Induction of cardiac myogenesis in avian pregastrula epiblast: the role of the hypoblast and activin

The first mesodermally derived organ to form in many vertebrates is the heart, which initially arises as a simple tube consisting of myocardium and endocardium. The early appearance of the heart and its accessibility to experimental manipulation have made it a popular model for investigating mechanisms regulating cell differentiation and organogenesis. In chick, cells that will form the heart are found prior to gastrulation within the caudal half of the epiblast and become progressively localized toward the midline as gastrulation commences (Hatada and Stern, 1994; Rawles, 1943). Heart-forming cells involute during early- to mid-gastrula stages along a broad region of the primitive streak beginning just caudal to Hensen’s node (stage 3a,b; Garcia-Martinez and Schoenwolf, 1993). Following involution, bilateral precardiac regions form on either side of the streak. Anteriorward migration of precardiac cells combined with overall folding of the embryo bring heart-forming regions together along the ventral midline, where they fuse to form myocardium surrounding an endocardial tube (DeHaan, 1963a,b; Rosenquist, 1966).

Specification maps in several species have revealed considerable disparity regarding the timing of myocardial cell specification, which may be partly attributable to differences in rates of development from gastrulation to heart tube formation. Some urodele amphibians, for example, can require a week or more to proceed from gastrulation to formation of a beating heart. In these species, precardiac mesoderm is not capable of self-differentiating in explant culture until mid- to late-neurula stages (reviewed in Jacobson and Sater, 1988). Anuran amphibians such as Xenopus, in contrast, can require less than 24 hours to progress from gastrulation to heart formation. Myocardial cell specification is underway by mid-gastrula stages in Xenopus, as precardiac region mesoderm is capable of differentiating in culture as soon as cells can be isolated following involution (Sater and Jacobson, 1990).

Birds also progress rapidly from gastrulation to formation of a beating heart. Explantation and cell culture studies in chick and quail have shown that myocardial cell specification is well underway by mid-gastrulation (H&H stage 4; Gonzalez-Sanchez and Bader, 1990; Antin et al., 1994; Montgomery et al., 1994; Gannon and Bader, 1995). While this would suggest that signaling interactions regulating establishment of myocardial cell lineages are initiated prior to or during early stages of gastrulation, virtually all studies in birds have focused on the potential role of anterior lateral (AL) endoderm from late gastrula stages onward in regulating myocardial cell specification. Numerous reports have detailed the role of AL endoderm in increasing cell proliferation, rate of myocyte differentiation and degree of heart morphogenesis (DeHaan, 1964; Lough et al., 1990; Antin et al., 1994; Sugi and Lough, 1994; Gannon and Bader, 1995). A definitive role for endoderm in the specification of myocardial cells, however, has been more difficult to demonstrate. Cardiac-inducing activity in AL endoderm was reported by Schultheiss and colleagues (1995), who showed that AL endoderm from stage 5 quail embryos can induce cells within the posterior-most region of chick stage 2-4 primitive streak to become cardiac myocytes. It was suggested that this activity might reflect continuation of an earlier specification signal present within emerging endodermal cells.

Studies in chick suggest that pregastrula hypoblast may be a source of early mesoderm-inducing signals. The hypoblast is involved in the formation of the embryonic axis in the epiblast (Waddington, 1932, 1933; Azar and Eyal-Giladi, 1981; Mitrani and Eyal-Giladi, 1981; see however Khaner, 1995) and may produce early mesoderm-inducing signals similar to those provided by the Xenopus pregastrula vegetal region. In support of this possibility, FGF-2 and activin, two signaling molecules with mesoderm-inducing capacity in Xenopus, are also produced by chick hypoblast (Mitrani and Shimoni, 1990; Mitrani et al., 1990a,b; Ziv et al., 1992). FGF-2 is necessary prior to stage XII for red blood cell development in chick (Gordon-Thomson and Fabian, 1994), and activin can induce axial organization in avian pregastrula epiblast (Mitrani and Shimoni, 1990; Mitrani et al., 1990a,b), as well as the appearance of notochord, skeletal and smooth muscle in fragments of pregastrula and gastrula stage epiblast (Stern et al., 1995).

