In the embryonic zebrafish, skeletal muscle fibres are formed from muscle progenitors in the paraxial mesoderm. The embryonic myotome is mostly constituted of fast-twitch-specific fibres, which are formed from a fast-specific progenitor cell pool. The most lateral fraction of the fast domain in the myotome of zebrafish embryos derives from the Pax7-positive dermomyotome-like cells. In this study, we show that two genes, belonging to the sine oculus class 1 (six1) genes (six1a and six1b), are both essential for the regulation of Pax7+ cell proliferation and, consequently, in their differentiation during the establishment of the zebrafish dermomyotome. In both six1a and six1b morphant embryos, Pax7+ cells are initially formed but fail to proliferate, as detected by reduced levels of the proliferation marker phosphohistone3 and reduced brdU incorporation. In congruence, overexpression of six1a or six1b leads to increased Pax7+ cell number and reduced or alternatively delayed fibre cell differentiation. Bone morphogenetic protein signalling has previously been suggested to inhibit differentiation of Pax7+ cells in the dermomyotome. Here we show that the remaining Pax7+ cells in six1a and six1b morphant embryos also have significantly reduced pSmad1/5/8 levels and propose that this leads to a reduced proliferative activity, which may result in a premature differentiation of Pax7+ cells in the zebrafish dermomyotome. In summary, we show a mechanism for Six1a and Six1b in establishing the Pax7+ cell derived part of the fast muscle and suggest new important roles for Six1 in the regulation of the Pax7+ muscle cell population through pSmad1/5/8 signalling.
The formation of muscle cells during zebrafish development depends on the combined function of several signalling pathways. The myotome is derived from the paraxial mesoderm and muscle differentiation is depending on the expression of the myogenic regulatory factors (MRFs), including MyoD, Myf5 and myogenin (reviewed by Bryson-Richardson and Currie, 2008). Initially, the myogenic progenitors have the competence to form both slow and fast fibre cells depending on the signals they receive during their development. Hedgehog signalling induces slow fibre differentiation in the muscle progenitors closest to the notochord known as adaxial cells (Blagden et al., 1997; Du et al., 1997). Fast muscle fibre are formed from non-adaxial myoblasts and will constitute the majority of muscle cells in the zebrafish myotome (Blagden et al., 1997; Devoto et al., 1996). In a subpopulation of cells at the somite anterior border the paired box protein Pax7 can be detected, this region has been referred to as the zebrafish equivalent to the amniote dermomyotome (Devoto et al., 2006; Hollway et al., 2007; Stellabotte et al., 2007), and will give rise to the most laterally positioned fast fibres (Stellabotte et al., 2007). Furthermore, the zebrafish dermomyotome can be divided into two domains, where the central dermomyotome and the peripheral most ventral and dorsal domains of the dermomyotome precursors are suggested to respectively constitute functionally and genetically distinct subpopulations (Windner et al., 2012). The Pax7+ cells have a stem cell like character and part of the Pax7+ cell population has been suggested to remain in an undifferentiated state throughout development and constitute an equivalent to the mammalian muscle satellite cells (Devoto et al., 2006; Hollway et al., 2007). When differentiation of Pax7+ cells is induced, MyoD and subsequently myogenin becomes upregulated (Relaix et al., 2006). The mechanism behind the regulation of differentiation and proliferation within the Pax7+ cell population has however to date not been determined.
The bone morphogenetic protein (BMP) pathway is an important player in zebrafish myogenesis where it defines subdomains within the myotome by counteracting hedgehog signalling from the notochord and floorplate (Dolez et al., 2011; Maurya et al., 2011). BMP receptors signal by phosphorylation of regulatory (R) Smads 1, 5 and 8. The activated R-Smads translocate to the nucleus together with co-Smad4 to regulate BMP target genes (reviewed by Massagué et al., 2005). In mammals BMP signalling has been shown to regulate cell proliferation by activating Smad1/5/8, which can mediate transcriptional regulation and influence cell cycle progression (Pardali and Moustakas, 2007). BMP signalling can influence amniote myogenesis by activating Smad1/5/8 through phosphorylation (Yamamoto et al., 1997) and thereby repress myogenic progenitor differentiation (Amthor et al., 1998; Dahlqvist et al., 2003; Frank et al., 2006). In addition, inhibition of BMP signalling leads to a decreased number of Pax7+ cells during fetal development in chick (Wang et al., 2010), whereas overexpression of BMP leads to increased Pax7+ cell number in zebrafish embryos at 24 hours post fertilization (hpf) (Patterson et al., 2010).
