Lamb1a regulates atrial growth by limiting excessive, contractility-dependent second heart field addition during zebrafish heart development

During early vertebrate heart development, the heart transitions from a linear tube to a complex asymmetric structure. This process includes looping of the tube and ballooning of the emerging cardiac chambers, which occur simultaneously with growth of the heart. A key driver of cardiac growth is deployment of cells from the Second Heart Field (SHF) into both poles of the heart, with cardiac morphogenesis and growth intimately linked in heart development. Laminin is a core component of extracellular matrix (ECM) basement membranes, and although mutations in specific laminin subunits are linked with a variety of cardiac abnormalities, including congenital heart disease and dilated cardiomyopathy, no role for laminin has been identified in early vertebrate heart morphogenesis. We identified dynamic, tissue-specific expression of laminin subunit genes in the developing zebrafish heart, supporting a role for laminins in heart morphogenesis. lamb1a mutants exhibit cardiomegaly from 2dpf onwards, with subsequent progressive defects in cardiac morphogenesis characterised by a failure of the chambers to compact around the developing atrioventricular canal. We show that loss of lamb1a results in excess addition of SHF cells to the atrium, revealing that Lamb1a functions to limit heart size during cardiac development by restricting SHF addition to the venous pole. lamb1a mutants exhibit hallmarks of altered haemodynamics, and specifically blocking cardiac contractility in lamb1a mutants rescues heart size and atrial SHF addition. Furthermore, we identify that FGF and RA signalling, two conserved pathways promoting SHF addition, are regulated by heart contractility and are dysregulated in lamb1a mutants, suggesting that laminin mediates interactions between SHF deployment, heart biomechanics, and biochemical signalling during heart development. Together, this describes the first requirement for laminins in early vertebrate heart morphogenesis, reinforcing the importance of specialised ECM composition in cardiac development.

2dpf onwards, with subsequent progressive defects in cardiac morphogenesis characterised by a failure 23 of the chambers to compact around the developing atrioventricular canal. We show that loss of lamb1a 24 results in excess addition of SHF cells to the atrium, revealing that Lamb1a functions to limit heart size 25 during cardiac development by restricting SHF addition to the venous pole. lamb1a mutants exhibit 26 hallmarks of altered haemodynamics, and specifically blocking cardiac contractility in lamb1a mutants 27 rescues heart size and atrial SHF addition. Furthermore, we identify that FGF and RA signalling, two 28 conserved pathways promoting SHF addition, are regulated by heart contractility and are dysregulated 29 in lamb1a mutants, suggesting that laminin mediates interactions between SHF deployment, heart 30 biomechanics, and biochemical signalling during heart development. Together, this describes the first 31 requirement for laminins in early vertebrate heart morphogenesis, reinforcing the importance of 32 specialised ECM composition in cardiac development. 33

Introduction 35
Tissue morphogenesis requires tight coordination of changes in cell shape and organisation, gene 36 expression, and tissue patterning, together with the integration of both intrinsic and extrinsic signalling 37 blood flow is also required for heart morphogenesis. The forces generated by blood flow results in 75 localised induction of transcription factors such as klf2a which functions in the endocardium to promote 76 regionalised cell shape changes [32] and proliferation associated with early valve development [24,25]. 77 Thus a complex interplay of biochemical and biomechanical cues are required for robust heart 78 morphogenesis during development. Here we identify tissue-specific expression of multiple laminin subunits in the developing zebrafish 125 heart during early cardiac morphogenesis. Through targeted mutagenesis of the highly conserved 126 laminin genes lamc1 and lamb1a, we identify two novel requirements for laminins during heart 127 development: promoting heart looping morphogenesis and restricting cardiac size. We show that 128 Lamb1a controls atrial growth by limiting SHF addition to the venous pole, and demonstrate that the 129 excessive atrial SHF addition in lamb1a mutants can be rescued by blocking heart contractility. Finally, 130 we demonstrate that loss of either contractility or lamb1a disrupts expression of FGF and RA-131 responsive genes in the heart, supporting a role for laminins in coupling mechanical force, intercellular 132 signalling, and cardiac growth. Together, this study presents the first reported role for laminins in the 133 early development of the vertebrate heart and highlights a crucial requirement for the ECM in 134 coordinating distinct aspects of tissue morphogenesis. 135 136

