The bHLH transcription factor Hand2 plays parallel roles in zebrafish heart and pectoral fin development

In vertebrate embryos, derivatives of the lateral plate mesoderm (LPM) include a number of discrete tissue types such as the heart, endothelium, blood, connective tissue, smooth muscle and chondrogenic portion of the limbs. The genetic regulation of LPM patterning and differentiation is undoubtedly complex, and the precise steps that convert undifferentiated LPM into these distinct organs remain unknown.

The patterning of the cardiogenic region of the LPM is especially intricate. In zebrafish, for example, a number of genes (e.g. gata4, gata5, gata6, hand2 and tbx5) are expressed bilaterally in a large portion of the LPM from the completion of gastrulation (Serbedzija et al., 1998; Reiter et al., 1999; Ruvinsky et al., 2000; Begemann and Ingham, 2000). During early somitogenesis stages, a specific anterior section of the LPM also expresses the NK-class transcription factor gene nkx2.5, considered an early marker of precardiac mesoderm (Chen and Fishman, 1996; Lee et al., 1996). By mid-somitogenesis, an anterior subset of these nkx2.5-expressing cells go on to express myocardial-specific genes (Yelon et al., 1999) and contribute to the myocardium (Serbedzija et al., 1998). These myocardial precursors are further subdivided prior to cardiac fusion into preventricular and preatrial cells, as evidenced by the restricted expression of chamber-specific markers (Yelon et al., 1999). Thus, a detailed pattern of gene expression is established within the anterior LPM well before heart tube formation.

Only a few regulators of pattern formation within the cardiogenic LPM have been identified. The secreted growth factors Bmp2 and Fgf8 and the transcription factor Gata5 contribute to the regulation of nkx2.5 expression (Kishimoto et al., 1997; Schultheiss et al., 1997; Reifers et al., 2000; Reiter et al., 1999). Signals from the embryonic midline may influence the induction of the myocardial precursors from among the nkx2.5-expressing precardiac cells (Goldstein and Fishman, 1998; Serbedzija et al., 1998). Finally, the subdivision of the myocardial precursors into preventricular and preatrial populations is likely to be influenced by the localization of retinoic acid (Chazaud et al., 1999; Niederreither et al., 1999; Xavier-Neto et al., 1999; Yelon and Stainier, 1999) as well as the transcription factor Irx4 (Bao et al., 1999; Bruneau et al., 2000).

In an effort to understand the genetic regulation of pattern formation within the LPM in greater detail, we have identified a number of zebrafish mutations that disrupt myocardial patterning (Alexander et al., 1998). Here, we demonstrate that the zebrafish hands off (han) locus plays an important role in the differentiation, patterning and morphogenesis of two distinct LPM derivatives-the myocardium and the pectoral fin mesenchyme. nfa2.5-expressing precardiac mesoderm forms in han mutants, but these cells cannot differentiate into properly patterned myocardial tissue. Similarly, the pectoral fin buds in han mutants arise but do not differentiate normally and fail to exhibit proper anteroposterior (AP) patterning.

Molecular analysis of two han alleles indicates that the han locus encodes the bHLH transcription factor Hand2, also known as dHAND/Thing-2/Hed (Cross et al., 1995; Hollenberg et al., 1995; Srivastava et al., 1995). hand2 is expressed in the LPM of zebrafish, chick, frog, and mouse embryos (Cross et al., 1995; Hollenberg et al., 1995; Srivastava et al., 1995; Angelo et al., 2000). While targeted gene inactivation in the mouse previously implicated Hand2 in the later differentiation of the right ventricle (Srivastava et al., 1997), its broader role in the early stages of myocardial development as well as its role in forelimb development were not recognized. Thus, our studies reveal several essential early functions of Hand2. Moreover, the phenotypic analysis of han mutants highlights an intriguing genetic parallel between heart and forelimb development.

Adult fish and embryos were maintained and staged as previously described (Westerfield, 1995). Mutations were maintained by outcrossing heterozygous adults to standard wild-type strains; homozygous or /ransheterozygous mutant embryos were generated by intercrossing heterozygotes. Some lines carried the golden mutation, which inhibits melanophore pigmentation and thereby facilitates inspection of internal organs (Streisinger et al., 1986).

