|Home | About | Journals | Submit | Contact Us | Français|
The function of an organ relies upon the proper relative proportions of its individual operational components. For example, effective embryonic circulation requires the appropriate relative sizes of each of the distinct pumps created by the atrial and ventricular cardiac chambers. Although the differences between atrial and ventricular cardiomyocytes are well established, little is known about the mechanisms regulating production of proportional numbers of each cell type. We find that mutation of the zebrafish type I BMP receptor gene alk8 causes reduction of atrial size without affecting the ventricle. Loss of atrial tissue is evident in the lateral mesoderm prior to heart tube formation and results from the inhibition of BMP signaling during cardiac progenitor specification stages. Comparison of the effects of decreased and increased BMP signaling further demonstrates that atrial cardiomyocyte production correlates with levels of BMP signaling while ventricular cardiomyocyte production is less susceptible to manipulation of BMP signaling. Additionally, mosaic analysis provides evidence for a cell-autonomous requirement for BMP signaling during cardiomyocyte formation and chamber fate assignment. Together, our studies uncover a new role for BMP signaling in the regulation of chamber size, supporting a model in which differential reception of cardiac inductive signals establishes chamber proportion.
Organ function is the cumulative effect of coordinating each of an organ's individual subunits. In the vertebrate heart, the propulsion of circulation depends upon the serial contraction of each of its independent pumps, the atrial and ventricular cardiac chambers. Therefore, attainment of proper cardiac chamber proportionality is crucial for effective cardiac function.
Little is known about the mechanisms that regulate the relative sizes of the cardiac chambers. Atrial and ventricular cardiomyocytes are distinct cell types that differ both histologically and physiologically (Moorman and Christoffels, 2003). For example, in the zebrafish heart tube, chamber cardiomyocytes have distinct cell morphologies, with atrial cardiomyocytes appearing more squamous and ventricular cardiomyocytes appearing more cuboidal (Rohr et al., 2008). Additionally, zebrafish atrial and ventricular cardiomyocytes express the chamber-specific myosin heavy chains, atrial myosin heavy chain (amhc) and ventricular myosin heavy chain (vmhc) (Berdougo et al., 2003; Yelon et al., 1999). Although several transcription factors have been implicated in the regulation of chamber-specific characteristics (Bruneau, 2002), the signaling pathways that trigger initial chamber fate assignment and generate the appropriate numbers of atrial and ventricular cardiomyocytes are poorly understood.
Analysis of the mechanisms regulating chamber size is relatively convenient in the simple two-chambered zebrafish heart, in contrast to the four-chambered mammalian heart, which arises from multiple heart fields (Buckingham et al., 2005; Cai et al., 2003; Kelly et al., 2001; Meilhac et al., 2004). Fate map experiments in zebrafish reveal that cardiac progenitors are located at the lateral margin of the embryo at the onset of gastrulation (Keegan et al., 2004). Even at this early stage, chamber progenitor populations are already spatially organized. Within the lateral marginal zone of the blastula, atrial progenitors are located more ventrally, and ventricular progenitors are located more dorsally; additionally, ventricular progenitors tend to be closer to the margin than are the atrial progenitors (Keegan et al., 2004). The relative organization of chamber progenitors seems to be maintained during and after gastrulation (Keegan et al., 2004; Schoenebeck et al., 2007). Cardiac specification is thought to take place during the time window in which the migrating progenitor cells form the lateral plate mesoderm, culminating in robust expression of precardiac markers like nkx2.5 around the 6-8 somite stage (Thomas et al., 2008). Myocardial differentiation then initiates around the 13-14 somite stage (Yelon et al., 1999), and chamber-specific gene expression is first evident at the 14 somite stage with the expression of vmhc (Yelon et al., 1999) and at the 19 somite stage with the expression of amhc (Berdougo et al., 2003).