137
Laminins display dynamic, tissue-specific expression during early zebrafish heart morphogenesis 138 To investigate the role of laminin complexes in early vertebrate heart morphogenesis we probed a 139 previously-published transcriptomic analysis of cardiac gene expression to identify laminin subunit 140 genes expressed in the heart tube [52], and in combination with an in situ hybridization screen identified 141 a subset of laminin subunits with cardiac expression during early stages of heart looping (Fig 1,  142 Supplemental Fig S1). 143 Zebrafish share ten of the twelve mammalian laminin subunits, with no orthologs for lamb3 or lamc2 144 and possess two duplicated beta subunits -lamb2l and lamb1b [53]. At 30hpf (hours post fertilisation, 145 Fig 1A), during early heart tube morphogenesis, six laminin subunits are expressed in the zebrafish 146 heart: two alpha chains, lama4 and lama5, (Fig 1C,E); three beta chains, lamb1a, lamb1b, and lamb2 147 ( Fig 1G,I,K) and a single gamma chain: lamc1 (Fig 1M). Since specific laminin isoforms can exhibit 148 tissue-specific deposition, we carried out two colour fluorescent in situ hybridization at 30hpf to identify 149 whether the myocardium and endocardium have a specific laminin expression profile (Supplemental 150 Fig S1). This identified two endocardial laminin subunit genes: lama4 and lamb1b (Supplemental Fig  151   S1A-B), two myocardial laminins: lama5 and lamb2 (Supplemental Fig S1C-D) and two laminin 152 subunits that are expressed in both the myocardium and endocardium: lamb1a and lamc1 (Supplemental 153 Fig S1E-F). 154 While at 30hpf the majority of laminin genes are expressed along the length of the heart tube, following 155 initial heart looping morphogenesis at 55hpf (Fig 1B), the expression of most laminin subunits has 156 become restricted to the ventricle and atrioventricular canal (Fig 1D,F,H,L,N) with the exception of 157 lamb1b which is expressed only in the atrioventricular canal ( Fig 1J). This dynamic spatiotemporal 158 control of specific laminin subunit expression during early heart development suggests that individual 159 endocardial or myocardial-derived laminin complexes may drive early heart tube morphogenesis. 160 161 lamc1 and lamb1a regulate heart morphology and size during cardiac development 162 Having identified multiple potential laminin complexes expressed in the heart during early 163 morphogenesis, we wanted to examine the role of laminins during heart development. Laminins are a 164 heterotrimeric complex of a single alpha, beta and gamma chain ( Fig 1O) which are assembled in the 165 cell prior to deposition into the ECM, thus the removal of a single subunit is sufficient to prevent 166 secretion of the complex from the cell [54]. Therefore, to investigate the requirement for laminins during 167 heart looping morphogenesis we targeted the single gamma subunit lamc1 (the homolog of human 168 LAMC1), expressed in both the myocardium and endocardium (Fig 1M-N  tube by 26hpf indicating laminin is not required for initial heart tube formation. At 2dpf lamc1 mutants 177 exhibit mild pericardial oedema, suggesting possible defects in heart looping morphogenesis 178 (Supplemental figure S2D). We therefore assessed the impact of loss of Lamc1 function on heart 179 morphology by mRNA in situ hybridization analysis of myl7 expression (Fig 2A-D), quantifying 180 looping ratio and heart size at 55hpf and 72hpf after initial heart looping morphogenesis has occurred 181 heart looping morphogenesis (Fig 2A-B), displaying a significant reduction in heart looping ratio 183 compared to control embryos (Fig 2E), although the size of the heart appears relatively normal (Fig 2F). 184 By 72hpf, in addition to abnormal cardiac morphology ( Fig 2D, and Supplemental Fig S2E) lamc1 F0 185 crispant hearts also appear significantly larger than in control embryos ( Fig 2H, Supplemental Fig 2F). 