Voltage-gated calcium channels (VGCC) mediate Ca2+ influx into cells upon activation by membrane depolarization (Catterall,2000). In excitable cells, such as muscles, neurons and endocrine cells, voltage-gated calcium channels play important roles in a variety of processes, including excitation–contraction coupling, synaptic transmission, and hormone secretion (Sheng et al.,1994; Rettig et al.,1997; Schredelseker et al.,2005). In addition, Ca2+ entering through voltage-gated calcium channels can serve as a second messenger in signaling pathways to regulate gene expression in developmental processes, including mesoderm patterning, neural induction, process outgrowth, neuronal migration, and cardiac cell differentiation (Komuro and Rakic,1992; Moorman and Hume,1993; Moreau et al.,1994; Leclerc et al.,1995,1997,2000; Brosenitsch et al.,1998; Haase et al.,2000; Palma et al.,2001; Rottbauer et al.,2001).
Each VGCC usually consists of a pore-forming α1 subunit, a cytoplasmic β subunit, an extracellular α2 subunit associated with a transmembrane δ subunit, and a transmembrane γ subunit (Dolphin,2003). The α1 subunit contains four repeated domains, each having six transmembrane segments, which is sufficient to form the voltage-dependent Ca2+ channel. Among the auxiliary subunits, the β subunit modulates membrane targeting and the electrophysiological properties of the channel (Castellano et al.,1993; Herlitze et al.,2003). Coexpression of the β subunit with the α1 subunit is able to increase membrane expression and modify the pharmacological and biophysical properties of the channel (Bichet et al.,2000). How β subunits modulate these properties depends on the combination of α1 and β subunit types. For example, β subunits facilitate the voltage-dependence of the activation of L-type VGCCs but not non-L-type VGCCs. On the other hand, β subunits cause a hyperpolarizing shift of the inactivation of non-L-type VGCCs but not L-type VGCCs (Walker and De Waard,1998). In mammals, there are four VGCC β subunit genes, CACNB1-CACNB4 (Birnbaumer et al.,1998). Each VGCC β subunit protein (CAB) has five distinctive domains (D1–D5). D2 is a SH3 (Src Homology 3) -like domain and D4 is a MAGUK-like domain, which are found in the members of the membrane associated guanylate kinase (MAGUK) protein family (Dolphin,2003). These two domains are highly conserved among CABs in different species. A motif at the beginning of D4 domain, known as β interaction domain (BID), is also highly conserved in all known VGCC β subunits and critical for the interaction between the VGCC α1 and β subunits. In contrast, D1, D3, and D5 are more variable.
Only a few nonmammalian CACNB genes have been identified (Dolphin,2003). Recently the cacnb1 gene was identified in zebrafish (Schredelseker et al.,2005; Zhou et al.,2006). The loss of function of zebrafish cacnb1 causes the immotile phenotype of the zebrafish relaxed mutant, demonstrating the essential role of this gene for normal muscle function. Zebrafish cacnb1 is expressed in skeletal muscle and the nervous system, in agreement with its mammalian orthologs, but no obvious defect in motor output from the central nervous system (CNS) was detected by electrophysiological recordings of skeletal muscle after sensory stimulation of relaxed embryos (Zhou et al.,2006). This raises the possibility that other members of the VGCC β subunit family may compensate for the loss of cacnb1 in the nervous system of relaxed embryos as other β subunits are able to partially restore VGCC function in β1-deficient muscle in mammals (Beurg et al.,1999).
As an initial step toward the characterization of signaling within zebrafish neural circuits, we identified six additional zebrafish CACNB homologs by a sequence homology search of the zebrafish expressed sequence tag (EST) and genome databases. These CACNB homologs encode proteins that share the same domain structures with mammalian CABs and are mainly expressed in the nervous system with cacnb4a also expressed in the embryonic heart. The expression patterns within the nervous system of CACNB genes overlap with that of cacnb1 suggesting that they may compensate for the loss of function of cacnb1 in the nervous system of relaxed mutants.
RESULTS AND DISCUSSION
We queried the EST (http://www.ncbi.nlm.nih.gov) and zebrafish genomic (http://www.ensembl.org/Danio_rerio/) databases with the sequence of zebrafish cacnb1c (GenBank accession no. DQ198172) and identified six additional homologous sequences from zebrafish. The predicted protein sequences along with those of human and murine β subunits were aligned and analyzed phylogenetically to determine the relationship of the zebrafish genes to the mammalian counterparts (Fig. 1). The results suggested that the β2, β3, and β4 subunit genes were duplicated in zebrafish and thus they were named cacnb2a and b, cacnb3a and b, and cacnb4a and b.
