The yolk syncytial layer regulates myocardial migration by influencing extracellular matrix assembly in zebrafish

Specific regions of the vertebrate extra-embryonic tissues have been shown to function as important signaling centers for early body patterning. In mammals, for example, the extra-embryonic anterior visceral endoderm (AVE)plays essential roles in forebrain induction (reviewed by Beddington and Robertson, 1999; de Souza and Niehrs, 2000). In addition, reciprocal cell-cell interactions between the extra-embryonic ectoderm and the epiblast play crucial roles in early anterior-posterior patterning and mesoderm induction during mouse embryogenesis(Brennan et al., 2001). Similarly, in zebrafish, the extra-embryonic yolk syncytial layer (YSL) is known to function as an important signaling center for mesoderm induction and early dorsoventral patterning (Mizuno et al., 1996; Chen and Kimelman,2000) (reviewed by Sakaguchi et al., 2002). However, little is known about the roles the extra-embryonic tissues might play in later developmental events, such as organogenesis.

The heart is the first organ to form and function during vertebrate development. In all vertebrates, the heart tube develops from bilateral populations of precursor cells in the anterior lateral plate mesoderm(reviewed by McFadden and Olson, 2002). In zebrafish, these myocardial precursors migrate in between the pharyngeal endoderm and the extra-embryonic YSL toward the midline and subsequently fuse to form the functional heart tube(reviewed by Stainier, 2001; Trinh and Stainier, 2004). Large-scale forward genetic screens in zebrafish have identified mutations in eight loci that disrupt this migration process, resulting in the formation of two separate hearts, a phenotype referred to as cardia bifida(Chen et al., 1996; Stainier et al., 1996; Alexander et al., 1998). Previous analyses have demonstrated that the endodermal layer is essential for myocardial migration, as all mutants that lack pharyngeal endoderm show cardia bifida (Alexander et al., 1999; Reiter et al., 1999; Kikuchi et al., 2000). Cell transplantation analyses have further shown that wild-type endoderm can rescue the myocardial migration defects in the casanova/sox32 mutant(David and Rosa, 2001),further supporting the idea that the pharyngeal endoderm is essential for myocardial migration. Cellular analyses have indicated that the epithelial integrity of the myocardial precursors is also important for their migration,as evidenced by the cardia bifida mutation natter, which affects fibronectin 1 and appears to disrupt the polarity of the myocardial precursors (Trinh and Stainier,2004). This finding is consistent with previous observations in mouse and chick embryos that Fibronectin, a major constituent of the extracellular matrix (ECM), is required for heart tube formation(Linask and Lash, 1988; George et al., 1993). The mechanisms that lead to cardia bifida in the miles apart mutant are still unclear, although the mutant exhibits dysmorphic pharyngeal endoderm(Kupperman et al., 2000). The importance of the pharyngeal endoderm in myocardial migration has been well documented; however, a role for the extra-embryonic YSL in this process has not been identified from the forward genetic analyses.

The YSL is the cortical cytoplasm region of the yolk cell, and it contains hundreds of nuclei (Trinkaus,1992). The YSL is formed at the 500- to 1000-cell stage as marginal blastomeres collapse onto the yolk cell(Kimmel and Law, 1985). Although the YSL usually persists up to 9 days post-fertilization, it does not contribute cells or nuclei to any adult tissues(D'Amico and Cooper, 2001). Therefore, once established at the blastula stage, the YSL takes an exclusively extra-embryonic cell fate.

Several zebrafish mix-family genes, such as bonnie and clyde (bon), mtx1 (mxtx1 - Zebrafish Information Network), and mtx2 (mxtx2 - Zebrafish Information Network), are expressed in the YSL(Alexander et al., 1999; Hirata et al., 2000; Kikuchi et al., 2000). bon and mtx2 are expressed in the marginal mesendodermal cells and the YSL, while mtx1 is expressed exclusively in the YSL during blastula and gastrula stages(Alexander et al., 1999; Hirata et al., 2000). In higher vertebrates, mix-family genes such as mixer and Mixl1 play important roles in endoderm and axial mesoderm development(Henry and Melton, 1998; Hart et al., 2002). In zebrafish, bon plays a crucial role in early endoderm differentiation(Kikuchi et al., 2000), and mtx2 appears to regulate epiboly movements(Bruce et al., 2005). However,the function of mtx1 remains unexplored.

