Autosomal recessive centronuclear myopathy (CNM2), caused by mutations in bridging integrator 1 (BIN1), is a mildly progressive neuromuscular disorder characterized by abnormally centralized myonuclei and muscle weakness. BIN1 is important for membrane sensing and remodeling in vitro in different cell types. However, to fully understand the biological roles of BIN1 in vivo and to answer critical questions concerning the muscle-specific function of BIN1 in vertebrates, robust small animal models are required. In this study, we create and characterize a novel zebrafish model of CNM2 using antisense morpholinos. Immunofluorescence and histopathological analyses of Bin1-deficient zebrafish skeletal muscle reveal structural defects commonly reported in human CNM2 biopsies. Live imaging of zebrafish embryos shows defective calcium release in bin1 morphants, linking the presence of abnormal triads to impairments in intracellular signaling. RNA-mediated rescue assays demonstrate that knockdown of zebrafish bin1 can reliably examine the pathogenicity of novel BIN1 mutations in vivo. Finally, our results strongly suggest that the phosphoinositide-binding domain of BIN1, present only in skeletal muscle isoforms, may be more critical for muscle maturation and maintenance than for early muscle development. Overall, our data support that BIN1 plays an important role in membrane tubulation and may promote skeletal muscle weakness in CNM2 by disrupting machinery necessary for excitation–contraction coupling in vertebrate organisms. The reproducible phenotype of Bin1-deficient zebrafish, together with the generalized advantages of the teleost system, makes this model readily adaptable to high-throughput screening strategies and may be used to identify therapies for CNM2 and related neuromuscular diseases.
Centronuclear myopathies (CNMs) are a heterogeneous group of congenital disorders characterized by muscle weakness and abnormal nuclear centralization in myofibers. Several genetic forms have now been described, and vary in terms of age of onset, severity of clinical symptoms and mode of inheritance. Pathogenic CNM mutations are most commonly reported in the phosphoinositide (PI) phosphatase myotubularin (MTM1) and the large GTPase dynamin 2 (DNM2) genes, resulting in neonatal X-linked (CNMX; MIM #310400) and autosomal dominant (CNM1; MIM #160150) forms of the disease, respectively (1,2). Variants in other genes, including the ryanodine receptor calcium release channel (RYR1), titin (TTN) and myotubularin-related protein 14 (MTMR14), have also been identified in rare cases with centralized nuclei on muscle biopsy (3–6). Autosomal recessive CNM (CNM2; MIM #255200) is most often associated with homozygous partial loss-of-function mutations in the bridging integrator 1 (BIN1) gene (7). Compared with MTM1- and DNM2-related CNM, pathological mechanisms underlying BIN1-related CNM remain particularly unclear, owing to fewer available human biopsies as well as a lack of faithful and accessible animal models.
BIN1 is a scaffolding protein involved in the tubular invagination of membranes and in membrane trafficking (8–10). Alternative splicing of the BIN1 gene generates at least 12 isoforms with diverse functions (11–14). All known protein isoforms possess an N-terminal amphipathic helix, a BAR (Bin/Amphiphysin/Rvs) domain able to sense and promote membrane curvature through dimerization and a C-terminal SH3 domain important for protein–protein interactions (15–17). Variability arises in the protein's central region, where a clathrin-binding domain is present in neuronal isoforms and a PI-binding domain is found almost exclusively in skeletal muscle isoforms (11,13,18). Five of the six partial loss-of-function BIN1 mutations in CNM described to date are located in the ubiquitously expressed and evolutionarily conserved BAR and SH3 domains (Fig. 1D). BAR missense mutations strongly decrease BIN1 membrane tubulating properties in cultured cells, whereas SH3 truncating mutations impair interactions with DNM2 (7). Only recently was the first human BIN1 splice mutation affecting the muscle-specific PI-binding domain identified, and found to result in an unusually progressive form of the myopathy (19). The molecular basis of the muscle specificity of CNM2 remains largely unresolved.
BIN1 expression increases dramatically during muscle cell growth and differentiation (12,20). In humans, the subcellular localization of BIN1 protein shifts during skeletal muscle development. BIN1 immunoreactivity appears as longitudinal striations along the muscle fiber in neonates and switches to a transverse orientation by 3 months of age (18). BIN1 co-localizes with transverse (T)-tubules throughout this reorganization, which are deep invaginations of the sarcolemma that together with the sarcoplasmic reticulum (SR) compose the triad and facilitate excitation–contraction coupling (ECC) (13,18). Studies conducted in vivo provide evidence to also implicate BIN1 in the biogenesis of these structures. Null mutations in the Drosophila melanogaster orthologue of mammalian BIN1 result in viable, flightless flies with severely disorganized and reduced number of T-tubules (21). Similar disruptions in skeletal muscle, such as the mislocalization of T-tubules to Z-lines, have been observed in early Bin1 null mouse embryos and reproduced in isolated adult murine myofibers (22,23). Short hairpin RNA silencing of Bin1 gene expression in mouse skeletal muscle leads to T-tubule structural abnormalities, disrupted expression patterns of nascent T-tubule markers, and defects in intracellular calcium release (23). Studies in vitro and in cultured cells have attributed the localization patterns and membrane tubulating properties of BIN1 to the PI domain specifically found in skeletal muscle isoforms (7,24,25).
