The MEK-ERK signaling pathway promotes maintenance of cardiac chamber identity

The vertebrate heart contains two types of cardiac chambers with distinct functions: atrial chambers collect and supply blood to the ventricles, whereas ventricular chambers forcefully propel blood to the periphery. The morphological, physiological and molecular differences between the ventricular and atrial myocardium underlie the functional characteristics of each chamber: for example, the trabecular architecture of ventricular tissue differs from the thinner atrial walls, characteristic action potentials and calcium dynamics distinguish ventricular and atrial conductive properties, and each chamber expresses sets of genes encoding specific myosin isoforms and ion channels (Asp et al., 2012; Cao et al., 2023; DeLaughter et al., 2016; Li et al., 2016; Moorman and Christoffels, 2003; Singh et al., 2016; van Weerd and Christoffels, 2016; Wessels and Sedmera, 2004). Although the operational importance of chamber-specific traits is well established, we do not yet fully understand how ventricular and atrial cardiomyocytes acquire and maintain their characteristic features.

Fate mapping studies in chick, zebrafish and mouse have suggested that chamber fate specification occurs in the early embryo, where spatial segregation of ventricular and atrial myocardial lineages is evident before or during gastrulation (Bardot et al., 2017; Garcia-Martinez and Schoenwolf, 1993; Ivanovitch et al., 2021; Keegan et al., 2004; Rosenquist, 1970; Stainier et al., 1993). During these stages, the signaling pathways that pattern the embryonic axes, including Nodal, FGF and BMP signaling, also appear to influence patterning of the cardiac progenitors (Keegan et al., 2004; Marques and Yelon, 2009; Marques et al., 2008; Reiter et al., 2001). Despite early specification of chamber progenitors, the chamber identities of differentiated cardiomyocytes require active maintenance as development proceeds (Martin and Waxman, 2021; Yao et al., 2021). For example, several transcription factors have been shown to be required for maintenance of chamber-specific gene expression patterns. In mouse, COUP-TFII (also known as NR2F2) and TBX5 are essential for atrial identity maintenance: both factors sustain the expression of atrial genes and repress the expression of ventricular genes in differentiated atrial cardiomyocytes (Sweat et al., 2023; Wu et al., 2013). Likewise, in zebrafish, nr2f1a, a functional homolog of mammalian COUP-TFII, prevents atrial cardiomyocytes from developing ectopic ventricular and pacemaker cell identities (Martin et al., 2023). Conversely, studies in mouse and chick indicate crucial roles of Hey2 and Irx4 in ventricular identity maintenance, as they repress expression of atrial genes in the ventricular myocardium (Bao et al., 1999; Bruneau et al., 2001; Koibuchi and Chin, 2007; Xin et al., 2007). Although these transcription factors clearly promote maintenance of chamber identities, less is known about the regulatory pathways that control their chamber-specific expression.

It is therefore intriguing that previous studies have indicated a role for FGF signaling in promoting the expression of Nkx transcription factors, which serve as upstream regulators of Hey2 and Irx4. In zebrafish, mouse and human embryonic stem cell models, Nkx2-5 homologs are required to drive expression of Hey2 and Irx4 homologs in the ventricle (Anderson et al., 2018; Bruneau et al., 2000; Targoff et al., 2013). Furthermore, in zebrafish, loss of function of the Nkx factors nkx2.5 and nkx2.7 disrupts ventricular identity maintenance, causing the acquisition of expression of the atrial gene atrial myosin heavy chain (amhc; also known as myh6) in the ventricle, coupled with the loss of expression of the ventricular gene ventricular myosin heavy chain (vmhc; also known as myh7) (Targoff et al., 2013). Similarly, inhibition of FGF signaling after the initiation of myocardial differentiation results in ectopic expression of amhc and reduced expression of vmhc in the ventricle, as well as reduced expression of nkx2.5 and nkx2.7 (Pradhan et al., 2017). As overexpression of nkx2.5 can partially rescue the ventricular defects caused by inhibition of FGF signaling (Pradhan et al., 2017), the synthesis of these data demonstrates that FGF signaling plays an important role, upstream of Nkx factors, in maintaining ventricular chamber identity. However, it remains unclear how FGF signaling regulates the expression of Nkx factors and whether additional genes also mediate the role of FGF signaling in reinforcement of ventricular traits.

In many developmental processes, FGF signaling is known to execute its functions by activating the MEK1/2-ERK1/2 pathway (Brewer et al., 2016; Dailey et al., 2005). For example, FGF4 functions through the ERK1/2 pathway to activate primitive endoderm specification within the inner cell mass of the mouse embryo (Kang et al., 2013; Krawchuk et al., 2013; Nichols et al., 2009; Soszyńska et al., 2019), where live visualization of ERK activity has revealed distinct ERK signaling dynamics in different lineages (Simon et al., 2020). As another example, studies in multiple vertebrate models have indicated a role of ERK1/2 downstream of FGF signaling in somite segmentation (Akiyama et al., 2014; Delfini et al., 2005; Niwa et al., 2011). In contrast, FGF signaling in other developmental contexts is mediated by different signal transduction mechanisms including the PI3K-AKT/PKB and PLCγ-PKC pathways, as well as pathways using MAPK proteins other than ERK1/2 (Brewer et al., 2016; Dailey et al., 2005). For example, in Xenopus, FGF signaling acts via p38 MAPK to regulate the initial expression of nkx2.5 in cardiac progenitors (Keren-Politansky et al., 2009). FGF signaling has also been shown to function through PI3K pathways in the development of GnRH-secreting neurons and through JNK pathways during bile acid biosynthesis (Brewer et al., 2016; Holt et al., 2003; Hu et al., 2013). It therefore seems likely that one of these established downstream signal transduction pathways facilitates the role of FGF signaling in the maintenance of ventricular identity.

