A change in cis-regulatory logic underlying obligate versus facultative muscle multinucleation in chordates

In vertebrates, multinucleated myofibers are formed through fusion of mononucleated myoblasts. Myomaker (Mymk) is a transmembrane protein required for myoblast fusion and muscle multinucleation (Millay et al., 2013). In tunicates, the sister group to vertebrates (Delsuc et al., 2006; Putnam et al., 2008), Mymk is also required for myoblast fusion and muscle multinucleation (Zhang et al., 2022). Mymk from tunicate species such as Ciona robusta can rescue cell fusion in Mymk CRISPR knockout myoblasts in diverse vertebrate species, suggesting highly conserved function (Zhang et al., 2022). Phylogenomic analyses indicate that Mymk (previously referred to as Tmem8c) arose in the last common ancestor of tunicates and vertebrates through duplication of an ancestral Tmem8 gene, and is not found in other invertebrates, including cephalochordates (Zhang et al., 2022).

However, unlike mammalian Mymk, which is expressed in all skeletal muscles, Ciona Mymk is exclusively expressed in the differentiating precursors of multinucleated, post-metamorphic (i.e. juvenile/adult) muscles and not in those of mononucleated larval tail muscles (Zhang et al., 2022). Transcription of Mymk in mammalian skeletal myoblasts is carried out by myogenic regulatory factor (MRF) family members, especially MyoD1, which function as the molecular switch for muscle specification and differentiation (Zhang et al., 2020). Most tunicates, including Ciona, have a biphasic life cycle transitioning from a swimming larval phase to a sessile filter-feeding adult phase (Karaiskou et al., 2015). Their larvae have muscles in their tail that are specified by the Ciona MRF ortholog (Meedel et al., 2007). However, unlike the post-metamorphic muscles of the adult body wall and siphons, they do not express Mymk and do not undergo cell fusion or multinucleation (Zhang et al., 2022). We therefore sought to understand the molecular mechanism underlying this muscle subtype- and life cycle stage-specific activation of Mymk and myoblast fusion in tunicates.

Here, we describe the cis- and trans-regulatory bases of Mymk expression specifically in the post-metamorphic muscles of Ciona. First, we show that, as in vertebrates, MRF is required for Mymk activation in Ciona. However, in Ciona the transcription factor early B-cell factor (Ebf, also known as Collier, Olf1, EBF or COE) is required in combination with MRF to activate Mymk transcription. Mis-expressing Ebf together with MRF in the larval tail and other tissues is sufficient to activate ectopic Mymk transcription. We show that these effects are recapitulated even by using human MYOD1 and EBF3 in Ciona, whereas Ciona MRF alone is sufficient to activate human MYMK in cultured myoblasts. Finally, we identify the likely binding sites for MRF and Ebf in the Ciona Mymk promoter, suggesting that differences in cis (promoter sequences), not in trans (transcription factor protein-coding sequences), are the primary drivers of evolutionary change between facultative and obligate Mymk expression and myoblast fusion in the chordates.

Transcription of Mymk in mammalian skeletal myoblasts is carried out by bHLH transcription factors of the MRF family (Zhang et al., 2020), which are part of the major molecular switch for muscle specification and differentiation (Hernández-Hernández et al., 2017). In Ciona, MRF is the sole ortholog of human MRF family members MYOD1, MYF5, MYOG and MRF4, and is necessary and sufficient for myoblast specification in the early embryo (Hernández-Hernández et al., 2017; Meedel et al., 2007, 1997, 2002). In tunicates, MRF is expressed in larval tail muscles and in the atrial and oral siphon muscles (ASMs and OSMs, respectively) of the post-metamorphic juvenile/adult (Razy-Krajka et al., 2014). Yet Mymk is expressed only in the ASMs/OSMs (Zhang et al., 2022), suggesting the regulation of Mymk in tunicates is different from that of vertebrates as MRF alone is not sufficient to activate Mymk in larval tail muscle (Fig. 1A,B).