Robust animal models are necessary to fully elucidate the biological roles of BIN1 in vivo and to answer critical questions concerning the muscle-specific function of BIN1 in vertebrates, similar to the animal modeling already accomplished for the MTM1- and DNM2-related CNM forms. Knockout mice generated by targeted mutagenesis of Mtm1 and Labrador retrievers with MTM1 mutations exhibit progressive motor deficits and triad defects (26,27). The zebrafish model for X-linked CNM generated by morpholino (MO) knockdown of the mtm1 gene displays similar developmental delays and ultrastructural abnormalities, including mislocalized nuclei and disorganized organelles (28,29). Tissue-specific excision of exon 4 in mouse Mtm1 suggests that the muscle phenotype observed in animal models of myotubularin deficiency is because of loss of Mtm1 function in skeletal muscle (26). Regarding autosomal dominant CNM, Dnm2 knockout mice die before embryonic day 10, but knock-in mice heterozygous for the most common pathogenic mutation, p.R465W (c.1393C>T), present with a mild myopathic phenotype (30,31). Skeletal muscle weakness and disruptions of mitochondrial and T-tubule networks are also seen in a second murine model created by intramuscular adeno-associated virus (AAV) injections of DNM2–R465W into adult skeletal muscle tissue, as well as in zebrafish models overexpressing mutated human DNM2 (32,33).
The Inherited Myopathy of Great Danes (IMGD) canine model recently characterized by Jocelyn Laporte and colleagues is the first animal model to reproduce CNM2 human pathology, as the perinatal lethality of Bin1 null mouse embryos precludes extensive studies in skeletal muscle (19,22). Although larger mammals such as mouse and dog are beneficial and necessary to study the effects of a few lead therapeutic candidates, large-scale drug screening is not feasible in these models. There is thus significant benefit to developing small vertebrate models, especially for a disease still in the earliest stages of identifying therapeutic strategies. Zebrafish are greatly amenable to high-throughput chemical screens owing to their ease of husbandry and breeding, rapid ex utero development and high spawning productivity. Zebrafish embryos are also permeable to small molecules, and their transparent chorions allow for simple developmental observations (34). For the study of skeletal muscle pathologies in particular, zebrafish offer important research advantages. Genes essential for muscle development and function, as well as the physical ultrastructure of the sarcomere, are highly conserved between mammals and zebrafish (35). Additionally, muscular disorganization and locomotive impairment can be quickly and non-invasively quantified in the zebrafish over the first few days after fertilization using birefringence and touch-evoked escape behavior assays (36).
Therefore, to develop a small vertebrate model of CNM2 suitable for high-throughput chemical screens, we created and characterized the first zebrafish model of CNM2 using antisense oligonucleotides. Bin1-deficient zebrafish reliably reproduce the morphological abnormalities, impaired early motor functions, defects in skeletal muscle ultrastructure and histopathology observed in human CNM2 patients. Our bin1 knockdown morphants offer powerful insights into the biological functions of BIN1 in vivo as well as in CNM2 disease pathogenesis.
MO knockdown of zebrafish bin1
The Ensembl genome browser identifies two zebrafish genes as orthologues to human BIN1: bridging integrator 1 (bin1; NM_001004602.1) and myc box-dependent-interacting protein 1-like (LOC563561; XM_001921399.4). Of these two, only the 1878 bp zebrafish bin1 located on chromosome 6 (NM_001004602.1) shares strong synteny with the human gene, suggesting it to be the truest orthologue of human BIN1 (Supplementary Material, Fig. S1). Zebrafish bin1 encodes a 364 amino acid protein that is highly conserved (67% identity, 84% similarity) with the longest human BIN1 isoform (Fig. 1D). Reverse transcriptase-polymerase chain reaction (RT–PCR) and western blots confirmed that bin1 is actively expressed in the developing fish (Fig. 1B and C) and whole-mount in situ hybridizations localized this expression to skeletal and cardiac muscle (Fig. 1A). Expression of bin1 in zebrafish striated muscle is consistent with in situ data generated previously using high-throughput analyses (37) and made available on the Zebrafish Model Organism Database (http://zfin.org). To ablate the function of bin1 in zebrafish, two independent MOs were designed to the translational start site (ATG MO) and to the splice donor site of exon 8 (Ex8 MO) (Fig. 2A and C). The ATG MO was predicted to block bin1 translation and the Ex8 MO was expected to introduce premature stop codons to the transcript by retaining the adjacent intron. Both MOs were independently injected into 1-cell stage embryos. Embryos were phenotypically analyzed through 4 days post fertilization (dpf). A standard MO directed against human β-globin, a sequence of nucleotides not found in the zebrafish genome, was used in all experiments to control for injection-related non-specific effects in wild-type (WT) embryos.
