Atoh8, a bHLH Transcription Factor, Is Required for the Development of Retina and Skeletal Muscle in Zebrafish

Background Math6/atoh8, a bHLH transcription factor, is thought to be indispensable for early embryonic development and likely has important roles in vertebrate tissue-specific differentiation. However, the function of Atoh8 during early development is not clear because homozygous knockout causes embryonic lethality in mice. We have examined the effects of the atoh8 gene on the differentiation of retina and skeletal muscle during early development in zebrafish. Results We isolated a Math6 homologue in zebrafish, designated as zebrafish atoh8. Whole -mount in situ hybridization analysis showed that zebrafish atoh8 is dynamically expressed mainly in developing retina and skeletal muscle. Atoh8-MO knock-down resulted in reduced eye size with disorganization of retinal lamination. The reduction of atoh8 function also affected the arrangement of paraxial cells and differentiated muscle fibers during somite morphogenesis. Conclusion Our results show that Atoh8 is an important regulator for the development of both the retina and skeletal muscles necessary for neural retinal cell and myogenic differentiation during zebrafish embryogenesis.


Introduction
Basic helix-loop-helix (bHLH) genes control both the initial neuronal fate determination and subsequent neuronal differentiation/maturation during neurogenesis [1]. Previous studies have suggested that several neurogenic bHLH genes such as Mash1, Math3 and Neurogenin facilitate neuronal fate determination, rather than the glial cell fate [2,3]. In contrast, bHLH genes Hes1, Hes5 and Hesr2 functionally antagonize neurogenic bHLH genes and promote gliogenesis [4]. Misexpression of bHLH genes in murine cerebral cortex was shown to affect cell fate choices and neuronal survival [5]. In mouse retina, glial cell fate specification is modulated positively by the bHLH gene Hes5 [4,6]. bHLH genes such as NeuroD and Math2 (Nex1) are expressed in differentiating neurons after neuronal fate determination. Mutations of these genes in neurons cause defects in terminal differentiation/maturation [3].
Atoh8, a novel member of the bHLH gene family, has an obvious structural similarity to the Drosophila proneural gene atonal [7]. The human Atoh8 contains 321 amino acids with one bHLH domain. Atoh8 might be a transcriptional activator and a permissive factor for terminal podocyte differentiation [8]. It was implicated in differentiation of neuronal cell lineages in the brain by the observation that overexpression of atoh8 in the developing retina induced neurogenesis while inhibiting gliogenesis [9]. Recently, it has been found that Atoh8 may also be involved in the development of kidney [8], liver [10] and pancreas [11].
In the present study, we cloned and characterized the zebrafish atoh8 gene. In addition, we investigated the distribution and time course of the expression of zebrafish atoh8 with special attention paid to the role of this gene in the development of the retina and somites of embryonic zebrafish.

Zebrafish atoh8 is highly expressed in retina and skeletal muscle
To gain insights into the role of the zebrafish atoh8 gene, we analyzed its spatiotemporal expression pattern during embryo development using whole-mount in situ hybridization ( Figure 2). The results showed that zygotic atoh8 mRNA is expressed at as early as epiboly stage, and is broadly distributed till early segmentation stage (Figure 2A-B). Later, the signals are expressed dynamically and mainly in developing retina and skeletal muscle (Figure 2A-M).
In the eye, atoh8 expression was detected in the optic primordium at the 6-somite stage when the optic vesicle is first distinguished ( Figure 2B). Atoh8 was steadily expressed during the invagination of the optic vesicle from its initiation to its expansion to the lens rudiment ( Figure 2C-F, 2L-M).
In skeletal muscle, a faint atoh8 signal could be detected in muscle precursor cells at the early segmentation period ( Figure 2B-F), and staining of ventral somites was visible from the 21-somite to Figure 1. Alignments of bovine, human, rat, monkey and zebrafish Atoh8. Amino acids are shaded according to the degree of conservation: black (100% similarity), gray (80% similarity) and light gray (60% similarity). The amino acid sequences of Bos taurus (Bovin), Homo sapiens (Human), Rattus norvegicus (Rat), Macaca mulatta (Monkey) and Danio rerio (Zebrafish) were obtained from the GenBank database with Accession Nos. XP_873292, NP_116216, NP001102711, XP_001084825 and EU272033, respectively. During this period, the distribution of atoh8 extended to the dorsal somites ( Figure 2K). In addition, the atoh8 signal also appeared in the brain and neural crest cell ( Figure 2G-J).