Tissue interactions and growth factor signals regulating early steps of avian heart muscle cell development have yet to be identified. A major limitation has been the lack of suitable in vitro assays. Ideally, a myocardial cell specification assay should provide both responding and inducing cell layers which, when combined, give rise to heart muscle. Additionally, responding and inducing cell layers should be derived from developmental stages prior to onset of specification. Here we report an assay for investigating signaling interactions leading to the establishment of myocardial cell lineages in birds that meets these criteria. Following a report by Gordon-Thomson and Fabian (1994) indicating that explants from the posterior region of stage XI-XIV blastoderms give rise to beating heart cells, we conducted a survey of the cardiogenic potential of different regions of chick and quail blastoderms. We found that the posterior, but not anterior, region of stage XI-XIV blasto-derms formed heart muscle at high frequency when explanted in defined medium. Since this region contains cells that are normally fated to form heart, a detailed examination was undertaken of the regulation of cardiac myogenesis in these explants. We find that heart muscle cells arise from posterior region epiblast in response to a signal from the hypoblast and that this signal(s) is required prior to stage 3. Heart-inducing capacity of the hypoblast is qualitatively distinct from heart-inducing capacity previously identified in stage 5 AL endoderm. Experiments also indicate that activin can substitute for hypoblast to induce cardiac myogenesis in pregastrula epiblast and is necessary for heart muscle development in intact posterior region explants.

Embryos were removed from fertile chick (Rosemary Farms, Santa Maria, CA; or SPAFAS, Inc., Preston, CT) and quail eggs (Strickland Quail Farm, GA) following 0-24 hours of incubation at 37°C and staged according to Eyal-Giladi and Kochav (EG&K; Eyal-Giladi and Kochav, 1976) for pregastrula stages (stages I-XIV) and according to Hamburger and Hamilton (H&H; Hamburger and Hamilton, 1951) for stages from the beginning of gastrulation (stage 2) onward. Embryos were removed from the egg, placed in 123 mM NaCl and cleaned of yolk using a hair brush. Anterior/posterior orientation was determined by the presence of Koller’s sickle and forming hypoblast. Posterior or anterior regions were excised under a dissecting microscope using a hair bristle. Separation of explants into epiblast and hypoblast cell layers was performed at the time of excision by peeling back the uppermost cell layer(s) with a hair bristle. Care was taken to remove all cells adhering to the epiblast. Recombinations were performed at initiation of culture by overlaying separated cell layers. Heart-forming regions (HFR) of stage 2-5 embryos were excised using a tungsten needle. Explants were transferred to fibronectin-coated Lab Tek chamber slides (Naperville, IL) and cultured in defined medium (75% DMEM:25% McCoy’s medium, supplemented with 10⁻⁷ M dexamethazone and 50 μg/ml gentamycin) in a humidified incubator at 37°C and 7% CO₂ for 72 hours, unless otherwise noted. Human recombinant activin A (Genentech, South San Francisco, CA; and National Hormone and Pituitary Program, Bethesda, MD) was diluted with defined medium plus 0.5 mg/ml bovine serum albumin (BSA; Sigma, St. Louis, MO). Follistatin (National Hormone and Pituitary Program) was prepared as a 1 mg/ml stock in 0.1 NHOAc, and diluted to working concentrations in defined medium plus 0.5 mg/ml BSA. Solutions of follistatin and activin were prepared immediately prior to use. Concentrations used were consistent with effective concentrations reported by others.

