Nodal signaling establishes cardiac L/R asymmetry by increasing myocardial migration rates on the left
We have previously shown that the sidedness of Nodal signaling directs the first asymmetry evident in the heart; a left, anterior-directed movement of atrial cells within the zebrafish “cardiac cone” 
, also see ). To gain insight into how Nodal signaling establishes the laterality of cell trajectories, we analyzed cell behaviors in time lapse images of cardiac development during conversion of the cardiac cone into the linear heart tube, and find that the left-biased asymmetry of cell movements in WT embryos is established by differences in cell migration velocities along the L/R axis. Cells on the left of the cardiac cone, which in WT embryos are exposed to the zebrafish Nodal southpaw
), migrate with an average rate of 9.2 nm/s (). Cells on the right side of the cone that do not exhibit Nodal target gene expression in WT embryos migrate with significantly slower rates of 5.9 nm/s (p<10−3
; ; Video S1
). Thus, exposure to Nodal signaling appears to induce increases in cell velocities within the cardiac cone. This L/R asymmetry in migration rates consistently leads to a left direction of cardiac jog by 24 hours post-fertilization (hpf) in WT embryos (3/3 embryos; ).
Trajectories and average velocities of migrating cells within the cardiac cone.
To confirm that Nodal signaling is responsible for the increase in migration rates in cells on the left of the WT cardiac cone, we analyzed no tail
) morpholino-injected embryos (morphants), which express spaw
bilaterally in the lateral plate mesoderm 
. We find that ntl
morphants exhibit a statistically significant increase in cell velocities compared with WT, with the left and right cells migrating an average of 10.9 nm/s and 10.2 nm/s, respectively (p<10−3
; ; Video S2
). This bilateral increase in migration rates leads to a loss of asymmetry in cell trajectories and subsequent loss of asymmetry in heart position at 24 hpf (3/3 embryos analyzed) (; and data not shown). We have further confirmed that this velocity increase is Nodal-dependent by analyzing migration rates in spaw
morphants and embryos treated with a Nodal-inhibitor drug, SB-505124, that blocks activity of the Nodal Type I receptors 
. In both conditions, migration rates are significantly reduced compared to WT (p<10−3
for each), but with no statistical difference in average cell velocities between spaw
morphant and drug-treated embryos (p
0.713) (; Video S3
). These migration rates are also consistent with what has been reported for late zygotic one-eyed pinhead
), which lack the essential Nodal co-receptor 
. Importantly, the loss of biased asymmetry in migration rates in spaw
mutants, and drug-treated embryos also results in predominant loss of biased asymmetry in cardiac jogging (see below) (), indicating that differences in cell velocity along the L/R axis of the cardiac cone are required for consistent establishment of asymmetry in cardiac jog.
The Bmp pathway directs random jogging asymmetry in the absence of Nodal signaling
While asymmetric spaw
is required to establish differences in cardiac velocities along the L/R axis, the heart can respond to additional laterality cues in the absence of Nodal signaling. Loss of either Spaw or the essential Nodal co-receptor Oep leads to randomized jogging (left, right or midline), not midline jogging, suggesting residual, randomized L/R signals function in the absence of Nodal pathway activation () 
. Evidence from the literature suggests that the Bmp pathway may provide this additional signal, as ubiquitous over-expression of Bmp ligands or global inhibition of the Bmp pathway both lead to alterations in cardiac laterality 
. If Bmp signals do provide the remaining asymmetric information in the absence of Nodal signaling, we hypothesized that combined inhibition of both the Nodal and Bmp pathways would remove all asymmetric information to the heart and lead to predominantly midline jogging. Bmp4 has been implicated as the Bmp ligand required during cardiac laterality determination 
, so we utilized the bmp4Y180
null mutant 
to analyze the jogging phenotypes of embryos with diminished Nodal and Bmp pathway activities.
Consistent with our hypothesis, loss of both Spaw and Bmp4 produces predominantly midline hearts by 24 hpf, with 89% of embryos lacking L/R asymmetry in cardiac jog (midline jog; ). Previous work has suggested that Bmp signals provide the dominant laterality cue to the heart and that Nodal signaling plays a secondary role, required only to ensure a consistent bias in Bmp pathway activity 
. However, in contrast to this existing view, we find that maternal-zygotic (MZ) MZbmp4Y180
mutants on their own do not exhibit significant jogging defects (). This result suggests that Bmp4 is only required to provide asymmetric cues in the absence of Nodal and is otherwise dispensable for cardiac laterality. Additionally, these data argue that Spaw, not Bmp4, provides the dominant laterality cue to the developing heart. The differences in our results from those previously published are possibly because those studies utilized overexpression of Bmp2b, producing non-physiological levels of Bmp signaling via a ligand that is not expressed in the cardiac field at this stage of development. Interestingly, we find that in the absence of Spaw, cardiac cells are highly sensitive to even small changes in the dosage of bmp4
, further supporting the idea that the levels of Bmp signaling can influence this process. While embryos containing only one copy of the bmp4Y180
mutation do not exhibit jogging defects (data not shown), when Nodal signaling is inhibited in these embryos, 76% display midline hearts at 24 hpf (). Taken together, our results provide strong evidence that Nodal signals dominantly influence cardiac laterality. Importantly, we find Bmp4 to only be critical in the absence of Spaw to direct jogging, at which point cardiac asymmetry is highly sensitive to the overall level of Bmp signaling present in the embryo.
