Cited3 activates Mef2c to control muscle cell differentiation and survival

Skeletal muscle is a major tissue type comprising around 40 percent of the adult mass, controlling locomotion and body-support, and accounting for most of an individual's daily energy consumption. Loss of skeletal muscle mass and/or strength occurs during aging, disease, or inactivity (Evans, 2010). When the function of the muscle tissue is impaired it results in various metabolism- and motility-related defects. One of the therapeutic avenues for treating muscle diseases is cell-based therapies, where functional muscle cells will be generated and amplified in culture, and finally delivered into patient tissue. However, generation of functional muscle cells requires a detailed understanding of how muscle cells differentiate in the first place.

As in other tissue types, specification of muscle cells does not rely on a linear regulatory pathway but rather on a network of interlinked signaling pathways and transcriptional regulators (Parker et al., 2003; Berkes and Tapscott, 2005; Ochi and Westerfield, 2007; Bryson-Richardson and Currie, 2008). Hence, elucidating this process can serve as a model for understanding development of other tissues. To identify the transcriptional regulators that are upregulated during the initial stages of muscle development in vivo, we microdissected zebrafish paraxial mesoderm into different spatiotemporal domains and performed microarrays (Ozbudak et al., 2010). Our recent study, for the first time, revealed the in vivo temporal induction order of transcriptional regulator genes as cells commit to muscle differentiation.

cited3, which is a member of the CITED (CBP/p300-interacting transactivator with glutamic acid/aspartic acid-rich tail) family transcriptional coactivator, was among the differentially expressed transcriptional regulators. Zebrafish cited3 is orthologous with mammalian Cited4. Expression of Cited4 was previously detected in human skeletal muscle tissue (Tews et al., 2007). Although the functional roles of Cited genes in skeletal muscle development have not been studied, they are important for mammalian development and have been implicated in cardiac (Sperling et al., 2005; Boström et al., 2010; MacDonald et al., 2012) and skeletal muscle (Pescatori et al., 2007; Sáenz et al., 2008; Tobimatsu et al., 2009) defects. Cited2 RNA levels were significantly reduced in Duchenne muscular dystrophy and limb-girdle muscular dystrophy 2A human patients (Pescatori et al., 2007; Sáenz et al., 2008) and overexpression of Cited2 counteracts glucocorticoid-induced muscle atrophy in C2C12 myotubes in cell culture (Tobimatsu et al., 2009). On the other hand, an increased level of Cited4 was previously associated with cardiomyocyte growth and proliferation in mice (Boström et al., 2010). Therefore, we decided to investigate the functional role of cited3 in skeletal muscle development.

An elegant lineage-tracing study in zebrafish previously demonstrated that slow muscle cells differentiate from their progenitor adaxial cells residing next to the midline axis in the somites (Devoto et al., 1996; Ochi and Westerfield, 2007; Stellabotte and Devoto, 2007; Bryson-Richardson and Currie, 2008). By in situ hybridization, we identified that cited3 is expressed in the precursors of slow myofibers and this expression is depended on Hedgehog signaling via MyoD and Myf5.

We knocked down the expression of cited3 by injecting two different antisense morpholino oligonucleotides that block splicing of cited3; this impaired muscle cell differentiation and growth, increased the number of apoptotic muscle cells and eventually led to total immotility. Overexpression of cited3 mRNA rescued all the cited3 loss-of-function phenotypes. In situ hybridization for various muscle-specific genes placed cited3 into a gene regulatory network, where it acts downstream of myoD/myf5 but upstream of mef2c. We identified that mef2c is the main target of Cited3 as its phenotype is rescued by restoring expression of mef2c. These data make a new and unprecedented connection between Cited3 and Mef2c – a crucial transcription factor for both heart and skeletal muscle tissues (Lin et al., 1997; Nakagawa et al., 2005; Hinits and Hughes, 2007; Potthoff et al., 2007; Potthoff and Olson, 2007). This is the first report demonstrating a Cited-family transcriptional coactivator functioning in the early stages of muscle development.

