Contraction of individual cardiomyocytes in zebrafish is coincident with the formation of the heart tube (Stainier et al., 1993). The heartbeat is created by the coordinated contraction of multiple cardiomyocytes that iteratively take-up and release calcium (Berridge et al., 2003; Ebert et al., 2005). As development proceeds, changes in chamber shape, cardiac looping, and the development of atrioventricular valve leaflets, as well as changes in electrical excitability, all contribute to the development of a synchronized heartbeat. Voltage-gated L-type calcium channels (LTCC), located on the cardiomyocyte cell surface, contribute to the cardiac action potential by mediating calcium entry into the cell, and triggering calcium-induced calcium release (Fabiato, 1983). The resulting high concentrations of intracellular calcium facilitate the next systolic contraction (Bers, 2002). The cardiac LTCC is an oligomeric channel composed of a large pore-forming α subunit and two auxiliary proteins (the α2δ and β subunits). The transmembrane α subunit houses the voltage sensor which responds to membrane depolarization by opening the channel, allowing Ca2+ entry into the cell. The membrane-associated α2δ regulatory subunit increases the calcium current and the propensity of the α subunit to associate with the plasma membrane (Davies et al., 2007). The β subunits, encoded by the CACNB genes, are responsible for most regulation of channel expression and gating (Catterall, 2000). Ca2+ conduction, channel expression and recovery from inactivation are increased in the presence of the β subunit (Singer et al., 1991; Jeziorski et al., 2000).
Mammals express four β subunits (β1–β4) that interact nonexclusively with the α subunit(s). Embryonic expression of β subunits is temporally and spatially segregated (with some overlap; Schjott et al., 2003; Acosta et al., 2004). β subunit mRNA and protein are found in all excitable tissues as well as some nonexcitable tissues including kidney and liver (Ebert et al., 2008a, b; Link et al., 2009). The wide distribution of β subunits throughout the body is consistent with their important roles in maintaining physiological functions. Moreover, genetic variation in CACNB genes has been linked with several human disorders affecting heart and brain, including cardiac arrhythmia (early repolarization syndrome, Brugada syndrome, and familial sudden cardiac death syndrome; Antzelevitch et al., 2007; Hedley et al., 2009; Burashnikov et al., 2010), cardiovascular disease (Levy et al., 2009), congenital cardiac morphology (Waleh et al., 2010), coronary artery disease (Davis et al., 2010a, b), bipolar disorder (Lee et al., 2011), and common migraine (Nyholt et al., 2008).
In mammals, each of the four β genes is associated with unique patterns of embryonic and adult channel expression and electrophysiological properties, yet their overall similar protein structure classifies them into the membrane associated guanylate kinase (MAGUK) family of proteins (Buraei and Yang, 2010). The β proteins, like other MAGUKs, contain an SH3 domain as well as a catalytically inactive guanylate kinase (GK) domain, but lack the PDZ domain (Funke et al., 2005). By means of their SH3 and GK protein:protein interaction domains, MAGUK proteins can perform scaffolding functions by establishing a link between transmembrane proteins and cytoskeletal elements at cellular junctions, or serve as intermediates in signal transduction pathways (Jelen et al., 2003; Velthuis et al., 2007; Hara and Saito, 2009; Marasco et al., 2009). As one example relevant to this report, the multifunctional ZO proteins of the MAGUK family are essential for correct organization of tight junctions, but also associate with G-proteins and factors that regulate gene expression (Bauer et al., 2010).
Recent reports have identified additional roles and novel subcellular areas of expression that support the hypothesis that β proteins function as multifunctional MAGUKS rather than just auxiliary channel subunits (Colecraft et al., 2002; Alvarez et al., 2010; Zhang et al., 2010). First, the β subunits can indirectly modulate the overall stability of the calcium channel complex by regulating α subunit availability within the cell. In the absence of β subunits, α subunits become ubiquitinated by RFP2 and are targeted for degradation by proteasomes (Altier et al., 2011). Second, a possible role for β2 subunits in regulation of transcription is suggested by localization of a particular splice variant of β4 to the nucleus (Colecraft et al., 2002; Subramanyam et al., 2009), and its interaction with a chromatin binding protein (Hibino et al., 2003; Xu et al., 2011). High levels of β4 associated with the yolk syncytial nuclei in zebrafish may be related to the failure of β4-depleted embryos to complete epiboly (Ebert et al., 2008c). Finally, at the cell periphery, the role of β proteins may extend beyond direct modulation of the calcium channel. The actin binding protein AHNAK and β2 share multiple binding sites, and the sequestration of β2 by AHNAK may indirectly repress calcium channel current (Haase et al., 2005; Haase, 2007).
The β2 subunit is the predominant β gene expressed in the adult murine heart (Zhou et al., 2008; Link et al., 2009). Several β2 transcript variants exist in the zebrafish as well as mammals, adding to the potential complexity of this protein's function (Hullin et al., 1992; Ebert et al., 2008a; Zhou et al., 2008). Its role as regulator of the L-type calcium channel in the adult heart is well documented in mammals. β2 is also expressed in the embryonic hearts of both mice and fish, although not necessarily at high levels (Serikov et al., 2002; Ebert et al., 2008a; Zhou et al., 2008).
Loss of β2 expression led to vascular and cardiac deformities in mice (Weissgerber et al., 2006). These phenotypes were associated with embryonic lethality by E10.5. Hearts of gene targeted β null mutant mice exhibited compromised cardiac performance including decreased cardiac output and slower heartbeats. The complexity of the yolk sac vasculature was reduced, most likely as a secondary consequence of altered hemodynamic forces.