Efficacy of the ATG MO to interfere with translation of zebrafish bin1 was verified by western blot. Reduced levels of endogenous zebrafish Bin1 protein were observed in 3 dpf bin1 morphants (Fig. 2B). Efficacy of the Ex8 MO to alter bin1 mRNA processing and stability was first examined by RT-PCR using primers targeted to flanking exons (Fig. 2D). DNA sequencing showed retention of the 105 bp intron between exons 8 and 9 in 3 dpf bin1 morphants and contains at least two putative stop codons likely to result in a truncated protein. Further analysis by western blot and quantitative real-time RT-PCR confirmed that injection of the Ex8 MO at high concentrations (10 ng) results in a lower-molecular-weight, presumably non-functional, form of Bin1 protein and that expression of the full-length bin1 transcript decreases with increasing concentrations of Ex8 MO (Fig. 2E and F).
Loss of zebrafish Bin1 results in abnormal morphology and impaired locomotor activity
Zebrafish embryos undergo rapid skeletal muscle development. Multinucleated myofibers are present by 1 dpf and muscle cells fully differentiate by 2 dpf. Morphological abnormalities in bin1 MO-injected embryos arise within this same time frame. Both MOs yielded indistinguishable phenotypes throughout early zebrafish development (Fig. 3). As early as 1 dpf, bin1 morphants display a dorsal curvature that is absent from WT controls (Fig. 3A). This curvature is reminiscent of the kyphosis/scoliosis often seen in CNM2 patients (38,39), and has been observed in other zebrafish models of congenital myopathy, including the closely related X-linked form of CNM (28). By 3–4 dpf, bin1 morphants develop a more pronounced kyphosis and exhibit bent and/or foreshortened tails, together resulting in an overall S-shaped appearance (Fig. 3B and C). Mild bradycardia was also noted in bin1 MO-injected 4 dpf larvae (Supplementary Material, Fig. S2). This finding is both consistent with earlier findings of diminished calcium transients and impaired ventricular contractility in Bin1-deficient zebrafish hearts (40), as well as with cardiac arrhythmias reported in human CNM2 patients (41).
Consistent with defects in morphology, bin1 morphants display deficits in early motor functions. The first recognizable muscle-dependent motor activity in zebrafish is spontaneous embryo coiling, detectable between 17 and 26 hours post fertilization (hpf). On average, WT embryos coil 10.6 ± 0.9 times per 15 s whereas bin1 morphants coil only 7.7 ± 0.8 times within the same time interval (n = 20) (Fig. S3; see also Supplementary Material, Movies S1 and S2). Motor functions in early zebrafish development can also be evaluated by measuring the rate of chorion hatching. By 60 hpf, 95.9 ± 0.7% of WT zebrafish embryos hatch naturally from their protective chorions, as opposed to only 46.0 ± 1.7% of bin1 morphants. This atypical delay suggests a mild early muscle weakness in the bin1 morphant phenotype (n = 20) (Fig. S3).
Zebrafish embryos swim in response to touch by 26 hpf, and the frequency of muscle contractions during swimming increases to a value comparable to that of adult zebrafish shortly thereafter (42,43). By 3 dpf, bin1 morphants display markedly impaired locomotion. WT larvae respond to tactile stimuli by swimming quickly out of the field of view, whereas bin1 morphants exhibit fluttering motions that either propel the larvae in circles or for only a few short lengths (WT: 6.3 ± 0.3 cm/0.1 s; bin1 MO: 0.30 ± 0.05 cm/0.1 s; n = 20) (Fig. S3; see also Supplementary Material, Movies S3 and S4). The highly diminished motor behaviors of bin1 morphants are indicative of reduced muscle function and overall skeletal muscle weakness.
Bin1 deficiency results in disorganized myofibers and molecular abnormalities similar to CNM2
The skeletal muscle architecture in Bin1-deficient morphants was examined to see if their abnormal morphologies and diminished locomotive behaviors result from structural defects in this tissue. Hematoxylin/eosin (H&E) staining of skeletal muscle in 4 dpf larvae revealed that the sarcomeric organization of myofibers is regularly disrupted around myonuclei in bin1 MO-injected embryos. Moreover, bin1 morphant myonuclei are mislocalized, rounded and grouped. Morphant myonuclei are also frequently larger and filled with darkly staining nucleoli compared with the smaller WT myonuclei, which are typically singular, elongated and located toward the fiber periphery (Fig. 4A–F).
Whole-mount direct immunostaining of 3 dpf larvae with phalloidin, a high-affinity probe for filamentous actin, showed that myofibrillar organization is disrupted in bin1 morphants compared with WT controls. Major trunk skeletal muscle fibers overlap and often cross over one another in bin1 morphants, whereas WT myofibers generally appear parallel (Fig. 4G–J). Such gross observations may at least partially explain the curved bodies of bin1 morphants, and are in accordance with the extensive myofibrillar disorganization reported in CNM2 patients (41). Additional immunofluorescence studies confirmed that dystrophin expression at the myosepta and the number and appearance of neuromuscular junctions in 3 dpf Bin1-deficient zebrafish skeletal muscle are normal (Supplementary Material, Fig. S4). Together, these data suggest that the bin1 morphant phenotype is a primary defect within myofibers, and does not arise from abnormalities at the sarcolemma and/or extracellular matrix.
