Tbx6, Mesp-b and Ripply1 regulate the onset of skeletal myogenesis in zebrafish

Skeletal musculature in the vertebrate embryo originates in the paraxial mesoderm, which develops on either side of the neural tube and notochord, and subsequently segments into somites. Segmentation of paraxial mesoderm into somites and myogenesis within each somite are sequentially initiated, starting posterior to the head and progressing towards the tail bud, which simultaneously produces more paraxial mesoderm to elongate the body axis (Buckingham and Vincent, 2009; Oates et al., 2012). How the onset of myogenesis is coordinated with the formation of somites is not yet understood.

Expression of the paired domain transcription factor pax3 marks the entry of paraxial mesoderm cells into the myogenic program before their incorporation into a somite (Kusakabe and Kuratani, 2005; Buckingham and Relaix, 2007). Subsequently, Pax3-expressing cells give rise to two spatially segregated myogenic tissues. The dermomyotome, a transient epithelial structure containing proliferative muscle progenitors, is located at the lateral somite surface; the myotome, consisting of differentiated muscle fibers, lies underneath/medial to the dermomyotome. Muscle progenitors retain pax3 expression and upregulate pax7, whereas differentiating muscle fibers downregulate pax3 and express myogenic regulatory factors like myoD and myf5 (Buckingham and Rigby, 2014).

The zebrafish myotome consists of spatially segregated slow and fast muscle fibers, which are specified in the pre-somitic paraxial mesoderm. First, Hedgehog signaling from the notochord and ventral neural tube induces slow muscle fibers in the adjacent medial paraxial mesoderm (Blagden et al., 1997; Du et al., 1997). Then, immediately before somite boundary formation, Hedgehog and Fgf8 signaling promote differentiation of fast fibers in the pax3-expressing lateral paraxial mesoderm, specifically in the posterior half of the future somite (Groves et al., 2005; Hammond et al., 2007). After somite formation, pax3-expressing cells in the anterior half of the lateral somite upregulate pax7 and translocate to the surface of the developing myotome (Hollway et al., 2007; Stellabotte et al., 2007). It is unknown how the onset of myoD expression becomes restricted to the posterior half of the lateral somite. Similarly, it is unknown what promotes the development of Pax3⁺/Pax7⁺ dermomyotome cells in the anterior half of the lateral somite.

The establishment of anterior and posterior half-somite domains has been studied in the context of segmentation, because the polarized/alternating expression of genes in the pre-somitic mesoderm precedes somite boundary formation (Dahmann et al., 2011). The patterning process begins in the tailbud, where single cells initiate oscillating expression of members of the Her/Hes family of bHLH transcriptional repressors, which are subsequently synchronized among neighboring cells by Delta-Notch signaling (Pourquié, 2011; Oates et al., 2012). Before morphological boundary formation, these oscillations, known as the segmentation clock, are translated into non-oscillating expression of genes specific to either prospective anterior or posterior half-somites (Saga, 2012).

tbx6/fused somites encodes a T-box transcription factor required for the maintenance and readout of these oscillations in the anterior pre-somitic mesoderm, and for the formation of somite boundaries in zebrafish (van Eeden et al., 1996; Nikaido et al., 2002; Oates et al., 2005; Brend and Holley, 2009). We recently discovered that Tbx6 is also required for the development of a subpopulation of dermomyotome cells (Windner et al., 2012). tbx6 mutants ectopically upregulate myoD in the anterior half-somite, which correlates with a gain in fast fiber nuclei and a loss of the central domain of the Pax3⁺/Pax7⁺ dermomyotome, whereas peripheral (dorsal and ventral) dermomyotome develops normally. These findings suggest that Tbx6 inhibits premature differentiation of dermomyotome cells, and provide an unexpected link between myogenesis and somite boundary formation.

In mouse, expression of the bHLH transcription factor Mesp2 is induced by the combination of Tbx6 and spatially restricted Notch signaling, specifically in the anterior half of the future somite (Yasuhiko et al., 2006). Mesp2 promotes morphological somite boundary formation via activation of Eph/Ephrin signaling and activates co-repressors of the Ripply family, which terminate the patterning network by negatively regulating Tbx6 function (Nakajima et al., 2006; Takahashi et al., 2010). mesp and ripply genes are expressed in a Tbx6-dependent manner and also regulate anterior-posterior somite polarity in zebrafish (Sawada et al., 2000; Kawamura et al., 2005); however, it is unknown whether they play a role in myogenesis.

