Lamb1a regulates atrial growth by limiting second heart field addition during zebrafish heart development

Tissue morphogenesis requires tight coordination of changes in cell shape and organisation, gene expression and tissue patterning, together with the integration of intrinsic and extrinsic signalling cues. Cardiac development represents an excellent example of such complex morphogenesis, in which the linear heart tube undergoes growth, local tissue deformation and functional regionalisation. The importance and complexity of heart morphogenesis are evident in the prevalence of congenital heart defects (CHDs), which occur in at least 1% of live births and are the leading cause of birth defect-related deaths worldwide (Triedman and Newburger, 2016).

Heart looping and chamber ballooning are key stages in cardiac development, during which the heart tube undergoes a complex morphological rearrangement resulting in a helical looped tube in mouse and a planar looped tube in zebrafish (Danio rerio) (Desgrange et al., 2018). This is concomitant with an increase in myocardial cell number, primarily achieved through cell addition to the poles of the developing heart from a progenitor pool in the adjacent mesoderm, termed the ‘second heart field’ (SHF) (Kelly, 2012). During cardiogenesis in mouse and chick, SHF addition generates a significant proportion of cardiac tissue, including the right ventricle, atria and inflow and outflow tracts (OFT) (Kelly et al., 2001; Mjaatvedt et al., 2001; Waldo et al., 2001). In zebrafish, the SHF makes a similar contribution to the inflow tract at the base of the atrium, the single ventricle and the OFT (Hami et al., 2011; Lazic and Scott, 2011; de Pater et al., 2009; Zhou et al., 2011). The signaling pathways required for SHF addition are highly conserved across vertebrates (Knight and Yelon, 2016; Rochais et al., 2009), with Fgf8 promoting SHF addition and an opposing retinoic acid (RA) gradient limiting SHF addition to the arterial pole (Ryckebusch et al., 2008; Rydeen and Waxman, 2016; Zaffran et al., 2014). SHF addition and cardiac morphogenesis are tightly linked, with defects in SHF addition leading to heart malformations and congenital heart disease (Francou and Kelly, 2016).

Mechanical forces are also linked with SHF addition. Recent studies demonstrated that the SHF epithelium is under tension, which is proposed to regulate cell orientation and cell division, driving extension of the linear heart tube (Francou et al., 2017); it has also been suggested that heart tube contractility could contribute to this tension. Finally, because heart function begins once the heart tube is formed, cardiomyocyte contractility, blood flow, SHF addition and morphogenesis all occur simultaneously (Beis et al., 2005; Dietrich et al., 2014; Heckel et al., 2015; Kalogirou et al., 2014; Samsa et al., 2015; Vermot et al., 2009), resulting in a complex interplay of biochemical and biomechanical cues driving robust heart morphogenesis.

The extracellular matrix (ECM) is an important signalling centre that influences biochemical signalling between cells and provides biomechanical stimuli. Numerous studies have highlighted the importance of the cardiac ECM during heart development (Derrick and Noël, 2021); however, comparatively little is known about the specific roles that individual ECM components play in promoting heart looping, chamber morphogenesis and cardiac growth. Laminins are large heterotrimeric complexes deposited early during ECM construction. Consisting of an alpha, beta and gamma chain, laminin trimers are an essential component of the basement membrane, where they interact with integrin receptors on the cell membrane and facilitate ECM organisation in the interstitial matrix (Domogatskaya et al., 2012; Mouw et al., 2014). Multiple alpha, beta and gamma subunits are encoded in the genome and assemble into a variety of trimer isoforms, which often exhibit tissue-restricted expression and play specific roles in different tissue contexts (Schéele et al., 2007). Previous studies suggest that laminins have important roles in human heart development and function. Deleterious mutations in LAMA4 have been identified in patients with dilated cardiomyopathy (Knöll et al., 2007), a mutation in LAMA5 has been associated with a multi-systemic disorder that includes cardiac abnormalities (Sampaolo et al., 2017), and ∼30% of patients with Dandy–Walker Syndrome (DWS), a rare brain malformation linked to mutations in LAMC1, also present with CHDs (Darbro et al., 2013). Although these studies suggest requirements for laminins in cardiac form and function, mechanistically little is known about the roles that they play during cardiac morphogenesis. Direct evidence supporting a role for laminins in heart development comes from Drosophila, in which laminins promote the formation and integrity of the dorsal vessel (Haag et al., 1999; Yarnitzky and Volk, 1995). Current vertebrate models have provided limited insights into the role of laminins in cardiac development. Lama4 mutant mice survive postpartum without overt cardiac defects, although, over time, pups develop enlarged hearts with larger cardiomyocytes (Thyboll et al., 2002; Wang et al., 2006). However, interrogating the role of laminins more broadly in heart development is challenging because of the early lethality in both Lamc1 and Lamb1 mutant mice (Miner et al., 2004; Smyth et al., 1999).

In this study, we identify two novel functions for laminins during heart development in zebrafish, promoting heart looping morphogenesis and restricting cardiac size. We show that Lamb1a controls cardiac growth by limiting SHF addition and demonstrate that excessive atrial SHF addition to the venous pole in lamb1a mutants is rescued by blocking heart contractility. Finally, we demonstrate that loss of lamb1a disrupts expression of RA-responsive genes in the heart in a contractility-dependent manner, supporting a role for laminins in coupling mechanical force, intercellular signalling and cardiac growth. Thus, this study presents the first reported role for laminins in early vertebrate heart development.

To investigate the role of laminin complexes in early vertebrate heart morphogenesis, we probed a previously published transcriptomic analysis of cardiac gene expression to identify laminin subunit genes expressed in the heart tube (Derrick et al., 2021) and, in combination with an in situ hybridisation screen, identified a subset of laminin subunits with cardiac expression during early stages of heart looping (Fig. 1; Fig. S1).

At 30 hours post fertilisation (hpf) (Fig. 1A), during early heart tube morphogenesis, six laminin subunits are expressed in the zebrafish heart: two alpha chains (lama4 and lama5; Fig. 1C,E); three beta chains (lamb1a, lamb1b and lamb2; Fig. 1G,I,K) and a single gamma chain (lamc1; Fig. 1M). Given that specific laminin isoforms can exhibit tissue-specific deposition, we carried out two-colour fluorescent in situ hybridisation at 30 hpf to identify whether the myocardium and endocardium have a specific laminin expression profile (Fig. S1). This identified two laminin genes expressed in the endocardium lama4 and lamb1b (Fig. S1A,B); two laminin genes expressed in the myocardium: lama5 and lamb2 (Fig. S1C,D); and two laminin subunits expressed in both myocardium and endocardium: lamb1a and lamc1 (Fig. S1E,F).

Whereas at 30 hpf the majority of laminin genes were expressed along the length of the heart tube, following initial heart looping morphogenesis at 55 hpf (Fig. 1B), the expression of most laminin subunits became restricted to the ventricle and atrioventricular canal (Fig. 1D,F,H,L,N), with the exception of lamb1b, which was expressed only in the atrioventricular canal (Fig. 1J). This dynamic, spatiotemporal control of specific laminin subunit expression during early heart development suggests that individual endocardial- or myocardial-derived laminin complexes may play a role in early heart morphogenesis.