The class of sine oculis (six) 1 and 4 genes have previously been linked to muscle development in several species (Bessarab et al., 2004; Pandur and Moody, 2000; Spitz et al., 1998). Six1 and Six4 are partially redundant in mammals where Six1−/− knock out mice die at birth and show severe muscle hypoplasia, whereas Six1−/− Six4−/− mice completely lack fast muscle (Grifone et al., 2005; Laclef et al., 2003). In zebrafish, there are two six1 genes, named six1a and six1b. Studies have been performed on six1b under the previous name six1a (Bessarab et al., 2008; Lin et al., 2009). In contrast to mammalian Six1, the zebrafish six1b morphant embryos alone show a severe muscle phenotype, in particular in the craniofacial region, where they completely lack fast muscle fibre (Lin et al., 2009). six1b morphants have a reduced expression of several fast muscle markers, such as isoforms of the fast muscle specific myosin heavy chains and a-tropomyosin (tpma) (Bessarab et al., 2008). In addition, Six1b has been shown to be essential for the formation of cranial and pectoral fin muscle (Lin et al., 2009). Six1 in mammals has also been shown to regulate the expression of Pax3 (Grifone et al., 2005), which is expressed together with Pax7 in the majority of the dermomyotomal cells (Relaix et al., 2005), but the specific role of Six1 in the Pax7+ cells in zebrafish has not been determined. The function of zebrafish Six1a (previously called Six1b) has to date not been studied and the potential role of Six1a and Six1b during Pax7+ cell formation, proliferation and differentiation has not been established.
In this study, we show that Six1a and Six1b both are important for proper formation and growth of fast specific skeletal muscle. We also show that Six1a and Six1b are essential for regulating proliferation and consequently the differentiation of Pax7+ cells in the dermomyotome. Finally, we propose a mechanism where Six1 is acting via phosphorylated Smad1/5/8 in Pax7+ cells in the zebrafish peripheral dermomyotome.
Zebrafish Six1 influences fast fibre formation
Skeletal fast specific fibres are formed from the non-adaxial cells in the paraxial mesoderm and genes encoding fast specific components of the sarcomere can be detected in these cells after the 8 somite stage (8 ss) (Burguière et al., 2011). Similarly, expression of six1b can be detected in the same region starting at the 6 ss (Bessarab et al., 2008). The expression of six1a has previously not been studied. We examined the expression patterns of both six1a and six1b in the myotome of zebrafish embryos during the period of fast fibre differentiation and compared them with the expression of the slow myosin heavy chain 1 (smyhc1) to identify adaxial versus non-adaxial muscle progenitors. Expression of both six1a and six1b was detected in the fast muscle progenitors but was absent in the adaxial cells and in the adaxially derived slow fibres (Fig. 1). Interestingly, the expression patterns of six1a and six1b differed within the fast progenitors, where six1a was downregulated in the anteriormost somites and conversely six1b was expressed more intensely (Fig. 1A,B). six1b was also expressed in the presomitic mesoderm and the newly formed somites at 12 ss (Fig. 1B). These expression patterns remained through the 18 ss, but were less pronounced at 24 hpf, where both six1a and six1b were widespread throughout the whole fast domain within the trunk (Fig. 1). At 12 ss, six1a was expressed in the rostral and medial part of the somite (Fig. 1A). six1b was uniformly expressed in the most mature somites, at 12 ss (Fig. 1B), indicating that the expression of both genes was, at least partially, corresponding to the Pax7+ dermomyotome cells (Stellabotte et al., 2007). However, in newly formed somites expression of six1b was restricted to the caudal part of the somite (Fig. 1B), which is consistent to what Bessarab et al. (Bessarab et al., 2008) found at 6 and 9 ss.
In order to inhibit Six1a and Six1b during skeletal muscle development, we used morpholinos targeting six1a and six1b respectively, both individually and in combination. We used established morpholinos targeting six1b (Bessarab et al., 2008) and designed a new morpholino targeting the six1a exon1/intron1 splice junction. The six1a splice morpholino introduced an early splice site in the six1a transcript and led to an out of frame translation product with a premature translational stop (supplementary material Fig. S1). The efficiency of the six1b morpholino was tested by injection into a transgenic line, six1b:GFP, where part of the six1b promoter drives the expression of GFP (described below). The morpholino is designed to block translational start and thus inhibits translation of GFP in the transgenic line (supplementary material Fig. S1). Knock down of Six1a and Six1b respectively both resulted in embryos with changes in the trunk morphology that indicated impaired skeletal muscle development (supplementary material Fig. S2). In addition to having an imperative effect on cranial muscle formation as previously described for Six1b (Lin et al., 2009), knock down of both Six1a and Six1b reduced the fast muscle domain significantly (Fig. 2A,B). This effect was seen both for Six1a and Six1b respectively, but was found most significant in embryos injected with a combination of six1a and six1b morpholinos (Fig. 2). An indication of a reduction of the fast domain was first observed after 24 hpf, but was found more pronounced after 3 days post fertilization (dpf), where the fast domain was significantly reduced in the morpholino treated embryos (Fig. 2B). Co-injections using p53 morpholinos together with the six1a and six1b morpholinos did not significantly change the six1 morpholino induced phenotypes (Fig. 2B,C). A reduction of cell number in the fast domain was less evident, but still significant in six1a/six1b morpholino treated embryos after 3 dpf (Fig. 2C). However, the number of nuclei per fibre was not affected by the morpholino treatment (data not shown). Neither was the number of slow fibres in the Six1 morphants at 24 hpf, where control embryos had 55.7 slow fibres/somite, six1a and six1b morphants both had 56.0 slow fibres/somite and embryos injected with six1a and six1b morpholino in combination had 55.6 fibres/somite (Fig. 2D). Nor did the six1 impairment significantly affect slow fibres in embryos at 3 dpf; control embryos had 61.8 slow fibres/somite, embryos injected with six1a morpholino had 60.4 slow fibres/somite, six1b morphants had 58.6 slow fibres/somite and embryos injected with six1a and six1b morpholino in combination had 61.0 fibres/somite (Fig. 2D).
