Abstract
The dorsal axial muscles, or epaxial muscles, are a fundamental structure covering the spinal cord and vertebrae, as well as mobilizing the vertebrate trunk. To date, mechanisms underlying the morphogenetic process shaping the epaxial myotome are largely unknown. To address this, we used the medaka zic1/zic4-enhancer mutant Double anal fin (Da), which exhibits ventralized dorsal trunk structures resulting in impaired epaxial myotome morphology and incomplete coverage over the neural tube. In wild type, dorsal dermomyotome (DM) cells, progenitors of myotomal cells, reduce their proliferative activity after somitogenesis and subsequently form unique large protrusions extending dorsally, potentially guiding the epaxial myotome dorsally. In Da, by contrast, DM cells maintain the high proliferative activity and form mainly small protrusions. By combining RNA- and ChIP-sequencing analyses, we revealed direct targets of Zic1 which are specifically expressed in dorsal somites and involved in various aspects of development, such as cell migration, extracellular matrix organization and cell-cell communication. Among these, we identified wnt11r as a crucial factor regulating both cell proliferation and protrusive activity of DM cells. We propose that the dorsal movement of the epaxial myotome is guided by DM cells and that Zic1 empowers this activity via Wnt11r to achieve the neural tube coverage.
Introduction
Active locomotion, which is powered by skeletal muscles in vertebrates, is critical for animals to survive. Vertebrate skeletal muscles consist of axial muscles (head, trunk and tail muscles) and appendicular muscles (limb muscles). Axial muscles first arose in the chordates to stabilize and enable side-to-side movement of the body axis. In jawed vertebrates, the subdivision of axial muscles into epaxial (dorsal) and hypaxial (ventral) muscles led to an increased range of movement: dorsoventral undulation in fish and lateral movements in terrestrial vertebrates (Glass and Goodrich, 1960; Romer and Parsons, 1986; Fetcho, 1987; Clifton, 2002; Sefton and Kardon, 2019). Among these muscles, epaxial muscles are characterized by their unique anatomical structure which extends dorsally and surrounds the vertebrae. This morphology also ensures mechanical support and protection of the vertebrae and the spinal cord inside. While we have a detailed understanding of how the myotome, precursors of epaxial and hypaxial muscles, differentiates from the somites (Kalcheim and C, 1999; Gros, Scaal and Marcelle, 2004; Hollway and Currie, 2005), we only begin to understand the cellular and molecular mechanisms of the subsequent morphogenetic processes generating the epaxial muscles. Previous studies in rats and mice suggested that myocytes of the epaxial myotome do not actively migrate dorsally but are guided by external forces (Deries, Schweitzer and Duxson, 2010; Deries et al., 2012). However, what exerts such forces to drive extension of epaxial myotome is still unclear.
Fish have been extensively utilized to study myotome development thanks to the transparency of their bodies throughout embryonic development (Nguyen et al., 2017; Ganassi et al., 2018). Additionally, their epaxial trunk muscles have a simple structure consisting of only one anatomical unit (reviewed in (Sefton and Kardon, 2019)). Like other vertebrates, fish myotomes, on either side of the neural tube, extend dorsally after somite differentiation and eventually cover the neural tube by the end of embryonic development (Figure 1A). The spontaneous medaka (Oryzias latipes) mutant Double anal fin (Da) displays a particular epaxial myotome morphology, in which the dorsal ends of the left and right epaxial myotome fail to extend sufficiently and thus do not cover the neural tube at the end of embryonic development. Previous studies demonstrated that the dorsal trunk region of the Da mutant is transformed into the ventral one, including not only the myotome but also the body shape, skeletal elements, pigmentation and fin morphology (Figure 1B, D) (Ishikawa, 1990; Ohtsuka et al., 2004). Given the unique morphological features, the medaka Da mutant is an excellent model to study the morphogenesis of epaxial myotome. Genetic analysis of the Da mutant revealed that this phenotype is due to a dramatic reduction of the expression of the transcription factors zic1 and zic4 in the dorsal somites (Figure 1C, E), and identified zic1/zic4 as master regulators of trunk dorsalization (Ohtsuka et al., 2004; Kawanishi et al., 2013). The down-regulation of zic1/zic4 specifically in the dorsal somites is caused by the insertion of a large transposon, disrupting the dorsal somite enhancer of zic1/zic4 (Moriyama et al., 2012; Inoue et al., 2017). While the function and downstream targets of Zic1 and Zic4 have been studied in the nervous system (Aruga and Millen, 2018), the molecular mechanism of how these Zic genes control dorsal trunk morphogenesis has not been investigated so far.
Here we describe the morphogenetic process of the formation of the epaxial myotome of the back, which we termed “dorsal somite extension” (Figure 1A). By in vivo time-lapse imaging we uncovered its cellular dynamics; during dorsal somite extension, dorsal dermomyotome (DM) cells reduce their proliferative activity and subsequently form unique large protrusions extending dorsally, potentially guiding the epaxial myotome dorsally. In the Da mutant, by contrast, DM cells keep their high proliferative activity and mainly form small protrusions. Mechanistically, we identify a Zic1 downstream-target gene, wnt11r, as a crucial factor for dorsal somite extension. We demonstrate that Wnt11r regulates cellular behavior of dorsal DM cells by promoting protrusion formation and negatively regulating proliferation.
Results
Our previous study showed that zic1 and zic4 expression starts at embryonic stages and persists throughout life (Kawanishi et al., 2013). Phenotypic analysis of homozygous adult Da mutants implies long-term participation of Zic-downstream genes in the formation of dorsal musculatures, which eventually affects the external appearance of the fish adult trunk. Here, we examined the initial phase of this long-term dorsalization process. In the following of the study, we will focus on zic1, since zic1 and zic4 are expressed in an identical fashion with overlapping functions in trunk dorsalization of medaka, and zic4 is expressed slightly weaker than zic1 (Moriyama et al., 2012; Kawanishi et al., 2013).
The dorsal myotome of the Da mutant fails to cover the neural tube
In wild type (Wt) medaka, the dorsal ends of the myotomes first came in contact at 7 days post fertilization (dpf, stage 37) and formed the tight, thick myotome layer covering the neural tube at the end of embryonic development (9 dpf, stage 39) (Figure 1F-G, Figure 1 - figure supplement 1A-F). In the ventralized Da mutant, however, the dorsal ends of the myotomes did not extend sufficiently and failed to cover the neural tube at the end of embryonic development (Figure 1H-I, Figure 1 - figure supplement 1G-L). The ends of the ventralized dorsal myotome in the mutant displayed a round shape (not pointed as found in Wt) which resembled the morphology of the ventral myotome.
