The bHLH transcription factor hand2 is essential for noradrenergic differentiation of sympathetic neurons

The autonomic nervous system controls body function homoeostasis in vertebrates and is essential for adaptation to internal and external changes. Noradrenergic sympathetic neurons play a major role, for instance by regulating vascular tone, exocrine glands and gut function. To understand the molecular control of differentiated functions of noradrenergic neurons, it is useful to study the mechanisms involved in the initiation of sympathetic neuron differentiation. To this end, considerable progress has been made in recent years in the understanding of mechanisms leading to the generation and differentiation of sympathetic neurons from neural crest precursor cells(Goridis and Rohrer, 2002; Rohrer, 2003).

The development of autonomic neurons depends crucially on the presence of bone morphogenetic proteins (BMPs), extrinsic differentiation factors that are produced in the environment of the primordia of autonomic ganglia. In vivo loss-of-function experiments revealed that BMPs are essential for the generation of noradrenergic neurons of the paravertebral sympathetic chain ganglia (Schneider et al.,1999) as well as for parasympathetic ciliary ganglion neurons(Müller and Rohrer,2002). Conversely, additional sympathetic and parasympathetic neurons are produced in response to local BMP overexpression(Reissmann et al., 1996; Müller and Rohrer, 2002). The activation of BMP receptor signaling leads to the sequential expression of Ascl1, a proneural gene expressed by sympathetic and parasympathetic neurons, and of the paired-homeodomain transcription factors Phox2band Phox2a, which are expressed throughout the autonomic nervous system. The elimination of Ascl1 produces a strong impairment of sympathetic ganglion development, in particular in noradrenergic differentiation (Guillemot et al.,1993; Hirsch et al.,1998). The absence of Phox2b results in the complete lack of the autonomic nervous system, including neurons of the enteric nervous system (Pattyn et al., 1999). Phox2a is expressed downstream of both Ascl1 and Phox2b during sympathetic ganglion development(Lo et al., 1998; Pattyn et al., 1997; Hirsch et al., 1998), but its physiological function in the initial stages of this lineage is unclear in view of the recent observation that Phox2a is unable to rescue the effect of the Phox2b knockout on sympathetic neuron development(Coppola et al., 2005).

One of the most interesting properties of the Phox2 transcription factors is their ability to directly activate the transcription of dopamine-β-hydroxylase (Dbh) and tyrosine hydroxylase (Th) genes(Swanson et al., 1997; Yang et al., 1998; Zellmer et al., 1995; Kim et al., 1998; Lo et al., 1999). Th and Dbh control essential steps in the synthesis of noradrenaline and represent characteristic marker genes for differentiated noradrenergic neurons. The finding that Phox2 factors control cell type specification in autonomic progenitors as well as their differentiation provided a mechanism to explain noradrenergic differentiation in both the peripheral nervous system (PNS) and the central nervous system (CNS). However, as Phox2a/b are also expressed in non-noradrenergic neurons in both PNS and CNS(Morin et al., 1997; Pattyn et al., 2000a; Pattyn et al., 2000b; Pattyn et al., 1997), it is evident that the action of Phox2 on the expression of noradrenergic marker genes requires additional co-regulators that are differentially expressed or modified between noradrenergic and non-noradrenergic neurons. Indeed, a number of additional transcriptional regulators have been implicated in the control of Th and Dbh in sympathetic precursor cells, including Creb(Swanson et al., 1997; Adachi and Lewis, 2002),Ap-2α (Barallo-Gimeno et al.,2004; Holzschuh et al.,2003; Knight et al.,2003; O'Brien et al.,2004) (also known as Tfap2a), Gata2/3(Lim et al., 2000; Tsarovina et al., 2004) and Hand2 (also known as dHand) (Howard et al., 1999; Howard et al.,2000; Xu et al.,2003; Rychlik et al.,2003).

Hand2 has been identified as a basic helix-loop-helix (bHLH)transcription factor expressed in heart, autonomic nervous system and neural crest derivatives (Srivastava et al.,1995; Firulli,2003). In the autonomic nervous system of chick and mouse, Hand2 is expressed by sympathetic neurons, adrenal chromaffin cells and enteric neurons (Firulli,2003; Howard and Cserjesi,1996; Howard et al.,1999). During development, Hand2 expression is induced by BMPs but is first observed after the onset of Ascl1 and Phox2b expression in chick sympathetic ganglion priomordia(Howard et al., 2000). This finding, together with the lack of Hand2 expression in the Phox2b knockout and the maintained expression in the Gata3knockout (Tsarovina et al.,2004), implicates a function of Hand2 downstream of Phox2b and upstream of Gata2/3 in the group of BMP-induced transcription factors. Hand2 overexpression was found to induce the generation of catecholaminergic neurons from neural crest precursor cells in vitro and in vivo (Howard et al.,1999; Howard et al.,2000), acting in concert with Phox2b, Phox2a and Ascl1, which are ectopically expressed after Hand2overexpression. Also upon forced Phox2a expression, Ascl1and Phox2b are induced, suggesting cross-regulation between these transcription factors, which are thus considered a network rather than a regulatory cascade (Stanke et al.,1999; Stanke et al.,2004). This cross-regulatory action prohibits firm conclusions from being drawn on the epistatic relationships between these transcription factors, based on overexpression experiments. Nor is it possible to address the individual functions of the factors in sympathetic neuron differentiation,i.e. whether Hand2 may control specific aspects of the sympathetic neuron subtype, such as Th and Dbh expression.