Recent work in zebrafish supports an emerging idea regarding the acquisition of chamber identity: the signals responsible for cardiac induction may also be responsible for the subdivision of the heart field into atrial and ventricular territories. These studies demonstrated that mild inhibition of cardiac inductive signals can lead to differential reduction of atrial and ventricular chambers. For example, although complete inhibition of Nodal signaling eliminates all cardiac mesoderm (Gritsman et al., 1999), modest loss of Nodal signaling results in reduced heart size with a particularly striking reduction of the ventricle (Reiter et al., 2001). Fate mapping experiments revealed that Nodal promotes ventricular fate assignment in blastomeres nearest to the margin (Keegan et al., 2004), consistent with a vegetal-animal gradient of Nodal signaling that positions the highest levels of Nodal signaling closest to the margin (Chen and Schier, 2002). A similar trend is observed in experiments inhibiting the FGF signaling pathway, which has also been implicated in promoting cardiac induction (Dunwoodie, 2007). Zebrafish fgf8 mutants form small hearts with disproportionately small ventricles (Reifers et al., 2000), and temporal inhibition of FGF signaling has demonstrated its particular importance for ventricular cardiomyocyte formation during gastrulation stages (Marques et al., 2008), consistent with a dorsal-ventral gradient of FGF signaling that correlates with the more dorsal location of ventricular progenitors (Fürthauer et al., 1997; Keegan et al., 2004). Together, these studies suggest that differential reception of cardiac inductive signals within a uniform pool of cardiac progenitors could dictate chamber identity assignment.
BMP signaling has a conserved role in cardiac induction (Zaffran and Frasch, 2002). In Drosophila, dpp mutant embryos lack expression of tinman, an essential cardiac transcription factor gene, and do not form a dorsal vessel (Frasch, 1995). Conversely, ectopic Dpp can induce formation of ectopic tinman-expressing cells (Yin and Frasch, 1998). In chick, exogenous application of BMP also induces expression of the tinman homolog NKX2-5 (Nakajima et al., 2002; Schlange et al., 2000; Schultheiss et al., 1997). Additionally, excess BMP signaling in mouse Nkx2-5 mutants is thought to be responsible for the overspecification of cardiac progenitors (Prall et al., 2007). As in Drosophila, severe loss of cardiac tissue is also observed when BMP signaling is inhibited in vertebrates. In frog, inhibition of BMP signaling results in the absence of a morphologically recognizable heart (Shi et al., 2000). In mouse Bmp2 mutants, expression of Nkx2-5 is reduced or absent (Zhang and Bradley, 1996), and conditional removal of Bmpr1 from precardiac mesoderm with a Mesp-1 Cre driver dramatically reduces the heart field (Klaus et al., 2007). Similarly, zebrafish swirl (bmp2b) mutants lack expression of nkx2.5 (Kishimoto et al., 1997; Nguyen et al., 1998; Reiter et al., 2001).
While strong reduction of BMP signaling severely compromises heart development, mild reduction results in formation of a dysmorphic heart (Chocron et al., 2007). Zebrafish lost-a-fin (laf) mutants exhibit a weak dorsalized phenotype characterized by loss of the ventral tail fin, due to disruption of the type I BMP receptor gene alk8 (Bauer et al., 2001; Mintzer et al., 2001). Maternal-zygotic laf mutants are morphologically similar to swirl (Mintzer et al., 2001). In contrast to the dramatic loss of cardiac tissue observed in swirl mutants, zygotic laf mutants form hearts with abnormal atrial morphology (Chocron et al., 2007). Because of the correlation between the relatively ventral location of atrial progenitors and the ventral-dorsal gradient of BMP signaling during gastrulation stages (Tucker et al., 2008), we hypothesized that the abnormal atrial morphology in laf mutants could reflect a role for BMP in atrial cardiomyocyte generation.
Here, we analyze the effects of BMP signaling on cardiac chamber size in the zebrafish embryo. We find that laf mutants exhibit a significant reduction in the number of atrial cardiomyocytes. Atrial tissue reduction occurs before the heart tube forms and results from inhibition of BMP signaling during cardiac progenitor specification stages. Conversely, high levels of BMP signaling result in a preferential increase of atrial size. Through mosaic analysis, we show that BMP signaling is cell autonomously required for cardiomyocyte formation and chamber identity acquisition. Together, our data reveal a crucial role for BMP signaling in regulating cardiac chamber proportion through differential effects on atrial and ventricular progenitor populations.