Our data suggests that the hearts that jog directionally in Spaw morphants (left and right; ) are responding to asymmetric information provided by Bmp4. We predict this should result in asymmetries in cell migration velocities which produce the resulting directional jog. However, despite their randomized jogging phenotype, asymmetries in cell velocities across the L/R axis were not detected, and both the global population of spaw morphants and individual morphant embryos display bilaterally reduced cell velocities. Due to the profoundly diminished migration rates in spaw morphants, there is a subsequent delay in the conversion of the cardiac cone into the linear heart tube. Thus, it is likely that our time lapse movies were simply not long enough to observe an establishment of L/R asymmetry in cell velocities in the subset of spaw morphants with directional cardiac jog (3/7 embryos). In addition, even when directional jog is established in these morphants, the heart tubes are displaced from the midline much less significantly than are left-jogged hearts in WT embryos. Therefore, even when asymmetries are established in cell velocities along the L/R axis, we would anticipate those differences to be substantially less than the asymmetry observed in WT cardiac migration rates and, therefore, potentially below the threshold for significance used in our statistical analysis.
Bmp signaling negatively influences cardiac migration rates
To determine how Bmp signaling influences cell migration during jogging, we analyzed cardiac cell migration in bmp4Y180−/+
embryos injected with spaw
MO. We find that combined loss of Nodal signaling and just one functional copy of bmp4
leads to significantly increased average migration rates (6.08 nm/s on left and 5.82 nm/s on right) compared with loss of Nodal signaling alone (4.8 nm/s left and right; p<10−3
; ; Video S4
). This increase in cardiac migration rate is even more pronounced in embryos injected with bmp4
MOs, with left- and right-sided cells displaying average velocities of 7.8 nm/s and 7.4 nm/s, respectively (; Video S5
). As with all other conditions in which significant L/R asymmetries in migration rate are lost, nearly all embryos with combined inhibition of bmp4
also exhibit loss of directional cardiac jog (midline jog −3/5 bmp4Y180−/+
embryos injected with spaw
MO; 4/5 bmp4/spaw
double morphants). The increase in cardiac migration velocities upon inhibition of the Bmp pathway suggests that Bmp signaling is normally required to limit the migratory ability of cardiac cells.
Bmp signaling activity is asymmetrically increased on the left of the WT cardiac cone
The negative influence of Bmp signaling on cell migration rate, along with the significantly slower velocities of right-sided cardiac cells in WT embryos, suggests that the Bmp pathway normally acts on the right of the cone to influence cardiac laterality. However, previous work has established that bmp4
is expressed with a left
bias at 20 hpf and that the laterality of this expression is altered in embryos with defects in jogging asymmetry 
. Given the inconsistencies between earlier reports and our cell migration data, we were interested in determining the specific localization and potential asymmetry of Bmp activity within the heart. To this end, we analyzed Bmp pathway activation in WT embryos by immunofluorescence for the activated Bmp intracellular effectors Smads 1, 5 and 8 (phospho-Smad1/5/8 or p-Smad1/5/8). Consistent with previous reports 
, we find that Bmp pathway activity is asymmetrically increased on the left of the cardiac cone at 20 hpf, as indicated both by increased p-Smad1/5/8 fluorescence intensity in cells on the left compared with right (p<10−3
) and by greater numbers of p-Smad1/5/8 positive cells present on the left of the cardiac cone (p
Quantitation of Bmp pathway activity by average fluorescence intensity and number of p-Smad1/5/8 positive cells.
