The zebrafish (Danio rerio) possesses two mechanosensory organs believed to be homologous to each other: the inner ear, which is responsible for the senses of audition and equilibrium, and the lateral line organ, which is involved in the detection of water movements. Eight zebrafish circler or auditory/vestibular mutants appear to have defects specific to sensory hair cell function. The circler genes may therefore encode components of the mechanotransduction apparatus and/or be the orthologous counterparts of the genes underlying human hereditary deafness. In this report, we show that the phenotype of the circler mutant, mariner, is due to mutations in the gene encoding Myosin VIIA, an unconventional myosin which is expressed in sensory hair cells and is responsible for various types of hearing disorder in humans, namely Usher 1B syndrome, DFNB2 and DFNA11. Our analysis of the fine structure of hair bundles in the mariner mutants suggests that a missense mutation within the C-terminal FERM domain of the tail of Myosin VIIA has the potential to dissociate the two different functions of the protein in hair bundle integrity and apical endocytosis. Notably, mariner sensory hair cells display morphological and functional defects that are similar to those present in mouse shaker-1 hair cells which are defective in Myosin VIIA. Thus, this study demonstrates the striking conservation of the function of Myosin VIIA throughout vertebrate evolution and establishes mariner as the first fish model for human hereditary deafness.
Received 31 May 2000; Revised and Accepted 11 July 2000.
During the last few years human geneticists have made remarkable progress in cloning genes which are responsible for both syndromic and non-syndromic forms of deafness (1). Expression patterns of these deafness genes in rodents have begun to hint at the roles of various cell types within the cochlea. Six of fourteen genes responsible for isolated deafness in human and two additional genes responsible for isolated deafness in mouse are expressed specifically in sensory hair cells. Despite the existence of a collection of inbred mouse strains with defects in hearing, many of the human hearing disorders still do not have a corresponding animal model. Recently a large-scale mutagenesis screen in zebrafish identified eight genes that are essential for balance and hearing (2,3). The zebrafish circler mutants swim in vertical loops or corkscrew paths and in some cases do not respond to acoustic/vibrational stimuli. Analysis of the defects in the zebrafish circler mutants suggests that the proteins encoded by the circler genes play specific roles in sensory hair cell function. Therefore, these zebrafish mutants may provide some of the animal models necessary for the study of the pathogenesis caused by the lesions in deafness genes.
myosin VIIA is expressed in both vestibular and auditory hair cells and mutations in myosin VIIA cause balance and/or hearing impairment in patients suffering from Usher 1B syndrome and autosomal dominant and recessive forms of non-syndromic deafness (4–7). Myosin VIIA is also defective in mouse shaker-1 mutants (8). Studies of the shaker-1 phenotype indicate that stereocilia are disorganized and that shaker-1 hair cells do not endocytose ototoxic antibiotics and lack an electrophysiological response (9,10). In mice, however, sensory hair cells are located deep within the petrous temporal bone and are not accessible to direct experimental manipulation nor simple observation. We therefore considered the zebrafish as an alternative model to study the molecular basis of development and function of sensory hair cells. In zebrafish, sensory hair cells are clearly visible within the inner ear during development (11) which occurs ex utero, and the sensory hair cells of the lateral line organ, believed to have an ancestral origin common with that of the inner ear (12), are located at the surface of the skin (13). Anatomical studies indicate that the structure of the vestibular inner ear is highly conserved between teleost fish and higher animals, and fish sensory hair cells are morphologically and physiologically similar to hair cells in higher vertebrates (14–16).
Among the circler mutants, mariner displays inner ear hair cell bundle defects, lack of acoustic vibrational sensivity and reduced or abolished microphonic potentials (3). In addition, recent data demonstrated that calmodulin-dependent apical endocytosis is defective in mariner (17). Due to the phenotypical resemblance between the shaker-1 mouse and mariner, we hypothesized that myosin VIIA could be the gene deficient in mariner. Here, we show that zebrafish Myosin VIIA, which is highly homologous to the mammalian protein, is expressed in sensory hair cells of the inner ear and the lateral line. We observed, by mapping, the co-localization of myosin VIIA with the mariner mutations, and identified myosin VIIA point mutations in the five alleles of mariner. We also investigated the phenotype–genotype correlation in these mutants. The results establish mariner as the first fish model for human hereditary deafness, thus providing the possibility of in vivo study of the role of Myosin VIIA in the sensory hair cells.
