Zebrafish tbx5 paralogs demonstrate independent essential requirements in cardiac and pectoral fin development

Authors


Correspondence to: Deborah M. Garrity, Colorado State University, Department of Biology, 1878 Campus Delivery, Fort Collins, CO 80523. E-mail: deborah.garrity@colostate.edu

Abstract

Background: T-box genes constitute a large family of transcriptional regulators involved in developmental patterning. Homozygous mutation of tbx5 leads to embryonic lethal cardiac phenotypes and forelimb malformations in vertebrate models. Haploinsufficiency of tbx5 results in Holt-Oram syndrome, a human congenital disease characterized by cardiac and forelimb defects. Homozygous mutation of zebrafish tbx5a leads to lethal defects in cardiac looping morphogenesis, blocks pectoral fin initiation, and impairs outgrowth. Recently, a second zebrafish tbx5 gene was described, termed tbx5b. Results: Our phylogenetic analyses confirm tbx5b as a paralog that likely arose in the teleost-specific whole genome duplication ∼270 MYA. Using morpholino depletion studies, we find that tbx5b is required in the heart for embryonic survival, and influences the timing and morphogenesis of pectoral fin development. Because tbx5a hypomorphic mutations are embryonic lethal, tbx5a and tbx5b functions in the heart must not be completely redundant. Consistent with this hypothesis, simultaneous depletion of both tbx5 paralogs did not lead to more severe phenotypes, and injection of wild-type mRNA from one tbx5 paralog was not sufficient to cross-rescue phenotypes of the paralogous gene. Conclusions: Collectively, these data indicate that, despite similar spatio-temporal expression patterns, tbx5a and tbx5b have independent functions in heart and fin development. Developmental Dynamics 242:475–492, 2013. © 2013 Wiley Periodicals, Inc.

INTRODUCTION

Genes of the T-box superfamily of transcription factors are present throughout nearly all of the animal Kingdom, from primitive invertebrates (including Hydra and jellyfish) through all tetrapods (Agulnik et al., 1995; Technau and Bode, 1999). T-box genes encompass five subfamilies: Brachyury, T-brain1, Tbx1, Tbx2, and Tbx6 (Papaioannou, 2001). Members of the same subfamily most likely resulted from duplication events throughout evolution. T-box genes have diverse functions in development, including regulation of gastrulation, patterning of body axes, morphogenesis, and organ development (Stennard and Harvey, 2005). In general, T-box genes help to specify the tissue in which they are expressed (Naiche et al., 2005; Greulich et al., 2011). T-box genes are associated with human disorders including DiGeorge syndrome and cancer, among others (Naiche et al., 2005). They may play a role in metastasis of certain cancers by promoting epithelial–mesenchymal transition. Moreover, recent studies indicate that expression of Brachyury in early tumor stages predicts poor patient prognosis (Naiche et al., 2005; Kilic et al., 2011; Suzuki et al., 2011). All genes of this family include a highly conserved “T-box” region that encodes a DNA binding domain which regulates expression of target genes. The transcriptional activation domain of T-box proteins is typically located in the C-terminus, outside of the T-domain (Zaragoza et al., 2004). The specificity of T-box proteins is conferred in part by their interactions with other transcriptional regulators, in some cases involving an N-terminal interaction domain distinct from the T-domain itself (Wilson and Conlon, 2002). In cardiac development, members of two T-box subfamilies are required for normal heart formation, including Tbx1, Tbx18, and Tbx20 of the Tbx1 subfamily, and Tbx2, Tbx3, and Tbx5 of the Tbx2 subfamily (Plageman and Yutzey, 2005; Hoogaars et al., 2007).

Tbx5 is one of the earliest genes to be expressed in the cardiac field during embryonic development (Bruneau et al., 1999). Mutation of mammalian Tbx5 produces dominant phenotypes affecting cardiac septation as well as electrophysiological defects including atrioventricular conduction block (NewburyEcob et al., 1996; Bruneau et al., 2001). Holt-Oram syndrome (HOS) is a human congenital disease that results from mutations in Tbx5 (Mori and Bruneau, 2004; Piacentini et al., 2007). HOS is characterized by cardiac and forelimb morphological defects, and progressive cardiac conduction disease. Mammalian models of HOS indicate an essential role for Tbx5 in cardiac morphogenesis (Basson et al., 1999; Showell et al., 2004). In zebrafish, the tbx5a gene is expressed in the embryonic heart, dorsal optic cup, and pectoral fins (Begemann and Ingham, 2000). The tbx5a/heartstrings (hst) mutant displays a loss-of-function phenotype that recapitulates aspects of HOS in both the heart and forelimb (Garrity et al., 2002). Despite overtly normal heart tube formation and slight bradycardia, the hearts of homozygous tbx5a mutant embryos later fail to loop and progressively deteriorate, leading to heart failure and lethality. The cardiac phenotypes in tbx5a mutants differ somewhat from those reported for Tbx5 mutants in mouse, which particularly affect formation of the venous end of the developing heart (Bruneau et al., 2001; Ghosh et al., 2009; Parrie and Garrity, unpublished data). In addition, homozygous tbx5a mutation prevents initiation of pectoral fin formation (Garrity et al., 2002). Strikingly, conditional deletion of mouse Tbx5 in the forelimb buds similarly results in a complete lack of forelimb bud formation (Rallis et al., 2003).

Recently, a second zebrafish tbx5 paralog has been described, termed tbx5b (Albalat et al., 2010). Because a tbx5b-like gene is present in all known teleost genome sequences, this gene likely arose during the teleost-specific genome duplication event approximately 270 million years ago (Hurley et al., 2007); however, phylogenetic analyses have yet to confirm this hypothesis (Albalat et al., 2010). Based on in situ hybridization (ISH) studies, Albalat et al. (2010) reported that zebrafish tbx5b, similar to tbx5a, is expressed in the heart and dorsal embryonic eye. ISH was not sufficient to detect tbx5b expression in either the caudal-most portion of the lateral plate mesoderm, which contributes to pectoral fin mesenchyme, or within the forming pectoral fins themselves (Albalat et al., 2010).

The overlapping expression of both zebrafish tbx5 paralogs in the embryonic heart raised the possibility that these genes might share certain functions during cardiogenesis. However, Tbx5b functions cannot be fully redundant with those of Tbx5a because depletion of the latter by mutation or morpholino knockdown leads to lethal cardiac phenotypes (Ahn et al., 2002; Garrity et al., 2002), despite the presence of wild-type Tbx5b. What, then, is the function of tbx5b in the embryonic heart? Following gene duplication, paralogous genes typically do not retain identical functions over extended evolutionary time (Ohno, 1970). A more common outcome is that one paralog becomes a pseudogene, or that the two paralogs become different in function by means of the processes of subfunctionalization or neofunctionalization (Ohno, 1970; Taylor and Raes, 2004). We considered here the potential functional consequences of having two distinct tbx5 genes expressed in the zebrafish embryo. It is possible that, although coordinately expressed in the heart, the functions of the tbx5 paralogs in this organ are principally independent. If true, then the phenotypes of embryos depleted for tbx5b should differ from those of tbx5a mutants. An alternative possibility is that, although the functions of tbx5a and tbx5b at postlooping stages are likely nonredundant (based on lethality of tbx5a mutants), the genes might still share an essential earlier function such as facilitating cardiomyocyte specification or heart tube assembly. If this hypothesis were correct, then the dual depletion of both paralogs could create more severe (and potentially earlier) phenotypes than depletion of single tbx5 genes. Moreover, co-injection of mRNA from the paralogous tbx5 gene might be able to partially rescue defects of tbx5 singly depleted embryos.

To examine the functional consequences of tbx5 duplication for zebrafish development, we first confirmed the status of zebrafish tbx5b as a tbx5 paralog resulting from the teleost-specific whole genome duplication event ∼270 mya and examined its amino acid divergence from other tbx5 genes using phylogenetic analysis. Next, we determined the embryonic functions of tbx5b during development using both morpholino knockdown and rescue approaches. Collectively, the findings in this report support a model of independent essential functions for Tbx5a and Tbx5b in cardiac and pectoral fin morphogenesis.