Here, we modulate the MEK1/2-ERK1/2 signaling pathway, using both pharmacological and genetic approaches, and evaluate the impact on chamber identity maintenance. We find that inhibition of MEK-ERK signaling following myocardial differentiation results in ectopic amhc expression as well as reduced vmhc expression in ventricular cells, similar to the effects of inhibition of FGF signaling. Additionally, MEK-ERK signaling and FGF signaling appear to be required for ventricular identity maintenance during similar timeframes, correlating with stages when phosphorylated ERK (pERK) is detectable in the myocardium. Epistasis analysis indicates that a constitutively active form of MEK1 can partially rescue the effects of FGF pathway inhibition and that overexpression of nkx2.5 can partially rescue the effects of MEK pathway inhibition. Together, these data support a model in which MEK-ERK signaling acts downstream of FGF signaling and upstream of Nkx factors to promote ventricular identity maintenance. Our results also indicate that the role of this FGF-MEK-ERK-Nkx pathway is context-dependent: some regions of the ventricle are more sensitive than other regions to inhibition of FGF or MEK signaling. Additionally, in the atrium, although endogenous levels of MEK-ERK signaling are not sufficient to induce ventricular traits, heightened MEK signaling can cause ectopic ventricular gene expression. Thus, our studies indicate distinct functions of MEK-ERK signaling in ventricular and atrial identity maintenance.

To investigate the pathways that might mediate FGF signal transduction during ventricular identity maintenance, we began by evaluating the effects of a set of kinase inhibitors on gene expression patterns in the ventricle. In these experiments, we used the expression of amhc and vmhc as proxies for atrial and ventricular identities, with the understanding that these two genes reflect only a small part of the distinctions between the chambers (Yao et al., 2021). Although amhc and vmhc definitely do not represent all aspects of atrial and ventricular identity, alteration of the amhc or vmhc expression patterns certainly can indicate a failure to maintain chamber identity properly.

We found that inhibition of PI3K or JNK, using LY294002 or SP600125, failed to induce any ectopic amhc in the ventricle (Fig. S1). In contrast, treatment with PD0325901, an inhibitor of MEK1/2 activity that prevents the phosphorylation of ERK1/2 (Sebolt-Leopold, 2008), induced both ectopic amhc transcripts (Fig. 1B) and Amhc protein (Fig. 1E) in the ventricle. Moreover, the pattern of ectopic atrial marker localization in PD0325901-treated embryos was highly reminiscent of the phenotype that we had previously observed in embryos treated with the FGFR inhibitor SU5402 (Fig. 1C,F) (Pradhan et al., 2017). Additionally, we found that MEK pathway inhibition reduced the extent and intensity of vmhc expression in the ventricle (Fig. 1G,H), with complementary patterns of ectopic amhc and decreased vmhc expression (Fig. S2), similar to the effects of FGF pathway inhibition with SU5402 or with expression of a dominant-negative form of FGFR1 (Fig. 1I; Pradhan et al., 2017). Together, these data suggest that MEK-ERK signaling, like FGF signaling, acts both to prevent atrial gene expression and to promote ventricular gene expression within the ventricle.

Extending our comparison of the influences of MEK and FGF signaling on the ventricular myocardium, we examined whether these pathways act during the same time interval. Treatment with PD0325901 or SU5402 starting at 18 h postfertilization (hpf), when ventricular and atrial myocardial differentiation are already underway (Berdougo et al., 2003; Yelon et al., 1999), effectively induced ectopic amhc expression in the ventricle (Fig. 1A-F). However, treatment at later time points, such as 26 hpf, was less potent, causing ectopic amhc expression only in a few cardiomyocytes in or near the outflow tract (OFT) (Fig. 1J-L). Thus, our data suggest that MEK and FGF signaling are both required during a comparable timeframe to promote ventricular identity maintenance.

Not all of the regions in the ventricle appeared to be equally sensitive to the inhibition of MEK or FGF signaling. Typically, we noted that cells in the ventricular inner curvature (IC) and OFT more readily expressed ectopic amhc, whereas amhc expression was less frequently observed in the ventricular outer curvature (OC) (Fig. 1B,C,E,F). This general trend was consistent with our previous observations of the effects of FGF pathway inhibition (Pradhan et al., 2017). Investigating this differential sensitivity more systematically, we found that low concentrations of SU5402 or PD0325901 frequently induced ectopic amhc expression in the OFT and in the ventricular apex (Fig. 2A,B,F,G). Higher concentrations led to the additional induction of ectopic amhc in the IC (Fig. 2C,H) and the central portion of the ventricle (Fig. 2D,I). With the highest concentrations tested, ectopic amhc was also observed in the OC (Fig. 2E,J), such that all regions of the ventricle were expressing amhc. Correspondingly, we observed greater loss of vmhc expression when treating embryos with higher concentrations of SU5402 or PD0325901 (Fig. 2M-V).