Comparing multinucleated post-metamorphic and mononucleated larval muscles, one key molecular difference between the two that we hypothesized might determine the selective regulation of Mymk is the expression of Ebf in the former, but not the latter (Stolfi et al., 2010) (Fig. 1A). Ebf orthologs have been frequently associated with myogenic activity throughout animals. In models such as Drosophila and Xenopus, Ebf orthologs are upstream of or in parallel to MRF in muscle development (Dubois et al., 2007; Enriquez et al., 2012; Green and Vetter, 2011). In Ciona, Ebf specifies post-metamorphic muscle fate (Stolfi et al., 2010; Tolkin and Christiaen, 2016) and activates both MRF and ASM-specific gene expression (Razy-Krajka et al., 2014). Therefore, MRF and Ebf were the prime candidates for post-metamorphic muscle-specific activation of Mymk.

To test whether MRF is necessary for Mymk expression in post-metamorphic Ciona muscles, we targeted the MRF locus using tissue-specific CRISPR/Cas9-mediated mutagenesis (Stolfi et al., 2014). We specifically targeted the B7.5 lineage that gives rise to ASMs using the Mesp promoter (Davidson et al., 2005) to drive Cas9 expression in this lineage. To target MRF, we used a combination of two sgRNAs (U6>MRF.2 and U6>MRF.3) that had been previously designed and validated (Gandhi et al., 2017). We allowed animals to develop into metamorphosing juveniles at 46 h post-fertilization (hpf) and scored the expression of a previously published Mymk>GFP reporter plasmid in Mesp>mScarlet+ ASMs (Fig. 1C). In MRF CRISPR juveniles, Mymk>GFP expression is nearly extinguished, as we observed only 1% of mScarlet+ juveniles showing GFP expression in the MRF CRISPR juveniles compared with 95% in the negative control condition (Fig. 1D). These data strongly suggest that MRF is necessary for Mymk transcription in Ciona post-metamorphic muscles, as in vertebrate skeletal muscles.

We were not able to use the same approach to test the requirement of Ebf for Mymk expression, as Ebf is required for MRF activation in post-metamorphic muscle progenitors (Razy-Krajka et al., 2014), and CRISPR/Cas9-mediated disruption of Ebf results in loss of MRF expression, converting siphon muscles to heart cell fate (Stolfi et al., 2014). Instead, to test whether the combination of Ebf and MRF is sufficient to activate Mymk expression, we forced their combined expression in different larval cells. In addition to being expressed in the progenitors of multinucleated post-metamorphic muscles, Ebf is also expressed in the central nervous system of the larva, where it is important for cholinergic gene expression and motor neuron development (Kratsios et al., 2012; Popsuj and Stolfi, 2021). In contrast, MRF is expressed in the tail muscles of the larvae, where Ebf is not expressed. Therefore we sought to ectopically express Ebf in MRF⁺ tail muscles, and MRF in Ebf⁺ neural progenitors.

We first drove ectopic Ebf expression in the larval tail muscles using the extended MRF promoter (Zhang et al., 2022). Ectopic expression of EBF in larval tail muscles with MRF>Ebf resulted in transcription of the Mymk reporter in the tail muscles of 100% of successfully transfected larvae compared with 0% in negative control larvae (Fig. 1E,F). Similarly, we overexpressed MRF in the larval central nervous system using an Ebf promoter (Stolfi and Levine, 2011). In this case, we tested two different transcript variants (Tv) of MRF, MRF-Tv1 and MRF-Tv2, which differ in the length of the encoded C termini with slightly different functional properties (Izzi et al., 2013). Strong ectopic Mymk reporter expression was seen in the larval nervous system with either variant of MRF (MRF-Tv1: 100% Mymk>GFP+; MRF-Tv2: 98% Mymk>GFP+) (Fig. 1G,H). In negative control larvae, scattered weak Mymk>GFP expression was visible in the nervous system in only 7% of transfected larva. In contrast, overexpression of MRF in the larval tail muscles using an MRF>MRF-Tv1 construct resulted in Mymk>GFP expression in only 8% of larvae, mostly comprising strong expression in certain neurons (presumably Ebf⁺) that also had leaky MRF promoter activity (Fig. S1). We conclude that increased MRF dose does not adequately replace the function of Ebf. Taken together, our results suggest that MRF and Ebf co-expression is sufficient for Mymk activation in Ciona, in the absence of any repressive inputs.