Zebrafish cacnb2a and cacnb2b
We identified an EST clone (GenBank accession no. CN324195) for cacnb2a and complete sequencing of this EST (GenBank accession no. DQ372944) revealed that this cDNA encoded a protein of 377 amino acids that lacked most of the D5 domain. Mammalian CACNB2 is alternatively spliced into many variants (Birnbaumer et al.,1998; Colecraft et al.,2002; Takahashi et al.,2003) including ones varying in the D1 domain that gave rise to distinct subcellular localizations and modulatory effects on L-type VGCC gating (Takahashi et al.,2003). However, no splice variant of β2 lacking the D5 domain was reported. Because the D5 domain is not required for interactions with the α1 subunit (Qin et al.,1997), the protein encoded by zebrafish cacnb2a could still bind the α1 subunit, but the physiological function of this β2 isoform is unknown.
The cacnb2b gene was identified from the zebrafish genome database in Zv6_scaffold1007 and Zv6_scaffold3352. Two cDNA sequences (GenBank accession no. DQ372945) for cacnb2b were cloned by reverse transcriptase-polymerase chain reaction (RT-PCR). cacnb2b.1 encodes a protein (CAB2b.1) of 598 amino acids and cacnb2b.2 encodes for an identical protein except that it has an in-frame deletion of 27 amino acids, suggesting that the two forms are derived from the same gene (Fig. 2A). Because the deletion disrupts the critical BID domain, the isoform encoded by cacnb2b.2 is predicted not to interact with VGCC α1 subunits. cacnb2a and cacnb2b are most closely related to mammalian CACNB2 genes. The proteins encoded by these two genes (zebrafish CAB2a and CAB2b.1) have 50% and 67% similarity with human CAB2.
We determined the chromosomal location of the two genes by radiation-hybrid mapping. cacnb2a was mapped to Chromosome 2, 16.37cR from EST maker fc08d03 (LOD = 10.8) and cacnb2b to Chromosome 7 at the location of EST marker fa93e09 (LOD = 16.5). In the mouse and human genomes, the CACNB2 gene is located close to the signal transducing adapter molecule (STAM) gene on Chromosome 2 (mouse) and Chromosome 10 (human), respectively (http://www.ncbi.nlm.nih.gov/Genomes/). In the zebrafish genomic sequence assembly (http://www.ensembl.org/Danio_rerio/), we found a zebrafish homolog of STAM (ENSDARG00000002127) located on the same contig (CT027716.7) containing cacnb2a. This genomic synteny suggests that cacnb2a is an ortholog of mammalian CACNB2. We did not obtain sufficient syntenic information for cacnb2b because the genomic region containing cacnb2b is poorly assembled in the genomic database but the high level of sequence homology (63.9% identity) between CAB2b.1 and CAB2a suggests that cacnb2b is also an ortholog of mammalian CACNB2.
cacnb2a was expressed prominently by the trigeminal ganglia and the epiphysis at 24 hours post fertilization (hpf; Fig. 2B,C). At 48 hpf, cacnb2a expression in the brain increased with notable expression in the optic stalk (Fig. 2D). At 72 hpf, cacnb2a was still detectable in the epiphysis and was strongly expressed in the retina (Fig. 2E). No expression of cacnb2a was detected in the spinal cord from 24 hpf to 72 hpf (not shown).
cacnb2b was expressed extensively throughout the brain at higher levels compared with cacnb2a as well as in the spinal cord at 24 hpf (Fig. 2F,G). The expression of cacnb2b increased in the brain by 48 hpf (Fig. 2H,I) and remained high at 72 hpf (Fig. 2J). In addition to the CNS, there was weaker expression of cacnb2b in the olfactory placodes (Fig. 2F,H).
In mammals the VGCC β2 subunit is expressed in many tissues including heart, brain, lung, kidney, and pancreas (Hullin et al.,1992; Perez-Reyes et al.,1992). β2 is the predominant β subunit in the heart and targeted knockout of β2 results in prenatal death due to cardiac failure in mice (Ball et al.,2002). During embryonic development of the rat, however, β2 subunit in not detected in the heart until fetal day 15 and its abundance increases steadily with the maturation of the heart until birth (Haase et al.,2000). Similarly, as assayed by in situ hybridization expression of neither cacnb2a nor cacnb2b was detected in the developing cardiac tissue during the earliest stages of heart development in zebrafish.