186 This demonstrates that laminins promote initial heart looping morphogenesis and may subsequently 187 function to restrict cardiac size. 188 189 Distinct laminin complexes play varied yet specific roles in different developmental contexts [40]. Since 190 lamb1a (the homolog of human LAMB1) exhibits similar expression dynamics and tissue-specificity as 191 lamc1 (Fig 1G,H,M,N), this suggested that Lamb1a and Lamc1 may together form part of the specific 192 laminin complexes required for heart development. To confirm this role for lamb1a we generated two 193 stable mutant alleles lamb1a ∆19 and lamb1a ∆25 using CRISPR-Cas9 mediated genome editing 194 (Supplemental Fig S2G-M). At 55hpf lamb1a mutants display a much less severe heart looping defect 195 when compared to lamc1 crispants (Fig 2B,J), but instead present a mild increase in heart size ( Fig 2N,  and sleepy sa379 (sly, lamc1) mutants revealed identical phenotypes to our novel loss of function models 202 (Supplemental. Fig S3). Together, these data demonstrate two previously uncharacterised requirements 203 for laminins in vertebrate heart morphogenesis, where Lamc1-containing trimers promote heart looping 204 and restrict cardiac size, while Lamb1a-containing trimers primarily function to restrict heart size. 205

206
The difference in phenotypes between lamb1a and lamc1 mutants suggested other laminin beta subunits 207 may act during early heart morphogenesis, or functionally compensate for loss of lamb1a. We examined 208 the impact of loss of lamb1a on the expression of the two other identified laminin beta subunits (lamb1b 209 and lamb2) in the heart at 30hpf (Supplemental Fig S4A-D). At 30hpf a striking upregulation and 210 expansion of lamb1b expression was observed in lamb1a Δ25 mutants (Supplemental Fig S4A-B), whilst 211 lamb2 expression was unaffected (Supplemental Fig S4C-D). This suggested that lamb1b could 212 compensate for loss of lamb1a, resulting in the weaker looping morphogenesis phenotype in lamb1a 213 mutants when compared to loss of lamc1. To investigate this we used CRISPR-Cas9-mediated genome 214 editing to generate two lamb1b promoter deletion alleles: lamb1b ∆183 and lamb1b ∆428 (Supplemental 215 Fig S4E-F). We generated lamb1b;lamb1a double heterozygous adult fish, incrossed them to obtain 216 lamb1b;lamb1a Δ25 double mutant embryos and examined lamb1b expression at 30hpf by in situ 217 hybridization to confirm the absence of lamb1b transcript (Supplemental Fig S4G-J). Analysis of heart 218 size and morphology in lamb1b;lamb1a Δ25 double mutants at 55hpf revealed that loss of lamb1b did not 219 modify the lamb1a mutant phenotype, demonstrating that despite its upregulation in lamb1a Δ25 mutants, 220 does not compensate for loss of lamb1a (Supplemental Fig S4K-N). This further supports the hypothesis 221 that lamc1 and lamb1a play distinct roles in cardiac development. 222 223 lamb1a limits SHF addition to the venous pole 224 We have identified a requirement for laminins in driving looping morphogenesis and restricting organ 225 size during heart development (Fig 2). These two processes are uncoupled in the lamb1a mutant, an 226 interesting finding as often defects in heart size are coupled with a severe impact on looping 227 morphology, such as the loss of Cerebral Cavernous Malformation (CCM) pathway components where 228 cardiac chambers are larger but morphology is also severely disrupted [60-62]. The lamb1a mutant thus 229 represents an excellent model to investigate the mechanism by which laminins limit heart growth 230 independent of the morphogenesis of the tissue. not enlarged in lamb1a Δ25 mutants, we therefore hypothesised that cardiomegaly in lamb1a mutants 247 results from increased cell number, and quantified DsRed-positive cardiomyocytes in sibling and 248 lamb1a Δ25 mutant embryos at 55hpf and 72hpf. We observed no profound increase in cell number in the 249 ventricle at either stage; however, we did find a significant increase in the number of atrial cells in 250 lamb1a Δ25 mutant embryos at both 55hpf and 72hpf when compared to sibling embryos (Fig 3L,M). 251 Together this suggests that Lamb1a controls atrial size through regulating cell number. 