Absence of heartbeat in the Xenopus tropicalis mutation muzak is caused by a nonsense mutation in cardiac myosin myh6

Mechanisms coupling heart function and cardiac morphogenesis can be accessed in lower vertebrate embryos that can survive to swimming tadpole stages on diffused oxygen. Forward genetic screens in Xenopus tropicalis have identified more than 80 mutations affecting diverse developmental processes, including cardiac morphogenesis and function. In the first positional cloning of a mutation in X. tropicalis, we show that non-contractile hearts in muzak (muz) embryos are caused by a premature stop codon in the cardiac myosin heavy chain gene myh6. The mutation deletes the coiled-coil domain responsible for polymerization into thick filaments, severely disrupting the cardiomyocyte cytoskeleton. Despite the lack of contractile activity and absence of a major structural protein, early stages of cardiac morphogenesis including looping and chamber formation are grossly normal. Muz hearts subsequently develop dilated chambers with compressed endocardium and fail to form identifiable cardiac valves and trabeculae.


Introduction
Formation of the heart is highly conserved in vertebrate species. Genes relevant to human cardiac development and disease can be studied in lower vertebrate models whose externally-developing embryos are easily accessible during heart forming stages and survive for several days on passively-diffused oxygen if cardiac function is compromised experimentally. Xenopus researchers have combined classical embryological explant and transplant approaches with over-and mis-expression of gene products (Warkman and Krieg, Open Access under CC BY 3.0 license. 2007) to examine early steps in heart formation, including specification of the heart field (Sater and Jacobson, 1989), transcriptional regulation of cardiac identity (Evans et al., 1995;Fu et al., 1998;Grow and Krieg, 1998), and signaling pathways underlying cardiac asymmetry (Branford et al., 2000;Hyatt and Yost, 1998;Ramsdell and Yost, 1999). In zebrafish, heart development studies have built on loss-of-function genetic tools, as well as the optical properties of the embryos for microscopy, to analyze cardiac morphogenesis and valve formation (Beis et al., 2005;Stainier, 2001). As teleost fish are the most diverse vertebrates, due in part to the ancestral genome duplication and subsequent shuffling of gene functions (Force et al., 1999;Postlethwait et al., 2000), comparative studies in other models will help identify developmental mechanisms shared broadly among tetrapods. Loss-of-function studies in X. laevis have previously been limited to injection of dominant negative constructs (Grow and Krieg, 1998;Shi et al., 2000) and, more recently, antisense morpholino oligonucleotides (Peterkin et al., 2007;Small et al., 2005). Large-scale genetic approaches are impractical in X. laevis due to its pseudotetraploid genome and long generation time, but are well-suited to its diploid relative Xenopus tropicalis. X. tropicalis reaches maturity in a relatively short 4-6 months, and its small, canonically-organized tetrapod genome (1.5×10 9 bp in 10 chromosomes) is supported by extensive sequence resources including a high-quality draft genome assembly (http:// genome.jgi-psf.org/Xentr4/Xentr4.home.html), over one million ESTs, and a meiotic linkage map of Simple Sequence Length Polymorphisms (SSLPs) (http://tropmap.biology.uh.edu/ index.html) Klein et al., 2006;Klein et al., 2002).
In a pilot screen for chemically-induced mutations in X. tropicalis, we recovered several phenotypes with decreased cardiac function (Goda et al., 2006). Here we show that the lack of cardiac contractility in the muzak mutant is caused by a nonsense mutation truncating the cardiac myosin heavy chain gene myh6. Despite this defect in a major structural component of sarcomeres resulting in absence of myofibrils and contractility, looping and chamber formation appear surprisingly normal. Muz hearts subsequently display dilated ventricles and atria and malformed endocardium, segments of which appear collapsed with little or no lumen. Later steps in cardiac development, such as valve formation and trabeculation, are not detected, but it is beyond the scope of this study to determine whether these are direct or indirect effects of the mutation. This report describes the first positional cloning of a mutation in X. tropicalis.