Wild-type AB and TB fish strains were raised in the fish facility. Loss-of-function experiments of FGF signaling was performed using transgenic fish with heat-shock-driven expression of dn-fgfr1 Tg (hsp70: dn-fgfr1-gfp) (Lee et al., 2005). Loss of function of Wnt signaling was accomplished by using transgenic fish with heat-shock-driven expression of Δtcf: Tg (hsp70: Δtcf-gfp), (Lewis et al., 2004). Gain of function of Wnt signaling was performed by using transgenic fish with heat-shock driven expression of wnt8: Tg (hsp70: wnt8-gfp) (Weidinger et al., 2005). myf5^hu2022 mutant were obtained from the Wellcome Trust Sanger Institute. Fish were bred and maintained at 28.5°C on a 14–10 h light/dark cycle as described (Westerfield, 1993).

Capped RNA for microinjection was synthesized by in vitro transcription according to Manufacturer's protocol (Ambion, the RNA Company, Texas, US) from linearized plasmid of pCS2⁺-cited3-254 (coding for full-length 254 amino acids), pCS2⁺-cited3-200 and pCS2⁺-Mef2c (Mouse cDNA, kind gift from Stephen Tapscott). Antisense cited3 (S1) morpholino oligonucleotide (MO) ‘TCTCCCTTCATTAAAACACACAAGA’ and cited3 (S2) MO ‘GCATCATCATGTGCTCTGCCATAGA’, are synthesized to block the proper splicing of cited3 at two different domains, were injected at a concentration of 0.8 or 12 ng/embryo, respectively, and both produced similar phenotypes. myoD MO ‘ATATCCGACAACTCCATCTTTTTTG’, was injected at 0.4 ng/embryo according to Hinits et al. (Hinits et al., 2009). Equivalent concentration of control MO (Gene Tools) was injected as a control. Fertilized eggs at one- to four-cell stages were microinjected with 80–200 pg RNA/embryo. Embryos were fixed at the desired stages in 4% paraformaldehyde fixative overnight at 4°C and washed with phosphate-buffered saline and stored in methanol at −20°C until use.

The coding sequence of cited3 was cloned into pDrive (Qiagen) and pCS2⁺ vectors by using cited3F: ATGGCAGAGCACATGATGATGCC and cited3R: TCAGCAGCTAACGGTGCTCGG primers. The C-terminally truncated cited3 variant is created by placing a stop codon after the first 199^th amino acids. Whole-mount in situ hybridization was carried out as previously described (Thisse and Thisse, 2008). Antisense RNA probes were synthesized by in vitro transcription from linearized plasmids using RNA polymerase T7 or T3 enzyme and digoxigenin RNA labeling mix (Roche). The following in situ hybridization probes were used: myod, myf5 (cb641 from ZIRC), mef2ca, prox1, stnnc and smyhc1 (kind gift from Phillip Ingham), alpha actinin, myogenin, myhz1, myhz2, mylc2, tnnt1 and tnnt3b (kind gift from Monte Westerfield), mef2d (kind gift from Simon Hughes), cmlc2 (kind gift from Deborah Yelon). Mouse Mef2c mRNA (kind gift from Stephen Tapscott) was injected in rescue experiments. Double fluorescent in situ hybridization was done following the protocol of Brend and Holley (Brend and Holley, 2009).

The following primary antibodies were used: F59 IgG (1:10 dilution) mainly detects slow muscle specific myosin heavy chain, S58 (1:10 dilution) detects slow muscle specific myosin heavy chain, MF20 (1:10 dilution) labels all differentiated muscle fibers and F310 (1:10 dilution) labels fast muscle specific myosin heavy chain were obtained from DSHB. Anti-Mef2 (1:100 dilution) was obtained from Santa Cruz. The following secondary antibodies were used at 1:800 dilutions: Alexa Fluor 488 or 555 goat anti-mouse IgG1 or IgA, Alexa Fluor 568 or 647 goat anti-rabbit IgG, Alexa Fluor 647 goat anti-mouse IgG2b (Molecular Probes). Immunostaining was performed following the method of Bird et al. (Bird et al., 2012).

Embryos were pulsed for different time periods in 1 mM BrdU (Sigma), fixed in 4% PFA and stained using anti-BrdU antibody (Sigma) following the protocol of Barresi et al. (Barresi et al., 2001) with minor modification for whole mount and on sections. To identify apoptotic cells, in situ cell death detection kit (Roche) was used. Embryos fixed (4% PFA) at different stages were treated with Proteinase K 1 µg/ml according to the stage and washed with PBST and stained with TMR red for 30–60 minutes at 37°C.