Transmission electron microscopy was performed at 4 dpf to visualize the effect of bin1 knockdown on underlying skeletal muscle ultrastructure. The architectural integrity of Bin1-deficient myofibers is largely preserved. Bin1-deficient fish showed normal organization of the contractile apparatus, with no necrotic fibers or apoptotic nuclei, consistent with a non-degenerative myopathy. The most remarkable pathological defect of bin1 morphant muscle was the presence of malformed and highly disorganized triad structures. The majority of triads were abnormal in bin1 morphants, with distortions in size and/or shape, in comparison with WT controls (Fig. 4K–N; quantified in Fig. 7E). Terminal cisternae of the SR, which are localized adjacent to T-tubules, appeared to be collapsed in most bin1 morphant myofibers, whereas no significant abnormalities were detected in SR longitudinal vesicles. Other ultrastructural features unique to bin1 morphant skeletal muscle and absent from WT controls were large whorled membranous structures of ambiguous origin localized to perinuclear regions (Fig. 4O–R). Similar structures and T-tubule deformities have been reported in human CNM biopsies and in myotubularin-deficient zebrafish (18,28,29). Overall, these data support the hypothesis that BIN1 is important for membrane remodeling and maintaining the structural integrity of skeletal muscle, and that bin1 morphant zebrafish are a suitable model for human CNM2 on both the morphological and molecular levels.
Bin1 deficiency disrupts localization of triad markers and calcium signaling in zebrafish skeletal muscle
BIN1 is localized to T-tubules in human skeletal muscle and is hypothesized to play a role in the formation and/or maintenance of these structures. To test whether the triad defects observed in bin1 morphants are a direct consequence of Bin1 deficiency at T-tubules, localization of Bin1 in zebrafish skeletal muscle was examined in cultured 3 dpf zebrafish myofibers. Indirect immunofluorescence revealed that zebrafish Bin1 protein is expressed in a distinctive striated pattern in WT embryos that overlaps with that of the dihydropyridine receptor (Dhpr), an established T-tubule marker (Fig. 5A, top). The presence of Dhpr protein at T-tubule membranes was reduced in bin1 morphants (Fig. 5A, bottom). Knockdown of bin1 also perturbed expression patterns of other triad markers, including triadin (Trdn), an SR transmembrane protein (Fig. 5B). However, it did not appear to affect non-triad muscle proteins, such as the Z-line marker alpha-actinin (Actn) (Fig. 5C). These data agree with studies of BIN1 in human skeletal muscle (44), and importantly suggest that BIN1 is involved in maintaining the structural organization of triads in vivo.
The triad is the membrane structure controlling ECC in skeletal muscle by regulating the release of calcium from the SR to the cytosol. To directly examine whether disorganized triads in bin1 morphants may lead to impaired ECC and defects in calcium homeostasis in vivo, calcium transients were imaged in 3 dpf zebrafish embryos mosaically expressing an α-actin-driven GCaMP3 Ca2+ reporter (45,46) (Fig. 6A). Skeletal muscle contraction was evoked by the application of potassium chloride, which depolarizes cells and activates voltage-gated calcium channels. Peak calcium release in fluorescing myofibers expressing the EGFP-tagged GCaMP3 Ca2+ reporter was found to be significantly reduced in bin1 morphants compared with WT controls [WT: 3.8 ± 0.4 relative fluorescence units (RFU); bin1 MO: 1.8 ± 0.2 RFU; n = 12] (Fig. 6D). Fluorescence propagation through skeletal muscle over a fixed time interval was also qualitatively assessed in the examined embryos, and was noticeably diminished in bin1 morphants (Fig. 6B and C; see also Supplementary Material, Movies S5 and S6). These findings provide strong in vivo evidence to suggest that defective ECC and abnormal calcium signaling in BIN1 deficiency, resulting from structural alterations at triads, may be a primary cause of CNM2. The Bin1-deficient zebrafish is the first animal model to demonstrate this important link in skeletal muscle.
PI-binding domain is not crucial for Bin1 function during early zebrafish development
RNA-mediated rescue studies were performed to confirm that the muscular phenotype observed in bin1 morphants directly results from Bin1 deficiency. Human BIN1 encodes at least 12 different alternatively spliced transcripts, and those specifically expressed in skeletal muscle are distinguished by the presence of a PI-binding domain encoded by exon 11. The polybasic residue sequence of exon 11 (RKKSKLFSRLRRKKN) is required for BIN1-induced membrane tubulation when exogenously expressed in cultured cells (7,24). However, until recently, all human BIN1 mutations had been reported in ubiquitously expressed exons (19). To investigate the functional significance of exon 11 in vivo, we exogenously overexpressed two distinct human BIN1 isoforms [Fig. 1D (II and III)] in bin1 morphant embryos and evaluated the ability of each to rescue the skeletal muscle phenotype of the fish in terms of morphology, birefringence assay and electron microscopy at 4 dpf. Intriguingly, overexpression of either isoform 8 (with exon 11) or isoform 9 (without exon 11) rescues the severe skeletal muscle morphology of bin1 morphants, and both do so to a similar degree (Fig. 7A, B and D). Ubiquitous overexpression of either BIN1 isoform also significantly improves birefringence, a non-invasive measure of skeletal muscle organization (36). As the pronounced dorsal and tail curvature of bin1 morphants did not allow for accurate birefringence quantification of the whole fish, birefringence was quantified in five posterior somites consistently displaying flat orientation across all captured images. Bin1 MO-injected embryos displayed 53.8 ± 3.7% of the maximal birefringence observed in WT controls, whereas co-injection of WT human BIN1 mRNA and bin1 MO increased the mean pixel intensity of the selected somites to 78.4 ± 7.9% (isoform 8) or 75.8 ± 6.9% (isoform 9) (Fig. 7E). Skeletal muscle ultrastructure was largely corrected in RNA-rescued bin1 morphants. The fraction of abnormal triads in bin1 MO-injected embryos decreased from 74.4 ± 2.3% (n = 129) to 26.2 ± 5.0% (n = 151) when rescued with isoform 8 mRNA, and to 31.0 ± 4.3% (n = 161) when rescued with isoform 9 mRNA (Fig. 7C and E). This result strengthens the hypothesis that BIN1 plays a role in T-tubule biogenesis, and may do so independently of the PI-binding domain. Our rescue data demonstrate the specificity of MO-mediated bin1 knockdown and suggest that the BIN1 PI-binding domain may be more critical for skeletal muscle maturation and maintenance, rather than for early development.