Frog Strains
The original mutagenesis and fertilization to produce mutant founder F1 animals was performed on the N (Nigerian) strain (kind gift of Enrique Amaya, Manchester University, United Kingdom); polymorphic crosses used for mapping were generated using the IC (Ivory Coast) strain (kind gift of Robert Grainger, University of Virginia, Charlottesville, USA). Mutant and wt embryos used for mapping and phenotyping were generated from a cross of an F2 muz/+ N/IC female and an F3 muz/+ male produced by crossing an F2 N/IC female to an N/PacBio (wild-caught animals of unknown origin obtained from Pacific Biological Supply, Inc.) male carrying the mlc2GFP transgene.

Mapping
Gynogenesis was performed as described previously (Goda et al., 2006). AFLP reactions were performed using the AFLP Analysis System I kit (Invitrogen, 10544-013). PCR products were resolved on 6% denaturing acrylamide gels and visualized by autoradiography. SSLP markers were amplified and resolved as described on the tropmap website (http://tropmap.biology.uh.edu/polyprotocol.html). SSLP markers from the meiotic map 040E09, 018E09, and 026G09 can be found on the tropmap database (http:// tropmap.biology.uh.edu/) and have the following sequences:

Whole Mount In Situ Hybridization
Embryos were staged according to Nieuwkoop and Faber. Fixing and WISH were carried out as described previously (Sive et al., 2000).
WISH probes for myh6 and myh6.2 were made by cloning RT-PCR products into the PCRII-TOPO vector using the TOPO TA Cloning Kit (Invitrogen, K4600-40). Probes were prepared by linearizing with XhoI and transcribing with SP6. Primers used were:

RT-PCR
RNA was prepared using Trizol (Invitrogen). cDNA was prepared and amplified with the Enhanced Avian HS RT-PCR kit (Sigma HSRT-100) using the following primers:

Immunoblotting
Dissected hearts from st. 40 tadpoles were collected on ice, resuspended in a modified SDSsample buffer, boiled for 1 minute, resolved by 6% PAGE, transferred to membrane, and immunoblotted as described previously (Ehler et al., 1999)

Silver Staining
Silver staining of proteins on SDS-PAGE gels was performed according to manufacturer's instructions using the Silver Stain Plus Kit (Bio-Rad, 161-0449)

Morpholino injections
Morpholinos were purchased from GeneTools LLC. A total of 12ng of each morpholino was injected into both cells of a two-cell embryo. Morpholino sequences were as follows:

Immunohistochemistry
Stage 42 tadpoles were fixed in 1% paraformaldehyde for 1 hour, washed in PBS, blocked in PBS+10% sheep serum, 2mg/ml BSA and 0.2% saponin for 1 hour at room temperature (RT), then incubated with primary antibody in block solution at 4°C overnight, washed in PBS containing 0.2% saponin and incubated in block solution containing Alexa Fluor 488conjugated anti-mouse IgG secondary antibody (Invitrogen, A21202) for 2 hours at RT. After washing in PBS with 0.2% saponin, the tadpoles were incubated with 1:20 dilution of Alexa Fluor 568 phalloidin (Invitrogen, A12380) in block solution, washed again, then hearts were dissected and visualized with a Zeiss LSM5 Pascal confocal microscope.

Plastic Sections and 3-D modeling
Embryos were fixed o/n in Bouin's fixative (BDH Laboratory Supplies,28087 4V), dehydrated in ethanol, embedded in JB-4 resin (Polysciences Inc.), 3μm sections cut with a Leica RM 2165 microtome, and stained with Hematoxylin and Eosin (both Sigma). Sections were visualized on a Zeiss axiocam microscope, serial images were converted into 8bit greyscale stacks and loaded in Amira 3D Visualisation software Mercury Computer Systems, Germany) and heart structures were manually outlined and annotated. 3D models were generated using the surface rendition tool in Amira.

Results
The muzak mutation affects heart function Homozygous muz embryos were identified by complete lack of cardiac contractility at heart looping stages (Movie S1). Embryonic blood fails to circulate in muz tadpoles, and erythrocytes pool in the ventral blood islands where they form. The tadpoles swim normally, indicating that the mutation does not affect skeletal muscle, and other tissues are not visibly affected. By stage 43 (3 days post fertilization), muz embryos develop cardiac edema, and absence of heart function persists until at least feeding tadpole stage (5 days post fertilization). No phenotype was observed in heterozygotes, suggesting that the muz allele behaves in a simple recessive fashion.