We wondered if there are other morphological differences between Wt dorsal myotome and the ventralized dorsal myotome of the Da mutant. Indeed, the cross-sectional area in the Da mutant was significantly larger compared to Wt (Figure 1L). Possible explanations for a larger cross-sectional area in the Da mutant could be a larger myofiber diameter or a higher number of myofibers which make up the myotome. We measured the diameter of dorsal myotome muscle fibers in Wt and Da mutant embryos but could not observe a difference, suggesting that dorsal myotome of the Da mutant has a higher number of myotomal cells (Figure 1 - figure supplement 1M).
Proliferative activity of the dorsal DM cells is enhanced in the Da mutant
In fish, as in other vertebrates, the DM gives rise to muscle precursor cells which ultimately differentiate into myofibers. In medaka the DM is a one cell-thick, Pax3/7-positive cell layer encompassing the myotome (Figure 2A-B’’) (Hollway et al., 2007; Abe et al., 2019). A high proliferative activity of the dorsal DM could explain a larger cross-sectional area of the dorsal myotome of the Da mutant. To test this, we performed immunohistochemistry against the mitotic marker phosphor-histone H3 (PH3) and the DM marker Pax3/7 on Wt and Da embryos (Figure 2C-E). In both Wt and Da embryos, PH3-positive cells were randomly distributed in the dorsal DM without obvious bias (Figure 2C-D). At the 12-somite stage (12 ss, 1.7 dpf stage 23), when zic1 expression in the somites becomes restricted to the dorsal region (Kawanishi et al., 2013), the number of PH3-positive DM cells per dorsal somite was not significantly different between Wt and Da. Remarkably, from 16 ss (1.8 dpf, stage 24) onwards, the number of PH3-positive cells became reduced in the Wt, whereas in Da, no such reduction was observed (Figure 2E). At 35 ss (3.4 dpf, stage 30), PH3-positive cells increased both in the Wt and Da, but the mutant DM cells were more proliferative (Figure 2E). The number of PH3-positive cells in the ventral DM was not significantly different in Wt and Da embryos at 22 ss (2.75 dpf, stage 26) (Figure 2—figure supplement 1A). These results suggest that zic1 reduces proliferative activity of the DM, which becomes evident following the confinement of its expression to the dorsal somite region.
Wt dorsal DM cells form numerous large, motile protrusions at the onset of dorsal somite extension
The epaxial myotome, on either side of the neural tube, extends dorsally to cover the neural tube by the end of embryonic development. To examine the behavior of zic1-positive cells underlying this dorsal somite extension, we performed in vivo time-lapse imaging of dorsal somites using the transgenic line Tg(zic1:GFP; zic4:DsRed), which expresses GFP under the control of the zic1 promoter and enhancers to visualize the dorsal somitic cells (Kawanishi et al., 2013) (hereafter called Tg(zic1:GFP) since the DsRed fluorescence was negligible in the following analyses). Intriguingly, around 22 ss and onwards, cells at the tip of the dorsal somites started to form numerous large protrusions extending dorsally towards the top of the neural tube (Figure 3A, Video 1). We defined the beginning of protrusion formation as the onset of dorsal somite extension. Close-up views of the time-lapse images (Figure 3A, Video 1) showed that these protrusions were motile and dynamically formed new branches at their dorsal tips (Figure 3J). Immunohistochemistry revealed that the protrusion-forming cells belong to the DM (Figure 3 - figure supplement1 A-A’’).
To characterize the protrusions, we classified them according to their length into small (< 8 μm, Figure 3C arrowheads) and large (≥ 8 μm, Figure 3C arrow) protrusions (Figure 3 - figure supplement 1B-C). Based on their shape, we reasoned that the small protrusions correspond to lamellipodia (Figure 3C, arrowheads), while the large protrusions appeared more complex. To investigate the nature of large protrusions, we injected Actin-Chromobody GFP mRNA to visualize the actin skeleton. The large protrusions were found to exhibit a complex architecture consisting of a lamellipodia-like core structure (Figure 3D, bracket) with additional multiple bundles of filopodia (protrusions with linear arranged actin filaments) branching out from their dorsal tips (Figure 3D, arrows, summarized in Figure 3E).
Interestingly, time-lapse in vivo imaging of Tg(zic1:GFP);Da showed that in the Da background, protrusions started to form later (Figure 3B, Video 2) and the number of large protrusions and protrusions in total was significantly lower than in Wt (Figure 3I-I’’). In addition, protrusions were transient and mostly failed to form new branches at their dorsal tips (Figure 3K). While no difference in the actin skeleton of small protrusions could be observed, filament bundles branching out from large protrusions of Da DM cells contained fewer and shorter filopodia compared to Wt (Figure 3G, arrowheads, summarized in Figure 3H). These results indicate that the protrusive activity, especially the ability to form large protrusions, is significantly reduced in the Da mutant. The large protrusions of the dorsal DM cells continuously appeared at later stages of dorsal somite extension, too (Figure 3 - figure supplement 2).
Taken together, the unique large protrusions of the dorsal DM might be involved in guiding the epaxial myotome dorsally and zic1 might promote this function.
Video 1: Onset of dorsal somite extension in Tg(zic1:GFP).
Dorsal view of time-lapse in vivo imaging of 24 ss Tg(zic1:GFP) embryo. 15th somite is positioned in the center, z-stacks were imaged every 10 min, time is displayed in min.
Anterior = left, scale bar = 50 μm.
Video 2: Onset of dorsal somite extension in Tg(zic1:GFP);Da.
Dorsal view of time-lapse in vivo imaging of 24 ss Tg(zic1:GFP);Da embryo. 15th somite is positioned in center, z-stacks were imaged every 10 min, time is displayed in min. Bright cell at the bottom migrating to the right is a melanophore. Anterior = left, scale bar = 50 μm.