In this study, we show that the YSL-specific gene mtx1 is required for myocardial migration. We also show that mtx1 regulates natter/fibronectin 1 expression, and that mtx1 and natter/fibronectin 1 interact genetically. Based on these and other data, we propose that the extra-embryonic YSL regulates cardiac morphogenesis,and does so in part by regulating ECM assembly in embryonic tissues.

Wild-type zebrafish (Danio rerio) embryos were obtained from natural crosses of the Oregon AB or AB/TU heterozygote backgrounds. Homozygous or heterozygous natter^tl43c mutant embryos were obtained from incrosses and/or outcrosses of natter heterozygous fish. Collected embryos were maintained at 28.5°C and staged according to age[hours post-fertilization (hpf) at 28.5°C] and morphological criteria(Kimmel et al., 1995).

Three independent antisense morpholinos (MOs) for the mtx1 gene(Gene Tools, LLC) were injected into the YSL at the 1000-cell stage (3 hpf) as previously described (Sakaguchi et al.,2001). Eight nanograms/embryo of mtx1 MO(A)(5′-CATCAATAGTGCTGTCTTTCCACAT-3′), 4 ng/embryo of mtx1MO(B) (5′-GCGTCTTCACTGGTGGAATCCTGGA-3′), 8 ng/embryo of mtx1 MO(C) (5′-TAAACATGACAGCCCACCTGTATGC-3′), and 8 ng/embryo of 4 bp mismatch control mtx1 MO(A)(5′-CAaCAAaAGTGCTGTCTaTCCtCAT-3′) were injected into the YSL. Two nanograms/embryo of laminin c1 MO(5′-TGTGTCCTTTTGCTATTGCGACCTC-3′)(Parsons et al., 2002) was injected into 1-cell stage embryos.

Capped sense mtx1 mRNAs were synthesized using the mMessage mMachine in vitro transcription kit (Ambion), and purified as previously described (Sakaguchi et al.,2001). One hundred picograms/embryo of mtx1 mRNA was injected into the YSL at the 1000-cell (3 hpf) or dome (4.3 hpf) stage.

For RT-PCR of mtx1 cDNA, wild-type embryos and mtx1 MO(C)injected embryos were collected at shield stage and total RNA was isolated. RT reactions were performed using the SuperScriptII RT-PCR kit (Gibco BRL) with the mtx1 specific RT primer(5′-TCCTCAGTGGAGTGGTTATTTAGTC-3′) according to the manufacturer's recommendations. PCR was performed using the primer pair(5′-GAGAGTCTTTCCCAAACCACAG-3′ and 5′-CATATTGTTGTAGGCTGGCAAGT-3′) to amplify mtx1 cDNA fragments.

Genotyping by RFLP of natter^tl43c was performed by PCR as previously described (Trinh and Stainier, 2004).

Whole-mount in situ hybridizations were performed as described(Sakaguchi et al., 2001). Immunohistochemistry was performed as described(Trinh and Stainier, 2004). We used the following antibodies: rabbit polyclonal anti-Fibronectin (Sigma)(Trinh and Stainier, 2004) at 1:200; mouse IgG₁ anti-β-catenin (Sigma) at 1:500; rabbit polyclonal anti-Laminin 1 (Sigma) at 1:200. For whole-mount anti-Laminin 1 staining, we used the ABC staining kit (Vector Laboratory) after treating with an anti-rabbit IgG biotin-conjugated antibody. Fluorescence confocal images were acquired using a Zeiss LSM5 Pascal confocal microscope.