Muz maps to an interval containing cardiac myosin heavy chain gene
When we began linkage studies to identify the gene underlying the muz phenotype, no meiotic map was available. In a map-independent initial strategy, bulk segregant pools of DNA from gynogenetic muz and wild type siblings were used to obtain a set of Amplified Fragment Length Polymorphism (AFLP (Vos et al., 1995) markers linked to the mutant locus. 5 bands which amplified from wild type but not muz DNA ( Figure S1A) were extracted, reamplified, sequenced, and placed on the X. tropicalis genome assembly in Version 4 scaffolds 554, 91, 567, 289, 158 (http://genome.jgi-psf.org/Xentr4/ Xentr4.home.html). The subsequent release of an X. tropicalis meiotic map of SSLP markers (http://tropmap.biology.uh.edu) located these scaffolds in a ~12 cM interval on Linkage Group 1 (LG1). Linkage of the mutation to SSLP markers in these scaffolds was confirmed by bulk segregant analysis of pools of mutant and wild type embryos from a conventional cross of heterozygous carrier siblings (see Figure S1B for an example).
To define the genetic interval containing the muz locus, individual muz embryos from a conventional sibling cross were genotyped with SSLP markers from LG1 of the meiotic map. Analysis of 3200 meioses placed muz between two flanking markers, 040E09 in scaffold 91 (40 recombination events, Figure 1A) and 018E09 in scaffold 554 (77 recombination events). We tested the set of recombinant embryo DNAs further with a marker between the flanking markers, 026G09 (scaffold 256), and found a subset of the recombinants with 018E09 were still recombinant with this polymorphism, whereas all the recombinants with 040E09 were homozygous for the wild type 026G09 allele, suggesting that muz was located between the latter two markers. As the X. tropicalis genomic sequence assembly was fragmented in this region, and many scaffolds are not represented on the meiotic linkage map, we compared syntenic regions in well-characterized mammalian genomes to generate an in silico hypothetical local scaffold assembly. By examining syntenic human and mouse genomic regions that overlapped the termini of scaffolds 256 and 91, we identified candidate intervening scaffolds 439, 792 and 972 in the muz interval.
Analysis of SSLP markers 1.439.1 (two recombination events), 1.439.3 (no recombination events) and 1.439.10 (1 recombination event) confirmed this local assembly and placed the mutation in scaffold 439. Further analysis refined the muz interval to a 370kb region between Single Strand Conformation Polymorphism (SSCP) markers SSCP439.1 (two recombination events) and SSCP439.15 (one recombination event) on scaffold 439 containing 12 gene models on the JGI assembly ( Figure 1A and Table S1). The sequence interval containing muz was then inspected for candidate genes.
Compellingly, two gene models in this interval, myh6 and myh6.2, were annotated as myosin heavy chain (MHC), with >88% identity to the human cardiac MYH6 and MYH7 proteins, the major MHC genes expressed in mammalian hearts. These genes are known to be required for normal heart function in humans, with mutations in MYH6 and MYH7 implicated in atrial-septal defects and familial hypertrophic cardiomyopathies respectively (Ching et al., 2005;Geisterfer-Lowrance et al., 1990). In human, mouse, and rat these gene pairs are chromosomally adjacent, and are thought to have arisen by tandem duplication before these species diverged, some 70 million years ago (Mahdavi et al., 1984;Mahdavi et al., 1982). Of the two X. tropicalis MHC genes on scaffold 439, the centromere-proximal is orthologous to MYH6 based on mutual best BLAST as well as its strong expression in wild type hearts (Figure 2A, black arrow); weaker expression is also seen in jaw muscles ( Figure  2A, white arrow). The distal gene, annotated myh6.2, is expressed in developing jaw muscle but not heart ( Figure 2C), and hence is unlikely to be responsible for the muz phenotype.
To assess whether a defect in myh6 might underlie the muz phenotype, we sequenced cDNA from mutant and unrelated wild type embryos, and found a C to T transition creating a premature stop codon at position 3187 of the coding sequence. Genomic DNA from adult muz carrier animals was also found to be heterozygous for this lesion. The resulting truncated protein (1062 aa vs 1996 aa wild type, Figure 2E) is likely to be nonfunctional as it deletes the coiled-coil tail required for dimerization and aggregation into functional thick filaments.