A function of Hand2 in the control of Dbh expression has been suggested by the finding that Hand2 cooperates with Phox2a in activating transcription from Dbh promotor reporter constructs(Xu et al., 2003; Rychlik et al., 2003; McFadden et al., 2002; Howard, 2005; Firulli, 2003). There is also in vivo evidence implicating a role for Hand2 in the maintenance of Th and Dbh expression in autonomic neurons(Müller and Rohrer,2002). Developing chick ciliary neurons, by contrast to sympathetic neurons, express Th and Dbh only transiently. As Hand2 is not expressed by the vast majority of ciliary neuron precursors (Müller and Rohrer,2002; Lee et al.,2005), and ectopic Hand2 expression maintains the noradrenergic phenotype of ciliary neurons, Hand2 may be required for the continued expression of noradrenergic properties(Müller and Rohrer,2002).

So far, the proposed effects of Hand2 on sympathetic neuron development could not be analysed by in vivo loss-of-function approaches. Hand2^-/- mouse embryos die too early to be analysed for sympathetic ganglion development(Yamagishi et al., 1999). We have now studied the hand2^-/- zebrafish mutant hands off (Yelon et al., 2000),as embryos in this species survive for a sufficient number of days to analyse noradrenergic differentiation in cervical sympathetic ganglia (CSG) and the CNS. We observed a strong decrease in the expression of th, dbh,gata2 and tfap2a, with no effect on phox2b- and zash1a-expressing sympathetic precursor cells. As the zash1agene encodes a protein very similar to the mammalian achaete scute homolog Mash-1 (Allende and Weinberg,1994), we refer to zash1a as ascl1. Interestingly, the expression of the early neuronal marker elavl3(also known as HuC) (Kim et al.,1996) was not affected in hands off embryos. Together,these results demonstrate for the first time an essential and selective function for hand2 in the expression of noradrenergic marker genes dbh and th in the autonomic nervous system. The similar selective effect on noradrenergic but not generic neuronal differentiation in hand2 and tfap2a zebrafish mutants(Holzschuh et al., 2003),together with the reduced tfap2a expression in the hands offmutant, implies that tfap2a is a downstream effector of hand2 in sympathetic neurons.

The development of central noradrenergic neurons of the locus coeruleus is initated by a similar sequence of inducing signals, i.e. BMP-dependent ascl1 and phox2a/2b expression. The absence of effects on th and dbh expression in the locus coeruleus of hands off embryos confirms previous conclusions that after common initial development later steps of noradrenergic maturation are controlled differentially in the PNS and CNS.

Zebrafish adults and embryos were reared at the Ohio State University zebrafish facility under standard conditions at 28.5°C. Embryos were staged according to morphological criteria(Kimmel et al., 1995) and hours post-fertilization (hpf). The han^s6 hand2 mutant was kindly provided by Dr Deborah Yelon, Skirball Institute, NYU. The han^s6 mutant results from the deletion of the entire hand2 locus (Yelon et al.,2000). To better visualize internal structures such as sympathetic ganglia without interference from superficial melanophores, embryos were typically incubated in 0.2 mmol/l 1-phenyl-2-thiourea (PTU; Sigma) to prevent pigment formation.

Analysis of mRNA expression for hand2(Yelon et al., 2000), dbh,tfap2a (Holzschuh et al.,2003), th (T. Look, Dana Farber Cancer Institute), phox2a (S. Guo, USCF, San Francisco), phox2b (full-length construct by Jochen Holzschuh, University of Freiburg, Germany, unpublished), zash-1a (T. Look, Dana Farber Cancer Institute), gata2 (T. Look, Dana Farber Cancer Institute) and elavl3 (HuC)(Kim et al., 1996) was performed in whole-mount preparations according to Thisse et al.(Thisse et al., 1993), with minor modifications. A detailed protocol will be provided upon request.