To test whether this MRF-Ebf cooperativity we observe is unique to the Ciona proteins, we replaced Ciona Ebf and MRF in the above experiments with their human homologs EBF3 and MYOD1. We then assayed whether they can cooperate with endogenous Ciona MRF or Ebf and activate the Ciona Mymk reporter in the larval tail muscles or central nervous system. Remarkably, both replacements resulted in strong ectopic Mymk reporter expression (Fig. 1E-H). This implies that the species origin of the proteins themselves does not seem to matter as long as an MRF ortholog and an Ebf ortholog is expressed in the same cell. This suggested that the difference in obligate versus facultative Mymk expression is likely due to changes in cis (i.e. different binding sites in the Mymk promoter) rather than in trans (i.e. change to MRF or EBF family proteins) between tunicates and vertebrates.

Using a similar logic from the previous experiment, we introduced Ciona MRF (Tv2) and/or Ebf into CRISPR-generated, MYOD1-deficient human myoblasts to see whether they (alone or in combination) are sufficient to activate the expression of human MYMK (Fig. 2A). Remarkably, transfected Ciona MRF alone resulted in nearly identical levels of MYMK mRNA expression as human MYOD1 (Fig. 2B). In contrast, when Ciona Ebf was expressed alone, no significant MYMK mRNA expression was detected, and Ebf in combination with MRF resulted in a significant reduction of MRF efficacy, appearing to hamper its activation of MYMK. These data further support the idea that changes in cis, and not in trans, underlie the differential requirement of Ebf in activating Ciona Mymk.

Ebf has been established as an important regulator of post-metamorphic muscle fate in Ciona (Kaplan et al., 2015; Razy-Krajka et al., 2018, 2014; Stolfi et al., 2010; Tolkin and Christiaen, 2016). When we ectopically expressed Ebf in larval tail muscle cells, we observed a striking change in their morphology (Fig. 3A). Typical larval tail muscles are mononucleated with defined polygonal shapes and cell-cell junctions (Passamaneck et al., 2007), but Ebf expression suppressed these features. Instead, the tail muscle cells became more reminiscent of post-metamorphic siphon and body wall muscles, becoming elongated and myofiber like, losing their characteristic polygonal shape, and gaining more and smaller nuclei. Although we could not detect any clear instances of tail muscle cells fusing, these observations suggest that the combination of MRF and Ebf might be activating the expression of additional determinants of post-metamorphic muscle-specific morphogenesis.

To identify other putative targets of MRF-Ebf cooperativity, we performed bulk RNAseq to compare the transcriptomes of wild-type embryos with embryos in which Ebf was overexpressed in the tail muscles. We extracted RNA at 11 hpf from whole embryos that were transfected with either MRF>Ebf or MRF>CD4::GFP as a negative control. Using DESeq2 to analyze the resulting RNAseq data, we observed significant upregulation of many genes in the MRF>Ebf condition. Of the 16,433 genes detected, 163 were significantly upregulated (P<0.00001, logFC>0), of which 33 showed a logFC greater than 1.0 (Fig. 3B, Table S2). In contrast, 155 genes were significantly downregulated (P<0.00001, logFC<0). Interestingly, high-ranking genes that had both a significant P-value and log fold change ≥1 include Tropomyosin.a (Tpm.a), Col24a-related, myosin heavy chain.c (Myh.c) and myosin light chain.d (Myl.d), which have all been confirmed as upregulated specifically in the ASMs by in situ hybridization (Razy-Krajka et al., 2014). Mymk was 25th in this list, confirming that endogenous Mymk (and not only the Mymk>GFP reporter) is ectopically activated in tail muscles upon Ebf overexpression. This suggests that Ebf is sufficient to partially convert larval tail muscle cells into post-metamorphic, atrial siphon-like muscles.

