TBLASTN searches of the GenBank database at NCBI using human β2 subunit sequences suggested that zebrafish encode two β2 homologues 43. We therefore designed primers to highly conserved sequences within the SH3 and GK protein domains, and performed 5' and 3' RACE-PCR on RNA extracted from embryos aged 1–3 dpf. Using RACE and reverse-transcriptase PCR (RT-PCR), we isolated cDNAs representing two zebrafish β2 genes, termed β2.1 and β2.2 (see Figs. 1, 2 and Additional File 1). The β2.1 gene showed a near perfect match with genomic sequences located within Genbank zebrafish chromosome 22 (NC_007133.1), whereas β2.2 matched genomic sequences found on chromosome 2 (NC_007113.1) 434445. Alternative splicing occurred at the N-terminus for both genes and within the HOOK domain for β2.1 (see Figs. 1 and 2D for descriptions of transcript variants). In the initial RACE analysis for β2.1, we recovered four β2.1_tv1 clones, and five β2.1_tv6 clones, suggesting these may be the most abundant transcript variants. One clone each was found for the β2.1_tv2, 3, 4, 5, 7 and 8 transcripts, suggesting they may be less abundant. For β2.2, we recovered two β2.2_tv1 and three β2.2_tv2 5' RACE clones. In additional RT-PCR analysis using primers closely flanking the HOOK domain, we confirmed that β2.1 transcript variants containing more than one exon between exons 6 and 10 (i.e. β2.1_tv3, 4, 7 and 8) could be detected in the RNA samples (see Additional file 2). Conversely, we confirmed that no alternative splicing occurred in the β2.2 HOOK domain (see Additional File 2). Several 3' RACE clones were of a single variant for β2.1 and a single variant for β2.2, suggesting that no alternative splicing occurred in the GK or C-terminal regions of the genes.

To better assess the conservation of the new genes, we performed a comparative search for orthologous β2 sequences in the pufferfish Tetraodon nigrovidis and Fugu rubripes genomes 454647. We identified a single β2 subunit gene in each pufferfish genome (Fig. 2A). All four teleost β2 genes contained highly conserved SH3 and GK domains characteristic of MAGUK family proteins (Fig. 2B). The size and number of exons comprising these domains are nearly identical in teleosts and mammals (Fig. 2A, yellow and green sections). Pairwise comparisons of amino acid sequences within these core (SH3 through GK) domains, using the most similar transcript variants available, indicated the zebrafish β2.1, Fugu and Tetraodon β2 genes shared ~87% identity with other vertebrate β2 genes (Figs. 2C; see Fig S3 for alignment of core regions). We extended this analysis to five teleost species by including ESTs or proteins derived from genomic sequences. Three crystallographic studies on mammalian CACNB genes have identified a total of 21 amino acid residues as critical for interaction of the β subunit with the α subunit AID (alpha interaction domain). In our five different teleost species, we observed that all 21 residues were nearly 100% conserved with mammals (see Additional File 3) 484950.

In contrast, the more divergent zebrafish β2.2 core region shares only ~62% amino acid identity (see Additional File 4 for alignment of core regions). The most extreme sequence divergence in the zebrafish β2.2 gene occurs at the 5' end of the gene, both 5' of and within the SH3 domain, although high levels of divergence exist throughout the entire protein. We identified a single EST, termed DW608729, from three-spined stickleback as a possible ortholog to the zebrafish β2.2 gene. Within the core domain, DW608729 is 74% identical to zebrafish β2.2 (EU301442) but only 52% identical to zebrafish β2.1 (EU301434). In addition, DW608729 maps to a genomic contig, AANH01005391.1, which contains sequences homologous to zebrafish β2.2 exon 1 (encoding MFCCGLGHWRREQSTY) and β2.2 exon 2 (encoding MPPKKK)(Fig. 3). In the β2.2 gene(s), regions of high divergence encompass sequences both within and outside of inferred secondary structures 32. Nevertheless, 19 of 21 β subunit residues noted for interaction with the α subunit AID are conserved in zebrafish β2.2 (see Additional File 4) 484950. High sequence divergence in β2.2 relative to other vertebrate β genes is reflected in the branch lengths on the phylogeny (Fig. 6). This pattern contrasts with that seen in the β4 group, where the two zebrafish paralogs have experienced similar rates of amino acid substitution.