Cloning of zebrafish myosin VIIA cDNA
We hypothesized that a number of the zebrafish circler genes may be the same as those responsible for human hearing disorders. Since the mariner mutants were reported to have a bundle defect (3), we focused on the candidate gene, myosin VIIA. Accordingly, we cloned and characterized the zebrafish homolog of myosin VIIA. The complete zebrafish myosin VIIA cDNA is 7590 nucleotides long: it contains a 161 nucleotide 5′ untranslated fragment, 6540 nucleotides of coding sequence and an 889 nucleotide 3′ untranslated region. Homology with human and mouse nucleotide sequences is 74% overall (77% in the translated region). The predicted zebrafish Myosin VIIA primary structure contains 2179 amino acids. Comparison between the human and mouse proteins reveals a very high conservation between species with an overall identity of 85%. Strikingly, all the domains that have been described for mammalian Myosin VIIA are highly conserved in the zebrafish protein (Fig. 1).
Analysis of the expression of myosin VIIA mRNA in the zebrafish larvae
To confirm that we had cloned the corresponding ortholog, myosin VIIA mRNA expression was studied by whole-mount in situ hybridization in zebrafish larvae. A strong signal was detected in the otic vesicle of the 24 h post-fertilization (h.p.f.) embryo (Fig. 2A and B) in two regions located at each opposite end of the vesicle; the labeling was associated with the tether cells which anchor the forming otoliths and are the first sensory cells to emerge (18). Myosin VIIA mRNA expression remained strong later on and expanded as sensory patches grew within the developing inner ear of the zebrafish from stage 24 h.p.f. to 5 days post-fertilization (d.p.f.) (Fig. 2C, E and G). At 48 h.p.f., expression was detectable in the neuromasts of the anterior and posterior lateral lines (Fig. 2C and D) and persisted until later stages (Fig. 2E–G). These results provided further evidence that we had cloned the correct homolog which is expressed in hair cells.
myosin VIIA co-localizes with mariner
We then mapped the zebrafish homolog of myosin VIIA, using radiation hybrid cell lines (19). Linkage was established to linkage group 18 next to an expressed sequence tag (EST) encoding a swelling-dependent chloride channel, icln (Fig. 3a). Synteny is very high in this region between the mouse and the corresponding human chromosome along a 28 cM stretch containing the neighboring genes of icln and myosin VIIA. Using simple sequence length polymorphic (SSLP) markers, we meiotically mapped marinerrtc320b, establishing a close linkage to the markers z13329 and z11944 within the syntenic region of the linkage group, 18 (Fig. 3a).
Identification of mutations in myosin VIIA in the five alleles of mariner
As suggested by the mapping data and the expression pattern of myosin VIIA mRNA in the zebrafish inner ear and lateral line organ, myosin VIIA was an excellent candidate for the mariner gene. The five alleles of mariner were therefore screened for mutations in myosin VIIA by sequencing of the cDNA obtained by RT–PCR on homozygous mutant larval mRNA. Three nonsense and two missense mutations were identified (Fig. 3b). The nonsense mutations are as follows: (i) K533X (marinerty220d) is located in the head region of Myosin VIIA before the actin-binding domain; (ii) Y846X (marinertc320b) is located within the fifth IQ motif; and (iii) C2144X (marinertn4503) is located at the end of the tail, 3′ to the second Band 4.1/Ezrin/Radaxin/Moesin protein (FERM) domain. The missense mutations are as follows: (i) F516S (marinertn3540) is located in the head region before the actin-binding domain; and (ii) G2073D (marinertr202b) is located within the second FERM domain, at its C-terminal end.
Hair bundle phenotype of the mariner mutants
Homozygous mutant mariner larvae are characterized by an inner ear hair cell bundle defect, the lack of acoustic vibrational sensivity and reduced or absent extracellular hair cell potentials (3). Analysis of the fine structure of mariner hair bundles revealed pronounced splaying of the stereocilia (Fig. 4). Phalloidin-labeled wild-type hair bundles were invariably intact and conical in shape (Fig. 4a). Hair bundles in homozygous larvae (5 d.p.f.) carrying the nonsense mutations showed the greatest degree of splaying (Fig. 4b–d). In contrast, marinertr202b larvae had relatively few splayed bundles (Fig. 4f) and the splaying defect in mutant marinertn3540 larvae was also less severe (Fig. 4e). In addition, we examined stereociliary bundles using transmission electron microscopy. A representative example of sections of marinertc320b and marinertr202b hair bundles is shown in Figure 4 (right panels). Large gaps between the kino- and stereocilia were observed in marinertc320b specimens (Fig. 4h). Obviously splayed bundles were infrequently observed in marinertr202b larvae carrying the missense mutation in the second FERM domain (Fig. 4i). The hair bundles also appeared to be more intact in marinertc320b larvae fixed at an earlier stage (3.5 d.p.f.), suggesting that the splaying defect is progressive (data not shown).