Embryos that are injected with control MO, cited3 MO or coinjected with cited3 RNA were tested for motility. Touch evoked response was recorded using live imaging of embryos from 2 days-post-fertilization (dpf) to 5 dpf using a Leica microscope and the speed of movements were analyzed using Sony Vegas movie Studio HD 9.0 software.

In situ hybridized embryos and immunostained embryos were scanned under a Leica 205MA microscope and Carl Zeiss Imager Z2 equipped with apotome and axiovision 4.8 Rel. Serial sections of fluorescent images of antibody staining were taken at 1–3 µm intervals with Apotome and then 2D images were created by stacking all pictures taken at different focal planes.

Mef2, Tunel and BrdU positive cells were counted using Image J. Cell counts were performed on z-stacks spanning an entire segment on one side of the embryo or on sections. Somites in 5–10 animals were scored and the experiments were reproduced minimum three times. Fluorescent images of different samples were taken on the same microscope with the same objective and exposure duration. The fluorescent intensity was quantified on the unmodified original images and the background intensity was subtracted by using Image J. F59 staining was diminished significantly after 2 dpf, while S58 staining was reduced at a slower pace and lost after 5 dpf. Therefore, we used S58 staining to count the number of slow fibers. Nevertheless, as S58 immunostaining was reduced in the morphants, we pseudo increased the intensity of staining in order to count the number of fibers. Fibers were counted on whole mount embryos on one hemi-segment in the anterior 10–12 somites and posterior 16–18 somites. Total number of fibers were counted from the 3D images acquired using axiovision software and verified with fluorescent intensity plots from plot profile using Image J (Barresi et al., 2001). All P-values were calculated by performing two-tailed Student's t-test in Excel. A P-value below 0.05 was considered statistically significant.

Cell transplantation was done following the protocol of Zeller and Granato et al. (Zeller and Granato, 1999). Donor embryos were injected with fixable tetramethyl rhodamine dextran fluorescent dye (Molecular Probes, D7162) with or without morpholinos. Cells from donor embryos were transplanted at shield stage into host embryos; chimeric embryos were identified, fixed, immunostained with F310 or F59 and Mef2 antibodies, and imaged.

cited3 is expressed in the polster (arrow), which is the precursor of hatching gland, and in the notochord (red arrow) and chordo-neural hinge (green arrowhead) at 10 hpf (Fig. 1A). At 13 hpf, cited3 is expressed in the slow muscle precursor adaxial cells (black arrowhead in Fig. 1B) on either side of the notochord. Subsequently, cited3 transcripts become abundant in slow muscle cells at 14 hpf to 18 hpf (Fig. 1C) when they are known to differentiate and migrate to the periphery. Expression of cited3 is detected in the hatching gland (black arrow), neural crest cells (red arrowhead), brain and branchial arches at 24 hpf (Fig. 1D). We performed double fluorescent in situ hybridization for cited3 and myod on the cross-sections of embryos that are at 24 hpf. At this stage, differentiating slow myofibers are located lateral to the differentiating fast myofibers. We demonstrated that cited3 is expressed only in the slow myofibers, while myod is expressed in both myofiber types (Fig. 1E–H). Double fluorescent in situ hybridization also revealed that cited3 is expressed both in the atrium and ventricle heart precursors (Fig. 1I–K) overlapping with cmlc2 expression (Yelon et al., 1999).