Having identified potential laminin complexes expressed in the heart during early morphogenesis, we examined the role of laminins during heart development. Laminins are heterotrimeric complexes comprising single alpha, beta and gamma chains (Fig. 1O), which are assembled intracellularly prior to deposition into the ECM; thus, the removal of a single subunit is sufficient to prevent extracellular secretion of the complex (Yurchenco et al., 1997). Therefore, to investigate the requirement for laminins during heart looping morphogenesis, we targeted the single gamma subunit lamc1, expressed in both the myocardium and endocardium (Fig. 1M,N; Fig. S1F). Using CRISPR-Cas9 mutagenesis, we generated transient F0 lamc1 mutants (F0/crispants) (Burger et al., 2016). lamc1 crispants recapitulated the morphological phenotype of stable lamc1 mutants with high efficacy, whereas uninjected embryos or injection controls [guide RNA (gRNA) only or Cas9 only] were morphologically normal (Fig. S2A-D) (Odenthal et al., 1996; Parsons et al., 2002). lamc1 crispants formed a beating heart tube by 26 hpf and, at 2 days post fertilisation (dpf), exhibited mild pericardial oedema, suggesting defects in heart looping morphogenesis (Fig. S2D). We assessed the impact of loss of Lamc1 on heart morphology by myl7 expression analysis (Fig. 2A-F), quantifying looping ratio and heart size at 30 hpf, 55 hpf and 72 hpf (Fig. 2G-L; Fig. S2E-F). lamc1 crispant heart tubes were smaller than control hearts at 30 hpf (Fig. 2A,B,H) and, at 55 hpf, lamc1 crispants had failed to undergo correct heart looping morphogenesis (Fig. 2C,D), displaying a significant reduction in heart looping ratio compared with controls (Fig. 2I). Interestingly, lamc1 crispant hearts were of comparable size to those of their control siblings at 55 hpf (Fig. 2J) whereas, by 72 hpf, in addition to abnormal cardiac morphology (Fig. 2F,K; Fig. S2E), lamc1 crispant hearts appeared significantly larger than those in controls (Fig. 2E,L; Fig. S2F). These results reveal that laminins promote initial heart looping and might regulate cardiac size throughout morphogenesis.

Distinct laminin complexes play varied yet specific roles in different developmental contexts (Schéele et al., 2007). Given that lamb1a exhibited similar expression dynamics and tissue specificity as lamc1 (Fig. 1; Fig. S1), this suggested that Lamb1a and Lamc1 may form part of the laminin complexes required for heart development. To investigate this, we generated two stable mutant alleles, lamb1a^Δ19 and lamb1a^Δ25, using CRISPR-Cas9-mediated genome editing (Fig. S2G-M). In contrast to lamc1 crispants, at 30 hpf, lamb1a^Δ25 mutant hearts were comparable in size to their sibling controls (Fig. 2M,N,T). However, at 55 hpf, lamb1a mutants also displayed a mild yet significant reduction in heart looping compared with siblings (Fig. 2O,P,U), although this defect was less severe than that observed in lamc1 crispants (compare Fig. 2D with Fig. 2P; unpaired t-test of looping ratio between mutants: P<0.0001). lamb1a mutants also exhibited a mild increase in heart size at 55 hpf (Fig. 2V; Fig. S2N,O,R), a progressive defect resulting in significantly cardiomegaly at 72 hpf (Fig. 2Q,R,X; Fig. S2P,Q,S). Although mutant alleles for both lamc1 and lamb1a have been previously described (Hochgreb-Hägele et al., 2013; Odenthal et al., 1996; Parsons et al., 2002), defects in heart development have not been reported. Analysis of heart morphology in grumpy^tj299a (gup, lamb1a) and sleepy^sa379 (sly, lamc1) mutants revealed similar phenotypes to our loss-of-function models (Fig. S3), although lamb1a^Δ25 mutants presented with slightly more severe defects in looping morphology at 55 hpf compared with grumpy^tj299a mutants. Together, these data demonstrate two previously uncharacterised requirements for laminins in vertebrate heart morphogenesis: promoting heart looping and restricting cardiac size.

The difference in phenotypes between lamb1a and lamc1 mutants suggested that other laminin beta subunits may either act during early heart morphogenesis or functionally compensate for loss of lamb1a. We examined the impact of loss of lamb1a on the expression of the other laminin beta 1 paralog (lamb1b) in the heart at 30 hpf and 55 hpf, revealing a striking upregulation and expansion of lamb1b expression in lamb1a^Δ25 mutants (Fig. S4A-D). This suggested that upregulation of lamb1b could compensate for loss of lamb1a (El-Brolosy et al., 2019), resulting in the weaker looping morphogenesis phenotype in lamb1a mutants compared with loss of lamc1 models. To investigate this, we generated two lamb1b promoter deletion alleles: lamb1b^Δ183 and lamb1b^Δ428 (Fig. S4E,F). We incrossed lamb1b;lamb1a^Δ25 double-heterozygous adult fish to obtain lamb1b;lamb1a^Δ25 double-mutant embryos and confirmed the absence of lamb1b transcript at 30 hpf (Fig. S4G-J). Analysis of heart size and morphology in lamb1b;lamb1a^Δ25 double mutants at 55 hpf revealed that loss of lamb1b did not modify the lamb1a mutant phenotype in particular with respect to looping ratio, demonstrating that, despite its upregulation in lamb1a^Δ25 mutants, lamb1b does not compensate for the loss of lamb1a (Fig. S4K-N). We next investigated the expression of another laminin beta subunit, lamb2, in lamb1a^Δ25 mutants. At both 30 hpf and 55 hpf, lamb2 expression levels in the hearts of lamb1a^Δ25 mutants were comparable with those in their siblings (Fig. S5A-D). To rule out the possibility that endogenous lamb2 compensates for loss of lamb1a, we generated lamb2 F0 crispants (Fig. S5E), either in a sibling or lamb1a^Δ25 mutant background. lamb2 crispants did not exhibit gross morphological defects, in line with previously published lamb2 mutants (Jacoby et al., 2009), and we used PCR analysis to confirm the successful mutagenesis of the multiple lamb2 target sites. Similar to our functional analysis of lamb1b, analysis of heart size and morphology in lamb2 crispants at 55 hpf revealed that loss of lamb2 alone did not impact cardiac morphology or size, did not modify the lamb1a cardiac phenotype, and did not recapitulate the lamc1 crispant looping defect (Fig. S5F-O). Together, these results suggest that lamc1 and lamb1a may play functionally or temporally different roles in cardiac development, or may represent different dynamics of maternal deposition of lamc1 and lamb1a.

The progression from reduced heart size at 30 hpf to increased heart size at 72 hpf in lamc1 crispants (Fig. 2) suggests that laminins may regulate heart size differently at early and late stages of cardiac development, and that the early lamc1-dependent requirement for laminin in cardiac size is closely linked to looping morphogenesis. Conversely, lamb1a mutants exhibited relatively mild defects in initial heart looping morphogenesis, but developed pronounced cardiomegaly. This represents an interesting model because defects in heart size are often coupled with a severe impact on looping morphology, such as the loss of cerebral cavernous malformation (CCM) pathway components, in which cardiac chambers are larger, but morphology is also severely disrupted (Mably et al., 2006). Therefore, to understand how laminin regulates the growth of the heart specifically subsequent to tube formation, we focused our analysis on the lamb1a mutant.

