a functional interaction between the erbB receptor family and IGF-I receptor signal path-ways (35,36). Injection of IGF-I alone into the heart did not induce expansion of cushion mesenchyme (Table I and Fig. 3C). However, the combination of both IGF-I and NRG-1 consistently reduced the AV cushion distance to an average of 22 Ϯ 2 m (Fig. 3D), suggesting a synergistic effect on expansion of cushion mesenchyme in the developing heart of treated embryos in culture. In addition, the combined injection of NRG-1 and IGF-I induced partial fusion of the cushions, which occurred in 50% of treated embryos. In normal development in vivo, comparable fusion of the cushions is not observed prior to E13 (30).

NRG-1 and IGF-I Synergistically Promote Compact Zone Expansion and Activates Ventricular Muscle Cell Proliferation-
The three-dimensional morphology of the ventricular chamber was examined using scanning electron microscopy on the parietal and septal sections of the developing heart. From these images, the depth of the compact zone of the free wall of the ventricles can be determined. Measurements of the thickness of the compact zone were acquired at a location of the right ventricular free wall immediately dorsal to the apex of the chamber. The average distance of the ventricular compact zone was 16 Ϯ 1 m (1-2 cell layers) in untreated embryos (Fig. 4A) and was not modified by NRG-1 or IGF-I injections (Fig. 4, B and C). A remarkable effect on the ventricular compact zone was observed by combined injections with both IGF-I and NRG-1. The two growth factors together consistently expanded the compact zone to an average of 26 Ϯ 2 m (Fig. 4D), which represents an increase to 3-5 layers.
The morphological changes induced by combined injection of NRG-1 and IGF-I may be due in part to stimulation of proliferation of the respective cardiac lineages in the ventricular chambers and endocardial cushion tissue (ventricular myocytes and endocardial-derived mesenchyme). Therefore, we analyzed proliferation by monitoring nuclear incorporation of exogenous BrdUrd. No differences in cardiomyocyte BrdUrd incorporation were observed between control, PBS-injected, NRG-1-injected, or IGF-I-injected embryos (Fig. 5, A-C). Combined NRG-1/IGF-I injections in developing embryos resulted in active BrdUrd incorporation in 56 Ϯ 6% of trabecular cardiomyocytes (Fig. 5D) versus 24 Ϯ 4% of BrdUrd positive cells in trabeculae of control or single factor-injected embryos (Fig.  5, A-C). Therefore, the novel observation on the expansion of the compact zone (Fig. 5D) induced by combined NRG-1/IGF-I injection was accompanied by an increased DNA synthesis as assessed by BrdUrd incorporation (Fig. 5D, arrows). Differentiation of cardiac chambers, as evidenced by proper downregulation of the atrial form of myosin light chain, MLC-2a in ventricular cardiomyocytes (37), occurred normally in NRG-1and IGF-I-injected embryos (data not shown). Collectively, these results indicate that the potentiation of NRG-1 and IGF-I  activities increase the ability of cardiomyocytes to proliferate and differentiate in developing embryos. Concerted Activation of Cultured Cardiomyocyte Growth by NRG-1 and IGF-I-The analysis of the function of NRG-1 and IGF-I on cardiac muscle cells in the absence of additional effects from nonmuscle cells was carried out in isolated neonatal rat cardiomyocytes, which are readily purified from other cells present in the heart. The specific nuclear incorporation of BrdUrd in cardiomyocytes, an index of DNA synthesis, was determined subsequent to treatment with NRG-1 and/or IGF-I. As shown in Fig. 6A, the low DNA synthesis rate in cardiomyocytes in culture is increased in serum-containing medium. Significantly higher stimulation of BrdUrd incorporation was achieved in cardiomyocytes grown in the combined presence of NRG-1 and IGF-I versus untreated or NRG-1 or IGF-I or serum-stimulated cells (Fig. 6A). The effect of combined NRG-1/ IGF-I on BrdUrd incorporation was prevented by wortmannin, a specific inhibitor of the PI 3-kinase (Fig. 6A). Changes in DNA synthesis were quantified by BrdUrd incorporation as well as [ 3 H]thymidine uptake in cell lysates (Fig. 6B). A 2-fold stimulation of DNA synthesis was observed in cardiomyocytes treated with NRG-1 or IGF-I alone. The level of [ 3 H]thymidine incorporation observed in cardiomyocytes treated with NRG-1 and IGF-I was increased by 10-fold versus untreated cardiomyocytes, representing a higher incorporation compared with the 6-fold induction observed in serum-stimulated cardiomyocytes. The level of DNA synthesis induced by combined NRG-1 and IGF-I was consistent with a synergistic response of cardiomyocytes to both growth factors. Two major signaling pathways that are triggered by the activation of either erbB2 or IGF-I receptors are the MAP kinase and the PI 3-kinase pathways (38 -40). Therefore, to dissect the cascade of events triggered by the combined treatment of NRG-1 and IGF-I, we used wortmannin, an inhibitor of PI 3-kinase and the MAP kinase, mitogen-activated extracellular-signal regulated kinase specific inhibitor PD098059 and by assessing the effect of dominant negative molecules of either MAP or PI 3-kinase pathways. NRG-1/IGF-I-stimulated DNA synthesis was blocked by addition of wortmannin (Fig. 6B), whereas it was not prevented by PD098059, which blocks the response of cardiomyocytes to the individual effects of NRG-1 and IGF-I (Fig. 6B).