Altogether, for a given concentration of either SU5402 or PD0325901, more embryos expressed ectopic amhc in the more sensitive regions, such as the OFT, whereas fewer embryos expressed amhc in the less sensitive regions, such as the OC (Fig. 2L). Additionally, higher concentrations of either inhibitor generally resulted in more embryos expressing amhc in any ventricular region (Fig. 2L). These dose-dependent responses suggest differential degrees of plasticity for ventricular identity in different portions of the ventricle. Moreover, the phenotypic similarities between SU5402-treated and PD0325901-treated ventricles suggest that FGF and MEK signaling could act in the same pathway to regulate chamber-specific characteristics.

Our previous work has suggested that FGF signaling is required cell-autonomously within ventricular cardiomyocytes to maintain their identity (Pradhan et al., 2017). Ectopic amhc expression in the fgf8a mutant ventricle suggests that Fgf8a is one of the crucial signals involved (Pradhan et al., 2017), and fgf8a is expressed in the myocardium during the 18-26 hpf time interval that appears to be essential for ventricular identity maintenance (Reifers et al., 2000). We therefore hypothesized that FGF signaling activates MEK-ERK signaling within the ventricular myocardium during these stages. In order to visualize MEK-ERK signaling activity, we examined the localization of pERK in the cardiac cone at 20 hpf and in the heart tube at 24 hpf.

Heart tube assembly begins with the fusion of bilateral populations of myocardial precursors to form the cardiac cone at the embryonic midline, with ventricular cardiomyocytes at the center of this shallow cone and atrial cardiomyocytes at the periphery (Berdougo et al., 2003; Staudt and Stainier, 2012). In wild-type embryos, we detected pERK localization in multiple cells within the myocardial cone, as well as within its endocardial core (Fig. 3A,B). This pERK signal was dramatically diminished in PD0325901-treated embryos (Fig. S4), validating the specificity of our anti-pERK immunofluorescence protocol. A similar reduction of myocardial pERK signal was observed in SU5402-treated embryos (Fig. S5). The number and location of pERK⁺ cardiomyocytes varied between wild-type embryos, as did the intensity of the pERK signal (Fig. 3B,E; Fig. S6). Qualitative analysis did not identify any consistent regional patterns of pERK localization within the myocardium: there was no consistently detectable enrichment of pERK levels in the more central ventricular cells or in the more peripheral atrial cells, nor was there any consistent enhancement of pERK signal on the left or right sides or in the anterior or posterior portions of the cardiac cone (Fig. S6). Altogether, the localization of pERK appeared to be unique in each embryo, and we could not correlate any pattern of pERK distribution with the differential sensitivity of distinct ventricular regions. Given the broad variability of pERK distribution, we suspect that MEK-ERK signaling is highly dynamic within the myocardium of the cardiac cone.

We continued to observe pERK in the myocardium after heart tube assembly was complete (Fig. 3C-E). Although myocardial pERK was present in all embryos examined at 24 hpf, endocardial pERK was less frequently observed at this stage (only in 6/14 embryos). Additionally, although we did detect pERK in both the ventricular and atrial portions of the heart tube, we typically observed enriched pERK localization in the atrial myocardium (Fig. 3C,D). Thus, ERK signaling appears to be more broadly distributed in the cardiac cone than in the heart tube, with ERK signaling in the ventricular myocardium more prevalent in the cone than in the tube, correlating with the timeframe when FGF and MEK signaling are required for ventricular maintenance. At the same time, the presence of pERK in the atrial myocardium is evidently not sufficient to induce ventricular traits.

To test whether MEK signaling acts downstream of FGF signaling during ventricular identity maintenance, we next examined whether constitutive MEK1 activity could rescue the effects of FGF pathway inhibition. For this purpose, we generated stable lines carrying the transgene Tg(hsp:MEK1^S219D-mCherry) [hereafter referred to as Tg(hsp:caMEK1)], which expresses a constitutively active form of MEK1 (caMEK1) in response to heat shock. Ser219 is predicted to be a key phosphorylation site during activation of zebrafish MEK1, and the substitution of Ser to Asp to mimic phosphorylation has been shown to increase ERK phosphorylation (Bolcome and Chan, 2010; Chou et al., 2015). Correspondingly, we found that heat-induced expression of Tg(hsp:caMEK1) significantly increased the number of pERK⁺ cardiomyocytes and the levels of pERK signal in the myocardium, although the degree of increase varied (Fig. 4).

We evaluated whether expression of Tg(hsp:caMEK1) would prevent the appearance of ectopic amhc in the ventricular myocardium of SU5402-treated embryos. Indeed, following heat shock and SU5402 treatment, Tg(hsp:caMEK1) embryos displayed less ectopic amhc than their wild-type siblings (Fig. 5), suggesting that MEK-ERK signaling functions downstream of FGF signaling in the context of ventricular maintenance. When interpreting these results, we considered that SU5402 has also been shown to cause mild inhibition of VEGF signaling (Molina et al., 2007) and that the MEK-ERK pathway can also be activated by VEGF (Shin et al., 2016). However, treatment with a VEGFR inhibitor, SU4312, did not affect ventricular identity maintenance (Fig. S7), suggesting that our epistasis analysis reflects the importance of an FGF-MEK-ERK signaling pathway in repressing ectopic amhc expression in the ventricle.