252 253 During morphogenesis, the heart grows primarily through addition of cells to the poles of the heart from 254 the second heart field (SHF), a conserved population of cells situated in the mesoderm adjacent to the 255 heart. While previous studies have demonstrated that SHF addition to the arterial pole of the heart is 256 sensitive to perturbations in ECM composition [33], comparatively less is known about SHF addition 257 to the venous pole, and in particular how the ECM may regulate this process. Since lamb1a Δ25 mutants 258 exhibit increased atrial cell number in the heart during the window of SHF addition, we hypothesised 259 that loss of Lamb1a-containing laminin trimers leads to increased cardiac size through elevated SHF 260 addition. We visualised SHF addition in Tg(myl7:eGFP);Tg(myl7:DsRed) double transgenic sibling 261 and lamb1a Δ25 mutant embryos, where cardiomyocytes derived from the original linear heart tube/first 262 Excessive second heart field addition to the venous pole in lamb1a mutants is dependent on heart 281 contractility 282 Our finding that loss of lamb1a leads to increased SHF addition to the atrium without expansion of the 283 SHF domain suggested a SHF specification-independent mechanism of altered SHF addition. Our 284 previous findings that lamb1b upregulation in lamb1a mutants does not appear to be triggered by 285 compensatory pathways and does not play a functional role (Supplemental Fig S4) may provide clues 286 to the mechanisms underlying increased SHF addition in lamb1a mutants. 287 In the time frame in which heart morphogenesis and SHF addition occur, lamb1b is expressed 288 throughout the endocardium at 30hpf (Fig S1B) and by 55hpf is restricted to the atrioventricular canal, 289 the site of atrioventricular valve development ( Fig 1J). This expression domain and dynamic is highly 290 similar to other genes that are required for valvulogenesis such as notch1b, the expression of which is 291 regulated by sensation of blood flow [24,63-66]. Furthermore, fibronectin 1b, a key ECM component 292 required for valvulogenesis, is expressed in a similar domain to lamb1b at 55hpf and is also dependent 293 on haemodynamic forces [34]. We therefore hypothesized that lamb1b expression is also flow-294 dependent, and that the misexpression of lamb1b throughout the endocardium of lamb1a Δ25 mutants 295 may reflect changes in cardiac function or flow sensitivity upon loss of laminin. 296 To investigate this, embryos from a lamb1a Δ25 heterozygous incross were injected at the 1-cell stage 297 with a translation-blocking morpholino oligonucleotide (MO) targeting troponin T type 2a (cardiac) 298 (tnnt2a, previously silent heart) to block heart contractility and abolish blood flow [28]. Expression of 299 lamb1b was then examined at 30hpf (Fig 5A-F), when uninjected and control tp53 MO injected sibling 300 embryos have low levels of endocardial lamb1b expression (Fig 5A,B), and uninjected and control 301 injected lamb1a Δ25 mutants misexpress lamb1b throughout the endocardium (Fig 5D,E) as previously 302 described (Supplemental Fig 4). As expected, morpholino-mediated knockdown of tnnt2a in sibling 303 embryos results in a loss of lamb1b expression in the endocardium (Fig 5C). Similarly, loss of heart 304 contractility in lamb1a Δ25 mutants results in reduced endocardial expression of lamb1b compared to 305 uninjected and control mutants (Fig 5F). Together this demonstrates that endocardial expansion of 306 lamb1b in lamb1a Δ25 mutants is dependent on heart contractility. 307 308 To confirm altered flow-responsiveness in lamb1a Δ25 mutant endocardium, we analysed the expression 309 of klf2a, a key transcription factor whose expression is regulated by turbulent flow [24]. We injected 310 embryos from a lamb1a Δ25 heterozygous incross with tnnt2a MO and examined klf2a expression at 311 30hpf. In sibling uninjected and control embryos, klf2a expression is localised predominantly to the 312 arterial pole endocardium (Fig 5G, H), whereas uninjected and control injected lamb1a mutants 313 misexpress klf2a more broadly throughout the heart and also have prominent upregulation of klf2a in 314 the head (Fig 5J,K). Morpholino-mediated knockdown of tnnt2a in sibling embryos results in almost 315 total loss of klf2a expression in the endocardium compared to controls (Fig 5I). Similar to the effect on 316 lamb1b expression, loss of heart contractility in lamb1a Δ25 mutants results in reduced endocardial 317 expression of klf2a compared to control mutants ( Fig 5L). Together these data suggest that loss of 318 Lamb1a may result in perturbations to the response to heart contractility and/or blood flow during heart 319 looping morphogenesis. 320

321