In this report, we use the zebrafish model system to study β2 function in cardiac development. Because zebrafish embryos aged 6 days or less receive oxygen by diffusion rather than cardiac output, development of the cardiovascular system can be tracked independently of the effects of hypoxia. The zebrafish genome encodes two β2 genes, β2.1 and β2.2, consistent with the duplication of the zebrafish genome 450 million years ago (Jaillon et al., 2004; Ebert et al., 2008a). Of these, β2.1 is the most similar to the human β2 (87% amino acid homology compared with 62% for β2.2). We report that depletion of β2.1 in zebrafish embryos results in reduced cardiac looping, decreased cardiac output, and weak contractility. The cardiac chambers are dysmorphic with dilated atria and collapsed ventricles. To explore the mechanistic basis for this phenotype, we investigated the specification of cardiac precursors within the bilateral heart fields, the rate of cell proliferation in the early heart tube, and the functional properties of the maturing cardiomyocytes. All of these properties were altered in β2.1-depleted embryos, identifying the β2.1 protein as an essential component of embryonic heart development.
Depletion of β2.1 Results in Lethality Due to Heart Failure
To determine the function of β2.1 in vivo we depleted its expression in zebrafish embryos using either of two morpholino oligonucleotides. Because β2.1 is alternatively spliced at the 5′ end, we designed one morpholino (MO1) to target the splice donor site of exon 5, an internal exon that is common to all splice variants (Fig. 1A). Systematic studies using splice-blocking morpholinos indicate that targeting an internal exon usually results in an exon-deletion event (Morcos, 2007), suggesting exon 5 should be deleted in β2.1-targeted transcripts. Mis-splicing of the β2.1 transcripts is predicted to encode a frame-shift leading to a premature termination codon which would truncate the protein within exon 6, resulting in a protein of 149 amino acids compared with the 579 residues of wild-type β2.1. If translated, the truncated protein would encode the SH3 domain; however, the conserved residues of the GK domain that are critically required for β subunit interaction with the α subunit would be missing. Hence, the truncated β2.1 subunit (if made) is not anticipated to interact with the calcium channel.
Reverse transcriptase-polymerase chain reaction (RT-PCR) was used to investigate the efficacy and specificity of MO1. Embryos injected with MO1 showed a strong reduction in the stability of full-length β2.1 mRNA (Fig. 1B). Gene-specific primers located in exons 3 and 10, which flank the MO1 target site, produced a 428-bp amplicon in controls, consistent with the wild-type β2.1 sequence. In morphant embryos, flanking primers were predicted to amplify a smaller 291-bp product lacking exon 5. However, no stable transcripts could be recovered from these reactions, suggesting that mis-spliced transcripts were subject to nonsense-mediated decay (NMD). Consistent with this hypothesis, the introduction of a premature translation codon is a common trigger for NMD (Muhlemann, 2008; Moraes, 2010). In addition, RT-PCR indicated that MO1 was specific to β2.1 because mRNA transcripts for the closely related β2.2 gene were not detectably reduced in size or stability (Fig. 1B).
A second morpholino (MO2) was designed to block the translation initiation codon for β2.1 transcript variant 6 (β2.1_tv6). Previous RT-PCR studies had indicated that β2.1_tv6 was the predominant form of β2.1 expressed in the embryonic heart at 72 hours postfertilization (hpf; Ebert et al., 2008a). The embryonic heart also expresses two additional β2.1 transcripts derived from an alternatively spliced first exon that would encode β2.1 proteins using an alternate translation initiation codon; however, these transcripts were present in very low abundance. Therefore, we hypothesized that blocking translation of the β2.1_tv6 transcript would create phenotypes similar to MO1.
Because embryonic β2.1 expression occurs in the developing brain and heart (Ludwig et al., 1997; Zhou et al., 2008; Link et al., 2009), we examined these tissues for phenotypes. Injection of 750 μM of β2.1 (32 ng total) of MO1 resulted in cardiac defects in 70% of injected embryos (Fig. 2A,B). With this dose, off-target effects such as cell death throughout the CNS or altered body axes were observed in fewer than 5% of the embryos. By 48 hpf, defects in chamber shape, cardiac looping, and contractility were apparent in the heart. At later stages (72 hpf), morphant hearts remained linear, and were surrounded by cardiac edema. Hydrocephaly in the hindbrain was observed at 48 hpf, but became less severe with time. Injection of MO2 produced similar lethal cardiac defects in 56% of embryos (Supp. Fig. S1, which is available online).
In preparation for injection rescue experiments, we quantified the extent of cardiac looping by measuring the “looping angle”; that is, the angle formed by the plane of the atrioventricular junction (AVJ) relative to the anterior/posterior (A/P) axis of the embryo (Fig. 3). In normal embryos, cardiac looping is progressive: at 34 hpf (near the onset of looping), wild-type heart tubes displayed an average looping angle of 47°. By 48 hpf, this angle had decreased, and by 52 hpf, wild-type hearts showed an average looping angle of only 13 degrees, reflecting the substantial re-shaping of the heart tube along the A/P axis (Fig. 3A). In MO1 morphants, hearts demonstrated a greater average looping angle (50°) at 48 hpf, compared with 20° in control embryos, indicating morphants were significantly less looped (P = 0.009; Fig. 3B). MO2 morphant hearts likewise showed a significantly decreased average looping angle (52°) at 48 hpf (P < 0.001). MO1 and MO2 morphant embryos showed similar development of fin and otic placode structures when compared with their wild-type siblings, indicating that the cardiac looping phenotypes were not caused by developmental delay. As a rigorous test for morpholino specificity, we co-injected wild-type β2.1 tv_6 cRNA along with MO1 to determine whether cardiac phenotypes could be rescued. Looping phenotypes were rescued to wild-type ranges in 72% of co-injected embryos. Based on these data we concluded that β2.1 is essential for cardiac morphogenesis and survival. In addition, these studies pinpoint β2.1 tv_6 as the specific splice variant primarily responsible for cardiac functions of β2.1.