Myh6 expression is strongly reduced in muzak hearts
We then evaluated how the mutation affected expression of the two MHC genes in the interval. Whole Mount In Situ Hybridization (WISH) showed a significant decrease in myh6 expression in muz embryos compared to wild type ( Figure 2A, B), possibly due to nonsensemediated decay (Peltz et al., 1993;Whitfield et al., 1994). Expression of the neighboring paralog myh6.2 in jaw muscle was unaffected by the mutation (Figure 2C, D, black arrow).
Levels of cardiac MHC protein were assayed by immunoblotting with the A4.1025 antibody, which recognizes an epitope shared by sarcomeric myosin heavy chain head domains (Dan-Goor et al., 1990) retained in the muz allele. A band of ~220kDa is observed in extracts of dissected wild type but not muz hearts ( Figure 2F). The mutant protein of predicted size ~120kDa is not detected, possibly due to depletion of the mRNA by nonsense-mediated decay, as suggested by WISH. Given the deletion of the tail domain required for thick filament formation and the severe reduction in expression levels, muz is likely to be a strong hypomorph or null allele of myh6.

Myh6 antisense morpholinos phenocopy the muz mutation
To confirm that a defect in myh6 could produce the muz cardiac phenotype, we designed morpholino antisense oligonucleotides to deplete the endogenous protein. Both translationblocking and splice-blocking morpholinos, when injected into both blastomeres of a two-cell embryo, affected cardiac contractility with high penetrance (76/79 and 94/100 injected embryos respectively). In contrast, heart looping and chamber formation were unaffected. Approximately 50% of myh6-depleted embryos had no detectable heartbeat, mimicking the muzak phenotype, while the remainder exhibited faint twitching insufficient for blood circulation (Movie S2). Injected embryos were otherwise morphologically normal, with tadpole motility unaffected, indicating that the morpholinos did not interfere with off-target skeletal MHCs. Control morpholino injections had no effect on cardiac function (85/85 wild type). Knockdown efficacy was assayed by immunoblotting protein extracts from dissected morphant hearts with the A4.1025 antibody. Both myh6 morpholinos strongly depleted cardiac MHC compared to control morpholino ( Figure 2G). These gene knockdown data confirm a requirement for myh6 in cardiac function, strongly supporting the conclusion that a defect in this gene underlies the muz phenotype.

Myh6 is the major cardiac sarcomeric MHC at swimming tadpole stages and is necessary for myofibril formation
Myh6 is likely to be the principal functional sarcomeric MHC in tadpole hearts, based both on the failure of the A4.1025 antibody to detect any immunoreactive species in muz heart extracts and the penetrance of the morphant phenotype. However, since the antibody may not recognize all Xenopus MHC proteins, and some morpholino-injected embryos retained faint twitching, we asked whether other sarcomeric MHC mRNAs were expressed in stage 40 hearts or upregulated in muz. RT-PCR of dissected stage 40 hearts confirms that myh6 is expressed strongly in wild type and at much reduced levels in muz hearts ( Figure 3A). Myh6.2 was amplified from stage 40 whole embryo mRNA, consistent with its expression in jaw muscle, but not from wild type or muz embryonic hearts, nor from adult heart ( Figure  3B). The Xenopus genome is not thought to contain an ortholog of mammalian MYH7 (Garriock et al., 2005), and it is likely that myh6.2 derives from a separate tandem duplication from the one which gave rise to mammalian MYH6 and MYH7. A third cardiac MHC, myh15/vMHC (an inactive pseudogene in human), has been found in chicken (Oana et al., 1998), as well as X. laevis (Garriock et al., 2005) where it is not expressed until after chamber formation. We found no myh15/vMHC expression in hearts of either wild type or muz stage 40 embryos by RT-PCR, although it is detected in adult heart ( Figure 3B), consistent with previously described onset of expression in X. laevis at stage 43. Similarly, no expression in embryonic heart was observed for the skeletal MHCs myh1,2,3 or 4 (data not shown). However, two MHCs present in mammalian heart EST collections, the slowtonic myh7B and myh8, were detected at comparable levels in both wild type and muz dissected hearts ( Figure 3A). Absence of myh6 protein in muz does not appear to induce expression of non-cardiac MHCs or up-regulate myh7B and myh8 mRNAs which, although present in muz hearts, are not sufficient to rescue the phenotype.
We then examined sarcomere formation in muz to see whether the remaining myh7B and myh8 could organize myofibrillar structures. Stage 42 wild type and muz embryos were stained with the A4.1025 antibody, counterstained with phalloidin, and their hearts dissected and visualized by confocal microscopy. Consistent with the depletion of myh6 mRNA and protein levels, anti-MHC immunostaining is greatly diminished in muz and is not organized in striated myofibrils (Figure 4). Significantly, phalloidin staining shows that actin does not form myofibrils in the mutant heart confirming the lack of any cardiac MHC proteins capable of assembling into sarcomeres in muz embryos. Myofibrils were absent at stage 35 (data not shown) when contractions begin as well as stage 40, making it unlikely that muz hearts have sarcomeres at any stage of development.