Here, we analyze the Tbx6–Mesp-b–Ripply1 regulatory network in zebrafish and reveal its role at the onset of skeletal myogenesis. We show that both mesp-b paralogs and ripply1 are direct targets of Tbx6. Mesp-b is necessary and sufficient for development of the central domain of the Pax3⁺/Pax7⁺ dermomyotome, similar to Tbx6. Mesp-b inhibits myoD and myf5 expression, and by inducing the dermomyotome marker meox1 in the anterior half of future somites. By contrast, Ripply1 negatively regulates mesp-b expression by eliminating Tbx6 protein. This negative-feedback loop, which is also promoted by Mesp-b, is required for upregulation of pax3/7 as well as myoD and myf5, suggesting a Tbx6-dependent inhibition of myogenic gene expression. Our results provide an explanation for the spatially restricted development of myogenic progenitors within the somites, and for the temporal coordination of myogenesis with the formation of somite boundaries.

Tbx6 is required for the expression of mesp-b and ripply1 in the paraxial mesoderm during somite formation, and for the specification of the central Pax3⁺/Pax7⁺ dermomyotome (Sawada et al., 2000; Kawamura et al., 2005; Windner et al., 2012). At the 10-somite stage (14 h post fertilization), tbx6 mRNA was highly expressed throughout the anterior pre-somitic mesoderm, and at decreasing levels in the recently formed somites SI-SIII, whereas, surprisingly, Tbx6 protein was downregulated prior to somite formation (Fig. 1A,B). Specifically, Tbx6 protein was reduced in the posterior proportion of S-I and virtually undetectable throughout the S-0 domain (antibody specificity shown in supplementary material Fig. S1). mesp-ba and ripply1 mRNA spatially overlapped with Tbx6 protein (Fig. 1C,D): mesp-ba was expressed in two precise stripes in the future anterior half-somite in S-II and S-I; ripply1 expression was initiated in the future posterior half-somite in S-I, highly upregulated throughout the anterior and posterior of S-0 and SI, and gradually downregulated and restricted to the anterior half in the maturing somites. myoD was co-expressed with Tbx6 in the developing slow muscle fibers along the medial somite surface, but did not overlap with Tbx6 in the lateral paraxial mesoderm (Fig. 1E).

We carried out chromatin immunoprecipitation, followed by high throughput sequencing (ChIP-seq) of DNA associated with Tbx6. Because the Tbx6 antibody did not work for ChIP (data not shown), we used a transgenic zebrafish line with heat shock-inducible Myc-tagged Tbx6, Tg(hsp70l:tbx6^myc)^v8, that can rescue the tbx6 mutant phenotype (Windner et al., 2012), and immunoprecipitated with an anti-Myc antibody 1 h after transgene induction. We identified regions of Tbx6 binding in close proximity to the mesp-ba, mesp-bb and ripply1 genes (Fig. 1F; supplementary material Table S1), many of which match those found computationally using the T-box consensus motif (Cutty et al., 2012). De novo motif discovery showed that the Tbx6^Myc in vivo binding sequences included the consensus T-box previously identified for T-box proteins in zebrafish and mouse (Kispert and Hermann, 1993; Garnett et al., 2009) (Fig. 1G; supplementary material Table S2). The binding site near the transcriptional start site of mesp-ba has been confirmed previously by Kawamura et al. (2008); we confirmed the binding site ∼6700 bp upstream of the transcription start site by ChIP-PCR (supplementary material Fig. S2).

Combined, these data suggest that Tbx6 directly regulates the mesp-b and ripply1 genes in vivo. The expression of mesp-b and ripply1 precedes the development of dermomyotome cells in the anterior half-somite and fast muscle fibers in the posterior half-somite (Fig. 1H).

To test whether mesp-b regulates dermomyotome development in the anterior half-somite, we knocked down Mesp-b function using morpholinos (MOs) against both zebrafish paralogs (Mesp-ba and Mesp-bb MOs, henceforth referred to as Mesp-b MO). Knockdown of either Mesp-ba or Mesp-bb alone did not result in a dermomyotome phenotype (data not shown), suggesting functional redundancy between mesp-b paralogs. We found that translation-blocking and splice-blocking Mesp-b MOs produced similar phenotypes, and validated the functionality of the splice-blockers by RT-PCR (supplementary material Fig. S3). As previously reported, ripply1 mRNA was expressed in the pre-somitic mesoderm in Mesp-b morphants; however, expression was not maintained after somite boundary formation (Kawamura et al., 2005) (Fig. 2A).