Cardiac Morphology and Function Progressively Worsens in β2.1 Morphant Embryos
In zebrafish, cardiac looping occurs concomitantly with chamber ballooning (Auman et al., 2007). In the course of these morphological reshaping events the atrial and ventricular chambers acquire their distinct “kidney” shapes by differential growth of the cells in the inner (IC) and outer curvature (OC) of each chamber (Auman et al., 2007). We delineated regions of OC and IC cardiomyocytes within morphant hearts according to Auman et al. 2007. Individual ventricular myocytes in the OC of β2.1 morphant hearts did not exhibit the usual increase in surface area that accompanies chamber ballooning and looping. Morphant OC ventricular myocytes were significantly smaller in size than those of wild-type at 48 hpf (P = 0.003; Fig. 4A). In the wild-type embryo, OC cardiomyocytes transition from a smaller, rounder shape in the linear heart tube (27 hpf) to a larger, elongated shape in the ballooning heart (Fig. 4B,C). β2.1 morphant OC ventricular cardiomyocytes failed to undergo this transition in cell shape, and instead resembled the linear heart tube cardiomyocytes with their smaller, more rounded cell shape. These data indicate the stalling of chamber ballooning and cell growth in hearts of β2.1 morphant embryos by approximately 48 hpf.
To study how heart function was compromised in β2.1-depleted embryos, we assayed heart rate, stroke volume, and cardiac output. We recorded videos of live wild-type and β2.1 morphant ventricles at 30 and 48 hpf, chosen as representative points before or after the onset of chamber morphogenesis. Heart rates were calculated based on the number of ventricular contractions over time (beats per min [bpm]). At 30 hpf, no significant difference was noted in the average heart rate of β2.1 morphants relative to wild-type embryos (Fig. 5A). By 48 hpf, however, the average heart rate was approximately 30% slower in β2.1 morphant hearts (97 bpm) compared with control hearts (140 bpm; Fig. 5A; P < 0.001). Co-injection of wild-type β2.1 mRNA along with the morpholino (MO1) restored the heart rate of morphant embryos to levels similar to wild-type (138 bpm; P = 0.64).
To evaluate the heart's capacity for cardiac contraction, the dimensions of the ventricle (width and height) were calculated by outlining the ventricle in selected video frames chosen at end-diastole. Based on the formula of a prolate spheroid, ventricular volume was estimated (see the Experimental Procedures section). At 30 hpf, the β2.1-depleted ventricle did not differ in size from the wild-type ventricle (P = 0.48), but by 48 hpf the ventricle was substantially smaller than wild-type (Fig. 5B; P < 0.05). Indeed, the diastolic β2.1 morphant ventricle at 48 hpf remained quite similar to its size at 30 hpf (P = 0.40), suggesting little growth had occurred in this chamber over the 18-hr period. “Stroke volume” represents the volume of blood pumped from the ventricular chamber with each heartbeat. At 30 hpf, β2.1 morphants did not differ from wild-type in stroke volume (P = 0.81), but by 48 hpf they showed significant reductions relative to wild-type (Fig. 5C; P < 0.05). Within the β2.1 morphant ventricle, stroke volume failed to increase significantly from 30 to 48 hpf (P = 0.70). Another rubric of cardiac contractility is “cardiac output”, the product of stroke volume and heart rate, which represents the volume of blood pumped by the heart per min. Relative to wild-type, at 30 hpf, cardiac output in the β2.1 morphant ventricle was not changed, but by 48 hpf it was significantly decreased (Fig. 5D; P < 0.01). Within the β2.1 morphant ventricle, cardiac output failed to increase significantly from 30 to 48 hpf (P = 0.44). Taken together, these physiological measurements suggest that cardiac function was relatively normal in heart tubes of β2.1 morphants up to 30 hpf before chamber ballooning. However, by the onset of chamber ballooning around 48 hpf, heart function was significantly impaired, ventricular cell growth was lagging, and expected transitions in ventricular cell shape had failed to occur.
Pharmacological Assessments of LTCC Function in Cardiac Development
In considering the mechanisms of β2.1 function, we questioned whether its contribution to heart development was based on its role as a cytoplasmic subunit of the oligomeric L-type calcium channel (LTCC). Alternatively, as noted in the Introduction, several recent reports suggest that β subunits can have functions in the cell that are independent of the LTCC. We hypothesized that if β2.1 cardiac phenotypes were LTCC-dependent, then pharmacological inhibition of LTCC should produce a similar impairment of cardiac function and morphology, i.e., a “phenocopy.” We impaired LTCC function with nifedipine, a well-characterized dihydropyridine (DHP) LTCC antagonist. In a converse approach, we stimulated LTCC function with BayK (a.k.a. BayK8644), a potent LTCC agonist. The predication was that if β2.1 roles are LTCC-dependent, then nifedipine would phenocopy the effects of β2-depletion on cardiac morphology and function, whereas BayK might rescue these effects.
Wild-type embryos exposed to 20 μM nifedipine from 24 to 48 hpf developed pericardial edema and dysgenic, unlooped hearts similar to β2.1 morphant embryos (Fig. 2C). Lethality occurred around day 6. Moreover, treatment with 20 μM nifedipine depressed contractility and reduced the average heart rate from 140.4 bpm in untreated controls to 106 bpm (P < 0.001), a rate that was statistically similar to β2.1 morphants (97 bpm; P = 0.31; Fig. 5A). The effects of nifedipine were dose-responsive (data not shown). Thus, blocking LTCC activity by means of nifedipine did affect heart morphology and function in a manner similar to β2-depletion.
Next, we exposed embryos to 40 μM of BayK for 30 min before scoring the heart rate. BayK treatment increased the average heart rate of nifedipine-treated embryos to 126 bmp, a significant difference (P < 0.001), although still below the rate of untreated controls. BayK treatment also significantly increased the average heart rate β2 morphant embryos (118 bpm; P < 0.001). The recovery heart rates after BayK stimulation were statistically similar for 20 μM nifedipine-treated and β2 morphant embryos (P = 0.29). Thus, the heart function phenotypes produced by depression of LTCC function or by β2.1-depletion could be ameliorated by stimulating the LTCC with the BayK agonist. These pharmacological studies provide support for the hypothesis that the contractile phenotypes of β2.1 morphant hearts are due at least in part to depressed calcium channel function.