DM cells delaminate and accumulate between opposing somites during the late phase of dorsal somite extension
We continued to trace the behavior of the DM tip cells until the somites reach the top of the neural tube. Indeed, in vivo imaging of Tg(zic1:GFP) revealed that dorsal DM cells continued to form protrusions, and additionally, some of them delaminated to become Zic1-positive mesenchymal cells accumulating in the space between the dorsal ends of the left and the right somites (Figure 4A, star-shaped cells, arrowheads) from 4.5 dpf (stage 33) onwards. This is consistent with previous observations of strongly zic1 expressing mesenchymal cells in Wt at late embryonic stages (Ohtsuka et al., 2004). As dorsal somite extension proceeded, the number of these mesenchymal cells increased, filling the space between the two myotomes (Figure 4A-D, arrowheads, Video 3-4, arrowheads indicate representative mesenchymal cells). These mesenchymal cells formed protrusions towards neighboring mesenchymal cells and DM cells at the tip of somites, creating a dense cellular network between the dorsal ends of the somites. Mosaic cell-labeling demonstrated that the mesenchymal cells originated from the DM (Figure 4 - figure supplement 1A-F, arrowheads). Interestingly, while mesenchymal cells dynamically formed protrusions, they showed no extensive migratory behavior and were rather stationary (Video 3-4, Figure 4 – figure supplement 2). This could suggest that these protrusions fulfill a non-migratory function. When the opposing somites came in contact with each other at 8 dpf (stage 38, one day before hatching), the mesenchymal cells tended to attach to the nearest DM cells at the tip of the somite, bridging the gap between the left and right DM cells (Figure 4K-K’’’’).
In Da mutants, mesenchymal DM cells were also detected in the space between the two myotomes, but the timing of their appearance was delayed, i.e. between 5 dpf (stage 34) and 5.5 dpf (stage 35) (Figure 4E-H) (4.5 dpf in Wt). Additionally, Da mesenchymal cells exhibited a rounder morphology and formed significantly fewer protrusions compared to Wt (Figure 4I-J, Figure 4 - figure supplement 1G-L).
Collectively, dorsal DM cells and the mesenchymal cells derived from them seem to actively participate in the entire process of dorsal somite extension, from its onset to neural tube coverage at the end.
Video 3: Mesenchymal DM cells during dorsal somite extension at 4.5 dpf.
Dorsal view of time-lapse in vivo imaging of 4.5 dpf Tg(zic1:GFP) embryo. 10th somite positioned in center, z-stacks were imaged every 10 min, time is displayed in min.
Arrowhead indicates representative mesenchymal DM cell. Anterior = left, scale bar = 50 μm.
Video 4: Mesenchymal DM cells during dorsal somite extension at 5.5 dpf.
Dorsal view of time-lapse in vivo imaging of 5.5 dpf Tg(zic1:GFP) embryo. 10th somite is positioned in center, z-stacks were imaged every 10 min, time is displayed in min.
Arrowhead indicates representative mesenchymal DM cell. Anterior = left, scale bar = 50 μm.
Zic1 regulates the expression of dorsal-specific genes during somite differentiation
We then addressed the molecular machinery controlling the dorsal somite extension investigated above. Since somite extension is impaired in the zic1-enhancer mutant Da (Moriyama et al., 2012; Kawanishi et al., 2013), we reasoned that downstream genes of Zic1 are regulators of this process.
First, we identified genes which are specifically expressed in the dorsal somites. For this, we dissected somites of the transgenic line Tg(zic1:GFP) and FACS sorted them into cells from the dorsal (GFP+) and ventral (GFP-) somites, including the DM, and performed RNA-seq and ATAC-seq on both cell populations (Figure 5A, Figure 5 - figure supplement 1A). The RNA-seq identified 1,418 differentially expressed genes. Among them 694 genes showed higher expression in the dorsal somites (termed hereafter dorsal-high genes), and 724 genes showed higher expression in the ventral somites (termed hereafter dorsal-low genes) (Figure 5C). We confirmed that zic1 and zic4 were found among the dorsal-high genes (Figure 5C, Supplementary table 1).
Next, we identified potential direct Zic1-target genes by investigating Zic1 binding sites. Since there were no suitable antibodies available to perform ChIP-seq against medaka Zic1, we created a transgenic line expressing a Myc tagged zic1 in the Da background under the control of zic1 promoter and enhancers (Tg(zic1:zic1-Myc;zic4:DsRed);Da, called Tg(zic1:zic1-Myc);Da hereafter). This transgene (zic1:zic1-Myc) efficiently rescued the ventralized phenotype (dorsal and anal fin shape, pigmentation pattern, and body shape) of the Da mutant (Figure 5 - figure supplement 1B-C), verifying the full functionality of the tagged Zic1 protein. Somites of the transgenic line Tg(zic1:zic1-Myc);Da were dissected and subjected to ChIP-seq using antibodies against Myc (Figure 5B). Since somites contain a low number of cells, we applied ChIPmentation (Schmidl et al., 2015) to identify genome-wide Zic1 binding sites. From two biological replicates, 5,247 reliable ChIP peaks were identified, and we confirmed that the enriched DNA motif among these peaks was showing high similarity with previously identified binding motifs of ZIC family proteins (Figure 5 - figure supplement 2A-B). Then, we associated each Zic1 peak to the nearest gene within 50 kb and identified 3,232 genes as Zic1 target genes. By comparing the ChIPmentation with RNA-seq data, we found that Zic1 target genes are overrepresented in dorsal-high and dorsal-low genes. While 47% of the dorsal-high genes and 27% of the dorsal-low genes were found to be Zic1 target genes, only 12% of genes which showed no differential expression in dorsal or ventral somites were potential Zic1 downstream target genes (Figure 5D, Supplementary table 1). This suggests that Zic1 can function as transcriptional activator and repressor of versatile genes, but the former role seems to be dominant.
Wnt11r is a direct downstream target of Zic1 and down-regulated in the dorsal somites of the Da mutant
To identify potential regulators of dorsal somite extension, we further investigated the differentially expressed Zic1 target genes. Gene Ontology (GO) analysis indicated that both dorsal-high and dorsal-low gene groups, regardless of whether they are Zic1 targets or non-Zic1 targets, were significantly enriched in development related GO terms (Figure 6A, Figure 6 - figure supplement 1A, Supplementary table 2, 3). This indicates that Zic1 regulates a number of developmental genes both directly and indirectly. These results are consistent with the fact that Zic1 regulates various dorsal-specific morphologies of somite-derivatives (Kawanishi et al., 2013).