Knock-down of Atoh8 influenced embryonic development
To further investigate atoh8 function in the development of the retina of zebrafish, two morpholinos (MOs) were designed., Atoh8-MO1 (MO1) and Atoh8splI1E2 (MO2), targeting the translation starting point and the splicing between intron 1 and exon 2 region of atoh8, respectively. Quantitative RT-PCR showed that MO2 significantly reduced (p,0.05) levels of matured atoh8 mRNA by approximately 50% (Figure 3). Both MOs dose dependently caused developmental deficiency in zebrasifh embryos (Figure 4, 5 and 6A) shown by two phenotypes, mild type and severe type. The severe phenotype displayed a shrunken head, reduced eye size and fused somites ( Figure 4A, C, F and Figure 5A, D). Most embryos with this phenotype died within 3 days. On the other hand, embryos with the mild phenotype displayed an abnormal brain shape with smaller eyes, a curved body axis with indistinct somite boundaries and frequently reduced or no circulation ( Figure 4D,G,I,K,M,O and Figure 5B, E, G). In addition, the yolk extension was hardly visible starting at the beginning of tail straightening from the 17-somite stage ( Figure 4C, D, F, G, I and Figure 5A, B, D, E, G). The tail also appeared to be less flexible than that of the controls. In addition, the branchial arches were greatly reduced ( Figure 4O). Knock-down of Atoh8 in mylz2-GFP transgenic zebrafish expressing skeletal-muscle-specific GFP provide more evidence for the effect ( Figure 4K-P). The translational blocking morpholino (MO1) caused 51.562.9% and 70.960.75% severe as well as 26.661.2% and 14.861.3% mild deficient phenotypes at the concentrations of 4 ng/embryo and 8 ng/embryo, respectively. The splicing blocking morpholino (MO2) caused 22.561.1% and 52.661.2% severe as well as 23.761.5% and 18.761.4% mild deficient phenotypes at the concentration of 4 ng/embryo and 8 ng/embryo, respectively. It is interesting to note that treatment with the two MOs caused similar percentages of mild deficiency phenotype but the translation blocking morpholino (MO1) resulted in much higher rate of severe type than the splicing blocker (MO2) (p,0.0001 for 4 ng groups and p,0.002 for 8 ng groups, Figure 6A). While it is possible that MO1 may have higher non-specific toxicity than MO2, another explanation is that some atoh8 transcripts are maternally supplied. To investigate this possibility, RT-PCR was conducted on total RNAs collected from unfertilized eggs and embryos of 64-cell stage as well as oblong-sphere stage. As shown in Figure 7, maternally supplied atoh8 mRNA was evidently present in both unfertilized oocytes and 64-cell stage but not the oblong-spheres, indicating preexisting of atoh8 mRNA that was not affected by the spicing blocker MO2 but was prevented from protein synthesis by the translation blocker MO1.
To rule out the possibility that the effect of Atoh8-MO was mediated through a general p53-induced apoptosis [12][13], all Atoh8-MOs were co-injected with a p53-MO (at a concentration ratio of 1:1.5). Furthermore, there were no differences in atoh8 morphants with or without p53-MO demonstrated by a separate experiment (data not shown). Thus, the abnormality in zebrafish development caused by Atoh8 knock-down is probably independent of p53 activity.