To determine whether the growth of a specific chamber was impacted by the loss of lamb1a, we examined chamber size at 55 hpf and 72 hpf by analysis of myh7l and myh6 expression in the ventricle and atrium, respectively. lamb1a mutants displayed a significant increase in the size of both chambers (Fig. 3A-F), with progressive enlargement between 55 hpf and 72 hpf, suggesting that laminin limits the growth of both chambers. Two mechanisms could account for increased cardiac size: cardiomyocyte hypertrophy or increased cell number. Given that loss of lama4 and integrin-linked kinase (the intracellular effector of laminin-integrin signalling) in zebrafish has previously been associated with dilated cardiomyopathy (Knöll et al., 2007), this suggested that cardiomegaly in lamb1a mutants may result from enlarged cardiomyocytes. We quantified the internuclear distance in both chambers in wild-type (WT) and lamb1a^Δ25 mutant Tg(-5.1myl7:DsRed2-NLS) embryos [henceforth Tg(myl7:DsRed)], in which myocardial nuclei express DsRed2, at 55 hpf and 72 hpf (Fig. 3G-I). In contrast to our expectation that loss of lamb1a would result in enlarged cardiomyocytes, lamb1a^Δ25 mutant embryos did not exhibit increased internuclear distance compared with siblings, demonstrating that Lamb1a is not restricting cardiomyocyte size (Fig. 3J,K). Therefore, we hypothesised that cardiomegaly in lamb1a mutants results from increased cell number; quantification of DsRed-positive cardiomyocytes in sibling and lamb1a^Δ25 mutant embryos at 55 hpf and 72 hpf revealed a significant increase in atrial cell number in lamb1a^Δ25 mutant embryos at both stages compared with sibling embryos (Fig. 3L,M). Together, these results suggest that Lamb1a controls atrial size by regulating cell number. Analysis of DsRed-positive cell number at 30 hpf revealed comparable numbers of cardiomyocytes in lamb1a^Δ25 mutants and siblings (Fig. S6A-C), suggesting that initial cell number in the heart tube is not affected and that increased cardiomyocyte number in lamb1a^Δ25 mutants is a progressive defect.

During morphogenesis, the heart grows primarily through addition of cells to the poles of the heart from the SHF. Although previous studies demonstrated that SHF addition to the arterial pole of the heart is sensitive to perturbations in ECM composition (Derrick and Noël, 2021), comparatively less is known about SHF addition to the venous pole and how the ECM may regulate this process. Given that lamb1a^Δ25 mutants exhibited increased atrial cell number in the heart during the window of SHF addition, we hypothesised that loss of Lamb1a-containing laminin trimers would lead to increased cardiac size through elevated SHF addition. We visualised SHF addition in Tg(myl7:eGFP);Tg(myl7:DsRed) double-transgenic sibling and lamb1a^Δ25 mutant embryos, in which cardiomyocytes derived from the linear heart tube/first heart field were marked by both GFP and DsRed expression, whereas cells recently added from the SHF were GFP positive only (de Pater et al., 2009) (Fig. 4C,C′). At 55 hpf, lamb1a^Δ25 mutants appeared to have a larger GFP+;DsRed− area at the poles of the heart, suggesting an increased number of SHF cells (Fig. 4A-B″). Quantification of GFP+;DsRed− cell number in the atrium at 55 hpf revealed a significant increase in SHF cells at the venous pole of lamb1a^Δ25 mutant embryos compared with siblings (Fig. 4D,E), demonstrating that Lamb1a limits atrial size by restricting SHF addition to the venous pole. Analysis of SHF addition to the venous pole of lamb1a^Δ25 mutant hearts at 30 hpf revealed no significant differences compared with siblings (Fig. S6D), suggesting the excess atrial SHF cells in lamb1a mutants are added throughout looping morphogenesis. Comparable analysis of cardiomyocyte number in lamc1 crispants revealed no obvious defects in DsRed+ cell number in the atrium at 55 hpf (Fig. S6E-I) but did suggest an increase in SHF addition to the venous pole (Fig. S6J). This further supports a role for laminins once the heart tube has formed in regulating SHF addition to the venous pole during looping morphogenesis.

Given that lamb1a^Δ25 mutants also exhibit an increase in ventricular size at 55 hpf (Fig. 3F), we examined SHF addition to the arterial pole of the heart. Cardiomyocytes were more densely packed in the arterial pole than in the venous pole (Fig. 3J,K); therefore, instead of cell number, we quantified the amount of GFP+;DsRed− tissue distal to the first DsRed+ cell in the arterial pole. lamb1a^Δ25 mutants exhibited a significant increase in the amount of GFP+ SHF tissue at the arterial pole (Fig. 4F), suggesting that, similar to the venous pole, laminin limits SHF addition to the ventricle during heart looping morphogenesis.

Increased SHF addition to the atrium of lamb1a^Δ25 mutants could result from a larger SHF progenitor pool at the venous pole. Thus, we examined expression of the transcription factor isl1a, which is expressed in SHF cells and required for SHF addition to the venous pole (de Pater et al., 2009). At both 24 hpf and 55 hpf, the expression domain and levels of isl1a were comparable between lamb1a^Δ25 mutant and sibling embryos (Fig. S7A-D′), demonstrating that increased SHF addition in lamb1a^Δ25 mutants was not the result of enlargement of the SHF domain. Conversely, we observed a mild increase in the ventricular expression of spry4, an FGF signalling response gene (Fig. S7E,F), in line with previous studies demonstrating roles for FGF signalling in cell addition to the arterial pole (Felker et al., 2018; de Pater et al., 2009). Although the heart grows almost exclusively through SHF addition between 1 dpf and 2 dpf (de Pater et al., 2009), we wished to rule out an increase in proliferation in lamb1a mutants driving increased cell number. Quantification of the number of phospho-histone H3 (pH3) cells in sibling and lamb1a^Δ25 mutant embryos at 55 hpf revealed no increased cell proliferation in lamb1a^Δ25 mutants (Fig. S7G), further supporting the hypothesis that cardiomegaly in lamb1a mutants is driven by the increased addition of SHF cells.

The increased SHF addition to the atrium in lamb1a mutants without expansion of the SHF domain suggested a SHF specification-independent mechanism. Our finding that lamb1b upregulation in lamb1a mutants was not triggered by compensatory pathways and did not play a functional role (Fig. S4) may provide clues to the mechanisms underlying increased SHF addition in lamb1a mutants.

During heart morphogenesis and SHF addition, lamb1b is expressed throughout the endocardium at 30 hpf and, by 55 hpf, is restricted to the atrioventricular canal, the site of atrioventricular valve development (Fig. 1). This expression dynamic is similar to genes required for valvulogenesis, such as notch1b and fibronectin 1b, which are regulated by the sensation of blood flow (Steed et al., 2016; Vermot et al., 2009). We hypothesized that lamb1b expression is also flow dependent, and that the misexpression of lamb1b throughout the endocardium of lamb1a^Δ25 mutants may reflect changes in cardiac function or sensitivity to blood flow upon loss of laminin.