The dermomyotome marker meox1 and the myotome marker myoD were similarly expressed in pre-somitic mesoderm in control and Mesp-b MO-injected embryos, but showed complementary changes within the somites: in newly formed somites, meox1 was reduced and myoD expanded to the anterior half-somite, whereas in more mature somites, neither mRNA was maintained in the lateral somite (Fig. 2B,C). myoD expression in slow muscle progenitors along the medial somite surface was not affected in Mesp-b morphants. The transient upregulation of myoD in anterior and posterior half-somites resembled the pattern of myoD expression in tbx6 mutants (van Eeden et al., 1996; Windner et al., 2012), and, together with a downregulation of meox1, suggests a loss of dermomyotome and an expansion of fast muscle development.

Indeed, the number of Pax7⁺ dermomyotome cells was reduced in Mesp-b morphants, specifically in the central domain of the trunk, identical to the dermomyotome phenotype in tbx6 mutants at the 24-h stage (Fig. 2D-F). The loss of central, but not peripheral, dermomyotome in tbx6 mutant and Mesp-b knockdown embryos was consistent with the restriction of mesp-b expression to the central proportion of the paraxial mesoderm (supplementary material Fig. S4). We conclude that, downstream of Tbx6, Mesp-b is required for central dermomyotome development.

To test whether Mesp-b can rescue central dermomyotome development in the absence of tbx6, we created transgenic fish with a heat shock-inducible promoter driving ubiquitous expression of Myc-tagged Mesp-ba, Tg(hsp70l:mesp-ba^myc). Transgene expression in tbx6 mutants during segmentation stages caused an increase in the number of Pax7⁺ central dermomyotome cells at the 24-h stage, and rescued medio-lateral patterning of slow and fast muscle fibers, which is disrupted in tbx6 mutants (Fig. 2G-K). At the 36-h stage, the number of Pax7⁺ dermomyotome cells and the size of the myotome in transgenic tbx6 mutants were increased when compared with non-transgenic siblings, and resembled those in wild-type embryos (Fig. 2L-N). Thus, ubiquitous Mesp-ba expression is sufficient to rescue myogenic phenotypes found in tbx6 mutants (Windner et al., 2012).

tbx6 mutants do not form somite boundaries (fused somites phenotype) (van Eeden et al., 1996). By contrast, Mesp-b morphants formed irregular somites (Fig. 2A-C) and showed clearly segmented myotomes, as indicated by expression of the myogenic enhancer factor Mef2 and myosin heavy chain protein (MF20) in differentiated muscle fibers (Fig. 2D-F). Conversely, ubiquitous Mesp-ba expression did not fully rescue somite boundary formation in tbx6 mutants, but resulted in the formation of partial and irregularly spaced myotome boundaries (Fig. 2M).

Combined, these knockdown and rescue experiments suggest that Mesp-b is necessary and sufficient for central dermomyotome development downstream of Tbx6, but not necessary/sufficient for somite boundary formation.

To further examine the role of Mesp-b in central dermomyotome development, we analyzed the immediate effects of ubiquitous Mesp-ba^Myc expression on genes characterizing the onset of dermomyotome development (meox1, pax3) and myogenic differentiation (myoD, myf5) in wild-type embryos. Transient overexpression of Mesp-ba led to the opposite effect of Mesp-b knockdown; in Tg(hsp70l:mesp-ba^myc) embryos 15 min after heat shock, the expression of myoD and myf5 in the lateral paraxial mesoderm was downregulated, and the expression of meox1 was upregulated (Fig. 3A-C; supplementary material Fig. S5). The rapid changes in mRNA levels occurred similarly in all somites, independent of their maturation state, and did not affect the developing slow muscle fibers in the medial paraxial mesoderm. This suggests that specific effects on myogenic marker genes are immediately downstream of Mesp-b.

pax3 is expressed in myogenic progenitors in future anterior and posterior somite domains before being restricted to the developing dermomyotome (Groves et al., 2005). We could not detect differences in pax3 mRNA levels between transgenic embryos and non-transgenic siblings at this stage (Fig. 3D), indicating that Mesp-ba does not directly regulate pax3 expression, and that the total population of dermomyotome and fast muscle fibers was not altered by ubiquitous Mesp-ba expression.

Six hours post heat shock (24-somite stage), the dorsal-to-ventral extent of myoD expression was reduced in Tg(hsp70l:mesp-ba^myc) embryos, whereas meox1 was still upregulated, particularly in somites formed during transgene expression (Fig. 3E,F). Smaller myotomes in Tg(hsp70l:mesp-ba^myc) embryos showed a deficit specifically in differentiated fast muscle fibers, which expressed myosin heavy chain protein (MF20) and low levels of Mef2 in non-transgenic siblings (Fig. 3G,H). By contrast, slow muscle fiber differentiation was comparable in transgenic and non-transgenic embryos.