Zebrafish Myosin VIIA is highly homologous to the human and mouse proteins. As shown in Figure 1, all interacting and signaling domains were conserved during evolution suggesting their importance for the function of Myosin VIIA in sensory hair cells. Recessive point mutations within these conserved regions account for the auditory/vestibular defects in the zebrafish circler mutant, mariner. Sequencing of the cDNA from homozygous larvae revealed that three alleles, marinertn4503, marinerty220d and marinertc320b, contained nonsense mutations within the head, neck or tail domains of the protein, and two alleles, marinertn3540 and marinertr202b, contained missense mutations within the head or tail domains. The two missense substitutions in marinertn3540 and marinertr202b affect amino acids that were highly conserved during evolution since these amino acids are present in all identified myosin VIIA sequences (Fig. 1) and are thus likely to be essential for the function of the protein. The substitution of a hydrophobic for a polar residue (F516S, marinertn3540) may affect the structure of the head of Myosin VIIA and lead to a less efficient association at the nearby actin-binding site. The replacement of a neutral with an acidic residue (G2073D, marinertr202b) at the C-terminal end of the second FERM domain may hinder the binding of Myosin VIIA to its ligands. In humans, 54 mutations in the myosin VIIA gene have been reported to be responsible for syndromic and non-syndromic forms of deafness: (i) 49 mutations in patients affected by Usher 1B syndrome were found in both conserved and non-conserved domains (4,20–25); (ii) four mutations including two substitutions within the motor domain were present in patients affected by the isolated autosomal recessive form of deafness, DFNB2 (5,6); and (iii) a small deletion within the coiled-coil domain was found in affected members of a family carrying the dominant form of isolated hearing loss, DFNA11 (7). In mice, seven mutations affecting either conserved domains or leading to null mutations in myosin VIIA are also responsible for the shaker-1 phenotype (8). Interestingly, of the five mutations reported here, C2144X (marinertn4503) is the zebrafish equivalent to the shaker-1 mutation, Myo7a3336SB (8). This nonsense mutation, located at the C-terminal end of the protein, leads to the deletion of the last 35 amino acids which have been shown to impair the stability of the protein in the mouse mutant (26).
The strong alleles of mariner which may give rise to truncated proteins or may represent null mutations cause severe splaying of sensory hair cell bundles and, as shown previously, affect mechanotransduction(3). In contrast, the presence of relatively intact bundles in marinertr202b larvae suggests that either the second FERM domain is not important for bundle integrity or the mutation does not impair its ability to bind to ligands that are involved in bundle integrity. Extracellular hair cell potentials in marinertr202b larvae are reduced by approximately two-thirds (3), suggesting that the second FERM domain may still play a role in hair bundle mechanics or, alternatively, the subtle defect detected in phalloidin-labeled bundles (less conical in shape than wild-type bundles) may have a profound effect on mechanotransduction. Interestingly, blocking mechanotransduction also inhibits apical endocytosis in zebrafish larval hair cells (17), raising the possibility that a ligand which binds to Myosin VIIA may be involved in both processes. Although larvae carrying the different alleles of mariner can be distinguished from each other based on the severity of splaying of hair bundles, they cannot be distinguished at the behavioral level, i.e. the severity of the balance defect and lack of an acoustic startle reflex appear to be very similar among the five mariner mutants.