Alignment of residues in the C terminus of the β subunit genes shows that sequences 3' to the GK domain are not highly conserved even among the teleost species (see Additional File 5). It has previously been noted among multiple vertebrate β subunit genes that the C terminal regions are not highly conserved. Specific functions for C terminal domains are poorly defined, although studies of truncated β2 proteins suggest the C terminus could contribute to protein function 35.

The zebrafish β2.1 and β2.2 genes undergo alternative splicing within the N-terminus and within the internal HOOK domain (Figs. 1, 2A, and 2D). The β2.1 and β2.2 genes each encode two mutually exclusive N termini (Fig. 3). Zebrafish β2.1 and β2.2, as well as the Tetraodon and Fugu β2 genes, share some 5' exons in common with mammals (Fig. 2A, lines 4 and 8; Fig. 3). In contrast, the β2.1 exon 1 sequence could not be found in any mammalian, Xenopus or chick databases, but was present in the Tetraodon and Fugu genomes, suggesting this exon may be specific to teleosts (Fig. 2A, line 6). The β2.2 exon 2 was found only in one other sequence, the stickleback EST DW608729, although the small size of this exon (6 amino acids) could make it difficult to recognize if it had diverged in other species (Fig. 2A, line 5). Additional 5' exons occur in mammals that were not observed in any teleost cDNAs or in genomic sequences available to date, although the limitation of this analysis is that complete genomic sequences are not available for all teleost species examined, and that our data relies in part on genome mining rather than extensive cDNA analysis. However, our data support the hypothesis that each of the teleost β2 loci have evolved to contain a unique combination of 5' exons, each of which is predicted to encode an alternative protein N-terminus.

Alternative splicing also occurs internally for one zebrafish β2 subunit gene. β2.1 encodes three alternatively spliced HOOK domain exons (exons 7, 8 and 9; Figs. 1 and 4). The high conservation of β2.1 exons 8 and 9 with mammalian counterparts (Fig. 2A, lines 15 and 16), and exon 8 with several other teleost genes (Fig. 4), suggests that these internal sequences could have functional relevance. The β2.1 exon 7, which appears to be unique to zebrafish, includes a premature in-frame stop codon expected to truncate the protein in the HOOK domain (Fig. 2A, line 14). In β2.1 transcripts that contain both exon 8 and 9, the reading frame is altered such that a premature in-frame stop codon truncates the protein in exon 10.

No alternative splicing was observed for β2.2 transcripts in the HOOK domain; instead, all variants encode a short exon (exon 7) specific to fish that contains several positively charged residues (Fig. 2A, line 13). Likewise, the Tetraodon and Fugu genomes contain only one recognizably homologous exon (exon 9), similar to human exon 11 (Fig. 2B, line 15). The variety of β2 transcript variants is expected to encode an array of different β2 subunit proteins in teleosts.

Remarkably, the introns that separate the first few 5' exons of the β2 loci are among the largest introns known in the zebrafish, pufferfish or human genomes. Each Tetraodon and Fugu β2 locus contains one intron over 10 kb, a size that ranks within the top 5% of the largest introns in pufferfish 51. In addition, the two pufferfish genes, the stickleback β2.2-like gene, and both zebrafish β2.2 and β2.1 each contain introns of > 5000 bp that separate 5'-most exons (Fig. 5). In pufferfish, the modal value for intron size is 79 bp, with 75% of introns < 425 bp in length 51. Given the compact nature of the pufferfish genomes 52, it was not surprising that introns of the human β2 locus exceeded the size of those in fish. Nevertheless, the trend of megasized introns in the N-terminus of β2 loci extends to the human genome. Three giant introns ranging in size from 60–100 kb each separate the human β2 exons in the N-terminus. Thus, when human exon 3 is spliced to exon 7 (beginning GSAD...), the splicing machinery must exclude over 250 kb of intronic sequence. In contrast, introns in the remaining part of the human β2 gene averaged 3261 bp in size. In humans, the mean size of introns adjoining coding sequences is 3749 bp, but 75% of introns are smaller than 2609 bp 515354. A recent analysis of the genomes of Arabidopsis thaliana, Drosophila, mouse and human indicated that the median sizes of introns separating 5' UTR (non-coding) sequence are significantly larger than introns separating coding sequence 54. Even so, the median size of human 5' UTR introns (8223 bp, 54) is still much smaller than the human 5' β2 introns, which separate coding sequences.