To investigate this, embryos from a lamb1a^Δ25 heterozygous incross were injected at the one-cell stage with a translation-blocking morpholino oligonucleotide (MO) targeting troponin T type 2a (cardiac) (tnnt2a) to block heart contractility and abolish blood flow (Sehnert et al., 2002). Expression of lamb1b was then examined at 30 hpf (Fig. 5A-F), when uninjected and control tp53 MO-injected sibling embryos have low levels of endocardial lamb1b expression (Fig. 5A,B), and uninjected and control-injected lamb1a^Δ25 mutants misexpress lamb1b throughout the endocardium (Fig. 5D,E). As expected, MO-mediated knockdown of tnnt2a in sibling embryos resulted in a loss of lamb1b expression in the endocardium (Fig. 5C). Similarly, loss of heart contractility in lamb1a^Δ25 mutants resulted in reduced endocardial expression of lamb1b compared with control mutants (Fig. 5F). Together, these results demonstrate that endocardial expansion of lamb1b in lamb1a^Δ25 mutants is dependent on heart contractility.

To confirm altered flow responsiveness in lamb1a^Δ25 mutant endocardium, we analysed expression of klf2a, a transcription factor the expression of which is regulated by turbulent flow (Vermot et al., 2009). We injected embryos from a lamb1a^Δ25 heterozygous incross with tnnt2a MO and examined klf2a expression at 30 hpf. In sibling uninjected and control embryos, klf2a expression was localised predominantly to the arterial pole endocardium (Fig. 5G,H), whereas uninjected and control-injected lamb1a mutants misexpressed klf2a more broadly throughout the heart (Fig. 5J,K). MO-mediated knockdown of tnnt2a in sibling embryos resulted in almost total loss of klf2a expression in the endocardium compared with controls (Fig. 5I). Similar to the effect on lamb1b expression, loss of heart contractility in lamb1a^Δ25 mutants resulted in reduced endocardial expression of klf2a compared with control mutants (Fig. 5L). Together, these data suggest that loss of Lamb1a results in perturbations to the response to heart contractility and/or blood flow during heart looping morphogenesis.

The increase in SHF addition and altered expression of haemodynamic-responsive genes suggest that cardiac function may play a role in the failure to restrict heart size in lamb1a mutants. We investigated whether perturbing contractile and/or haemodynamic forces rescued cardiomegaly in lamb1a^Δ25 mutants by injecting the tnnt2a MO and measuring heart size (Fig. 6A-D). Blocking cardiac contractility in lamb1a^Δ25 mutant embryos significantly reduced heart size at 55 hpf and 72 hpf compared with control lamb1a^Δ25 mutants, suggesting that excess SHF addition is mediated by contractility upon loss of lamb1a (Fig. 6E; Fig. S8A). To confirm this, we examined the impact of loss of heart contractility specifically on SHF addition to the venous pole of the heart at 55 hpf in Tg(myl7:eGFP);Tg(myl7:DsRed) transgenic embryos (Fig. 6F-I′). In line with our previous data, uninjected and control-injected lamb1a^Δ25 mutant embryos displayed an increase in the number of newly added SHF cells in the atrium at 55 hpf (Fig. 6K). However, in lamb1a^Δ25 mutants injected with tnnt2a MO, addition of SHF cells to the atrium was rescued to levels comparable with siblings (Fig. 6K). Surprisingly, we also observed that loss of heart contractility in sibling embryos resulted in a subtle, yet significant, increase in GFP+;DsRed+ cells in the atrium at 55 hpf (Fig. 6J). This suggests more broadly that heart contractility may limit the timing of SHF addition to the atrium. Taken together, these data demonstrate that Lamb1a is required to limit excessive, contractile-dependent SHF addition to the atrium during heart looping morphogenesis.

To further investigate the interaction between loss of lamb1a and heart function, we analysed heart rate in lamb1a mutants. Heart rate at 2 dpf and 3 dpf was not significantly different between sibling and lamb1a^Δ25 mutant embryos (Fig. S8B), suggesting that increased rate of heart contractility is not driving aberrant, contractility-mediated SHF addition. Additionally, having observed changes in the expression of flow-responsive genes in lamb1a^Δ25 mutant embryos, we examined the role of shear stress in the generation of cardiomegaly in lamb1a mutants. We reduced blood viscosity by injecting embryos from an incross of lamb1a^Δ25 heterozygous adults with a MO targeting the transcription factor gata1a, a master regulator of erythropoiesis (Brownlie and Zon, 1999; Hsu et al., 2019). Knockdown efficiency was confirmed by expression analysis of the haemoglobin subunit hemoglobin beta embryonic-1.1 (hbbe1.1) (Quinkertz and Campos-Ortega, 1999) in control and gata1a-injected embryos at 55 hpf (Fig. S8C-E), and heart morphology was analysed at 72 hpf by myl7 expression. Quantification of heart area revealed that loss of Gata1a function and the resulting reduction in blood viscosity did not rescue lamb1a^Δ25 mutant heart size (Fig. S8F). Therefore, these data demonstrate that loss of lamb1a results in excessive SHF addition to the atrium through a contractile-dependent, shear stress-independent mechanism.

The molecular pathways underlying SHF patterning and addition are conserved among vertebrates. Complex, antagonistic, FGF8 and RA signalling networks act across the atrioventricular axis of the cardiac-forming region (Rochais et al., 2009). We showed an increase in atrial cell number and volume of SHF-derived tissue at the arterial pole, which correlated with a mild upregulation of the FGF-response gene spry4 in the ventricle of lamb1a mutants at 55 hpf (Fig. S7E,F). Given that cardiac function has been implicated in regulating the expression of aldh1a2 (formerly raldh2), a key enzyme in the RA synthetic pathway (Morton et al., 2008), we examined aldh1a2 (involved in RA synthesis and a RA-signalling target) expression in lamb1a^Δ25 mutants (Fig. 7). At 30 hpf, lamb1a^Δ25 mutants exhibited a marked upregulation of aldh1a2 expression throughout the endocardium, including in the venous pole/atrium (Fig. 7A,B), similar to the upregulated expression of lamb1b and klf2a (Fig. 5). This upregulation persisted at 55 hpf in the ventricle of lamb1a mutants (Fig. 7C,D), similar to the upregulation of lamb1b (Fig. S4). Therefore, to investigate whether the impact of loss of lamb1a on RA signalling is also contractility dependent, we examined aldh1a2 expression in sibling and lamb1a^Δ25 mutant embryos injected with tnnt2a MO. Whereas uninjected and control injected lamb1a^Δ25 mutant embryos had a clear expansion of aldh1a2 (Fig. 7H,I), MO-mediated knockdown of tnnt2a abrogated endocardial aldh1a2 expression in lamb1a^Δ25 mutants (Fig. 7J). Together, these results suggest not only that disruption to RA signalling in lamb1a^Δ25 mutants is partly regulated by heart function, but also that cardiac contractility itself influences activity of the pathways regulating SHF addition.