The increase in SHF addition and altered expression of haemodynamic-responsive genes in lamb1a Δ25 322 mutants suggest that cardiac function may play a role in the failure to restrict heart size in lamb1a 323 mutants. We therefore investigated whether perturbing contractile and/or haemodynamic forces rescues 324 cardiomegaly in lamb1a Δ25 mutants by injecting the tnnt2a MO and measuring heart size (Fig 6A-D, 325 Fig S6A & E). Blocking cardiac contractility in lamb1a Δ25 mutant embryos significantly reduced heart 326 size at 55hpf and 72hpf compared to control lamb1a Δ25 mutants, suggesting that excess SHF addition is 327 mediated by contractility upon loss of Lamb1a (Fig 6E, Fig S6A). To confirm this, we then examined 328 the impact of loss of heart contractility specifically on SHF addition to the venous pole of the heart at 329 55hpf in Tg(myl7:eGFP);Tg(myl7:DsRed) transgenic embryos. In line with our previous data, 330 uninjected and control-injected lamb1a Δ25 mutant embryos display an increase in the number of newly-331 added SHF cells in the atrium at 55hpf (Fig 6K). However, in lamb1a Δ25 mutants injected with tnnt2a 332 MO, addition of SHF cells to the atrium is rescued to comparable levels with siblings ( Fig 6K). 333 Surprisingly, we also identified that loss of heart contractility in sibling embryos results in a subtle, yet 334 significant increase in the number of GFP+/DsRed+ cells in the atrium at 55hpf ( Figure 6J). This 335 suggests that more broadly, heart contractility may be required to limit the rate of SHF addition to the 336 atrium. Taken together, this demonstrates that Lamb1a is required to limit excessive, contractile-337 dependent SHF addition to the atrium during heart looping morphogenesis. 338

339
Since heart contractility drives excessive SHF addition in lamb1a Δ25 mutants, this could be due to 340 increased heart rate. Importantly, analysis of heart rate at 2dpf and 3dpf did not reveal any significant 341 differences between sibling and lamb1a Δ25 mutant embryos (Supplemental Fig S6B), suggesting 342 increased rate of heart contractility itself is not driving the aberrant contractility-mediated SHF addition. 343 Additionally, since we observed changes in the expression of flow-responsive genes in lamb1a Δ25 344 mutant embryos we wished to rule out that increased shear stress itself is responsible for cardiomegaly 345 in lamb1a mutants. We lowered blood viscosity by injecting embryos from an incross of lamb1a Δ25 346 heterozygous adults with a MO targeting the transcription factor gata1a, a master regulator of  analysed at 72hpf by myl7 expression. Quantification of heart area at 72hpf revealed that loss of Gata1a 351 function and reduction in blood viscosity does not rescue lamb1a Δ25 mutant heart size (Supplemental 352 Fig S6F), demonstrating that cardiomegaly in lamb1a mutants is not due to increased flow sensing. 353 Therefore, together our data demonstrates that loss of lamb1a results in excessive SHF addition to the 354 atrium through a contractile-dependent, shear stress-independent mechanism. 355 356 Retinoic Acid treatment during early SHF addition partially rescues cardiomegaly in lamb1a mutants 357 The molecular pathways underlying SHF patterning and addition are conserved among vertebrates, and 358 include balanced, antagonistic, FGF8 and RA signalling across the arterial-venous axis of the cardiac-359 forming region [14]. Furthermore, cardiac function has been implicated in regulating the expression of 360 aldh1a2 (formerly raldh2), a key enzyme in the RA synthetic pathway [70], suggesting cross-talk 361 between cardiac contractility and SHF patterning. We therefore examined the expression of aldh1a2 362 (RA synthesis and RA signalling target) and spry4 (FGF signalling responsive) in lamb1a Δ25 mutants 363 by mRNA in situ hybridization (Fig 7). At 30hpf and 55hpf, lamb1a Δ25 mutants exhibit a marked 364 upregulation of aldh1a2 expression (Fig 7A-D) together with increased expression of spry4 at 55hpf 365 with variable penetrance (Fig 7E,F). As aldh1a2 expression is sensitive to levels of RA in the embryo, 366 where low levels of RA result in increased aldh1a2 expression [71-73], this suggests that loss of 367 Lamb1a alters the balance of FGF-RA signalling in the embryo during the window of SHF addition. To 368 investigate whether the impact of loss of lamb1a on FGF and RA signalling is also contractility-369 dependent, we examined aldh1a2 and spry4 expression in sibling and lamb1a Δ25 mutant embryos 370 injected with tnnt2a MO. While uninjected and control injected lamb1a Δ25 mutant embryos have a clear 371 expansion of aldh1a2 (Fig 7J,K), MO-mediated knockdown of tnnt2a abrogates endocardial aldh1a2 372 expression in lamb1a Δ25 mutants (Fig 7L). Surprisingly, blocking contractility in either sibling or 373 lamb1a mutant embryos also resulted in altered spry4 expression, with an expansion of spry4 into the 374 atrium that was most prominent upon loss of contractility in the lamb1a Δ25 mutant (Fig 7M-R). Together 375 this suggests, not only that disruptions to FGF-RA signalling in lamb1a Δ25 mutants are partly regulated 376 by heart function, but also that cardiac contractility itself can influence activity of the pathways 377 regulating SHF addition. perturbed, this is not the only pathway driving increased heart size, and that heart function likely affects 395 SHF addition through additional mechanisms. 