Normal Specification of Early First Heart Field Cells in β2.1 Morphants.
Cardiac morphogenesis is a multistep process involving specification, patterning and differentiation of an initial set of cardiomyocyte precursors derived from the first heart field, followed by additional contribution of cells derived from the second heart field (Bakkers, 2011). We assessed the progression of β2.1-depleted hearts through these stages using a series of markers in in situ hybridization (ISH) experiments. To assess the initial specification of a pool of cardiac progenitors, we examined expression of fgf8, one of the earliest markers of primary cardiac progenitors expressed in the bilateral heart fields by the four-somite stage. fgf8 has the ability to induce the expression of other cardiac genes that contribute to the growth and differentiation of the heart (Alsan and Schultheiss, 2002; Marques et al., 2008). We find that at the four-somite stage, β2.1 morphant and wild-type embryos showed no differences in fgf8 expression in the bilateral heart fields (Fig. 6A,B), suggesting that a normal pool of precursors was initially specified in β2.1-depleted embryos.
Normal Differentiation at the Venous and Arterial Poles in β2.1 Morphants
Following formation of the linear heart tube in zebrafish (and other vertebrates), additional cells augment the length of the heart tube at both the venous and arterial poles (de Pater et al., 2009; Hami et al., 2011; Lazic and Scott, 2011). At the venous pole, wild-type embryos express bone morphogenetic protein 4 (BMP4) in a population of cells marking the sinus venosus by 48 hpf. Islet-1 mutants, which show reduced differentiation at the venous pole, failed to express BMP4 in the sinus venosus (de Pater et al., 2009). Mutation of the Bmp type 1 receptor alk8 in laf embryos leads to hearts with a reduced atrium but normal-sized ventricle, further supporting the role of BMP signaling in venous pole differentiation (Chocron et al., 2007; Marques and Yelon, 2009). To assess differentiation at the venous pole in β2.1 morphants, we evaluated BMP4 expression in the sinus venosus. At 48 hpf, both wild-type and β2.1 morphants expressed BMP4 normally in the venous pole (Fig. 6C,D). Although β2.1 morphants showed bradycardia, they did not exhibit the frequent sinus pauses, AV block, or other arrhythmias that are frequently associated with impairment of pacemaker activities located in the sinus venosus region (Dobrzynski et al., 2007; de Pater et al., 2009; Arrenberg et al., 2010).
At the arterial pole between 24 and 48 hpf, additional cardiomyocytes differentiate in an fibroblast growth factor and Mef2cb-dependent process that elongates the distal portion of the ventricle and the outflow tract (de Pater et al., 2009; Lazic and Scott, 2011). At 30 hpf, after the linear heart tube has been formed, additional myocytes differentiating in the “late ventricular region” are added to the arterial pole (Lazic and Scott, 2011). These myocytes express vhmc and are arrayed in a splayed manner adjacent to the arterial pole of the heart tube (Lazic and Scott, 2011). The late ventricular region resembles the secondary or anterior heart field that extends the heart tube in amniotes (Kelly and Buckingham, 2002; Hami et al., 2011; Lazic and Scott, 2011). To assess differentiation at the venous pole in β2.1 morphants, we evaluated vhmc expression in 30 hpf embryos. Both wild-type and β2.1 morphants expressed vhmc in a splayed array of cells adjacent to the arterial pole of the heart tube (Fig. 6E,F). Therefore, embryos depleted for β2.1 did not appear to be defective in the population of late-differentiating myocytes at the cardiac poles.
Morphant Hearts Contain Fewer Cardiomyocytes Due to a Reduction in Rate of Proliferation
To investigate whether reduced cardiac numbers contributed to β2.1 phenotypes, we quantified the total population of cardiomyocytes in each cardiac chamber at 48 hpf. A Tg(myl7:nDsRed2) transgenic line allowed the visualization of cardiac myocytes. In the atrium, no significant difference in cardiomyocyte number was observed between wild-type and morphant at 48 hpf (Fig. 7). However, the ventricles of morphant hearts contained 25% fewer cardiomyocytes than wild-type hearts (n = 20 hearts per group).
To determine whether the decrease in cell number was due to apoptosis, TUNEL (terminal deoxynucleotidyl transferase–mediated deoxyuridinetriphosphate nick end-labeling) assays were used. Overall, the rate of apoptosis was very low within the hearts (fewer than five cells per heart), and no differences were detected between wild-type and morphant hearts (data not shown).
To determine whether cell proliferation rates were reduced in β2.1 morphant hearts, we assayed for actively proliferating cells by bromodeoxyuridine (BrdU) uptake within 6 hr windows ranging from 24 hpf through 48 hpf. BrdU is incorporated into newly synthesized DNA of replicating cells during S phase of the cell cycle. β2.1-depleted hearts exhibited significantly fewer proliferative cardiomyocytes during early looping stages (24–36 hpf) and during chamber ballooning stages (42–48 hpf; Fig. 8). However, embryos did not exhibit any difference in BrdU-uptake from 36–42 hpf. These data, together with the counts taken on transgenic lines, suggest that depletion of β2.1 led to a reduction in the number of cardiomyocytes in the ventricle, and that one mechanism of β2.1 function involves the regulation of the cell replication at several stages of heart tube morphogenesis.