Having established that excessive SHF addition in lamb1a mutants is dependent on heart contractility, and that RA signalling appears dysregulated upon loss of laminin in a contractility-dependent manner, we investigated whether we could rescue cardiomegaly in lamb1a mutants through modulation of RA. Studies in mouse have shown that loss of RA signalling in aldh1a2 mutants results in hypoplastic atria (Niederreither et al., 2001; Sirbu et al., 2008), that upregulation of aldh1a2 is a consequence of insufficient RA signalling (Chen et al., 2001; Dobbs-McAuliffe et al., 2004; Niederreither et al., 1997), and that increased spry4 expression suggests overactive FGF signalling, which is antagonised by RA signalling (Rochais et al., 2009). Therefore, we hypothesised that addition of exogenous RA during early heart looping morphogenesis may rescue heart size in lamb1a mutants. We treated embryos from a lamb1a^Δ25 heterozygous incross from 24 hpf to 55 hpf with 100 nM RA, washed the drug off and allowed the embryos to develop to 72 hpf. We confirmed the efficacy of our drug treatment by expression analysis of the RA-responsive gene dhrs3a (Waxman et al., 2008), which was upregulated at 55 hpf upon RA treatment (Fig. S9A-F). Analysis of heart size in sibling and lamb1a^Δ25 mutant embryos at 72 hpf following 100 nM RA treatment revealed only a partial rescue of cardiac size in lamb1a mutants (Fig. S9G-P). This suggests that, although contractility-dependent RA signalling in lamb1a mutants is perturbed, this is not the only pathway driving increased heart size, and that heart function likely affects SHF addition through additional mechanisms.

Thus, we have shown for the first time the requirement for laminins in regulating early vertebrate heart morphogenesis, promoting heart morphology and restricting heart size through restriction of SHF addition. Furthermore, our data suggest that the ECM and cardiac contractility function together to regulate the balance of SHF-related signalling pathways.

We provide the first evidence that laminin restricts heart growth during looping morphogenesis by limiting the number of SHF cells incorporated into the venous pole of the heart. Previous studies identified roles for ECM components, such as versican and fibronectin (Fn), in promoting SHF addition to the arterial pole of the heart (Kern et al., 2007; Mittal et al., 2013, 2019; Mjaatvedt et al., 1998; Yamamura et al., 1997); however, we identified an opposing role for Lamb1a in restricting excessive SHF addition to both poles. Highlighting the importance of cell-ECM interactions in the SHF, loss of Tbx1, a master regulator of SHF addition, results in reduced expression of integrin, loss of focal adhesion markers, and impaired filopodia formation in the SHF (Alfano et al., 2019; Francou et al., 2014).

Heart function is tightly linked to heart morphology during development, and previous studies focussed on the impact of contraction on regionalised ventricular cell shape change, valvulogenesis and trabeculation (Auman et al., 2007; Bartman et al., 2004; Cai et al., 2019; Samsa et al., 2015; Staudt et al., 2014). Our finding that excessive atrial SHF addition in lamb1a mutants can be rescued by abolishing heart contractility (Fig. 6) suggests that laminin may alter the physical force of heart contractility. lamb1a and lamc1 are expressed broadly throughout the zebrafish embryo during the window of SHF migration, both within the heart and in the surrounding tissues in which the SHF resides. Therefore, whether the role of Lamb1a in restricting SHF addition is autonomous or non-autonomous to the heart tube remains an open question.

The upregulation of flow-sensitive klf2a expression in lamb1a mutants suggests that the dynamics of myocardial wall contraction may be altered. Given that laminins coordinate ECM assembly, loss of laminin may alter ECM stiffness, which has been shown in vitro to impact cardiomyocyte contractility (Bhana et al., 2010; Engler et al., 2008), whereas ECM composition can also affect the organisation of contractile apparatus in cardiomyocytes (Bildjug and Pinaev, 2014; Hilenski et al., 1989; Vanwinkle et al., 1996). Alternatively, loss of laminin could affect how SHF cells interact with the underlying ECM. Studies in mouse have shown that epithelial tension in the posterior SHF is accompanied by nuclear localisation of the mechanosensitive transcription factor YAP (Francou et al., 2017), a phenomenon associated with increased ECM stiffness (Dobrokhotov et al., 2018), and recent in vitro studies demonstrated that laminin itself promotes nuclear YAP shuttling in keratinocytes (De Rosa et al., 2019). Activation of YAP/TAZ signalling in the SHF at the venous pole of the heart is conserved in zebrafish (Fukui et al., 2018), and the increased atrial SHF addition in 55 hpf in lamb1a mutants is similar to that observed in lats1/lats2 double mutants, which have a global increase in activity of YAP/TAZ signalling (Fukui et al., 2018). This suggests that ECM-mediated mechanotransduction in the SHF may represent a conserved mechanism regulating SHF addition that is disrupted in lamb1a mutants. How this is impacted by cardiac contraction is unclear, although it has been speculated that cardiac function could contribute to SHF tension (Francou et al., 2017), and it is conceivable that cardiomyocyte contractility contributes to the balance of pulling and pushing forces regulating SHF incorporation into the OFT in mice (Li et al., 2016). Mechanical loading has been implicated as a moderator of ECM content in other contexts, such as bone (Humphrey et al., 2014). Therefore, it is possible that loss of heart contractility could affect the composition of the cardiac ECM by, for example, upregulating ECM components, which could restore a suitable environment for SHF addition in lamb1a mutants. Importantly, although blocking contractility rescued excessive cell addition to the venous pole in lamb1a mutants, morphology of the inflow tract appeared impaired (Fig. 6), suggesting that cardiac contraction is also required to shape the venous pole.