Nine hours post heat shock (24-h stage), Tg(hsp70l:mesp-ba^myc) embryos had an increased number of mature, Pax7⁺ dermomyotome cells (Fig. 3I,J), which spatially correlated with the decrease in myoD expression 6 h after heat shock (Fig. 3E). However, myotome sizes were similar in transgenic and non-transgenic embryos at this stage. The recovery of fast muscle indicates that dermomyotome-derived myogenesis is not disrupted in Tg(hsp70l:mesp-ba^myc) embryos and compensates for the initial muscle deficit.

Together, these results show that transient ubiquitous expression of Mesp-ba causes immediate changes in dermomyotome and myotome marker gene expression, which correlate with an inhibition of fast muscle development and an increase in the number of Pax7⁺ dermomyotome cells.

To determine whether Ripply1 negatively regulates Tbx6 protein, as seen in the context of somite formation in mouse (Takahashi et al., 2010), we knocked down Ripply1 function using MOs. As previously shown (Kawamura et al., 2005), knockdown of Ripply1 caused a loss of somite boundary formation, similar to the segmentation phenotype seen in tbx6 mutants.

In Ripply1 morphants during segmentation stages, tbx6 mRNA and protein were present at high levels throughout the paraxial mesoderm (Fig. 4A,B). Reflecting the expression of Tbx6 in Ripply1 morphants, both mesp-b paralogs were expressed in one to three stripes in the presomitic mesoderm (PSM), similar to their expression in control embryos, but remained upregulated in the anterior paraxial mesoderm (Fig. 4C,D). These expression patterns are consistent with previous work (Kawamura et al., 2005), and suggest that Ripply1 restricts the expression of Tbx6 to the pre-somitic mesoderm, where spatially restricted co-regulators further confine mesp-b expression to the future anterior half-somite.

Ubiquitous expression of Ripply1 in Tg(hsp70l:ripply1^myc) embryos eliminated Tbx6 protein in the pre-somitic mesoderm, while only slightly reducing tbx6 mRNA (Fig. 4E,F), thus supporting a role for Ripply1 in regulating Tbx6 at the protein level. Consistent with a lack of Tbx6 protein, mesp-ba expression was downregulated and myoD mRNA was upregulated in transgenic embryos (Fig. 4G; supplementary material Fig. S6A). To independently test whether Ripply1 promotes degradation of Tbx6 protein we injected either control or Ripply1 MOs into Tg(hsp70l:tbx6^myc)^v8 embryos, heat-shocked for 30 min during segmentation stages and fixed 15 min post heat shock (supplementary material Fig. S6B). In Ripply1 morphants, endogenous Tbx6 protein was present in paraxial mesoderm cells along the entire anterior-posterior axis; transgene expression led to additional Tbx6 protein in the notochord and surface ectoderm. Control MO-injected embryos also show high levels of Tbx6 expression in the notochord and surface ectoderm, but not in the paraxial mesoderm. Thus, Tbx6 protein is immediately removed specifically from the paraxial mesoderm in a Ripply1-dependent manner.

Ubiquitous Mesp-ba expression in Tg(hsp70l:mesp-ba^myc) embryos increased ripply1 mRNA levels in wild type and induced ripply1 expression in tbx6 mutants (Fig. 4H,I), suggesting that Tbx6 regulation of ripply1 is also mediated by Mesp-b in zebrafish, as in mouse (Morimoto et al., 2007).

In Tg(hsp70l:mesp-ba^myc) embryos after heat shock, Mesp-ba^Myc protein levels remained high throughout the paraxial mesoderm for several hours after transgene induction (Fig. 3H). However, mesp-ba mRNA was highly upregulated only in the intermediate mesoderm and neural tube of transgenic embryos, whereas mRNA levels in the paraxial mesoderm were reduced, including expression from the endogenous mesp-ba gene (Fig. 4J). Conversely, knockdown of Mesp-b function caused an increase in mesp-b mRNA expression (Fig. 4K).

Together, these data suggest that Mesp-b downregulates its own mRNA, possibly via induction of Ripply1 and consequent Tbx6 protein degradation.

In contrast to transient upregulation of Mesp-ba in Tg(hsp70l:mesp-ba^myc) embryos, knockdown of Ripply1 causes continuously high levels of mesp-b expression (Fig. 3 and Fig. 4C,D) (Kawamura et al., 2005). We hypothesized that this would amplify the myogenic phenotypes seen in transgenic embryos and further increase the number of Pax7⁺ dermomyotome cells.