After in situ hybridization, the stained embryos were transferred to 100%glycerol. After equilibration for at least 2 hours, the yolk and the tail were removed and used for genotyping. To take pictures at the microscope, the neural tube and parts of the head overlying the cervical sympathetic ganglion(CSG) were removed. In addition, the dorsal part of the somites (about one-third) was removed, using feather scalpels. In embryos stained for ascl1 the gut was also removed. Pictures were taken with a Zeiss Axiophot 2 with a SPOT digital camera. Of each specifically stained cell cluster in the CSG, one representative focal plane was imaged at 20×magnification. The image was processed and analysed using Metamorph 6.0 Software (Visitron). For planimetric analysis of the pictures, the area to be measured was manually thresholded. The size of CSG was determined as stained areas in μm², combining area measurements of all cell groups that belong to the ganglion. This morphometric method has been used before to study sympathetic ganglion development on sections of chick embryos (e.g. Schneider et al., 1999; Howard et al., 2000; Tsarovina et al., 2004). The variation between repeated quantitation of the same ganglion is below 10%. Student's t-tests were performed with at least six (6-29) embryos for 2, 3 or 4 dpf embryos and each in situ probe, respectively. As the area measurements from these whole-mount in situ hybridizations depend crucially on probe- and stage-specific parameters affecting penetration, background and staining intensity, they represent relative rather than absolute values. Thus,the data have been normalized for each age and gene analysed, comparing the data of hands off mutants to data obtained with the same probe in control wild-type or heterozygous animals.

Embryos were genotyped by PCR after Proteinase K digest of the yolk, tail and parts of the removed CNS for 10 hours at 55°C. The primers used were fgf14-exon1-F (ACATGGCAGCGGCGATTGCC), fgf14-exon1-R(AGCCCAACGCCACAAGTCCC), which gave a band of 324 bp, and hand2-UTR-F(AATTTCCCACTACGGACATTGGA) and hand2-UTR-R(AGAGACAGAAATAGATAATGAACGT), which gave a band of 225 bp. PCR conditions were as follows: 94°C for 5 minutes, 40 × 94°C for 30 seconds,56°C for 30 seconds, 72°C for 1 minute and 72°C for 5 minutes. We could not distinguish between wild-type and heterozygous fishes.

Capped synthetic hand2 mRNA was generated using a pCS2: hand2 construct (Yelon et al.,2000) and the mMessage machine system (Ambion) using standard procedures. Embryos from crosses of identified hand2 mutant heterozygotes were injected with 50-250 pg hand2 mRNA at the one- to four-cell stage. Injected embryos were raised for 2-4 dpf, fixed and processed for in situ hybridization. All experimental embryos were subsequently genotyped.

Mutations in the zebrafish hands off locus were discovered in a screen for genes affecting heart induction and patterning(Alexander et al., 1998). This locus was later shown to affect differentiation, patterning and morphogenesis of myocardium, jaw and pectoral fin (Yelon et al., 2000). Molecular analysis of two alleles indicated that the hands off locus encodes the bHLH transcription factor hand2 (Yelon et al.,2000). To study the role of hand2 in sympathetic neuron development, the han^s6 mutation is used, in which the entire hand2 locus is deleted(Yelon et al., 2000). Although mutant embryos display a severe heart phenotype, they are able to survive up to 4 dpf, the latest stage analysed in the present study. As sympathetic ganglion development in the rostral trunk is initiated just before 2 dpf, the hands off mutation provided the opportunity to investigate the role of hand2 in sympathetic ganglion development. We focus mainly on the initial stages of ganglion development in 2 and 3 dpf embryos, involving the specification and differentiation of sympathetic neurons(Guo et al., 1999; Holzschuh et al., 2001; An et al., 2002; Stewart et al., 2004).

Whole-mount in situ hybridization for dbh demonstrates the early generation of cervical sympathetic ganglia (CSG)(Fig. 1A,B, arrowheads) in wild-type embryos (Guo et al.,1999; Holzschuh et al.,2001; An et al.,2002) and a massive reduction of dbh-expressing cells in hands off mutant embryos (Fig. 1C,D, arrowheads). By contrast, the dbh-expressing cells of the medulla oblongata/area postrema(Holzschuh et al., 2001; Holzschuh et al., 2003) are not affected (asterisks). To be able to analyse CSG development in wild-type and hands off mutant embryos in more detail, yolk sac and dorsal neural tube were dissected after in situ hybridization. Sympathetic ganglia were viewed and quantified morphometrically from dorsal, combining area measurements of all cell groups that belong to the ganglion (Figs 2, 3, 4, 5 and 6; see Fig. S1 in the supplementary material).