In the dorsal-high Zic1 targets, GO terms related to cell migration showed higher enrichment (e.g. “chemotaxis” (P=5.62E-10), “locomotion” (6.65E-15), “ameboidal-type cell migration” (P=9.32E-10)) than dorsal-low genes or non-Zic1 target genes (Supplementary table 3). Interestingly, Wnt signaling pathway genes (e.g.: axin2, wnt11r, sp5, lrp5, fzd10, prickle1a) and semaphorin-plexin signaling pathway genes (e.g.: sema3a, sema3c, sema3g, plxna1, plxna2, plxnb2, plxnb3, nrp2) were included in these gene groups. Indeed, the GO terms “Wnt signaling pathway” and “Semaphorin-plexin signaling pathway” themselves were also significantly enriched (P=4.45E-7 and 1.94E-8, respectively) in dorsal-high Zic1 target genes (Supplementary table 2). This suggests that Zic1 directly regulates Wnt pathway, semaphorin and plexin genes, possibly to regulate cell movement in the dorsal somites. We also noticed that the terms “extracellular matrix organization” (P=7.74E-7; e.g.: adamts20, fbln1), “cell communication” (P=1.95E-5; e.g.: efna5, epha3) are enriched, suggesting that these genes also affect dorsal somite cell behavior.
Wnt signaling pathway components also ranked high among the dorsal-high Zic1 target genes by pathway enrichment analysis (Figure 6B, Figure 6 - figure supplement 1B). Among them, we focused on wnt11r for further analyses due to the following reasons: First, wnt11r was one of the most differentially expressed genes in dorsal somites (Figure 5C, Supplementary table 1). Second, previous studies implicated Wnt11 in protrusion formation and cell migration (Ulrich et al., 2003; De Calisto et al., 2005; Garriock and Krieg, 2007; Matthews et al., 2008). This is particularly interesting since DM cells also exhibit protrusions and migration activity during dorsal somite extension, which are defective in Da mutants.
At the wnt11r locus, peaks of the Zic1-ChIP overlapped with intergenic open chromatin regions downstream of wnt11r. These sites were more accessible in dorsal somites than in ventral somites, suggesting that Zic1 regulates wnt11r via enhancers (Figure 6C).
Additionally, in situ hybridizations against wnt11r performed on Wt and Da embryos indicated that wnt11r expression is significantly reduced at the dorsal tip of Da dorsal somites including the DM (Figure 6 - figure supplement 1C-D). Already when the zic1 expression becomes restricted in the dorsal somites (12 ss) and with proceeding development when zic1 expression gets further restricted to the most dorsal part of the somites and mesenchymal DM cells (Ohtsuka et al., 2004), the expression of wnt11r in the dorsal somites of the Da mutant was reduced compared to Wt (Figure 6D-F, Figure 6 - figure supplement 1C-F’).
Taken together, we identified wnt11r as promising downstream target gene of Zic1, which could be a novel somite dorsalization factor and play a role in dorsal somite extension.
Wnt11r regulates cell behavior of dorsal DM
Our previous RNA-seq dataset revealed that the expression of wnt11r starts before gastrulation and increases as development proceeds (Figure 7 - figure supplement 1A)(Nakamura et al., 2021). This makes it challenging to examine the role of wnt11r during late embryonic development using loss-of-function experiments. We took two different approaches to knock-down Wnt11r during dorsal somite extension. Firstly, we used a wnt11r anti-sense morpholino (wnt11r MO) and determined a concentration which resulted in maximal knock-down effects during dorsal somite extension with minimal gastrulation phenotypes (we excluded any embryos showing gastrulation defects in our experiments). Secondly, we performed temporally controlled knock-down of Wnt11r after gastrulation using photo-cleavable Photo-Morpholinos (PhotoMOs) (Tallafuss et al., 2012) (Figure 7A). Strikingly, in wnt11r Photo-Morphants with severe phenotypes the dorsal myotome failed to cover the neural tube (n = 7, Figure 7B-C), a phenotype similar to the Da mutant myotome. Additionally, the myofibers of the Photo-Morphants at 9 dpf were shorter and less organized. This phenotype is consistent with a previously reported role of Wnt11 during early myogenesis, where it regulates the elongation and orientation of myoblasts (Gros, Serralbo and Marcelle, 2008).
Next, we investigated the proliferative activity of dorsal DM cells in wnt11r morphants and found that knock-down of wnt11r (either by conventional MO or by PhotoMO) induced significantly more PH3-positive cells per somite, compared to control embryos at 22 ss (Figure 7D). These findings are similar to the observations previously made in the dorsal DM of Da mutants, and suggest that Wnt11r is negatively regulating proliferation of dorsal DM cells.
We then explored whether the onset of dorsal somite extension in wnt11r morphants is similarly impaired as in the Da by time-lapse in vivo imaging. Remarkably, we observed a protrusion formation behavior of wnt11r morphant DM cells similar to Da DM cells, namely delayed onset of protrusion formation and the formation of fewer, shorter protrusions (Figure 7E, Video 5). Quantification of protrusions indicated that wnt11r morphants have significantly fewer large protrusions and protrusions in total, compared to control embryos (Figure 7F-H’’).
Overall, the knock-down of the Zic1 target gene Wnt11r recapitulated the phenotype of Da DM cells (Figure 3G-G’’), showing the essential role of Wnt11r in regulating protrusion formation of DM cells.
Video 5: Onset of dorsal somite extension in wnt11r morphant embryo.
Wnt11r was knocked-down using the PhotoMO approach (Figure 7A) in a Tg(zic1:GFP) embryo. Dorsal view of time-lapse in vivo imaging of 24 ss Tg(zic1:GFP) embryo. 15th somite is positioned in the center, z-stacks were imaged every 10 min, time is displayed in min. Anterior = left, scale bar = 50 μm.
To further confirm the importance of Wnt11r during dorsal somite extension, we performed rescue experiments in Da embryos. At 18 ss (2.1 dpf, stage 25), we injected a mix of human recombinant Wnt11 (hrWnt11) protein (or BSA in the control group) and Dextran Rhodamine onto the top of the 10th somite of Tg(zic1:GFP);Da embryos (Figure 8A, B). Strikingly, the dorsal DM of Da embryos injected with Wnt11 formed significantly more large protrusions and more protrusions in total compared to Da embryos injected with BSA only (Figure 8C-C’’). This indicates that the ventralized Da DM protrusion phenotype can be partially rescued by Wnt11 protein injections and further emphasizes the importance of Wnt11r in somite dorsalization and during dorsal somite extension.