396 Together, we have shown the first requirement for laminins in regulating early vertebrate heart 397 morphogenesis, promoting heart morphology and restricting heart size through restriction of SHF 398 addition to the venous pole. Furthermore, our data indicate that the ECM and cardiac contractility 399 together function to regulate the balance of SHF-related signalling pathways. rescued by abolishing heart contractility, but not shear stress (Fig 6, Fig S6), suggests that laminin may 414 alter or dampen the physical force of heart contractility. lamb1a and lamc1 are expressed broadly 415 throughout the zebrafish embryo during the window of SHF migration, both within the heart and in the 416 surrounding tissues where the SHF resides (Fig 1). Therefore whether the role of Lamb1a in restricting 417 represent a conserved mechanism regulating SHF addition that is disrupted in lamb1a mutants. How 434 this is impacted by cardiac contraction is unclear, although it has been speculated that cardiac function 435 could contribute to SHF tension [22], and it is conceivable that cardiomyocyte contractility contributes 436 to the balance of pulling and pushing forces regulating SHF incorporation into the OFT in mouse [23]. 437 Importantly, while blocking contractility rescues excessive cell addition to the venous pole in lamb1a 438 mutants, morphology of the inflow tract appears impaired (Fig 6) we observed a mild upregulation of the FGF-response gene spry4 in the ventricle of lamb1a mutant 448 hearts at 55hpf (Fig 7). However, levels of FGF activity are also balanced by antagonistic RA signalling, 449 the activity of which is regulated by the family of Cyp26 enzymes. This balance is crucial, as 450 exemplified by the loss of ventricular SHF addition in Cyp26a1/Cyp26c1 deficient embryos, despite 451 SHF progenitors being correctly specified and maintained [16]. We observe an upregulation of the RA-452 responsive RA-synthesising enzyme aldh1a2 in lamb1a mutants as early as 30hpf (Fig 7), together 453 suggesting that RA signalling is impaired upon loss of laminin, leading to a dysregulation of FGF 454 activity. Providing some support for this hypothesis, timed RA treatments during early SHF addition 455 partially rescued the increased heart size of lamb1a mutants at 3dpf (Supplemental Fig S7), although 456 since the balance of RA-FGF is important, the global upregulation of RA is likely too broad to restore 457 this balance completely. RA signalling during early development has recently been proposed to define 458 the rate of cardiac progenitor differentiation in the anterior lateral plate mesoderm since disruption of 459 RA signalling from 6hpf onwards results in a reduction in ltbp3 expression and a loss of isl1a-positive 460 pacemaker cells at the inflow tract [99]. Importantly, we do not observe changes in the size of ltbp3 and 461 isl1a expression domains at either the venous or arterial pole respectively at 30hpf (Fig S5), suggesting 462 that altered FGF-RA signalling is not affecting the size of the SHF progenitor populations themselves 463 and that RA signalling in lamb1a mutant hearts is disrupted after SHF specification. Furthermore, 464 analysis of aldh1a2 expression in lamb1a mutants in which cardiac contractility has been abrogated 465 reveals that the aldh1a2 upregulation in lamb1a mutants is dependent on heart function. Surprisingly, 466 we also observe that loss of contractility disrupts the patterning of spry4 expression, similar to a study 467 demonstrating that muscle contractility regulates FGF signalling in the developing chick limb [100]. 468 This suggests that dysregulation of these pathways may be secondary to altered contractility in lamb1a 469 mutants, highlighting the complexity of interaction between the ECM, cardiac function, cell signalling, 470 and SHF addition. Alternatively, other pathways regulating timely SHF differentiation may be altered 471 in lamb1a mutants -for example lamc1 promotes the correct localization of HSPGs, which in turn 472 patterns BMP signalling during myotome development in zebrafish [101]. Together, a complex picture 473 is emerging around the mechanisms underlying SHF deployment during heart development, and how 474 the ECM regulates this process. While blocking contractility rescues excess SHF addition in lamb1a 475 mutants, loss of contractility in a wild type embryo does not reduce cardiomyocyte or SHF number (Fig  476   6), suggesting that under wild type conditions heart function is not required for SHF addition to the 477 atrium. However, we identify a mild but significant increase in the number of 'older' cardiomyocytes 478 in the atrium of tnnt2a morphants, suggesting that contractility could play a role in the timing of SHF 479 addition. 480