Adenoviral infection of cardiomyocytes mediated highly efficient delivery of dominant negative mutants of the MAP kinase pathway (Ras, and Raf) or of the PI 3-kinase pathway (Rac) (data not shown). NRG-1/IGF-I induced BrdUrd incorporation to an average of 18 Ϯ 3% positive cells/100 cells and was significantly inhibited to basal levels by dominant negative Rac1 (⌬Rac, 1 Ϯ 0.4%) and partially inhibited by dominant negative Ras (⌬Ras, 7.5 Ϯ 1%) but not by dominant negative Raf (⌬Raf) (Fig. 6, C and D). Because Rac1/Akt is a downstream mediator of activated PI 3-kinase, this result confirms the requirement of PI 3-kinase for induced BrdUrd incorporation, an index of DNA synthesis. In addition, BrdUrd incorporation was partially induced by an activated form of Rac1 (Rac*) to an average of 12 Ϯ 2% versus 1 Ϯ 0.5% in control cardiomyocytes, which were infected with adenovirus expressing GFP (Fig. 6C) or an activated form of Raf (Raf*) (data not shown).
Concordant with previous observations (41), a rapid phosphorylation of erbB2 and erbB4 receptors was induced by NRG-1 in treated cardiomyocytes in culture (Fig. 7A). In the presence of NRG-1, the cardiomyocytes showed an elongated shape with an increased and highly organized sarcomere structure (Fig. 7B). Although IGF-I has been shown to induce IGF-I receptor phosphorylation (42), it had no apparent effect on cardiomyocyte growth and morphology. The presence of both NRG-1 and IGF-I dramatically induced cardiomyocyte growth, which acquired a more organized myofilament structure (Fig.  7B), indicative of an effect to promote differentiation. Although NRG-1-stimulated myofibrillogenesis was abolished by the MAP kinase inhibitor, PD098059, or the PI 3-kinase inhibitor wortmannin, the combined effect of NRG-1/IGF-I was only prevented by wortmannin (data not shown). To determine whether biochemical interactions between receptors could be involved in the NRG-1/IGF-I-mediated PI 3-kinase activation, the phosphorylation level of the erbB receptors was assayed in the presence of NRG-1 or IGF-I, because this is the earliest event occurring upon activation. As shown in Fig. 7C, erbB2 phosphorylation was detected in the presence of NRG-1 but not IGF-I, suggesting that there is no cross-activation of the erbB2 receptor by IGF-I. However, a weak signal corresponding to the ␣ and ␤ forms of the p85 regulatory subunit of the PI 3-kinase was only detected in erbB2 immunoprecipitates and peaked at 10 min of NRG-1/IGF-I treatment (data not shown). In this regard, a direct interaction with PI 3-kinase may require phos- phorylation of additional tyrosine residues of NRG-1-activated erbB2 receptor in the presence of IGF-I. The activated Rac1 mutant could induce DNA synthesis to a comparable, however not identical, level of BrdUrd incorporation observed in combined NRG-1/IGF-I-treated cardiomyocytes. In addition, Ras activation appears to be involved in the cellular phenotypic changes induced by combined NRG-1/IGF-I treatment, which could occur via a convergence on activation of PI 3-kinase and unlikely to occur through MAP kinase because inhibition of mitogen-activated extracellular-signal regulated kinase did not modify the cardiomyocyte response to NRG-1/IGF-I. Based on these results, the synergistic effect of NRG-1 and IGF-I is likely mediated through convergence of signaling cascades via a direct-and Ras-mediated activation of PI 3-kinase. DISCUSSION The Whole Mouse Embryo Culture System Allows Identification of Potential Paracrine Signals for Cardiac Morphogenesis in Intact Embryos-A growing body of evidence suggests that muscle cell nonautonomous mechanisms mediated by paracrine factors may play a critical role in cardiac chamber morphogenesis (43,44). Therefore, an ideal system for assaying the effects of potential growth factors in the heart would be one that allows for subsequent cardiac development to take place in the intact embryo where cell-cell communication is maintained. As has been previously noted, mouse embryos are able to predictably develop in culture and even respond to grafted cells (19,45). The intracardiac delivery of peptide growth factors combined with the ability to monitor the maturation of specific cardiac compartments (conotruncal ridges, AV cushions, trabeculae, and compact zone) enables the identification and functional relevance of secreted molecules that act on different cardiac cell populations. In addition, this system will be a valuable tool in performing rescue studies of mutant phenotypes following delivery of defined factors.