In normal wild-type embryos, hand2, th and dbh are expressed in the CSG by 2 dpf (Fig. 2), as shown by in situ hybridization analysis. CSG are located ventrally to the notochord between somites 1 and 4 in somewhat variable arrangements. Whereas at 2 dpf several smaller cell aggregates are present(Fig. 2A-C; see Fig. S1 in the supplementary material for a higher magnification image), larger cell groups are commonly found in 3 dpf embryos (Fig. 2G-I). In addition, single cells or small cell groups were detected more caudally up to somite 6. The differentiation of sympathetic precursors at somites 3 to 4 seems to precede the onset of differentiation of more rostral cell groups, as individual embryos with preferential th/dbh expression in the caudal group were observed, whereas embryos with normal rostral expression always displayed expression in the caudal group. hand2, by contrast to th and dbh, was also expressed in enteric neurons that are located ventrolaterally and posteriorly to sympathetic ganglia. These cells seem to further migrate between the second and third day of development, as their posterior and ventral location became more evident in 3 dpf embryos (Fig. 2A,G). Enteric neurons formed a continuum of cells throughout the length of the gut (see Fig. S2 in the supplementary material) that could be clearly distinguished from the more dorsally located cells of the CSG, which were restricted to somites 1-4.

In the hands off mutant, hand2 expression was completely absent from sympathetic ganglia (Fig. 2D,J), as well as enteric ganglia, heart, branchial arches and fin primordia (Yelon et al., 2000)(data not shown). The mutant also demonstrated a very dramatic reduction in the expression of dbh at both 2 and 3 dpf(Fig. 2E,K). thexpression was also reduced to very low levels at 2 dpf in hands offmutant embryos but was expressed in a slightly larger proportion of cervical sympathetic neurons at 3 dpf (Fig. 2F,L). It was apparent that, by contrast to the complete absence of hand2 expression, th and dbh remained detectable in a small number of cervical sympathetic neurons. Quantification of the area of dbh-positive cells revealed a more than 90% reduction in the expression for both ages (Fig. 2M). th-expression was reduced at 2 and 3 dpf to <5%and 30% of controls, respectively (Fig. 2N).

To demonstrate that the effect observed in hands off embryos is caused by the lack of hand2, mutant embryos were injected with hand2 mRNA. Injection of synthetic hand2 mRNA enhanced the number of th-expressing sympathetic ganglion cells in a considerable fraction of injected hands off mutant embryos (see Fig. S3 in the supplementary material). Quantitative analysis revealed a highly significant>3-fold increase in th-positive cells in rescued hand2^-/- mutants compared with uninjected mutant embryos(Fig. 2N). Similar rescue of dbh-expressing cells was also observed (data not shown). hand2 injection in zebrafish embryos (n>100) did not result in ectopic noradrenergic neurons, by contrast to previous Hand2 overexpression in the chick embryo(Howard et al., 2000).

To address the question of whether noradrenergic differentiation, and th-expression in particular, is delayed rather than prevented in the absence of hand2, sympathetic ganglia were also analysed in 4 dpf embryos. At 4 dpf, in addition to the CSG neurons, a group of strongly th/dbh-positive cells that probably represent chromaffin cells(An et al., 2002) are observed ventrolaterally to the caudal CSG (Fig. 3A,C arrowheads). These cells are devoid of elavl3 (HuC)(Fig. 6)(An et al., 2002). In the hands off mutant, th and dbh expression was strongly reduced both in CSG neurons and in the chromaffin cells(Fig. 3B,D). The total number of th- and dbh-expressing cells was reduced to 5-10%compared with wild-type embryos (Fig. 3E; dbh 9% of controls), demonstrating a continued reduction of th and dbh expression in the absence of hand2. A similar reduction was observed when dbh was analysed specifically in the rostral CSG (up to somite 3), where chromaffin cells are absent (dbh 23% of controls). In conclusion, these findings demonstrate an essential function for Hand2 in noradrenergic differentiation during sympathoadrenergic development.

The observed reduction in the number of th- and dbh-expressing cells could either reflect a lack of sympathetic ganglion cells or an effect on the expression of these genes. The latter possibility implies that sympathetic neuron precursors would still be present but, in the absence of hand2, would be unable to further differentiate. To address this issue, phox2b and ascl1expression in sympathetic precursor cells was analysed in control and hands off mutant embryos.

phox2b expression in sympathetic ganglion primordia of wild-type embryos was equivalent to the expression of hand2, th and dbh at 2 and 3 dpf (Fig. 4A,E) but decreased to undetectable levels at 4 dpf (not shown). Ascl1 was expressed only very transiently in sympathetic ganglia(Fig. 4B,F). The ascl1signal was already reduced to <10% in 3 dpf embryos, compared with 2 dpf ganglia. Enteric neurons were also strongly positive for phox2b (e.g. Fig. 3C; see Fig. S1 in the supplementary material) and ascl1 (not shown). Due to the very strong ascl1 expression in enteric neurons, the gut was removed in embryos stained for ascl1 (Fig. 4B,D,F,H).