We used CLUSTALW software to align the β2 subunit core domain amino acid sequences and TreePuzzle and PAUP* to construct phylogenetic trees using maximum likelihood and maximum parsimony, respectively 55. By using only core domains (~SH3-GK), we minimized differences due to alternative splicing (Fig. 6). The Fugu and Tetraodon β2 subunit genes, zebrafish β2.1, zebrafish β2.2, and other vertebrate β2 genes form a monophyletic group (MLQP = 87%, MPBP < 50%). The sequence of β2.2 is substantially more diverged from other vertebrate β2 genes than is β2.1, both in amino acid replacements (see branch lengths in Fig. 6) and insertions/deletions. Portions of the gene were sufficiently divergent to be essentially random in sequence with respect to all other taxa, and the region 5' of the SH3 domain also contained an 18-amino acid deletion. No convincing region of synteny could be established between any of the teleost β2 genes and human β2, despite the fact that human chromosome 10 and mouse chromosome 2 share a region of synteny inclusive of the β2 genes. Thus, a syntenic approach was not useful to confirm or refute the orthology of zebrafish β2.1 or β2.2 with mammalian β2 genes.

To determine whether β2 genes are expressed in a stage- or tissue-specific manner in early embryogenesis, we performed RT-PCR on RNA isolated from embryos of several stages in early development. To track the expression patterns of specific transcript variants (Fig. 2D), we used forward primers specific to the 5' exons 1 or 2, and reverse primers in exon 10 (the GK domain) in RT-PCR experiments (Fig. 7). Surprisingly, amplification of β2.1 and β2.2 transcripts occurred in embryos as young as the 4-cell and 1000-cell stages (Fig. 7A). Since zebrafish zygotic transcription does not initiate until the 10^th cell division (~the 1000-cell stage) 56, the presence of mRNA in 4-cell embryos indicates that the transcripts are of maternal origin. The β2.2 transcript variant 2 was expressed steadily from the 4-cell stage through 72 hours post-fertilization (hpf). In contrast, β2.2 transcript variant 1 showed a pulse of expression in early epiboly stages. β2.1 transcript variant 6 was robustly detected from 26 hpf through at least 3 dpf. Other β2.1 transcript variants (1 and 2) were detected more sporadically or were undetectable in this assay, consistent with their rare recovery in RACE reactions. Thus, both β2.1 and β2.2 are expressed from the earliest stages of embryogenesis, but show significant heterogeneity in patterns of transcript variant expression throughout the first three days of embryogenesis.

To determine which transcript variants of the zebrafish β2 genes are expressed in the embryonic heart, we performed RT-PCR on RNA extracted from cardiac tissue at 72 hpf (Fig. 7B). These data indicate that only a subset of β2.1 and β2.2 transcript variants are expressed in the embryonic heart relative to the adult heart. In humans, at least seven β2 transcript variants are expressed in the heart, including β2a and β2e 343539. In zebrafish, the β2.1 transcript variant 6 (resembling β2e at the N-terminus) was expressed only in the embryo, whereas the β2.2 transcript variant 1 (resembling human β2a at the N-terminus) was expressed in the adult but not embryonic heart.

The expression of β2 subunits in several adult organs or tissues was assayed. In most adult tissues, we observed transcript variant-specific patterns of expression for the β2.1 and β2.2 genes (Fig. 7C). Mutually exclusive expression of β2.1 transcript variants 1 and 6 occurred in the heart and brain, respectively. Some other tissues express more than one β2.1 transcript variant. Many adult tissues express both β2.2 transcript variants 1 and 2, but a few tissues (muscle, gill and skin) expressed only β2.2 transcript variant 2. Thus, adult tissues also show significant heterogeneity in expression of β2 subunit transcript variants.