We noted that expression of Tg(hsp:caMEK1) was most potent in rescuing the less plastic ventricular regions, including the OC and central ventricle (central V), of SU5402-treated embryos, whereas very little or no rescue was observed in the more plastic regions, such as the OFT and the ventricular apex (Fig. 5E). This pattern of partial rescue suggests the possibility that lower levels of MEK-ERK signaling are required for identity maintenance in the less plastic regions of the ventricle, whereas higher levels of MEK-ERK signaling are required in the more plastic regions.

When considering effector genes that could function downstream of the FGF-MEK-ERK pathway to promote ventricular maintenance, we kept in mind our previous finding that FGF signaling acts upstream of Nkx factors during this process (Pradhan et al., 2017). Similar to the effects of FGF pathway inhibition (Pradhan et al., 2017), MEK pathway inhibition resulted in reduced expression of both nkx2.5 and nkx2.7 (Fig. 6A-D), suggesting that MEK-ERK signaling might act in between FGF signaling and Nkx factors in ventricular identity maintenance. To test this, we examined whether overexpression of nkx2.5 could prevent ectopic amhc expression in PD0325901-treated embryos. Following heat shock and PD0325901 treatment, we found that expression of Tg(hsp:nkx2.5) led to less ectopic amhc than was seen in wild-type siblings (Fig. 6E-I). Additionally, overexpression of nkx2.5 achieved greater rescue in the less plastic regions of the ventricle, such as the OC and the central V, whereas the observed rescue in the OFT and the ventricular apex was negligible (Fig. 6I). Together with our other results, these data suggest a model in which different regions of the ventricle have differential degrees of requirement for activation of an FGF-MEK-ERK-Nkx pathway in order to promote ventricular identity maintenance.

We wondered whether the differential regional sensitivity within the ventricle could reflect differences in mechanical conditioning caused by the forces associated with cardiac function. For example, perhaps the OC is more resistant than the IC to the effects of inhibiting the FGF-MEK-ERK pathway because these two curvatures are exposed to different biomechanical cues as they develop. To test this hypothesis, we examined the effects of PD0325901 on tnnt2a mutants, which lack myocardial contractility and blood flow (Sehnert et al., 2002). Comparing the patterns of ectopic amhc expression, we found that the tnnt2a mutant ventricle exhibited the same regional tendencies that were observed in the wild-type ventricle (Fig. S9). For example, the OC of the tnnt2a mutant ventricle appeared to be just as resistant to MEK inhibition as the wild-type OC, and the tnnt2a mutant IC appeared to be generally as sensitive to MEK inhibition as the wild-type IC (Fig. S9E). Thus, it appears that biomechanical forces do not underlie the regional vulnerability to the inhibition of MEK-ERK signaling in the ventricle, and further studies will be necessary to identify the basis for regional differences in the plasticity of ventricular identity.

While evaluating the impact of Tg(hsp:caMEK1) in the ventricle, we were intrigued to find that constitutive MEK1 activity also led to an interesting atrial phenotype. Specifically, we observed ectopic vmhc expression in the atrium of Tg(hsp:caMEK1) embryos (Fig. 7A-D). Although amhc expression appeared to be normal (Fig. 7A,B), individual vmhc-expressing cells, or small clusters of such cells, were scattered throughout the atrium, without any discernible regional pattern (Fig. 7C,D): ectopic vmhc expression was seen near the atrioventricular canal, near the inflow tract or in the middle of the atrium, and each embryo exhibited a unique distribution of atrial vmhc. Ectopic vmhc expression emerged by 30 hpf, when multiple atrial vmhc-expressing cells were present (Fig. 7E,F) and appeared to be scattered in a manner similar to that seen at 48 hpf (Fig. 7D). Expression of Tg(hsp:caMEK1) robustly induced ectopic vmhc expression in the majority of embryos when heat shock was initiated between 15 and 18 hpf (Fig. S10). In contrast, the severity and penetrance of the phenotype were substantially reduced when heat shock was initiated before 15 hpf or after 18 hpf (Fig. S10), demonstrating a discrete time window when constitutive MEK1 activity influences atrial expression of vmhc. Ectopic vmhc expression could also be induced by a second, independently derived, allele of Tg(hsp:caMEK1), indicating that this phenotype is caused by constitutive MEK1 activity and not by positional effects of the transgene (Fig. S11). Taken together, these data suggest that high levels of MEK-ERK signaling can disrupt the maintenance of atrial chamber characteristics.