Cell Adhesion Is Compromised in β2.1 Morphant Hearts
We hypothesized that β2.1, as a MAGUK protein, might be involved in a type of scaffolding function affecting the integrity of cytoskeletal or cell junction components in cells that comprise the heart tube. Following heart tube formation around 26 hpf, contractions in the embryonic heart increase in force and frequency over the next few days of embryonic development (Stainier et al., 1993; Baker et al., 1997). The continued development of cardiac morphology and cardiac function requires that cardiomyocytes form and maintain complex internal cytoskeletal structures as well as appropriate cell-to-cell connections (Luo and Radice, 2003; Bagatto et al., 2006). To test the structural integrity of heart tubes, we tested their ability to withstand force applied by overlaying a glass coverslip in situ. The weight of the coverslip was sufficient to flatten wild-type heart tubes, which nevertheless remained intact. In contrast, coverslips gently laid over morphant heart tubes caused the cells to lyse and disperse in 30% of the embryos (Fig. 9). These data suggest that morphant heart tubes are comprised of cells with an inherently weaker cellular ultra-structure or less robust intercellular attachments.
To further investigate the integrity of morphant cardiomyocytes, we examined the structure of the actin cytoskeleton. In differentiating cardiomyocytes, cellular actin is assembled into stress fibers that spread beneath the sarcolemma, lending structural rigidity and shape to cells (Pollard and Cooper, 2009). Actin is also in integral component of the sarcomere without which muscular contraction would be impossible. We examined cytoplasmic actin in cardiomyocytes using rhodamine-labeled phalloidin to determine if stress fibers were reduced or misaligned in morphant hearts. By 36 hpf, wild-type cardiomyocytes contained actin stress fibers that ran the length of the cell below the membrane (Fig. 10A). The Tg(myl7:EGFP-HsHRAS) transgenic line, which exhibits green fluorescent protein (GFP) localized at the plasma membrane, provided a visualization of the cell periphery (Fig. 10B,D,F,H). By 48 hpf, actin was clearly present in regular condensed units characteristic of sarcomeres, which develop initially in zebrafish in the perimembrane region (Fig. 10C arrow; Huang et al., 2009). In β2.1-depleted hearts aged 36 to 72 hpf, stress fiber localization and organization at the cell periphery was indistinguishable from that of wild-type hearts. β2.1 morphant hearts, however, failed to display the characteristic banded pattern of sarcomeric actin at any of the three time points, suggesting that sarcomeres in morphant cardiomyocytes were fewer or less organized (Fig. 10A,C,E).
Within the embryonic heart, adherens junctions are considered to be the most “load bearing” of cellular attachments between neighboring cells (Noorman et al., 2009). N-cadherin (cdh2) is the major classical cadherin that has been identified in adherens junctions of cardiac cells (Noorman et al., 2009). N-cadherin mediates calcium-dependent homophilic cell adhesion. The cytoplasmic domain of the cadherin protein is directly linked to the actin cytoskeleton by means of scaffolding proteins, including MAGUK family proteins (Lilien et al., 2002; Nechiporuk et al., 2007). Some evidence suggests that myofibrillogenesis in embryonic cardiomyocytes depends on N-cadherin. Disrupting N-cadherin function can inhibit the expression of sarcomeric proteins such as α-actin in and myosin (MacGrogan et al., 2011). We therefore assayed the distribution of N-cadherin protein at cell membranes of wild-type and β2.1-depleted hearts using an anti-pan-cadherin antibody in immunohistochemistry experiments. N-cadherin protein localized uniformly near the cell periphery of the cardiomyocytes in wild-type hearts, similar to the expression of membrane associated GFP in the Tg(myl7:EGFP HsHRAS) line (Fig. 10G,H). In contrast, β2.1-depleted hearts displayed substantially less N-cadherin localized to the cell periphery (Fig. 10G). Overall, 66% of β2.1-depleted hearts showed a strong reduction in N-cadherin signal localized at the cell periphery (n = 18). Taken together, our data suggests that the presence of β2.1 in the cell impacts the amount of N-cadherin localized to the cell periphery. We propose that this in turn adversely affects the maintenance of adherens junctions and integrity of the heart tube.
β2.1 is an auxiliary calcium channel β subunit expressed in zebrafish embryonic heart. To determine the role of β2.1 in cardiac development, we depleted β2.1 gene products in the embryo by morpholino knockdown. Because mammalian β2 is a major regulator of the cardiac LTCC, we predicted its knockdown might lead to impaired cardiac function in the zebrafish embryo. A significantly slower heart rate and weak cardiac contractions in β2.1-depleted embryos confirmed this hypothesis. Here we show that β2.1 is essential for growth of the ventricular chamber, for progression of heart tube looping and cell shape changes associated with chamber ballooning, and for the structural integrity of the heart tube. A particular splice variant, β2.1 tv_6, is sufficient to mediate these effects. The mechanistic basis for the β2.1-depletion phenotypes involves several aspects. First, the initial specification of cardiac precursors, the assignment of chamber-specific cell fates, cell survival, and addition of cells to the heart tube at arterial and venous poles all occur normally in β2.1-depleted hearts. β2.1 is, however, required for appropriate cardiomyocyte proliferation in the ventricle throughout most of the period of cardiac looping and chamber morphogenesis. In addition, although cardiac cells in β2.1-depleted embryos appear healthy and generate normal actin stress fibers, N-cadherin is severely depleted from the cell periphery in morphants, suggesting that the attachments of neighboring cardiac cells by means of adherens junctions are compromised in these embryos.
A major finding of this report is that embryos depleted for β2.1 form a heart tube comprised of significantly fewer ventricular cardiomyocytes compared with wild-type. In β2.1-depleted hearts, fewer cardiomyocytes enter S phase of the cell cycle, as assayed by BrdU incorporation over the course of cardiac development. To our knowledge, this is the first report to demonstrate an effect of a β subunit on cell proliferation, but additional studies will be needed to define whether this effect is direct or indirect.