In order to identify the signaling pathways regulating cited3 expression, we blocked the hedgehog signaling by treating embryos with hedgehog inhibitor cyclopamine or with ethanol as a control from 5 hpf to 15 hpf. As a control, we checked the transcription of myod, which is known to be lost only in the adaxial region upon loss of Hedgehog signaling (Lewis et al., 1999; Osborn et al., 2011). Transcription of both myod and cited3 was lost in cyclopamine treated embryos in the adaxial region (Fig. 1M,O) but myod transcription was retained in precursors of fast myofibers and that of cited3 was retained in the chordo-neural hinge, compared to ethanol treated embryos respectively (Fig. 1L,N). These results indicate that cited3 is activated in response to Hedgehog signaling only in slow muscle precursors. Then we blocked the activity of Hedgehog signaling for a shorter duration starting at 13 hpf (when transcription of cited3 is already turned on in the adaxial cells) until 18 hpf. This resulted in loss of myod transcription in the adaxial cells throughout the axis (Fig. 1P,Q), while cited3 transcription is reduced only in the posterior-most part (presomitic mesoderm) of the adaxial region but not in the somites (Fig. 1R,S). Incubation of embryos with cyclopamine only for 2 hours from 13 hpf to 15 hpf reduced the transcription levels of myod drastically but not that of cited3 (data not shown). The results suggest that myod is a direct target of Hedgehog signaling, while cited3 might be indirectly regulated by Hedgehog signaling.

Then we investigated the regulatory relationship between cited3 and muscle-regulatory-factors myod and myf5. We knocked down the expression of myod with the myod MO (Hinits et al., 2009). Expression of cited3 was not affected in the myod morphants (Fig. 1T,U). Similarly, we checked the expression of cited3 in embryos from the mating of heterozygous myf5^hu202 mutants and found that its expression was not affected (data not shown). MyoD and Myf5 often regulate their targets redundantly. To investigate this possibility, therefore, we knocked down the expression of myod in embryos from myf5^hu202 heterozygous incross and observed that expression of cited3 was not affected in 75% of the embryos (Fig. 1V,V′) but was lost in the remaining 25% of embryos (Fig. 1W,W′). Hence, these results demonstrate that Myod and Myf5 redundantly activate transcription of cited3. Furthermore, they also suggest that Hedgehog signaling might activate the expression of cited3 via activating the expression of myod and myf5 (Lewis et al., 1999; Osborn et al., 2011) in the adaxial cells.

As Wnt and Fgf signaling are known to regulate muscle development, we tested the effect of these signaling factors on the expression of cited3. We perturbed the activities of Wnt and Fgf signaling in a time-resolved fashion by applying a pulse of heat-shock to transgenic embryos, where heat shock-responsive hsp70 promoter activates the expression of various transgenes (supplementary material Fig. S1). However, we found that neither Wnt nor Fgf signaling regulate transcription of cited3 (supplementary material Fig. S1).

We performed loss-of-function experiments for cited3 to investigate its functional role during development. We designed two splice-blocking morpholino oligonucleotides (MO) targeting the first intron-exon boundary (S1-MO and S2-MO) of cited3. As a control, we used the standard control MO that does not target any zebrafish gene. Subsequently, we confirmed the efficacy of two cited3 MOs by RT-PCR: injection of each cited3 MO prevented the splicing of the first intron significantly, whereas it is efficiently spliced out in the embryos injected with the control MO (supplementary material Fig. S2).

Knockdown of cited3 resulted in obvious morphological phenotypes compared to wild-type embryos or embryos injected with the control MO starting after 1 dpf. In the cited3 morphants, the heart was dilated, the body axis did not grow as much as it should (Fig. 2A,B), and the myosin heavy chain staining that is detected with F59 was reduced (Fig. 2E,F,I,J).

The specificity of the phenotype was verified by coinjection of full-length spliced mRNA of cited3 along with cited3 MOs. Full-length cited3 mRNA rescued the cited3 morphant phenotype (Fig. 2C,G,K).

Cited proteins are transcriptional coactivators. They interact with histone acetylases with their conserved C-terminal CR2 domain (Bragança et al., 2002). To assess whether the function of Cited3 protein depends on this conserved domain, we coinjected C-terminally truncated cited3 mRNA with cited3 MO. The truncated cited3^1–200 mRNA could not rescue the morphant phenotype; it also failed to rescue the F59 staining (Fig. 2D,H,L). Hence, we conclude that the transcriptional coactivator domain of Cited3 protein is necessary for its function during development.

Fig. 2M is the graphical representation of intensity measurements from F59 staining. Intensity of F59 staining was significantly reduced in cited3 morphants when compared to that in wild-type (P-value<2×10⁻⁴); the staining was significantly rescued when the full-length cited3 (P-value<2×10⁻²) but not the truncated cited3^1–200 RNA was co-injected (P-value>0.7). Interestingly, injection of full-length cited3 RNA alone into wild-type embryos increased the myosin heavy chain expression significantly compared to that in wild-type embryos (data not shown).