Phenotypic analysis of the mariner mutants along with previously published data on the phenotype of the mouse mutant, shaker-1 (9,10,17), shows that recessive mutations within the conserved domains of Myosin VIIA result in similar phenotypes among different species. Shaker-1 mice carrying alleles which lead to motor dysfunction or loss of Myosin VIIA protein also have splayed and disorganized stereociliary bundles, and no electrophysiological response (9). In addition, two studies report that sensitivity to ototoxic antibiotics is absent in both shaker-1 mutants and mariner mutants (10,17). Sensitivity is due to accumulation of antibiotics which presumably are endocytosed at the apical surface of wild-type hair cells. Both animal models suggest that Myosin VIIA plays a crucial role in sensory hair cell bundle integrity and may be important for endocytosis at the apical end of the sensory hair cell. This hypothesis is supported by studies on the subcellular localization of Myosin VIIA which reveals that the protein is concentrated within stereocilia and the pericuticular necklace, a putative vesicle trafficking zone at the apical surface (27). In addition, a role in phagocytosis for the Dictyostelium discoideum homolog of Myosin VIIA was recently described, confirming the notion that Myosin VIIA is important for events requiring membrane internalization (28). Since all mutant mariner larvae are insensitive to ototoxic antibiotics and do not internalize a marker of endocytosis, FM1-43 (17), the missense mutation within the second FERM protein domain in zebrafish marinertr202b larvae suggests that this C-terminal domain may bind to ligands which mediate vesicle trafficking. Alternatively, the missense mutation may reduce the binding efficiency of a single ligand involved in both endocytosis and bundle integrity, but the consequence is less severe for bundle integrity.
Our study provides evidence that the sequence and function of the zebrafish homolog of Myosin VIIA is highly conserved. It is interesting to note that mutations in myosin VIIA affect both inner ear and lateral line hair cells, suggesting that myosin VIIA was essential for hair cell function from an early time point onwards, before these two populations diverged and took on different roles. As a lower vertebrate, the zebrafish offers some advantages with respect to studying hair cell function. In mammals, sensory hair cells of the inner ear are inaccessible during development. In contrast, zebrafish sensory hair cells are either clearly visible or superficial, therefore mariner may be a more useful animal model for the in vivo study of the pathogenesis caused by mutations in myosin VIIA. Moreover, 80% (16/20) of myosin VIIA missense mutations described as being responsible for Usher 1B syndrome (4,20–25) affect amino acids that are conserved in the zebrafish cDNA sequence, thus making the zebrafish an excellent model system for studying the effect of mutations identified in this particular human disease. In addition, large-scale allele screens are feasible in the zebrafish, offering the exciting possibility of genetically analyzing the various ligand binding domains of Myosin VIIA whose functions in sensory hair cells are unknown.
MATERIALS AND METHODS
The following zebrafish strains (Tübingen background) were used for this study: mariner alleles tc320b, tr202b and ty220d (3). In addition, two new alleles, tn4503 and tn3540, were isolated in a small-scale screen for vibration-insensitive larvae. Heterozygous fish were crossed to generate homozygous larvae which were identified based on their behavioral phenotype of defective balance and insensitivity to vibration. Larvae were raised at 30°C in E3 medium (containing 5 mM NaCl, 0.17 mM KCl, 0.33 mM CaCl2 and 0.33 mM MgCl2). Specimens designated as wild-type were the phenotypically wild-type siblings of homozygous mutants.
Whole larvae (day 5) were anesthetized with 0.02% MESAB (3-aminobenzoic acid ethyl ester) and then fixed by immersion in 2.0% glutaraldehyde and 1.0% paraformaldehyde in normal solution (containing 145 mM NaCl, 3 mM KCl, 1.8 mM CaCl2, 10 mM HEPES, pH 7.2) overnight to several days at 4°C. Specimens were fixed with 1.0% OsO4 in H2O for 10 min on ice, followed by fixation and contrast with 1.0% uranyl acetate for 1 h on ice, and then dehydrated with several steps in ethanol and embedded in Epon. Ultrathin sections of mutant marinertc320b specimens (n = 11) and mutant marinertr202b specimens (n = 4) were stained with lead citrate and uranyl acetate.
Hair bundle staining
Whole larvae (day 5) were stained with Oregon green 488 phalloidin (Molecular Probes, Eugene, OR) by fixing as described above and were then permeabilized with 2% Triton X-100 overnight at 4°C. The larvae were rinsed with normal solution and then incubated with 2.5 mg/ml Oregon green 488 phalloidin for >3 h at 4°C. After several rinses, the larvae were mounted onto dishes with fine glass fibers and confocal images were made using a 100× oil lens on a Leica DM IRBE inverted microscope. For all experiments, n ≥ 5 for each allele.