Human BIN1 mutations are pathogenic in zebrafish model of Bin1 deficiency
In vivo RNA-mediated overexpression experiments were also used to test the ability of mutant human BIN1 to rescue the morphological abnormalities observed in bin1 morphants. Two pathogenic BIN1 mutations affecting the most highly evolutionarily conserved BIN1 protein domains were independently introduced into the WT human BIN1 gene by site-directed mutagenesis and the mutant mRNA transcribed in vitro [Fig. 1D (III)]. The p.K35N (c.105G>T) missense change alters the charge of the BIN1 BAR domain N-terminus, and is predicted to compromise BIN1 interactions with negatively charged membranes and its capacity to generate membrane curvature (7). The p.K575* (c.1723A>T) nonsense mutation removes the last alpha helix and two beta strands of the SH3 domain. This change alters the three-dimensional structure of the SH3 region and disrupts important BIN1 protein–protein interactions, such as those with large GTPase DNM2, a causative gene of autosomal dominant CNM (7,47). Overexpression of either mutant BIN1 was not able to rescue 4 dpf bin1 morphant morphology or birefringence (Fig. 7B, D and E). These data show that our bin1 morphant fish is a reliable tool for characterizing the pathogenicity of BIN1 mutations in vivo, and may be especially useful to quickly analyze novel BIN1 mutations as they are identified.
Bin1-deficient zebrafish as a small vertebrate model for CNM2
Although small animal models have been carefully characterized for the MTM1- and DNM2-related forms of CNM, the perinatal lethality of Bin1 null mice has hindered progress on this front to date (22). In this study, we use a MO-mediated knockdown approach to examine the role of Bin1 in early zebrafish development and to develop a faithful small vertebrate model of autosomal recessive CNM (CNM2). Our data demonstrate that Bin1 localizes to T-tubules in zebrafish skeletal muscle, similar to published observations in mouse, canine and human tissue (13,18,19). The loss of Bin1 in zebrafish disrupts this characteristic expression pattern and leads to similar morphological abnormalities and skeletal muscle defects as those reported in CNM2 patients. Furthermore, because Bin1-deficient zebrafish have a reproducible and clear phenotypic readout, this model is particularly suited for drug discovery applications. Zebrafish models have already been successfully used to demonstrate efficacy and toxicity of potential chemical therapeutics for human disease (48–50).
PI-binding domain of BIN1 may be more critical for late skeletal muscle maintenance rather than for early development
The molecular basis of the muscle specificity of CNM2 remains unclear. Of the 12 known alternatively spliced BIN1 isoforms, a PI-binding domain encoded by exon 11 is present only in isoforms expressed in skeletal muscle (11,13,18). In vitro studies have shown that the exon 11-encoded PI-binding motif is required for the membrane tubulating properties of BIN1 in cultured muscle cells (24). Its unique muscle-specific expression may be of particular relevance to CNM, as defects in the remodeling of PI-rich membranes are common to all of the major genetic forms (51), yet almost all CNM2 missense mutations described to date have been found in ubiquitously expressed exons and result in mild-to-moderate disease progression (7,38,39,41). Only recently has the first human splice mutation involving the skipping of the muscle-specific exon 11 been documented in a patient with rapidly progressive CNM2 (19).
The BIN1 PI-binding domain is not required to rescue the bin1 phenotype in zebrafish skeletal muscle. Exogenous ubiquitous expression of human BIN1 transcript variants with and without the canonical PI-binding domain both drastically improved bin1 morphant morphology, as well as muscle birefringence and ultrastructure, and did so to a similar degree (Fig. 7). These data reinforce the notion that BIN1 PI-binding ability may play a greater role in muscle's later maturation and maintenance, rather than in early muscle formation, and agree with the progressive integration of exon 11 in BIN1 mRNA during skeletal muscle cell differentiation and development (12,44). These observations also explain why exon 11 acceptor splice site mutations in dogs and humans present as a highly progressive phenotype later in life but are unaffected at birth and during early childhood. Additionally, the PI phosphatase activity of MTM1 is not required to completely rescue the X-linked CNM phenotype of myotubularin-deficient mice (52). Our rescue studies in zebrafish, coupled with recently published work in the CNM2 dog model (19), build on earlier findings that MTM1 is essential for skeletal muscle maintenance rather than for myogenesis by extending this functional role to another protein implicated in human CNM (26).