Knockdown of cited3 resulted in shorter embryos; this led us to test whether muscle cell growth is impaired and/or muscle cells are dying in the absence of cited3. First, we investigated muscle cell death in cited3 morphants by using in situ cell death detection kit, TMR Red. These embryos were stained with anti-Mef2 antibody to identify muscle cells. cited3 morphants had increased cell death in the muscle (Fig. 3D–F) compared to control MO injected embryos (Fig. 3A–C) and this phenotype was rescued in the embryos coinjected with cited3 RNA (Fig. 3G–I). We quantified the number of apoptotic non-muscle versus muscle cells in all three treated groups. At 1 dpf, apoptotic cells were not found in the muscle of all three groups of embryos (data not shown). However, we found significantly higher number of apoptotic muscle cells in cited3 morphants at 2 dpf and 3 dpf (Fig. 3M). Simultaneously, we did BrdU pulses in these embryos from 30 hpf to 42 hpf and 54 hpf to 66 hpf. We found that the numbers of BrdU positive cells were not significantly different in all the three treated groups (Fig. 3J–L; data not shown). Correlated with this finding, we still detect generation of thin secondary myofibers due to cell proliferation in cited3 morphants after 1 dpf (data not shown). Hence, these results demonstrate that the absence of cited3 leads to muscle cell apoptosis but it does not affect proliferation of the cells.

Next, we quantified the number of slow myofibers through 1–4 dpf by using slow muscle specific S58 antibody staining. cited3 knockdown resulted in statistically significant reduction in the number of slow myofibers starting at 3 dpf, which was rescued by coinjecting cited3 RNA (Fig. 4). cited3 knockdown also prevented the growth of myofibers; the width and length of individual fibers and the width of somites were reduced significantly at 5 dpf (supplementary material Fig. S3). Correspondingly, the heights of somites were shorter in embryos that were injected with the cited3 MO compared to those that were injected with the control MO (supplementary material Fig. S3). Altogether, these defects resulted in reduced motility in cited3 morphants, which became totally immotile at 5 dpf as assessed by touch-evoked escape response test (supplementary material Fig. S4). The motility defect was significantly rescued by the coinjection of cited3 RNA together with the cited3 MO (supplementary material Fig. S4).

In order to identify where cited3 fits in the gene regulatory network controlling muscle cell differentiation, we assessed expression levels of a set of muscle genes via in situ hybridization. We found that knockdown of cited3 does not affect transcriptions of muscle regulatory genes myod (Fig. 5A,B) or myf5 (Fig. 5C,D). But, differentiation marker myogenin transcripts were retained in cited3 morphants at a stage when it is downregulated in control embryos (Fig. 5E,F). A similar effect on myogenin expression was previously detected when the expression of zebrafish mef2 genes were lost (Hinits et al., 2009). Therefore, we checked the expressions of zebrafish mef2 genes. Transcription of mef2ca was reduced in the cited3 morphants compared to the control MO-injected embryos (Fig. 5G,H). Coinjection of cited3 RNA along with cited3 MO restored the mef2ca transcription (Fig. 5I). On the other hand, the transcription of mef2d was not perturbed (Fig. 5J–L).

We also validated our findings by immunostaining with the general Mef2 antibody that recognizes both Mef2c and Mef2d proteins in zebrafish (Hinits and Hughes, 2007). We found a significant reduction in the levels of total Mef2 proteins in cited3 morphants and this was rescued by coinjection of cited3 RNA (Fig. 5M–O). We also detected a similar phenotype in the heart, where Mef2 protein levels were reduced in heart cardiomyocytes of the cited3 morphants. This phenotype was also rescued by coinjection of cited3 RNA (Fig. 5M′–O′). Altogether, these results demonstrate that cited3 functions upstream of mef2ca in the heart and skeletal muscle.