In hands off embryos, phox2b⁺ and ascl1⁺ sympathetic precursors were unaffected and were found in their normal rostral position in 2 and 3dpf embryos. Quantification of the areas of phox2b-positive and ascl1-positive CSG cells revealed no significant difference between hand2-deficient and control embryos. Thus, the absence of hand2 has no immediate effects on the survival of phox2b⁺/ascl1⁺-sympathetic precursor cells and the level of phox2b and ascl1 expression up to 3 dpf, the latest stage that could be analysed using these markers. Taken together, these results confirm the epistatic relationship between phox2b, ascl1 and hand2, with hand2 as a downstream member in this group of transcription factors.

The timing of Hand2 expression in avian sympathetic primordia,together with the results of Hand2 overexpression experiments suggested that Hand2 is expressed upstream of Phox2a and may control Phox2a expression (Howard et al., 2000). We have now observed that phox2aexpression was not affected in hands off mutant embryos up to 3 dpf(Fig. 5A,D,G,J,M). phox2a expression was, however, reduced to 34±6% of wild-type controls in 4 dpf embryos (mean±s.e.m.; P<0.007; n=37) (data not shown). These results demonstrate that hand2is not required for the initiation of phox2a expression. The decrease at 4 dpf may be explained by a loss of ganglion cells and/or a decreased phox2a expression. The presence of wild-type numbers of elavl3-expressing CSG neurons at 4 dpf (see below) supports the latter possibility.

The Zn-finger transcription factors Gata2 and Gata3 are essential for sympathetic neuron development in chick and mouse, respectively, and belong to the BMP-controlled network of transcription factors in sympathetic precursors(Lim et al., 2000; Tsarovina et al., 2004). Gata2 expression in chick sympathetic ganglia is detectable after the onset of Hand2 expression, suggesting a function downstream of Hand2. The continued expression of Hand2 in sympathetic precursor cells of Gata3-knockout mice is in agreement with the notion that Gata2/3 are acting downstream of Hand2(Tsarovina et al., 2004). This is now confirmed by the massive reduction of gata2 expression in the hands off mutant (Fig. 5B,E,H,K,N). Whereas at 3 dpf the reduction of gata2expression was virtually complete (about 7% of control; Fig. 5K,N), some gata2-expressing cells were detectable in the rostral part of the CSG at earlier stages (Fig. 5E,N).

The Ap-2α transcription factor has been implicated in the control of noradrenergic differentiation, because it binds to the Th and Dbh promotor and stimulates transcription from Th- and Dbh-reporter constructs. In addition, the analysis of the zebrafish mutation mont blanc/tfap2a provided in vivo evidence for a function of tfap2a in noradrenergic differentiation(Holzschuh et al., 2003; O'Brien et al., 2004). However, it was unclear at which point tfap2a interacts with members of the BMP-induced genetic cascade. The strong reduction in tfap2aexpression in hands off embryos(Fig. 5C,F,I,L,O) demonstrates a direct or indirect regulation by hand2. Therefore, Ap-2αseems to act, together with Gata2/3, downstream of hand2 in the control of noradrenergic development.

With the exception of tfap2a, the transcription factors involved in the control of autonomic neuron specification display pleiotropic roles,i.e. the elimination of Ascl1, Phox2b and Gata-2/3 resulted in the simultaneous impairment of noradrenergic and generic neuronal genes(Guillemot et al., 1993; Hirsch et al., 1998; Pattyn et al., 1999; Tsarovina et al., 2004; Holzschuh et al., 2003). Under these conditions, the resulting immature neuronal cells subsequently died by apoptosis (Pattyn et al.,1999; Tsarovina et al.,2004). To address this issue for hand2, the expression of the generic neuronal gene elav3l (HuC) was studied in the hands off mutant. Interestingly, elav3l (HuC) expression was not impaired in the hands off mutant up to 4 dpf, and even transiently increased at 3 dpf (Fig. 6A-G). Similar observations were made using whole-mount immunostaining for Hu protein expression (data not shown). The comparison of elavl3 and th/dbh expression in 4 dpf embryos (compare Fig. 3B,D with Fig. 6F) demonstrates the lack of elavl3 in the chromaffin cell cluster. More importantly, the massive decrease in th and dbh, by contrast to the maintained elavl3 expression, supports the notion of a selective function of hand2 in the development of the noradrenergic neurotransmitter phenotype.