RNA was isolated from single or multiple explants from pregastrula chick embryos at time of excision or following 72 hours of incubation in defined medium according to the method of Chomczynski and Sacchi (1987) or using an RNeasy kit (Qiagen, Chatsworth, CA). RNA was treated with 1 U RNAse-free DNAse (Statagene, La Jolla, CA) for 15 minutes and then repurified. Reverse transcription (RT) reactions were performed in 30 μl using random hexamers (Boehringer Mannheim, Indianapolis, IN), 1 U AMV reverse transcriptase (Promega, Madison, WI), 1 U RNAsin (Promega) and 1 μg of total RNA at 37°C for 60 minutes and stored at 4°C. Controls lacking reverse transcriptase were also performed. 50 μl reactions containing Taq buffer (Perkin Elmer, Foster City, CA), 0.5 U Taq polymerase (Perkin Elmer), 0.1 μl [³²P]-α-dCTP and 0.15 μM each of the appropriate primer pair were combined with 3 μl of RT or no-RT reaction mix. GAPDH (51°C), MyoD (55°C), myogenin (55°C), cTnC (56°C) and cNkx-2.5 (51°C) primer sequences, annealing temperatures and amplification conditions are as described in Schultheiss et al. (1995). Cycle number for each primer pair was chosen to fall within linear range of amplification using positive control RNA samples, as assessed using an Instant Imager (Packard Instrument Co., Meriden, CT). Following amplification, 10 μl of each sample was electrophoresed on a 2% Nusieve (FMC Corp, Rockland, ME) gel. Dried gels were exposed to X-ray film or quantitated using Instant Imager software.

Explants were fixed in 4% paraformaldehyde in phosphate-buffered saline (PBS) for 15 minutes at room temperature, then incubated in PBS plus 0.2% Triton X-100, 0.02% sodium azide for 30-45 minutes at room temperature. Explants were incubated in 3% normal goat serum (NGS) in PBS for 15 minutes, then incubated overnight at 4°C in anti-LMM (anti-light meromyosin; a gift from Howard Holtzer, University of Pennsylvania), a fluorescein-conjugated rabbit antibody recognizing striated muscle myosin heavy chain isoforms (Antin et al., 1994), in PBS plus 3% NGS. Explants were rinsed in PBS and incubated for 1 hour at room temperature in 5 μg/ml DAPI (Sigma) in PBS. Following fixation as above, slides of chimeric recombinations were incubated for two hours at 37°C with anti-QCPN (Developmental Studies Hybridoma Bank, Iowa City, IA), a monoclonal antibody that recognizes a quail nuclear antigen not present in chick cells. Explants were then washed in PBS and incubated overnight at 4°C in a combination of anti-LMM and sheep anti-mouse antibody conjugated to biotin (Amersham International, Arlington Heights, IL) in 3% NGS, followed by incubation for 30 minutes at room temperature in 5 μg/ml DAPI and Texas Red-conjugated streptavidin (Amersham International) in PBS. Explants were then washed in PBS, mounted in 90% glycerol containing 1 mg/ml p-phenylenediamine and viewed with a Leitz Diaplan microscope equipped with epifluorescence optics. Explants were scored as positive for cardiac myocytes when containing at least one cluster of four or more brightly fluorescing anti-LMM-stained cells. Positive explants generally had 15-200 contiguous anti-LMM-positive cells arranged in one or more muscular regions. Explants with fewer than four fluorescing cells were scored as negative for heart muscle. Quail cells were distinguished from chick in chimeric explants by observing strong nuclear anti-QCPN staining. Pairwise comparisons of proportions were conducted using a 2-tailed Z test with a pooled estimate of the standard error.

To investigate the cardiac myogenic potential of avian embryos prior to gastrulation, defined regions of EG&K stage XI-XIV quail or chick blastoderms consisting of epiblast and hypoblast (Fig. 1) were cultured for 72 hours and assayed for the appearance of cardiac myocytes. Screens were performed using a polyclonal antisera against myosin heavy chain isoforms present in skeletal and cardiac muscle (anti-LMM) in combination with a monoclonal antibody that recognizes an epitope found specifically in skeletal muscle (12101 antigen; Kintner and Brockes, 1984). While cardiac myocytes were almost never observed in explants from anterior regions of stage XI-XIV blastoderms cultured for 72 hours, one or more aggregates of heart muscle cells were present in almost 80% of quail posterior region explants that included Koller’s sickle and the posterior marginal zone (Figs 2, 3) and in 60% of chick posterior region explants (data not shown). Cardiac myocytes within many cultures exhibited spontaneous rhythmic contractions. Skeletal muscle cells were never observed in anterior or posterior region explants cultured for 72 hours.