In order to identify whether Cited3 activates the expression of mef2c cell-autonomously or not, we carried out cell transplantation experiments. As the anti-Mef2 antibody recognizes both Mef2c and Mef2d proteins and there are cell-to-cell variations in its staining, we quantified the intensity of Mef2 staining in both the transplanted and 8–10 closest neighboring slow myofibers that are located in the same somite. We transplanted presumptive slow muscle cells from donor wild-type embryos that are injected with the rhodamine dextran fluorescent dye into control MO injected host embryos of similar stage (Fig. 5P–R′). There was no significant difference in Mef2 protein levels between the transplanted wild-type slow myofiber and the slow myofibers in host embryos (Fig. 5Q; see supplementary material Fig. S5 for quantification). Similarly, we transplanted presumptive slow muscle cells from wild-type donors into cited3 morphant hosts (Fig. 5S–U′) and found that Mef2 protein levels were retained in the wild-type transplanted cells that are surrounded by the morphant cells in which Mef2 protein levels were lower (Fig. 5T and quantified in Fig. 5Y). Whereas donor fibers that are transplanted from cited3 MO injected embryos into wild-type hosts (Fig. 5V–X′) depicted a reduction in Mef2 protein levels (Fig. 5W and quantified in Fig. 5Z). Thus, these data demonstrate that cited3 activates the expression of mef2c cell-autonomously in slow myofibers.

Expression levels of prox1, smyhc1, alpha actinin, myhz1, myhz2 and mylc2 were not affected at 19 hpf (supplementary material Fig. S6). As differentiation proceeds further, transcription levels of smyhc, stnnc, tnnt1, myhz2, mylc2, tnnt3b and myhz1 are declining in wild-type embryos at 2 dpf. However, their transcription levels were retained in cited3 morphants at the same stage (Fig. 6A–N). Interestingly, fast-muscle specific F310 myosin heavy chain expression is reduced after 2 dpf in cited3 morphants (Fig. 6P) compared to control MO injected embryos (Fig. 6O).

We detected transcription of cited3 only in the slow fiber cells both by regular and fluorescent in situ hybridization. However, starting after 2 dpf, we have detected disturbances in the expression of slow myofiber specific genes smyhc, stnnc and tnnt1 (Fig. 6A–F), fast myofiber specific genes myhz2, mylc2 and tnnt3b (Fig. 6G–L), and myhz1 (which is expressed in both fiber types) (Fig. 6M,N). Expression of fast myosin heavy chain that is detected by F310 antibody staining is also reduced in cited3 morphants at the same stage (Fig. 6O,P). In order to identify whether Cited3 regulates expression of fast-myofiber specific myosin heavy chain cell-autonomously or not, we transplanted presumptive muscle cells from donor wild-type embryos that are injected with rhodamine dextran fluorescent dye into host embryos of similar stage that are injected with the cited3 MO and stained embryos with F310 at 2 dpf. The transplanted wild-type fast myofibers express F310 at low levels that are comparable to their neighboring morphant fast myofibers (supplementary material Table S1; Fig. S7). This result suggests that fast-fiber myosin heavy chain expression is non-cell-autonomously upregulated by cited3. We performed the opposite experiment by transplanting cells from cited3 morphants into wild-type hosts. The transplanted morphant cells expressed fast myosin heavy chain at comparable levels to their neighboring wild-type fast myofibers. Altogether, this result suggests that Cited3 acts non-cell-autonomously to promote terminal differentiation of fast myofibers.

To determine whether mef2c transcription factor is the main target of Cited3, we coinjected mouse Mef2c RNA together with cited3 MO into embryos. We found that Mef2c rescues the morphological phenotypes of the cited3 morphants (Fig. 7A–C). Similarly, Mef2c also rescued expression of myosin heavy chain as detected by F59 staining (Fig. 7D–F). Furthermore, restoring the expression of Mef2c also prevented the increased cell death phenotype that was observed upon knockdown of cited3 as revealed by the tunel staining (Fig. 7G–O). Quantification of apoptotic muscle cells confirmed that cited3 maintains the muscle cells by activating the expression of mef2c as Mef2c could rescue the cell-death phenotype of cited3 morphants (Fig. 7P).