The han^s6 mutation was identified in a screen for ENU-induced mutations affecting MF20 and S46 immunostaining in haploid embryos (Alexander et al., 1998). We used a postmeiotic mutagenesis protocol that can induce an array of genetic lesions, including point mutations, deletions and translocations (Imai et al., 2000). The han^c99 mutation was identified in an independent screen for gamma ray-induced mutations as part of the Zebrafish Deletion Project. Both mutations behave in a Mendelian recessive fashion with complete penetrance and embryonic lethality. han^s6 and han^c99 fail to complement each other: 157/601 embryos from c99 /+ x s6/+ matings display the mutant phenotype.

Whole-mount immunofluorescence, in situ hybridization and sectioning were performed as previously described (Yelon et al., 1999). The monoclonal antibodies MF20 (Bader et al., 1982) and S46 (generous gift of Dr Frank Stockdale) were used. MF20 was obtained from the Developmental Studies Hybridoma Bank, maintained by the Department of Biological Sciences, University of Iowa, under contract NO1-HD-2-3144 from the NICHD.

Genomic DNA extraction was performed as previously described (Reiter et al., 1999) or with a Nucleon kit (Scotlab). PCR strategies were standard, except for the use of eLONGase enzyme mix (BRL) for long-range amplification. For cDNA analysis, total RNA was isolated using Trizol (BRL) and reverse transcribed using Superscript II (BRL). 5’ RACE (SMART RACE, Clontech) was performed in order to confirm the 5’ end of the hand2 mRNA and to investigate the splice variants in han^c99 mutants. All sequencing was performed using an ABI 377 system.

A zebrafish P1 artificial chromosome (PAC) library (Amemiya and Zon, 1999) was screened by PCR on pools of clones and by hybridization to high-density filters (RZPD, Berlin). Further analysis of hand2-containing PACs was conducted in an effort to estimate the size of the han^s6 deletion. For example, we sequenced the ends of the hand2-containing PAC 57N7 and determined by PCR that both ends are present in han^s6 genomic DNA. Since 57N7 contains a ∼100 kb insert (determined by PFGE), we conclude that the han^s6 deletion removes less than 100 kb of genomic DNA.

We placed han^s6 on zebrafish linkage group 1 using half-tetrad analysis (Johnson et al., 1995) and mapped hand2 to the same region (between z9409 and z21548) using the Goodfellow zebrafish radiation hybrid panel (Geisler et al., 1999; see http://wwwmap.tuebingen.mpg.de/). han^s6 linkage testing was performed using a combination of four PCR primers that simultaneously amplify a hand2 fragment and a fragment of the end of PAC 57N7 from individual diploid or haploid embryos. Oligonucleotides used were 5’-AATATTGAACTTGCAAACATA-CAAGC-3’, 5’-GTCTATATGAATTACACTCTAGTGG-3’, 5’-AATT-TCCCACTACGGACATTGGA-3’, and 5’-AGAGACAGAAATAGA-TAATGAACGT-3’. han^c99 linkage testing was performed using a combination of three PCR primers that simultaneously amplify a fragment of hand2 that spans the han^c99 insertion and a fragment of the insertion from individual diploid or haploid embryos. The locations of these primers are indicated in Fig. 5C and typical results are shown in Fig. 5D. Oligonucletides used were 5’-TGATCACCCG-TTAATGTTCTTG-3’, 5’-CCCATGAAAAAGGTAAGAGTGAA-3’, and 5’-CGATTCAGACACCAACTGTCTC-3’. The same oligonucleotide mixes were used to genotype han^s6 or han^c99 embryos following in situ hybridization.

Capped synthetic hand2 and lacZ mRNA were generated using pCS2:hand2, SP64TK:lacZ and the mMessage mMachine system (Ambion). Embryos were injected with 100-500 pg of hand2 or lacZ mRNA at the 1- to 4-cell stage.