These results surprised us, as the induction of ectopic vmhc expression by constitutive MEK1 activity contrasts with our previous observation that increased FGF signaling, driven by expression of Tg(hsp:cafgfr1), did not induce ectopic vmhc expression in the atrium (Pradhan et al., 2017). Revisiting and extending those experiments, we again failed to find ectopic expression of vmhc in Tg(hsp:cafgfr1) embryos after heat shock at 14, 16 or 18 hpf (Fig. S12). We wondered whether Tg(hsp:caMEK1) might be more effective than Tg(hsp:cafgfr1) at increasing MEK-ERK signaling, but we did not find evidence for this in our examination of pERK localization. Expression of Tg(hsp:cafgfr1) robustly increased the number of pERK⁺ cardiomyocytes and heightened the intensity of the pERK signal (Figs S13, S14), without any apparent deficiency compared with the effects of Tg(hsp:caMEK1) (Fig. 4; Figs S13, S14). However, we note that our analysis of pERK levels at a single timepoint would not reveal differences in the duration or dynamics of ERK activity, and we recognize that the kinetics of ERK activity are affected by multiple negative feedback mechanisms (Kim and Bar-Sagi, 2004; Lake et al., 2016; Szybowska et al., 2021; Thisse and Thisse, 2005). We expect that increased ERK signaling in response to Tg(hsp:cafgfr1) would be subject to negative feedback regulation by endogenous pathways that modulate MEK and ERK activity, including the FGFR endocytosis machinery and negative regulatory proteins such as Spry and Spred (Kim and Bar-Sagi, 2004; Lake et al., 2016; Szybowska et al., 2021; Thisse and Thisse, 2005), whereas ERK signaling in response to Tg(hsp:caMEK1) would not be attenuated by these factors. This presumed difference in susceptibility to negative feedback pathways could potentially account for the distinct impacts of Tg(hsp:caMEK1) and Tg(hsp:cafgfr1).

We hypothesized that the ectopic vmhc expression observed in Tg(hsp:caMEK1) embryos represented atrial cardiomyocytes that were aberrantly expressing vmhc, rather than ventricular cardiomyocytes that had migrated into the atrium. Consistent with this idea, we found that the ectopic vmhc expression colocalized with a myocardial marker and that these vmhc-expressing cells resided within the myocardial wall of the atrial chamber (Fig. 8A-I). Furthermore, we did not find evident loss of amhc expression in Tg(hsp:caMEK1) embryos (Fig. 7A,B), and the ectopic vmhc-expressing cells appeared to maintain normal levels of Amhc protein (Fig. 8J-R). These data, taken together with the scattered location of the ectopic vmhc-expressing cells, suggested that these cells were atrial cardiomyocytes in the process of acquiring ventricular characteristics. Thus, although endogenous levels of pERK are not sufficient to induce vmhc expression in the atrium, it appears that heightened levels of MEK-ERK signaling can override the mechanisms that normally enforce the maintenance of atrial chamber identity.

Our data indicate a previously unappreciated role for MEK-ERK signaling in enforcing the maintenance of ventricular chamber identity: MEK-ERK signaling is required after the onset of myocardial differentiation to repress ectopic induction of atrial genes and promote expression of ventricular genes in the developing ventricle, acting downstream of FGF signaling and upstream of Nkx factors. Our studies have also shown that the requirement for FGF-MEK-ERK signaling during chamber identity maintenance is context-dependent. Different portions of the ventricle exhibit differential degrees of sensitivity to inhibition of the FGF or MEK pathways, suggesting a regionalized pattern for ventricular plasticity. Meanwhile, in the atrium, we have uncovered a surprising ability of excessive MEK-ERK signaling to induce ventricular gene expression in atrial cardiomyocytes, even though endogenous levels of MEK-ERK signaling do not normally disrupt atrial identity. Taken together, our findings demonstrate distinct roles of MEK-ERK signaling in ventricular and atrial chamber identity maintenance.

Going forward, it will be important to identify additional components of the FGF-MEK-ERK-Nkx pathway that promotes ventricular identity maintenance. It is not yet clear how directly ERK signaling impacts nkx2.5 and nkx2.7 expression or which effector genes might mediate this interaction. ETS transcription factors are interesting candidates for relevant effectors downstream of ERK signaling, as FGF signaling has been suggested to act through the Pea3 family of ETS factors in order to regulate the formation of nkx2.5-expressing cardiac progenitors in the early zebrafish embryo (Znosko et al., 2010). In addition to the identification of ERK-dependent regulators of nkx gene expression, it will be valuable to investigate whether there are other effector genes that act downstream of FGF-MEK-ERK signaling, in parallel with nkx genes and potentially including hey2 and irx4a, to regulate the maintenance of chamber-specific patterns of gene expression.

Analysis of additional players in the FGF-MEK-ERK-Nkx pathway should also be coupled with further investigation of the dynamics of MEK-ERK signaling within the myocardium. The observed myocardial localization of pERK is consistent with a cell-autonomous role for MEK-ERK signaling in the maintenance of ventricular cardiomyocyte characteristics, but the variable distribution and intensity of the pERK signal raise interesting questions about the duration and fluctuation of myocardial MEK-ERK signaling. In future studies, it will be beneficial to perform live imaging using ERK activity reporter transgenes, such as ERK-KTR sensors (De Simone et al., 2021; Qi et al., 2022; Simon et al., 2020), in order to evaluate precisely how myocardial MEK-ERK signaling is sustained or oscillates over time. These specific dynamics of MEK-ERK signaling may reveal regional patterns of signal intensity that were not apparent in our individual snapshots of pERK localization, and these patterns may be relevant to the differential plasticity of particular ventricular regions.