In addition to facilitating cell-ECM interactions, laminins are crucial for ECM assembly, interacting with ECM components such as heparan sulfate proteoglycans (HSPGs). In turn, HSPGs interact with additional ECM components, such as Fn, and signalling molecules, such as FGF (Mouw et al., 2014). HSPG and Fn both regulate FGF signalling during SHF addition to the OFT/arterial pole in mice (Mittal et al., 2010; Zhang et al., 2015); therefore, laminin may interact with HSPGs to regulate extracellular signalling promoting SHF addition. Supporting this, we observed a mild upregulation of the FGF-response gene spry4 in the ventricle of lamb1a mutant hearts at 55 hpf (Fig. S7). However, levels of FGF activity are also balanced by antagonistic RA signalling, and the upregulation of the RA-responsive RA-synthesising enzyme aldh1a2 in lamb1a mutants (Fig. 7) suggests that RA signalling is impaired, leading to a dysregulation of FGF activity. Supporting this hypothesis, timed RA treatments during early SHF addition partially rescued heart size in lamb1a mutants at 3 dpf (Fig. S9). However, global upregulation of RA is likely too broad to restore the balance of RA-FGF levels, and, therefore, it is difficult to interpret the specific contribution of disrupted RA signalling to the increased SHF addition and cardiomegaly in lamb1a mutants, given the complex antagonistic interactions. RA signalling during early development was recently proposed to define the rate of cardiac progenitor differentiation in the anterior lateral plate mesoderm, because disruption of RA signalling from 6 hpf onwards results in a reduction in ltbp3 expression and a loss of isl1a-positive pacemaker cells at the inflow tract (Duong et al., 2021). Importantly, we did not observe changes in the size of the isl1a expression domains at the venous pole (Fig. S7), suggesting that altered FGF-RA signalling does not affecting the size of the SHF progenitor populations in lamb1a mutants and that RA signalling is disrupted after SHF specification. Furthermore, analysis of aldh1a2 expression in lamb1a mutants in which cardiac contractility had been abrogated revealed that aldh1a2 upregulation in lamb1a mutants is dependent on heart function. This suggests that dysregulation of RA signalling is secondary to altered contractility in lamb1a mutants, highlighting the complexity of interaction between the ECM, cardiac function, cell signalling, and SHF addition. Our observation that the size of the isl1a expression domain was not increased in lamb1a mutants, coupled with our finding that proliferation was not upregulated at 55 hpf, supports the requirement for lamb1a in specifically regulating the timing or rate at which SHF cells are added to the venous pole. Although we suggest that this addition alone is sufficient to drive the increase in atrial size observed in lamb1a mutants, it is possible that SHF cells exhibit a transient increase in proliferation after they have been added to the heart or there is a subtle increase in the proliferative index of the myocardium, which we were unable to capture in this study.

Both LAMB1 and LAMB2 are detectable in human heart samples at gestational weeks 8/9, and are deposited into the ECM that surrounds the cardiomyocytes and the basement membrane of the endocardium (Roediger et al., 2010). Mutations in LAMC1 are linked with DWS, a rare CNS disorder associated with congenital heart defects (Darbro et al., 2013). We showed a conserved requirement for zebrafish lamc1 in heart morphogenesis, with crispants also displaying hydrocephalus (Fig. 2; Fig. S2), another symptom associated with DWS. Furthermore, our finding that loss of laminin may lead to altered contractility and heart morphology at relatively early stages of cardiac development could shed further light on the mechanisms underlying the progression of dilated cardiomyopathy in individuals with LAMA4 mutations (Knöll et al., 2007), given that contractile dysfunction is a factor in the initiation of cardiac remodelling and cardiomyopathies (Vikhorev and Vikhoreva, 2018). This conservation of laminin function highlights the value of zebrafish as a model for understanding the role of ECM dysfunction in human cardiac diseases.

Thus, we describe the first direct evidence that laminins promote morphogenesis and growth during early vertebrate heart development, uncovering a novel role for laminin in restricting contractility-dependent SHF addition to the venous pole. This work also identifies new links between ECM composition, mechanical and biochemical cues in shaping the heart, reinforcing the importance of the extracellular environment during organ morphogenesis.

Adult zebrafish (D. rerio) were maintained according to standard laboratory conditions. The following, previously described lines were used: WT (AB), Tg(myl7:eGFP) (Huang et al., 2003), Tg(myl7:lifeActGFP) (Reischauer et al., 2014), Tg(-5.1myl7:DsRed2-NLS)^f2 (Rottbauer et al., 2002), grumpy^tj299a (Odenthal et al., 1996) and sleepy^sa379 (Kettleborough et al., 2013). The lines generated for this study were: lamb1a^Δ19 (lamb1a^sh589), lamb1a^Δ25 (lamb1a^sh590), lamb1b^promΔ183 (lamb1b^sh587) and lamb1b^promΔ428 (lamb1b^sh588). Embryos were maintained in E3 medium (5 mM NaCl, 0.17 mM KCl, 0.33 mM CaCl₂, 0.33 mM MgS0₄) at 28.5°C and were staged according to Kimmel et al. (1995). Embryos older than 24 hpf were transferred into E3 medium containing 0.003% 1-phenyl 2-thiourea (PTU, Sigma-Aldrich P7629) to inhibit pigment formation and aid imaging. Animal work was approved by the local Animal Welfare and Ethical Review Body (AWERB) at the University of Sheffield, conducted in accordance with UK Home Office Regulations under PPLs 70/8588 and PA1C7120E, and in line with the guidelines from Directive 2010/63/EU of the European Parliament on the protection of animals used for scientific purposes.

To generate lamb1a (ENSDART00000170673.2) mutant zebrafish, lamb1a-targeting gRNAs were designed using CHOPCHOP (Labun et al., 2016; Montague et al., 2014) and, following selection of suitable gene-specific sequences, the first two nucleotides were converted from NG/GN to GG and the PAM sequence (NGG) was removed. A single gRNA targeting exon 6 (5′-GGATCCTCAATCCTGAAGGCAGG-3′) was selected. The reverse complement of the resulting sequence was inserted into an ultramer scaffold sequence (Hruscha et al., 2013) containing a T7 promoter (AAAGCACCGACTCGGTGCCACTTTTTCAAGTTGATAACGGACTAGCCTTATTTTAACTTGCTATTTCTAGCTCTAAAACxxxxxxxxxxxxxxxxxxxxCTATAGTGAGTCGTATTACGC), where ‘x’ denotes insertion site for gene-specific targeting sequence. The template was amplified by PCR (F: 5′-GCGTAATACGACTCACTATAG-3′, R: 5′-AAAGCACCGACTCGGTGCCAC-3′) and used as a template for in vitro transcription using MEGAshortscript T7 kit (Ambion/Thermo Fisher Scientific AM1354). gRNA was injected together with Cas9 protein (New England Biolabs M0386T) and Phenol Red (Sigma-Aldrich P0290) into the yolk at the one-cell stage. Each embryo was injected with 2 ng gRNA, 1.9 nM Cas9 protein and 10% Phenol Red. CRISPR Cas9-injected embryos were raised to adulthood (F0) and outcrossed to WT to identify F0 individuals with germline transmission of deletions that result in a frameshift and subsequent premature termination codon. Embryos from F0 outcrosses were genotyped by PCR to amplify the region of lamb1a targeted for mutagenesis (F: 5′-CTTCTGTCTCTCATGGGCCA-3′, R: 5′-TGCCTTTACTTTGAATTCTGGGG-3′), and mutations were analysed by Sanger sequencing. Two lamb1a-coding sequence deletion alleles were recovered: lamb1a^Δ19 (lamb1a^sh589) and lamb1a^Δ25 (lamb1a^sh590). F0 founders transmitting these mutations were outcrossed to WT (AB) and offspring were raised to adulthood. Phenotypic analyses were carried out using F2 or F3 adults.

Two gRNAs were designed to target the upstream of the annotated promoter of lamb1b (ENSDARG00000045524) according to the Eukaryotic Promoter Database (Dreos et al., 2015, 2017) [lamb1b CRISPR RNA (crRNA) 1: 5′-TTGTTAATAGCATAGTACATTGG-3′; underlining denotes the protospacer adjacent motif (PAM)] and downstream of the annotated initiation codon (lamb1b crRNA 2: 5′-GGAGAACAAGCAAAACGATGAGG-3′; underlining denotes the PAM). Sequence-specific crRNAs were synthesised by Merck, resuspended in MilliQ (MerckMillipore) water to 500 µM and dilutions made for working stocks. Next, 2 nl of a Cas9-gRNA ribonucleoprotein complex was injected into the yolk of one-cell-stage embryos. Each embryo was injected with 61.2 nM crRNA, 122.5 nM tracrRNA, 3.9 nM Cas9 and 14% Phenol Red.