In the course of a screen for zebrafish mutations affecting myocardial development (Alexander et al., 1998), we identified s6, an ENU-induced mutation that causes severe heart, jaw and pectoral fin defects. We named the locus represented by this mutation hands off (han), primarily because of its lack of pectoral fins, which are the teleost equivalents of tetrapod forelimbs (Grandel and Schulte-Merker, 1998). A gamma ray-induced non-complementing allele (see Materials and Methods), h an^c99, causes a very similar array of defects that are generally less severe. Here, we use these two mutant alleles to investigate the role of han in heart and forelimb development; the role of han in jaw development will be discussed elsewhere (C. Miller, D. Y., D. Y. R. S. and C. Kimmel, unpublished data).

Cardiovascular defects are morphologically apparent in han^s6 mutant embryos by 24 hours postfertilization (hpf). It is difficult to detect any contracting tissue or circulating blood and a mild pericardial edema is present (Fig. 1A,B). Immunodetection of cardiac myosin heavy chain molecules (Fig. 1C,D) and examination of cardiac myosin light chain 2 (cmlc2) (Fig. 1E,F) expression revealed that extremely little myocardial tissue forms in han^s6 mutants. Furthermore, while wild-type embryos form a single midline heart tube, han^s6 mutants develop two small lateral clusters of myocardial cells that never fuse together at the midline (Fig. 1C-F). The wildtype heart is clearly divided into two distinct chambers-an anterior ventricle and a posterior atrium-but the han^s6 myocardial tissue is primarily atrial, as indicated by S46 immunostaining (Fig. 1C,D). Correspondingly, the expression of ventricular myosin heavy chain (vmhc) is strongly affected; some han^s6 mutants do not express vmhc at all, while others show an irregular distribution of vmhc in a few myocardial cells (Fig. 1G,H).

These defects in the production and patterning of myocardial tissue are evident from an early stage in han^s6 mutants. Even at the onset of myocardial differentiation at the 15-somite stage (16.5 hpf), han^s6 mutants have noticeably fewer cmlc2-expressing myocardial cells than their wild-type siblings (Fig. 1I,J). As in older han^s6 mutant embryos, small myocardial populations are clustered bilaterally and these cells rarely express vmhc (Fig. 1I-L).

Many myocardial genes are expressed robustly within the small populations of han^s6 myocardial cells, including bmp4, mef2c, gata4, gata5, gata6 and all of the cardiac contractile genes examined except vmhc (data not shown). The T-box transcription factor gene tbx5 (Tamura et al., 1999; Ruvinsky et al., 2000; Begemann and Ingham, 2000) is a notable exception. Expression of tbx5 in the LPM is initiated normally in han^s6 mutants during early somitogenesis (data not shown), but the mutant myocardial cells fail to maintain tbx5 expression. By 24 hpf, tbx5 expression is significantly reduced in han^s6 myocardial cells (Fig. 1M,N), while another myocardial T-box gene, hrt (Griffin et al., 2000), continues to be expressed at normal levels (Fig. 1O,P).

Thus, specific aspects of myocardial differentiation (production of cmlc2-expressing cells and maintenance of tbx5 expression), patterning (vmhc expression) and morphogenesis (heart tube formation) are defective in han^s6 mutants.

Although myocardial development is severely disrupted in han^s6 mutants, the initial establishment of the precardiac mesoderm appears to proceed normally. nkx2.5 is normally expressed in bilateral populations of precardiac cells from early somitogenesis stages (Chen and Fishman, 1996; Lee et al., 1996); han^s6 mutants and wild-type siblings are indistinguishable in this regard (Fig. 2A,B). In wild-type embryos, most, but not all, of these nCr2.5-expressing cells go on to express myocardial genes like cmlc2 (Yelon et al., 1999) and contribute to the myocardium (Serbedzija et al., 1998). However, in han^s6 mutants, very few nkx2.5-expressing cells go on to express cmlc2 or other myocardial genes (Fig. 1I,J). Therefore, the failure of han^s6 mutants to produce normal numbers of myocardial cells is not due to a deficiency of precardiac mesoderm. The initiation of myocardial differentiation in a subset of nkx2.5-expressing cells represents a specific developmental transition that does not proceed efficiently in han^s6 mutants.