As the YSL is an important signaling center during embryogenesis (reviewed by Sakaguchi et al., 2002), we screened for genes that are expressed in the YSL at the late blastula stage and examined their function by MO(Heasman, 2002) knockdown experiments. We found that an MO designed to knock down the 124H4clone caused heart malformations and subsequently determined that the 124H4 clone is identical to the previously reported transcription factor gene mtx1 (Hirata et al.,2000). mtx1 is a Mix-type homeobox gene that is expressed exclusively in the YSL from the oblong stage (3.7 hpf) until 80% epiboly (8.3 hpf) (Hirata et al., 2000)(data not shown). No mtx1 transcripts can be detected after gastrula stage, or in any embryonic tissues (Hirata et al., 2000). However, it remains theoretically possible that low levels of mtx1 transcripts are present in embryonic tissues. In all experiments, we injected mtx1 MOs into the YSL at the 1000-cell stage(3 hpf) to knock down mtx1 function in this tissue. We found that mtx1 MO(A) injected embryos displayed pericardial edema(Fig. 1B, arrowhead) and a flattened hindbrain (Fig. 1B,arrow) at 34 hpf. Injecting a 4 bp mismatch control MO into the YSL did not cause any phenotype (Fig. 1A). Interestingly, when we investigated the cause of the pericardial edema, we found that approximately 30% (14/51) of mtx1 MO(A)-injected embryos showed cardia bifida at 34 hpf (Fig. 1D, arrowheads). Co-injection of Rhodamine Dextran with mtx1 MO(A) allowed us to visualize the distribution of the MO and showed it to be mostly restricted to the YSL even as late as 34 hpf(Fig. 1E,F). These results indicate that the YSL regulates myocardial migration.

We further analyzed the cardia bifida phenotype of mtx1MO(A)-injected embryos by examining the expression of the myocardial-specific marker cmlc2 (myl7 - Zebrafish Information Network)(Yelon et al., 1999). We first examined myocardial migration at the 21-somite stage (19.5 hpf), a time at which the two groups of myocardial precursors initially fuse at the midline(Yelon et al., 1999). We found that 100% (14/14) of mtx1 MO(A)-injected embryos exhibited two bilateral clusters of myocardial cells at this stage, indicating a failure in their migration to the midline (Fig. 1H). By contrast, none of the 4 bp mismatch control MO-injected embryos exhibited a myocardial migration defect(Fig. 1G). Time-lapse observations of myocardial migration with Tg(cmlc2:GFP) embryos(Huang et al., 2003) indicated that the myocardial precursors, which failed to fuse at the 21-somite stage,had reached the midline by 24 hpf in 4/6 of mtx1 MO(A)-injected embryos (Fig. 1N), while in the other two cases they remained bilateral(Fig. 1O). The embryos showing delayed myocardial fusion sometimes established circulation by 34 hpf, whereas the embryos showing bilateral myocardial precursors at 24 hpf developed clear cardia bifida by 34 hpf. Consistent with these observations, we found that at 38 hpf the cmlc2-expressing myocardial precursors remained bilateral in approximately 30% (4/11) of mtx1 MO(A)-injected embryos(Fig. 1Q). These results indicate that the migration of the myocardial precursors was severely compromised in mtx1 MO(A)-injected embryos.

To test the specificity of MO(A) for mtx1, we designed and injected two additional mtx1 MOs [MO(B) and (C)] into the YSL. Both MO(B) and MO(C) caused the same phenotype as MO(A), and myocardial cells failed to fuse at the midline at the 21-somite stage in all injected embryos(16/16, 18/18, respectively; Fig. 1I,J), suggesting that the mtx1 MOs function in a target-specific manner.

As cytoplasmic bridges connect the yolk cell to marginal blastomeres until sphere stage (4 hpf) (Cooper and D'Amico,1996), MOs injected into the YSL before sphere stage can diffuse into embryonic tissues. As it remained formally possible that mtx1was expressed at very low levels in marginal blastomeres, and to assess the specificity of the MOs, we wanted to test whether restoring mtx1 mRNA specifically in the YSL would rescue the myocardial migration phenotype of mtx1 MO-injected embryos. We injected mtx1 MO(C) into the YSL at the 1000-cell stage (3 hpf) and subsequently injected some of these embryos with mtx1 mRNA into the YSL at dome stage (4.3 hpf). While all (n=8) of the embryos injected with mtx1 MO(C) showed delayed myocardial migration at the 21-somite stage(Fig. 1K), 50% (3/6) of the embryos subsequently injected with mtx1 mRNA into the YSL exhibited wild-type-like myocardial migration (Fig. 1L). These data indicate that mtx1 functions in the YSL to regulate myocardial migration.