We employed the following zebrafish mutations and transgenes: lost-a-finsk42 (laf/alk8), Tg(cmlc2:DsRed2-nuc)f2 (Mably et al., 2003), Tg(cmlc2:egfp)twu277(Huang et al., 2003) and Tg(hsp701:dnBmpr-GFP)w30 (Pyati et al., 2005). The lafsk42 mutation is a recessive lethal allele that was identified in a screen for ethylnitrosourea-induced mutations that affect cardiac chamber morphology (Auman et al., 2007). lafsk42 fails to complement laftm110b, and the phenotypes of lafsk42 and laftm110b mutants appear indistinguishable (Bauer et al., 2001; Chocron et al., 2007; Mintzer et al., 2001; Mullins et al., 1996). All zebrafish work followed protocols approved by the NYU School of Medicine IACUC.
We counted cardiomyocytes in embryos carrying the transgene Tg(cmlc2:DsRed2-nuc), using immunofluorescence to detect DsRed in cardiomyocyte nuclei and to detect atrial myosin heavy chain (Amhc) in atrial cells, as described previously (Marques et al., 2008). Student's t-test (homocedastic, two-tail distribution) was used to analyze the differences between the means of cell number data sets.
Anti-sense probes for amhc, vmhc, nkx2.5, hand2, and gata5 were used as previously described for in situ hybridization (Berdougo et al., 2003; Chen and Fishman, 1996; Reiter et al., 2001; Yelon et al., 1999; Yelon et al., 2000).
A 10 mM stock of dorsomorphin (Calbiochem, #171261) in DMSO was diluted to a working concentration of 5 μM in E3 medium (Nusslein-Volhard and Dahm, 2002), a concentration previously described to cause dorsalization (Yu et al., 2008). Up to 10 embryos were treated in a final volume of 1 mL in a glass vial. Vials were kept on a nutator in the dark at 28.5°C. Control embryos were treated with a corresponding dilution of DMSO.
Embryos from outcrosses of fish heterozygous for Tg(hsp701:dnBmpr-GFP) were exposed to heat shock at desired stages through placement in 40 mL of embryo medium in a Petri dish on top of a covered heat block for 1 hour at 37°C. Following heat shock, transgenic embryos were identified by gfp expression and body morphology. Transgenic embryos heat shocked during gastrulation exhibit a previously described mild dorsalization, while embryos heat shocked at the 8-somite stage are grossly comparable to wild-type (Pyati et al., 2006; Pyati et al., 2005).
alk8CA and alk8(K232R) mRNAs were in vitro transcribed from previously described plasmids using the SP6 mMessage mMachine System (Ambion) (Bauer et al., 2001). Embryos were injected at the 1-cell stage with 15 pg of alk8CA mRNA or 20 pg of alk8(K232R) mRNA. Injection of either mRNA causes a range of phenotypes; embryos were sorted into categories according to previously described scales of ventralization and dorsalization (Kishimoto et al., 1997; Mullins et al., 1996).
Transplantation was performed at midblastula stages as previously described (Ho and Kane, 1990; Parker and Stainier, 1999). 10-15 donor blastomeres were placed into the margin of a wild-type host embryo. Donor embryos carried either Tg(cmlc2:egfp) or fluorescein as a lineage tracer to allow detection of donor-derived cells in the host at 48 hours post fertilization (h.p.f.). Donor embryos were either wild-type, injected with 1.5 pg of alk8CA mRNA, or injected with 2 ng of anti-smad5 morpholino (5′-ATGGAGGTCATAGTGCTGGGCTGC -3′), the characterization of which will be described elsewhere (J.A. Tucker, V.H. Nguyen, A.P. Wiemelt, and M.C. Mullins, manuscript in preparation). At this concentration, the anti-smad5 morpholino causes severe dorsalization without apparent nonspecific defects. Fisher's exact test was used to analyze the differences between frequencies of contribution to the myocardium.
Images were captured with Zeiss Axioplan and M2Bio microscopes equipped with Zeiss Axiocam digital cameras and AxioVision software. Adobe Photoshop CS3 software was used to process images.