During the initiation of myocardial differentiation in zebrafish, the anterior LPM undergoes an apparent expansion. At the completion of gastrulation, LPM markers like gata4 are expressed in narrow bilateral stripes of cells (Fig. 2C); wildtype and han^s6 mutant embryos appear identical in their expression of LPM markers from the tailbud stage until the 10-somite stage (data not shown). Between the 10-somite (14 hpf) and 15-somite (16.5 hpf) stages in wild-type embryos, these gene expression patterns change; the domains of LPM marker expression gradually expand mediolaterally. By the 15-somite stage, LPM genes are expressed in significantly broader bilateral sheets of cells in wild-type embryos (Fig. 2D). This transition from narrow to broader domains of LPM gene expression is profoundly affected in han^s6 mutants. For example, at the 15-somite stage, han^s6 mutants still exhibit narrow bilateral stripes of gata4 expression; no expansion has occurred (Fig. 2E). By 24 hpf, this tissue appears narrow and dysmorphic (Fig. 2G) in comparison to the continuous sheet of gata4-expressing cells found in wild-type siblings (Fig. 2F).

Having identified significant defects in the cardiogenic portion of the han^s6 LPM, we proceeded to examine the formation of other LPM derivatives in han^s6 mutants and found profound defects in pectoral fin development. In wild-type embryos, pectoral fin mesenchyme proliferates to form bilateral fin buds (Grandel and Schulte-Merker, 1998), but this process is significantly delayed in han mutants (Fig. 3A,B). The wildtype fin gradually elongates and comes to be composed of a central chondrogenic condensation flanked by myogenic mesenchyme (Grandel and Schulte-Merker, 1998); in contrast, han^s6 mutants have a small bud of undifferentiated mesenchyme that fails to elongate or form a chondrogenic condensation (Fig. 3C,D). The mutant pectoral fin mesenchyme expresses tbx5, although, in comparison to wild type, expression is not well maintained after 24 hpf (Fig. 3A-D and data not shown). The pectoral fin defects are unlikely to be secondary consequences of the lack of circulation, since many zebrafish mutants with severe cardiac defects exhibit normal pectoral fin development (Chen et al., 1996; Stainier et al., 1996).

In zebrafish and other vertebrates, limb outgrowth is dependent upon proper limb bud patterning (Johnson and Tabin, 1997; Martin, 1998). The expression of sonic hedgehog (shh) in a posterior region of the pectoral fin bud (the zone of polarizing activity, or ZPA) is an important influence on fin AP patterning (Neumann et al., 1999; Schauerte et al., 1998). han^s6 mutants never express shh in their pectoral fin buds at any stage, but shh expression in han^s6 endoderm and ventral neuroectoderm is normal (Fig. 3E). It is not known precisely how the posterior restriction of shh expression is regulated, although some early indications of limb AP patterning have been shown to be shh-independent in the zebrafish (Neumann et al., 1999). For example, initiation of hoxd-11, hoxd-12 and bmp2 expression in a posterior portion of the pectoral fin mesenchyme occurs normally in zebrafish sonic you (syu) mutants that lack shh activity (Neumann et al., 1999). In contrast, han mutants do not display any features of AP patterning within the pectoral fin mesenchyme (Fig. 3F-I and data not shown).

We also examined the pectoral fin-forming region of the LPM at stages prior to fin bud formation. In wild-type embryos, LPM markers such as tbx5 are first expressed in narrow stripes of the fin-forming region of the LPM (Ruvinsky et al., 2000; Begemann and Ingham, 2000). During segmentation stages, these expression domains expand mediolaterally (Fig. 3J,K), in a manner similar to the expansion of gene expression within the cardiogenic LPM.

This process of expansion does not occur efficiently in han^s6 mutants (Fig. 3J,K).

Examination of the less severely affected han^c99 mutants confirmed the roles of han indicated by the han^s6 phenotype. For example, the role of han in the generation of myocardial tissue is evident in han^c99 homozygotes and han^s6/c99transheterozygotes (Fig. 4A-D). The lack of myocardial cells is most extreme in han^s6 mutants (Fig. 4D) and most mild in han^c99 mutants (Fig. 4B), with transheterozygotes affected to an intermediate degree (Fig. 4C). Similarly, the failure to maintain myocardial tbx5 expression is more severe in han^s6 mutants (Fig. 4H) than in han^c99 mutants (Fig. 4F) or transheterozygotes (Fig. 4G), confirming that han regulates tbx5 maintenance. Furthermore, comparisons of vmhc expression indicated reduced amounts of ventricular tissue in han^c99 mutants and in transheterozygotes (data not shown), reinforcing the conclusion that han is essential for ventricular differentiation.