These observations were confirmed using RT-PCR (Fig. 4). mRNAs encoding cNkx-2.5 (tinman), the product of a homeobox-containing gene first expressed in myocardial cells a few hours prior to the onset of differentiation (H&H stage 6; Schultheiss et al., 1995), and cardiac troponin C (cTnC), a marker of differentiated cardiac and embryonic skeletal muscle cells (Hastings et al., 1991), were not detectable in posterior region explants at the time of explantation, but were readily detectable in explants following 72 hours of culture (Fig. 4, lanes 1,2). In contrast, MyoD and myogenin, basic helix-loop-helix (bHLH) transcription factors expressed in skeletal myogenic cells (Pownall and Emerson, 1992), were not detectable in posterior region explants at these times. These results demonstrate that explants from the posterior region of avian pregastrula embryos give rise to cardiac, but not skeletal, myocytes during the first 72 hours of culture.

A detailed time course analysis was performed to determine the temporal appearance of cardiac myocytes in quail explants during the initial 72 hour culture period (Fig. 2). Heart muscle cells were first observed after 36 hours of culture and, by 48 hours, greater than 70% of posterior region explants contained one or more dense aggregates of cardiac myocytes. The timing of heart muscle differentiation observed in explants corresponds to the time interval (33-36 hours) between stage XI-XIV and the onset of heart muscle differentiation in vivo at stages 8-10.

The stage XI-XIV blastoderm consists of a dorsal epiblast layer and a forming ventral hypoblast layer (Fig. 1B). To determine whether an interaction between hypoblast and epiblast is necessary for cardiac myogenesis in posterior region explants, quail posterior hypoblast, including Koller’s sickle and ‘middle layer’ cells (Stern and Canning, 1990; Izpisua-Belmonte et al., 1993), was carefully separated from the epiblast and each layer cultured separately for 72 hours. Whereas 78% of intact explants gave rise to heart muscle cells (Fig. 2), separated posterior hypoblast or epiblast gave rise to heart muscle in only 8% (n=86) and 24% (n=54) of cases, respectively. When isolated posterior hypoblast and epiblast layers were recombined, after 72 hours in culture, heart muscle cells were observed in 67% (n=30) of cases. Cells within isolated and recombined layers replicated at roughly comparable rates, and necrotic cells were not observed, indicating that reduced incidence of cardiac myogenesis in explants of isolated hypoblast or epiblast was due neither to the failure of cells to replicate nor to cell death.

Fate mapping studies have shown that the entire embryo proper arises from epiblast while hypoblast gives rise to extraembryonic structures (Vakaet, 1970, 1984; Hatada and Stern, 1994). To verify that heart muscle cells arise from the epiblast layer of posterior region explants, chimeric recombinations between chick hypoblast and quail epiblast were cultured for 72 hours and assayed using anti-LMM and anti-QCPN (quail-specific nuclear antigen). Examination of chimeric recombinations revealed that nuclei of cardiac myocytes invariably bound the QCPN antibody (Fig. 5A-C). The reverse recombination using quail hypoblast and chick epiblast gave rise to heart muscle cells that failed to bind the QCPN antibody (not shown). These findings show that heart muscle cells arise from posterior epiblast in response to an interaction with posterior hypoblast.