At first glance, this left-biased increase in pSmad1/5/8 in cells with faster velocities appears contradictory to data from our time lapse experiments which strongly support a role for the Bmp pathway in negatively regulating myocardial migration rates. Results from our time-lapse analyses, coupled with our genetic data demonstrating BMP signaling is dispensable for generating asymmetry in the heart in the presence of asymmetric Nodal signaling, suggest that regardless of the increase in p-Smad1/5/8 on the left, the cue from Nodal for cells to increase migration rates is stronger than the influence of Bmp signals on these same cells. By contrast, as cells on the right of the cone do not receive inductive cues from Spaw, Bmp activation on the right significantly diminishes migration rates. In WT embryos, this leads to cell velocities on the right of the cone being reduced compared to those on the left. Thus, when Nodal signaling is absent (as in spaw morphants and SB-505124-treated embryos), all cells in the cone respond to repressive cues from the Bmp pathway and both left and right myocardial cell velocities are substantially reduced. Likewise, when the Nodal pathway is activated on both sides of the cone (as in ntl morphants) cell velocities are increased to higher rates than those observed in WT cells exposed to Nodal, presumably because bilateral Nodal signaling diminishes the repressive effects of Bmp on myocardial migration rates.
Spaw negatively regulates Bmp responsiveness on the left of the cardiac cone
While we observe an increase in p-Smad1/5/8 on the left side of the cardiac cone, we argue that Nodal signaling increases migration on the left and overrides repressive cues from Bmp. Consistent with this hypothesis, analysis of Bmp activity in the hearts of spaw
morphants reveals a significant increase in the fluorescence intensity of p-Smad1/5/8 immunostaining in both right (p<10−3
) and left (p
0.002) cardiac fields compared with WT (). These results indicate that Nodal signaling limits the level of Bmp pathway activity, potentially by competing for the common intracellular effector Smad4, a mechanism of Nodal/Bmp antagonism that is known to occur during earlier stages of L/R patterning in zebrafish and other species 
. Interestingly, despite having higher intensities of p-Smad1/5/8 fluorescence, we find that the average number of p-Smad1/5/8 positive cells is significantly diminished in spaw
morphants compared to WT (p
0.02) (), which may indicate a role for Nodal signaling in both positive and negative regulation of Bmp pathway activation within the cardiac cone. These results suggest that Nodal signaling ensures the establishment of differential migration rates along the cardiac L/R axis in two ways; first, by directly increasing cell velocities on the left and second, by limiting the level of Bmp activity on the left. Ultimately, robust development of jogging asymmetry appears to require left-restricted activation of the Nodal pathway to increase migration rates and response to the Bmp pathway on the right of the cone, where Bmps serve to diminish migration velocities.
Bmp activation occurs in endocardial cells in the heart
In our immunofluorescence experiments, we noticed that the GFP staining in myocardial cells did not significantly colocalize with the p-Smad1/5/8 present in the heart field (). As endocardial cells are also localized to the midline by this stage of development, we hypothesized that Bmps signal more predominantly to the endocardium at 20 hpf. To address this possibility, we performed p-Smad1/5/8 staining in embryos with GFP expressed from the kdrl
promoter, which labels endothelial and endocardial cells 
. We observed significant colocalization of GFP and p-Smad1/5/8 in these embryos, indicating that Bmp activity is primarily upregulated within endocardial cells (). By contrast, all direct Nodal targets that have been identified in the heart are expressed within the myocardial population 
. Thus, the Nodal and Bmp pathways appear to act in parallel to establish asymmetries in cell migration velocities within the cardiac cone: Nodal pathway activation in the myocardium on the left leads to increases in migration rates while Bmp signaling in the endocardium limits myocardial migration, primarily on the right. Precedence for cross-regulation between the endocardium and myocardium during the earlier migration events leading to cone formation have been described 
. Thus, interplay between these two tissues is critical for at least two migrations during cardiac development.
FoxH1 is required for responsiveness of cardiac cells to both Nodal and Bmp signals
While loss of the ligand Spaw or the co-receptor Oep results in randomized jogging (), embryos with a nonsense mutation in the Nodal transcription factor FoxH1 
display 78% midline hearts (). These results suggest that FoxH1 performs both Nodal-dependent and independent functions within the heart. Interestingly, midway
jogging defects are strikingly similar to those of embryos lacking both Nodal and Bmp signaling () suggesting that FoxH1 may be required for cardiac cell responsiveness to both TGFβ pathways. To address this possibility, we analyzed cardiac cell migration in midway/foxH1
mutants. Cells on the left and right of the cardiac cone in midway
mutants migrate with average velocities of 7.8 nm/s and 7.1 nm/s, respectively (; Video S6
). These migration rates are significantly faster than those of cardiac cells in embryos lacking Spaw (p<10−3
) or treated with the Nodal-inhibitor drug SB-505124 (p<10−3
), confirming a Nodal-independent function for FoxH1 in establishing cardiac laterality (). Importantly, midway
cardiac velocities are not statistically different than those of bmp4/spaw
double morphants (p
0.580), suggesting that FoxH1 is necessary for cardiac cells to respond to both Nodal and Bmp signals. Consistent with this idea, we observe a significant, bilateral decrease in p-Smad1/5/8 in the hearts of midway
mutants compared with WT, both in fluorescence intensity (p<10−3
) and in the number of p-Smad1/5/8 positive cells (p<10−3
) (). Interestingly, midway
mutants exhibit a more profound decrease in Bmp pathway activation than that observed in bmp4/spaw
double morphants in both fluorescence intensity and number of p-Smad1/5/8 positive cells (p<10−3
), indicating that loss of FoxH1 activity profoundly blocks cardiac cell responsiveness to Bmp signals ().