Phenotypic comparisons also indicated the role of han in the regulation of cardiac fusion (Fig. 4A-H). While the bilateral myocardial populations in han^s6 mutants never fuse (Fig. 4D and data not shown), the myocardial cells in han^c99 mutants (Fig. 4B,F) and transheterozygotes (Fig. 4C,G) fuse slowly in comparison to wild-type embryos and form small heart tubes (data not shown).

Examination of younger embryos demonstrated that han function regulates the extent of expansion of gata4 expression in the LPM (Fig. 4I-L). Slightly expanded domains of gata4 expression do form in han^c99 mutants (Fig. 4J) and transheterozygotes (Fig. 4K) show signs of irregular LPM morphogenesis. Furthermore, analysis of cmlc2 expression at this stage verified that han controls the early production of myocardial precursors (Fig. 4M-O). The number of cmlc2-expressing cells is clearly reduced in han^c99 mutants relative to wild type (Fig. 4M,N), although not as dramatically as in han^s6 mutants (Fig. 4O).

In contrast to their distinct cardiac phenotypes, the han^s6 and han^c99 pectoral fin phenotypes are quite similar. Neither the han^c99 nor the han^s6 pectoral fins elongate, maintain tbx5 expression effectively, express shh, hoxd-11 and hoxd-12, or exhibit normal expansion of tbx5 expression in the fin-forming LPM (data not shown). Maintenance of gene expression in the apical epidermal fold of the pectoral fin is regulated by shh in the zebrafish (Neumann et al., 1999), so neither han^s6 nor han^c99 mutants maintain apical fold gene expression well (Fig. 4P-R). Even so, han^c99 mutants have detectable dlx2 expression in the apical fold more often than han^s6 mutants do (8 of 8 han^c99 cases and only 5 of 10 han^s6 cases). This distinction suggests that some level of mesenchyme differentiation and/or patterning may be initiated, but not maintained, in han^c99 fin buds.

Altogether, the relationships between the han^s6, han^c99 and transheterozygote phenotypes lead to the hypothesis that han^s6 represents a null allele, while han^c99 retains partial han function.

In search of the genetic defect responsible for the han mutant phenotypes, we considered candidate genes known to be expressed in the heart, forelimb and jaw. The bHLH transcription factor gene hand2 is expressed in these tissues in mouse and chick embryos (Cross et al., 1995; Hollenberg et al., 1995; Srivastava et al., 1995). Moreover, we mapped the han^s6 mutation and a zebrafish hand2 homolog (Angelo et al., 2000) to the same region of zebrafish linkage group 1 (see Materials and Methods). We therefore proceeded to test whether han encoded zebrafish Hand2.

The genomic structure of zebrafish hand2 is relatively simple: the entire transcribed region (1.6 kb) is contained within two exons separated by a ∼200 bp intron (Fig. 5A). Extensive PCR and Southern blot analyses indicate that the entire hand2 locus is deleted from han^s6 genomic DNA; no fragment of the hand2 genomic sequence can be amplified from han^s6 genomic DNA by PCR (Fig. 5B and data not shown), and no portion of hand2 can hybridize to han^s6 genomic DNA on a Southern blot (data not shown). Furthermore, the deletion of hand2 segregates with the han^s6 mutant phenotype in 480 meiotic events examined (see Materials and Methods). Analysis of hand2-containing PACs indicates that the han^s6 deletion removes less than 100 kb of genomic DNA (see Materials and Methods).

Molecular analysis of the han^c99 mutation provided further evidence that a loss of Hand2 function was responsible for the han mutant phenotypes. The coding and untranslated sequences of the hand2 gene in han^c99 genomic DNA are intact, with the exception of a ∼5 kb insertion between bases 346 and 347 of the 5’ untranslated region (UTR) (Fig. 5C,D). This insertion segregates with the han^c99 mutant phenotype in 480 meiotic events examined (see Materials and Methods). We examined the 5’ end of hand2 mRNA in han^c99 mutants by 5’ RACE and identified insertion-containing messages as well as a splice variant not seen in wild-type embryos. This splice variant eliminates 269 bp of 5’ UTR and 76 bp of coding sequence (Fig. 5C) and could encode a truncated Hand2 protein, with the ATG that normally encodes the forty-sixth residue functioning as the initiation codon.