The muscle regulatory factors (MRFs) MyoD, Myf5, Myogenin and Mrf4 initiate myogenesis (Parker et al., 2003; Berkes and Tapscott, 2005; Hinits et al., 2009). Mef2 proteins are expressed during terminal differentiation of muscle cells (Black and Olson, 1998). Expression of Mef2c is activated by the MRFs both in mice and zebrafish (Wang et al., 2001; Hinits et al., 2009). Later on, Mef2 proteins interact with the MRFs to cooperatively activate late-stage muscle differentiation genes (Penn et al., 2004). Mef2c is required for heart looping morphogenesis and formation of the right ventricle in mice (Lin et al., 1997). mef2c gene is duplicated in zebrafish. Morpholino-mediated knockdown of mef2ca (Ghosh et al., 2009) or mef2cb (Lazic and Scott, 2011) are reported to result in pericardial edema or cell addition defects from the secondary-heart field, respectively. mef2ca^−/−; mef2cb^−/− double mutants had reduced expression of sarcomeric genes in zebrafish heart (Hinits et al., 2012). Skeletal muscle-specific deletion of Mef2c in mice results in pups that were slightly smaller than the wild-type littermates and the mutant mice died within the first day after birth. At earlier stages of development, Mef2c-deficient myocytes differentiate and fuse normally; however, muscles gradually become disorganized at later stages (Potthoff et al., 2007). Double knockdown of mef2ca and mef2d did not affect the initial differentiation and fusion of myoblasts, however, the myoblasts failed to complete sarcomere assembly and double morphant fish larva also exhibit motility defects starting at 1 dpf (Hinits and Hughes, 2007). This phenotype exhibits similarities to the skeletal muscle specific loss of Mef2c in mice.

Here, we showed that expression of cited3 is activated by MyoD/Myf5 and Hedgehog signaling (Fig. 1). As transcription of myod switches off much faster than that of cited3 upon blockage of Hedgehog signaling, it is highly likely that Hedgehog signaling activates expression of cited3 via activating myod and myf5 (Lewis et al., 1999; Osborn et al., 2011). Although the short time interval between reductions of transcription levels of myod to that of cited3 in cyclopamine treated embryos suggests that Myod (and Myf5) directly activates transcription of cited3, we do not have a proof of a direct regulation at the moment. In turn, Cited3 activates expression of mef2ca (Fig. 5G–I, Fig. 8), which results in retainment of myog transcription at 22 hpf as previously observed in zebrafish (Hinits and Hughes, 2007). This results in retainment of expression of muscle structural genes encoding various thin and thick filaments in myofibers at 2 dpf (Fig. 6).

Knockdown of cited3 did not prevent initial differentiation and fusion of myoblasts at 1 dpf; it also did not perturb expression levels of various muscle structural genes at 19 hpf (supplementary material Fig. S6). However, cited3 morphants failed to complete muscle differentiation, had muscle growth and maintenance defects and impaired motility (Figs 1–Fig. 2,Fig. 3,4; supplementary material Figs S3, S4). The onset of muscle differentiation and motility defects in the cited3 morphants was at 2 dpf (Fig. 2; supplementary material Fig. S4), which is later compared to those in the double knockdown of mef2ca and mef2d (Hinits and Hughes, 2007). Likewise, expression levels of various muscle structural genes were perturbed at later stages (F59 staining at 30 hpf, F310 staining and transcription of other genes at 2 dpf) in cited3 morphants (Figs 2,6). This could be due to the persistence of mef2d transcripts in the cited3 morphants (Fig. 5J–K). Strikingly, expression of mouse Mef2c rescued the cited3 morphant phenotype (Fig. 7) suggesting that mef2c is the main transcriptional target of Cited3 in skeletal muscle differentiation. Interestingly, transcription of mef2ca starts at 14 hpf (Hinits and Hughes, 2007), which is soon after cited3 is expressed in the adaxial cells (Fig. 1B) suggesting that mef2ca might be a direct target of Cited3.

Cited3 morphants also had a dilated heart. Due to its small size, fish embryos can develop and survive in the absence of a functional heart and blood circulation as oxygen is obtained via diffusion (Bakkers, 2011). Therefore, the skeletal muscle phenotype is unlikely to result from a defect in circulation. As cited3 is expressed in the heart (Fig. 1) and regulates the levels of Mef2 proteins in the heart (Fig. 5), and mouse Mef2c rescues loss of function phenotype of cited3 in the heart (Fig. 7), mef2c is likely to be the major target of Cited3 also in the heart.