Wnt11r acts through the Wnt/Ca2+ signaling pathway at the onset of somite extension
Finally, we investigated through which signaling pathway the non-canonical Wnt11r acts. In Xenopus, Wnt11r acts through the Wnt/Ca2+ pathway regulating the migration of cells from the dorsal somite and the neural crest into the dorsal fin fold (Garriock and Krieg, 2007). To examine whether this signaling pathway also plays a role during dorsal somite extension, we inhibited the Wnt/Ca2+ signaling pathway using KN-93 in Tg(zic1:GFP) embryos from 4 ss (1.3 dpf, stage 20) to 22 ss (Figure 9A). KN-93 is known to inhibit CaMKII, a component of the Wnt/Ca2+ pathway (Wu and Cline, 1998; Garriock and Krieg, 2007; Rothschild et al., 2013). Embryos treated with KN-93 showed a higher number of PH3-positive dorsal DM cells per somite, compared to embryos in the control group (Figure 9B). Furthermore, DM cells at the tip of the dorsal somite of embryos treated with KN-93 formed significantly fewer large and fewer protrusions in total, compared to embryos in the control group (Figure 9C-C’’), although the effect was less significant compared to the wnt11r morphants (Figure 7).
From these results, we suggested that Wnt11r potentially acts through the Wnt/Ca2+ signaling pathway at the onset of somite extension. Wnt/Ca2+ signaling pathway is known to regulate actin polymerization (Choi and Han, 2002; Kohn and Moon, 2005) which could explain the dynamic protrusive activity of the DM cells (Figure 3), emphasizing the importance of Wnt11r during epaxial myotome morphogenesis.
Discussion
Here, we elucidate the key developmental process underlying the epaxial myotome morphogenesis in teleost fish, medaka. At the initiation of dorsal somite extension, DM cells at the tip of the dorsal somite form unique large, motile protrusions, which extend dorsally, and potentially play a pioneering role in guiding the myotome towards the top of the neural tube. By analyzing dorsal somite extension in the ventralized Da mutant, we demonstrated that zic1 is essential for this process. DM cells of the Da dorsal somite have a higher proliferative activity and form fewer and shorter protrusions during dorsal somite extension. These altered cellular properties of Da dorsal DM cells, together with a delayed onset of dorsal somite extension, potentially cause the incomplete coverage of the neural tube by the epaxial myotome at the end of embryonic development.
Furthermore, we investigated the molecular background of dorsal somite extension and identified a direct downstream target of Zic1, the non-canonical Wnt wnt11r, as a crucial factor for dorsal somite extension. Wnt11r reduces the proliferative activity of DM cells in the dorsal somites, but instead, promotes the formation of large, motile protrusions of dorsal DM cells at the tip of the somite. Indeed, wnt11r morphants recapitulate the phenotype of the Da mutant. Additionally, the protrusion phenotype of Da dorsal DM cells can be partially rescue by injection of Wnt11 proteins. Based on these findings, we propose a model for dorsal somite extension shown in Figure 10.
In the present study, we described the characteristic behavior of dorsal DM cells and suggest their guiding role for the epaxial myotome moving towards the top of the neural tube. During dorsal somite extension, DM cells at the tip of the dorsal somites form large, motile protrusions which contain multiple bundles of filipodia-like protrusions dynamically branching out from the tip of the protrusions. These long protrusions could be beneficial for invading the restricted open space between the neural tube and the ectoderm. Additionally, previous studies showed that in migrating mesenchymal cells, the formation of lamellipodia is associated with higher migratory speed, whereas filopodia play an exploratory role and are associated with high directionality (Leithner et al., 2016; Innocenti, 2018). Large protrusions of wild-type DM cells consist of a lamellipodia-like core and multiple filopodia, which could account for fast dorsal somite extension with high accuracy. The detailed structure and function of the large protrusions needs to be further investigated in the future studies.
At later stages of dorsal somite extension, we observed that DM cells delaminate from the tip of the dorsal somites and progressively occupy the space between the opposing dorsal somites. These mesenchymal DM cells actively form protrusions towards neighboring mesenchymal DM cells and DM cells of the somites, thus forming a dense cellular network on top of the neural tube. This could provide a communication platform for the opposing somites to meet at the right position, exactly on top of the neural tube.
The ability of the motile DM cells to guide the underlying myotome dorsally assures complete coverage of the neural tube by the epaxial myotomes. Considering the fact that DM cells are in close cell-cell contact with the underlying myotome, it is plausible that they are biomechanically coupled with myotomal cells, facilitating dorsal myotome extension. A similar mechanism has been investigated in the mouse neural tube formation. During mouse neural tube closure, the surface ectoderm, overlying the neuroepithelium, forms cell protrusions towards the midline. Disruption of membrane ruffles, a form of lamellipodia, in the surface ectoderm results in incomplete fusion of the subjacent neuroepithelium (Rolo et al., 2016).
In medaka dorsal somites, Wnt11r exerts its effect through promotion of protrusion formation and down-regulation of cell proliferation in the dorsal DM. Regarding the protrusion formation, previous studies of the zebrafish mutant silberblick and migrating neural crest cells consistently reported that Wnt11 is involved in the oriented elongation and stabilization of protrusions (Ulrich, 2003; De Calisto et al., 2005; Matthews et al., 2008). Similarly, myocardial cells in the heart of Wnt11 mutant mice also form fewer protrusions (Zhou et al., 2007). Thus, regulation of cell protrusions is a conserved function of Wnt11, observed in various developmental processes. However, the relationship between Wnt11 and cell proliferation could be context-dependent, as it negatively regulates proliferation in mouse neonatal hearts (Touma et al., 2017), while it promotes cell proliferation in mouse intestinal epithelial cell culture (Ouko et al., 2004). Hence, the function of Wnt11r in medaka somites is unique in that it promotes tissue elongation by regulating a balance between proliferative and migrative activity.
By inhibiting the Wnt/Ca2+ signaling pathway we showed that Wnt11r probably acts through this non-canonical Wnt signaling pathway during dorsal somite extension. Likewise, previous studies in Xenopus have shown that Wnt11r acts through this pathway during cell migration into the dorsal fin fold. There it regulates epithelial–mesenchymal transition in a distinct dorsal somite cell population which, together with a population of neural crest cells, contribute to the mesenchyme of the dorsal fin fold (Garriock and Krieg, 2007). Furthermore, during convergent extension in vertebrate gastrulation, Wnt/Ca2+ pathway can regulate cell adhesion by promoting actin polymerization (Choi and Han, 2002; Kohn and Moon, 2005).