Zebrafish cacnb3a and cacnb3b
The zebrafish cacnb3a gene was identified from genomic contig CR936486.7 in the Zv7 genome assembly. The cDNA (Genbank accession no. DQ372946) cloned by RT-PCR encodes a protein of 439 amino acids (CAB3a) that is most closely related to mammalian CAB3 (60% similarity; Fig. 3A). However, CAB3a has a shorter C-terminal domain than human and mouse CAB3 and could represent a splice variant of cacnb3a. cacnb3a was mapped by radiation-hybrid mapping to Chromosome 23, 5.23 cR from EST marker fd02b09 (LOD = 18.0).
cacnb3b was identified from genomic contig CT573349.6. The partial cDNA (GenBank accession no. DQ372947) cloned by RT-PCR encoded for 332 amino acid protein (CAB3Lb). The predicted protein sequence shared 55% similarity with human CAB3 and 72% similarity with zebrafish CAB3a. The nucleotide sequence identity between cacnb3a and cacnb3b was 79.8%, suggesting they were duplicated genes. cacnb3b was mapped to chromosome 23, 10.09 cR from SSLP marker Z13363 (LOD = 10.9), approximately 85.16 cR from cacnb3a.
Although the D3 region is variable among the β subunits, a highly conserved motif (AKQKQKQ/S/V) within D3 is found in neuronal β1, β3, and β4 but not in β2 (Dolphin,2003). We found this conserved motif is also present in both zebrafish CAB3a and CAB3b (Fig. 3A).
Synteny between the region of mammalian and zebrafish genomes containing VGCCβ3 suggests that zebrafish cacnb3a and cacnb3b are orthologs of mammalian CACNB3. Mouse and human CACNB3 are closely linked to adenylate cyclase 6 (ADCY6), DEAD polypeptide 23 (DDX23) and Rho family GTPase 1 (RND1). Zebrafish cacnb3a is found in the contig located in chromosome 23 containing zebrafish homolog of ADCY6 (ENSDARG00000010558) and DDX23 (ENSDARG00000021945). The cacnb3b is also found closely linked to another zebrafish homolog of ADCY6 (ENSDARG00000027797) and a RND1 homolog (ENSDARG00000004218) in chromosome 23. The synteny thus suggests that cacnb3a and cacnb3b may be duplicated orthologs of mammalian CACNB3.
At 24 hpf, cacnb3a was expressed in two groups of cells located in the forebrain immediately adjacent to the olfactory placodes (Fig. 3B). In the hindbrain, it was expressed by rhombomerically distributed pairs of cells that were likely to be reticulospinal neurons (Fig. 3C). In the spinal cord, cacnb3a was expressed by large, dorsally located cells likely to be mechanosensory Rohon-Beard neurons. In addition, cacnb3a expression was also detected in segmentally distributed neurons with ventrally projecting axons. These neurons are ventral to the Rohon-Beard neurons and likely to represent commissural neurons in the dorsal spinal cord (Kuwada et al.,1990; Fig. 3D). By 48 hpf, cacnb3a was expressed in the brain (Fig. 3E) with high levels of expression in groups of cells in the hindbrain (Fig. 3F), the trigeminal ganglion and the posterior lateral line ganglion (Fig. 3G). At 72 hpf, cacnb3a was expressed throughout the brain and in the trigeminal and other cranial sensory neurons (Fig. 3H).
At 24 hpf, expression of cacnb3b could be detected in the trigeminal ganglia, two groups of sensory neurons in the otic vesicles (Fig. 3I) and Rohon-Beard neurons in the dorsal spinal cord (Fig. 3J). By 48 hpf cacnb3b was detected in the retina, the trigeminal ganglia, the otocysts and the posterior lateral line ganglia (Fig. 3K). Sections showed that cacnb3b was expressed by cells in the otocyts at 72 hpf (Fig. 3L). Thus, cacnb3b was expressed mainly by sensory neurons.