Altogether, these molecular analyses indicate that han^s6 is a null allele and han^c99 is a hypomorphic allele, consistent with the differences between the han^s6 and han^c99 mutant phenotypes. In an effort to test the hypothesis that a deficiency of hand2 is sufficient to cause the defects observed in han^s6 mutants, we examined the ability of wild-type hand2 mRNA to rescue aspects of the han^s6 mutant phenotype. Injecting 100-500 pg of synthetic hand2 mRNA at the 1- to 4-cell stage can enhance the production of myocardial precursors in a small fraction of injected han^s6 mutant embryos (9 out of 95 injected mutants; Fig. 5E-G). This phenotype appears to represent a partial rescue of han function, since we never observed this number of cmlc2-expressing myocardial precursors in han^s6 uninjected or lacZ-injected siblings (>150 mutant embryos examined).

The early expression of hand2 in the LPM and the later expression of hand2 in the heart and pectoral fins correspond well with the timing and location of the defects observed in han mutants (Fig. 6). In wild-type zebrafish embryos, hand2 expression begins at the completion of gastrulation (tailbud stage) in a large portion of the LPM, with its anterior extent just posterior to the head and its posterior extent wrapped around the tailbud (Fig. 6A-C). As convergence and extension proceed, hand2 expression persists in the LPM (Fig. 6D,E); during somitogenesis, a small gap in hand2 expression appears (Fig. 6D, arrowheads), producing anterior and posterior expression domains. As the cardiogenic LPM expands, hand2 transcripts are found in mediolaterally wider domains of the anterior LPM (Fig. 6F), and expression persists in the myocardium as the heart tube forms (Fig. 6K). A mediolateral expansion of hand2 expression also occurs in the pectoral fin-forming portion of the LPM (Fig. 6G). In this portion of the LPM and in the pectoral fin buds, hand2 expression appears to overlap with the posterior portion of tbx5 expression (Fig. 6G-J).

Thus, all of the tissues affected in han mutants express hand2 at the time that defects are apparent. As expected from the molecular analysis of the mutant locus, han^s6 mutants never express hand2 at any stage (Fig. 6L and data not shown). han^c99 mutants exhibit a normal hand2 expression pattern, although expression levels are significantly and consistently reduced (Fig. 6M and data not shown).

Our phenotypic and molecular analyses of han^s6 and han^c99 mutants provide compelling evidence that the han mutant phenotypes are caused by disruption of the hand2 gene. All of the defects identified in han^s6 mutants are also present in han^c99 mutants, although han^c99 mutants are less severely affected. In agreement with the severity of the phenotypes, the han^s6 deletion removes the entire hand2 gene and the han^c99 insertion alters hand2 splicing. Furthermore, the expression pattern of hand2 correlates well with the timing and location of the han defects. We therefore conclude that the han locus encodes Hand2.

The present data, however, cannot rule out that the han^s6 and han^c99 mutations affect other genes besides hand2. In particular, there could be a neighboring gene that is removed or influenced by the han^s6 deletion that is also affected at a distance by the han^c99 insertion. Without additional han alleles available at this time, we favor the simplest interpretation that the han^s6 and han^c99 phenotypes reflect the role of Hand2 alone.

These studies of han mutants provide the first evidence that a hand gene is required for specific early transitions during myocardial development. Although a normal number of nkx2.5-expressing precardiac cells form without Hand2, Hand2 function is essential for their differentiation into cmlc2-expressing myocardial precursors. Furthermore, Hand2 is important for generation of vmhc-expressing ventricular cells, maintenance of myocardial tbx5 expression and formation of a midline heart tube. Additionally, Hand2 is critical for an early transition of the cardiogenic LPM, during which the expression domains of LPM genes expand mediolaterally. We hypothesize that this expansion of gene expression reflects the morphogenesis of the LPM through a combination of cell movements and proliferation.