We have identified four new calcium channel β2 subunits in teleosts, including two in zebrafish and one gene each in the Fugu and Tetraodon genomes. Like the mammalian calcium channel β subunits, the four teleost β2 subunit genes encode characteristic MAGUK family proteins, with SH3 and GK domains that are highly conserved with other vertebrates. We show that several alternatively spliced transcripts arise from both β2.1 and β2.2 subunits, which are expressed in the embryo and adult but subject to both temporal and spatial regulation. Only selected transcript variants of each β2 subunit gene are expressed in the embryonic heart, and these differ from adult transcript variants. Thus, the heterogeneity of β subunits and their transcript variants in teleost species is extensive.

Alternative splicing occurs within the N-terminus as well as internally in the β2.1 and β2.2 subunit genes. Notably, each zebrafish β2 subunit gene encodes at least two mutually exclusive 5' exons, each with a separate translation initiation site. Each gene has one 5' exon conserved with mammals, and one 5' exon unique to fish. Similarly, some of the internal alternatively spliced exons are similar to those in mammals, while others are unique to teleosts. This variety of transcript variants potentially encodes an array of different β2 subunit proteins. Previous studies have demonstrated that 5' variation due to alternative splicing can be functionally significant in β subunits. When expressed in HEK 293 cells, human 5' β2 subunit variants β2a and β2e (starting with human exons 4 or 6, respectively) showed differential sub-cellular localization compared to transcript variants β2b, β2c and β2d (starting with human exons 4, 1 and 2, respectively) 57. Moreover, the five human 5' β2 subunit variants differentially affected open probability, peak current and availability of L-type channels 5758. Likewise, in Xenopus oocyte expression studies, human β4 subunits with alternatively spliced N-termini showed functionally distinct electrophysiological properties 5960. Domain swapping experiments further indicate the functional significance of the N-terminal sequences in β subunit proteins. Replacing the β1b N-terminus with β2a N-terminus created a chimeric β protein with slow inactivation kinetics 61. Conversely, replacing β2a N-terminus with the non-palmitoylated β3 N-terminus created a chimeric β protein with β3-like inactivation kinetics 62. Post-translational modifications such as phosphorylation or palmitoylation, present on exons near the N-terminus, may account for some of the functional diversity, although this functional relationship has been experimentally demonstrated for only a few genes to date 6364. Intriguingly, a recent report indicates that the length of the β2 N-terminus is, independent of sequence, an important factor in mediating the magnitude of channel modulation 58. Further functional studies using the zebrafish model system will be necessary to determine whether the array of zebrafish β2 subunit proteins indeed have different biochemical functions within the cell.

A curious feature of at least five species of teleost β2 subunit genes, as well as the human β2 subunit gene, is that some of the largest introns of the genome separate the several short, alternatively spliced 5' coding exons. Do these large introns have any adaptive significance for β2 subunit biology? In several instances, introns have been shown to incorporate enhancer or repressor elements that influence transcription 65. Expanding this idea, we propose that the unusual β2 gene structure may provide a mechanism for independent cis-regulation of each transcript variant; that is, sequences within individual introns might independently direct the tissue and temporally-specific expression of transcript variants utilizing the associated ATG-containing 5' exon. This hypothesis is supported by the observation that transcript variant-specific patterns of expression do in fact occur in both embryo and adult. However, it is increasingly becoming appreciated that, beyond regulatory elements, introns can impact mRNA metabolism in a number of other ways. Potential effects include modulating transcription rates by regulating DNA accessibility, modulating editing and polyadenylation of the pre-mRNA, and affecting nuclear export, translation and mRNA decay rates 6566.

The inclusion of different subsets of 5' exons (and their associated introns) among each of the pufferfish or zebrafish β2 subunit genes is interesting from an evolutionary standpoint. A canonical view is that variations in protein structure and function form the basis of evolutionary innovation and phenotypic divergence. However, as more genomes are sequenced and annotated, it is becoming evident that alternative splicing substantially increases the proteome in many species. (For example, about 40–50% of human and mouse genes contain alternative promoters 67). A corollary to this hypothesis is that the regulatory circuitry regulating alternative splicing or the expression of particular transcript variants is also an important source of evolutionary innovation 6768. The mechanisms by which new introns arise during evolution, and the impact of large intron size per se on cis-regulation or other adaptive phenomena, remain actively debated areas of research 697071.