The essential role of Hand2 in the control of noradrenergic differentiation in sympathetic neurons raised the question of whether Hand2 would also be required for the generation of central noradrenergic neurons. Although in previous expression studies Hand2expression was not detected in the locus coeruleus(Goridis and Rohrer, 2002),the possibility remained that Hand2 levels below the detection level of the in situ hybridization technique might be sufficient to maintain noradrenergic differentiation. The present study, showing that th and dbh expression in the locus coeruleus is unchanged and develops to normal size in hand2-deficient zebrafish(Fig. 7), excludes this possibility. These results indicate that the molecular control of noradrenergic differentiation differs between the PNS and the CNS, although the initial set of regulatory factors is identical, i.e. BMPs as extrinsic signals and Ascl1/Phox2 as downstream transcription factors.

The bHLH transcription factor Hand2 belongs to the network of regulatory factors implicated in the specification and differentiation of sympathetic neurons. Using the zebrafish hand2 mutant hands off, we now demonstrate an essential in vivo function of hand2 in the expression of dbh and th in sympathetic neurons. Gata2 and tfap2a expression is also strongly impaired in the hands offmutant. This suggests that Gata2 and Ap-2α may mediate, at least in part, the effects of Hand2 on noradrenergic differentiation.

Hand2 has no effect on the initial stages of sympathetic neuron development, including the expression of the proneural gene ascl1 and the pan-autonomic regulatory genes phox2b and phox2a, nor on generic neuronal differentiation as assessed by the expression of elavl3. Therefore, hand2 seems to be required selectively for initial noradrenergic but not pan-neuronal differentiation.

Hand2 has been considered an important member of the transcriptional network controlling sympathetic neuron development due to its ability to induce the generation of noradrenergic/catecholaminergic neurons in gain-of-function experiments (Howard et al., 1999; Howard et al.,2000). Overexpression is an important tool to identify candidate target genes, but in the sympathetic lineage different effects were observed upon transcription factor overexpression compared with in vivo loss-of-function approaches. Gata2/3, for instance, is essential for Th expression during normal development of sympathetic neurons but has virtually no effect on Th expression in overexpression experiments (Tsarovina et al.,2004). A further complication is the induction of upstream genes by the forced ectopic expression of transcription factors(Stanke et al., 1999; Stanke et al., 2004). Thus, it is essential to confirm the functions of candidate target genes identified from overexpression experiments using loss-of-function approaches.

The role of Hand transcription factors in the development of catecholaminergic cells has been addressed in quail neural crest cultures,using antisense oligonucleotides directed against Hand2 and Hand1 (Howard et al.,1999). Inhibitory effects were observed only with the combined interference of both Hand1 and Hand2. As Hand2, but not Hand1, seems to be involved in sympathetic neuron generation in vivo (Howard et al., 2000), it was unclear to what extent the neural crest cultures reflect the in vivo situation. To address the physiological function of Hand2 during in vivo development of sympathetic ganglia, Hand2-deficient mice were not informative due to the early death of mouse embryos between embryonic day 9.5 and 10.5, caused by impaired heart development(Yamagishi et al., 1999).

Zebrafish mutations in the hands off locus display similar malformations as observed for Hand2^-/- mouse embryos(Yelon et al., 2000; Yamagishi et al., 1999). The dramatic reduction in ventricular precursors is even more severe in hands off zebrafish than that observed in Hand2 knockout mice,possibly because of the lack of compensatory activity of a second Hand gene in this organism. However, by contrast to the mouse, the zebrafish embryos survive sufficiently long to permit the analysis of the development of cervical sympathetic ganglia, which differentiate several days before trunk sympathetic neurons (An et al., 2002). Cervical sympathetic neurons are detected in 2 dpf embryos by the expression of phox2b, phox2a, ascl1, hand2, gata2, tfap2a,th, dbh and elavl3 (HuC) between somites 1 to 4. The location of cells varies from individual to individual but generally becomes organized in two major cell groups with increasing age.

The massive loss of th and dbh expression in cervical sympathetic ganglia of hands off zebrafish embryos demonstrates an essential role for hand2 in sympathetic neuron development. hand2 is essential for th/dbh expression, not only in sympathetic ganglia but also in cells with the characteristics of chromaffin cells that become apparent in 4 dpf embryos(An et al., 2002). As the number of ascl1-, phox2b- and phox2a-expressing sympathetic precursor cells is initially not affected, a function of Hand2 in the control of th and dbh expression is the most probable mechanism of action, and indirect effects on cell migration can be excluded. A direct action is supported by the interaction and synergistic function of Phox2a and Hand2 at the Dbh promotor (Xu et al., 2003; Rychlik et al.,2003). Although evidence for the function of Hand2 at the Th promotor is still lacking, it is tempting to speculate that Hand2 and Phox2a act together to induce both Dbh and Th in this lineage. However, the results on the transactivation of the Thpromotor by Phox2a are controversial [cf. Zellmer et al.(Zellmer et al., 1995) and Yang et al. (Yang et al.,1998)] and there is also evidence for differential regulation of Th and Dbh expression in developing sympathetic ganglia(Tsarovina et al., 2004).