Furthermore, the regional differences in sensitivity to FGF-MEK-ERK pathway inhibition may reflect inherent distinctions between the differentiation states of specific portions of the ventricle. For example, as OFT cardiomyocytes initiate their differentiation at a later stage than other groups of ventricular cardiomyocytes (de Pater et al., 2009; Hami et al., 2011; Lazic and Scott, 2011; Zhou et al., 2011), the late-differentiating OFT cells could be more vulnerable to modulation of the FGF-MEK-ERK pathway than the early-differentiating cells that are more advanced along their differentiation trajectory. This notion is consistent with our observation that inhibition of FGF or MEK signaling at 26 hpf still induces amhc expression in OFT cells without significantly impacting the OC, central V or IC (Fig. 1K,L). A similar logic could apply to the relative sensitivities of the OC and IC, as OC cardiomyocytes are thought to be more advanced in their differentiation than the relatively primitive cardiomyocytes in the IC (Auman et al., 2007; Christoffels et al., 2000; Habets et al., 2002; Jensen et al., 2013; Moorman and Christoffels, 2003) and may therefore require lower levels of FGF-MEK-ERK signaling to maintain their ventricular identity. In the future, it will be interesting to identify the differences in gene expression patterns or signaling pathway activities that distinguish portions of the ventricle and provide the molecular basis for regional differences in ventricular plasticity.

It is also intriguing to contemplate why the endogenous levels of pERK found in the atrial myocardium do not normally interfere with atrial identity maintenance, whereas heightened MEK signaling can induce ectopic expression of vmhc in the atrium. Perhaps there are atrium-specific repressive pathways that usually inhibit the effects of endogenous MEK-ERK signaling on chamber identity, such that only excessive levels of ERK signaling can override these repressive influences. Components of pathways that are known to regulate the maintenance of atrial identity, including the COUP-TFII, Nr2f1a or TBX5 pathways (Martin et al., 2023; Sweat et al., 2023; Wu et al., 2013), might serve as such repressors, and it will be worthwhile to evaluate this possibility in future studies. We note that the sparse expression of vmhc in the atrium of Tg(hsp:caMEK1) embryos contrasts with the broader expression of amhc in the ventricle of PD0325901-treated embryos. As we did not observe a universal increase of pERK intensity in all of the cardiomyocytes in Tg(hsp:caMEK1) embryos, we speculate that only the atrial cells with the most sustained or highest cumulative ERK activity over time went on to initiate vmhc expression in Tg(hsp:caMEK1) embryos. This idea appears to be consistent with the relatively small number of ectopic vmhc-expressing cells observed and the lack of a regional pattern for their location in the atrium. Such particularly high levels of ERK activity may even induce vmhc expression in the atrium through downstream effectors distinct from those that act in the ventricle; along these lines, we note that the overexpression of nkx2.5 in Tg(hsp:nkx2.5) embryos appears to be insufficient to generate ectopic vmhc in the atrium (George et al., 2015). Further experimentation will be needed to identify the chamber-specific signaling dynamics and molecular context that distinguish the role of MEK-ERK signaling in the atrium from its role in the ventricle. In addition, it will be interesting for future studies to evaluate whether the observed pERK in the atrium reflects a different, separable role of MEK-ERK signaling during atrial development.

Over the long term, we envision that strategies for cardiac tissue engineering will benefit from an enhanced understanding of how signaling pathways maintain cardiac chamber identities. There is ongoing interest in cardiac disease modeling and personalized drug discovery using engineered tissue derived from pluripotent cells (Huang et al., 2021; Lewis-Israeli et al., 2021; Parrotta et al., 2020; Protze et al., 2019). It is therefore crucial to develop robust approaches for producing populations of ventricular and atrial cardiomyocytes that will stably maintain their chamber-specific characteristics (Ng et al., 2010; Zhao et al., 2020; Zhuang et al., 2022). Whereas existing protocols are able to direct the initial differentiation of ventricular and atrial cells (Cyganek et al., 2018; Goldfracht et al., 2020; Lee et al., 2017; Zhao et al., 2019), the ongoing investigation of the mechanisms underlying chamber identity maintenance, including the roles of MEK-ERK signaling, will shape future approaches for continually enforcing the specific traits of ventricular and atrial tissue.

We used the following zebrafish strains: Tg(myl7:EGFP)^twu34 (Huang et al., 2003), Tg(kdrl:Hsa.HRAS-mCherry)^s896 (Chi et al., 2008), Tg(hsp70l:nkx2.5-EGFP)^fcu1 (George et al., 2015), tnnt2a^b109 (Sehnert et al., 2002) and Tg(hsp70:cafgfr1)^pd3 (Marques et al., 2008). Embryos carrying Tg(myl7:EGFP)^twu34 were identified by GFP fluorescence. Embryos carrying Tg(kdrl:Hsa.HRAS-mCherry)^s896 were identified by mCherry fluorescence. Embryos carrying Tg(hsp70l:nkx2.5-EGFP)^fcu1 were identified by GFP fluorescence following heat shock. Embryos homozygous for tnnt2a^b109 were identified by their lack of cardiac contractility. Embryos carrying Tg(hsp70:cafgfr1)^pd3 were either identified by dsRed fluorescence in the lens at 48 hpf or by PCR genotyping for dsred following immunofluorescence procedures, using the primers 5′-CAGTACGGCTCCAAGGTGTA-3′ and 5′-GTCCTCGAAGTTCATCACGC-3′, as in Figs S13 and S14. All zebrafish work followed protocols approved by the UCSD IACUC.