Zebrafish β2 duplication may be the consequence of a genome-wide duplication thought to have occurred in teleosts approximately 300 million years ago 727374. Alternatively, duplication could be the result of an ancient region-specific duplication. The β2 subunit gene does not appear to be duplicated in either the Fugu or Tetraodon genomes. The existence of single β2 subunit homologs was of itself not particularly surprising since phylogeny and synteny data suggest that the common ancestor of zebrafish and pufferfish underwent a large-scale gene or complete genome duplication event 75. Subsequently, pufferfish may have lost many duplicates that were retained in the zebrafish 75. Data from the other β gene paralog groups (β1 and β4) are equivocal, as zebrafish have one described copy of β1 and two copies of β4 4176. The fate of most duplicated genes is the accumulation of degenerative mutations in coding or regulatory regions that leads to gene loss or silencing 73. Alternatively, genes may acquire separable patterns of expression or separable functions that then require the maintenance of functional copies of both genes in the genome (subfunctionalization) 73. Since the two zebrafish β2 genes are both robustly but differentially expressed, our data are more consistent with subfunctionalization.

Calcium is an important signal in the embryo even prior to gastrulation 77. Calcium gradients, waves and pulses have been described in the blastula and early gastrula that may represent key pattern forming events. The mechanism underlying these calcium signals, and whether they encompass voltage-gated calcium channels, is not clear. Surprisingly, both β2 subunits show maternal as well as the early zygotic expression of several transcript variants. Our recent study of two zebrafish β4 subunits, which are also expressed maternally and zygotically in the gastrulating embryo, indicates that these genes are essential for normal epiboly. Initially, we predicted that β subunits might be involved in the gastrulating embryo voltage-gated calcium channel-related functions. However, our study showed that the zebrafish β4 subunits, at least, operate independently of voltage-gated calcium channels in mediating epiboly 76.

In addition to possible early roles, we find that expression of β2.1 or β2.2 is under strong temporal control in the post-gastrula embryo and in adult tissues. Of particular interest, given the cardiac lethality observed in mouse β2-deficient embryos, is the observation of differential subsets of β2.1 and β2.2 transcript variants expressed in the embryonic and adult heart. This observation supports the hypothesis, previously proposed by others, that the heterogeneity of β subunit transcript variant expression within the heart may provide a mechanism for fine-tuning the cardiac voltage-activated current as the organism progresses through embryonic, juvenile and adult stages 3839. In addition, the expression levels of particular β2 transcript variants has also been linked to pathophysiology of heart failure in humans 787980. Clearly, the first step in interpreting the function of β subunits in any model system is simply to understand when and where the various transcript variants are expressed.

This study used the WIK zebrafish strain. Zebrafish care was provided in accordance with animal care policies of Colorado State University. Embryos were staged as described 81.

Total RNA was prepared from embryos aged 24–72 hpf using Trizol (Invitrogen, Carlsbad, CA) as per manufactuer's protocol. Poly(A)⁺ mRNA was prepared using magnetic beads (μMACS mRNA kit Miltenyi Biotec), and resuspended in RNase free dH₂O at a concentration of 1 mg/ml. RNA was stored in aliquots as an ethanol precipitate at -80 degrees Celsius. RACE Ready cDNA was created using the SMART RACE cDNA synthesis kit (Clontech, Palo Alto, CA). RACE reactions were carried out as per manufacturer's protocol using gene-specific primers designed to hybridize within the SH3 and GK domain encoding regions of β subunit genes identified by BLAST searches of the zebrafish genome (Fig. 1 and Additional File 2A). RACE PCR products were separated in ethidium bromide-stained agarose gels, excised and purified over a silica matrix column (Zymoclean Gel DNA Recovery Kit, Zymo Research). cDNA fragments were cloned into pCR 2.1-TOPO using the TOPO TA cloning method (Invitrogen, Carlsbad, CA).