481
The tissue-restricted expression of individual laminin subunits we have identified during early heart 482 morphogenesis suggests that different cell types in the heart may deposit specific laminin trimers which 483 play distinct roles in heart morphogenesis and growth. We have identified multiple requirements for 484 more broadly-expressed beta and gamma subunits during heart development, however our identification 485 of tissue restricted alpha subunit expression (Figure 1 and S1) suggests that these alpha subunits may 486 confer tissue-specificity to the multiple roles laminins play in the heart. Laminin deposition into the 487 ECM by either myocardial or endocardial cell types likely facilitates signalling between these two tissue heart, correlating with increased tissue stiffness and a dramatic decline in regenerative potential [106]. 513 Our data demonstrating that lamb1a limits cardiomyocyte migration into the heart provides further 514 avenues for investigation into how targeted modulation of the cardiac ECM can promote cardiomyocyte 515 recruitment to the injury site, and improve regenerative potential. 516 517 Together, we describe the first direct evidence that laminins promote morphogenesis and growth during 518 early vertebrate heart development, uncovering a novel role for laminin in restricting contractility-519 dependent SHF addition to the venous pole. This work also identifies new links between ECM 520 composition, mechanical and biochemical cues in shaping the heart, reinforcing the importance of the 521 extracellular environment during organ morphogenesis. 522 523

Imaging and image quantification 617
Prior to quantification files were blinded using an ImageJ Blind_Analysis plugin (modified from the 618 Shuffler macro, v1.0 26/06/08, Christophe LeTerrier). Looping ratio, heart area and chamber area were 619 quantified as previously described [52]. 620 Total heart cardiomyocyte cell number and internuclear distance was quantified from Tg(myl7:DsRed) 621 transgenic hearts. Z-stacks of fixed hearts were imaged on a Nikon A1 confocal, using a 40x objective 622 with a z-resolution of 1µM. The DsRed channel of each heart was used to generate a depth-coded z-623 projection of the z-stack, using the temporal colour code function in Fiji. Cell number in the atrium and 624 ventricle were quantified from these z-projections. Internuclear distance was quantified by measuring 625 the distance between DsRed+ nuclei with the same or similar depth-coding in the projection. Six cells 626 were selected per chamber, and from each cell the distance to the four nearest neighbours with similar 627 z positions was measured. Average internuclear distance was then calculated for each chamber in each 628

embryo. 629
Atrial second heart field addition was quantified similar to previous methods [13]. Stacks were opened 630 in Fiji and converted to Maximum Intensity Projections. Using the DsRed channel only, the intensity 631 was increased to maximum and the number of atrial DsRed+ nuclei were quantified using the ROI 632 Manager, the GFP channel was used to confirm position in the heart. Using the GFP channel only, the 633 intensity was increased to maximum and GFP+ nuclei not previously counted in the ROI Manager were 634  injected controls (55hpf: n=44; 72hpf: n=44) and lamc1 F0 crispants (55hpf: n=47; 72hpf: n=44). lamc1 662 crispants exhibit reduced heart looping at 55hpf and 72hpf, and an increased area of myl7 expression at 663 72hpf. Horizontal bars indicate median with interquartile range, comparative statistics performed using 664 Kruskal-Wallis test. I-L: mRNA in situ hybridization analysis of myl7 expression in sibling (I,K) and 665 lamb1a Δ25 mutants (J,L) at 55hpf and 72hpf. M-P: Quantitative analysis of looping ratio (M,O) and myl7 666 area (N,P) in sibling (55hpf: n=70; 72hpf: n=56) and lamb1a Δ25 mutant embryos (55hpf: n=25; 72hpf: 667 n=34). lamb1a Δ25 mutants exhibit a mild reduction in heart looping at 55hpf, and an increased area of 668 myl7 expression at 55hpf and 72hpf. Scale bars: 50μm. Comparative statistics performed using Mann 669 Whitney test, **** = p < 0.0001, *** = p < 0.001, ** = p < 0.01, * = p< 0.05, ns = not significant.