Of the three mtx1 MOs tested, two of them [MO(A) and MO(B)] were designed to block translation initiation of mtx1 mRNA, and the other[MO(C)] was designed against the splice donor site of the second exon in order to interfere with splicing (Fig. 2A). RT-PCR analysis revealed that MO(C) blocked the endogenous splice site of mtx1 and, as a result, an additional 20 bp was inserted into the mtx1 mRNA, leading to a frame shift and premature translation stop (Fig. 2A,B). This result further illustrates the efficiency and specificity of mtx1 MOs.

As endoderm formation is known to be essential for myocardial migration in zebrafish (reviewed by Stainier,2001), we first asked whether endoderm development in mtx1 MO-injected embryos was affected. Expression of early endoderm markers, such as sox17 (Alexander and Stainier, 1999) and casanova/sox32(Dickmeis et al., 2001; Kikuchi et al., 2001; Sakaguchi et al., 2001),appeared unaffected in mtx1 MO(A)-injected embryos(Fig. 3A-D), indicating that the loss of mtx1 function does not affect early endoderm formation. Later in development, myocardial cells migrate underneath a sheet of pharyngeal endoderm toward the midline(Trinh and Stainier, 2004). Therefore, we examined pharyngeal endoderm development in mtx1MO(A)-injected embryos and it appeared unaffected, as assessed by examining the pharyngeal endoderm-specific GFP expression in Tg(her5:GFP)^ne206 embryos(Tallafuss and Bally-Cuif,2003) at the 18-somite stage(Fig. 3E,F). Despite the absence of obvious pharyngeal endoderm defects, these embryos exhibited myocardial migration defects at the 21-somite stage (data not shown),indicating that mtx1 had been knocked down in these embryos. We also examined Tg(her5:GFP)^ne2067; Tg(cmlc2:GFP) embryos in cross-sections and did not detect any obvious defects in her5:GFPexpression in mtx1 MO-injected embryos at the level of the cardiac mesoderm at the 17-somite stage (17.5 hpf)(Fig. 3G,H). These data suggest that the YSL regulates myocardial migration independently of pharyngeal endoderm formation.

Although the pharyngeal endoderm did not appear to be disrupted in mtx1 MO-injected embryos, we found that more posterior endoderm,which gives rise to the liver and pancreas, appeared to be affected. The posterior endoderm forms a wide sheet-like structure at early segmentation stages and subsequently condenses into a rod-like structure at the midline by 24 hpf (Field et al., 2003; Ober et al., 2003). We were able to visualize these condensing foregut endodermal cells at the 15-somite stage with a sox17-promoter-GFP transgenic line, Tg(sox17:GFP)^s870 (Y.K. and W. Chung, unpublished)(Fig. 3I). In mtx1MO(A)-injected embryos at the 15-somite stage, foregut endodermal cell migration appeared to be delayed, as evidenced by its sheet-like appearance(Fig. 3J). By 24 hpf, foregut endodermal cells, marked by foxa1 expression, have already formed a rod in control embryos (Fig. 3K), whereas this cell migration was delayed in mtx1MO(A)-injected embryos (12/15; Fig. 3L). Due to this cell migration delay, approximately 30% (6/18) of mtx1 MO(A)-injected embryos developed duplicated hepatic and anterior pancreatic buds, as assessed by Tg(gutGFP)^s854(Field et al., 2003; Ober et al., 2003) expression at 54 hpf (Fig. 3M,N). As endodermal cells migrate on the YSL to coalesce at the midline, the YSL is probably also regulating this process.