Six1 inhibits differentiation of Pax7+ cells in the dermomyotome
To further study the expression of six1b, we generated a transgenic line, tg(six1b:GFP)J1 where 6.5 kb upstream the six1b translation start drives the expression of GFP. At 12 ss, 15 ss and 18 ss six1b:GFP expression was found strongest in the most anterior somites where the expression partly overlapped with the expression of Pax7 (Fig. 3A). By comparing six1b:GFP expression with six1b mRNA expression at 24 hpf, we found similar expression patterns in the myotome, the lateral line placode and the cranial placodes (Fig. 3B), confirming that the transgenic line phenocopied six1b expression. At 26 hpf, six1b:GFP was expressed in the majority of the dermomyotomal Pax7+ cells as well as in fast muscle fibres (Fig. 3C). At later stages, 48 hpf and 3 dpf, expression of six1b:GFP was found in the otic vesicle, olfactory vesicle, adenohypophyseal placode, lateral line neuromasts and pectoral fins, as well as in the fast muscle fibres (Fig. 3D,E).
Muscle differentiation requires the expression of myoD and myf5 (Hinits et al., 2009; Tapscott, 2005; Tapscott et al., 1988). Both slow and fast fibres are significantly reduced or fail to form in embryos where MyoD and Myf5 are knocked down following morpholino treatment (Fig. 4A). Interestingly, we found that the mRNA expression of six1b remained in myoD/myf5 morphant embryos. In cross-sections of 24-hpf myoD/myf5 morphant embryos, the six1b expression was detected in cells surrounding the residual slow and fast fibres (Fig. 4A). The six1b expression domain coincided with the location of the Pax7+ muscle progenitor cells, which was increased in number in the myod/myf5 morphant embryos (Fig. 4A) The six1b:GFP was also expressed in the Pax7+ cells in wt and in the myod/myf5 morphant embryos (Fig. 4B), confirming our previous observations of overlapping Pax7 and six1 expression.
In order to examine a putative function of Six1a and Six1b in the formation of Pax7+ cells, we analysed the expression of Pax7 in six1a and six1b morphant embryos. At 24 hpf the average number of Pax7+ cells was significantly reduced both in six1a and six1b morphants as well as in six1a and six1b combined morphant embryos (Fig. 5A,B). The number of Pax7+ cells was found to be 46.0 per somite in control embryos, six1a morphants had 32.8 Pax7+ cells per somite, in six1b morphants the number of Pax7+ cells was reduced to 27.2 per somite and the six1a/six1b double morphant embryos had 27.1 Pax7+ cells per somite (Fig. 5A,B). To examine if the reduced number of Pax7+ cells was due to a lack of initial cell formation or inhibited cell proliferation we analysed the number of phosphorylated histone H3 positive (pH3+) cells in six1a and b morphants at 24 hpf and found that the number of pH3+ cells was reduced from 6.02 per somite in the control to 3.78 in the six1a morphants, 2.55 in the six1b morphants and 2.30 in the six1a/six1b double morphant embryos (Fig. 5C,D), indicating a lowered cell proliferation following Six1a and/or Six1b inhibition. To verify that the proliferation of Pax7+ cells was affected, embryos were subjected to brdU treatment, which was initiated at 15 ss and was terminated at 24 hpf. Thereafter, the brdU incorporation specifically in Pax7+ cells was analysed. In the control embryos, 67.5% (23.7/35.1) of the Pax7+ cells were brdU+ (Fig. 5E,F). The proportion of cells double labelled with Pax7 and brdU decreased significantly to 32.5% in six1a morphants, 26.3% in six1b morphants and 14.8% in six1a/six1b double morphants (Fig. 5E,F), which indicates that the proliferation of Pax7+ cells is significantly inhibited in the six1 morphant embryos.