The developmental functions for β2 proteins should be considered in light of both potential calcium channel-related or channel-independent activities. β2 subunits might affect cellular processes through protein:protein interactions mediated by their MAGUK domains, especially the SH3 and GK regions. Precedent exists for MAGUK-family proteins in regulation of epithelial cell proliferation and density through the stabilization of cyclins during mitosis and through interaction with transcription factors necessary to carry out cell division (Balda et al., 2003; Capaldo et al., 2011). Alternatively, as auxiliary subunits to cardiac LTCC, the influence of β2 proteins on the expression and gating of LTCC may impact calcium signaling which subsequently could lead to altered cell proliferation and gene transcription. As one example, calcineurin is a protein phosphatase responsive to changes in intracellular calcium following signal transduction events. Cardiac-specific mutagenesis of calcineurin in mice led to altered expression of calcium-handling genes in the embryonic heart, and was associated with a reduced number of cardiomyocytes (Maillet et al., 2010). In 2008, Meissner and Noack showed that blocking LTCC conductivity with channel antagonists in cultured human epithelial cells resulted in reduced proliferation, whereas treating with channel agonists rescued the cell number (Meissner and Noack, 2008). Likewise, mutations in zebrafish that reduce LTCC function have been associated with fewer cells in the developing heart tube (Rottbauer et al., 2001). The zebrafish island beat (isl) allele encodes a null mutation in the alpha subunit of the Cav1.2, CACNA1C (Rottbauer et al., 2001). Homozygous isl mutant embryos contained 43% fewer ventricular cardiomyoctes. The atria of isl embryos fibrillate while the ventricles fail to contract at all. Although atrial fibrillation was never observed in β2.1-depleted hearts, contractility was weakened, and heart function compromised. Thus, there is good precedent that LTCC-mediated calcium signaling can be a significant regulator of cardiomyocyte proliferation, leading us to propose that β2.1 may affect cell proliferation in a calcium channel-dependent manner.
In support of the idea that β2.1 operates in a calcium channel-dependent manner, we found that inhibition of LTCC activity by means of nifedipine generated cardiac phenotypes that resemble β2.1 knockdown. Moreover, LTCC function could be recovered either in nifedipine-treated or in β2.1-depleted embryos by application of the LTCC agonist BayK, which resulted in a partial rescue of cardiac phenotypes and further supported a link between β2.1 activity and LTCC function. Of note, LTCC activity appeared not to be completely eradicated in β2.1-depleted hearts, nor in hearts exposed to 20 μM nifedipine, because cardiomyocytes do still contract. A possible explanation for this observation in morphants is that other β genes expressed in the heart, including β4.1, β4.2, or β2.2, partially compensate for loss of β2.1, so that some LTCC activity remains. It is also possible that some aspects of β2.1 function are calcium channel-dependent, while others (such as the cell integrity functions) are calcium channel-independent. More detailed studies involving mutated or truncated constructs will be necessary to fully confirm exactly which β2.1-depletion phenotypes are LTCC-dependent.
Studies in zebrafish have indicated that multiple mechanisms determine heart size. Initially, a restricted population of cells in the lateral plate mesoderm is specified to become cardiomyocytes (Yelon, 2001). The recruitment of new cardiomyocytes and regulated cell division, combined with localized changes in cellular morphology, result in the growth and shaping of the developing embryonic heart (de Pater et al., 2009). Genetic analysis suggests that calcium signaling, paracrine communication, and transcription factor signaling cascades are major mechanisms that collectively regulate these morphogenetic processes in the forming heart in early development (Morin et al., 2000; Rottbauer et al., 2001; Ebert et al., 2005; Marques et al., 2008; Caprioli et al., 2011; Suzuki, 2011). In zebrafish, the cardiac mitogenic factor fgf8 is expressed as early as the three-somite stage in the incipient heart fields in the lateral plate mesoderm (Reifers et al., 2000). Genetic analysis has shown that fgf8 is one of the earliest players involved in specification of cardiomyocytes, and that it is required for the induction and patterning of myocardial precursors (Reifers et al., 2000). Embryos homozygous for fgf8 mutant alleles showed a decrease in expression of chamber specific markers at the 21-somite stage, and later developed a heart with significantly fewer cells at 48 hpf (Marques et al., 2008). In β2.1-depleted embryos, we found no difference in fgf8 expression at the four-somite stage, suggesting that the changes responsible for the reduction in cardiomyocytes most likely occur after the initial specification of the heart field. As development progressed, the ventricle appeared to be more strongly impacted by β2.1 depletion than the atrium. The ventricle developed with 25% fewer cells by 48 hpf. A lessor rate of proliferation was observed in the ventricular chamber at several stages before 48 hpf, which likely accounts for this difference.
The specification of secondary heart field cells to the venous or arterial poles appeared not to be impacted by β2.1 depletion. We note, however, that normal specification of cells at the venous or arterial poles does not necessarily suggest that these cells actually join into the developing heart. Indeed, we have inferred that β2.1 may play a role in cell junction formation, which could perturb migration of these cells into the heart and thus affect the total number of cardiomyocytes in the ventricle or atrium. To examine this alternative hypothesis, multicolor transgenic tools such as those developed by de Pater et al. (2009) or Lazic and Scott (2011), which can accurately track the differentiation and addition of cells to the poles of the heart, would be most helpful.
Cardiomyocytes that are actively contracting must maintain strong attachments to the ECM and to adjacent cardiomyocytes to facilitate mechanotransduction (Noorman et al., 2009). Several emerging studies suggest that cell junction integrity among cardiomyocytes plays important roles in regulating myocyte growth and survival, cardiac morphogenesis, and in establishment or maintenance of the normal cardiac contractile rhythm (Grossmann et al., 2004; Samarel, 2005; Bugorsky et al., 2007; Al-Amoudi and Frangakis, 2008). Although cardiomyocytes express several different cellular junctions, adherens junctions are thought to provide the major structural integrity to the chambers, which are subjected to constant contractions (Noorman et al., 2009). A major component of adherens junctions, the transmembrane protein N-cadherin, is connected to the intracellular actin cytoskeleton by means of catenins (Ebnet, 2008). The disruption of N-cadherin in the zebrafish glass onion mutant led to lethal cardiac phenotypes that depressed both cardiac function and morphology, and resulted in rounder, more loosely associated cardiomyocytes within the heart tube (Bagatto et al., 2006). In general, a large degree of actin remodeling accompanies the maturation of adherens junctions as they connect to the cytoskeleton (de Mendoza et al., 2010). In our study, embryonic hearts depleted of β2.1 protein showed a drastic reduction of N-cadherin protein localized to the cell periphery, suggesting that far fewer adherens junctions were present at the interface of adjacent cardiomyocytes. In a functional assay for adhesion, these same hearts demonstrated reduced ability to maintain tissue integrity under stress.