Cardiac chamber morphology and valve development in muzak
In addition to depletion of the myh6 protein, a major structural component of myocardial cells, the muz mutation results in abrogation of contractile activity (thought to be required for various steps in cardiac morphogenesis, as well as loss of sarcomeres (known to play

Author Manuscript
Author Manuscript

Author Manuscript
Author Manuscript signaling as well as mechanical roles in cardiac function (Nicol et al., 2000)). We wished to describe how these deficits affect the major morphogenetic steps in heart development.
As in other vertebrates, the Xenopus heart initially forms as a linear cardiac tube comprising a muscular myocardial layer surrounding an inner endocardial channel. After undergoing rightward looping, this tube balloons out into chambers separated by cardiac valves. The final stages of heart development in Xenopus include trabeculation of the ventricular myocardium and septation of the atrium into two chambers (Kolker et al., 2000;Mohun et al., 2000). To characterize how these processes are affected by absence of myh6 and the resulting lack of sarcomeres and contractility, muz hearts were subjected to histological analysis. Plastic sections of the cardiac region of wild type and mutant embryos were obtained at stages relevant to specific tissue formation processes: stage 35 (heart looping), 40 (onset of chamber formation), and 42 (valve formation). Figure 5 shows stage 40 wild type and muz hearts; sections are numbered to indicate their position in the stack beginning at the ventral side of the cardiac cavity. Outlines of myocardial and endocardial layers in the image stacks were then used to generate 3D projections ( Figure 5A-H, see also movies S3, S4 for a rotating view). Regions of the heart are indicated by colour: the myocardium of the outflow tract (blue) was defined by morphological position; thinner myocardium in dorsal sections (green) forms the atrium; thicker myocardium (red) in ventral sections is clearly ventricular (e.g. Figure 5 section 32 'v'); however, in malformed mutant hearts, where atrial and ventricular chambers showed little difference in wall thickness, the precise border was assigned arbitrarily.
Cardiac chambers in muz are dilated, and at stage 40 the myocardial wall appears thinner than wild type throughout. Segments of the endocardial tube, notably in outflow tract and atrioventricular canal (AVC), appear constricted with little lumen (white arrowheads, section 23 and F, H). The expanded 'peri-endocardial' region between the distended myocardium and the constricted endocardium distorts their alignment (white arrowhead, section 14). The cardiac tube at AVC level, spanned by black arrowheads in sections 27 and 23, is narrower in the mutant (black arrowhead in C and G). Dorsally the muz atrium is usually distended (white arrowhead, section 41). No blood cells are seen in muz hearts at this stage due to lack of circulation. Many of these abnormalities are already present prior to chamber differentiation in earlier looped cardiac tube (stage 35) muz tadpoles, including the dilated outflow tract, collapsed endocardial tube, and the narrow cardiac tube at the level of the AVC ( Figure S2).
At slightly later stages, valve formation begins in Xenopus; this process is not thought to occur in the absence of contraction in zebrafish (Bartman et al., 2004). We therefore examined plastic sections of stage 42 wild type and muz tadpoles ( Figure 6). A spiral valve can be distinguished in the outflow tract of wild type embryos (black arrowhead, sections 14 and 23), and the 'endocardial cushion' valve precursors are forming in the AVC (black asterisks, section 23). In stage 42 muz hearts, as at earlier stages, the endocardial tube is often narrower (white arrowheads, sections 54, 58 and F, H) and no valve formation can be discerned. Transverse sections more clearly show endocardial cushions forming in the AVC region of wild type ( Figure 6I, white arrowhead) but not muz hearts ( Figure 6J). Since it is difficult to unambiguously identify valve-forming AVC and outflow tract positions in the morphologically-distorted mutant hearts, we have also examined complete stacks of cardiaclevel sections from 10 muz embryos without detecting identifiable cushions at any position (data not shown). Endocardial cushions were clearly visible in 10/10 sibling wild type embryos.
Another important process, trabeculation, in which the ventricular myocardium takes on a spongiform appearance, is also occurring at this stage. In wild type hearts, myocardial cells can be seen proliferating and protruding into the lumen ( Figure 6I, black arrowheads); interestingly, these cells also take on a vacuolated appearance that may be integral to the mechanism of trabeculation (black arrows). No trabeculae are seen in muz ventricular myocardium, which is very thick but retains abundant vacuole-like structures similar to wild type ( Figure 6J, black arrows).
3-D modeling at stage 42 reveals that mutant cardiac morphology is becoming progressively more distorted (Figure 6, see also movies S5, S6 for a rotating view). Whereas in wild type the outflow tract rises sharply out of the ventricle towards the dorsal side of the embryo ( Figure 6 section 14 and A), in muz hearts it gently loops out of the end of the elongated ventricle ( Figure 6, 34 and E). The narrow cardiac tube at AVC level seen at stages 35 and 40 becomes more pronounced at stage 42 (Black arrowhead, G). Again, blood cells are absent in the muz heart except for a few in the atrium and inflow tract (Figure 6 section 58). The qualitative morphological abnormalities described here are consistently present in muz embryos at stages 40-42 (>10 mutant and wild type hearts examined in plastic sections, and 6 mutant and wild type examined by High Resolution Episcopic Microscopy (HREM)(data not shown)).
Analysis of histological sections of muz hearts demonstrates that later steps in heart development such as valve formation and trabeculation do not occur in the absence of myh6/ contractility and sarcomeres. The morphology of heart chambers is altered; dilated ventricles and atria are observed as early as stage 35, and become progressively more pronounced. The endocardium is likewise severely malformed, with segments of lumen highly constricted. It is beyond the scope of this analysis to conclude that these late effects are direct consequences of the mutation in myh6. However, early steps in cardiogenesis, such as looping and chamber formation, are relatively unaffected by absence of contractility and blood flow.