Based on our previous observations that six1b:GFP expression was not found in all dermomyotomal Pax7+ cells (Fig. 3), and that the six1 morphants had an inhibited proliferation of Pax7+ cells (Fig. 5A–F) we explored if six1 is expressed specifically in the proliferating Pax7+ cells as opposed to the subset of Pax7+ cells that are non-proliferative. six1b:GFP embryos were incubated in brdU from 20 ss to 24 hpf, whereafter cells positive for Pax7, BrdU and GFP were analysed (Fig. 5G,H). We found that an average of 20.6 cells per somite were brdU+ and 3.6 lacked brdU among the six1b:GFP negative Pax7+ cells. However, among the six1b:GFP positive Pax7+ cells an average of 17.2 cells per somite were brdU+, whereas none (0.0) of the six1b:GFP positive Pax7+ cells (n = 379) were brdU negative, indicating that all Pax7+ cells expressing six1b:GFP had entered the cell cycle (Fig. 5G,H), thus suggesting that six1 is keeping the Pax7+ cells in a proliferative state.
If six1 is promoting Pax7+ cell proliferation, can overexpression of six1 increase the proliferative activity of the Pax7+ cells? To investigate this we injected six1a and six1b mRNA into one-cell stage embryos and analysed the number of Pax7+ cells at 24 hpf (Fig. 6A). Overexpression of six1a resulted in 47.4 Pax7+ cells per somite and embryos injected with six1b RNA had 49.2 Pax7+ cells per somite, which is a moderate but still significant increase compared with control embryos that had 42.7 Pax7+ cells per somite. These results support our hypothesis that six1 is involved in keeping the Pax7+ cells in a proliferative state.
To further analyse the effects of six1a and six1b overexpression in muscle progenitors during myogenesis, we used a heat-shock inducible construct that drives the expression of six1a or six1b together with a reporter gene upon activation. One-cell stage embryos were microinjected either with an empty control vector expressing only gfp after heat-shocked, or with vectors also bi-directionally driving six1a or six1b expression. At 18 ss the proportions of GFP+ fibres versus GFP+ non-differentiated myotomal cells were quantified (Fig. 6B). Embryos injected with the construct driving GFP expression alone had 57.3% GFP+ fibres and 42.7% non-differentiated GFP+ myotomal cells. When expression of six1a was induced together with GFP, the proportion of GFP+ fibres dramatically decreased to 11% and upon six1b overexpression, only 4.4% of the GFP+ cells had adopted a fibre-like morphology (Fig. 6B,C) indicating that the overexpression of six1a or six1b resulted in an inhibition or delay of fibre differentiation. To further examine the effects of six1 overexpression on fibre differentiation, newly fertilized mylz2:GFP embryos, that express GFP specifically in differentiated fast fibres, were injected with the same heat shock constructs but with tdtomato (tdtom) as the reporter gene. These embryos were analysed at 16 ss where the mylz2:GFP reporter has started to mark newly differentiated fast muscle cells (Fig. 6D). The control construct expressed tdTom in numerous muscle cells that had adopted a fibre-like morphology and several which also expressed the mylz2:GFP (Fig. 6D). In contrast, tdtom+/six1a+ or tdtom+/six1b+ cells rarely had a fibre-like morphology (Fig. 6D), Taken together, these results further strengthens our hypothesis that the role of six1 expression in Pax7+ cells is to keep them proliferating and consequently inhibiting the Pax7+ cells from forming muscle fibres.
Six1 regulates pSmad1/5/8 in Pax7+ cells
Proliferation of Pax7+ cells has previously been linked to BMP signalling, where overexpression of Bmp2b led to an increased number of Pax7+ cells via elevated levels of phosphorylated Smad (pSmad) in the zebrafish embryonic dermomyotome (Patterson et al., 2010). To examine if Six1a and Six1b act on proliferation through the pSmad pathway, we analysed pSmad1/5/8 in six1a and six1b morphant embryos specifically in the Pax7+ cells. We found that 9.0 out of 46.7 Pax7+ cells (19.3%) were pSmad1/5/8+ in uninjected control embryos at 24 hpf (Fig. 7A,B). Most of these cells were situated in the most dorsal and ventral regions, the peripheral dermomyotome, but were excluded from the central dermomyotome (Fig. 7A,C). As described above, the number of Pax7+ cells is significantly decreased in all types of six1 morphant embryos at 24 hpf (Fig. 5A,B; Fig. 7B). In addition, the pSmad1/5/8+ proportion of the Pax7+ cells was significantly decreased to 8.9% (2.5/28.2) in six1a morphants, 7.2% (2.3/31.9) in six1b morphants and 6.1% (1.8/29.6) in six1a/six1b double morphants (Fig. 7B). Interestingly, while pSmad1/5/8 was significantly reduced in Pax7+ cells, we found that the pSmad1/5/8 expression in the Pax7 negative myotomal cells remained unaffected in the six1 morphants (Fig. 7C), indicating that the Six1 effect on pSmad1/5/8 is constrained to cells in the dermomyotome.