RESULTS

Zebrafish Tbx5b Protein Structure

We isolated and sequenced full-length tbx5b mRNA from WIK wild-type Danio rerio embryos (Fig. 1; GenBank accession no. HQ822122). The predicted Tbx5b protein is 430 amino acids in length, 55 residues shorter than D. rerio Tbx5a. This reduced protein length reflects a substantially shorter, highly diverged C-terminus. Within the T-box region, most T-box genes typically display 95–99% amino acid identity (Holland et al., 1995; Bamshad et al., 1997; Horb and Thomsen, 1999; Papaioannou, 2001). However, the D. rerio Tbx5b T-box domain showed 83% sequence identity with D. rerio Tbx5a, and 87.5% with H. sapiens (Fig. 1A). Therefore, the degree of conservation within the D. rerio Tbx5b T-domain was notably lower than reported for other T-box gene family members. A comparison of the genomic structure at the D. rerio Tbx5a and Tbx5b loci showed that exon sizes of the two tbx5 genes were quite similar within the T-box region (Fig. 1B), as expected from the high degree of T-box exon size conservation across long evolutionary timescales (Wattler et al., 1998). Outside of the T-box domain, however, the number and size of exons were substantially different between the two genes, with tbx5b having fewer, shorter exons.

Figure 1.

Alignment and genomic structure of zebrafish Tbx5a and Tbx5b. A: The Tbx5b T-box (AA 47–246, underlined in red) was highly conserved (86.5%), but the paralogs showed an overall amino acid identity of only 55.2%. Colors indicate similar types of amino acids: green, polar; orange, nonpolar; blue, basic; purple, acidic. B: Schematic comparison of genomic structure of the zebrafish tbx5 paralogs. Numbers in blue boxes are exon sizes in basepairs. Blue arrow denotes the location of the start codon. Blue arrowheads indicate location of the wild-type termination signal. Red underline identifies the region encoding the T-box DNA-binding domain. Orange arrow indicates the location of the nonsense mutation in the tbx5am21 allele.

tbx5a and tbx5b Are Paralogs From the Teleost Whole Genome Duplication

Our phylogenetic analysis of 35 Tbx4 and Tbx5 sequences from 16 species confirms that zebrafish tbx5b is most closely related to the other teleost tbx5b paralogs, as reported previously (Albalat et al., 2010). More generally, our analysis recovered a tbx4 clade and a tbx5 clade, the latter of which consisted of a teleost tbx5a clade, a teleost tbx5b clade, and a nonteleost tbx5 clade. Relationships among the tbx5as, tbx5bs, and tbx5s conflicted with those reported by Albalat et al. 2010, and both our initial results and those of Albalat et al. were inconsistent with the teleost tbx5a and tbx5b genes dating back to the teleost-specific whole genome duplication. For example, Albalat et al. found the fish Tbx5a clade to be more closely related to the nonfish Tbx5 clade rather than the fish Tbx5b clade. This is inconsistent with the fish genes having resulted from the teleost whole genome duplication event. However, a potential problem in attempting to align all 35 Tbx4 and Tbx5 sequences was that the protein regions that could be readily aligned among these genes (i.e., the T-box domain and very few other regions) were highly conserved across species, thus severely limiting the phylogenetic signal present in the dataset. To partially overcome this lack of signal, we aligned and analyzed just the Tbx5 sequences, resulting in a longer alignment with fewer invariable amino acid positions. The relatively low bootstrap proportions for all nodes (i.e., ≤70%) reflect low numbers of shared amino acid substitutions in a protein sequence composed of domains that are either extremely highly conserved or extremely divergent. Phylogenetic analysis of these 19 Tbx5 sequences from 14 species confirmed that the tbx5a and tbx5b sequences do indeed result from a gene duplication event in the common ancestor of our focal teleosts, consistent with the teleost-specific whole genome duplication (Hoegg et al., 2004; Hurley et al., 2007) (Fig. 2).

Figure 2.

Maximum likelihood phylogenetic tree of Tbx5 amino acid sequences from 14 vertebrate species. Numbers at nodes are bootstrap proportions. Teleost tbx5b and tbx5a clades are each other's closest relatives, consistent with paralogs formed during the teleost-specific whole genome duplication. Danio rerio tbx5b has the highest degree of amino acid sequence divergence.

Sequence Divergence of tbx5b

Across the five teleosts included in our analyses, rates of amino acid replacement were consistently higher in Tbx5b than in Tbx5a. Within the Tbx5b paralog group, rates of amino acid replacement were substantially higher in the zebrafish than in any of the other teleosts. More generally, relative rates of amino acid replacement among species differed between Tbx5a and Tbx5b; although the zebrafish had the most rapidly evolving Tbx5b sequence, it also had the most slowly evolving Tbx5a sequence. Accelerated rates of amino acid replacement in the Tbx5b protein are consistent with relaxed purifying selection (i.e., decreased functional constraint) and/or positive selection for novel/modified protein function following gene duplication.

Morpholino Knockdown and Analysis of Cardiac Phenotypes

Whole-mount ISH indicated tbx5b is expressed in the embryonic heart (Fig. 3A), as others have reported (Albalat et al., 2010). Two different tbx5b RNA probes also detected weak signal in the developing pectoral fins (Fig. 3B and data not shown), suggesting possible low levels of expression. However, the level of fin expression relative to background made conclusions about fin expression tentative. To functionally determine whether depletion of Tbx5b is associated with cardiac and/or fin phenotypes, morpholinos that block tbx5b translation (tbx5bMO) or interfere with tbx5b pre-mRNA splicing (tbx5bSD-MO) were designed for tbx5b knockdown. To validate the efficacy of tbx5bMO, we created a Tester mRNA consisting of the tbx5b morpholino target sequence fused to sequence encoding the GFP reporter. Whereas embryos injected with 1,000 ng/μl of capped Tester mRNA displayed green GFP fluorescence, 87% of embryos co-injected with Tester mRNA along with 100 μM of tbx5bMO lacked any detectable GFP fluorescence by 24 hours postfertilization (hpf), indicating that tbx5bMO efficiently targeted the expected mRNA sequences (Fig. 3C,D). The second morpholino, tbx5bSD-MO, was designed to block the splice donor junction of exon 2, an event predicted to delete exon 2 from the final mRNA transcript, thereby changing the reading frame and introducing a stop codon shortly thereafter. To validate the efficacy of tbx5bSD-MO, primers located in flanking exons were used in reverse transcriptase-polymerase chain reaction (RT-PCR) reactions (Fig. 3E). In RNA samples extracted from 24 hpf wild-type embryos, PCR experiments amplified a 513 bp product representing wild-type tbx5b message. In RNA samples extracted from 24 hpf tbx5bSD-MO morphant embryos, the 513 bp product was undetectable, whereas a smaller 417 bp product appeared. Sequencing of the 417 bp product indicated a specific deletion of 96 basepairs corresponding to exon 2, precisely as predicted. In 42 hpf samples, the 513 bp product was barely detectable, whereas the 417 bp product remained abundant. These data suggest that that tbx5bSD-MO provided a complete knockdown of tbx5b through early heart tube stages, and that by day 2 of development wild-type transcript production had not recovered to any substantial degree.

Figure 3.

tbx5b expression and confirmation of morpholino knockdown using green fluorescent protein (GFP) -tagged mRNA containing the tbx5b morpholino (MO) target site or reverse transcriptase-polymerase chain reaction (RT-PCR). A,B: tbx5b in situ hybridization of 36 hpf embryos, showing (A) a ventral view with cardiac expression and (B) a dorsal view with signal in the pectoral fins. C,C′: The efficacy of tbx5b morpholino knockdown was confirmed using a Tester mRNA constructed by inserting the tbx5b MO target site in front of EGFP coding sequences. At 36 hpf, 100% of control embryos injected with 1,000 ng/μl Tester mRNA displayed normal body morphology and expressed GFP (n = 30). D,D′: Embryos co-injected with 100 μM morpholino and 1,000 ng/μl Tester mRNA displayed normal body axes and central nervous system development, but (D′) 87% failed to detectably express GFP (n = 30), indicating that the tbx5bMO efficiently targeted the expected binding site. E: RT-PCR–amplified RNA from buffer-injected control (wild-type [WT]) or tbx5bSD-MO (MO) injected embryos, with eF1a loading controls. Expected amplicon sizes are 513 bp for wild-type transcripts and 417 bp for transcripts lacking exon 2. M, marker.