Prior studies of zebrafish laf (alk8) mutants have shown that loss of laf function results in formation of a linear, dysmorphic heart (Chocron et al., 2007). By 48 h.p.f., both the atrium and the ventricle are visibly misshapen in laf mutants (Fig. 1A,B). Quantification of the number of cardiomyocytes in each chamber reveals that the laf mutant atrium contains many fewer cardiomyocytes than the wild-type atrium (Fig. 1D), accounting for its shrunken appearance. In contrast, ventricular cell number is normal in laf mutants (Fig. 1D). The shortage of atrial cells in laf mutants is evident even prior to heart tube assembly (Fig. 2A,B), as shown by the reduced number of cells expressing amhc. Consistent with ventricular size at 48 h.p.f., expression of vmhc is relatively normal at earlier stages in laf mutants (Fig. 2D,E). The diminished population of atrial cardiomyocytes in laf mutants implicates BMP signaling in the regulation of atrial chamber size.
The reduced number of atrial cells in laf mutants could be the consequence of an early defect in atrial progenitor specification or a later defect in atrial growth or differentiation. To determine whether the laf cardiac phenotype originates prior to myocardial differentiation, we examined molecular markers of cardiac progenitor populations in the anterior lateral plate mesoderm (ALPM) at the 10-somite stage. At this stage, cardiac progenitors reside within the hand2 expression domain of the ALPM, with ventricular progenitors occupying a more medial territory defined by nkx2.5 expression and atrial progenitors occupying a more lateral territory that does not yet express nkx2.5 (Fig. 3A) (Schoenebeck et al., 2007). In keeping with previous reports (Chocron et al., 2007; Hogan et al., 2006), we find that nkx2.5 expression in laf mutants is comparable to that seen in wild-type embryos (Fig. 3B,E). However, the hand2 expression domain is considerably narrower in laf mutants (Fig. 3C,F). Likewise, gata5 expression, which is normally expressed throughout the ALPM (Schoenebeck et al., 2007), appears in thinner strips of this tissue in laf mutants (Fig. 3D,G). Together, these expression patterns point to a defect in the width of the laf mutant ALPM, which also appears abnormally thin when visualized using Nomarski optics (SRM and DY, unpublished data). Moreover, the normal width of nkx2.5 expression in laf mutants suggests that the missing portion of the LPM may be a lateral portion of the heart field, which normally contains atrial progenitors. Thus, laf function seems to influence formation of the atrial lineage from an early stage, well before myocardial differentiation begins.
Based on the laf ALPM phenotype, it seems likely that bmp2, bmp4, and bmp7, all of which are expressed during gastrulation on the ventral side of the embryo (Dick et al., 2000; Martínez-Barberá et al., 1997), could be involved in establishing appropriate atrial cell numbers. Alternatively, bmp4 could also influence cardiac development at later stages, when it is expressed in the ALPM and cardiomyocytes (Martínez-Barberá et al., 1997). To confirm the importance of BMP signaling prior to myocardial differentiation, we temporally inhibited signaling using dorsomorphin, a pharmacological inhibitor of type I BMP receptors (Yu et al., 2008). Consistent with previous studies (Yu et al., 2008), we found that exposure to 5 μM dorsomorphin from 50% epiboly onward results in the loss of ventral tail fin (Fig. 4A,B), mimicking the laf phenotype (Bauer et al., 2001; Mintzer et al., 2001). This period of dorsomorphin exposure also results in a significant reduction of atrial cardiomyocyte number, without a significant change in ventricular cell number (Fig. 4D), again reminiscent of laf mutants (Fig. 1D). Dorsomorphin exposure from tailbud stage onward causes mild ventral tail fin defects (Fig. 4C) and cloaca defects (SRM and DY, unpublished data), consistent with previously described roles for BMP signaling (Pyati et al., 2006), but this treatment does not affect cardiomyocyte number (Fig. 4D). We obtained similar results with a different method of inhibiting BMP signaling, employing Tg(hsp701:dnBmpr-GFP) transgenic embryos that express a heat-inducible dominant negative form of Alk8 (Pyati et al., 2005). Heat shock at 65% epiboly causes loss of ventral tail fin (Pyati et al., 2005), as well as fewer atrial and ventricular cardiomyocytes, with a more striking deficit of atrial cells (Fig. 4E). However, heat shock at later stages did not have a significant impact on cardiomyocyte number (Fig. 4E). Thus, both pharmacological and genetic inhibition point to an early requirement for BMP signaling, during cardiac progenitor specification stages, in the generation of the proper numbers of cardiomyocytes.