To examine the effect of six1a and six1b overexpression on pSmad1/5/8 in the dermomyotome, we analysed pSmad1/5/8 in Pax7+ cells at 24 hpf after heat-shock inducing six1a and six1b at 12 ss, but neither six1a nor six1b were able to induce pSmad1/5/8 ectopically (supplementary material Fig. S3), which suggests that zebrafish six1 is required but not sufficient to induce pSmad1/5/8 in Pax7+ cells.
We analysed the effect of dorsomorphin, which is an inhibitor of Smad1/5/8 phosphorylation (Yu et al., 2008), on the Pax7+ cell population. Embryos treated with dorsomorphin from shield stage and onwards until fixation at 24 hpf displayed a dorsalized phenotype (Fig. 7D) where the treated embryos also occasionally exhibited additional tail tissue as previously described in Yu et al. (Yu et al., 2008). However, the six1b expression remained unaffected in the dorsomorphin-treated embryos, indicating that six1b is acting either upstream or in a parallel pathway (Fig. 7D). Furthermore, the number of Pax7+ cells was significantly decreased at 24 hpf in dorsomorphin-treated embryos compared with DMSO treated controls when inhibiting Smad1/5/8 phosphorylation from 2–3 hpf (Fig. 7E,G). To assess if the reduced Pax7+ cell number was caused by inhibition of proliferation of the Pax7+ cells, we also initiated dorsomorphin treatments at 10 ss, when the first Pax7+ cells are already formed (Feng et al., 2006; Hammond et al., 2007). Here, we observed a similar reduction in Pax7+ cell number (Fig. 7F,G), reinforcing our observation that reduced levels of pSmad1/5/8 results in reduced Pax7+ cell number and support our hypothesis that the reduced pSmad1/5/8 influences the proliferation and consequently the differentiation of the Pax7+ cells rather than their initial specification.
To further analyse the relationship between Bmp signalling and Six1 during the proliferation of Pax7+ cells, we analysed six1 overexpression in dorsomorphin-treated embryos. Embryos subjected to dorsomorphin treatment, injected with six1a and six1b mRNA, had significantly higher numbers of Pax7+ cells than the dorsomorphin-treated uninjected embryos (Fig. 8A). We also overexpressed bmp2b in six1a and six1b morphant embryos and found a partial rescue of the Pax7+ population (Fig. 8B). Increased levels of bmp2b did however not lead to complete rescue of the Pax7+ population, leading to the conclusion that the proliferation of Pax7+ cells can be regulated either directly via Six1 or through pSmad1/5/8 (Fig. 8B,C), implying that there are dual Pax7+ populations with differing requirements. Collectively our data indicate that the proliferation of Pax7+ cells in the central dermomyotome, which lack pSmad1/5/8, is directly regulated by Six1, whereas the proliferation is modulated via pSmad1/5/8 in the peripheral dermomyotome cells (Fig. 8C)
By manipulating the cellular levels of Six1a and Six1b during the establishment and differentiation of cells in the zebrafish embryonic somite we have found that both genes have profound effects on fast muscle development and are important regulators of proliferation in the embryonic Pax7+ population.
Inhibition of Six1 leads to impaired muscle formation
We show that the domain in the myotome harbouring the differentiated fast fibres becomes reduced in six1a and six1b morphants (Fig. 2). This reduction is non-significant after 24 hpf, but becomes more evident at 3 dpf. The number of nuclei per fibre was not affected in the morphant embryos and even though the number of fast-twitch fibres decreased in 3-dpf six1a and six1b morphants, the primary cause of the reduced fast domain does not seem to be due to a decreased fibre number, but rather a condense in cell sizes, which links Six1 to functions related to hypertrophy. At 24 hpf six1a and six1b morphant embryos have a reasonably normal morphology and the effects on the myogenic cells in the trunk are not causing any severe phenotypes at this stage. The numbers of surface slow fibres and muscle pioneer cells remain normal (Fig. 2D) and the engrailed positive cells are seemingly unaffected (data not shown). This observation is in line with previous studies using six1b morphant embryos (Bessarab et al., 2008). In addition to the reduced fast muscle domain, we also observed a drastic reduction of cranial muscle using both six1a and six1b morpholinos. The function of Six1b in cranial muscle development has previously been described by Lin et al. (Lin et al., 2009), our data show that inhibition of Six1a has a similar, but less severe effect as that of Six1b in cranial muscle development. The strongest effect, both on the cranial muscle and the fast domain in the myotome, was observed in the six1a/six1b double morphant embryos (Fig. 2), which suggests that there is some redundancy between the zebrafish six1 genes. However, both genes are individually essential for normal muscle development in the craniofacial as well as the myotome region.