The mammalian CACNB3 gene is mainly expressed in the brain but also in a variety of other tissues, including the aorta, trachea, lung, heart, pancreas, and adrenal gland (Hullin et al.,1992; Castellano et al.,1993). CACNB3 knockout mice displayed reduced nociception (Murakami et al.,2002), which was attributed to the reduction of high voltage-gated Ca2+ currents in dorsal root ganglion cells. A similar reduction of voltage-gated Ca2+ currents was also observed in sympathetic neurons in CACNB3 knockout mice (Namkung et al.,1998). Thus, CACNB3 plays essential roles in the nervous system. The expression patterns of the two zebrafish CACNB3 homologs are consistent with the functional studies in mammals. Furthermore, the expression patterns of cacnb3a and cacnb3b are partially overlapping. Both genes are expressed in mechanosensory neurons, such as Rohon-Beard neurons, trigeminal ganglion, and posterior lateral line ganglion neurons. Therefore, the two genes could act redundantly in these sensory neurons. However, cacnb3a but not cacnb3b is strongly expressed in the brain and some interneurons in the dorsal spinal cord, while cacnb3b is expressed in otic sensory cells. The diverged expression may allow for the analysis of CACNB3 function in different sets of neurons. Zebrafish mutant gemini displays auditory–vestibular defects due to a mutation in Cav1.3a, which encodes an L-type VGCC α subunit localized at the ribbon synapses of inner ear hair cells (Sidi et al.,2004). It would be interesting to investigate whether CAB3b is preferentially associated with Cav 1.3a channels and facilitates its function in the neurons in inner ears.
Zebrafish cacnb4a and cacnb4b
Our database search yielded two CACNB4 homologs in zebrafish. cacnb4a was initially identified from genomic contig BX548038.9 and the full-length cDNA (GenBank accession no. DQ372948) was cloned by RT-PCR. The predicted cacnb4a gene product (CAB4a) consists of 485 amino acids that shares 94% similarity with human CAB4 (Fig. 4A). Two cDNA clones (GenBank accession no. BQ260456 and CK362454) were first identified for cacnb4b and the genomic sequence of this gene was located in genomic contig BX072556.9. The cDNA cloned by RT-PCR (GenBank accession no. DQ372949) encoded for a protein of 489 amino acids (CAB4b) that was most homologous with human CAB4 (88% similarity; Fig. 4A). Previous structural analyses have shown that the D1 domain of the β4 subunit was unique compared with the D1 domains of the other VGCC β subunits (Vendel et al.,2006). This β4-specific D1 domain and the AKQKQKQ/S/V motif were found in both CAB4a and CAB4b (Fig. 4A).
We mapped cacnb4a to Chromosome 9, 30.11 cR from SSLP marker Z1273 (LOD = 12.0) and cacnb4b to Chromosome 6, 4.81 cR from EST marker fb33h05 (LOD = 15.9). The genomic regions containing CACNB4 in mouse and human were highly syntenic. In these regions at least 7 genes (NMI, TNFAIP6, RIF1, NEB, ARL5A, CACNB4, and STAM2) are clustered in identical order on mouse Chromosome 2 and human Chromosome 2. We found a predicted zebrafish STAM2 homolog (ENSDARG00000005318) located immediately adjacent to cacnb4a in the Zv6 genome assembly. Thus the synteny corroborated the assignment of cacnb4a as a CACNB4 ortholog.
cacnb4a and cacnb4b were widely expressed in the CNS from 24 to 72 hpf. At 24 hpf, cacnb4b appeared to be expressed at higher levels in the brain compared with cacnb4a as assayed by in situ hybridization (Fig. 4B,C,H,I), but by 48 hpf expression of the two genes were quite similar (Fig. 4E,J). At 72 hpf, expression of cacnb4a was strong in the retina and various clusters of cells in the brain (Fig. 4F) while expression of cacnb4b was strong in the retina and presumptive cerebellum (Fig. 4K) as well as in segmentally distributed cells in the spinal cord (Fig. 4L). Beyond these differences a major distinction between cacnb4a and cacnb4b was that cacnb4a but not cacnb4b was expressed in the developing cardiac tube (Fig 4B).