Importantly, our RNA-seq and ChIPmentation analyses revealed that Zic1 has diverse downstream target genes including various developmental genes. This suggests that the dorsal myotome is established via pleiotropic actions of Zic1; Wnt11r may not be a sole factor for dorsal somite extension, although it was shown to be essential in the present study. The semaphoring-plexin pathway may play a role, since previous studies suggested, besides its implication in axon guidance and neural cell migration, a role in non-neural cell migration (Alto and Terman, 2017). Furthermore, cell migration and tissue deformation are often linked with the extracellular matrix (ECM), which provides guiding or restraining cues influencing cell movements. Previous studies showed that during mouse epaxial myotome development, ECM composition changes dynamically, which is tightly accompanied by epaxial muscle morphogenesis; while the laminin content in the ECM decreases, increasing tenascin and stable fibronectin contents potentially promotes the alignment of myofibers and their final organization (Deries et al., 2012). Furthermore, recent studies suggested that cells can remodel the surrounding ECM and thereby increase their own motility. For example, during zebrafish gastrulation, the metalloproteinase mmp14 is expressed by migrating endoderm cells and degrades laminin and fibronectin, components of the ECM (Hu et al., 2018). In this context, of particular importance is our identification of Adamts20, encoding a proteoglycanase, as a dorsal-high Zic1-target gene in the somites (Supplementary table 1). Since Adamts20 is known to play a pivotal role in embryonic melanoblast migration by remodeling the dermal ECM (Rao et al., 2003; Silver et al., 2008), Adamnts20 could thus facilitate migration of Wnt11r-expressing DM cells through remodeling of the ECM. Further functional analysis of dorsal-high Zic1-target genes identified in somites will provide useful insight into the molecular network driving dorsal somite extension.
Finally, in vertebrates and especially in fish, body shape and muscle morphology are closely linked, since a majority of the body mass consist of muscular tissue. Previous studies in fish populations have shown that speciation and adaptation to a specific aquatic habitat are associated with a changes in body depth, a measurement of the trunk dorsoventral axis (Tobler et al., 2008; Elmer et al., 2010; Weese, Ferguson and Robinson, 2012; Fruciano et al., 2016). In this context, the external appearance of the adult Da mutant is intriguing, in that it exhibits a teardrop body shape (typical for fish swimming in the middle layer like tuna), instead of a dorsally flattened one (typical for surface swimming fish like medaka). Since our study shows that the activity of wnt11r could influence the body shape by regulating cell proliferation and the behavior of muscle progenitor cells, Wnt11r could be one of the crucial factors in evolution and diversity of body shape in fish. Furthermore, the expression of zic1 (Ohtsuka et al., 2004; Sun Rhodes and Merzdorf, 2006; Houtmeyers et al., 2013) and wnt11r (Marcelle, Stark and Bronner-fraser, 1997; Olivera-Martinez, Thelu and Dhouailly, 2004; Garriock et al., 2005; Matsui et al., 2005) are strongly conserved among vertebrates, and we hypothesize that Wnt11r-mediated morphogenesis of the somites represents an evolutionarily conserved mechanism that acts across vertebrates.
Materials and Methods
Key resource table
Animals and transgenic lines
Fish were raised and maintained under standard conditions. All experimental procedures and animal care were performed according to the animal ethics committee of the University of Tokyo. Sex was randomly assigned to experimental groups. Medaka d-rR stain was used as wild-type, the Da mutant used in this study was previously described (Ohtsuka et al., 2004). The pre-existing transgenic line Tg(zic1:GFP/zic4:DsRed) (Kawanishi et al., 2013) was used, and the transgenic line Tg(zic1:GFP/zic4:DsRed);Da was created by crossing Da mutants with Tg(zic1:GFP/zic4:DsRed). The transgenic line Tg(zic1:zic1-Myc/zic4:DsRed);Da was generated by modifying the BAC used to generate Tg(zic1:GFP/zic4:DsRed) (Kawanishi et al., 2013), by replacing the ORF of GFP with the ORF of zic1 containing a sequence for a Myc-tag fused to its C-terminus. To establish the transgenic line, the BAC(zic1:zic1-Myc/zic4:DsRed) was co-injected with I-SceI Meganuclease (NEB) into 1-cell stage Da embryos, as previously described (Thermes et al., 2002).
Visualization of actin skeleton of protrusions
The AC-TagGFP2 sequence from the Actin-Chromobody plasmid (TagGFP2) (Chromotek) was cloned into the pMTB vector for mRNA generation. To investigate the actin skeleton of protrusions, cells of Wt and Da embryos were mosaically labelled using Actin-Chromobody-GFP (AC-GFP) mRNA. Embryos were injected at 1-cell stage with 152 ng/µl membrane-mCherry mRNA. At 4-cell stage, one cell was injected with 184 ng/µl AC-GFP mRNA.
Lineage tracing of mesenchymal DM cells
Mosaic labelling of cells was achieved by co-injecting 20 ng/µl pMTB-memb-mTagBFP2 plasmid with Tol2 into a 1-cell stage embryo of the transgenic line Tg(zic1:GFP/zic4:DsRed). Embryos were raised until 5 dpf and labelled cells were continuously observed until 8 dpf.
Morpholino injection
Microinjections into medaka embryos were performed using 12.5 µM wnt11r Morpholino antisense oligonucleotides (MO). Injections were performed into 1-cell stage embryos.
Photo-Morpholino mutagenesis
Wnt11r Sense-Photo-Morpholino (PhotoMO) and wnt11r antisense Morpholino were annealed in a ratio 2:1. Microinjection of 25 µM of the annealed oligonucleotides was performed into 1-cell stage embryos and embryos were raised until 4 ss in the dark. Photo-cleavage was performed using the 10x objective and the DAPI filter of a Keyence BZ-9000 Biorevo microscope (Keyence). Embryos were mounted, dorsal side facing up, in 1 % Methylcellulose in a glass bottom dish (Wako) and illuminated for 30 min. After Photo-cleavage, embryos were dechorionated and raised until the desired stage for subsequent analysis.
Injection of human recombinant Wnt11 protein into Da mutant somite
To immobilize embryos, embryos from the Tg(zic1:GFP/zic4:DsRed);Da transgenic line were injected with 25 ng/µl α-bungarotoxin mRNA at 1-cell stage. At 18 ss embryos were mounted in 1 % low melting agarose in 1x Yamamoto’s Ringer Solution and oriented with the dorsal side facing upwards. Embryos were injected on top of the 10th somite with a mix containing Dextran Rhodamine (Thermo Fisher) and 1.7 ng hrWnt11 protein (R&D Systems) or BSA (Sigma-Aldrich) and raised to 24 ss, followed by in vivo imaging and analysis.