To generate transgenes for heat-activated expression of MEK1^S219D, a constitutively active variant of zebrafish map2k1 (ZDB-GENE-040426-2759), we first amplified the MEK1^S219D coding sequence from a plasmid received as a gift from Dr Chia-Hsiung Cheng (Chou et al., 2015) and cloned the amplicon into the middle donor vector pDONR221 (Invitrogen, 12536-017) through a Gateway BP reaction using the primers 5'-GGGGACAAGTTTGTACAAAAAAGCAGGCTACCATGCAGAAAAGGAGGAAGCC-3′ and 5'-GGGGACCACTTTGTACAAGAAAGCTGGGTGCATTCCCACACTGTGAGTCG-3′. The transgene Tg(hsp70:MEK1^S219D-P2A-mCherry) was then generated through a Gateway LR reaction (Invitrogen, 12537-023) containing four plasmids: pDONR221-MEK1^S219D, p5E-hsp70l (Kwan et al., 2007), p3E-P2A-mCherry-pA (Covassin et al., 2009) and pDestTol2pA3 (Franco et al., 2019). We employed standard protocols to create transgenic founders (Soroldoni et al., 2009). Prospective founder fish were screened by evaluating mCherry fluorescence in their F1 progeny following heat shock and phenotypic analysis was performed on the F2 and F3 progeny of two separate founders carrying distinct integrations of Tg(hsp70:MEK1^S219D-P2A-mCherry), both of which exhibited qualitatively similar levels of mCherry fluorescence with relatively uniform brightness throughout the embryo. Similar phenotypes were observed in both transgenic lines. Data shown in Figs 4, 5, 7, 8, Figs S10 and S14 depict results from the line Tg(hsp70:MEK1^S219D-P2A-mCherry)^sd70, and data shown in Fig. S11 depict results from the line Tg(hsp70:MEK1^S219D-P2A-mCherry)^sd71. In all experiments, embryos carrying Tg(hsp70:MEK1^S219D-P2A-mCherry) were identified by mCherry fluorescence following heat shock.

Standard whole-mount in situ hybridization, fluorescent in situ hybridization, and fluorescent in situ hybridization in combination with immunofluorescence were performed as previously described (Pradhan et al., 2017). We used established probes for the following genes: amhc (myh6; ZDB-GENE-031112-1), vmhc (myh7; ZDB-GENE-991123-5), nkx2.5 (ZDB-GENE-980526-321) and nkx2.7 (ZDB-GENE-990415-179).

Regional location of ectopic amhc expression within the ventricle (Figs 2L, 5E, 6I, and Fig. S9E) was categorized while the treatment conditions or genotypes of the samples were anonymized. In each embryo examined, amhc expression was assigned to the regions schematized in Fig. 2K and illustrated in Fig. S3, based on a qualitative approximation of the location of expression relative to the outline of the ventricular chamber. Expression directly above the right side of the atrioventricular canal (AVC) constriction was assigned to the ventricular apex (V apex). Expression at the arterial pole of the heart was assigned to the OFT. Expression in the left edge of the chamber, extending between the AVC and the OFT, as well as expression in the left-most third of the chamber, was assigned to the IC. Expression in the right edge of the chamber, extending between the V apex and the OFT, as well as expression in the right-most third of the chamber, was assigned to the OC. Expression in the central third of the chamber, between the IC and the OC, was assigned to the central V. Assessment of regional location can be challenging in the context of a dysmorphic ventricle; even so, we found that we could discern the AVC constriction, chamber curvatures and arterial pole in nearly all ventricles examined. For the rare cases of completely linear hearts without any evident curvatures or AVC constriction, we declined to score the samples.

Whole-mount immunofluorescence was generally performed as previously described (Pradhan et al., 2017), using antibodies listed in Table S1. For detection of pERK, we modified other established protocols (Lei et al., 2021; Watterston et al., 2021). Specifically, embryos were fixed in freshly made 4% paraformaldehyde (PFA) in PBS with PhosSTOP (Roche, 4906845001) overnight at 4°C, then gradually dehydrated with a series of methanol/PBST (PBS with 0.1% Tween) washes before storage in 100% methanol at −20°C overnight. Embryos were then permeabilized in 50:50 acetone/methanol for 20 min at room temperature (RT) and re-hydrated through a series of methanol/PBST washes. Embryos were gradually warmed in 150 mM Tris-HCl (pH 9.0) buffer and were then kept at 70°C for 30 min. After cooling to RT, embryos were treated with Proteinase K (10 μg/ml) for 5 min (5 μg/ml if younger than 24 hpf). Embryos were next fixed in 4% PFA in PBS with PhosSTOP for 20 min at RT before being washed with PBST, blocked with dilution buffer (PBST with 2% bovine serum albumin and 10% goat serum) for at least 1 h at RT, and then incubated with primary antibodies for 2 nights at 4°C. Embryos were then washed with PBST, incubated with secondary antibodies overnight at 4°C in dark conditions, and washed in PBST at RT before imaging.