We used myoD/myf5 morphant embryos, which essentially lack differentiated muscle fibre, to examine if zebrafish six1b expression is dependent on the MRFs. Here, we found that six1b remains expressed in the somite, even when both the slow and the fast fibres are reduced to a minimum. Many of the remaining cells in the somite, following MyoD/Myf5 inhibition, are Pax7+ progenitor cells (Fig. 4) (Hammond et al., 2007). We found that the region expressing six1b in myoD/myf5 morphants overlapped with the location of the Pax7+ progenitor cells. A transgenic reporter line, where 6.5 kb upstream the six1b ATG drives expression of GFP, was established and analysed together with Pax7. This analysis identified cells coexpressing six1b:GFP and Pax7 and confirmed that zebrafish six1b is expressed in the Pax7+ progenitors (Fig. 3A,C), and collectively our data show that the six1 expression in Pax7+ cells is independent of MyoD and Myf5.
Zebrafish Six1 promotes Pax7+ cell proliferation
To follow up on the potential Six1 function in Pax7+ cells, we first analysed how inhibition of six1a and six1b affected Pax7+ cell number at 24 hpf and found significant reductions of Pax7+ cells using six1a and/or six1b morpholinos (Fig. 5A,B). However, a substantial number of Pax7+ cells was still present in the 24-hpf embryos after Six1 inhibition, which indicated that rather than being involved in the initial formation of the Pax7+ cells, zebrafish Six1 is involved in regulating Pax7+ cell number by either regulating Pax7+ cell proliferation or differentiation. We therefore analysed pHistone3, which marks cells in the M phase, in six1a and six1b morphants and found a general reduction (Fig. 5C,D). Previously, Bessarab et al. (Bessarab et al., 2008) showed that six1b inhibition resulted in minor changes in pH3+ cell number. The lack of effect could be explained by the early stage of the analysis, which was conducted at 9 ss and is very close to the onset of six1b expression. In our experiment, which was conducted at 24 hpf we found a very significant decrease in pH3+ cell number. In order to verify this and, in addition, to specifically analyse proliferation of Pax7+ cells, we brdU treated six1a and six1b morphant embryos between 15 ss to 24 hpf and analysed how many of the Pax7+ cells that had continued to proliferate after their initial formation. Here, we found that the proportion of brdU−/Pax7+ versus brdU+/Pax7+ cells increased dramatically in the six1a and six1b and particularly in the six1a/six1b double morphants, which indicates that Six1 is affecting proliferation of the Pax7+ cells (Fig. 5E,F). Control of cell proliferation has previously been established for Six1 in the development of sensory organs (Chen et al., 2009; Zheng et al., 2003) and in metastatic tumours (Coletta et al., 2004; Coletta et al., 2008), where Six1 acts as an activator of proliferation in line with our own findings. By contrast, Six1 has also been suggested to promote differentiation in adult cultured satellite cells (Yajima et al., 2010) and to regenerate muscle tissue and replenish the satellite cell pool following skeletal muscle trauma in adult mice (Le Grand et al., 2012), which indicates that six1 function may vary between cell types and be dependent on the cellular milieu.
In order to study overexpression of six1 during zebrafish muscle development, we used a heat-shock inducible vector, which we activated at late epiboly and analysed at 16 ss, when the first fast fibres have started to form, as marked by the fast muscle specific myosin light chain 2 reporter, tg(mylz2:GFP)i135 (Fig. 6C). In the control experiment, where the heat-shock vector exclusively expresses tdtomato, we found many cells that had an elongated fibre-like morphology, and a large proportion expressing the transgenic GFP as a consequence of their differentiated state. However, this was rarely observed in the six1a or six1b expression constructs, where very few cells in the myotomal region overexpressing six1a or six1b had adopted a fibre morphology, and even fewer coexpressing the mylz2:GFP at the examined stage. This indicated that overexpression of either six1a or six1b both lead to an inhibition or at least a delay of differentiation, which supports our finding that zebrafish Six1 promotes proliferation in a cell autonomous way.