We used human and zebrafish β2 sequences to identify homologous Fugu and Tetraodon β2 sequences by a multi-step process. First, we performed protein vs. translated DNA (TBLASTN) searches using human or zebrafish β2 sequences against the NCBI, Sanger Centre, UCSC Genome Bioinformatics and pufferfish genome databases 4344454647. This process identified the best matches among the Fugu or Tetraodon predicted proteins derived from contig assemblies and other sequences available for pufferfish. The homologous predicted Fugu or Tetraodon proteins were re-tested by TBLASTN against mammalian Refseq proteins to confirm they were more similar to β2 subunit genes than other β subunit family members. Next, we used the homologous Fugu or Tetraodon seqeunces in BLAT searches of UCSC Genome Bioinformatics database sequences to identify the genomic location of Fugu or Tetraodon β2 exons 3–14 and to determine conserved splice donor and splice acceptor sites 45. To identify the short N terminal exons (exons 1 and 2) or search for alternative HOOK domain exons in pufferfish, we used zebrafish or human exons in 1) BLAT searches of UCSC Genome Bioinformatics pufferfish (Fugu or Tetraodon) sequences, 2) MEGABLAST, peptide motif and BLASTP searches of NCBI, and 3) BLAST or BLAT searches of the Fugu Genome Browser, the Tetraodon Genome Browser, or the Sanger Center Ensemble Genomic databases 4344454647. Exons were numbered according to their sequential location in the genome, and probable splice donor and splice acceptor sites were identified for each exon. The pufferfish sequences were submitted to the Third Party Annotation database of NCBI. Additional teleost sequences from trout, medaka, and stickleback were obtained by searching the NCBI EST or WGS databases, or from JGI 82.

Amino acid sequences representing the β subunit "core" (i.e., from the amino acids "GSAD" just prior to the SH3 domain through the end of the GK domain) were aligned using ClustalW, varying gap opening and extension parameters; regions for which reliable homology could not be established because of indels were excluded from analysis 83. Portions of the zebrafish β2.2 gene were sufficiently divergent in sequence from all other genes (including those from Drosophila and mosquito) to be essentially random. To minimize the phylogenetic analytical problems associated with such extreme rate variation among lineages, we excluded regions of the alignment in which zebrafish β2.2 contained autapomorphic amino acid substitutions for ≥ 3 adjacent amino acid positions. Such exclusions resulted in an alignment of 252 amino acids. Maximum likelihood analyses were implemented in TreePuzzle 84 using the Müller & Vingron (VT) model of amino acid substitution with a mixed model of rate heterogeneity; amino acid frequencies, the gamma distribution parameter alpha, and the proportion of invariant sites were estimated from the data 85. Nodal support was assessed using quartet puzzling. Equally-weighted maximum parsimony analyses were carried out using PAUP* 55. A heuristic search was performed with 25 random additional replicates and TBR branch-swapping. Bootstrap proportions for clades were assessed with 1,000 pseudo-replicates. Resulting trees were rooted with Drosophila and mosquito. Teleost sequences other than pufferfish and zebrafish were not included because they only span a portion of the core domain.

As an RNA source, we used whole embryos, or individual organs or tissues dissected from adults aged approximately 1 year. Cardiac tissue from embryos was isolated on the basis of its GFP expression from the cmlc2:GFP transgenic line, in which the cmlc2 promoter drives expression of GFP only in the heart 86. RNA was extracted using the Trizol method and stored at -80°C. RT-PCR was performed for β2.1 and β2.2 genes using the Access RT-PCR System (Promega) using high-fidelity polymerases. The thermal cycling program was as follows: 48°C for 45 minutes (reverse transcription), followed by 94°C for 2 minutes (activation of PCR enzyme) and 40 cycles of 94°C for 30 seconds, 60°C for 1 minute, and 68°C for 2 minutes. A final step of 68°C for 7 minutes allowed for a final extension. Electrophoresis of the final product was performed on a 2% agarose gel containing ethidium bromide. Gels were imaged using a digital camera and imaging software (Scion Corporation, Frederick, MD).

The following Genbank sequences were used for the phylogenetic analysis:

CACNB1 genes

CACNB2 genes

CACNB3 genes

CACNB4 genes

other CACNB genes