Ripply1 knockdown did not have an effect in tbx6 mutants (supplementary material Fig. S7A), indicating that Ripply1 acts mainly through elimination of Tbx6 protein. Similar to Tg(hsp70l:mesp-ba^myc) embryos after heat shock, Ripply1 knockdown reduced expression of myoD in the lateral paraxial mesoderm (Fig. 3A; supplementary material Fig. S7B). Surprisingly, knockdown of Ripply1, although causing high levels of Tbx6, mesp-b and meox1, resulted in downregulation of pax3 mRNA (Fig. 5A,B; supplementary material Fig. S7C). In wild type, pax3 is expressed at the beginning of both dermomyotome and fast muscle development (Groves et al., 2005); thus, a lack of pax3 expression indicates a general inhibition of myogenesis in the lateral paraxial mesoderm.

We did not observe Pax7⁺ dermomyotome cells or differentiated fast muscle fibers in the presence of Tbx6 protein (Fig. 5C,D; supplementary material Fig. S7D). The elimination of Tbx6 protein in Ripply1 morphants took about 8 h, compared with 30 min in the presence of Ripply1, but similarly occurred in an anterior-to-posterior progression. At the 22-somite stage, Tbx6 protein had disappeared from the anterior trunk of Ripply1 morphants, but was still present in the posterior half of the embryo (Fig. 5C). At the same time, the first mature Pax7⁺ dermomyotome cells appeared in the anterior trunk (Fig. 5D). In the first ∼6 somites, which develop independently of Tbx6 function (Windner et al., 2012), tbx6, mesp-b and meox1 expression remained elevated for a longer period, and Pax7 expression was further delayed (Fig. 4A,C,D; data not shown).

At the 24-h stage, ectopic tbx6 was restricted to the most recently formed somites in the tail, whereas pax3 and pax7 were highly upregulated throughout the anterior and posterior trunk in Ripply1 morphants (Fig. 5E-G). Immunolabeling confirmed a dramatic increase in the number of Pax7⁺ dermomyotome cells, most prominent in the central domain of the trunk (Fig. 5H,I). Fast muscle differentiation was completely inhibited prior to upregulation of Pax7, whereas slow muscle fibers differentiated but remained in a position adjacent to the notochord (Fig. 5J,K; supplementary material Fig. S7E). This indicates that knockdown of Ripply1 leads to the induction of dermomyotome development throughout the lateral paraxial mesoderm, at the expense of fast muscle formation, and raises the possibility that, in the absence of Ripply1, fast muscle fibers are exclusively derived from Pax7⁺ dermomyotome cells.

Combined, these results suggest that continuous expression of Tbx6 and/or Tbx6-dependent genes inhibits myogenesis in the lateral paraxial mesoderm. The Ripply1-dependent elimination of Tbx6 protein prior to somite formation restricts dermomyotome development to anterior half-somites, which facilitates fast muscle differentiation in posterior half-somites.

We have explored a gene regulatory network in zebrafish that regulates some of the earliest steps in embryonic myogenesis. Our results significantly extend previous work on the interactions between tbx6, mesp-b and ripply1, and introduce novel roles for Mesp-b and Ripply1 in the specification and maturation of dermomyotome and myotome, respectively. We propose a model that explains how the readout of genes associated with segmentation leads to the spatially restricted specification of myogenic progenitors within the somites, and to the temporal coordination of myogenesis with the formation of somite boundaries (Fig. 6).

Our data suggest that Tbx6 directly regulates the expression of both mesp-b paralogs in vivo (see also Kawamura et al., 2008; Cutty et al., 2012). Binding of Tbx6 to the mesp-b promoter and the functional relevance of this interaction have been demonstrated in other vertebrate models, including medaka (Terasaki et al., 2006), Xenopus (Tazumi et al., 2008) and mouse (Yasuhiko et al., 2006, 2008), suggesting that this interaction is conserved. Mesp genes are expressed in one to three precise bands prior to somite boundary formation (Saga et al., 1997; Sawada et al., 2000). The restriction of mesp-b expression to the anterior half of future somites in mouse depends on the spatially restricted overlap of Tbx6 with active Notch signaling (Saga, 2012). Oscillating Notch activity is a fundamental component of the segmentation clock, also in zebrafish (Oates et al., 2012). Whether the restricted expression of zebrafish mesp-b represents a direct output of the segmentation clock is not known.