KN-93 treatment
Dechorionated 4 ss embryos were treated with 30 μM KN-93 (Wako, Japan) or DMSO (Sigma Aldrich, Germany) in 1x Yamamoto’s Ringer Solution at 28 °C, in the dark until 22 ss was reached.
In vivo imaging and in vivo time-lapse imaging
To immobilize embryos for in vivo imaging, embryos were injected at 1-cell stage with 25 ng/µl α-bungarotoxin mRNA (Swinburne et al., 2015; Lischik, Adelmann and Wittbrodt, 2019). Embryos were mounted in 1 % Low melting agarose (Sigma Adrich) in 1x Yamamoto’s Ringer Solution in a glass-bottomed petri dish (IWAKI) and oriented with the dorsal side facing down. Imaging was performed using a Zeiss LSM 710 confocal microscope system (Zeiss) equipped with an inverted stand and a Zeiss AXIO Observer Z1 and a T-PMT detector. The embryos were positioned with the 5th or 10th somite in the center and images were acquired using a 40x water objective. For the in vivo time-lapse imaging, the 10th or 15th somite was positioned in the center, z-stacks were imaged in a 600-sec interval for 10-15 h. Image analysis was performed in Fiji using the “Image Stabilizer” Plugin, the FFT Bandpass filter and the “Draw_arrows” Plugin to draw customized arrows (Li, 2008; Daetwyler, Modes and Fiolka, 2020).
Whole mount in situ hybridization
Whole mount in situ hybridization was performed as previously described with the following modifications (Takashima et al., 2007). Embryos were fixed in 4 % PFA/1.5x PTW at 4 °C, overnight. Hybridization was performed at 65 °C, overnight. Samples were treated with alkaline-phosphatide anti-DIG-AP Fab fragments (1:2000, Roche). Signals were developed using 4-nitro blue tetrazolium chloride (NBT, Roche) and 5-bromo-4-chloro-3-indolyl phosphate (BCIP, Roche).
Whole-mount immunohistochemistry
Embryos were fixed in 4 % PFA/PBS for 2 h at room temperature or at 4 °C overnight. Samples were permeabilized with 0.5 % TritonX-100 (Wako) in 1x PBS for 1-2 h and blocked in blocking solution (2 % BSA (Sigma-Aldrich), 1 % DMSO, 0.2 % TritonX-100 in 1x PBS) for 2-4 h at room temperature. Samples were incubated with respective primary antibody diluted in blocking solution at 4 °C, overnight. After an additional 4 h blocking step, samples were incubated with respective secondary antibodies diluted in blocking solution, at 4 °C, overnight. Samples were stored in 1x PBS at 4 °C until imaging.
Antibodies
Vibratome sectioning
Samples were mounted in 4 % agarose in 1x PBS. 40 μm or 200 µm sections were obtained by a Vibratome (Leica, Vibratome). The sections were mounted on a glass slide (Matsunami) in 60 % glycerol (Merck, Wako) and stored at 4 °C until imaging. Images were acquired using the 40x water objective of a Zeiss LSM 710 confocal microscope.
Histological sections
Dechorionated embryos were fixed in Bouin’s solution overnight, followed by a gradual dehydration using ethanol. Samples were embedded in Technovit 7100 (Heraeus Kulzer) and sectioned into 5-6 µm thick sections. Sections were stained with hematoxylin (Wako) and imaged using the 1.6x objective of a Leica M165 FC fluorescent stereo microscope.
Image processing and statistical analysis
Image processing was performed with the image processing software Fiji. The 3D-recreation of in vivo imaging date was created using FluoRender (Wan et al., 2009). Measurements of morphological features (distance between myotome tips, area of cross-section of dorsal somites, diameter of myofibers, distance between dorsal somite tip and the tip of neural tube, somite height) was performed by averaging the analysis of the feature from 3 consecutive Y planes. RStudio was used for the statistical analysis and representation of the data. In bar plots mean and error limits, defined by the standard deviation, are indicated. In box plots median first and third quantiles are indicated. Statistical significance was determined by un-paired t-tests, a p-value p < 0.05 was considered as significant. In the figure legends sample size (n) and number of individuals used in the experiment are stated. Sample sizes were not predetermined using statistical methods, but the sample sizes used are similar to those generally used in the field. To compare experimental groups, the allocation was performed randomly, without blinding.
Isolation of dorsal and ventral somite cells for ATAC-seq and RNA-seq
Yolk and head region were removed from 22 ss Tg(zic1:GFP/zic4:DsRed) embryos and embryos were incubated with 10 mg/ml pancreatin (Wako) at room temperature for 5-10 min. Epidermis and intestinal tissues were removed and somites were isolated from the neural tube. Cells were dissociated in 0.5% (w/v) Trypsin (Nacalai Tesque) at 37 °C for 10 min, the dissociation was stopped by adding the same volume of 15% (v/v) FBS / Leiboviz’s L-15 (Life Technologies). Dissociated cells were washed with PBS, and sorted into GFP positive (dorsal) and negative (ventral) cells using FACSAria III (BD Biosciences). Dead cells were detected by Propidium iodide (Life Technologies) and removed.
RNA-seq
Total RNA was extracted from sorted somite cells using RNeasy Mini kit (Qiagen). mRNA was enriched by poly-A capture and mRNA-seq libraries were generated using KAPA Stranded mRNA-seq Kit (KAPA Biosystems). Libraries were generated from two biological replicates, and sequenced using the Illumina HiSeq 1500 platform.
RNA-seq data processing
The sequenced reads were pre-processed to remove low-quality bases and adapter derived sequences using Trimmomatic v0.32 (Bolger, Lohse and Usadel, 2014), and aligned to the medaka reference genome (HdrR, ASM223467v1) by STAR (Dobin et al., 2013). Reads with mapping quality (MAPQ) larger than or equal to 20 were used for the further analyses.