The loss of th and dbh expression in the hands off mutant was significantly reduced by the injection of wild-type hand2 mRNA. This result is in line with the previously shown rescue of the hands off myocardial phenotype(Yelon et al., 2000) and confirms that the defects in the hands off(han^s6) mutant are due to a deficiency of hand2.Overexpression of hand2 in wild-type embryos by injection of hand2 mRNA did not notably affect the formation of sympathetic ganglia, nor cause the generation of ectopic noradrenergic neurons. By contrast, Hand2 overexpression in the chick embryo using RCAS-retroviral vectors resulted in ectopic neurons in the peripheral nerve(Howard et al., 2000). In avian and rodent peripheral ganglia and nerves, significant numbers of pluripotent neural crest stem cells have been identified (e.g. Le Lièvre et al., 1980; Duff et al., 1991; Bixby et al., 2002) (Binder,Tsarovina and H.R., unpublished). The lack of ectopic cells in zebrafish may be due to differences in the timing or expression levels of hand2mRNA or due to lower numbers/absence of neural crest stem cells in the developing zebrafish PNS. Although it is known that some pre-migratory neural crest cells in zebrafish are multipotent(Raible and Eisen, 1994), it is not currently known whether zebrafish neural crest cells at any embryonic stage possess stem cell properties.

We have found that the expression of gata2 and tfap2arequires hand2 function and gata2/3 and tfap2a have previously been shown to be essential for sympathetic neuron development(Lim et al., 2000; Tsarovina et al., 2004; Holzschuh et al., 2003). The mechanism of action for Gata2/3 is unclear, although a context-dependent function has been suggested based on the different effects of gain- and loss-of-function approaches (Tsarovina et al., 2004). By contrast, specific AP-2α-binding sites have been identified in the upstream regions of Dbh and Th genes in different species (Greco et al.,1995; Kim et al.,1998; Kim et al.,2001). Ap-2α transactivates Th and Dbhpromotor activities in non-catecholaminergic cells(Kim et al., 2001). The genetic link between hand2 and tfap2a revealed in the hands off mutant, together with the similar sympathetic neuron phenotype of hands off and mont blanc mutants(Holzschuh et al., 2003),suggests the possibility that Ap-2α functions downstream of hand2 and may, at least in part, mediate hand2 effects on th and dbh expression.

The analysis of the hands off mutant confirmed the proposed epigenetic relationship between hand2 and gata2 but not for hand2 and phox2a. Our data suggest that the initial phox2a expression is not controlled by hand2. The previously suggested scheme with hand2 upstream of phox2a was essentially based on the difference between the onset of expression of these genes, deduced from in situ hybridization analysis. The discrepant conclusions can be explained by a comparatively low sensitivity of the phox2a in situ hybridization probe, preventing the detection of phox2aexpression at early developmental stages. The decreased phox2aexpression in 4 dpf hands off embryos may reflect a hand2requirement for continued phox2a expression or a decrease in cell numbers. To address the question of whether sympathetic neuron number was reduced in 4 dpf mutant embryos, the expression of the generic neuronal marker elav3l (HuC) was studied.

The question of to what extent generic neuronal markers would be affected in the hands off mutant was also of interest, as Hand2 overexpression in chick neural crest cells results in the expression of both noradrenergic and generic neuronal genes (Howard et al.,2000). The analysis of the hands off mutants revealed that neuronal differentiation, i.e. elavl3 expression, was not impaired up to 4 dpf. The reason for the transiently increased area of elavl3-expressing cells in 3 dpf hands off embryos is unclear but may be due to effects of hand2 on the timing of neuronal differentiation or a transient inhibitory effect. By contrast to the maintenance or even increase of generic neuronal differentiation in hands off embryos, th and dbh expression is strongly reduced already at 2 dpf in the absence of hand2. This selective loss of noradrenergic properties is very similar to the phenotype of the tfap2a zebrafish mutation mont blanc in the CSG(Holzschuh et al., 2003). It supports the notion that hand2, acting through tfap2a, may selectively control the expression of subtype-specific noradrenergic genes in the sympathoadrenal lineage.