We provide the first evidence that Tbx6 directly regulates the expression of the transcriptional co-repressor ripply1. We also demonstrate that Mesp-ba can induce ripply1 in the absence of Tbx6, which is in accordance with studies showing direct activation of the orthologous mouse Ripply2 by Mesp2 (Morimoto et al., 2007). Further, we show that Mesp-b is not required for the expression of ripply1 in the anterior PSM but is required for the maintenance of ripply1 expression in the somites. Together, these results indicate that Tbx6 induces ripply1 expression in the pre-somitic mesoderm, whereas Mesp-b promotes continued expression within the somites. Our data do not indicate whether Mesp-ba regulation of ripply1 in zebrafish is direct or indirect.

We show that Ripply1 is necessary and sufficient for the elimination of Tbx6 protein prior to somite formation in zebrafish. Whether this is by ubiquitin-dependent protein degradation, as it is in mouse, remains to be determined (Oginuma et al., 2008; Takahashi et al., 2010). Ripply proteins recruit Groucho/TLE co-repressors to repress Tbx-mediated transcription of mesp genes in zebrafish and Xenopus (Kondow et al., 2007; Kawamura et al., 2008). Thus, Ripply family members might suppress Tbx6 function in multiple ways. The Ripply1-dependent inhibition of Tbx6 function restricts the activation of mesp-b genes to the pre-somitic mesoderm; additional negative feedback might be involved in the rapid downregulation of mesp mRNA before somite boundary formation.

The Tbx6-dependent expression of mesp-b genes precedes the upregulation of the dermomyotome markers pax3/7 in space and time (Fig. 6A,B). Pax3 is expressed throughout the lateral paraxial mesoderm immediately before somite boundary formation. The developing dermomyotome cells in the anterior half of the lateral somite maintain pax3, and subsequently express pax7, whereas downregulation of pax3 and activation of myoD and myf5 leads to differentiation of fast muscle fibers (Groves et al., 2005; Devoto et al., 2006). Signaling molecules, including Fgf8, regulate the onset of myoD expression, but not its spatial restriction to the posterior half of the lateral somite (Groves et al., 2005). We propose that mesp-b genes promote dermomyotome development in the anterior half-somite.

tbx6 mutants, which do not express mesp-b, upregulate myoD in the lateral paraxial mesoderm without spatial restriction (van Eeden et al., 1996). This phenotype correlates with an increase in fast fiber nuclei and a decrease in Pax3⁺/Pax7⁺ dermomyotome cells, whereas the total number of myogenic nuclei remains the same, thus suggesting ectopic development of fast muscle at the expense of dermomyotome cells in tbx6 mutants (Windner et al., 2012). Here, we show that mesp-b is not expressed in the dorsal and ventral periphery of the somite, and that mesp-b morphants recapitulate the specific loss of central dermomyotome seen in tbx6 mutants, whereas Mesp-ba overexpression rescues the dermomyotome and myotome phenotypes in tbx6 mutants. Thus, mesp-b gene expression regulates the establishment of the central dermomyotome cell population downstream of Tbx6. Normal development of peripheral dermomyotome in tbx6 mutants and Mesp-b morphants suggests different genetic regulation; whether peripheral dermomyotome cells have a different developmental origin remains to be determined.

We demonstrate that Mesp-b expression immediately upregulates the dermomyotome marker meox1 and inhibits the expression of myoD and myf5, specifically in the lateral paraxial mesoderm. Genome-wide transcriptional analysis after Mesp1 induction in mouse cell culture has shown that Mesp1 rapidly activates and represses specific sets of target genes, and thus acts as a key regulatory switch during cardiovascular progenitor cell specification (Bondue et al., 2008). Similarly, we suggest that mesp-b genes in the paraxial mesoderm act as an intrinsic determinant for the establishment of the central dermomyotome cell population by activating dermomyotome genes, while at the same time repressing genes promoting the alternative cell fate, the fast myotome. Thus, ubiquitous expression of mesp-b in Ripply1 morphants leads to widespread dermomyotome development and completely inhibits fast muscle differentiation throughout the paraxial mesoderm. However, we show that mesp-b genes do not immediately promote pax3/7 expression, which is consistent with in vivo analyses placing pax genes downstream of meox1 gene function (Mankoo et al., 2003), but could also indicate that mesp-b and pax3 genes synergistically regulate dermomyotome development.

Taken together, our data provide an explanation for the initial pattern of dermomyotome and fast muscle progenitors in the zebrafish somites. Whether the Mesp-dependent regulation of myogenesis is conserved in vertebrates remains to be determined. In mouse, it has been suggested that mesp genes regulate the ability of cells to contribute to the dermomyotome by affecting the process of epithelialization (Takahashi et al., 2005). Thus, it is possible that Mesp target genes effect dermomyotome development in various ways, including morphological changes in the paraxial mesoderm.