By 48 hpf, 79% (121/153) of embryos injected with 100 μM of ATG-blocking tbx5bMO or splice-blocking morpholino tbx5bSD-MO exhibited cardiac phenotypes, including defective cardiac looping, reduced contractility, and abnormal chamber morphology (Fig. 4). By 72 hpf, cardiac phenotypes for tbx5b morphant embryos overtly resembled those of homozygous tbx5a mutants (Fig. 4A and Garrity et al., 2002), including a collapsed, dysmorphic heart surrounded by pericardial edema. The cardiac “looping angle” has been defined as the angle created between the plane of the cardiac atrioventricular junction (AVJ) and the embryo anteroposterior (A/P) axis, as diagrammed (Fig. 4B) (Chernyavskaya et al., 2012). By 48 hpf, wild-type larvae undergoing normal morphogenesis had progressed to an average looping angle of 14° (Fig. 4C). In contrast, homozygous tbx5a mutant, tbx5bMO morphant and tbx5bSD-MO morphant embryos all demonstrated significantly larger average looping angles of over 54°, indicating greatly reduced looping. Note that, although the chambers in mutant hearts frequently remain linearly aligned, the plane of the AVJ still tilted in the direction expected for normal looping; reversed looping was not observed in any samples. At 42 hpf, heart rates of tbx5bMO or tbx5bSD-MO embryos were significantly reduced compared with controls, but not significantly different from tbx5a mutants or knockdown of tbx5b in tbx5a mutant embryos (Fig. 4D). Tbx5b-depleted embryos died by 6–7 days postfertilization. Thus, depletion of tbx5b by either of two morpholinos reveals an essential role for this gene in cardiac morphogenesis.

Figure 4.

tbx5a mutants and tbx5b morphants display cardiac defects at 48 hours postfertilization (hpf). At 48 hpf, the looping angles (defined as the angle created between the plane of the cardiac atrioventricular junction [AVJ] and the embryo anteroposterior [A/P] axis) were determined for tbx5a mutants and tbx5b morphants. A: Representative green fluorescent protein (GFP) -expressing hearts imaged in the Tg(myl7:EGFP-HsHRAS) transgenic line (ventral views). B: Diagram of the looping angle as measured for the hearts in (A). C: The average looping angle determined from at least 14 embryos per genotype. D: Heart rates were measured at early chamber morphogenesis stage (42 hpf), chosen as the optimal time for balancing maximal cardiac function before end-stage degeneration. The average heart rate in beats per minute (BPM) for 42 hpf embryos is presented (at least 11 embryos per genotype). Relative to controls, the mutants and morphants exhibited a statistical difference of **P < 0.001, or *P < 0.01. (a= anterior, defined here as 0°, p = posterior of the embryo).

To determine if double knockdown of tbx5a and tbx5b would result in noticeably earlier or more severe cardiac defects (such as a smaller or absent heart) than single tbx5 knockdowns, we injected tbx5bMO into tbx5a mutants or co-injected tbx5bMO and tbx5aMO. The end-stage cardiac phenotypes of zebrafish tbx5a/tbx5b double knockdown embryos were grossly similar to those of tbx5a mutant or tbx5b morphant embryos (Fig. 4A), displaying initially overtly normal heart tubes that later degenerated to dysmorphic, weakly contracting hearts surrounded by pericardial edema. The looping angle for embryos depleted for both tbx5a and tbx5b was 57°, indicating no difference from loss of either tbx5 alone. Thus, based on both morphological (heart looping) and functional (heart rate) parameters, knockdown of both zebrafish tbx5 paralogs together did not result in a noticeably earlier or more severe cardiac phenotype than observed for tbx5 singly depleted embryos. These data therefore argue against extensive overlap of cardiac functions between the two paralogs.

Comparative Analysis of Phenotypes Based on Molecular Markers of Cardiac Differentiation

We next used whole-mount ISH using several molecular markers to assay cardiac differentiation of tbx5b-depleted embryos. Although dual knockdown studies did not suggest extensively redundant functions, we were curious whether analysis of molecular markers on single mutants would support this conclusion, and, therefore, compared all samples with tbx5a phenotypes. To examine the effects of tbx5b knockdown on chamber specification, we assayed for the expression of atrial myosin heavy chain (amhc) and ventricular myosin heavy chain (vmhc) (Yelon et al., 1999; Berdougo et al., 2003). In tbx5b morphant embryos, chamber specification had occurred normally by 42 hpf (data not shown), as previously reported for tbx5a mutants (Garrity et al., 2002). Natriuretic peptide precursor A (Nppa) is a known target of zebrafish Tbx5a (Camarata et al., 2010) as well as a conserved Tbx5 target for other vertebrates (Bruneau et al., 2001; Hiroi et al., 2001; Habets et al., 2002; Garg et al., 2003; Stennard and Harvey, 2005). At 36 hpf, tbx5b morphants expressed nppa in both chambers in a pattern similar to controls, whereas tbx5a mutants expressed nppa only in the anterior portion of the ventricle (Fig. 5A–C). We investigated bmp4 expression as a marker for differentiation within the sinus venosus (SV) and AVJ. bmp4 expression in the SV at this time occurs concurrently with the addition of cells derived from the second heart field to the venous pole of the heart, and bmp4 expression in the SV is missing in mutants such as isl-1 that that lack this addition (de Pater et al., 2009). By 42 hpf in wild-type embryos, bmp4 expression becomes restricted to the AVJ (Fig. 5D, arrow) (Walsh and Stainier, 2001). Again, we found that tbx5b morphants expressed bmp4 normally in the AVJ, in contrast to inappropriately expanded bmp4 expression throughout the ventricle previously reported for tbx5a mutant hearts (Fig. 5D–F) (Garrity et al., 2002). Thus for these two markers (nppa and bmp4), tbx5b morphants showed normal expression at stages when tbx5a mutant expression was altered.

Figure 5.

Differential expression of putative target genes in tbx5a mutants versus tbx5b morphants. A–C: At 36 hours postfertilization (hpf), expression of nppa (a conserved Tbx5 target) was reduced in the atrium of tbx5a mutant embryos, but expressed throughout both chambers in tbx5b morphant embryos, similar to controls. D–F: At 42 hpf, bmp4 was misexpressed throughout the ventricle of homozygous tbx5a mutant embryos. In tbx5b morphant embryos, bmp4 expression was similar to wild-type. G–I: At 42 hpf, soon after looping has begun, wild-type embryos primarily expressed hand2 in the ventricle, but tbx5a mutant and tbx5b morphant embryos showed expanded expression encompassing both chambers. J–L: At 42 hpf, wild-type embryos showed vcana expression restricted to atrioventricular junction (AVJ) boundary, but mutant and morphant embryos showed expanded expression throughout the ventricle and atrium. Arrows, AVJ boundary; arrowheads, sinus venosus. Dotted lines flank the cardiac tissue.

However, tbx5b morphants themselves were not uniformly normal for all markers of cardiac differentiation. The hand2 transcription factor plays a role in the differentiation and polarization of myocardial precursors as well as in cardiac extracellular matrix assembly (Trinh et al., 2005; Barnes et al., 2011). versican a (vcana), a structural protein involved in cell adhesion and migration that interacts directly with the extracellular matrix (Matsumoto et al., 2003). Both hand2 and vcana expression were expanded, encompassing both chambers of tbx5b-depleted embryos, indicating potential chamber morphogenesis problems in these embryos. In contrast, wild-type embryos restricted hand2 expression to the ventricle only (Fig. 5G–I), and vcana expression to the AVJ only (Fig. 5J–L). Thus, for these latter two markers, tbx5b-depleted embryos generated abnormal phenotypes similar to those reported for tbx5a-depleted embryos (Garrity et al., 2002). This survey of gene markers indicated that on a molecular level, the tbx5b cardiac phenotypes were distinct from both tbx5a mutants and wild-type embryos. We, therefore, hypothesize that the functions of tbx5b in the heart are nonredundant with the functions of tbx5a.