Throughout gastrulation stages, a steep ventral to dorsal gradient of BMP signaling is present in the zebrafish embryo and can be visualized by the spatial distribution of BMP ligands and phosphorylated Smad5, a transcription factor activated by activation of Alk8 (Bauer et al., 2001; Hild et al., 1999; Martínez-Barberá et al., 1997; Mintzer et al., 2001; Tucker et al., 2008). Therefore, the relatively ventral atrial progenitors are likely to be exposed to higher levels of BMP signaling than are the more dorsal ventricular progenitors, suggesting that differential levels of BMP signaling could affect chamber progenitor specification.
To test whether increased levels of BMP signaling could result in atrial population expansion, we injected embryos with mRNA encoding a constitutively active form of Alk8 (alk8CA), which is able to signal independently of ligand and type II receptor (Bauer et al., 2001; Wieser et al., 1995). Injection of alk8CA mRNA creates a range of phenotypes, ranging from highly ventralized to relatively normal body morphologies (Bauer et al., 2001). The most ventralized class of injected embryos (the v2-v3 class, according to a previously described system of phenotypic classification (Kishimoto et al., 1997)) display an enlarged atrium, with a large surplus of atrial cardiomyocytes as well as a slightly enhanced number of ventricular cardiomyocytes (Figs. 1C, 1D, 2C, 2F). Together with the laf mutant phenotype, these results suggest a correlation between levels of BMP signaling and atrial size.
To examine a broader spectrum of BMP signaling levels, we compared a diverse array of embryos injected with mRNA encoding either alk8CA or a dominant negative kinase dead form of Alk8 (alk8(K232R)) (Bauer et al., 2001). Akin to the range of effects of alk8CA injection, alk8(K232R) mRNA injection creates a spectrum of dorsalized phenotypes (Bauer et al., 2001). We grouped both sets of injected embryos into previously described classes of ventralized or dorsalized body morphologies and quantified chamber cardiomyocyte number for each class (Fig. 5). The most ventralized embryos, characteristic of high levels of BMP signaling, displayed atrial enlargement (Figs. 1C, ,5A),5A), whereas the most dorsalized embryos, characteristic of low levels of BMP signaling, displayed small hearts (Fig. 5C). Comparing all classes of embryos, we find a correlation between level of BMP signaling and total cardiomyocyte number (Fig. 5D and Table 1). This positive correlation is also present for the number of atrial cardiomyocytes, which is significantly increased in the two most ventralized phenotypic classes, as opposed to the number of ventricular cardiomyocytes, which is not as enhanced in these embryos (Fig. 5D and Table 1). Likewise, antagonism of BMP signaling differentially affects the numbers of atrial and ventricular cardiomyocytes, with a stronger effect on the atrial population (Fig. 5D and Table 1). Thus, atrial size seems highly dependent on levels of BMP signaling, whereas the role of BMP signaling is less striking for the ventricle.