BIN1 functions in skeletal muscle and pathomechanisms of CNM
Immunohistochemical and ultrastructural analyses of Bin1-deficient zebrafish skeletal muscle revealed highly abnormal T-tubule morphology and mislocalization of triad proteins, supporting the hypothesis that BIN1 is essential for the formation and organization of these structures. Overall, Bin1 protein levels were greatly reduced in bin1 MO-injected zebrafish embryos, although not completely abolished (Fig. 2B). Residual Bin1 protein in bin1 morphant muscle fibers appeared to be mildly accumulated around myonuclei (Fig. 5). Similar findings were observed on biopsies from patients with BIN1 and MTM1 mutations (18), and are in accordance with the presence of a nuclear localization sequence in BIN1 (17). Studies performed in cultured cells demonstrate BIN1 to have both nuclear and cytoplasmic subcellular localizations under WT conditions, but point toward the primary role of BIN1 in the cytoplasm (13,17). However, it is feasible that BIN1 may be concentrated in the nucleus under certain functional or disease states (13). Our data are consistent with the hypothesis that CNM2 may not be caused by a generalized absence of BIN1 protein per se, but rather to a decreased association of BIN1 with triads (18).
Triad abnormalities are consistent with the known biochemical function of BIN1 and other members of the BAR protein family, which is to sense and induce membrane curvature through BAR domain dimerization and through the insertion of an amphipathic helix into membrane bilayers. Previous in vivo experiments have demonstrated that mutations in the Drosophila orthologue of mammalian BIN1 result in viable, flightless flies with disorganized triads (21). Murine Tibialis anterior injected with a U7 small nuclear RNA construct containing an antisense sequence to promote alternative splicing of Bin1 also display T-tubule abnormalities and skeletal muscle weakness (44). However, the Bin1-deficient zebrafish is the first animal model of CNM2 to provide a direct connection between malformed triads and aberrations in ECC. Calcium transients are reduced in bin1 morphant embryos compared with WT controls in terms of both peak calcium release and visualized signal propagation. Structural modifications to triads leading to disruptions in the transport of ion channels and subsequent intracellular calcium signaling may explain the skeletal muscle weakness and atrophy observed in CNM2.
Membrane remodeling defects very similar to those discovered in Bin1-deficient zebrafish have been reported in other published CNM animal models and in human patients. Disorganized triads and membranous whorls seen in bin1 morphant embryos are also found in myotubularin-deficient dog, mouse and zebrafish models, as well as in patients with MTM1 mutations involving protein loss (18,26–28,53). In agreement with our results for the Dhpr and triadin, abnormal triad markers have also been observed in MTM1- and DNM2-related CNM (18,54). Together, these commonalities suggest that MTM1, DNM2 and BIN1 may act in the same molecular pathway regulating membrane remodeling in skeletal muscle, and indicate a shared pathological mechanism between the different genetic forms of CNM.
MATERIALS AND METHODS
Fish and embryo maintenance
Zebrafish (Danio rerio) were bred and maintained as described previously (55). WT control embryos were obtained from the Oregon AB line and were staged by hours (hpf) or days (dpf) post fertilization at 28.5°C. All animal work was performed with approval from the Boston Children's Hospital Animal Care and Use Committee (11-05-1955R).
MO knockdown and mRNA rescue
Two antisense MOs, one targeting the translational start site (ATG MO) and one targeting the exon 8–intron 8 splice site (Ex8 MO), were designed to knockdown the zebrafish bin1 transcript (GeneTools LLC, Philomath, OR, USA). The MO sequences were bin1 ATG MO: 5′-TGACTCCTTTCCCAACCTCTGCCAT-3′ and bin1 Ex8 MO: 5′-TCTCTTATTATTGGCCTCACTTTGC-3′. An MO against human β-globin, which is not homologous to any sequence in the zebrafish genome by BLAST search, was used as a negative control for all injections (5′-CCTCTTACCTCAGTTACAATTTATA-3′). MOs were dissolved in 1X Danieau buffer with 0.1% phenol red and 1–2 nl (1–10 ng) injected into the yolk of 1-cell stage embryos.
For rescue experiments, full-length human BIN1 isoform 8 (NM_004305.3) and isoform 9 (NM_139350.2) cDNAs were cloned into a pCS2+ destination vector (a gift from Nathan Lawson) using Gateway technology (Invitrogen, Carlsbad, CA, USA). Substitutions p.K35N (c.105G>T) and p.K575* (c.1723A>T) were incorporated into BIN1 isoform 9 cDNA using GENEART site-directed mutagenesis (Invitrogen). mRNA for all constructs was synthesized in vitro using mMessage kits (Ambion, Austin, TX, USA). 50–200 pg of mRNA was injected into embryos at the 1-cell stage independently or in combination with ATG MO, and subsequent phenotypic analyses performed at 4 dpf.