Tbx5 Does Not Regulate Cardiomyocyte Number in Zebrafish

In various species, Tbx5 function impacts the numbers of cardiomyocytes as development proceeds. To evaluate cardiomyocyte numbers in zebrafish, we used a transgenic line in which the heart-specific myl7 (formerly cmlc2) promoter drives expression of the nuclear DsRed reporter gene (Tg(myl7:nDsRed2/myl7:EGFP) [Mably et al., 2003]). Cardiomyocytes were quantified in 48 hpf transgenic fish of several tbx5 genotypes, including: homozygous wild-type, tbx5b morphants, homozygous mutant tbx5am21, tbx5a morphants, tbx5a+b double-morphants, or tbx5bMO injected into tbx5a mutant embryos (Fig. 6). We observed no significant differences in total cardiomyocyte numbers at 48 hpf among any of these fish (P = 0.4), suggesting that normal cardiac cell numbers had developed for all genotypes. Thus, analysis of single mutant phenotypes suggests that neither tbx5b nor tbx5a is required for normal cell numbers in the heart; furthermore, analysis of double knockdown phenotypes suggests that these two genes do not share a redundant role in regulating cardiac cell number in the heart tube through early chamber morphogenesis stages (48 hpf).

Figure 6.

Cardiac cell number is not affected by the lack of functional Tbx5a or Tbx5b. Quantification of total cardiomyocyte numbers in wild-type (wt) and tbx5a or tbx5b-deficient embryos at 48 hours postfertilization (hpf). Data are expressed as means ± SE. There were no statistically significant differences among these treatments (*P > 0.05). For the columns listed left to right, n = 20, 11, 17, 11, 20, 14 embryos.

Comparative Analysis of Phenotypes Based on Gene Expression Using qPCR

The Tbx5 transcription factor has several well-characterized direct target genes in mammals, many of which are conserved in other species (Naiche et al., 2005; Stennard and Harvey, 2005; Greulich et al., 2011). If tbx5a and tbx5b carry out independent functions in the heart, as suggested above, then they would likely regulate independent subsets of target genes. To determine if tbx5b regulates some of the same genes targeted by tbx5a, we selected six well-characterized Tbx5 target genes and quantified their expression in zebrafish whole embryos using quantitative real-time PCR (qPCR). We compared mRNA expression levels for the six candidate genes in wild-type, tbx5a mutant (hst), and tbx5b morphant embryos. We tested two stages in development: 30 hpf (before overt cardiac phenotypes observed in tbx5a mutant and tbx5b morphant embryos) and 36 hpf (after the onset of the looping and contractile phenotypes but before the severe degeneration of the heart).

The six candidate target genes were: tbx5a, tbx5b, nppa, bmp4, hey2, and tbx2b. We evaluated expression levels of tbx5a and tbx5b to determine if depletion of one paralog spurred compensatory up-regulation of the alternate paralog. Additionally, we chose nppa because this gene is directly regulated by Tbx5 in mammals (Bruneau et al., 2001), and because nppa expression is reduced in tbx5a mutant embryos (Camarata et al., 2010) and Fig. 5B. We selected bmp4 because Tbx5 acts cooperatively with the Nkx2.5 transcription factor in mammals to directly regulate bmp4 expression (Puskaric et al., 2010). The fifth candidate, hey2, is a transcription factor involved in cardiac septation, proliferation inhibition, and cardiomyocyte differentiation (Plageman and Yutzey, 2005). The last candidate, tbx2b, is known to be directly regulated by Tbx5a together with Pdlim7 or Foxn4 (Chi et al., 2008; Camarata et al., 2010), and tbx5a mutant embryos showed a total absence of tbx2b expression at the AVJ (Camarata et al., 2010).

A known zebrafish housekeeping gene, ef1a, was used as a standard control for all qPCR assays. We found no difference in ef1a expression levels among the wild-type, tbx5a homozygous mutant, or tbx5b morphant embryos at 30 or 36 hpf by analysis of variance (ANOVA) analysis (P = 0.99), indicating consistent mRNA quantity and integrity among all reactions (Fig. 7G). In tbx5a homozygous mutant embryos, all of the candidate genes displayed decreases in log copy numbers of mRNA that were statistically significant at the level of P < 0.001 to P < 0.05 (Fig. 7A–F), reinforcing our expectation that tbx5a functions as a transcriptional activator, and that it would regulate these conserved vertebrate Tbx5 targets. In some cases, tbx5a mutant embryos showed a difference in target gene expression at both 30 hpf and 36 hpf (Fig. 7A,C,D), but for other genes the regulation was only apparent by 36 hpf (Fig. 7B,E,F), consistent with some functions of tbx5a arising later in embryogenesis. Previous ISH studies indicated tbx5a transcripts remained detectable by ISH in tbx5a mutants, even at 48 hpf (Garrity et al., 2002). Here, qPCR data indicated quantitatively that tbx5a transcripts were reduced in tbx5a mutants relative to wild-type levels (Fig. 7A). In tbx5a mutants, tbx5b expression at 36 hpf was down-regulated, suggesting that the tbx5b gene is possibly regulated by tbx5a (Fig. 7B). However, no T-box binding element (TBE) was detected using rVISTA analysis within 3 kb of sequence upstream of tbx5b. Embryos injected with tbx5b morpholino did not show a difference in endogenous levels of tbx5a mRNA, indicating that tbx5b does not appear to regulate expression of its paralog (Fig. 7B). Moreover, tbx5b morphants showed no differences from wild-type in transcript levels of any of the tested genes, either at 30 or 36 hpf, suggesting that tbx5b does not regulate any of these candidate genes at the time points tested. Note that morpholino efficacy studies support a high degree of tbx5b knockdown in morphant embryos even up to 42 hpf (Fig. 3E). Therefore, the tbx5a and tbx5b genes appear to demonstrate nonoverlapping profiles of gene regulation.

Figure 7.

Loss of Tbx5a or Tbx5b differentially affects the expression of conserved mammalian Tbx5 target genes. mRNA levels for candidate target genes were quantified by qPCR. Data are represented as the mean ± S.E. Differences in expression, measured in log-transformed values of copy number per ug template RNA, were analyzed by ANOVA. A Tukey-Kramer post hoc test was used to determine statistical differences between samples (*P < 0.05, **P < 0.001). A,B: The zebrafish Tbx5 paralogs. C–F: Candidate direct targets of Tbx5 based on conservation among vertebrates. G: A house-keeping gene used as a standard control.

tbx5b Depletion Results in Altered Pectoral Fin Morphology

Injection of tbx5bMO reduced pectoral fin size in 86% of embryos by 72 hpf (n = 282) (Fig. 8A,B). Injection of splice-blocker tbx5bSD-MO likewise reduced fin size in 36% of embryos (n = 106, data not shown), supporting the idea that small fin phenotypes are specific to tbx5b depletion. Injection of tbx5bMO occasionally resulted in upturned (6.2% of embryos) or unilateral (10.3%) pectoral fins (Fig. 8B). Upturned or unilateral fin phenotypes are rather unusual, but these phenotypes can be produced by modest reduction of tbx5a through low-dose morpholino injection; in contrast, homozyogous mutation of tbx5a completely blocked pectoral fin initiation (Garrity et al., 2002). To further determine how fin bud initiation or patterning was affected, we followed the timing of pectoral fin development from initiation through 72 hpf (Fig. 8C). Although fin bud swellings should be apparent by 28 hpf (Hoffman and Kaplan, 2002), 71% of tbx5bMO-injected embryos lacked any morphologically detectable fin buds at 30 hpf. However, fin bud formation was merely delayed, rather than absent in tbx5bMO-injected embryos, because scoring embryos at later time points indicated that smaller fin buds eventually formed in nearly all embryos (Fig. 8C). Fewer than 15% of tbx5bMO-injected embryos developed a normal pectoral fin by 72 hpf; instead, the majority of morphants developed stunted, up-turned, or otherwise abnormal pectoral fins (Fig. 8B). No defects were observed in the tail fins of morphants for either tbx5b morpholino. Therefore, embryos depleted for tbx5b delayed the initiation of pectoral fin buds, and later developed severe defects in fin bud outgrowth and patterning.

Figure 8.