Compounds (see Table S2) were dissolved with 100% DMSO to a concentration of 10 mM and were stored in small aliquots at −20°C. Stock solutions were diluted to working concentrations (as indicated in figure legends) in E3 buffer. During treatment, embryos were incubated at 28.5°C in dark conditions. Control embryos were treated with E3 containing the same dilution of DMSO.

To induce expression of Tg(hsp70:MEK1^S219D-P2A-mCherry), embryos were incubated in E3 buffer in a pre-warmed 37°C heat block for 1 h. To induce expression of Tg(hsp70l:nkx2.5-EGFP), embryos were incubated in E3 in a 37-37.5°C heat block for 30 min to 1 h. To induce expression of Tg(hsp70:cafgfr1), embryos were incubated in E3 in a pre-warmed 40°C incubator for 30 min, and they were then transferred to a pre-warmed 42°C heat block for 3 min.

All bright-field images, as well as fluorescent images of embryos carrying Tg(kdrl:Hsa.HRAS-mCherry), were captured using a Zeiss Axiocam and a Zeiss Axiozoom microscope and were processed using Zeiss AxioVision software. Confocal images were captured using a Leica SP5 or SP8 confocal laser-scanning microscope with a 25× water objective and a slice thickness of 1 µm, and confocal data were analyzed with Imaris software (Bitplane). To capture the full range of differences in levels of pERK signal, we carefully adjusted the laser power and exposure settings in order to avoid saturating the confocal detector. In each experiment comparing different datasets at 20 hpf (Fig. 4, Figs S4, S5, S13, S14), settings were selected using the set of samples with higher pERK levels and were then kept consistent throughout the experiment: for example, the DMSO-treated samples, and not the PD0325901-treated samples, were used for the experiments shown in Fig. S4, and the Tg(hsp:caMEK1) samples, and not the wild-type sibling samples, were used for the experiments shown in Fig. 4.

To evaluate pERK signal within the myocardium (Figs 3A′-D′, 4A′-D′, Figs S4A′-D′, S5A′-D′, S6, S13A′-D′, S14), we created a ‘myocardial mask’ based on the intensity of either the Tg(myl7:EGFP) or CH1 signals, using the ‘surface’ function in Imaris. Only pERK signal within the boundary of the surface was extracted; however, this masking strategy might still include signals from tissues closely attached to the myocardium, such as the endocardium. Using the Imaris ‘spot’ function within the myocardial mask, we identified all of the cardiomyocytes in each sample and then counted the number of pERK⁺ cardiomyocytes that had an intensity mean value for the pERK channel above a certain threshold (Figs 3E, 4E, Fig. S13E). For each set of data acquired with the same settings, this threshold was set to include every spot that could be easily distinguished by eye from background fluorescence. We used a diameter of 5 µm when creating spots for all image analyses. To evaluate the intensity of the pERK signal (Fig. 4F, Figs S13F, S14), we first used the ‘spot’ function to mark all of the EGFP⁺ nuclei within the myocardial mask, as the EGFP generated by Tg(myl7:EGFP) expression typically accumulates in the nucleus at the stages examined. Then, to compare the overall pERK intensity between samples (Fig. 4F, Fig. S13F), we obtained the mean of the intensity mean values for the pERK channel from all of the EGFP⁺ spots in each embryo. To examine the range of pERK intensity among cardiomyocytes (Fig. S14), we obtained intensity mean values for the pERK channel in each of the individual EGFP⁺ spots. Additionally, in order to compare the intensity of pERK signal between biological replicates (Fig. 4F, Figs S13F, S14), we normalized all of the values from wild-type samples and transgenic embryos such that the median of the wild-type values within each imaging session or clutch would be equal to 1 arbitrary unit (arb. unit). The myocardial cone is essentially a single layer of cardiomyocytes at 20 hpf (Holtzman et al., 2007), and we did not observe any notable differences in myocardial thickness among our samples.

Statistical analysis was performed using Graphpad Prism 9. Non-parametric two-tailed Mann–Whitney tests were conducted when data involved a continuous variable (Figs 3I, 4E,F, Figs S13E,F, S14). Fisher's exact test was used when data involved categorical variables (Figs 5E, 6I). Asterisks in graphs are used to indicate statistical significance: *P<0.05, **P<0.01 and ****P<0.0001. For in situ hybridization and immunofluorescence results without quantification, images shown are representative examples of multiple observations. The number of embryos examined is specified in the figure legends; in each case, the results typically represent at least two independent replicates of each experiment.

We thank C. H. Cheng for providing the MEK1^S219D construct, H. Kwan for generating the Tg(hsp70:MEK1^S219D-P2A-mCherry) plasmid, and J. Diaz for creating the illustration in Fig. 2K. We also thank T. Sanchez, A. Yarbrough, G. Avillion, and the UCSD Animal Care Program for excellent zebrafish care. Finally, we thank K. Cooper, X. Sun, N. C. Chi, J. Chen, G. Guerrero, and the members of the Yelon lab for thoughtful input.

The peer review history is available online at https://journals.biologists.com/dev/lookup/doi/10.1242/dev.202183.reviewer-comments.pdf