Slow muscle fibers are induced by Hedgehog signaling from notochord and ventral neural tube in the medial paraxial mesoderm and start differentiating prior to and independent of Tbx6 expression (van Eeden et al., 1996; Blagden et al., 1997; Du et al., 1997; Windner et al., 2012). Slow fibers are not displaced to the lateral somite surface in Ripply1 morphants, in which dermomyotome maturation is delayed, nor in tbx6 mutants, in which dermomyotome differentiates prematurely into fast muscle. Thus, the translocation of slow fibers depends on properly timed specification and maturation of dermomyotome and fast fibers in the lateral somite. Although Mesp-b expression can rescue the slow fiber phenotype in tbx6 mutants, it remains to be determined whether Mesp-b affects myotome patterning directly by regulating muscle progenitor translocation, and/or indirectly by regulating the balance between dermomyotome and fast muscle, which could subsequently shape the somite.

Mesp gene expression reliably marks the position of future somite boundaries in the anterior pre-somitic mesoderm and is implicated in regulating somite boundary formation via induction of components of the Eph signaling pathway (Nakajima et al., 2006; Dahmann et al., 2011). The role of Mesp genes in segmentation is well established in mouse (Saga, 2012); however, only mesp-1/mesp-2 double knockout leads to the complete loss of morphological somite boundaries (Oginuma et al., 2008). We find that irregular somite boundaries persist following knockdown of mesp-b in zebrafish (see also Akiyama et al., 2014). We suggest that this might reflect a functional redundancy with mesp-a genes, which show similar expression patterns and regulation (Sawada et al., 2000; Cutty et al., 2012). Conversely, transient ubiquitous expression of Mesp-ba is insufficient to completely rescue somite formation in tbx6 mutants, suggesting that segmental/polarized expression of mesp and/or other genes downstream of Tbx6 is required for proper somite boundary formation. The identification of Mesp-b target genes will provide insights into whether the different cellular processes attributed to mesp genes result from one or multiple pathways of gene activation.

Ripply1 is expressed in the future anterior and posterior half-somites prior to boundary formation and negatively regulates Tbx6 and mesp-b expression (Kawamura et al., 2005; this study). Our results show that Tbx6 protein has to be removed for the expression of pax3/7 and myoD in the lateral paraxial mesoderm, indicating that Tbx6 and/or Tbx6-dependent genes inhibit maturation of myogenic cells. Downstream of Tbx6, Mesp-ba promotes dermomyotome development but does not immediately promote pax3 expression, raising the possibility that mesp genes are among the Tbx6 targets that have to be downregulated for dermomyotome maturation. A similar mechanism has been shown in cell culture studies, in which transient expression of Mesp1 promotes the differentiation of cardiac progenitors, whereas continuous expression inhibits it (Bondue et al., 2008). We suggest that, by promoting Tbx6 protein degradation, Ripply1 triggers maturation of myogenic progenitors in the lateral paraxial mesoderm (Fig. 6).

In the context of somite boundary formation, the region in the anterior paraxial mesoderm where pre-somitic mesoderm markers are downregulated and somite-specific markers are induced has been termed the determination front/wavefront (Dubrulle et al., 2001). The molecular changes occurring at the front determine the position of future somite boundaries, and include arrest of the segmentation clock and induction of stable expression of genes characterizing future anterior and posterior half-somites (Holley, 2007; Aulehla and Pourquié, 2010; Oates et al., 2012). The appropriate axial position of the front is influenced by signaling gradients: Fgf and Wnt signaling from the tailbud and retinoic acid signaling from the maturing somites (Aulehla and Pourquié, 2010). In zebrafish, ripply1 expression is directly promoted by retinoic acid (Moreno et al., 2008), and thus constitutes a direct readout of the determination front. We propose that the Ripply1-dependent degradation of Tbx6 protein participates in the developmental switch occurring at the determination front, triggering the morphological segmentation of the paraxial mesoderm, as well as the formation of dermomyotome and myotome.

Although the major components of the determination front are conserved across different vertebrate models, some of the molecular mechanisms seem to be species-specific (Moreno et al., 2008). Future work will determine whether the inhibition of Tbx6 is a conserved feature of the front and coordinates the onset of myogenesis with somite boundary formation also in other vertebrates.

All animal experiments were undertaken according to protocols approved by the Wesleyan University Animal Care and Use Committee, assurance number A3956-01.