ATAC-seq
ATAC-seq was performed as previously described (Buenrostro et al., 2013) with some modifications. Approximately 4,000 sorted somite cells were used for each experiment. After washing with PBS, cells were resuspended in 500 μl cold lysis buffer (10 mM Tris-HCl pH 7.4, 10 mM NaCl, 3 mM MgCl2, 0.1% Igepal CA-630), centrifuged for 10 min at 500 g, supernatant was removed. Tagmentation reaction was performed as described previously (Buenrostro et al., 2013) with Nextera Sample Preparation Kit (Illumina). After tagmented DNA was purified using MinElute kit (Qiagen), two sequential PCRs were performed to enrich small DNA fragments. First, a 9-cycle PCR was performed using indexed primers from Nextera Index Kit (Illumina) and KAPA HiFi HotStart ReadyMix (KAPA Biosystems), amplified DNA was size selected for a size less than 500 bp using AMPure XP beads (Beckman Coulter). A second 7-cycle PCR was performed using the same primers as for the first PCR. PCR product was purified by AMPure XP beads. Libraries were generated from two biological replicates, and sequenced using the Illumina HiSeq 1500 platform.
ChIPmentation
Yolk and head region were removed from 22 ss Tg(zic1:zic1-Myc/zic4:DsRed);Da embryos, followed by an incubation with 10 mg/ml pancreatin (Wako) at room temperature for 5-10 min. Epidermis and intestinal tissues were removed and somites were isolated from the neural tube. ChIP was performed as previously described with the following modifications (Nakamura et al., 2014). Isolated somites were fixed with 1% formaldehyde for 8 min at room temperature then quenched by adding glycine (200 mM final) and incubating on ice for 5 min. After washing with PBS, cell pellets were stored at −80 ℃. Approximately 1.8×106 cells were thawed on ice, suspended in lysis buffer (50 mM Tris-HCl (pH 8.0), 10 mM EDTA, 1% SDS, 20 mM Na-butyrate, complete protease inhibitors, 1 mM PMSF) and sonicated 10 times using a Sonifier (Branson) at power 5. Chromatin lysates were collected by centrifugation and diluted 10-fold with RIPA ChIP buffer (10 mM Tris-HCl (pH 8.0), 140 mM NaCl,1 mM EDTA, 0.5 mM EGTA, 1% Triton X-100, 0.1% SDS, 0.2% sodium deoxycholate, 20 mM Na-butyrate, complete protease inhibitors, 1 mM PMSF) followed by an incubation with antibody/protein A Dynabeads (Invitrogen) complex at 4 °C, overnight, while rotating. Immunoprecipitated samples were washed three times with RIPA buffer (10 mM Tris-HCl (pH 8.0), 140 mM NaCl,1 mM EDTA, 0.5 mM EGTA, 1% Triton X-100, 0.1% SDS, 0.2% sodium deoxycholate) and once with TE buffer. After the washing steps 150 μl of Tris-HCl was added.
Library preparation for ChIPmentation was performed as previously described (Schmidl et al., 2015) with the following modifications. 24 μl of Tagmentation reaction mix (10 mM Tris-HCl pH8.0, 5 mM MgCl2, 10%(v/v) N,N-dimethyl formamide) and 1 μl of Tagment DNA Enzyme from Nextera Sample Preparation Kit (Illumina) were added to the DNA-beads complex and incubated for 70 sec at 37 °C. 150 μl ice-cold RIPA buffer was added and incubated for 5 min on ice. The DNA-beads complex was washed with RIPA buffer and TE buffer, suspended in 50 μl lysis buffer and 3 μl of 5 M NaCl, and incubated at 65 °C, overnight. The sample was incubated for 2 h with 2 μl of 20 mg/ml ProteinasK (Roche), and subjected to AMPure XP beads (Beckman Coulter) purification. The library was amplified by 18-cycle PCR using indexed primers from Nextera Index Kit (Illumina) and KAPA HiFi HotStart ReadyMix (KAPA Biosystems). For the input chromatin, tagmentation reaction was performed after DNA purification. Libraries were generated from two biological replicates, and sequenced using the Illumina HiSeq 1500 platform.
ChIPmentation and ATAC-seq data processing
The sequenced reads were pre-processed to remove low-quality bases and adapter derived sequences using Trimmomatic v0.32 (Bolger, Lohse and Usadel, 2014) and aligned to the medaka reference genome (HdrR, ASM223467v1) by BWA (Li and Durbin, 2009). Reads with mapping quality (MAPQ) larger than or equal to 20 were used for the further analyses. MACS2 (version 2.1.1.20160309) (Zhang et al., 2008) was used to call peaks and generate signals per million reads tracks using following options; ChIPmentation: -g 600000000 -B -- SPMR --keep-dup 2, ATAC-seq: --nomodel --extsize 200 --shift -100 -g 600000000 -q 0.01 - B --SPMR.
For ChIPmentation, peak regions called by two biological replicates were used as reliable peaks.
Motif analyses of Zic1 ChIPmentation peaks
Motifs enriched at reliable ChIPmentation peaks were analyzed by findMotifsGenome command of HOMER (Heinz et al., 2010) using default parameters.
Identification of Zic1 target genes
Differentially expressed genes were identified using DESeq2 (padj < 0.01). Each reliable ChIP peak was associated to the nearest TSS, and the gene was defined as Zic1-target gene if the distance between the peak and the TSS was closer than 50 kb.
Gene ontology and pathway analyses
The gene ontology enrichment analyses and pathway enrichment analyses were performed using the Gene Ontology Resource (Ashburner et al., 2000; Gene Ontology Consortium, 2021).
RT-PCR of cDNA generated from embryonic tails
To investigate the gene expression in tails of embryos, tails were dissected anterior from the first somite. Ten tails were pooled together and RNA was isolated using Isogen (Nippon Gene). RNA was purified using the RNeasy Mini kit (Qiagen) and reverse transcribed to cDNA using the Super Script III Kit (Invitrogen). RT-PCR was performed using the Thunderbird Sybr qPCR Mix (Toyobo) following manufacturer’s instructions and run in the Agilent Mx3000P qPCR System (Agilent). Normalization of relative quantities was performed against gapdh expression, followed by analysis with excel and RStudio.
Supplementary table 1: Gene expression profiles and distances to nearest Zic1 ChIP peak
Supplementary table 2: Full list of GO terms enriched in dorsal-high Zic1 target genes
Supplementary table 3: Full list of GO terms enriched in dorsal-low Zic1 target genes
Acknowledgments
We thank the members of the Takeda laboratory for constructive feedback and discussions on the project. We are greatful for Y. Yamagichi and M. Funato for fish husbandry. This work was supported by Japan Society for the Promotion of Science (JSPS) KAKENHI Grant Numbers JP15H05859 (H.T.), JP19K23741 (T.K.) and JP18K14620 (R.N.) and Japan Science and Technology Agency CREST Grant Number JPMJCR13W3 (H.T.).