Depletion of tbx5b delays pectoral fin initiation and reduces fin outgrowth. A: Lateral brightfield images of pectoral fin area taken of control or tbx5bMO-injected embryos; anterior is oriented to the left. Arrowheads denote the tip of the pectoral fin and are color-coded based on size (see B). B: Dorsal brightfield images of wild-type (wt) and tbx5bMO injected embryos at 72 hours postfertilization (hpf) illustrating the variability observed in pectoral fin morphology among morphant embryos. C: Quantification of pectoral fin size for tbx5bMO-injected embryos as development progresses. n = 21–46 embryos per time point. Dotted lines flank the fin bud tissue; bracket indicates area with missing fin.

tbx5b Affects Apical Fin Fold (AF) Specification

To investigate the patterning and differentiation of the fin bud on a molecular level, we performed ISH with a panel of markers indicative of differentiating ectoderm (apical fold) or mesenchyme. Zebrafish pectoral fin bud development begins with mesenchyme formation by 28 hpf, which is covered by an apical fold (AF, analogous to the apical ectodermal ridge of tetrapods) by 31 hpf (Grandel and Schulte-Merker, 1998; Neumann et al., 1999). Control embryos expressed hand2 in the pectoral fin bud mesenchyme at 36 hpf (Fig. 9A). In contrast, tbx5a mutant embryos do not express hand2 or other fin field markers consistent with the observation that pectoral fin buds never develop in these embryos (Fig. 9B,E,H) (Garrity et al., 2002). The normal expression of hand2 in the pharyngeal arches, or vcana and bmp4 in the otic placodes, provided internal controls to assure probe quality in these experiments. In contrast, tbx5b-depleted embryos did express hand2, but the region of expression was reduced, consistent with the smaller fin buds that later developed (Fig. 9C). Therefore, tbx5b morphant embryos were beginning to differentiate fin bud mesenchyme. To investigate AF differentiation, we assayed for bmp4 and vcana, which are normally expressed in the AF by 36–40 hpf (Rothschild et al., 2009). In control embryos, expression of vcana was present in the AF of the fin bud by 42 hpf (Fig. 9D). Although tbx5b morphants did initiate the outgrowth of pectoral fins, vcana expression remained absent or barely detectable even in embryos that had morphological fin bud swellings at 42 hpf (Fig. 9F). Assays for vcana in 36 hpf and 54 hpf embryos similarly failed to exhibit fin bud expression (data not shown). These data suggested that the AF was not properly specified in tbx5b morphants. In the wild-type fin, bmp4 expression occurred in the AF by 46 hpf and was faintly visible in the mesenchyme (Fig. 9G). In tbx5b morphants, bmp4 expression was detectable in the mesenchyme but usually absent from the AF, although in a small minority of cases it was observed in one developing pectoral fin but not the other (Fig. 9I). This mesenchymal expression pattern was similar in a 36 hpf pectoral fin (Fig. 9J), further suggesting that the pectoral fins of tbx5b morphants initiate but are developmentally delayed. Thus, tbx5a mutants and tbx5b morphants showed distinct differences in pectoral fin phenotypes on morphological and molecular levels.

Figure 9.

Differential phenotypes observed in fin differentiation for tbx5a mutants versus tbx5b morphants. A–J: Embryos at (A–C) 36 hpf; (D–F) 42 hpf; (G–I) 46 hpf; (J) 36 hpf. A: hand2 is expressed in mesenchyme of the wild-type fin bud. B: In tbx5a mutants, no pectoral fins develop; hand2 expression was absent in presumptive fin buds (brackets indicate region of expected expression) although present in pharyngeal arches. C: In tbx5b morphants, fin buds expressed hand2 in the mesenchyme in proportion to fin size. D: vcana expression in apical fold (AF) and otic placode (o). E: vcana expression missing in region of expected fin buds. F: vcana expression missing or barely detectable (*) in fin buds. G: bmp4 expression in eye, otic placode and AF of pectoral fin. H: No bmp4 expression in fin bud region. I: bmp4 expression can occur in AF (arrow); and was robustly detected in the mesenchyme of the fin bud. J: In younger embryos bmp4 was similarly detectable in mesenchyme and AF as it begins to form. AF, apical fold; e, eye; o, otic placode; p, pharyngeal arches; arrowhead, fin bud mesenchyme; arrow, AF; * barely detectable AF/mesenchyme expression; brackets, region of expected fin bud formation.

tbx5a and tbx5b mRNA Injections Rescue the Cognate but not the Paralogous Phenotypes

As a further test of the apparent divergence in tbx5a or tbx5b functions, we attempted to rescue the tbx5a-depletion phenotypes by overexpression of tbx5b mRNA and vice versa. For these “cross rescue” experiments, cardiac and pectoral fin phenotypes were scored separately. As a method to track the integrity of injected mRNAs, we created constructs encoding the tbx5a or tbx5b open reading frame fused with GFP. The presence of GFP in injected fish served to confirm the stability and translation of injected mRNAs. To avoid the possibility that fusion of GFP would interfere with Tbx5a or Tbx5b function, a viral 2A (V2A) peptide sequence was inserted between the open reading frames of tbx5 and GFP (Fig. 10A). V2A is a small, “self-cleaving” peptide, first described by (Ryan et al., 1991) as a domain of the foot-and-mouth disease (FMD) virus. Effectively, insertion of the V2A peptide allows bicistronic expression from a single transcript because the ribosome skips the synthesis of the glycyl-prolyl peptide bond at the C-terminus of the V2A peptide, but still continues its translation of downstream codons. This process ultimately results in the production of two separate proteins (Tbx5-V2A peptide and the GFP peptide) from sequences on the single transcript (Donnelly et al., 2001; de Felipe et al., 2010).

Figure 10.

mRNA injection rescue studies indicate that tbx5a and tbx5b are not functionally redundant in cardiac development. A: Schematic of mRNA structure for rescue constructs, in which the coding sequence for tbx5a or tbx5b was connected to enhanced green fluorescent protein (EGFP, with accompanying poly-adenylation signal) by means of a viral 2A peptide. Blue arrow indicates location of translation start site. Blue arrowhead indicates termination signal. B: Fluorescent images of 26 hours postfertilization (hpf) embryos previously injected with combinations of either buffer (in wild-type, wt), tbx5aMO or tbx5bMO, and tbx5a or tbx5b mRNA to evaluate mRNA integrity by means of production of translated GFP. C: Brightfield images of 72 hpf embryos showing representative cardiac phenotypes for each group of injected embryos. D: Quantification of morpholino injected embryos displaying a cardiac morphology phenotype at 54 hpf. (tbx5aMO 70.3% n = 91; tbx5a MO+ tbx5a mRNA 2.4% n = 84; tbx5a MO+ tbx5b mRNA 78.9% n = 133; tbx5b MO 69.1% n = 220; tbx5b MO+ tbx5a mRNA 84.6% n = 65, tbx5b MO+ tbx5b mRNA 40.6% n = 69).

As expected, embryos injected with rescue mRNAs consistently showed fluorescence associated with the GFP reporter protein, indicating the mRNAs were stable and appropriately translated (Fig. 10B). Injection of tbx5aMO generated embryos with the expected cardiac phenotypes in 70.3% of embryos (Fig. 10C,D). Co-injection of tbx5aMO along with 600 ng/μl tbx5a mRNA reduced the number of embryos with cardiac phenotypes to 2.4%, thereby demonstrating an efficient rescue of the cognate phenotype. Co-injection of tbx5b mRNA into tbx5a morphants resulted in 78.9% of embryos with cardiac defects, showing that tbx5b mRNA could not rescue tbx5a morphant cardiac defects. In the converse experiment, injection of tbx5bMO alone generated cardiac phenotypes in 69.1% of embryos. Co-injection of 775 ng/μl tbx5b mRNA into tbx5b morphants decreased the incidence of cardiac defects to 40.6%, demonstrating a partial rescue of cognate cardiac phenotypes (Fig. 10C,D). However, injection of tbx5a mRNA into tbx5b morphants resulted in 84.6% of embryos with cardiac phenotypes, showing that tbx5a mRNA could not rescue tbx5b morphant cardiac defects. These experiments demonstrate that, while zebrafish tbx5a and tbx5b mRNAs were capable of rescuing the cardiac phenotypes of their cognate morpholino, they were not sufficient to compensate for the cardiac function of the paralogous gene.