Discussion
The mapping of muzak marks the first identification of a sequence lesion underlying an induced mutation in X. tropicalis, an important step in establishing this species as a genetic model organism. The non-contractile heart phenotype is tightly linked to a nonsense mutation in the myh6 gene deleting the coiled-coil tail domain required for aggregation into functional thick filaments. This nonfunctional peptide, associated with severe reduction of mRNA and absence of detectable MHC protein and myofibrils, suggests that the muz allele is a strong hypomorph or null of myh6. Loss-of-function studies in Xenopus have previously been limited to morpholino knockdown and dominant negative strategies, where it can be difficult to obtain reproducible and complete deletions of specific activities. Precision loss- of-function tools are available in genetic systems such as mice and zebrafish. However, mutational analysis of cardiac development can be challenging in mammals, where heart function is required early in gestation; indeed, the null phenotype of mouse Myh6 has not been characterized due to early lethality (Jones et al., 1996). Genetic screens in fish have uncovered a large number of cardiac gene functions, but the basic structure of the twochambered fish heart differs significantly from the four-chambered mammalian heart. The ancestral teleost genome duplication has also led to wholesale reassignment and shuffling of gene functions (Force et al., 1999;Postlethwait et al., 2000), complicating orthology assignment and contributing to the diversity of developmental mechanisms. For example, zebrafish cardiac valves are thought to form by an atypical direct invagination of endocardial epithelia into leaflet structures (Scherz et al., 2008) rather than via a mesenchymal 'endocardial cushion' intermediate as has been described in other vertebrates (Armstrong and Bischoff, 2004;Eisenberg and Markwald, 1995) and indeed other fish (Gallego et al., 1997;Icardo et al., 2004). Genetic analysis of X. tropicalis, with its more conventionallyorganized tetrapod genome and array of functional assays, will help bridge studies of cardiac development from teleost models to amniotes.
In muzak embryos, the early processes of heart looping and chamber formation are remarkably successful despite the lack of myh6 protein and consequent absence of myofilaments, sarcomeres, heartbeat and blood flow. We have not ascertained which of these deficits is responsible for the later defects observed in chamber morphology, valve formation, and trabeculation, or whether these are direct or indirect consequences of the mutation. However, it is worth noting that mutant hearts never initiate detectable contraction and beating, and hence develop in the complete absence of blood flow-mediated pressure load and shear stress. The role of mechanical forces in cardiac morphogenesis has been studied extensively, with conflicting results (Taber, 2006). In diverse vertebrates, beating begins substantially prior to requirements for transport of blood-borne oxygen and nutrients, consistent with a role as a physical influence on early steps such as looping and chamber formation (Burggren et al., 2000;Mellish, 1994;Pelster and Burggren, 1996;Territo and Burggren, 1998); indeed, heart looping begins when the first myofibrils appear (Manasek et al., 1978). Mechanical or genetic perturbation of contraction and blood flow have supported a role in these early steps in some cases (Hove et al., 2003;Huang et al., 2003;Nishii et al., 2008), but not in others . Our histological analysis and 3-D modelling of muz hearts demonstrates that contractility and blood flow are not required for the key early steps of looping and chamber formation in this tetrapod.
Slightly later in heart development, chamber outgrowth or 'ballooning' is thought to be shaped by mechanical forces. Analysis of the chamber-specific MHC mutations weak atrium (atrial MHC, myh6) and half hearted (ventricular MHC, vmhc) show that blood flow promotes cardiomyocyte elongation in specific regions of the linear heart tube in the zebrafish embryo, while contractility restricts cell size and elongation (Auman et al., 2007). The muzak cardiac tube still undergoes ballooning into ventricle and atrium, suggesting that factors other than fluid shear forces can initiate chamber outgrowth. Another striking feature of muz hearts is the constriction of the lumen seen in the atrioventricular canal and outflow tract segments of the endocardial tube. The developing heart has been compared to a specialized blood vessel; arteries are thought to remodel their lumen diameters to maintain shear stress near an optimal set point, decreasing diameter in response to decreased shear (Taber et al., 1995). It is possible that morphogenesis and inflation of these heart regions are particularly shear-dependent.