pSmad1/5/8 levels are affected by Six1 in Pax7+ cells in the dermomyotome
Transforming growth factor beta (Tgfβ) signalling, including bone morphogenetic proteins (BMPs) can act on myogenesis through phosphorylated Smad (Yamamoto et al., 1997), which can act as repressors of myogenic progenitor differentiation (Amthor et al., 1998; Dahlqvist et al., 2003; Frank et al., 2006). Further, inhibition of BMP signalling leads to decreased number of Pax7+ cells during fetal development in chick (Wang et al., 2010). In zebrafish, overexpression of Bmp2b leads to upregulated pSmad1/5/8 levels and results in an increased number of Pax7+ cells in the dermomyotome (Patterson et al., 2010). In the six1a and six1b morphant embryos, we found a significant reduction of pSmad1/5/8 specifically in the Pax7+ cells indicating that Six1 is necessary for pSmad1/5/8 expression in Pax7+ cells (Fig. 7). BMP induced pSmad1/5/8 in zebrafish Pax7+ cells has previously been suggested to inhibit differentiation (Patterson et al., 2010), which is in congruence with our finding in the six1a and six1b morphant embryos. However, six1 was not sufficient to activate pSmad1/5/8 ectopically when induced at the 12 ss (supplementary material Fig. S3), which could be explained by the fact that zebrafish six1 is already expressed both in the fast muscle progenitors as well as in the Pax7+ cells in the central part of the dermomyotome and that additional factors are required to induce pSmad1/5/8 in these regions. This also indicates that cells in the central dermomyotome respond differently than cells within the pSmad1/5/8 positive region. However, inhibition of pSmad1/5/8 through dorsomorphin treatment efficiently reduced the number of Pax7+ cells in the dermomyotome, but not to the same extent as to those found in the six1a/six1b double morphants (Figs 5, 7). This supports our hypothesis that Six1 regulates the proliferation and consequently, the differentiation of Pax7+ cells through pSmad1/5/8 in the peripheral dermomyotome region. Previous studies have shown that BMP signalling regulates the proliferation of Pax7+ satellite cells through pSmad1/5/8, where inhibition of pSmad1/5/8 lead to induced differentiation (Ono et al., 2011; Yanagisawa et al., 2011), which further supports our finding in the zebrafish Pax7+ cells.
Interestingly, both six1a and six1b overexpression were able to increase the number of Pax7+ cells in dorsomorphin-treated embryos, indicating that at least some cells are independent of pSmad1/5/8 (Fig. 8). Similarly, overexpression of bmp2b was able to partially rescue the number of Pax7+ cells in six1a and six1b morphant embryos (Fig. 8). Collectively this indicates that there are more than one population of Pax7+ cells with different requirements. A recent study by Windner et al. (Windner et al., 2012) shows that the zebrafish dermomyotome is divided in two domains, a central domain and a more peripheral domain, which is situated ventral and dorsal to the central domain. These domains are suggested to be genetically and functionally distinct, which concurs with our own observations. Our data indicate that Six1 regulates proliferation of Pax7+ cells via pSmad1/5/8 in the peripheral dermomyotome and that Six1 acts directly on proliferation in the central dermomyotome (Fig. 8C).
The function of Six1 during zebrafish muscle development is complex. Here, we show that both of the zebrafish six1 genes are important for normal muscle formation, both in the craniofacial region, the fast domain of the myotome and in the proliferation and differentiation of the Pax7+ dermomyotome cells. We show that zebrafish Six1 is promoting proliferation and simultaneously, as a consequence, inhibiting differentiation of Pax7+ cells into fast fibre through a process where we propose that pSmad1/5/8 is required in the peripheral dermomyotome to maintain the Pax7+ cells in a proliferative state. In addition to the role for Six1 in the Pax7+ cells, six1a and six1b are both expressed in the mature fast fibres. We found that inhibition of Six1 lead to a significant reduction of fibre cell size, which suggests that Six1 affects hypertrophy; however, the role for Six1 in the mature fast fibres is still not clear and remains to be examined further.
Materials and Methods
Zebrafish strains and maintenance
Embryos were obtained from wild-type zebrafish [Danio rerio, London wild type (LWT) and tupfel long fin (TL)]. Zebrafish were maintained by standard procedures in the Umeå University Zebrafish Facility. Embryos were staged up to the 15-somite stage by counting the number of somites, as described previously (Kimmel et al., 1995). Transgenic lines used were tg(mylz2:GFP)i135 (von Hofsten et al., 2008) and tg(six1b:GFP)J1 which was generated by isolating 6.5 kb upstream the six1b ATG (gene bank accession number: BX649231) to generate a stable transgenic line.
Whole-mount in situ hybridization
Zebrafish embryos were fixed in 4% paraformaldehyde (PFA) over night. Whole-mount in situ hybridization was performed as described previously (Thisse et al., 1993) with minor changes; 1% blocking reagent (Roche, IN) was used instead of 2% sheep serum and 2 mg/ml BSA. Two-colour whole-mount RNA in situ hybridization was performed as described previously (Westerfield, 2000) using DIG-labelled and fluorescein-labelled RNA probes, and using Fast Red (Roche, IN) and NBT/BCIP (Roche) for detection. For the fluorescent in situ hybridization anti-DIG POD (1∶10,000; Roche) was used with TSA-Cy3 (Perkin Elmer, Waltham, MA). RNA probes were fMyHC1.3 (gene bank accession number: EU115994), six1a (BC076015) and six1b (BC066396).