Because both Tbx5a and Tbx5b are required for pectoral fin outgrowth, we also used the rescue approach to investigate the degree of overlap in pectoral fin function. To evaluate pectoral fin rescue we quantified fin outgrowth at 72 hpf, by which time delayed fin buds should have initiated some outgrowth. As a control, embryos were injected first with 600 ng/μl tbx5a mRNA or 775 ng/μl tbx5b mRNA. Fewer than 5% of embryos showed any detectable phenotype (data not shown), suggesting that at these doses, rescue mRNAs injected singly rarely generated scorable phenotypes. When embryos were co-injected with tbx5aMO and 600 ng/μl tbx5a mRNA, we found that 65% of embryos still failed to develop any pectoral fins but 16% of embryos did develop a stunted (or “partial”) pectoral fin (Fig. 11). Thus, although tbx5a mRNA efficiently rescued heart development, it did not have a large ability to rescue fin initiation or fin outgrowth. Higher concentrations of tbx5a mRNA resulted in nonspecific body-axis defects or death at pregastrulation stages (data not shown). In the “cross-rescue” experiment, injection of 775 ng/μl tbx5b mRNA into tbx5a morphants showed 79% of total embryos failed to initiate pectoral fin initiation, showing that tbx5b mRNA likewise could not rescue pectoral fin initiation. As previously shown, injection of tbx5bMO alone resulted in a majority (86%) of embryos with stunted pectoral fins at 72 hpf. Injection of tbx5b mRNA along with tbx5bMO decreased the percentage of embryos with stunted pectoral fin phenotypes to 23%, with the remainder of embryos showing normal fins. Thus, tbx5b mRNA did substantially rescue fin phenotypes produced by tbx5bMO. In the “cross-rescue” experiment, injection of tbx5a mRNA into tbx5b morphants resulted in 76% of embryos that completely lacked pectoral fin bud initiation, indicating no ability of Tbx5a to function in the stead of Tbx5b in the fin. In fact, injection of tbx5a mRNA into tbx5b morphants made the fin phenotype more severe; most embryos failed to even initiate pectoral fins. In summary, we note that in all cases of morphants coinjected with the paralogous mRNA, no rescue occurred for either cardiac or fin phenotypes. Overall, these experiments are consistent with our other data indicating that, although both Tbx5a and Tbx5b are required in both the heart and the fin, the genes appear to carry out independent functions in the development of both structures.

Figure 11.

mRNA injection rescue studies indicate that tbx5a and tbx5b are not functionally redundant in pectoral fin bud outgrowth. A: Brightfield images of lateral pectoral fin region at 54 hours postfertilization (hpf). Dots outline the buds and are color-coded according to legend (see B). B: Percent of embryos exhibiting each phenotype (normal, partial or no pectoral fin outgrowth) at 54 hpf (tbx5a MO 82.5% n = 91; tbx5a MO+ tbx5a mRNA 78.6% n = 84; tbx5a MO+ tbx5b mRNA 89.5% n = 133; tbx5b MO 79.1% n = 220; tbx5b MO+ tbx5a mRNA n = 65, tbx5b MO+ tbx5b mRNA n = 69). Black or green dotted lines flank normal or truncated fin buds, respectively; bracket indicates area with missing fin.

DISCUSSION

The primary findings from this study are that (1) the zebrafish T-box transcription factor tbx5b is essential for cardiac development and pectoral fin morphogenesis; (2) that the zebrafish tbx5a and tbx5b paralogs have distinct, nonredundant functions for heart and pectoral fin development; and (3) that this functional divergence has evolved since these paralogs originated in the teleost-specific whole genome duplication ∼270 mya (Hurley et al., 2007).

The function of tbx5b was defined by means of morpholino knockdown and rescue studies. Using either of two morpholinos, tbx5b-depletion resulted in deficiencies in cardiac morphogenesis that affected cardiac looping, contractility, blood flow, and were ultimately lethal. In addition, reduction of tbx5b caused a delay in the onset of pectoral fin initiation, followed by irresolvable defects in fin patterning and outgrowth. These studies therefore define critical functions for tbx5b in cardiac and fin morphogenesis.

Several lines of evidence suggest that tbx5b functions are distinct from those of tbx5a, and that the two genes may regulate separable cohorts of targets. ISH studies evaluating several markers of differentiation indicated that tbx5b cardiac and fin phenotypes differ from those of tbx5a mutants, and differ from wild-type. qPCR studies supported this finding, indicating that depletion of tbx5b did not alter mRNA levels for any of six genes whose expression was decreased in response to tbx5a depletion. Because tbx5b phenotypes cannot be explained by alteration of tbx5a target genes, it is probable that the tbx5b transcription factor regulates separate targets. Dual depletion of both tbx5a and tbx5b by morpholinos did not result in more severe or additive cardiac phenotypes, such as severe depletion or lack of cardiac tissue, nor did it affect cardiomyocyte cell numbers more adversely than does single gene knockdown. Finally, the inability of each tbx5 paralog to “cross-rescue” the knockdown phenotype of the other paralog demonstrates minimal redundancy in function exists between the two paralogs.

Our data suggest two mechanisms that may contribute to the independent functions proposed for the tbx5 paralogs. First, sequence differences in the tbx5b T-box region might explain why tbx5b appears to regulate a different cohort of target genes than tbx5a. Indeed, D. rerio Tbx5b showed only 83% sequence identity with D. rerio Tbx5a, a value notably less than most T-box family members, which typically display 95–99% amino acid identity within this region (Holland et al., 1995; Bamshad et al., 1997; Horb and Thomsen, 1999). In addition, the D. rerio tbx5b gene showed substantial differences from tbx5a throughout its sequence, as well as in gene size and genomic organization. These differences may substantially affect the structure of the C terminal domain thereby leading to different protein:protein interactions.

Phylogenetic analysis of Tbx5 amino acid sequences shows that the teleost tbx5a and tbx5b clades are each other's closest relatives, and that the zebrafish tbx5b sequence has accumulated extensive amino acid substitutions since its origin during the teleost-specific whole genome duplication (Hoegg et al., 2004). Approximately 270 million years ago, teleost genomes underwent a whole-genome duplication event (Hoegg et al., 2004; Hurley et al., 2007). The presence of tbx5 paralogs in several teleost genomes, and the results of our phylogenetic analyses, suggest that tbx5a and tbx5b arose at this time. Following a duplication event, there are numerous possible outcomes for the newly created pair of paralogous genes (Innan and Kondrashov, 2010). One gene copy may accumulate degenerative mutations over time, ultimately becoming a nonfunctional pseudogene (van Noort et al., 2003). In other cases, the two paralogous genes can undergo subfunctionalization; both paralogs are retained in the genome because each retains a portion of the ancestral function of the single, preduplication gene (Force et al., 1999). Finally, paralogous gene pairs can undergo neofunctionalization, in which one duplicate evolves a new function (Walsh, 1995). These alternate fates are determined by complex interactions among mutation, natural selection, and genetic drift (Proulx, 2012). In the case of the zebrafish tbx5 paralogs, the regions of embryonic expression appear highly similar, and the organs of embryonic function are the same; thus, very little divergence has occurred between the paralogs with respect to spatio-temporal expression patterns. Yet, a more detailed analysis examining the expression of putative targets for these transcription factors suggests that, mechanistically, the functions of these two genes have diverged substantially. More detailed analyses of the preduplication function of tbx5 in basal ray-finned fishes are required to test whether this functional divergence reflects subfunctionalization or neofunctionalization.