Our results suggest that BMP signaling plays an early role in promoting the acquisition of cardiac identity, particularly atrial identity. However, these data do not indicate whether BMP signaling plays a direct role in this regard. Recent work in mouse has shown that BMP signal reception is required within Mesp-1-expressing mesoderm for cardiomyocyte generation (Klaus et al., 2007). We wondered whether individual cardiac progenitors require direct reception of BMP signaling in order to acquire cardiac identity and whether the level of BMP signaling received impacts chamber identity choice. To address this, we performed blastomere transplantation to compare the contributions of wild-type cells and cells deficient in BMP signal transduction to the cardiac chambers of wild-type host embryos. To inhibit BMP signal transduction, we used a morpholino (MO) to interfere with Smad5 translation. Embryos injected with anti-smad5 MO (smad5 morphants) exhibit more severely dorsalized bodies and stronger cardiac phenotypes than those observed in laf mutants (SRM and DY, unpublished data). Donor-derived cardiomyocytes were recognizable in host embryos due to their expression of the transgene Tg(cmlc2:egfp) (Fig. 6A,B). In wild-type-into-wild-type control transplants, we found donor-derived cardiomyocytes in 13.2% of our experiments (15 out of 114 transplants) (Fig. 6C). Strikingly, in smad5 morphant-into-wild-type transplants, donor-derived cells contributed to the myocardium much less frequently, becoming cardiomyocytes in only 1.2% of our experiments (2 out of 170 transplants), a significant difference from the effectiveness of wild-type donor cells (Fig. 6C). The poor ability of smad5 morphant cells to contribute to the heart does not reflect a general problem with the viability of cells from morphant donors: both wild-type and smad5 morphant donor-derived cells contribute equally well to other tissues, such as skeletal muscle (Fig. 6D-F). Taken together, our results indicate that BMP signaling is required cell autonomously for cardiomyocyte formation.
Since cell-autonomous receipt of BMP signaling is crucial for the generation of cardiomyocytes (Fig. 6A-C), we could not use smad5 morphant donors to evaluate conclusively whether BMP signal reception plays a cell-autonomous role during chamber fate assignment. Instead, we compared the contributions of wild-type cells and cells injected with alk8CA mRNA to the cardiac chambers of wild-type hosts. We first employed donor embryos injected with the same amount of alk8CA mRNA as in our other experiments (Figs. 1C, 2C, 2F, ,5A).5A). In these transplants, alk8CA mRNA-injected donor-derived cells were very inefficient in contributing to the myocardium (1.6% contribution frequency; 2 out of 122 transplants), but contributed normally to skeletal muscle (75.6% contribution frequency; 31 out of 41 transplants). Together with our smad5 morphant transplants, this result suggests that neither particularly high nor particularly low levels of BMP signaling allow donor-derived cells to become cardiomyocytes in the context of a wild-type host. Next, we injected donor embryos with a ten-fold lower concentration of alk8CA mRNA. In these transplants, alk8CA mRNA-injected donor-derived cells were more capable of contributing to myocardium (10.8% contribution frequency; 20 out of 185 transplants), closer to the effectiveness of wild-type cells (14.4% contribution frequency; 26 out of 181 transplants). If BMP signaling plays a cell-autonomous role in chamber identity acquisition, then we expected injection of this dose of alk8CA mRNA to affect the relative likelihood of donor-derived cells contributing to each chamber, favoring atrial over ventricular contributions. We found a significant difference in contribution to the ventricle, with alk8CA mRNA-injected cells contributing to the ventricle in only 5.9% of experiments (11 out of 185 transplants) and wild-type cells contributing to the ventricle in 12.2% of experiments (22 out of 181 transplants) (Fig. 7). In contrast, alk8CA mRNA-injected cells contributed to the atrium in 8.6% of experiments (16 out of 185 transplants), a subtle increase in contribution compared to that seen for wild-type cells, which contributed to the atrium in 6.6% of experiments (12 out of 181 transplants) (Fig. 7). Our data show that the likelihood of transplanted alk8CA mRNA-injected cells to contribute to one cardiac chamber or the other differs from the likelihood of wild-type contributions, indicating that the amount of BMP signaling received by transplanted cells affects their fate. Together, our data suggest that cell-autonomous receipt of BMP signaling influences chamber proportionality by biasing chamber fate assignment of progenitor cells.
Our studies uncover a previously unrecognized role of BMP signaling in atrial development and provide a new perspective on the roles of BMP signaling during heart formation. In addition to their established roles as cardiac inductive signals, we show that BMPs are also essential for the generation of the appropriate relative numbers of atrial and ventricular cardiomyocytes. Manipulations of BMP signaling that cause aberrations in total cardiomyocyte number also cause distortion of the atrial/ventricular ratio, supporting a model in which both functions of BMP signaling occur simultaneously during gastrulation stages. We propose that receipt of BMP signals by aspiring cardiac progenitor cells promotes specification of their cardiac identity and also biases their chamber assignment, with higher levels of BMP signaling favoring atrial over ventricular fate. Thus, through differential effects on atrial and ventricular lineages, BMP signaling influences chamber size and thereby chamber proportionality.