Another key step in cardiac development, remodeling of the ventricular myocardium to form trabeculae, is critical for increasing the surface area through which the muscle mass of the ventricle can diffuse oxygen prior to the development of coronary circulation (Sedmera, 2005). Trabeculation does not occur in muz; instead the non-trabeculating regions of the ventricular myocardial wall become very thick. Wild type myocardium undergoing trabeculation displays a vacuolated appearance that we also observe in muz. Failure to form trabeculae could be simply due to lower oxygen requirements of the inactive mutant heart; trabeculation could also depend structurally on sarcomere integrity, or require signals from the overlying endocardium (Gassmann et al., 1995;Grego-Bessa et al., 2007;Meyer and Birchmeier, 1995), some of which regulate myocyte proliferation. Interestingly, the nontrabeculating muz myocardial wall appears as thick as its wild type counterpart, suggesting that proliferation may still occur. Although endocardium does not express myh6, it is known to alter its gene expression in response to haemodynamic changes (Groenendijk et al., 2005); it remains to be seen whether specific trabeculation signals are affected in the mutant.
As the embryonic heart matures, efficient function depends on the formation of endocardial valves to prevent retrograde blood flow between chambers. Studies in Danio suggest that when contraction and/or blood flow is disrupted mechanically (Hove et al., 2003) or genetically (Bartman et al., 2004), valve formation is impaired, but this process is now thought to occur by an atypical mechanism of direct leaflet invagination in zebrafish (Scherz et al., 2008). We have seen no evidence of precursors or differentiated valves in muz embryos, consistent with a requirement for blood flow in valve formation mediated by more conventional endocardial cushion intermediates. However, in the absence of cushion-specific markers, which have not been described in Xenopus, morphological distortion of the muz endocardium makes it difficult for us to conclusively rule out the presence of ectopic cushion precursors.
Several other mutations affecting heart function have been identified in pilot genetic screens in X. tropicalis (Goda et al., 2006;Grammer et al., 2005;Noramly et al., 2005), rapid mapping strategies have been established ( (Khokha et al., 2009); see also Supplemental Figure S3 for an X. tropicalis genetic mapping strategy flowchart), and reverse genetic resources are being developed ( (Goda et al., 2006), http://www.sanger.ac.uk/Teams/Team31/ xtmr.shtml) from which mutants in known genes can be obtained. Heart development in X. tropicalis genetic models can be analyzed with a broad array of molecular, genomic, and embryological tools, including gain-of-function mRNA expression screens (Smith and Harland, 1992) to identify interacting suppressor or enhancer functions and sophisticated explant assays modeling differentiation to diverse tissue fates including beating cardiac muscle (Latinkic et al., 2003). Reinforced by these robust functional assays, genetic approaches in amphibians complement rapidly-advancing genomics technologies for dissecting tetrapod developmental processes. The work presented here demonstrates the feasibility of positionally cloning mutations in X. tropicalis, greatly increasing the range of genetic studies.    Coronal plastic sections of stage 40 wild type and muz hearts (top rows) numbered from ventral side of cardiac cavity, and indicated by white lines in 3D models (bottom rows). m= myocardium, e= inner endocardial tube, v= ventricle, ot= outflow tract , a= atrium. No blood cells are seen in the muz sections due to lack of circulation, and myocardial layer appears thinner throughout the muz heart compared to wild type. The muz ventricle is wider than in wild type (sections 7 and 11), while outflow tract and atrium are dilated (sections 14, 23 and 41). Abnormal muz chamber morphology is highlighted in 3D projections of outlines of myocardium (A, C, E, G, red=ventricle, blue=outflow tract, green=atrium) and endocardium (B, D, F, H, orange), including elongated ventricle, dilated outflow tract (black arrowhead in E) and narrow cardiac tube at AVC level (black arrow in G). muz endocardium is very compressed with drastically reduced lumen (white arrows in 23, F and H)