CRISPR Cas9-injected embryos were raised to adulthood (F0) and individual adults were outcrossed to WT to identify the germline transmission of suitable promoter deletions. Embryos collected from these outcrosses were genotyped by PCR to amplify the region of lamb1b targeted for mutagenesis (Forward: 5′-TCACACTAAGACATGGGGCA-3′, Reverse: 5′-ACCAAGCAACCAAAACACTGA-3′). Successful promoter deletion was identified by the presence of a smaller PCR fragment by gel electrophoresis and subsequent Sanger sequencing of the PCR fragment to confirm the deletion. Two separate lamb1b promoter deletion alleles were recovered: lamb1b^promΔ183 (lamb1b^sh587) and lamb1b^promΔ428 (lamb1b^sh588). F0 founders transmitting these mutations were outcrossed to lamb1a^Δ25 heterozygous adults and offspring were raised to adulthood. Heterozygous F1 lamb1a; lamb1b adults were genotyped using the relevant primers for each locus and used for experiments.

lamc1 and lamb2 F0 CRISPR mutagenesis was carried out as previously described (Burger et al., 2016; Wu et al., 2018). lamc1- and lamb2-targeting gRNAs were designed using CHOPCHOP. gRNAs were synthesised following the method described for lamb1a mutagenesis. gRNAs were designed to target the initiation codon of lamc1 (ENSDART00000004277.8) (lamc1 F0 gRNA1: 5′-GGCTTTCAATGCGACCGTGGTGG-3′; lamc1 F0 gRNA2: 5′-GGCGTGCAGTCACGGAGCGATGG-3′). gRNAs were designed to target exons 6, 12, 20 and 24 of lamb2 (ENSDART00000147326.2) (lamb2 F0 gRNA1: 5′-GGACAGTGTCCATGCCGACCTGG-3′; lamb2 F0 gRNA2: 5′-CGAGCCGTCGACAGAAGGAGAGG-3′; lamb2 F0 gRNA3: 5′-TGCCGGAAACTGTACCCCTGGGG-3′; lamb2 F0 gRNA4: 5′-AGACTGTCAGGAGAACCACTGGG-3′). The injection mix containing gRNA, Cas9 protein and Phenol Red was assembled on ice and incubated at 37°C for 5 min to aid Cas9-gRNA ribonucleoprotein complex formation for more efficient mutagenesis, prior to loading into a microinjection needle. Then, 1 nl of Cas9-gRNA was injected into the yolk of one-cell-stage embryos; each injection comprised either 500 pg of each gRNA for lamc1 or 214.3 pg of each gRNA for lamb2, in addition to 1.9 nM Cas9 protein and 14% Phenol Red. Mutagenesis was confirmed through PCR amplification of the targeted region of genomic DNA (lamc1 Forward: 5-ATCAAGACAGTGACGGTAGCAA-3′, Reverse: 5′-TGTGGCATGATTTAGTGACTCC-3′; lamb2 target 1 Forward: 5′-TGTGAATGCAGTTTAGAGGGCT-3′, Reverse: 5′-CAGCACACTCTCTGATTTTTGC-3′; lamb2 target 2 Forward: 5′-CTGGCAGGTGTATCGCTACTTT-3′, Reverse: 5′-ATCCTGATAGCAGGGTCAAGAA-3′; lamb2 target 3 Forward: 5′-ACCTCTGCACTTTTAGACCACC-3′, Reverse: 5′-TAACCAAATGTTCTCAGAGGGG-3′; lamb2 target 4 Forward: 5′-CATACAGTTTACAGGCCAGTGC-3′, Reverse: 5′-GGGAGAGAATCAAACCAGAAAA-3′); uninjected, gRNA-only injected and Cas9-only injected embryos were included as controls. Multiple lesions induced by CRISPR-mediated mutagenesis resulted in heteroduplex formation during PCR, which were resolved on a 4% agarose TBE gel. For initial experiments, mutagenesis was confirmed by Sanger sequencing of heteroduplex PCR products, and multiple lesions were identified at the target site. lamc1 mutagenesis was confirmed through heteroduplex analysis of the target loci in embryos displaying a morphological lamc1 mutant phenotype. Given that lamb2 mutants do not exhibit characteristic defects in overt morphology, efficacy was determined via heteroduplex analysis for all four lamb2 guides (gRNA1: 88%, n=30; gRNA2: 90%, n=20; gRNA3: 88%, n=30; gRNA4: 88%, n=30). However, in instances when mutagenesis was not observed at one gRNA target site, lesions were not observed at any of the four target sites in the same embryo, suggesting that the microinjection itself in those embryos was unsuccessful. Given that all four guides induced lesions in almost all injected embryos, successful mutagenesis by gRNA4 was used to genotype all subsequent experiments (efficacy=86%, n=137) and embryos in which lesions were not identified were discarded from analysis. Uninjected and gRNA-only injected controls were included in all heteroduplex analyses.

All MOs used are previously described: tp53-MO (Langheinrich et al., 2002), tnnt2a-MO (Sehnert et al., 2002) and gata1a-MO (Galloway et al., 2005). tp53 and tnnt2a MOs were purchased from GeneTools and resuspended in MilliQ to 1 mM. The following concentrations were used for knockdown: tp53 250 nM and tnnt2a 125 nM. The gata1a MO was a gift from J. Serbanovic-Canic (University of Sheffield, UK), and injected at 200 nM. tnnt2a or gata1a MOs were co-injected with the tp53 MO. Embryos were injected with 1 nl of MO solution into the yolk at the one-cell stage.

For chromogenic mRNA in situ hybridisation, embryos were fixed overnight in 4% paraformaldehyde (PFA, Cell Signalling Technology 12606); for fluorescence mRNA in situ hybridisation, embryos were fixed overnight in 4% PFA containing 4% sucrose. Following fixation, embryos were washed three times for 5 min each in PBS-Tween [PBST; 0.2% Tween (Sigma-Aldrich P2287) in PBS (Thermo-Fisher Scientific BR0014G)] and transferred into 100% MeOH for storage at −20°C. Chromogenic mRNA in situ hybridisation was carried out as previously described (Noël et al., 2013). Fluorescence in situ hybridisations were carried out using the PerkinElmer TSA kit (Welten et al., 2006). The fluorescence-labelled fli1 riboprobe was developed with Tyr-Cy5 (PerkinElmer NEL705A001KT), followed by the digoxigenin (DIG)-labelled gene of interest riboprobe developed with Tyr-Cy3 (PerkinElmer NEL704A001KT). Following probe signal amplification, embryos were fixed in 4% PFA with sucrose overnight and then washed into PBST for immunohistochemistry. The following previously published probes were used: lamb1b (Sztal et al., 2011), fli1 (Brown et al., 2000), myl7 (Yelon et al., 1999), myh6 (Derrick et al., 2021), myh7l (Derrick et al., 2021), aldh1a2 (Begemann and Meyer, 2001), spry4 (Fürthauer et al., 2001), ltbp3 (Zhou et al., 2011), klf2a (Novodvorsky et al., 2015) and hbbe1.1 (Quinkertz and Campos-Ortega, 1999). All other probes were generated for this study. Probe constructs were generated through PCR amplification of a DNA fragment from total zebrafish cDNA at 55 hpf, and ligated into either the pCRII-TOPO vector or pCR4-TOPO vector (Thermo Fisher Scientific 450640 and 450071). See Table S1 for probe primer and sequence details. Riboprobes were transcribed from a linearized template in the presence of DIG-11-UTP or Fluorescein-11-UTP (Roche).