To investigate the mechanisms by which the YSL regulates myocardial migration, we next focused on the natter/fibronectin 1 gene, which,like mtx1, is necessary for myocardial migration but appears to act independently from the endoderm (Trinh and Stainier, 2004). In addition, mtx1 MO-injected embryos exhibited similar phenotypes to those of natter/fibronectin 1mutants, such as a flattened hindbrain and a partially penetrant cardia bifida(Fig. 1B, Fig. 5C)(Trinh and Stainier, 2004). We analyzed natter/fibronectin 1 expression in mtx1MO(A)-injected embryos at several stages and detected a clear downregulation at the 14-somite stage (16 hpf) in the anterior lateral plate mesoderm, when the myocardial precursors start to migrate to the midline(Fig. 4A,B). This downregulation of natter/fibronectin 1 expression was also detected at the 21-somite stage (19.5 hpf), when the myocardial precursors first fuse at the midline (Fig. 4C,D). At the 21-somite stage, natter/fibronectin 1 is expressed in the laterally located myocardial cells and medially located endocardial cells(Trinh and Stainier, 2004). We observed a downregulation of natter/fibronectin 1 expression in both cell populations in mtx1 MO(A)-injected embryos(Fig. 4D). We also examined Fibronectin protein deposition in mtx1 MO(A)-injected embryos. In wild-type embryos at the 18-somite stage, Fibronectin is deposited on the basal side of the myocardial precursors as well as on the basal side of the pharyngeal endoderm (Fig. 4I)(Trinh and Stainier, 2004). In mtx1 MO(A)-injected embryos, Fibronectin deposition was greatly reduced around both the myocardial precursors and the pharyngeal endoderm(Fig. 4J).

As loss of mtx1 function downregulated natter/fibronectin 1 gene expression, we next asked whether mtx1 overexpression could alter natter/fibronectin 1 expression. Initially, we injected 100 pg mtx1 mRNA into 1-cell-stage embryos; however, these injected embryos showed severe gastrulation and epiboly defects. We then injected the same amount of mtx1 mRNA into the YSL at the 1000-cell stage to overexpress mtx1 preferentially in this tissue. We found that all these injected embryos appeared morphologically unaffected at 34 hpf (data not shown). However, interestingly, we found that natter/fibronectin 1expression in the anterior lateral plate mesoderm of these embryos appeared to be upregulated at the 18-somite stage (Fig. 4E-H). Fibronectin immunoreactivity was also clearly increased in embryonic tissues (Fig. 4K). These data suggest that the extra-embryonic YSL is regulating embryonic natter/fibronectin 1 expression in adjacent embryonic tissues.

To further investigate the relationship between mtx1 and fibronectin 1, we tested for genetic interaction. For this purpose,we injected mtx1 MO(A) into embryos obtained from crossing heterozygote natter parents, and surprisingly found that approximately three-quarters of these injected embryos exhibited a severely enhanced phenotype (Fig. 5B,D),while another quarter showed exactly the same phenotype as mtx1MO-injected embryos (Fig. 1B). Genotyping the affected and unaffected embryos revealed that injecting mtx1 MO(A) into heterozygous natter embryos enhanced the mtx1 MO injection phenotype (Fig. 5B): at 34 hpf, the head of these embryos appeared to be collapsed onto the yolk ball, the somite boundaries were not easily distinguishable, and pigmentation appeared to be delayed (Fig. 5B). These embryos survived at least until 50 hpf, suggesting that they were not dying simply from non-specific effects of the MO. These data indicate that mtx1 and natter/fibronectin 1 interact in a dosage-dependent manner, suggesting that they work in the same or interacting pathways.

Homozygous natter mutant embryos injected with mtx1 MO(A)frequently failed to complete epiboly and died after yolk rupture during gastrulation (data not shown). The embryos that survived appeared similar to mtx1 MO(A)-injected heterozygous natter embryos at 34 hpf(Fig. 5D).

As mtx1 MO-injected heterozygous natter embryos showed severe morphogenetic defects, we carried out a time-course analysis to determine how these defects developed during embryogenesis(Fig. 6). mtx1MO-injected heterozygous natter embryos appeared normal at the tailbud stage (Fig. 6A,F; 10 hpf). However, 2 hours later, at the 6-somite stage, the YSL at the hindbrain level started to darken (Fig. 6B,G; 12 hpf). By the 17-somite stage (17.5 hpf), the head structure appeared to be collapsing onto the YSL in mtx1 MO-injected heterozygous natter embryos (Fig. 6C-D,H-I). At 24 hpf, somite boundaries became hard to distinguish, especially in the anterior part(Fig. 6E,J). This result is consistent with previous observations that natter/fibronectin 1 and fibronectin 3 are required for somite formation(Julich et al., 2005; Koshida et al., 2005). By 34 hpf, the hindbrain was completely collapsed onto the yolk ball and most somite boundaries appeared to be missing (Fig. 5B). These data show that major defects appeared at late segmentation stages in mtx1 MO-injected heterozygous natterembryos.