Others have reported that down-regulation of N-cadherin is associated with cell rounding and inability to extend lengthwise (Knollmann and Roden, 2008; Hara and Saito, 2009). In wild-type hearts, we observed that cells of the ventricular OC transitioned as expected from a rounded to elongated cell shape around 48 hpf, as chamber ballooning and cardiac looping proceeded (Auman et al., 2007). However, in β2.1-depleted hearts, the cells of the OC failed to elongate and instead retained their rounded shape, suggesting that the molecules or cytoskeletal organization needed to make this transition were lacking. In contrast, hearts of β2.1-depleted embryos were not substantially different in their levels or global organization of F-actin, supporting the idea that cytoskeletal organization was not globally altered in hearts lacking β2.1. While we favor the hypothesis that alterations in cell shape and heart tube integrity are affected in β2.1 morphants by means of changes in N-cadherin-mediated cell–cell attachment, we cannot completely rule out the alternative possibility that subtle alterations in cytoskeletal structures or mechanisms of passive tension act as contributing factors in the failure of β2.1-depleted OC cardiomyocytes to elongate as expected. Indeed, contractility is affected in these embryos, raising the possibility that sarcomeres are negatively affected, which might in turn impact cell shape.
The conversion of nascent cell–cell junctions at the plasma membrane into mature linkages between cells requires the appropriate assembly of junction proteins by MAGUKS and other scaffolding proteins (Knollmann and Roden, 2008; Marasco et al., 2009). The assembly of cell–cell junctions is dynamic and regulated by multiple protein complexes depending on the state of maturation (Velthuis et al., 2007; Ebnet, 2008). Cytoplasmic scaffolding proteins are thought coordinate a mechanical link between the junction and the actin cytoskeleton (Funke et al., 2005). In some cases, recruitment of additional proteins into a mature adherens junction at the membrane is dependent on calcium signaling and calcium oscillations at the membrane. For example, treatment of fibroblasts with calcium channel blockers reduced cadherin-mediated adhesion by 60% (Ko et al., 2001). Thus, it is reasonable to propose that β2.1 might affect the integrity of adherens junctions indirectly, by means of modulation of LTCC calcium signaling. Alternatively, β2.1 protein might act by means of its MAGUK protein:protein interaction domains to recruit or stabilize proteins needed to establish or maintain adherens junctions.
We show here that β2.1 is essential for cardiac development and function. Specifically, β2.1 is critical for regulation of cardiomyocte proliferation and for the integrity of cardiomyocyte adhesions, mostly likely mediated through adherens junctions. Although potentially indirect, both of these functions represent novel roles for a β subunit.
Morpholino Design and Validation
The β2.1 gene includes multiple translational start sites encoded by alternatively spliced exons at the N terminus (Ebert et al., 2008a). To simultaneously knockdown all known splice variants of β2.1, a morpholino (Gene Tools) was designed to target the splice donor site of exon 5 (MO1: 5′-CCACCAGTCATTGTTAAACTTCT GT). Blocking the donor site is predicted to lead to the joining of exon 4 to exon 6, thus introducing a frameshift and a premature stop codon nine residues into exon 6. The truncated β2.1 protein would lack the GK domain, rendering it unable to interact with the α subunit. Analysis of a dilution series indicated that a dose of 750 μM produced the greatest proportion of cardiac phenotypes with a minimum of off-target effects. A second morpholino was designed to block the translation initiation codon of the β2.1_tv6 transcript (MO2: 5′-GCC TTCATCAGCCAATCCGATTAGA).
RNA was extracted from 48-hr wild-type and β2.1-depleted embryos using Trizol (Sigma). An oligo-dT primer (Invitrogen) was used to generate cDNA, which was subsequently amplified by PCR using gene specific primers for β2.1_tv6 EU301439.1 (5′-GCGCAACTCGACAAGGCCAAGAGT A [bp 439–463, in exon 3] and 5′-GCC CTTGAGTGAAGGTCCAACGAGA [bp 891–867, in exon 10; 428-bp amplicon]) and β2.2 NM_001081748.2 (5′-GGGG CCCACTGAGGAGAAGGAGAAGA [bp 452–472 in exon 3] and 5′-TAAGTGA AGGCCCCATGAGCACCAC [bp 919–895 in exon 10; 467-bp amplicon]) and control primers complimentary to EF1a (5′-CGGTGACAACATGCTGGAGG, 5′-ACCAGTCTCCACACGACCCA). The specificity of all primers was confirmed by sequencing RT-PCR products.
cRNA of zebrafish β2.1_tv6 (GenBank accession no.: EU301439) obtained using RT-PCR was used for all rescue experiments. RNA was made using the mMessage Machine RNA kit (Ambion) according to manufacturer's instructions.
To quantify the reduction in cardiac looping of β2.1 morphants, we compared morpholino-injected and buffer-injected embryos using the Tg(myl7:EGFP) line. At 48 hpf embryos were fixed in 4% paraformaldehyde (PFA) and mounted in 1% agarose with the heart optimally positioned for imaging. We quantified looping by measuring the angle formed between the anterior/posterior axis of the embryo and the cross-sectional plane of the AV junction. To asses cellular morphology, morphant and wild-type hearts were imaged at 48 hpf in the Tg(myl7:EGFP-HsHRAS) line. Cardiomyocytes in the outer curvature of the ventricle were designated according to specifications set by Aumen et al. (2007). The area of individual cardiomyoctes was measured using the ImageJ software.