To further investigate this muscle subtype fate change, we compared our EBF overexpression bulk RNAseq results with a published microarray analysis of Ebf overexpression in the trunk ventral cells (TVCs) that give rise to both heart and ASM progenitors (Razy-Krajka et al., 2014). Indeed, when comparing both datasets, many of the top genes in our list were also significantly upregulated by Ebf overexpression in the TVCs, resulting in a Pearson correlation coefficient (ρ) of 0.48 (Fig. 3C, Table S3). Although there are several genes that show discrepant changes in expression between the two datasets, this may reflect differences in the timing of RNA extraction and territory of Ebf overexpression (tail muscles at 11 hpf versus TVCs at 21 hpf). In fact, there are TVC-specific factors that are missing from the tail muscles and that could alter the response of the cell to Ebf overexpression (Davidson et al., 2006; Racioppi et al., 2019). Taken together, these data suggest that Mymk is only one of several genes that might be preferentially activated in post-metamorphic, multinucleated muscles by a similar MRF-Ebf combinatorial logic in Ciona.

Because we suspected vertebrate-tunicate differences in Mymk activation to be due primarily to differences in cis, we aligned the Ciona robusta Mymk promoter to the homologous sequence from the related species Ciona savignyi (Satou et al., 2008, 2019, 2022; Vinson et al., 2005) to identify potentially conserved transcription factor binding sites (Fig. 4A). We also used JASPAR (Castro-Mondragon et al., 2022), a predictive binding site search algorithm, to look for putative MRF and Ebf sites (Chaudhary and Skinner, 1999; Treiber et al., 2010). This led us to identifying MRF⁻¹³⁶ and Ebf⁻¹¹⁶ as conserved, high-scoring candidate binding sites to test (Fig. 4A, Table S4). We made mutations predicted to disrupt MRF or Ebf binding to these putative sites in the Mymk>GFP reporter plasmid, and scored the co-expression of these mutant reporters with a wild-type Mymk>mCherry reporter [Mymk(WT)>mCherry]. When observing juveniles electroporated with the Mymk>GFP reporter bearing the MRF⁻¹³⁶ mutation, it was clear that its activity was significantly reduced, with only 19% of Mymk(WT)>mCherry+ siphon muscles also faintly expressing GFP (Fig. 4B,C). As expected, we also observed dramatic reporter expression loss with the Ebf⁻¹¹⁶ mutation (20% GFP expression, although mutating both MRF⁻¹³⁶ and Ebf⁻¹¹⁶ did not further abolish the residual GFP expression) (Fig. 4B,C). This residual, faint GFP expression might represent basal plasmid expression, or even ‘cross-talk’ (i.e. transvection) with the co-transfected mCherry plasmid. In contrast, 100% of juveniles co-expressed wild type GFP and mCherry reporters. Similarly, mutation of a nearby poorly conserved, low-scoring predicted MRF site (MRF⁻¹⁵²) did not significantly reduce reporter activity, suggesting it is not required for activation and likely not a functional MRF-binding site (Fig. 4D,E). Taken together, these data suggest that Ciona Mymk activation is dependent on closely spaced conserved MRF and Ebf predicted binding sites in its proximal promoter region.