NRG-1 and IGF-I Act Synergistically on Cardiac Cushion Tissue Formation-Previous studies utilizing in vitro culture explants of the chicken atrioventricular region have established a paradigm in which signals that originate in the AV myocardium are required to trigger competent endothelial cells from the AV endocardium to form cushion mesenchyme (5). Through direct injection into the intact embryo, we now provide evidence that instructive signals for expansion of cushion mes- enchyme reside in cells of the endocardium. Thus, two factors, NRG-1 and IGF-I, whose expression is localized to endocardial and mesenchymal cells, respectively, can act synergistically to promote expansion of cushion mesenchyme. The consequent AV endocardial cushion growth was quantified by proximity of the superior and inferior cushions to each other via scanning electron microscopy. In some cases the two cushions appeared fused and indicated a precocious expansion comparable with cushion growth normally observed at E13 (30). Taken together, these results constitute a novel demonstration of a role for NRG-1 and IGF-I in cushion morphogenesis and maturation.
In the cushions, the effect of endocardial-derived NRG-1 may be mediated through erbB3 receptors present in mesenchymal cells adjacent to the endocardial cells (7,13). In these studies, we have detected expression of erbB4 in the AV cushions at later stages of embryo development, E13.5 (data not shown). It will be intriguing to examine whether erbB4 is indeed the erbB3 partner in cushion mesenchyme because the later can bind NRG-1 but lacks tyrosine kinase activity (15).

NRG-1 and IGF-I Synergistically Promote Compact Zone
Expansion and Activate Ventricular Muscle Cell Proliferation-In intact embryos in culture, the effect of exogenous NRG-1 was manifested by the appearance of organized projections of the ventricular wall, which were marked by the expression of AchE. In agreement with a requirement of NRG-1 for trabeculation, our results indicate that NRG-1 induces the highly proliferative cardiomyocytes of the compact zone down a differentiation pathway, thereby promoting the transition of compact zone myocardium into trabecular muscle without modification of basal levels of DNA synthesis.
A major transformation of the ventricular wall induced by the effect of combined NRG-1/IGF-I was the expansion of the compact zone, an event normally occurring after E13 (30). Previous work has indicated that IGF-I is involved in myofibrillogenesis of ventricular cardiomyocytes because inactivation of the IGF-I gene resulted in cardiomyocytes with reduced sarcomeric structures. A proliferative effect of IGF-I has been shown in transgenic mice overexpressing IGF-I in the mature heart (46). However, it had no apparent effect on cardiac development. The present study documents that the interaction of NRG-1 and IGF-I signaling pathways has a synergistic effect on cushion tissue and ventricular cardiomyocyte growth and maturation. The induced increase in DNA synthesis in ventricular cardiomyocytes was accompanied by normal differentiation, as assessed by down-regulation of MLC-2a. In this regard, proliferation and differentiation can simultaneously occur in cardiomyocytes, as opposed to skeletal muscle cells where these processes appear to be mutually exclusive (47,48). Indeed, the synergistic effect of combined NRG-1 and IGF-I dramatically increased both DNA synthesis and myofibril organization in cultured cardiomyocytes, which indicates the ability of cardiomyocytes to both proliferate and differentiate.