An important cellular function for Tbx5 in the heart concerns its effect on cardiomyocyte cell number. Evidence exists supporting either growth promoting activity or growth inhibiting activity of Tbx5 upon cardiomyocytes. For example, depletion of TBX5 by morpholino in Xenopus resulted in a ∼30–40% decrease of total cardiac cells in hearts undergoing early chamber formation (stage 37), indicating that TBX5 in frogs positively regulates cardiomyocyte proliferation in this system (Brown et al., 2005; Goetz et al., 2006). On the other hand, overexpression of wild-type TBX5 in chick (beginning at HH17–HH18, early cardiac looping stages) also reduced cardiomyocyte proliferation by 40% (Hatcher et al., 2001). These data support a model in which TBX5 functions in chick as a growth arrest signal that negatively regulates cardiomyocyte proliferation. In zebrafish, the single depletion of either Tbx5a or Tbx5b showed no differences from wild-type cardiomyocyte cell numbers, leading to the initial hypothesis that genetic redundancy between the paralogs might mask the function of Tbx5 in this regard. However, this hypothesis is not supported by the dual depletion experiments, which in every case indicate no difference from wild-type cardiomyocyte cell numbers. Thus in contrast to other organisms, it is possible that in zebrafish Tbx5 simply does not have a major impact on cardiomyocyte cell number. Alternatively, it has been proposed in frog that Tbx5 may have a dual role as an activator in early stages and a growth arrest signal in later stages (Goetz et al., 2006). If true, then the timing of when cell number is assessed in depletion or overexpression experiments could be an important factor in interpreting and comparing the phenotypes of different systems. In addition, data from mice indicate the presence of distinct Tbx5 alternatively spliced isoforms that show unique expression patterns and discrete biochemical properties suggestive of differential roles in cell proliferation (Georges et al., 2008).

Despite the probable low levels of tbx5b expression in the fin fields, we find that tbx5b function is indeed required for the normal outgrowth and patterning of fins. Importantly, the tbx5b morphant fin phenotypes are a specific outcome of tbx5b gene knockdown, because they can be rescued by co-injection of tbx5b mRNA. Although fin bud initiation was delayed in tbx5b morphants, the initiation defect was resolvable by 72 hpf when 85% of morphant embryos had developed at least a small fin bud. Outgrowth defects were not resolvable because the majority of pectoral fins remained permanently small and mispatterned. In contrast, the complete loss of tbx5a function definitively eliminates (rather than delays) pectoral fin initiation (Ahn et al., 2002; Garrity et al., 2002). Conceivably, if the knockdown of tbx5b by means of morpholino was incomplete during fin morphogenesis, we might see a hypomorphic effect on fin growth and patterning whereas the true null phenotype might have blocked fin initiation. However, we do not favor this explanation because 100% of embryos injected with tbx5bSD-MO (which can be verified by RT-PCR to provide a very strong knockdown) all developed scorable fins by the second day of development. In addition, differences in the expression of pectoral fin molecular markers suggest that different mechanisms may be impaired when Tbx5b versus Tbx5a is depleted. In particular, our results pinpoint a requirement for tbx5b function in differentiation of the AF, because embryos lacking tbx5b fail to express two markers for the AF. In contrast, tbx5b morphants expressed at least one marker of fin mesenchyme normally in their smaller pectoral fin buds. Of interest, bmp4 expression was missing or highly down-regulated in the AF but was potentially expanded within the fin bud mesenchyme, relative to controls (Fig. 9I). A similarly altered bmp4 expression pattern, combined with stunted and/or asymmetrical pectoral fin morphology, was reported for embryos depleted of CaMK-II by morpholino injection (Rothschild et al., 2009). CaMK-II is a type II multifunctional Ca2+/calmodulin-dependent protein kinase. Rothschild et al. determined that CaMK-II acts downstream of tbx5a in zebrafish. Whether tbx5b similarly regulates CaMK-II is currently unknown. In summary, Tbx5b now joins the cohort of several Tbx-family transcription factors with roles in limb patterning.

In conclusion, the tbx5b gene differs from its paralog tbx5a on several levels, including genomic organization, sequence, and function. Like tbx5a, we find that tbx5b has critical functions in both the embryonic heart and the pectoral fins. However, we find little evidence to postulate a mechanistic overlap or redundancy in function between these two genes. The paralogs appear to fit an evolutionary model of duplication followed by either neo- or subfunctionalization, although current data do not allow us to discriminate between these two models. Although both genes function in the embryonic heart and forelimb, as does human TBX5, the zebrafish Tbx5a protein appears to demonstrate a closer parallel to humans in its conserved regulation of mammalian Tbx5 target genes than does Tbx5b. However, transcriptional network studies for both genes may be required to fully comprehend the range of functions carried out by T-box genes during organogenesis.

EXPERIMENTAL PROCEDURES

Zebrafish Husbandry

Zebrafish were raised and staged as described previously (Westerfield, 1995; Nusslein-Volhard and Dahm, 2002) and as per Colorado State University Animal Care and Use Protocols. Developmental time at 28.5°C was determined from the morphological features of the embryo as described by (Kimmel et al., 1995).

Zebrafish Imaging

Live larvae were immobilized using MESAB and placed on agarose injection plates for imaging. All samples were visualized using an Olympus SZX12 Fluorescent stereo-dissecting microscope. Photographs were taken using a microscope-mounted Olympus U-CMAD3 digital camera and captured using SPOT software imaging (Diagnostic Instruments, Inc.). Postprocessing was done using Adobe Photoshop software to correct white balance and image sizes.

tbx5b WIK Allele Cloning and Sequencing

Wild-type tbx5b mRNA was extracted from WIK embryos using the Trizol method, reverse transcription and polymerase chain reaction were performed as described (Garrity et al., 2002; Ebert et al., 2005) using primers described by Albalat et al. (2010). Resulting cDNA was TA-cloned into the pCRII vector (Invitrogen) and sequenced using m13 universal primers.

tbx4 and tbx5 Sequences for Comparative Analyses

Sequences of tbx4 and tbx5 were obtained from diverse taxa (amphioxus, sharks, teleost fishes, amphibians, and several amniotes) from a variety of public databases. Genomic DNA and expressed sequence tag (EST) data were translated into amino acids.

Amino Acid Alignment

Amino acid alignments were performed using MUSCLE (Edgar, 2004). Amino acid sequences derived from EST data were examined for potential sequencing artifacts leading to frame-shifts in highly conserved domains; in two cases (G. aculeatus tbx4 and A. mexicanum tbx4), we conservatively restored the reading frame by adding an N or deleting a nucleotide in low-quality sequence. Identical sequences (i.e., taxa for which we only obtained portions of the highly conserved T-box domain) were excluded from further analysis, as were protein regions that could not be unambiguously aligned. We performed two alignments: one included both Tbx4 and Tbx5 sequences, and one included just Tbx5 sequences.

Phylogenetic Analysis

Phylogenetic trees were estimated using Maximum Likelihood, implemented in RAxML (Stamatakis, 2006; Stamatakis et al., 2008), using the GAMMA model of amino acid substitution and WAG protein substitution matrix with amino acid frequencies estimated from the data (Whelan and Goldman, 2001). Bootstrap support for nodes was calculated using rapid bootstrapping, with the number of bootstrap iterations defined by the Majority Rule Criterion (Stamatakis et al., 2008). Phylogenetic trees were estimated from (1) the alignment including all nonidentical Tbx4 and Tbx5 sequences (35 sequences, 251 amino acid positions), as well as (2) the alignment including only Tbx5 sequences (19 sequences, 311 amino acids). The Tbx5 alignment and resulting tree are deposited in TreeBASE (www.treebase.org) (TB2 ID number: S12622).

Morpholino Injections and RT-PCR Analysis

Morpholinos against tbx5b were designed by Genetools (California) to block translation (tbx5bMO: GGATTCGCCATATTCCCGTCTGAGT) or to target the splice donor site of exon 2 (tbx5bSD-MO: TTAAAAAACTAGGCACTCACCGGCC). The translation blocking tbx5a morpholino (tbx5aMO: GCCTGTACGATGTCTACCGTGAGGC) used here targets sequences slightly upstream of the initiation codon, thus making it possible to use wild-type mRNA (which does not contain the morpholino target site) for rescue experiments. This tbx5aMO morpholino was designed by (Lu et al., 2008) and the efficacy and specificity of the MO is demonstrated in that report. One- to four-cell embryos were injected with 100 μM tbx5bMO, 100 μM tbx5bMO, 150 μM tbx5bSD-MO, or coinjected with 75 μM tbx5bMO and 100 μM tbx5aMO together. For RT-PCR analysis of tbx5bSD-MO knockdown, RNA was extracted as described below and PCR performed using GoTaq (Promega) and primers #F363 5′-CGCGATGGATTTCGAGAGGGAT-3′ 54 and #R365 5′-TCGTCTGCCTTCACGATGTGCA-3′. For all morpholino experiments, embryos were co-injected with 0.1% tetramethyl–rhodamine dextran, used as an injection tracer, and only embryos showing robust levels of rhodamine fluorescence were scored or extracted.