As cited3 is functionally important for both heart and skeletal muscle development, Cited3 protein might interact with different tissue-specific transcription factors in each tissue or interact with the same transcription factor that is expressed in both tissues. At this point, we do not know which transcription factor Cited3 interacts with and also do not have a proof that Cited3 directly activates transcription of mef2ca. We will investigate these important questions and the detailed characterization of the heart phenotype in cited3 morphants in our future studies.

Although cited3 is expressed only in the precursors of slow myofibers, loss of its expression also resulted in defects in fast myofibers. In addition to slow myofibers, apoptotic cells are also detected among fast myofibers after 2 dpf (Fig. 3). Expression levels of fast-fiber specific genes are also perturbed starting at 2 dpf (Fig. 6). Our cell transplantation experiments revealed that Cited3 activates expression of Mef2 proteins cell-autonomously in the slow myofibers (Fig. 5). However, expression of fast fiber specific myosin heavy chains (detected by F310) is non-cell-autonomously regulated by Cited3 (supplementary material Fig. S7; Table S1). Altogether, this result suggests that Cited3 acts non-cell-autonomously to promote terminal differentiation and maintenance of fast myofibers. Slow fibers are the first differentiating fiber type in zebrafish and their differentiation trigger timely differentiation of the progenitors of fast myofibers (Henry and Amacher, 2004). Our results suggest that slow myofibers play an instructive role in terminal differentiation of fast myofibers even after their initial migration (Henry and Amacher, 2004).

Muscle diseases develop due to genetic mutations, metabolic disorders and aging. Development of cellular therapies in which functional muscle cells are introduced into the patient muscles is one of the most promising avenues for treating muscle diseases. Recent progress in the efficiency of direct transdifferentiation of cells into a different cell fate (Davis et al., 1987; Takeuchi and Bruneau, 2009; Forsberg et al., 2010; Harvey, 2010; Ieda et al., 2010; Vierbuchen et al., 2010) demonstrated the importance of identifying gene clusters functioning during embryonic development of each tissue type. Discovering functionally significant genes in muscle development, likewise, has the potential to result in significant cell-based therapeutic advancement (Parker et al., 2003; Snider and Tapscott, 2003; Tajbakhsh, 2009). Strikingly, Mef2c, which we now have identified to be the target of Cited3, was previously demonstrated to cooperate with MyoD and Myogenin to enhance transdifferentiation of fibroblasts into muscle cells (Molkentin et al., 1995).

The functional roles of Cited-family genes in skeletal muscle development have so far not been investigated in any organism. However, expression of Cited2 was reduced in Duchenne and limb-girdle muscular dystrophy patients (Pescatori et al., 2007; Sáenz et al., 2008). In contrast, overexpression of Cited2 prevented drug-induced muscle atrophy in cell culture (Tobimatsu et al., 2009). Interestingly, loss of Mef2 genes was also implicated in muscular atrophy in mammals (Yamakuchi et al., 2000; Uozumi et al., 2006; Tessier and Storey, 2010). This is the first report, where we demonstrate that expression of cited3 is critical for muscle cell differentiation and maintenance and it acts through activating the expression of mef2ca.

We thank Dr Monte Westerfield, Simon Hughes, Stephen Devoto, Stephen Tapscott, Deborah Yelon and Phillip Ingham for plasmids; Bruce Appel (for the hsp70:gal4VP16), José Campos-Ortega (for the UAS: myc-notch1a-intra∧kca3), Kenneth Poss (for the hsp70:dnfgfr1-EGFP), and Randall Moon (for the hsp70:wnt8a-GFP and hsp70l: ΔTcf-GFP) transgenic lines; Dr Stemple and the Wellcome Trust Sanger Institute for the myf5hu2022 mutant line. We also thank Burcu Guner-Ataman, Xin Li, Becky Weiss, Aaron Sarvet, Abdulvahap Sahin, Betul Sisman, Adem Demir, Rini Dcunha, Murat Isbilen, Cagri Kurt, Zeynep Ulupinar and Fatih Keles for technical help; Dr Florence Marlow for her support during the start-up phase of our lab; and members of Özbudak, Marlow and Jenny labs at the Albert Einstein College of Medicine for helpful discussions.