To characterize in more detail mtx1 MO-injected heterozygous natter embryos, we analyzed the expression of regional specification markers for the hindbrain, somites and heart. Although the morphology of mtx1 MO-injected heterozygous natter embryos appeared to be severely disrupted (Fig. 6),the expression of the midbrain-hindbrain boundary marker paxb(pax2a - Zebrafish Information Network)(Fig. 7A,B)(Krauss et al., 1991), the rhombomeres 3 and 5 marker krox20 (egr2b - Zebrafish Information Network) (Fig. 7A,B) (Oxtoby et al., 1993), and the adaxial cell and segmented somite marker myod (Fig. 7C,D) (Weinberg et al.,1996) appeared in a properly segmented manner, indicating that these embryos were being patterned properly. Also, although myocardial cells failed to migrate in these embryos, they continued to express the myocardial marker cmlc2 (Fig. 7E,F).

To further understand why mtx1 MO-injected heterozygous natter embryos could not maintain their morphology, we examined ECM assembly by analyzing Laminin 1 immunoreactivity(Parsons et al., 2002; Pollard et al., 2006). We found that mtx1 MO-injected heterozygous natter embryos exhibited greatly reduced Laminin 1 immunoreactivity in the hindbrain at 24 hpf, while it appeared unchanged in their trunk and tail(Fig. 7H,J). Confocal images of cross-sections at the level of the cardiac mesoderm further illustrated the fact that Laminin 1 immunoreactivity was reduced in mtx1 MO-injected heterozygous natter embryos (Fig. 7L). As mtx1 MO injections into wild-type embryos did not significantly reduce Laminin 1 immunoreactivity (data not shown), these data suggest that losing one copy of natter/fibronectin 1 in the absence of mtx1 function leads to a further disruption of ECM assembly.

To further analyze the role of the YSL in ECM assembly, we injected mtx1 MO(A) into the YSL of embryos previously injected with laminin c1 (lamc1) MO at the 1-cell stage. lamc1encodes Laminin γ1 and is essential for notochord differentiation and Laminin 1 immunoreactivity (reviewed by Stemple, 2005; Parsons et al., 2002; Pollard et al., 2006). Embryos injected with lamc1 MO at the 1-cell stage exhibited a shortened body axis and defects in notochord differentiation at 30 hpf(Fig. 7M)(Parsons et al., 2002). When we injected mtx1 MO(A) into the YSL of embryos previously injected with lamc1 MO, approximately 50% of embryos (47/90) appeared to die after yolk rupture during segmentation stages. Surviving embryos appeared to exhibit an enhanced phenotype, in which body structures were collapsing in the hindbrain region (22/90) (Fig. 7N) or in the entire head region (17/90)(Fig. 7O) at 30 hpf. These phenotypes were not seen in lamc1 MO-injected embryos or mtx1 MO(A)-injected embryos. These results further support the idea that the YSL influences embryonic ECM assembly.

We also injected lamc1 MO into embryos obtained from crossing heterozygote natter parents. Approximately a quarter of these injected embryos (14/44) exhibited a severely enhanced phenotype(Fig. 7P,Q), while another three-quarters showed exactly the same phenotype as lamc1 MO-injected embryos (Fig. 7M). Genotyping the affected and unaffected embryos showed that injecting lamc1 MO into homozygous but not heterozygous natter embryos enhanced their phenotype (Fig. 7P,Q). These results indicate that embryos lacking multiple components of the ECM can exhibit enhanced phenotypes.

Our data indicate that the YSL-specific transcription factor, Mtx1,regulates natter/fibronectin 1 gene expression in the lateral plate mesoderm. As natter/fibronectin 1 is required for myocardial migration (Trinh and Stainier,2004), loss of Mtx1 function also leads to cardia bifida. Interestingly, we also found a genetic interaction between mtx1 and natter/fibronectin 1. Taken together, these data indicate that the extra-embryonic YSL regulates cardiac morphogenesis, at least in part by the mtx1-natter/fibronectin 1 pathway.