To what extent is the function of Hand2 maintained in other lineages of the nervous system? hand2 is expressed in neurons of all parts of the peripheral autonomic nervous system, in cells of the sympathoadrenal lineage,i.e. sympathetic neurons and adrenal chromaffin cells, parasympathetic ganglia(Dai et al., 2004) (F.M. and H.R., unpublished) and enteric neurons (Wu and Howard, 2002; Dai et al.,2004). As there is a variable extent of noradrenergic differentiation in parasympathetic ganglia during development, Hand2expression and Th/Dbh expression may also correlate in parasympathetic ganglia, as observed for the chick sphenopalatine ganglion(Tsarovina et al., 2004) (F.M. and H.R., unpublished). Whether Hand2 is involved in the expression of (transient) adrenergic properties of rodent enteric neurons(Howard and Cserjesi, 1996; Wu and Howard, 2002) is presently unclear. This question could not be addressed in the present study because we did not detect th and dbh expression in phox2b⁺/phox2a⁺/hand2⁺ zebrafish enteric neurons. The th/dbh expressing group of cells in close vicinity of the CSG are considered to be chromaffin cells rather than enteric neurons (An et al., 2002; Holzschuh et al., 2001). It will be interesting to address the role of hand2 in the enteric nervous system using the hands off mutant in future studies.

Central noradrenergic neurons of the locus coeruleus have been shown to depend on similar inducers to those in the PNS. The development of locus coeruleus neurons is dependent on the transcription factors Ascl1,Phox2a and Phox2b (Morin et al., 1997; Hirsch et al.,1998; Pattyn et al.,2000b). The normal development of the locus coeruleus in hands off mutants now demonstrates that downstream of phox2bnoradrenergic differentiation is differentially regulated in PNS and CNS. This has also been shown for Gata2/3, which is required in the PNS but not in the locus coeruleus for Th and Dbh expression(Tsarovina et al., 2004).

Besides the nervous system, Hand2 is involved in the development of heart, branchial arches and limbs(Srivastava, 1999; Firulli, 2003). Although only a limited number of target genes are known so far, it is of interest that in several cases a maintenance function has been proposed. In hands offmutant embryos, the T-box transcription factor tbx5 is initiated in the developing heart and fins, but its expression is not maintained(Yelon et al., 2000). Another putative Hand downstream target in the heart is Irx4, which is initiated in Hand2-deficient mice but not maintained during development (Bruneau et al.,2000). In sympathetic neurons, a maintenance function of hand2 is now suggested for gata2 and phox2a, as their expression is more strongly affected with increasing age of the embryos. Also the presence of a small number of th/dbh-positive cells in hands off mutants may indicate that noradrenergic differentiation is initiated in newly born neurons but not maintained in the absence of hand2. An alternative possibility would be that hand2 is expressed and/or required for th/dbh expression in most but not all CSG cells. However, a definitive decision between an induction or/and maintenance function would require a conditional knockout of hand2 in differentiated sympathetic neurons.

A further common finding is that the function of Hand2 does not require DNA binding, i.e. Hand2 participates in a transcriptional complex, where the interaction with the promotor is mediated by the binding partner(s) of Hand2 rather than by Hand2 itself (McFadden et al., 2002; Xu et al.,2003; Rychlik et al.,2003). It is not clear whether the co-activation function of Hand2 is linked to the maintenance role, but it is tempting to speculate along this line. An additional characteristic observed in several lineages is the interaction of Hand2 with members of the Gata family of Zn-finger transcription factors. In the heart there is evidence suggesting a common action of Hand2 and Gata4 in the transcriptional control of a number of cardiac specific marker genes (Dai et al.,2002).

In conclusion, we demonstrate that hand2 is essential for the expression of noradrenergic marker genes th and dbh in zebrafish cervical sympathetic ganglia. The dramatic loss of dbh and th expression at early developmental stages is compatible with the notion, derived from previous work in the chick(Howard et al., 1999; Howard et al., 2000; Müller and Rohrer, 2002),that hand2 is involved in the induction and/or maintenance of noradrenergic characteristics. The effect of the hand2 loss of function on subtype-specific rather than generic neuronal genes correlates with the phenotype of the tfap2a mutant(Holzschuh et al., 2003) and implicates Hand2 and Tfap2a (Ap-2α) as selective regulators of the noradrenergic sympathetic neuronal phenotype.

We thank Sabine Richter for excellent technical assistance, Nina Tsarovina for helpful comments on the manuscript, and Jochen Holzschuh for introducing F.M. to zebrafish embryo dissection. The hands off mutant and zebrafish probes were obtained from D. Yelon, T. Look, S. Guo, J. Holzschuh and W. Driever. This work was supported by grants from the DFG (SFB269) to H.R. and NIH NS38115 to P.D.H., with additional support from NIH grant P30-NS045758.