Genetic and radiation hybrid mapping
Fish carrying the tc320b allele were used for mapping by crossing with a reference line, WIK (29). DNA was prepared from homozygous mutant and phenotypically wild-type sibling F2 larvae for meiotic mapping using PCR-able simple sequence (C·An) length polymorphisms (see mapping scheme at (http://www.eb.tuebingen.mpg.de/abt.3/geisler_lab/mapping.html ). Using DNA pooled from 48 homozygous mutants and 48 wild-type siblings, the mariner locus was mapped to linkage group 18 using a panel of 72 SSLP markers. Fine mapping of 48 single larvae was then performed with the following SSLP repeat markers: z1144, z3558, z13329 and z11944. Primers directed against the 3′-untranslated region of zebrafish myosin VIIA (5′-TGGCTTCTATGCATCCCAAAC-3′ and 5′-AAGCAGTTACTCATATCCCCCC-3′, respectively) were used with DNA samples of 94 zebrafish radiation hybrid cell lines (Goodfellow T51 panel) to determine the location of Myosin VIIA (see ref. 19 for a detailed protocol of radiation hybrid mapping).
myosin VIIA cDNA cloning and sequencing
A cDNA library was made with RNA extracted from 24 h.p.f. total zebrafish embryos using a Marathon cDNA amplification kit (Clontech, Palo Alto, CA). An initial fragment coding for zebrafish Myosin VIIA tail was cloned using degenerate oligonucleotides which were chosen after alignment of the human (HSU39226, HSU55208), mouse (MMU81453), Drosophila melanogaster (AC002502) and Caenorhabditis elegans (U80848) sequences available in the databases. The complete cDNA sequence (AJ404001) was cloned by 5′ and 3′ rapid amplification of cDNA ends. Double-stranded DNA was sequenced by the dideoxynucleotide method using the Big Dye Terminator RR kit (Perkin Elmer, Foster City, CA) or the DYEnamic ET terminator cycle sequencing premix kit (Amersham Pharmacia Biotech, Little Chalfont, UK) and an ABI PRISM 373 or 377 DNA sequencer (Perkin Elmer).
Analysis of the mariner mutations
RNA was prepared from homozygous mutant larvae of each allele (tc320b, tr202b, ty220d, tn4503 and tn3540) and from homozygous wild-type larvae (day 5) using an RNeasy kit (Qiagen, Hilden, Germany). RNA was reverse-transcribed using a primer from the 3′-untranslated region of zebrafish myosin VIIA cDNA. Oligonucleotides allowing for the amplification of the entire myosin VIIA cDNA in eight overlapping fragments of ∼1 kb each were chosen on the wild-type sequence. All PCR products were treated with shrimp alkaline phosphatase and exonuclease I, and then directly sequenced as described above. Point mutations were detected by comparison of each mutant cDNA sequence to the wild-type one (GenBank accession no. AJ404002) using the SeqLab program (Genetics Computer Group, Madison, WI). Each identified mutation was confirmed at least once with another independent sequence. To ensure that the missense mutations were not simply polymorphisms, we compared the nucleotide sequence of several wild-type strains at the relevant positions (Tübingen, WIK and GDS).
In situ hybridization
A digoxigenin-tagged antisense RNA probe was made from a 3350 bp fragment of cDNA coding for zebrafish Myosin VIIA tail (clone 110) using the DIG RNA labeling kit (Boehringer Mannheim, Mannheim, Germany). Whole-mount in situ hybridization was performed as described previously (30). pSTBlue-1 subcloning vector (Novagen, Madison, WI) was linearized with SalI and transcription was initiated with T7 RNA polymerase. Embryos were treated with 0.003% 1-phenyl-2-thiourea. Stained embryos were hand-dissected and mounted in phosphate-buffered saline/glycerol (1:1) for observation on a Leica Orthomat E microscope equipped with Nomarski optics. Semi-thin (5 µm) transverse sections of 5 d.p.f. embryos stained by in situ hybridization and embedded in epon/araldite were made.
S.E. wishes to thank members of the Stainier group (University of California, San Francisco, CA) for extensive training on zebrafish research during a sabbatical stay, and Philippe Herbomel (Institut Pasteur, Paris, France) for use of his microscope set-up for the high contrast in situ picture of the 24 h.p.f. embryo. T.N. thanks Ulrike Schönberger for her help with maintaining the fish strains and technical assistance, and the members of the EM facility at the Max-Planck-Institut für Entwicklungsbiologie for their advice and technical assistance. S.E. is supported by a fellowship from the Fondation pour la Recherche Médicale (France). T.N. was supported by a grant from the Sonderforschungsbereich 430.
To whom correspondence should be addressed. Tel: +49 7071 601377; Fax: +49 7071 601384; Email: email@example.com