Mouse CACNB4 is expressed extensively in brain with prominent expression in the cerebellum (Castellano et al.,1993). CACNB4 is essential for the function of neural circuits controlling motor behaviors because it is mutated in mouse lethargic mutants that exhibit an epilepsy-like phenotype (Burgess et al.,1997; Haase et al.,2000). Both zebrafish CACNB4 homologs are expressed widely in brain, consistent with the neural function of CACNB4. We detected a high level of cacnb4b transcript in the cerebellum in accordance with a role for CACNB4b in regulating motor behaviors. The β4 subunit is also expressed by the fetal heart in rats and precedes expression of the β2 subunit (Haase et al.,2000). Similarly in zebrafish cacnb4a was expressed by the embryonic heart during 24–72 hpf when neither cacnb2a nor cacnb2b expression was detected. Together, these data suggest that both zebrafish cacnb4a and cacnb4b are orthologs of mammalian CACNB4.
Expression of VGCC β Subunits in Zebrafish Retina
The zebrafish retina as with the other vertebrate retinas consists of three main layers: the ganglion cell layer (GCL) adjacent to the lens, the inner nuclear layer (INL) and the photoreceptors (PR; Pujic and Malicki,2004). In the zebrafish retina, the CACNB family genes are expressed in distinct patterns during embryonic development. cacnb2a expression is barely detectable at 48hpf (Fig. 2D), but by 96 hpf, it is specifically expressed in photoreceptors and the outermost half of INL while absent in the GCL and the innermost tier of INL (Figs. 2E, 5A). In contrast, cacnb2b expression in the retina was detectable as early as 24 hpf (Fig. 2F). After 48 hpf, cacnb2b is strongly expressed in the ganglion cell layer in addition to its weak and diffused expression in the INL (Figs. 2H,J, 5B). The two CACNB3 homologs exhibited partially overlapping expression patterns. At 72 and 96 hpf, both cacnb3a and cacnb3b were expressed in the GCL with additional expression of cacnb3a in the innermost tier of the INL that is immediately adjacent to the inner plexiform layer (Figs. 3G, M, 5C,D). The expression of the two CACNB4 homologs also appeared to be partially overlapping. cacnb4a expression in retina was restricted to the innermost tier of the INL (Fig. 5E) while cacnb4b was expressed in this region as well as the GCL (Fig. 5F). Interestingly, cacnb3a and cacnb4b were expressed in a similar pattern that was complementary to that of cacnb2a.
In mice, the β2 subunit is essential for the retinal formation. Knocking out β2 in the CNS leads to impaired vision, abnormal ERGs and morphological defects in the outer plexiform layer (Ball et al.,2002). Moreover, the expression of VGCC α1F in the outer segment (photoreceptor) of retina is abolished in these mutant mice. On the other hand, CNS knock-out of β1, β3, and β4 does not produce any obvious abnormality in the eye. Our results indicate that in the zebrafish retina the photoreceptors and the outermost layer of the INL express only the two β2 subunits while the other regions express multiple β subunits, suggesting that in the GCL and the inner layer of the INL these β subunits could function redundantly. This expression pattern is consistent with the loss of α1F expression in mice deficient for β2 in photoreceptors while there was little abnormality in mice deficient for the other β subunits.
Of the VGCC β subunits, cacnb1 appears to be the only one expressed in skeletal muscle in zebrafish (Table 1). This is consistent with the muscle phenotype found in relaxed mutants that is due to loss-of-function mutations in CACNB1(Schredelseker et al.,2005; Zhou et al.,2006). Of interest, CACNB1 is also expressed throughout the CNS, including the hindbrain and spinal cord, yet spinal motor output following tactile stimulation of mutants is normal. Because the hindbrain and spinal cord are sufficient for motor responses to tactile stimuli, normal neural responsiveness of relaxed mutants suggests that these regions of the CNS function normally despite the loss of cacnb1. The normal response of the CNS to tactile stimulation in relaxed mutants could be due to the redundant actions of the other six CACNB genes, which are also extensively expressed in the nervous system including the hindbrain and spinal cord. Additionally, a duplicate cacnb1 might also act redundantly. At present, neither genome database searches nor degenerate PCR targeting the β1-specific domain produced any sequence other than the known cacnb1 gene (data not shown) suggesting a lack of duplication of cacnb1 in zebrafish. It is notable that we were able to identify only one cacnb1 gene in Takifugu rubripes genome as well, which suggests that there may not be a duplication of CACNB1 in teleosts. However, because the zebrafish genome has not been completely assembled, we cannot rule out that a duplicate cacnb1 might exist.