Atoh8 is required for retinal lamination
To clarify the role of Atoh8 during retinal development, we examined a series of tissue cross-sections of the retinas of Atoh8 morphants that were treated with MO1 and survived for 3dpf and 7dpf. As shown in Figure 8, neuronal stratification had not appeared in the Atoh8 knock-down retina by 3dpf when retinal lamination in the control animals was already well developed ( Figure 8A-B). By 7 dpf ( Figure 8C-D), although lamination had formed in retinas of the treated embryos, the neural retina was significantly underdeveloped. The number of retinal ganglion cells was reduced by approximately 50% in atoh8 knock-down animals ( Figure 8E, 69.8617.0 vs.149621.1, n = 8, p,0.001). Moreover, the photoreceptor cells of the morphants had a less elongated appearance compared with the control. In addition, the portion of the photoreceptor layer ventral to the optic nerve seemed immature with a group of cells appearing from the inner nuclear layer ( Figure 8D, arrow).
To determine whether apoptosis played a role in the malformation of the eye caused by Atoh8 knock-down, TUNEL staining was performed in 48 hpf embryos with or without MO treatment. There was a substantial number of apoptotic cells in the retina of morphants ( Figure 9). More apoptotic cells were seen in severely mal-developed eyes than in those less affected by the Atoh8 knock-down.
Delayed retinal lamination in Atoh8 knock-down fish suggests the involvement of the gene in the fundamental mechanism of retinal development. Thus, we speculated that the levels of atoh8 may regulate the expression of major eye development markers such as Pax6 and atoh5. The Pax6 gene is expressed throughout the development of vertebrate eyes and controls many aspects of ocular formation [14][15] while the atoh5 gene is an earliest marker of neuronal production [16]. Transcriptional levels of both genes were measured in Atoh8 morphants. Both Pax6 and atoh5 were distributed in much smaller areas with a lower intensity of their signals in the Atoh8 knock-down embryos ( Figure 10). Expression of pax6a in the retinal was remarkably reduced (arrow) in this MOtreated embryo ( Figure 10A-D). Atoh5 expression was present throughout most of the neural retinal at 36 hpf in the control embryo ( Figure 10E) while present only in dorsal and temporal retina in the MO-treated embryos ( Figure 10F). By 48 hpf, expression of atoh5 was localized in all retina ganglion cells in the control ( Figure 10G) but present only in ventral retina in the MOtreated embryo ( Figure 10H). In addition, the signal of the two genes were much weaker than those of the control, indicating that normal level of atoh8 expression is required for adequate levels of both Pax6 and atoh5 in the developing retina. Atoh8 is important for myogenic differentiation of skeletal muscle To explore the function of atoh8 in the development of muscle, we examined the expression of several skeletal muscle markers in Atoh8 knock-down embryos using whole-mount in situ hybridization. We found that the expression of myogenic myoD in paraxial mesoderm had a tendency toward a ventral localization ( Figure 11A, G). By 36 hpf, the boundaries of the posterior somites were indistinct, even in the mildly abnormal embryos ( Figure 11). Embryos with a severe deficiency showed fewer posterior somites in myogenin at the 17-21 somite stage than did the controls (Figure 11 I,K). The transverse myoseptum of Atoh8 MO-treated embryos was ill-defined and incomplete. Thus, the somites formed a U or even a cuboidal shape instead of the chevron shape of the control (Figure 12A-D). The malformation of somites in Atoh8 knock-down mylz2-GFP embryos was even more evident ( Figure 4K-N). In addition, an examination of cross-sections of the embryos demonstrated that differentiated muscle fibers in myotomes were less compact and had a disordered arrangement ( Figure 12A-D). Furthermore, the markedly reduced GFP signal in the knock-down embryos may also reflect a deficiency in the related structures including the skeletal muscle of somites and branchial arches ( Figure 4K-P). Interestingly, unlike the retina, there was no detectable apoptosis in the muscle tissue of atoh8 knock-down fish (data not shown). Taken together, these data support the idea that atoh8 plays an important role in the organization of skeletal muscle during zebrafish development.