Muscle birefringence was analyzed as described previously (36). Quantification data were calculated for five posterior somites (numbers 12–16) that exhibited flat orientation in both bin1 morphants and WT controls. Mean pixel intensity was divided by the skeletal muscle area for each zebrafish birefringence image using ImageJ (National Institutes of Health, Bethesda, MD, USA), a value that was then normalized to WT controls.
Zebrafish embryos at 3 dpf were homogenized in buffer containing Tris–Cl (20 mm, pH 7.6), NaCl (50 mm), EDTA (1 mm), NP-40 (0.1%) and complete protease inhibitor cocktail (Roche Applied Sciences, Indianapolis, IN, USA). Following centrifugation at 11 000g at 4°C for 15 min, protein concentration in supernatants was determined by BCA protein assay (Pierce, Rockford, IL, USA). Proteins were separated by electrophoresis on 4–12% gradient Tris–glycine gels (Invitrogen) and transferred onto polyvinylidene difluoride membrane (Invitrogen). Membranes were blocked in PBS containing 5% casein/0.1% Tween 20, then incubated with either rabbit polyclonal antibridging integrator 1 (1 : 250, SAB1408547, Sigma, St. Louis, MO, USA) or mouse monoclonal anti-β-actin (1 : 1000, A5441, Sigma) primary antibodies. After washing, membranes were incubated with horseradish peroxidase-conjugated antirabbit (1 : 2500, 170-6515) or antimouse (1 : 5000, 170-6516) IgG secondary antibodies (BioRad, Hercules, CA, USA). Proteins were detected using the SuperSignal chemiluminescent substrate kit (Pierce).
Touch-evoked escape response assay
Mechanosensory stimuli were delivered to 3 dpf embryos by touching the yolk sac or tail with an insect pin as described previously (36). Motor behaviors were recorded using a SPOT RT3 digital camera system (SPOT Imaging Solutions, Diagnostic Instruments Inc., Sterling Heights, MI, USA) mounted on a Nikon SMZ1500 stereomicroscope (Nikon Instruments Inc., Melville, NY, USA). Video frame capture (30 Hz) was performed using ImageJ (NIH).
Reverse transcriptase-polymerase chain reaction
Total RNA was prepared from zebrafish embryos using RNeasy fibrous tissue mini kits (Qiagen, Valencia, CA, USA). cDNAs were synthesized from 1 to 2 μg of total RNA using Superscript III reverse transcriptase (Invitrogen) and random hexamers. To measure bin1 knockdown in 3 dpf embryos, quantitative PCR amplification of cDNAs was performed with a Taqman assay for bin1 exon 10–exon 11 (Applied Biosystems, Austin, TX, USA) on a 7300 Real Time PCR System (Applied Biosystems). Gapdh served as the control to normalize bin1 expression using the 2−ΔΔCt method. To detect relative bin1 expression levels in early zebrafish development, RT-PCR was performed on cDNAs from <1 to 7 dpf embryos on a Tetrad 2 thermocycler (BioRad). Primer sequences to zebrafish genes bin1 or ef1α were as follows: bin1 forward: 5′-TGTCTGGCAGAAATGTATGACC-3′; bin1 reverse: 5′-TATCACTCAGATTCTGGTTCAGTTTG-3′; ef1α forward: 5′-TCACCCTGGGAGTGAAACAGC-3′; and ef1α reverse: 5′-ACTTGCAGGCGATGTGAGCAG-3′.
Myofiber cultures and immunofluorescence
Mixed cell cultures from 3 dpf embryos were obtained as described previously (28). Fixed cells were blocked in 10% goat serum/0.3% Triton, incubated in primary antibody overnight at 4°C, washed in PBS, incubated in secondary antibody for 1 h at room temperature (RT), washed in PBS, then mounted with Vectashield Mounting Medium (Vector Laboratories, Burlingame, CA, USA). Primary antibodies used were rabbit polyclonal antibridging integrator 1 (1 : 50, SAB1408547, Sigma), and mouse monoclonal antisarcomeric alpha-actinin (1 : 100, A7732, Sigma), anti-Dhpr 1 alpha (1 : 100, ab58552, Abcam, Cambridge, MA, USA), antitriadin (1 : 100, T3569, Sigma). After washing in PBS several times, samples were incubated with antimouse fluorescein FITC (1 : 100, 115-095-003) or antirabbit rhodamine TRITC (1 : 100, 111-025-144) secondary antibodies (Jackson Immunoresearch, West Grove, PA, USA). Imaging was performed using an UltraVIEW VoX spinning disk confocal microscope (Perkin Elmer, Waltham, MA, USA).
Indirect immunofluorescence staining was performed on frozen sections from 3 dpf zebrafish embryos as described previously (56). Mouse monoclonal antidystrophin primary antibody (1 : 500, D8043, Sigma) was used in the work, and revealed by FITC antimouse secondary antibody at 1 : 100 dilution (Jackson Immunoresearch). Nuclei were stained with DAPI and imaged using a Nikon Eclipse 90i microscope. Whole-mount direct immunofluorescence was performed with Alexa Fluor® 546 phalloidin (1 : 100, A22283, Invitrogen) on 3 dpf embryos as described previously (56). Embryos were mounted in 70% glycerol and visualized using an UltraVIEW VoX spinning disk confocal microscope (Perkin Elmer).