Table 1. Expression Patterns of Mammalian CACNB Genes and Zebrafish Homologs
References are as follows: 1: Powers et al.,1992; 2, Ruth et al.,1989; 3: Pragnell et al.,1991; 4: Zhou et al.,2006; 5: Hullin et al.,1992; 6: Perez-Reyes et al.,1992; 7: Massa et al.,1995; 8:Hullin et al.,1992; 9: Castellano et al.,1993a; 10: Collin et al.,1994; 11: Murakami et al.,1996; 12: Castellano et al.,1993b; 13: Burgess et al.,1997.
The zebrafish VGCC β subunit homologs are similar with their mammalian counterparts in sequence, genomic synteny, and expression pattern. Yet significant divergence of this gene family exists in zebrafish. With the possible exception of β1, the other β subunits appear to have undergone gene duplication and specification during evolution, as each pair of putative duplicates share significant sequence similarity but only partially overlapping expression patterns. The expression of the duplicates in different cells may be useful for revealing the specific function of CACNB genes in the cells that selectively express each of the duplicates.
Fish Breeding and Maintenance
Zebrafish (Danio rerio) were bred and maintained in a breeding facility following established procedures that meet the guidelines set forth by the University of Michigan Animal Care and Use protocols. Embryos were collected after natural spawns, kept at 28.5°C, and staged according to hours post fertilization (hpf; Westerfield,1995).
A BLAST search for zebrafish CACNB gene homologs was done using the sequence of zebrafish CACNB1c (GenBank accession no. DQ198172). The GenBank database was searched for EST clones and the zebrafish genome database (http://www.ensembl.org/Danio_rerio/) for genomic sequences. Each EST clone was completely sequenced to determine whether it contained a complete open reading frame. The acquired genomic sequences were analyzed with Genscan software (http://genes.mit.edu/GENSCAN.html) to determine potential exons, which were used to design primers for RT-PCR.
RT-PCR and Cloning
Total RNA was isolated from 24 to 30 hpf embryos by using Tri-reagent (Invitrogen, Carlsbad, CA) and reverse-transcribed with oligo dT primers and Superscript II reverse transcriptase (Invitrogen) following the manufacturer's instructions (Superscript II manual, version 11-11-203). The PCR products were gel-purified, cloned into the pGEM T-easy vector (Promega, Madison, WI) and sequenced at the University of Michigan Sequencing Core. Sequence alignment (Clustal W method) and construction of phylogenetic tress (unrooted) were done with the Lasergene software (DNAStar, Madison, WI). The neuronal isoform of cacnb1 (CAB1c) was included as an outgroup in the phylogenetic analysis. The length of each pair of branches represents the distance between sequence pairs, while the units at the bottom of the tree indicate the number of substitution events. The dotted lines indicate the negative length of branches.
For degenerate RT-PCR of CACNB1 gene, the following primers were used for the amplification of β1-specific D2-D4 domain: forward primer, 5′-CCACCTCCAACTCCTTCGTNMGNCARGG, reverse primer, 5′-CATGGTCCGGTTCAGCARNGGRTTNGG. Other RT-PCR Primer sequences are available upon request to the authors.
In Situ Hybridization and Sectioning
In situ hybridization was carried out following standard protocols (Li et al.,2004). To prevent pigmentation after 24 hpf, embryos were transferred to water containing 0.2 mM of 1-phenyl-2-thiourea at 20 hpf and fixed at appropriate stages. The antisense digoxigenin (DIG) -labeled probes for zebrafish CACNB genes were synthesized in vitro from the cloned cDNA sequences. The sense probes were used as negative controls and produced no significant signals (not shown). In some cases, after in situ hybridization, embryos were washed in phosphate buffered saline, dehydrated with 25%, 50%, 75%, 85%, 95%, and 100% ethanol and embedded in JB-4 plastics (Polysciences). Sectioning was performed with a Leica RM2265 automatic microtome. For double labeling, in situ hybridization was carried out using Fast Red (Roche Applied Science) as a coloration substrate, followed by immunohistochemical labeling with anti-acetylated tubulin antibody (1:500; Sigma) and secondary anti-mouse IgG (1:2,000; Molecular Probes). Fluorescence imaging was acquired with a Leica SP5 confocal microscope.
Radiation Hybrid Mapping
The LN54 radiation hybrid panel was used for physical mapping of zebrafish sequences (Hukriede et al.,1999). Mapping primers were designed according to the genomic sequence information acquired by searching the zebrafish genome database. Primer sequences are available upon request to the authors.