The effect of BMP signaling on chamber fate assignment correlates nicely with the organization of myocardial progenitor cells within the lateral margin of the zebrafish blastula and its overlap with the described BMP signaling gradient. Within the lateral margin, Smad5 phosphorylation has been shown to be higher in the relatively ventral portion, which contains atrial progenitors, than it is in the relatively dorsal portion, which contains ventricular progenitors (Fig. 8A) (Keegan et al., 2004; Tucker et al., 2008). In a very simple model of how BMP might pattern the heart field, myocardial progenitors exposed to high levels of BMP signaling would become atrial progenitors, in parallel to the mechanism by which high levels of BMP signaling specify other ventral cell types (Kondo, 2007; Nguyen et al., 1998). Consistent with this model, we find that the atrial lineage is more susceptible than the ventricular lineage to both loss and gain of BMP signaling. This model would also predict that intermediate or low levels of BMP signaling would specify ventricular fate, in analogy to the role of BMP in specifying other lateral cell types (Nguyen et al., 1998; Stickney et al., 2007; Tucker et al., 2008). However, in our experiments, decreased levels of BMP signaling did not result in increased production of ventricular cardiomyocytes, and none of our alterations of BMP signaling provided results consistent with a fate transformation between atrial and ventricular lineages. Therefore, although BMP signaling has a potent influence on chamber proportionality, differential exposure to BMPs is not sufficient to pattern a uniform progenitor field into atrial and ventricular populations.
The lack of evidence for an atrial-ventricular fate transformation caused by manipulation of BMP signaling probably reflects the fundamental mechanism by which the cardiac progenitor lineages are separated. Instead of a uniform progenitor pool that is divided into two by differential reception of BMP signaling, we favor the notion of two separate progenitor pools with distinct requirements for BMP signaling (Fig. 8B). A shared pool model predicts that high BMP levels would substantially increase the atrial/ventricular ratio and that low BMP levels would substantially reduce it. However, our observed variations in proportionality are mild and do not support a shared pool model. Furthermore, in highly dorsalized embryos, the atrial/ventricular ratio is closer to the wild-type ratio than to the ratio in laf mutants. Therefore, our observations are more consistent with a separate pools model. The atrial and ventricular lineages are already separate by midblastula stages (Keegan et al., 2004; Stainier et al., 1993), so the two progenitor pools may already have distinct requirements for BMP signaling. These pools would not only receive distinct amounts of BMP signaling but also differ in their responsiveness to BMP signaling. Since alk8 and smad genes are expressed ubiquitously throughout gastrula stages (Dick et al., 1999; Müller et al., 1999; Yelick et al., 1998), the difference in responsiveness is not likely to be due to differences in these components of the BMP signaling pathway.
A separate pools model would also be consistent with our prior observations regarding the role of FGF signaling during chamber fate assignment (Marques et al., 2008). We have found that the ventricular lineage is more sensitive than the atrial lineage to loss of FGF signaling, correlating with the location of ventricular progenitors nearer to high levels of FGF signaling (Fürthauer et al., 1997; Keegan et al., 2004). However, low levels of FGF signaling do not result in increased production of atrial cardiomyocytes (Marques et al., 2008), again arguing against a shared pool model.
Together, our studies of BMP and FGF signaling suggest that chamber proportionality is established by a network of interacting pathways, which create a system buffered against the disruption of a single pathway. The BMP and FGF pathways are unlikely to be the only participants in this network, since chamber proportionality is not restored in fgf8;alk8 double mutants (SRM and DY, unpublished data). In addition to other ventricularizing influences like the Nodal signaling pathway, the network could also include other atrializing influences, such as the retinoic acid (RA) signaling pathway, which has been suggested to specify atrial identity at the expense of ventricular identity in chick (Hochgreb et al., 2003; Xavier-Neto et al., 2001). However, recent studies in zebrafish indicate that a loss of RA signaling increases both atrial and ventricular populations, without evidence of a fate transformation between the lineages (Waxman et al., 2008). Further studies will be necessary to identify additional signaling pathways that interact with the BMP pathway during atrial fate assignment.