In this study, we injected mtx1 MOs into the YSL at the 1000-cell stage (3 hpf) to knock down mtx1 function in the YSL. All three mtx1 MOs caused the same phenotype, as assessed by the expression of the myocardial marker cmlc2 at the 21-somite stage(Fig. 1H-J). These data strongly suggest that mtx1 MOs function in a target-specific manner. Injection of the mtx1 MO(C) into 1-cell-stage embryos caused cardia bifida, although at low frequency (data not shown). As the YSL contains hundreds of nuclei, we suspect that knocking down genes in the YSL requires a higher concentration of MOs, one achieved by YSL injections.

We propose that our mtx1 MO injections are specifically knocking down mtx1 function in the YSL for the following reasons: first, mtx1 appears to be exclusively expressed in the YSL(Hirata et al., 2000), and we injected mtx1 MOs into this tissue. Second, the same concentration of a 4 bp mismatch mtx1 MO(A) did not cause any obvious phenotypes when injected into the YSL (Fig. 1A,G). Third, the splice donor MO, mtx1 MO(C), leads to a frame shift and premature translational termination of mtx1(Fig. 2). Finally, restoring mtx1 mRNA in the YSL rescued the myocardial migration defect(Fig. 1L).

In zebrafish, myocardial precursors migrate in between the endodermal layer and the extra-embryonic YSL (Fig. 8A) (Trinh and Stainier,2004) (reviewed by Stainier,2001). Previous studies have demonstrated that the endodermal layer is essential for myocardial migration (reviewed by Stainier, 2001). This study clearly illustrates for the first time that the extra-embryonic YSL also plays important roles in myocardial migration.

As mtx1 is a strictly zygotically expressed gene, a maternal factor probably controls mtx1 gene expression(Fig. 8B). mtx2, which is expressed in the YSL and marginal cells(Hirata et al., 2000), appears to be regulated by the maternal factor Eomesodermin(Bruce et al., 2005). However,Eomesodermin does not appear to regulate mtx1 gene expression(Bruce et al., 2005),indicating that another maternal factor is involved. Interestingly, the phenotype of mtx1 MO-injected heterozygous natter embryos appears similar to that of the maternal mutant pollywog(Fig. 5B)(Wagner et al., 2004),suggesting that Pollywog is an upstream regulator of mtx1.

Once mtx1 expression is established in the YSL, it appears to regulate embryonic natter/fibronectin1 gene expression in a non-cell-autonomous manner (Fig. 4). It is striking that loss of mtx1 function leads to downregulation of natter/fibronectin 1 expression and that, by contrast, gain of mtx1 function leads to upregulation of natter/fibronectin 1 expression in the anterior lateral plate mesoderm. However, the natter/fibronectin 1 gene cannot be a direct target of Mtx1, as their expression patterns do not appear to overlap. No fibronectin 1 transcripts can be detected at shield stage(Trinh and Stainier, 2004),when mtx1 is expressed most strongly(Hirata et al., 2000). Zygotic fibronectin 1 expression appears to initiate at 60% epiboly (6.5 hpf)in the marginal blastomeres but not in the YSL(Trinh and Stainier, 2004),while mtx1 appears to be expressed only in the YSL until 80% epiboly(8.3 hpf) (Hirata et al.,2000). Therefore, the regulation of natter/fibronectin 1by Mtx1 is probably indirect. Our data suggest that factors unidentified to date are regulated by Mtx1 in the YSL and work non-cell-autonomously to regulate natter/fibronectin 1 expression in the embryonic lateral plate mesoderm at mid-segmentation stages(Fig. 7). Further investigation will be required to identify the molecules that function downstream of mtx1 and upstream of natter/fibronectin 1. The steroidogenic enzyme Cyp11a1 (Hsu et al.,2006) and Rac-GAP Chimerin 1(Leskow et al., 2006) have been reported to function in the YSL to regulate epiboly cell movement. It will be interesting to investigate the role of these proteins in natter/fibronectin 1 gene expression and myocardial migration.

mtx1 might be working together with casanova/sox32, which is expressed in the endoderm and YSL and is responsible for endoderm formation(Dickmeis et al., 2001; Kikuchi et al., 2001; Sakaguchi et al., 2001). Although casanova MO injections into the YSL did not cause any obvious morphological phenotype (Sakaguchi et al., 2001), co-injection of mtx1 and casanovaMOs into the YSL dramatically increased the penetrance of cardia bifida at 34 hpf (data not shown). More detailed analysis will be required to further investigate this potential interaction between casanova and mtx1.