A model for the generation of asymmetry in the zebrafish heart
Here, we report the parallel requirements for Nodal and Bmp pathways in establishing cardiac laterality through opposing influences on cardiac cell migration rates (). Additionally, we have uncovered a novel, Nodal-independent role for FoxH1 during this process in mediating cardiac cell responsiveness to Bmp signals (). The Nodal-dependent activity of FoxH1 in regulating cardiac cell migration is likely restricted to myocardial cells, as the Nodal targets lefty1, lefty2
all display myocardial expression 
, the latter of which has been implicated in previous work 
to influence myocardial migration. However, it is unclear at present whether the Nodal-independent function of FoxH1 is carried out within the myocardial or endocardial cells of the developing heart. FoxH1 can act as both transcriptional activator and repressor, and an inhibitory requirement for FoxH1 has been reported in the vasculature to limit expression of the VegF receptor, kdrl
. Consequently, FoxH1 may be required directly in the endocardium to block expression of Bmp antagonists or activate transcription of required components of the Bmp pathway. Alternatively, the influence of FoxH1 on Bmp activation may be more indirect. Recent work has reported that retinoic acid (RA) signaling is necessary for consistent asymmetry in bmp4
expression in the zebrafish cardiac cone, with inhibition of RA leading to bilateral bmp4
expression and an increase in midline jogging 
. Interestingly, a genome-wide screen for FoxH1 binding sites revealed aldh1a1
, a gene necessary for RA production, as a direct target of FoxH1 in mouse 
. While zebrafish lack an ortholog of this specific RA processing enzyme, there are a number of other aldh1a family members expressed throughout zebrafish development 
, making it compelling to speculate that the loss of Bmp responsiveness in midway
cardiac cells may be due to defects in RA signaling.
Model of the opposing roles for Nodal and Bmp signaling during asymmetric cardiac morphogenesis.
Our results support a model in which cardiac laterality is regulated by interactions and cross-regulations both between TGFβ pathways and between the myocardial and endocardial layers of the developing heart that regulate differential motility along the L/R axis. These interactions involve complex integrations between Nodal and Bmp pathways, and we demonstrate that cardiac cells are highly sensitive to the dosage of these TGFβ signals. Bilateral exposure to Spaw increases migration rates beyond what is observed in left cells of the WT cone, and loss of a single copy of bmp4 in addition to Nodal signaling significantly alters both jogging laterality and cardiac cell velocities. Moreover, the signals that can influence laterality in the heart likely involve additional members of the TGFβ family. We note that inhibition of Nodal signaling with the SB-505124 drug decreases cell velocities as expected. However, jogging laterality in these embryos is predominantly midline, which differs from loss of Spaw or Oep (). While this phenotype resembles that of embryos lacking Spaw and Bmp4 (), the cell migration rates in drug-treated embryos are consistent with loss of Nodal, but not Bmp signaling () and, indeed, we find that the Bmp pathway is still activated within the heart field upon SB-505124 treatment (data not shown). The spaw morpholino completely abolishes expression of spaw in the LPM, strongly suggesting that Spaw is absent in the hearts of these embryos. This, coupled with the similar phenotypes of spaw knockdown and LZoep mutants, suggests that the effect of the drug is not a result of more complete knockdown of Nodal signaling. SB-505124 acts intracellularly on the Alk 4/5/7 Type I receptors, which are utilized by both Nodal and TGFβ ligands. Overall, this suggests that another TGFβ molecule signaling through the Nodal receptors can affect the migration of cardiac cells and may be important for allowing the cardiac cells to respond to fluctuations in Bmp levels when Spaw is absent. Taken together, these results have implications for determining the underlying genetic lesions in CHD, as they suggest that heterozygous mutations in components of different TGFβ signaling pathways may synergize to produce severe phenotypes. Further analysis of integrations of signals within and between cardiac cells will provide insight into the general mechanisms driving asymmetric morphogenesis and will greatly enhance our understanding of the potentially complicated genetic interactions underlying the development of CHD in humans.