Concerted Activation of PI 3-Kinase Mediated by NRG-1 and IGF-I-Early events of NRG-1-or IGF-I-mediated cellular responses involves the phosphorylation of their cognate receptors inducing recruitment of SH2 domain adapter molecules that belong to the MAP kinase or PI 3-kinase cascades (38 -40, 48, 49). In the present study, the combined effect of NRG-1 and IGF-I appeared to converge on the activation of the PI 3-kinase and was reflected in a significant stimulation of DNA synthesis and differentiation both in cultured cardiomyocytes and in the intact heart in whole embryo culture. The synergistic effect of NRG-1 and IGF-I supports the concept that interplaying signals are required for complete induction of a given phenotypic change (50,51). The accumulated evidence indicates that transactivation of members of the erbB receptor family may occur by direct interaction or erbB2-mediated lateral signal transmission (52,53). Recently, association of gp130 subunits of the interleukin-6 receptor to erbB2 phosphorylation was observed in an interleukin-6-treated carcinoma cell line (54). In addition, activation of the epidermal growth factor receptor mediates intracellular transactivation of the IGF-I receptor and reconstitutes the poliferative and transforming ability of certain IGF-I receptor mutations (36).
It has been shown that either NRG-1 or IGF-I induces convergent signaling through interaction of Ras-MAP kinase and PI 3-kinase pathways, which mediate inducible gene expression (55) or protection from apoptosis (56). In our studies, phosphorylation of the erbB receptors in cardiomyocytes was mediated by NRG-1 alone but not by IGF-I alone, suggesting that there is no cross-activation of the erbB receptors by IGF-I. This may indicate that the presence of both NRG-1 and IGF-I may be required downstream of phosphorylation of erbB2 such as the p85 regulatory subunit of PI 3-kinase. Indeed, after 5 min of NRG-1/IGF-I treatment, erbB2 receptors co-immuno- FIG. 7. Detection of early activation of the erbB receptors and morphological changes induced by NRG-1 and NRG-1/IGF-I in cardiomyocytes. A, tyrosine phosphorylation of erbB2 and erbB4 receptors. Lysates of cardiomyocytes grown in serum-free medium or with NRG-1 for 4 min were subjected to immunoprecipitation (IP) with erbB2 affinity purified antibodies or erbB4 antiserum. Phosphorylation was detected in immunoblots with RC20H phosphotyrosine monoclonal antibodies. B, cardiomyocytes were grown in serum-free medium (Co), or in the presence of NRG-1 (N), IGF-I (I), or NRG-1/IGF (IN) for 72 h and immunostained for myomesin (red) or DAPI (blue). C, tyrosine phosphorylation of the erbB2 receptor was assayed in lysates of cardiomyocytes treated with NRG-1, IGF-I, or both. precipitated with the p85 regulatory subunit of PI 3-kinase (data not shown), suggesting that the biochemical interaction of erbB2 receptors with p85 may involve additional phosphorylation of erbB2 tyrosine residues mediated by combined NRG-1/IGF-I that was below the sensitivity of our phosphorylation assay. Alternatively, the NRG-1/IGF-I-induced phenotypic changes could be mediated by additional growth factors that could, in turn, activate PI 3-kinase.
The contribution of Ras to the NRG-1/IGF-I-mediated effects on cardiomyocytes may occur through convergent activation of PI 3-kinase or may directly affect the cellular response. This latter event would appear to be unlikely because inhibition of mitogen-activated extracellular-signal regulated kinase did not modify the cardiomyocyte response to NRG-1/IGF-I. In this regard, the partial inhibition mediated by dominant negative Ras and the complete block mediated by either dominant negative Rac1 or wortmannin treatment support the concept that convergence of the NRG-1 and IGF-I signaling pathways occurs at the level of PI 3-kinase pathway activation. In wortmannininjected embryos, preliminary results have shown a slight reduction of the basal level and an inhibition of the combined NRG-1/IGF-I-mediated stimulation of cardiomyocyte-incorporated BrdUrd, indicating that PI 3-kinase is indeed required for cardiomyocyte proliferation in the intact embryo. In this regard, it will be of interest to determine the morphologic effects of inhibiting PI 3-kinase using wortmannin and dominant negative Rac1 during normal cardiogenesis in whole mouse culture. Additionally, it will be of interest to examine, in detail, the requirement of the PI 3-kinase during cell proliferation and differentiation that ultimately results in formation and remodeling of cardiac structures during development. It is also intriguing to examine whether the synergistic effect of NRG-1 and IGF-I mediated by activation of PI 3-kinase plays a role in the balance between cell proliferation, survival, and morphological cellular changes during the stress-activated hypertrophic response in the mature heart (57).