Results presented thus far have shown that an interaction with posterior hypoblast is required for appearance of heart muscle cells from posterior epiblast. Posterior region epiblast contains cells that are fated to form heart, however, and therefore it cannot be determined from these experiments whether posterior hypoblast provides an inducing signal that is instructive, capable of changing the fate of cells, or permissive, allowing cells to express a previously attained cardiogenic potential. To distinguish between these possibilities, it is necessary to challenge posterior hypoblast with cells not normally fated to form heart. Recombinations were therefore produced between quail hypoblast and epiblast from stage XII-XIII posterior and anterior regions. 9% (n=22) of intact anterior explants, 0% (n=45) of isolated anterior epiblast (n=45) and 8% (n=24) of anterior epiblast-anterior hypoblast recombinations contained heart muscle cells after 72 hours of culture. However, when posterior hypoblast was recombined with anterior epiblast, or when anterior hypoblast was recombined with posterior epiblast, heart muscle cells were apparent in 40% (n=30) and 48% (n=48) of cases, respectively. For both of these recombinations, the frequency of heart muscle cell differentiation was lower (P<0.01) than observed with posterior homotopic recombinations (67% cardiac myogenesis), indicating that, although both anterior and posterior regions of the hypoblast have at least some inducing capacity and both anterior and posterior epiblast can respond to the inducing signal(s), inducing and responding capacities are highest in the posterior region hypoblast and epiblast, respectively. Regional differences in inductiveness and responsiveness may therefore restrict heart-forming capacity to the posterior region of the blastoderm.

The above results indicate that hypoblast produces an instructive signal that can alter the fate of cells that do not normally form heart. A recent study has shown that AL endoderm from precardiac regions of stage 4-6 quail embryos can induce cardiac myogenesis within chick stage 4 posterior primitive streak (Schultheiss et al., 1995). To compare the inducing capacities of these two cell layers, recombination experiments were performed between stage XII-XIII posterior hypoblast or stage 5 AL endoderm and the following candidate responding layers: (1) stage XII-XIII posterior epiblast (PE), (2) stage 3 posterior lateral epiblast (PLE), (3) stage 4 PLE, (4) stage 4 anterior lateral epiblast (ALE) and (5) stage 4 posterior primitive streak (PPS). In our hands, it was not possible to consistently separate stage 5 AL endoderm from overlying cardiogenic mesoderm, resulting in the occasional appearance of cardiac myocytes in explants of isolated endoderm. Recombinations using stage 5 AL endoderm were therefore performed using the responding cell layer from quail and AL endoderm from chick. Reverse recombinations were also performed. Care was taken to insure that the responding cell layers contacted the dorsal surface of the endoderm, which normally contacts precardiac mesoderm in vivo. For chick-quail chimeric recombinations, the QCPN antibody was used to distinguish between chick and quail cells.

As shown in Fig. 6, stage XII-XIII posterior hypoblast showed broad ability to induce heart muscle development in regions not normally fated to form heart. Recombinations between stage XII-XIII hypoblast and stage XII-XIII PE or stage 3 PLE elicited heart muscle in 67% and 73% of cases, respectively. By stage 4, responsiveness of ALE to stage XII-XIII hypoblast-inducing signal was relatively high (50% exhibiting cardiac myogenesis) while stage 4 PLE showed reduced responsiveness (24% cardiac myogenesis). Schultheiss et al. (1995) have shown that AL endoderm from stage 4-6 embryos can induce heart development in stage 4 PPS fragments, a region with no intrinsic heart-forming capacity. While none of 16 stage 4 PPS fragments contained heart muscle cells when cultured alone, PPS-derived cardiac myocytes were present in 18% of posterior hypoblast-PPS recombinations, indicating that hypoblast also possesses some capacity to convert cells within the PPS to heart muscle cell lineages (P<0.01).