Embryos were fixed overnight in 4% PFA containing 4% sucrose, washed three times for 5 min each in PBST and transferred into 100% MeOH for storage at −20°C. Embryos were rehydrated into PBST, washed briefly in PBST and twice for 5 min each in 0.2% Triton-X (Sigma-Aldrich T8787) in PBS (PBS-Triton). Embryos were incubated in blocking buffer [10% goat serum (Invitrogen 10000C) in PBS-Triton] at room temperature with gentle agitation for 1 h. Blocking buffer was removed and replaced with blocking buffer containing 1% DMSO and primary antibodies. Embryos were incubated overnight at 4°C with gentle agitation. Following removal of primary antibodies, embryos were extensively washed in PBS-Triton and incubated in blocking buffer containing 1% DMSO and secondary antibodies overnight at 4°C with gentle agitation. After removal of secondary antibodies, embryos were extensively washed in PBS-Triton at room temperature before being prepared for imaging. The following antibodies were used: chicken anti-GFP (1:500, Aves Labs GFP-1010), rabbit anti-DsRed (1:200, Takara Bio 632496), rabbit anti-PH3 (1:200, MerckMillipore 04-817), donkey anti-chicken-Cy2 (1:200, Jackson Immuno Research 703-225-155) and goat anti-rabbit-Cy3 (1:200, Jackson Immuno Research 111-165-003).

RA powder (Sigma-Aldrich R2625-50MG) was dissolved in DMSO (Sigma-Aldrich 276855) to a stock concentration of 10 mM, and aliquots were stored at −80C. Embryos were manually dechorionated prior to treatment, and ten lamb1a mutant and 13 lamb1a sibling embryos were placed in a glass petri dish. Stock RA was diluted 1:10,000 in E3-PTU to give a working concentration of 100 nM RA with 1% DMSO, and 8 ml was added to treatment dishes. Control embryos were incubated with either E3-PTU or E3-PTU with 1% DMSO. Embryos were incubated in RA or control medium from 24 hpf to 55 hpf, when the drug was removed by rinsing embryos three times for 5 min each in E3, and either fixed immediately or allowed to develop until 72 hpf and then fixed. Embryos were protected from light during the treatment window. Each RA treatment/control treatment was treated as one experimental unit for quantification, with an average value calculated from all embryos for each treatment. These treatment averages then formed one experimental replicate for statistical analyses, and treatments were replicated four times.

Prior to imaging at 2 dpf, embryos were sorted based on morphology into siblings and mutants. A pair of embryos (1 sibling, 1 mutant) was transferred in E3 medium from a 28.5°C incubator. A single embryo was positioned laterally on an agarose mould (2% agarose in E3) for imaging under a dissection microscope (11.5× magnification) attached to a high-speed camera (Chameleon3 USB3, FLIR Integrated Imaging Solutions) focussed on the heart. Image sequences (.tif) of 5 s were captured at 150 frames per second using SpinView Software (Spinnaker v. 2.0.0.147). This procedure was repeated for the remaining embryo of the pair and was then repeated at 3 dpf. Image sequences were imported into Fiji and converted to .avi movies. Movies were imported into, and heart rate quantified in, DanioScope (Noldus). Individual values represent an average heart rate over the 5 s imaging period.

Prior to quantification, files were blinded using an ImageJ Blind_Analysis plugin (modified from the Shuffler macro, v1.0 26/06/08, Christophe Leterrier, Aix-Marseille University, France). Looping ratio, heart area and chamber area were quantified as previously described (Derrick et al., 2021).

Total heart cardiomyocyte cell number and internuclear distance were quantified from Tg(myl7:DsRed) transgenic hearts. z-stacks of fixed hearts were imaged on a Nikon A1 confocal microscope, using a 40× objective with a z-resolution of 1 µm. The DsRed channel of each heart was used to generate a depth-coded z-projection of the z-stack, using the temporal colour code function in Fiji. Cell number in the atrium and ventricle were quantified from these z-projections. Internuclear distance was quantified by measuring the distance between DsRed+ nuclei with the same or similar depth coding in the projection. Six cells were selected per chamber and, from each cell, the distance to the four nearest neighbours with similar z positions was measured. Average internuclear distance was then calculated for each chamber in each embryo.

Atrial/venous pole SHF addition was quantified similar to previous methods (de Pater et al., 2009). Stacks were opened in Fiji and converted to maximum intensity projections. Using the DsRed channel only, the intensity was increased to maximum and the number of atrial DsRed+ nuclei was quantified using the ROI Manager in Fiji; the GFP channel was used to confirm position in the heart. Using the GFP channel only, the intensity was increased to maximum and GFP+ nuclei not previously counted in the ROI Manager were quantified as DsRed−. Returning to the original stack, individual slices were examined, together with the ROIs for DsRed+ and DsRed− atrial cells to ensure that no cells had been missed or miscounted. Cells derived from the ventricle or atrioventricular canal were discounted.

Ventricular/arterial pole SHF addition was quantified by determining the amount of GFP+ tissue at the arterial pole, distal to the last DsRed+ nucleus in the ventricle/OFT. Each sample z-stack was reoriented in Fiji to allow transverse reslicing into the arterial pole. Once the first DsRed+ cardiomyocyte was observed, all subsequent slices were discarded and the GFP channel selected, creating a small stack representing only the GFP+ SHF-derived component of the arterial pole. The 3D Object Counter Fiji plugin was used to threshold, identify and quantify the arterial pole SHF myocardium.

Statistical analyses of quantitative data were performed in Graphpad Prism 9. Data were first tested for normality using the D'Agostino-Pearson test to define whether a parametric or nonparametric test was required. Mean heart size, looping ratio or cell number were compared between two samples using an unpaired t-test or Mann Whitney U-test, assuming unequal variance. Means were compared between multiple samples using the Brown-Forsythe and Welch ANOVA, or Kruskal-Wallis with Dunn's mulitple comparisons test. Data were considered significant when P<0.05.

We thank Tanya Whitfield and Juliana Sánchez-Posada for comments on the manuscript. Additional imaging work was performed at the Wolfson Light Microscopy Facility using a Nikon A1 microscope. C.J.D. would like to thank Martin and Jane Derrick for financial support during 2020.

The peer review history is available online at https://journals.biologists.com/dev/article-lookup/doi/10.1242/dev.199691