The idea that mtx1 regulates natter/fibronectin 1 gene expression is also supported by the fact that mtx1 and natter/fibronectin 1 show dosage-sensitive genetic interactions(Fig. 5). The phenotype of mtx1 MO-injected heterozygous natter embryos(Fig. 5B) is much more severe than that of mtx1 MO-injected wild-type embryos(Fig. 1B) or uninjected homozygous natter mutant embryos(Fig. 5C), implying that there is another pathway downstream of mtx1, working in parallel to the mtx1-natter/fibronectin 1 pathway(Fig. 8). Based on our observation that Laminin 1 immunoreactivity is greatly downregulated in mtx1 MO-injected heterozygous natter embryos(Fig. 7G-L), while it is not significantly affected in mtx1 MO-injected wild-type embryos, ECM components other than Fibronectin also appear to be regulated by mtx1(Fig. 8). This idea is further supported by the observation that embryos injected with both lamc1and mtx1 MOs showed a much more severe phenotype than that of lamc1 MO-injected embryos (Fig. 7M), mtx1 MO-injected embryos(Fig. 1B), or an addition of these two phenotypes. These data suggest that lacking multiple components of the ECM can lead to synergistic defects. We therefore propose that the extra-embryonic YSL regulates heart organogenesis and body structure maintenance at least in part by controlling ECM assembly(Fig. 8).

Although mtx1 MO-injected embryos developed normal pharyngeal endoderm, they exhibited foregut endoderm migration defects(Fig. 3). This foregut endoderm phenotype is not seen in natter mutants(Trinh and Stainier, 2004),indicating that foregut endoderm migration is not regulated by the mtx1-natter/fibronectin 1 pathway. Vegfc has been shown previously to regulate foregut endoderm migration(Ober et al., 2004), and it will be interesting to investigate whether the YSL regulates Vegfc function. Furthermore, the YSL might be serving as a scaffold for many cell types,including endodermal cells and myocardial cells, which migrate on it. We hypothesize that mtx1 might be involved in establishing the YSL as a scaffold for cell migration, and therefore loss of mtx1 function would compromise the motility of many cell types that migrate on the YSL(Fig. 8). This hypothesis is consistent with the fact that mtx2 in the YSL regulates embryonic cell movements during epiboly (Bruce et al., 2005).

Several mutations in mouse, such as those in the fibronectin, Gata4,Mesp1, Furin and Foxp4 genes, are known to affect the migration of myocardial precursors and lead to cardia bifida(George et al., 1993; Kuo et al., 1997; Molkentin et al., 1997; Roebroek et al., 1998; Saga et al., 1999; Constam and Robertson, 2000; Li et al., 2004). In mouse embryos around embryonic day 7.5, myocardial precursors appear to lie adjacent to the extra-embryonic visceral endoderm(Kaufman, 1992). This scheme might be equivalent to that in zebrafish, in which myocardial precursors migrate on the extra-embryonic YSL. Interestingly, chimeric analyses in mouse have demonstrated that Gata4 function in the visceral endoderm is required for myocardial migration (Watt et al.,2004). Therefore, we speculate that the YSL in zebrafish has a role in myocardial migration equivalent to that of the visceral endoderm in mouse. Several genes, such as ferroportin1(Donovan et al., 2000), are expressed in the YSL in zebrafish and the visceral endoderm in mouse, further highlighting the similarities between these two tissues. It will therefore be interesting to assess the role of the visceral endoderm in establishing ECM assembly in mouse embryonic tissues.

We thank Won-Suk Chung for Tg(sox17:GFP)^s870 fish, Le Trinh for sharing unpublished information about natter mutants and helpful discussions, Rima Arnaout for critical reading of the manuscript, and the reviewers for constructive feedback. mtx1 constructs were kindly provided by Drs Tsutomu Hirata and Masahiko Hibi. T.S. was supported by fellowships from the JSPS, the TOYOBO Biotechnology Foundation, and the American Physiological Society, and this work was further supported by grants from the NIH and Packard foundation to D.Y.R.S. T.S. thanks Jun-ichi Sakiyama and Deborah Yelon for helpful discussions.