In contrast to the broad heart-inducing capacity of posterior region hypoblast, the ability of stage 5 AL endoderm to induce heart was restricted to stage 4 PPS (Fig. 6). Stage 5 AL endoderm-stage 4 PPS recombinations contained PPS-derived heart muscle cells in 42% of cases (Fig. 6). In contrast, stage 5 AL endoderm showed no ability to induce heart muscle in stage XII-XIII PE, stages 3 or 4 PLE or stage 4 ALE. One explanation for failure of stage 5 AL endoderm to induce heart is that the appropriate inducing signal(s) is produced but at levels below threshold for cardiac induction. To test this possibility, four fragments of stage 5 AL endoderm were recombined with one stage XII-XIII PE fragment. Each recombination was examined to ensure that the epiblast remained intimately associated with the endoderm fragments. Of 8 recombinations meeting these criteria, none contained cardiac myocytes. These results show that posterior pregastrula hypoblast possesses broad ability to induce heart in regions of the epiblast that are not normally fated to participate in cardiac myogenesis, while the ability of stage 5 AL endoderm to induce heart is restricted to cells within the posterior portion of the stage 4 primitive streak. The failure of even four fragments of stage 5 AL endoderm to induce heart muscle in stage XII-XIII posterior epiblast suggests that inducing signals of stage XI-XIV hypoblast and stage 5 AL endoderm are qualitatively distinct.

Explant and cell culture studies in chick and quail have shown that specification of myocardial cells is well underway by mid-gastrulation (stage 4; González-Sanchez and Bader, 1990; Antin et al., 1994; Montgomery et al., 1994; Gannon and Bader, 1995). To determine more precisely the timing of myocardial cell specification, epiblasts from heart-forming regions of stage XIV to stage 3 quail embryos were carefully excised and cultured in defined medium. At stage XIV, cells fated to form heart muscle are localized within the posterior epiblast roughly corresponding to the area excised in our explant studies (Hatada and Stern, 1994). By stage 3a,b (stage 3 subdivisions according to Schoenwolf, 1988), cells that will form the heart are localized to a broad region of the primitive streak posterior to Hensen’s node (González-Sanchez and Bader, 1990). When explanted and maintained in defined culture medium, stage XIV posterior epiblast or stage 2 posterior primitive streak epiblast showed little capacity to form heart muscle (Fig. 7). Epiblast isolated from the heart-forming primitive streak regions of stage 3a,b embryos formed heart muscle in more than 50% of cases. Explants consisting of the entire thickness of the primitive streak heart-forming region gave rise to cardiac myocytes at a similar frequency (62% of cases). These findings indicate that specification of myocardial cells in quail is underway by stage 3a,b of development.

To investigate whether activin is involved in regulating cardiac myogenesis in posterior region explants, explants were cultured in the presence of follistatin, a natural inhibitor of activin function (Kogawa et al., 1991; Nakamura et al., 1990). As shown in Fig. 8, follistatin significantly reduced the incidence of cardiac myogenesis. To determine whether activin could substitute for hypoblast to induce heart development in posterior region epiblast, quail stage XI-XIV posterior region epiblasts were cultured with increasing concentrations of activin and assayed for the presence of cardiac myocytes. Activin induced heart muscle cell development in a dose-dependent manner, with peak induction occurring at 25 ng/ml activin (Fig. 8). Results obtained with follistatin and activin demonstrate that activin, or an activin-like molecule, is necessary for an early step in cardiac myocyte development and can substitute for the hypoblast to induce cardiac myo-genesis in posterior region pregastrula epiblast.

We have found that explants from the posterior region of chick and quail pregastrula blastoderms give rise to cardiac, but not skeletal, myocytes during the first 72 hours of culture. The timing of cardiac myocyte differentiation in explants correlates with the interval between stage XI-XIV and onset of heart muscle differentiation in vivo, suggesting that mechanisms regulating the appearance of cardiac myocytes in explant cultures reflect mechanisms operating in vivo. Experiments demonstrate that appearance of heart muscle cells from the epiblast is dependent upon a signal from the hypoblast and that this signal is required prior to stage 3.