Zebrafish embryos were injected at the 1-cell stage with a plasmid encoding the GCaMP3–EGFP calcium reporter under control of an α-actin promoter (46) independently or in combination with ATG MO, resulting in WT or bin1 morphant embryos that mosaically expressed GCaMP3 within skeletal muscle. Before imaging at 3 dpf, tricaine was removed and embryos were allowed to recover in Evans solution (134 mm NaCl, 2.9 mm KCl, 2.1 mm CaCl2, 1.2 mm MgCl2, 10 mm glucose, 10 mm HEPES) for 10 min and then treated with 200 mmN-benzyl-p-toluene sulphonamide to inhibit contraction for 5 min. KCl, 100 μm, was added to initiate muscle contraction. GCaMP3–EGFP-expressing myofibers were imaged at 30 Hz using an UltraVIEW VoX spinning disk confocal microscope (Perkin Elmer). Perkin Elmer Volocity 6.3 confocal software was used to measure relative fluorescence intensity changes of induced Ca2+ transients in individual myofibers expressing GCaMP3–EGFP. Measurements were performed in four replicate WT and bin1 morphant fish and a minimum of three myofibers were chosen for analysis in each fish based on having similar levels of baseline EGFP expression (364.1 ± 16.4 RFU in WT controls and 346.1 ± 25.4 RFU in bin1 morphants).
For H&E sections, 4 dpf zebrafish embryos were anesthetized and fixed overnight at 4°C in 4% paraformaldehyde (PFA) in PBS, washed in PBS, dehydrated in alcohols and xylenes, and embedded in paraffin. Microtome sections were cut at 5 µm using a Leica 2255 cryostat (Leica Microsystems Inc., Wetzlar, Germany) and H&E was done per standard protocol. For electron microscopy, 4 dpf zebrafish embryos were fixed in formaldehyde–glutaraldehyde–picric acid in cacodylate buffer overnight at 4°C, followed by osmication and uranyl acetate staining. Subsequently, embryos were dehydrated in a series of ethanol washes and embedded in TAAB Epon (Marivac Ltd., Halifax, NS, Canada). Sections (95 nm) were cut with a Leica UltraCut microtome, picked up on 100 μm Formvar-coated copper grids, and stained with 0.2% lead citrate. Sections were viewed and imaged under a Philips Tecnai BioTwin Spirit electron microscope (Philips, Amsterdam, Netherlands) at the Harvard Medical School Electron Microscopy Core.
Whole-mount in situ hybridization
Riboprobes were constructed from the 3′ UTR of bin1 using adult zebrafish RNA. Total RNA was extracted from adult zebrafish muscle tissue using RNeasy mini kits (Qiagen). cDNAs were synthesized using the Superscript RT-PCR system (Invitrogen) and PCR-amplified with product ends carrying T7 or SP6 RNA polymerase recognition sequences. Sense or antisense digoxigenin-labeled riboprobes were synthesized by in vitro transcription using dig-labeling kits (Roche Applied Sciences). Whole-mount in situ hybridization was performed as described (57). Images were taken using a Nikon SMZ1500 stereoscope with a SPOT RT3 digital camera system.
Acetylcholine receptor labeling
Embryos were fixed in 4% PFA for 2 h at RT and dehydrated in 100% methanol overnight at −20°C. Embryos were rehydrated and permeabilized in 1× TBST (1% Triton) for 2 h at RT. Blocking was performed with 5% goat serum in TBST for 1 h at RT followed by incubation with fluorescent conjugated α-bungarotoxin (5 mg/ml) for 30 min at RT (Invitrogen). Embryos were washed several times in TBST and mounted in 70% glycerol and visualized using an UltraVIEW VoX spinning disk confocal microscope (Perkin Elmer).
Quantification and statistical analysis
Data were statistically analyzed by parametric Student's t-test (two-tailed) and were considered significant when P < 0.01. All data analyses were performed using the GraphPad Prism 6 software (GraphPad Software Inc., La Jolla, CA, USA).
This work was supported by the Muscular Dystrophy Association USA (MDA201302 to A.H.B.); the National Institute of Arthritis and Musculoskeletal and Skin Diseases (R01 AR044345 to A.H.B., K01 AR062601 to V.A.G.); the National Institute of Neurological Disorders and Stroke (F31 NS081928 to L.L.S.); A Foundation Building Strength; Cure CMD; and the AUism Charitable Foundation. Confocal microscopy and DNA sequencing were done through the Boston Children's Hospital IDDRC Cellular Imaging and Molecular Genetics Cores, respectively, supported by NIH P30 HD18655.
The authors thank Chris Lawrence and Jason Best from the Aquatics Resources Program at Boston Children's Hospital for their expertise in zebrafish husbandry. The authors also gratefully acknowledge Eric Horstick for his kind teaching expertise in in vivo calcium imaging, and Louise Trakimas from the Electron Microscopy Core at Harvard Medical School for her assistance with histology work. Microscopy imaging was performed at the Boston Children's Hospital IDDRC Cellular Imaging Core, with excellent help from Anthony Hill.
Conflict of Interest statement. None declared.