Zebrafish embryos were fixed in 4% PFA over night. Immunohistochemistry was performed using standard procedures. For brdU and pHistone3 detection, rehydrated embryos were treated with proteinase K for 10 minutes (10 µg/ml), re-fixed for 20 minutes in 4% PFA, washed and treated with 2 N HCl for 1 hour, followed by washing, blocking and antibody incubation. Primary antibodies used were F59 (slow myosin heavy chain, DSHB), F310 (myosin light chain 1 and 3F, DSHB) and anti-Pax7 (1∶10, 1∶50 and 1∶10, respectively, DSHB, University of Iowa, IA) anti-p-histone3 (1∶400, Merck Millipore KGaA, Darmstadt, Germany), anti-GFP (1∶200, Aves, OR) and anti-pSMAD1/5/8 (1∶30, Cell Signaling). Nuclei were stained using DAPI (Sigma). For brdU detection, a mouse anti-brdU antibody conjugated to either Alexa Fluor 555 or Alexa Fluor 488 was used (1∶20, BD Biosciences).
Microinjections and morpholino verification
For morpholino treatment, zebrafish embryos were injected at the 1–4 cell stage with 4 nl of myoD (atatccgacaactccatcttttttg, 0.6 nmol/µl) and myf5 (gatctgggatgtggagaatacgtcc, 0.6 nmol/µl), six1a (atgattaacgaagatctcacctctc, 1 nmol/µl) and six1b (tctcctctggatgctacgaaggaag, 1 nmol/µl), p53 (gcgccattgctttgcaagaattg, 1 nmol/µl) morpholinos (GeneTools, Philomath, OR). For double morpholino treatment, 4 nl of a mixture of six1a (0.5 nmol/µl) and six1b (0.5 nmol/µl) was used.
To verify the functionality of the six1a splice morpholino, mRNA from injected and uninjected embryos was isolated using TriZol (Life Technologies). After DNase treatment, extracted RNA was used as template for cDNA synthesis (forward primer 5′-cgcacaacccttatccatct-3′; reverse primer 5′-cgacatcagcgacttcactc-3′). cDNA was cloned into a TOPO vector (Invitrogen), sequenced and analysed.
six1a, six1b and bmp2b (Little and Mullins, 2009) RNA was generated by using mMESSAGE mMACHINE® Kit (Ambion, TX). 4 nl of RNA was microinjected into one-cell stage embryos at a concentration of 1 µg/µl for six1a and six1b and 3 ng/µl for bmp2b, fixed at desired stage in 4% PFA over night, dehydrated in 100% methanol and stored at −20°C until analysis using immunohistochemistry.
Embryos were dechorionated and incubated in 10 mM 5-Bromo-2′-deoxyuridine (brdU, Sigma) in embryo medium until they reached the desired developmental stage, fixed in 4% PFA over night, dehydrated in 100% methanol and stored at −20°C until analysis using immunohistochemistry.
Heat shock treatment
For heat-shock-induced expression, the complete six1a and six1b coding sequences were subcloned into a pSGH2 vector containing a bidirectional heat shock promoter (Bajoghli et al., 2004) that drives expression of both the fusion protein and fluorescent proteins tdTomato or GFP. One-cell stage embryos were microinjected with plasmid at a concentration of 40 ng/µl, when the embryos reached 70% epiboly or 12 ss, they were heat-shocked at 38°C for 2 hours. tdTomato/GFP positive embryos were fixed at 16 ss, 18 ss or 24 hpf in 4% PFA over night, dehydrated in 100% methanol and stored at −20°C until analysis using immunohistochemistry.
Dorsomorphin (Sigma) was dissolved in DMSO at a stock concentration of 5 mM and used at a concentration of 5 µM, 15 µM, 100 µM and 200 µM in embryo medium. Embryos were incubated in either dorsomorphin or the same volume of DMSO at 2–3 hpf for 5 µM, shield stage for 15 µM and at 10 ss for 100 µM and 200 µM, then fixed in 4% PFA at 24 hpf.
Cell counting and statistical analysis
To measure the F310+ area, the F310+ area on cross-sections, which excludes all cells but the fast specific fibres of 24-hpf and 3-dpf embryos was marked and measured using Adobe Photoshop CS4, similarly, the DAPI stained nuclei were counted in the F310+ area. Cross-sections were made at somite number 10 for embryos at 24 hpf and at number 8 for embryos at 3 dpf. The average number of cells positive for Pax7, brdU, GFP and pSMAD1/5/8 was determined by counting the average number of stained cells per somite number 10 to 12 or for pHistone3 somite number 10 to 18. The proportion of GFP+ fibres/non-differentiated myotomal cells after heat shock induced six1 overexpression was determined by counting GFP+ fibres and GFP+ non-differentiated myotomal cells at somite number 8 to 10. n = number of embryos or somites where applicable. Significance was determined by Student's t-test where P<0.05 is considered significant; *P<0.05, **P<0.01, ***P<0.001. All error bars represent the standard error of mean (s.e.m.).
We thank Mary C. Mullins and James A. Dutko for reagents and Edward Samuel Inbaraj for technical assistance.
H.N, L.S and J.v.H. conducted the experiments. H.N and J.v.H wrote the paper.
This project was funded by financial support from the Kempe foundation, the Åke Wiberg foundation, the Magnus Bergvall foundation, the Jeansson foundation and the Swedish Royal Academy of Science.