Nifedipine (Sigma product N7634) or BayK8644 (Sigma product B133) was diluted in E3 embryo media. Embryos were bathed in 20 μM nifedipine beginning at 24 hpf for 24 hr. At 48 hpf, a portion of embryos were additionally exposed to BayK (40 μM final concentration) for approximately 30 min before scoring heart rates and cardiac morphology.
Cardiac Performance Assays
Ten-second videos were taken of the hearts at 30 and 48 hpf on a Red Lake (Tallahassee, FL) digital video camera at 250 frames/sec. Frames were acquired from the video sequences using the Pinnacle Studio 8 program (Mountain View, CA). For digital analysis, individual frames were opened in SPOT Software imaging (Diagnostic Instruments Inc.). Considering the ventricular chamber as an ellipse, the major axis (length) and minor axis (width) were selected and measured (in microns). For each video, the 3 frames that best represented end-systolic points and 3 frames representing end-diastolic points were measured and the data averaged for that heart. Next, the estimated volume of the ventricular chamber in nanoliters (nl) was calculated based the formula of a prolate spheroid: 4/3 π × a × b2 where a = radius of the major axis and b = radius of minor axis, according to the method of (Jacob et al., 2002). The mean stroke volume was calculated as the difference in ventricle volumes at diastole and systole. Cardiac output was calculated as the stroke volume multiplied by the heart rate, and reflects blood flow through the chamber in nanoliters per min (nl/min). The physical parameters of heart rate, stroke volume, and cardiac output were determined for 10 morphant and 10 wild-type embryos.
We assessed the integrity of the heart tube in control and β2.1-depleted embryos of the Tg(myl7:nDsRed2) transgenic line. A 48 hpf embryo was placed on a slide in a 10-μl drop of E3 media (ZFIN) and a cover slip was gently lowered on top of it to slightly flatten the heart tube, but no further pressure was applied. Freshly prepared samples were imaged immediately using a Leica 5500 microscope. Assays were performed blinded and in a single sitting. A limitation of this assay is that hearts were not pharmacologically arrested to any uniform state of contraction before application of the coverslip. Hence, it is possible that hearts of individual embryos varied in their state of contraction at the instant of coverslip application. For this reason, several embryos (n = 20) were surveyed for each genotype.
In Situ Hybridization and Immunohistochemistry
Embryos younger than 24 hpf were staged by counting the number of somites (Westerfield, 1995). The following digoxigenin (Roche) -labeled probes were used for in situ hybridization: fgf8, vmhc, and myl7. Embryos were fixed in 4% PFA for 2 hr and subjected to permeablization and probe hybridization as previously described (Thisse and Thisse, 2008). Probe signal was detected using the NBT/BCIP chromogenic reagents (Roche).
TUNEL assay was performed according to manufacturer's protocols as per the ApopTagPlus Peroxidase In Situ Apoptosis Detection Kit (Millipore, catalog no. S7101). Embryos were then processed for immunohistochemistry using MF20 antibody to label cardiac specific myosin. BrdU analysis was used to quantify actively dividing cardiomyocytes at different times in development in conjunction with MF20 immunohistochemistry. The embryos were washed in 10 Mm BrdU reagent (Roche) and then developed for 6 hr. Afterward, they were fixed in DENTS fixative overnight and then rehydrated to PBT. Embryos were treated with Proteinase K to permeabilize tissues, followed by a 10-min fixation in PFA and 1-hr incubation in 2 N HCl. After blocking in bovine serum albumin (BSA) and sheep serum embryos were incubated in MF20 (1:50) overnight, followed by overnight incubation in secondary alkaline-phosphatase conjugated anti-mouse (1:3,000; Sigma) and an Alexa488 conjugated anti-BrdU (1:100; Caltag Labs). Detection of MF20 was achieved with the NBT/BCIP reagent. Hearts were dissected, mounted in a thick section, and imaged for Alexa488 fluorescence. Embryos older than 30 hpf were drained of blood before fixation to exclude erythrocytes that might have incorporated BrdU.
N-cadherin, the only classical cadherin known to be expressed in the heart, was detected using a pan-cdh Ab (Sigma; Noorman et al., 2009). Tg(myl7:EGFP-HsHRAS) control and morpholino injected hearts were dissected after overnight fixation in 1% formaldehyde. Hearts were subjected to MeOH and phosphate buffered saline-Triton X-100 (PBS-TX) washes followed by blocking in PBS-TX containing 1% (w/v) BSA. Samples were incubated in primary pan-cdh (1:300) overnight followed by secondary goat anti-rabbit Alexa546 (1:200; Invitrogen). All fluorescence labeling was imaged using the Leica 5500 fluorescent microscope and IP Lab imaging software.
F-Actin Labeling With Phalloidin
Wild-type and β2.1-depleted Tg(myl7:-EGFP-HsHRAS) embryos were fixed in 4% PFA. Explanted hearts were washed in PBS-TX twice and blocked in PBT with 1% dimethyl sulfoxide (DMSO) and 10% BSA (w/v) for 30 min. Rhodamine conjugated phalloidin was added in a 1:100 concentration for 30 min. Hearts were then rinsed in PBT and imaged with a Zeiss LSM 510 confocal microscope.
Histograms present the mean value ± SEM. Statistical analysis was performed using analysis of variance or Student's t-test. One asterisk represents a significant difference (P < 0.05), whereas two asterisks represents a highly significant difference (P < 0.001).
We thank Dr. Lindsay Parrie for assistance with experiments. In situ hybridization probes were generously provided by Drs Deborah Yelon, Qin Liu, and Kristin Artinger. D.M.G. was funded by National Science Foundation grant IOS 0719083.