To investigate the extent of responsive and inductive capacities of pregastrula epiblast and hypoblast, heterotopic recombinations were performed between anterior and posterior cell layers. Results show that anterior epiblast, which has no intrinsic heart-forming potential, will give rise to cardiac myocytes if recombined with posterior hypoblast. Posterior hypoblast therefore possesses instructive heart-inducing potential and all regions of the epiblast are capable of responding to this signal. Recombination experiments also show that anterior hypoblast can stimulate heart development in posterior epiblast, suggesting that all regions of the hypoblast possess at least some heart-inducing capacity. Inductive and responsive capacities therefore reside in anterior hypoblast and epiblast, even though intact anterior region explants do not give rise to heart muscle cells. These findings suggest that inductiveness and responsiveness are below threshold levels in anterior cell layers, and that differences in inductive and responsive capacities restrict cardiac myogenic potential to the posterior region of the embryo. The ability of anterior hypoblast to induce heart, and anterior epiblast to produce heart, argues against the possibility that one or both of these layers produces an inhibitory signal that represses cardiac myogenesis.

The hypoblast begins to form around the time of laying (EG&K stage X) from at least two populations of cells. One component migrates anteriorly from the posterior marginal zone and Koller’s sickle, while a second ingresses from the epiblast (Weinberger and Brick, 1982; Eyal-Giladi, 1984, 1991). In chick, a small group of cells have been identified, based upon expression of the goosecoid gene, which lies within a middle layer associated with Koller’s sickle (Izpisua-Belmonte et al., 1993). Fate-mapping studies have shown that some of these cells contribute to Hensen’s node and are therefore not part of the hypoblast proper. Grafts of Koller’s sickle that include the goosecoid-expressing cells can induce an ectopic axis, suggesting that they represent an early population of cells with organizer activity (Callebaut and Van Nueten, 1994; Izpisua-Belmonte et al., 1993). For our experiments, isolation of epiblast involved removal of all ventral cells, including the hypoblast layer, Koller’s sickle and, presumably, the middle layer goosecoid-expressing cells. Heart-inducing properties might therefore reside in the hypoblast or the middle layer cells, or both. The finding that anterior hypoblast can stimulate cardiac myogenesis in posterior epiblast, however, suggests that hypoblast is the source of at least one heart-inducing signal.

The central role of endoderm in the regulation of heart development has been recognized in both amphibians and birds (reviewed in Jacobson and Sater, 1988). In birds, AL endoderm can enhance the rate and degree of myocardial cell differentiation, myofibril assembly and heart tube formation (Lough et al., 1990; Sugi and Lough, 1994; Antin et al., 1994, Gannon and Bader, 1995). It is important, however, to distinguish between influences acting on previously specified myocardial cells from inductive signals leading to specification. In this regard, it has been difficult to reconcile a potential inductive role for stage 5-7 AL endoderm with results presented here and previous studies showing that specification of myocardial cells is well underway by stage 4 (Gonzalez-Sanchez and Bader, 1990; Antin et al., 1994; Montgomery et al., 1994; Gannon and Bader, 1995). Schultheiss and colleagues (1995) have demon-strated that stage 5 AL endoderm can induce heart muscle cell development in stage 4 PPS and suggested that this activity might reflect the continuation of a specification signal that is present within emerging endodermal cells at early stages of gastrulation.

We compared inducing capacities of stage 5 AL endoderm and posterior pregastrula hypoblast by performing a series of recombination experiments between either of these cell layers and various candidate responding cell populations. Results demonstrate qualitative differences between the ability of AL endoderm and hypoblast to induce cardiac myogenesis. While hypoblast possesses broad capacity to induce heart muscle cells in both pregastrula and gastrula stage epiblast, stage 5 AL endoderm shows no capacity to induce heart in epiblast cells. Of the candidate responding regions tested, stage 5 AL endoderm was capable of inducing heart only in stage 4 PPS, a cell population in which posterior hypoblast showed only modest heart-inducing capacity. These results suggest that, while stage 5 AL endoderm cannot induce heart muscle from epiblast, there is a period during which AL endoderm can shift the fate of emerging mesodermal cells towards a myocardial phenotype. Differences between the ability of these two cell layers to induce heart muscle in posterior epiblast appear to be qualitative, as four fragments of AL endoderm were not capable of inducing cardiac myocytes in pregastrula posterior epiblast.