Since hand2 is expressed broadly within the LPM (in more cells than just the myocardial precursors) and appears to influence the morphogenesis of an extensive portion of the LPM, it seems likely that the role of Hand2 during LPM development begins very early. Even as early as the tailbud stage, Hand2 activity could mediate the LPM response to intrinsic and extrinsic signals that control its subsequent morphogenesis, differentiation and patterning. It is interesting to consider our results from hand2 mRNA overexpression in light of this model. Overexpression of hand2 in wild-type embryos does not notably affect the formation of myocardial precursors (D. Y. and D. Y R. S., unpublished data); however, overexpression of hand2 in han^s6 embryos increases the production of myocardial precursors. These data are consistent with Hand2 playing more of a permissive, rather than instructive, role during the selection of myocardial precursors within the LPM. Thus, the lack of efficient or complete rescue of myocardial production in han^s6 mutants could be due to a requirement for continuous expression of hand2 within the LPM that cannot be restored by the transient introduction of short-lived mRNA molecules. Complete rescue of han^s6 mutants may therefore require stable transgenesis using tissue-specific promoters.

Loss of Hand2 function also inhibits pectoral fin outgrowth and differentiation, and Hand2 is essential for shh induction in the ZPA. In many respects, the han pectoral fin phenotype is similar to that observed in zebrafish syu mutants that lack shh expression (Neumann et al., 1999; Schauerte et al., 1998). However, there is a clear distinction between the han and syu phenotypes: Hand2 is essential for an earlier fundamental establishment of AP pattern, as revealed by hox gene expression, that is intact in syu mutants (Neumann et al., 1999). Additionally, Hand2 function, unlike Shh function (Begemann and Ingham, 2000), is required for the early expansion of the pectoral fin-forming portion of the LPM and for the maintenance of mesenchymal tbx5 expression from 24 hpf.

The similar themes of the han pectoral fin defects and the han myocardial defects-early disruption of morphogenesis followed by later defects in differentiation and patterning-reveal parallel roles of Hand2 in heart and forelimb development. Hand2 may promote an analogous state of developmental competence in both the cardiogenic and forelimb-forming portions of the LPM that allows critical morphogenetic transitions, like LPM expansion, to occur. We suspect that these morphogenetic transitions, in turn, would be necessary for proper differentiation. In this regard, it is interesting to note that the influence of Hand2 on pectoral fin development extends beyond the domain of hand2 expression in the fin-bud mesenchyme. While only a posterior portion of the fin-forming region of the LPM expresses hand2 (relative to broader tbx5 expression), reduced Hand2 function affects the morphogenesis of a larger LPM domain and also affects the maintenance of tbx5 expression throughout the fin bud. This discrepancy suggests that the critical stage for Hand2 function may occur during early somitogenesis, when hand2 expression extends throughout the fin-forming LPM. Alternatively, Hand2 may play a more specific role in the AP patterning of the fin bud; for example, absence of Hand2 in the posterior of the fin-forming LPM may prevent the localization of posterior determinants from an early stage, and the lack of proper AP coordinates inhibits general morphogenesis and differentiation processes.

Early parallel roles of Hand2 in myocardial development and limb development were not previously revealed by studies of mouse hand2 mutants (Srivastava et al., 1997). Targeted disruption of the mouse hand2 gene produces a less severe phenotype than disruption of zebrafish hand2. Mouse hand2 mutants proceed through the early stages of cardiogenesis well, generating a normal number of myocardial precursors that form a midline heart tube. However, at a later stage (E9.5), the right ventricle fails to differentiate, leading to embryonic lethality. No limb phenotypes have previously been reported for the mouse hand2 mutants.