Discussion
In this study, we isolated a zebrafish bHLH gene designated as atoh8, that is homologous to the mammalian Math6 gene and the Drosophila proneural gene atonal [7]. We have shown that atoh8 is   expressed in retinas and somites with a pattern closely associated with neuronal differentiation and myoseptum formation. Severe malformation of the somites and eyes were seen in Atoh8 knockdown embryos, indicated by both abnormal morphology and reduced marker gene expression. In addition, reduced Atoh8 function resulted in an alteration of retinal structures with a significantly reduced number of ganglion neurons.
In the present study, MO knockdown of atoh8 expression resulted in developmental deficiency. Both translation and splice blocking morpholinos gave rise to the same two phenotypes i.e. mild and severe abnormality. Since severe phenotype can be largely eliminated by overexpressing its mRNA, one cannot rule out that the mild phenotype may be caused by non-specific effect of the morpholinos. However, it is equally possible that the smaller difference in rates of mild abnormality between mRNA rescued and control groups was caused by part of severe phenotype becoming mild ones and the similar portion of mild type becoming normal type. More interestingly, there was a higher rate of severe phenotype by translation blocking morpholino than the splicing blocker. indicating the importance of maternally supplied atoh8 transcript and fundamental functions of this gene in early embryonic development.
Atoh8 is a member of the bHLH family that has been isolated from Drosophila, mosquitos, mice, rats and humans. There is a substantial similarity between the zebrafish and these other species in terms of the temporal and spatial expression pattern of atoh8 during embryogenesis [9], suggesting that the gene serves a similar function in embryonic development in all these species.
Many members of the bHLH gene family, such as Hes1, Hes5, Mash1 and Math5, are involved in Notch-Delta, Wnt and Nodal signal pathways which are known to be vital during retinogenesis [17], a process that involves the differentiation of different types of retinal cells in a fixed temporal sequence. Retinal ganglion cells (RGC) are the first to differentiate followed by bipolar cells and photoreceptors. Lastly, Müller glial cells are generated [18]. Particularly during the generation of RGC and phororeceptors, this process is regulated by transcription factors of the bHLH family. Inoue et al. demonstrated that Mis-expression of Math6 with retrovirus in the developing mouse retina induced neurogenesis, while inhibiting gliogenesis [9]. This is in agreement with our current findings that knockdown Atoh8 expression caused reduced number of ganglion cells in the retina. Further, substantial apoptotic cells were also seen in the retina, suggesting that the Atoh8 is also important in survival of neuronal cells. It was reported by Jean-Marc Matter et al. that cells with upregulated levels of ath5 expression initiate transcription of early RGC-specific traits, exit the cell cycle and express Neuro M as well as other postmitotic RGC-specific genes. Moreover, it was also reported that photoreceptor genesis involves two transcriptional pathways: ngn2RneuroDRraxL and ath5RneuroDRraxL [19]. Among these genes, ngn2 and ath5 function in progenitors and are responsible for cell type differentiation in vertebrate retina. NeuroD plays a central role in photoreceptor development. In the present study, we found that the expression of ath5 was significantly reduced in the retinas of Atoh8 knock-down zebrafish, suggesting that Atoh8 might be the upstream regulator of ath5 in the retina.
Atoh8 knock-down also resulted in distinct muscle abnormality. The malformations started at a very early developmental stage and most of the abnormal embryos shared the same phenotype. More importantly, we are the first to show that down regulation of atoh8 expression resulted in changes of the master myogenic transcription factor myoD in early myogenesis of zebrafish, indicated that atoh8 might play a crucial role in zebrafish during the early somite formation. Coutelle et al. found that Hh signaling was required for the expression of myoD and myf5 as well as for differentiation in zebrafish adaxial myogenesis [20]. It remains to be investigated whether atoh8 participates in the Hh signaling pathway during development somite and the detail function of atoh8 in development of somite is also not clear.
It was suggested that atoh8(Math6)-null mice might be embryonically lethal [11] because of the absence of homozygous embryos. This is in consistent with the high mortality and substantial retinal cell apoptosis seen in zebrafish embryos treated with Atoh8-MO. Given the high level of expression and wide distribution of atoh8 during early embryonic zebrafish development, it would not be surprising if this gene is involved in regulation of the expression of many downstream genes and also has an effect on the formation of many other organs besides retina and somites. For instance, evidence of heart defects, including altered jogging, abnormal expansion and pericardial edema, was seen in about half of the Atoh8 knock-down fish (data not shown).  To our knowledge, the present study is the first to demonstrate roles of endogenous atoh8 in early embryonic development in vertebrates. It is also the first time that atoh8 is found to regulate somite development. Our results suggest that atoh8 may be involved in many aspects of early embryonic development and that zebrafish is a good model system for studying the function of atoh8 in signal pathways governing cell differentiation in various organs.