Several potential models can explain these results. First, the posterior hypoblast (perhaps including Koller’s sickle and middle layer cells) might produce a signal(s) that leads directly to myocardial cell specification within a localized region of the emerging primitive streak prior to stages 2-3 of development. Once myocardial cells are specified, emerging endoderm within the cardiogenic region would maintain cells within myocardial lineages, perhaps against competing signals from other cell layers. AL endoderm would also increase proliferation and the rate of myocyte differentiation, and promote myofibril assembly and heart tube formation.

A second possibility is that hypoblast produces a signal that acts simultaneously with a signal from emerging endoderm to specify myocardial cells. This dual signal model proposes a mechanism that is consistent with the findings of Nascone and Mercola (1995) in which myocardial cell specification in Xenopus requires a signal from both Spemann’s Organizer and deep endoderm during early gastrula stages. In birds, those signals might arise from the early Hensen’s node and endoderm prior to stages 2-3.

A third possibility is that the hypoblast/middle layer cells might initiate processes that lead to the appearance of nascent mesoderm, and perhaps endoderm, within the early primitive streak. Emerging endoderm within the heart-forming region of the primitive streak might then provide a specification signal. In this case, the hypoblast-derived signal(s) would not directly specify myocardial cells but rather initiate a signaling cascade that ultimately leads to specification. This would be consistent with our results showing that although AL endoderm was unable to induce heart from epiblast, cells within the stage 4 PPS were receptive to endoderm-derived heart-inducing signals. Conversely, the ability of hypoblast, but not AL endoderm, to induce heart muscle cell development in epiblast cells suggests that the hypoblast signal precedes any potential signal from AL endoderm. Although additional experiments will be required to determine which scenario is most accurate, we feel that the data presently favors the third model. The largely complementary inducing capacities of the hypoblast and AL endoderm suggest that multiple, temporally distinct signals are involved in regulating early stages of cardiac myo-genesis.

An important unanswered question is the identity of potential heart-inducing molecules from the hypoblast. Of the known growth factors produced by the hypoblast, activin, has been shown to induce an embryonic axis in the epiblast as well as several mesodermal cell types in area opaca epiblast (Mitrani et al., 1990a; Stern et al., 1995). Induction of heart muscle cells by activin in birds, however, has not been previously reported. We find that activin can induce cardiac myogenesis in posterior epiblast in a dose-dependent manner at concentrations consistent with induction of other mesodermal cell types. Our experiments also show that follistatin, an inhibitor of activin function, can block cardiac myogenesis in posterior region explants. Collectively, these findings indicate that activin, or an activin-like molecule, is both necessary for cardiac myogenesis and is sufficient to induce heart muscle cell development in the epiblast. Activin may therefore represent at least one cardiogenic signal produced by the hypoblast. A recent study has shown that BMP-2 and FGF-4 are produced by AL endoderm, and in combination can induce cardiac myogenesis in stage 5-6 posterior mesoderm (Lough et al., 1996). Although activin is also produced by AL endoderm and can support survival and differentiation of pre-cardiac mesoderm (Kokan-Moore et al., 1991; Sugi and Lough, 1995), it cannot induce heart muscle cell development in posterior mesoderm (J. Lough, personal communication). Activin may therefore provide an early inductive signal and later play a supporting role in promoting development of specified premyocardial cells. Beginning with pregastrula epiblast, a two-step model for cardiac myogenesis might therefore involve an early activin-like signal followed by the combined action of BMP-2 and FGF-4. Additional growth factors produced by AL endoderm, including FGF-2, IGF-II, insulin and activin might further stimulate and enhance heart development (Parlow et al., 1991; Sugi et al., 1993; Sugi and Lough, 1995; Antin et al., 1996). Additional experiments will be required to more fully define the roles of these growth factors in cardiac myogenesis.

We thank John Lough for helpful discussions and comments on the manuscript. This work was supported by grants to P. B. A. from the NIH (HL54133 and HL20220) and the American Heart Association, Arizona Affiliate. A. N. L. is a Howard Hughes Medical Institute Pre-doctoral Fellow.