One possible explanation for this discrepancy is that the bHLH transcription factor gene Hand1, also known as eHAND/Thing-1/Hxt (Cross et al., 1995; Cserjesi et al., 1995; Hollenberg et al., 1995; Srivastava et al., 1995), may compensate for a lack of Hand2 function in mouse hand2 mutants. Functional redundancy for Hand1 and Hand2 has been suggested previously by antisense oligonucleotide experiments in chick: oligonucleotides directed against handl and hand2, in combination, but not singly, can disrupt cardiac looping (Srivastava et al., 1995). In the mouse embryo, the expression patterns of murine handl and hand2 overlap significantly, especially within the cardiogenic LPM (Cross et al., 1995; Cserjesi et al., 1995; Biben and Harvey, 1997; Srivastava et al., 1997). Targeted disruption of the mouse handl gene disrupts trophoblast development, making the analysis of cardiac development difficult (Firulli et al., 1998; Riley et al., 1998). However, tetraploid rescue experiments indicate that Hand1 is essential for the morphogenesis and differentiation of the midline heart tube (Riley et al., 1998). Therefore, we speculate that Hand1 and Hand2 can each compensate for the other’s absence during early stages of myocardial development in the mouse. Evidently, no such compensation for the loss of hand2 occurs in zebrafish han mutants, and we have not yet detected a zebrafish handl-like gene (Angelo et al., 2000; D. Y. and D. Y R. S., unpublished data).

Based on the phenotype of mouse hand2 mutants, previous models of myocardial Hand2 function have suggested that Hand2 regulates chamber-specific differentiation of the right ventricle (Srivastava et al., 1997) or that Hand2 prevents programmed cell death within the right ventricle (Srivastava, 1999; Yamagishi et al., 1999). Our analyses of zebrafish han mutants suggest that it is necessary to reevaluate these models, taking into account that Hand proteins appear to play a more general and early role in mediating LPM development, especially LPM morphogenesis and differentiation.

In addition to demonstrating a number of specific functions for Hand2 during zebrafish embryogenesis, our studies also highlight the parallels between heart and forelimb development. Several human congenital syndromes include developmental defects in both the heart and limbs (Wilson, 1998); one of these, Holt-Oram syndrome, is caused by mutations in TBX5 (Basson et al., 1997; Li et al., 1997). Tbx5 has also been shown to regulate cardiac differentiation in Xenopus (Horb and Thomsen, 1999) and forelimb differentiation in chick (Gibson-Brown et al., 1998; Isaac et al., 1998; Logan et al., 1998; Ohuchi et al., 1998). The han mutant phenotype, especially the poor maintenance of tbx5 expression in the heart and forelimb, suggests that Hand2, its cofactors and its targets may be relevant to a parallel pathway for heart and forelimb development, as well as to human disorders that occur when this pathway is disrupted.

Other factors are known to play parallel roles in heart and limb development. Fibroblast growth factors (FGFs) have been shown to regulate the induction of cardiac and limb primordia within the LPM. fgf8 influences nkx2.5 expression in the zebrafish precardiac mesoderm (Reifers et al., 2000), and a number of different FGFs can induce ectopic limbs in the chick (Martin, 1998). Initial investigations of a mouse mutant that cannot synthesize retinoic acid (RA) (Niederreither et al., 1999), as well as other studies of the functions of RA during embryogenesis (Chazaud et al., 1999; Xavier-Neto et al., 1999; Yelon and Stainier, 1999; Johnson and Tabin, 1997), demonstrate a common role for RA localization in the posteriorization of both the heart and limb. Further examination of Hand2 function is likely to reveal additional similarities between heart and forelimb development and the underlying genetic pathways; it will be especially interesting to examine the relationships between FGFs, RA and Hand2.

We thank C. Miller, X. Sun, G. Martin, A. Schier, A. Joyner, D. Chu and members of the Stainier laboratory for comments on the manuscript, A. Navarro for valuable technical assistance, J. Alexander, A. Rubinstein and M. Macurak for collaboration on genetic screens, M. Ekker, V. Prince, P. Sordino and K. Griffin for plasmids, F. Stockdale for the S46 antibody, and S. Alper for long-range PCR advice. D. Y. was an Amgen fellow of the Life Sciences Research Foundation and is currently the recipient of a Burroughs Wellcome Fund Career Award. M. E. H. acknowledges support from the NIH (DK55390-02). D. Y R. S. is supported by the NIH, AHA and the David and Lucille Packard Foundation.

See Charité et al. (Development127, 2463-2467) for a report of the limb phenotype in mice lacking Hand2. Additional studies of Hand2 function in the limb are reported by Fernandez-Teran et al. (Development127, 2133-2142).