Whether BMP signaling influences the choice between atrial and ventricular identities or the choice between cardiac and non-cardiac identities, it is clearly playing a direct role in the decisions made by cardiac progenitors. Because of the early requirement for BMP signaling and the early ALPM defects observed in laf mutants, we favor the notion that BMP signaling directly affects progenitor specification or migration into the ALPM. Although we cannot rule out the possibility that BMP signaling influences progenitor proliferation, it seems unlikely that BMP signaling regulates progenitor survival since we do not detect an increased number of apoptotic cells within the heart fields of laf mutants (SRM and DY, unpublished data). In future studies, it will be important to identify the downstream targets responsible for the effects of BMPs on chamber fate acquisition. These targets may be subsets of genes in the myocardial differentiation program that dictate chamber identity, especially since Smad binding sites are known to modulate expression of precardiac genes like Nkx2-5 in chick and mouse (Brown et al., 1998; Lee et al., 2004; Liberatore et al., 2002; Lien et al., 2002). Additionally, BMP targets might play important roles in directing the patterns of movement of progenitor cells as they create the heart fields in the ALPM. Abnormal migration of myocardial progenitors causes reduction of heart size (Scott et al., 2007; Zeng et al., 2007). During gastrulation, ventrally located BMP signals have a repulsive effect on lateral mesodermal cells, which fail to migrate dorsally in severely reduced BMP signaling scenarios (von der Hardt et al., 2007). Therefore, it is possible that the migration of the more ventral atrial progenitors into the ALPM is more affected by levels of BMP signaling than is the migration of ventricular progenitors.
Taken together, our results indicate that BMP signaling regulates heart size and chamber proportionality by differentially regulating atrial and ventricular cardiomyocyte number. Our data suggest a model in which BMP signaling regulates the acquisition of cardiac fate of separate pre-atrial and pre-ventricular pools, which behave similarly but differ in their sensitivity to BMP signaling. It is interesting to consider whether the role of BMP signaling in regulating chamber proportionality may be conserved in amniotes. Multiple studies have implicated BMP signaling at several developmental stages in distinct aspects of amniote heart development (van Wijk et al., 2007). During cardiac morphogenesis, it is clear that BMPs influence establishment of left-right asymmetry (Branford et al., 2000; Chocron et al., 2007; Mine et al., 2008; Smith et al., 2008), outflow tract morphogenesis (Délot et al., 2003; Liu et al., 2004; Stottmann et al., 2004), and valve formation (Délot et al., 2003; Liu et al., 2004; Solloway and Robertson, 1999; Sugi et al., 2004). At earlier stages, cardiac progenitors are clearly exposed to BMP signaling in both chick and mouse, as has been demonstrated in studies of the impact of BMPs on cardiac induction (Dunwoodie, 2007; Zaffran and Frasch, 2002). Although removal of BMP signaling has been shown to shrink or eliminate the mouse heart field (Klaus et al., 2007), it is not yet clear how BMPs impact chamber proportionality in mouse. More subtle manipulations of BMP signaling would be desirable for examination of this role. Analysis of chamber formation in amniotes is challenging due to the contributions of multiple myocardial lineages to the right ventricle and atria (Buckingham et al., 2005; Meilhac et al., 2004). Use of lineage-specific markers and tracers that distinguish between primary and secondary contributions to each chamber will help to determine the precise requirements for chamber identity acquisition in each amniote heart field. Such studies will be of great interest in the future, especially since the cell-autonomous impact of BMP signaling on chamber identity acquisition would be useful to harness for applications in regenerative medicine.
We thank J. Castro Lopes for support, H. Knaut and members of the Yelon laboratory for input, and M. Mullins and J. Tucker for fish, reagents, and advice. This work was supported by grants from the National Institutes of Health to DY. SRM received support from the GABBA Program and the Portuguese Foundation for Science and Technology (POCI 2010-FSE).
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.