Figure 6. Muz hearts become dilated and lack valves and trabeculae
Coronal plastic sections of stage 42 wt and muz hearts (top rows) numbered from ventral side of cardiac cavity, and indicated by white lines in 3D models (middle rows). v= ventricle, ot= outflow tract , a= atrium. Wild type hearts show a spiral valve in the outflow tract (sections 14, 23, black arrows), and thickening of endocardium preceding atrioventricular valve formation (section 23, black asterisk). Valve formation is not detected in muz hearts, and endocardial lumen is drastically reduced in outflow tract and AVC regions (white arrowheads sections 54, 58, also compare models B and F). Endocardial cushion formation in AVC can also be seen in transverse sections of stage 42 wild type (I, white arrowhead) hearts but not in muz (J). Trabeculation has initiated in the wild type ventricle (I, black arrowheads) but is absent in muz (J). At this stage the ventricular myocardium has a vacuolated appearance in both wt and mutant embryos (I, J black arrows). Middle two rows: 3D projections of outlines of myocardium (A, C, E, G) and endocardium (B, D, F, H) highlight abnormal muz chamber morphology; red = ventricle, green = atrium, blue = outflow tract, orange = endocardium. Muz ventricles are elongated relative to wild type (E, G white arrows). A narrow tube connects muz ventricle and atrium (section 54 and G, black arrowheads; compare to 23, C).