Cardiac Morphology

Forty-eight hpf larvae were immobilized using MESAB and placed in agarose-containing petri dishes for imaging. To quantify cardiac looping, wild-type and mutant lines were crossed with Tg(myl7:EGFP-HsHRAS) to image cardiac-specific fluorescence. Looping angles were determined by measuring the angle of the AV junction relative to the anterior/posterior axis of the embryo (Chernyavskaya et al., 2012). All samples were visualized using a Spot Insight IN1120 digital camera on an Olympus SZX12 fluorescent stereo-microscope. Postprocessing was completed with Adobe Photoshop software.

In Situ Hybridization

Embryos were fixed in 4% paraformaldehyde solution in PBS (PFA) and prepared for whole mount in situ hybridization as described (Thisse et al., 1993). Digoxigenin-labled riboprobes were synthesized as described previously (Ebert et al., 2008).

Cardiomyocyte Quantification

The number of cardiomyocytes in the heart was quantified in wild-type and mutant embryos crossed into a Tg(myl7:nDsRed2/myl7:EGFP) homozygous background (Mably et al., 2003). At 48 hpf, embryos were pressed between a glass coverslip and slide to gently flatten the heart. Intact flattened hearts were immediately imaged using a Leica 5500 microscope.

Capped mRNA Synthesis and Injection

For creation of the Tester mRNA, the tbx5bMO recognition site was fused to EGFP by means of PCR amplification with the following primers: #331TbMOGFPF 5′-ACTCAGACGGGAATATGGCGAATCAGGTGAGCAAGG GCGAGGAGCTGTTC-3′ and #304GFPR 5′-AAAAAAGGGCCCCCCGGGAAAAAACCTCCCACAC-3′. The resulting amplicon was TA-cloned into pCRII and confirmed by DNA sequencing. For the rescue mRNAs, the V2A peptide sequence was inserted 5′ of EGFP using a two-step PCR reaction and the following primers: #316F 5′-AAGTCTGCTAACATGCGGTGACGTCGAGGAGAATCCTGGCCCAACCATG GTGAGCAAG-3′, #317F2 5′-GGCGGCCGCTTGGCAGTGGAGAGGGCAGAGGAAGTCTGCTAACATG-3′. The #366 3′EV plasmid from the Chien Lab Gateway kit, which contains EGFP, was used as template for amplification using the EGFP-specific reverse primer and a forward primer containing a portion of the V2A and a portion of the 5′ end of GFP (primer #316F). The resulting band was gel-extracted and used as template for the second PCR using the same GFP-specific reverse primer and a second forward primer containing the remainder of the V2A sequence (primer #317F2). This primer set contained SmaI, ApaI, and NotI restriction sites to facilitate cloning in frame with tbx5a and tbx5b. The resulting amplicons were TA-cloned into pCRII (Invitrogen). The full-length constructs containing the tbx5 paralog, V2A peptide sequence, and EGFP sequence were confirmed by DNA sequencing and were termed pCRII-tbx5a-V2A-EGFP or pCRII-tbx5b-V2A-EGFP. Full-length mRNAs were prepared using mMessage Machine kit (Ambion) according to the manufacturer's instructions. mRNA was injected into the one-celled embryo at the following concentrations: tbx5bMOsite-GFP tester mNRA (1,000 ng/μl), tbx5aV2A-GFP (800 ng/μl), and tbx5bV2A-GFP (1,000 ng/μl).

Quantitative Real-Time PCR

Embryos were selected for RNA extraction based on stage (30 or 36 hpf) and genotype. Total RNA was isolated from 60 pooled embryos per treatment using TRIzol reagent (Invitrogen, Carlsbad, CA) according to the manufacturer's protocol. Briefly, total RNA was treated with DNase I (Invitrogen) to remove genomic DNA. After DNase treatment, total RNA was subjected to phenol:chloroform extraction, followed by precipitation of RNA to remove DNase, salts, and digested DNA. Total RNA was precipitated with 1 volume of isopropanol and resuspended in 30 μl of nuclease-free water. Nucleotide concentration was determined by absorbance at 260 nm and 3 μg was used for 20-μl cDNA synthesis reactions using AMV reverse transcriptase (Fisher Scientific) and Oligo(dT)12–18 primer (Invitrogen). Resulting cDNA was treated with RNase H (New England Biolabs, Ipswich, MA) to remove complementary RNA.

Quantitative analysis of tbx5b [GenBank Accession HQ822122; tbx5bF 5′-ATTCAGGACATCACGGAC GGAACA-3′, tbx5bR 5′-TTTGTAGCTGGGAAACATGCGTCG-3′], tbx5a [GenBank NM_130915 (Tamura et al., 1999); tbx5aF: 5′-GGAATT TAAGGCCTCACGGTA-3′, tbx5aR: 5′-GATTTGCTGACGGCTGCATTC TGT-3′], tbx2b [GenBank NM_131051 (Ruvinsky and Gibson-Brown, 2000); tbx2bF: 5′-GTCCCTTTCCCT TTCATCTGTCTC-3′, tbx2bR: 5′-CTGGGAGCTGATAAGGGTTGAATC-3′], hey2 [GenBank NM_131622 (Zhong et al., 2000); hey2F: 5′-GAAAGAAGCGGAGAGGGATCATTG-3′, hey2R: 5′-AGAAGTCCATGG CCAGAGAATGAG-3′], bmp4 [GenBank NM_131342 (Hammerschmidt et al., 1996); bmp4F: 5′-CACAGT ATCTGCTCGACCTCTA-3′, bmp4R: 5′-GATATGAGTTCGTCCTCTGGGATG-3′], nppa [GenBank NM_198800 (Berdougo et al., 2003); nppaF: 5′-CAGACACAGCTCTGACAGCAACAT-3′, nppaR: 5′-CTC TGTGTGTCAAATCCATCCGAG-3′], and eF1-alpha [GenBank NM_131263 (Gao et al., 1996); ef1aF: 5′-CGGTGACAACATGCTGGAGG-3′, ef1aR: 5′-ACCAGTCTCCACACGACCA-3′] was conducted using LightCycler FastStart DNA Master Plus SYBR Green I reaction mix and a LightCycler 480 thermal cycler (Roche Applied Science, Indianapolis, IN). Triplicate reactions consisted of 1 μl first-strand cDNA, 5 μl 2× SYBR Green Master Mix, 0.5 μl each of forward primer and reverse primer (10 μM each), and 3 μl PCR-grade water. PCR conditions included an initial denaturation at 95°C for 5 min, followed by 45 cycles of denaturation at 95°C for 10 s, annealing at 55°C for 15 s, and extension at 72°C for 15 s. At the end of each run, an analysis of the PCR product melting temperature was conducted. Amplicon specificity was also verified by sequencing. Transcript concentrations were calculated with LightCycler 480 software (Roche; version 1.2) using standard curves produced by serial dilution of purified PCR product (10 zg/μl to 10 ng/μl). Use efficiencies for all reactions were confirmed to be between 1.86 and 1.96 and transcripts were confirmed by sequencing. Statistical analysis was performed using JMP software v9.0.2 (SAS Institute, Inc., Cary, NC). qRT-PCR data were log transformed before analysis to reduce differences in variance from the mean. Means for transcript abundance were compared between wild-type, tbx5a, and tbx5b MO embryos using ANOVA. Post hoc multiple comparisons were made using Tukey-Kramer HSD tests. All data not plotted as individual points are presented as mean ± S.E. and the level of significance was set at α = 0.05 for all statistical analyses.

ACKNOWLEDGMENTS

The authors thank Tomoko Obaro, J. Mably, X. Xu, G. Begemann, P. Ingham, D. Yelon, C. Minguillon and D. Stainier for probes, plasmids, and transgenic lines, and Hazel Sive for her suggestion of the V2a sequence. We are also very appreciative of methodological support provided by Kathy Cozensa and Yelena Chernyavskaya.

Ancillary