We used wild-type zebrafish (Danio rerio) and the ti1 or te314a allele of fss/tbx6 mutants, each of which behaves as a null (Nikaido et al., 2002). We detected no differences between wild type and heterozygote tbx6 mutants, and for readability refer to heterozygous embryos as wild type. We used the Tg(hsp70l:tbx6^myc)^v8 line and generated stable transgenic lines, using full-length mesp-ba, mesp-bb (Cutty et al., 2012) and ripply1 (Kawamura et al., 2005) cDNAs, and gateway cloning as previously described (Kwan et al., 2007; Villefranc et al., 2007; Windner et al., 2012). All plasmids were verified by sequencing. Heat shocks were performed as previously described (Windner et al., 2012), for 1 h at 37°C, unless stated otherwise. Transgenics were identified using anti-cMyc antibody labeling or PCR genotyping. Embryos were cared for using standard procedures (Westerfield, 1995).

8 μg of a Tbx_6327-545 peptide fused to GST was injected into Balb/c mice; sera were screened via ELISA. Each antiserum with a positive signal was further tested for tissue-specific binding in 15-somite stage wild-type and tbx6 mutant embryos. Hybridoma cell lines were produced from one mouse; antibodies were purified from the supernatants. The antibody with highest signal-to-noise ratio was used for experiments (Clone: A83-1, IgG1).

MOs (GeneTools) were resuspended to obtain 1 µM stock solutions, and diluted 1:5 before co-injection with p53 MO into the yolk at the 1-cell stage. MO sequences are included in supplementary material Table S3.

For RT-PCR, embryos were injected with Mesp-ba or Mesp-bb splice-blocking and p53 MOs; total RNA was extracted from ∼50 embryos at bud stage with TRIzol reagent (Life Technologies), with additional RQ1 DNase I treatment (Promega). cDNA was produced using GoScript Reverse Transcriptase (Promega), primers used for PCR are shown in supplementary material Table S3.

Fixations, whole-mount RNA in situ hybridization and immunohistochemistry were carried out as previously described (Barresi et al., 2000; Patterson et al., 2010; Windner et al., 2012). Antibodies and RNA probes used are listed in supplementary material Table S3. Images of immunolabeled whole-mounts were taken with a Zeiss LSM510 confocal microscope, and confocal z-stacks were flattened. Cryostat cross-sections and in situ hybridization samples were imaged on a Zeiss Axioplan compound microscope, and on a Leica stereomicroscope, respectively. Embryos at 20-somite (19-h) stage and older are shown in lateral view, dorsal up, anterior left; younger stages are in dorsal view, anterior up.

Tg(hsp70l:tbx6^myc) and wild-type embryos were heat-shocked at the 10-somite stage, and fixed 1 h later. ChIP for 2000 embryos per replicate, using 10 µg goat anti-Myc tag antibody (Abcam, ab9132), was carried out as in Wardle et al. (2006), with the exception that purified immunoprecipitated DNA was directly prepared for Illumina sequencing using the TruSeq DNA sample prep kit, with DNA adapters diluted 1:100. Paired-end sequencing of ChIP and their matched input samples were sequenced on an Illumina HiSeq. Duplicate ChIP-seq experiments were performed for Tg(hsp70l:tbx6^myc) and a single control experiment in wild-type embryos.

To define Tbx6^Myc binding sites, paired end sequence fragments with unique outer coordinates were mapped to the danRer7 genome build using Bowtie (Langmead et al., 2009) with the following parameters: -5 3 -3 30 –n 3 –y –k 2 –m 2 –best. Peaks were called for each ChIP sample relative to their paired input sample using MACS (Zhang et al., 2008) with the following parameters: --gsize 1.3e+9 –tsize 67 -mfold 8 –pval 5e-10 for Tbx6^Myc ChIP samples and –mfold 4 for the wild-type Myc tag control ChIP. Reported peaks are present in both Tg(hsp70l:tbx6^myc) ChIP-seq experiment and absent in the wild-type control. Motif analysis was performed using Weeder v1.4.2 with default parameters (Pavesi et al., 2006), except that both strands were processed; Tbx6^Myc binding peaks which were significant in both experiments were used. All ChIP-seq data are stored in NCBI Gene Expression Omnibus under the accession number GSE57332.

For identification of potential T-boxes within genomic regions represented in Fig. 1F, a modified version of the Perl scripts in the TFBS suite was used (Lenhard and Wasserman, 2002) to identify matches to the position frequency matrix generated by Weeder v1.4.2 (see above).

The ripply1 cDNA was kindly provided by Shinji Takada. The Tbx antibody project was started by Lola Bajard.