Zebrafish maintenance
Zebrafish (AB strain) were maintained at 28.5uC on a 14-hour light/10-hour dark cycle as described by Westerfield [21]. Embryos were staged according to their morphological features and hours (h) or days (d) post-fertilization as described by Kimmel et al [22]. Embryos used in whole-mount in situ hybridization were cultured in Holtfreter solution with 0.003% phenylthiourea (PTU) (Sigma).

Cloning and analysis of the zebrafish atoh8 gene
To clone the atoh8 cDNA, total RNA was isolated from zebrafish embryos (36 hpf) using a Qiagen miniprep procedure. PCR amplification was carried out with primers 59-CGGGATCC-TGGACTTATTAGTCAGGCTGGA-39 (forward) and 59-GGAA-TTCGAGGTCCTTCTTGCTCCAGAT-39 (reverse) using the above cDNA as the template. The amplified product was cloned into pGEM-T Easy Vector (Promega) and sequenced. The fulllength cDNA of zebrafish atoh8 was submitted to GenBank under accession number EU272033. To determine online the predicted amino acid sequences of atoh8 and conserved domains, we used National Center for Biotechnology Information (NCBI) Blast Server 2.0 to create multiple alignments and ClustalW version 1.82 (http://www.ebi.ac.uk/clustalw/) to construct the phylogenetic trees.

Whole-mount in situ hybridization
To study the expression pattern of the atoh8 gene during embryogenesis, the embryos were fixed in phosphate buffered saline with 4% paraformaldehyde (PFA) and were then processed for whole-mount in situ hybridization as described by Westerfield [21]. The zebrafish atoh8 cDNA was cloned into pGEM-T easy vector, which contained T7 and SP6 promoters. The plasmid DNA was linearized with ApaI, and transcribed in vitro by T7 and SP6 RNA polymerases for the antisense and sense atoh8 RNA probe. This was labeled with digoxigenin (DIG) (Roche) and pGEM-T-atoh8 was used as the template. Other antisense RNA probes used in this study included myoD [23], myf5, myogenin (from P.J. Zhang), Pax6a and ath5 (provided by D. L. Stenkamp and R. Frey).
For rescue experiment, the coding sequences of zebrafish atoh8 were cloned into the expression vector pcDNA3.0. The capped mRNAs were synthesized in vitro using EcoR I linearized plasmid DNAs and the T7 promoter. When coinjection of mRNA with the MO was performed, the embryos was first injected with the mRNA and then with the morpholino at a dose described in the text.
All images were captured using an IP-20 (OLYMPUS) under an IX-12 (OLYMPUS).

Real-time quantitative PCR
The total RNA of control embryos and MO-treated embryos were extracted using Trizol(Invitrogen) method. For each sample the same quantity of total RNA (1 mg) was reverse transcribed into cDNA using the SMARTer TM PCR cDNA Synthesis Kit (Clontech). The real-time PCR was performed using SYBRH Green (Biotium) and 2 ml of cDNA template with RG-3000 realtime PCR apparatus (Corbett). Each individual reaction was performed in triplicate. A set of primer sequences for target genes were designed using Primer Primer5.0 and the sequences were sense: 59 CAGGCAAGTCCTCAAGTGTC 39, antisense: 59 TGCGGTGGTCTGGTGTAT 39, respectively. The PCR product was 379bp(373,751 in atoh8 mRNA). A set of control primers (b-actin) were used to normalize the abundance of cDNA in each reaction. The condition for Real-Time PCR was 95uC 5 min, then 95uC 10 sec, 60uC 15 sec, 72uC 30 sec for 40 Cycling, Melt: from 72uC to 99uC. For each comparison we calculated a 95% confidence interval about mean fold change, based on the expression level estimates across the 5 experimental replicates.

TUNEL assay
Embryos were fixed as for in situ hybridization and keep in methanol. After rehydration, embryos were sectioned and then permeabilized by proteinase K digestion, re-fixed in 4% PFA and washed in TBS. Apoptosis was detected in embryos by terminal transferase dUTP nick-end labeling (TUNEL), as described in the manufacturer's instructions (TdT-FragEL TM DNA Fragmentation Detection Kit: DAB, Merck).