When examining the Mymk promoter for potential transcription factor-binding sites, we noticed a conserved Ebox sequence further upstream in the Mymk promoter (Fig. 4A). We initially thought it could be an MRF-binding site, but JASPAR predictions revealed a much higher score for binding by Hairy Enhancer of Split (HES) transcriptional repressor family members (Fig. 4A). In Ciona, HES has been shown to mediate Delta/Notch-dependent repression of MRF expression and myogenic differentiation in the inner ASM precursor cells, prolonging their undifferentiated proliferative state (Razy-Krajka et al., 2014). This expression was shown to be downstream of Ebf via Delta signals from outer ASM progenitors, which are the first myoblasts to differentiate. In inner cells, the Delta/Notch-HES pathway temporarily downregulates MRF in and their descendants, which was proposed to promote a stem cell-like state (Razy-Krajka et al., 2014). In vertebrates, Delta/Notch signaling also represses MyoD expression and muscle differentiation (Delfini et al., 2000). In chick, HEYL (a HES homolog) binds to the Mymk promoter and inhibits its transcription, hinting at a deeply conserved strategy for restricting the onset of Mymk expression and fusion in developing myoblasts (Esteves de Lima et al., 2022). When we tested a Mymk GFP reporter plasmid carrying a mutation to disrupt this upstream Ebox, we observed a significant increase in the frequency of GFP expression compared with the wild-type reporter (Fig. 4F,G). Increased GFP expression suggests that this site is most likely bound by a repressor. Our results suggest that the direct repression of Mymk transcription by HES repressors (Fig. 4H) may have been an ancestral trait present in the last common ancestor of tunicates and vertebrates.

What might be the exact cis-regulatory change that result in the difference observed between tunicate and vertebrate Mymk regulation? Unfortunately, reporter constructs made using published human MYMK (Zhang et al., 2020) or chicken Mymk (Luo et al., 2015) promoters were not expressed at all in Ciona tail muscles (Fig. S2). This was not entirely surprising, given that orthologous promoters are frequently incompatible (i.e. ‘unintelligible’) even between different tunicate species, due to developmental system drift (Lowe and Stolfi, 2018). We therefore focused instead on testing different point mutations in the Ciona Mymk promoter that might result in ectopic activation in larval tail muscles.

Cis-regulatory logic can be complex with subtle changes in promoter sequences resulting in drastically different activation patterns (Spitz and Furlong, 2012). In tunicates, it has been shown that by making changes to the sequences flanking a given transcription factor binding site, one can increase its predicted binding affinity, resulting in higher expression levels or ectopic activation (Farley et al., 2015, 2016; Jindal et al., 2023). To test whether such ‘optimized’ MRF and Ebf-binding sites might result in activation of the Ciona Mymk reporter by MRF or Ebf alone (without the need for MRF-Ebf cooperation), we manipulated flanking sequences of putative MRF- or Ebf-binding sites, resulting in higher binding affinity scores predicted by JASPAR (Fig. 5A,C, Fig. S3).

Optimization of the conserved Ebf⁻¹¹⁶ site did not significantly increase Mymk>GFP activation in the central nervous system (Fig. 5A,B). This suggested that either the Ebf site is already ‘optimal’ or that Ebf binding affinity is not rate-limiting in this context. However, optimization of the conserved indispensable MRF⁻¹³⁷ site and/or the non-conserved, dispensable MRF⁻¹⁵² site resulted in significant Mymk>GFP expression in tail muscles (Fig. 5C-F). Interestingly, optimization of MRF⁻¹⁵² resulted in visible GFP expression in 95% of electroporated larval tails, whereas optimization of MRF⁻¹³⁶ resulted in GFP expression in a more modest 27% of tails (Fig. 5D). Because the increase in average predicted JASPAR score was most pronounced between the wild-type MRF⁻¹⁵² (JASPAR score 2.61) and its ‘optimized’ counterpart (JASPAR score 12.8, Fig. S3), this suggested that creating an additional high-scoring MRF-binding site is particularly effective for switching a combinatorial MRF-Ebf transcriptional logic to an MRF-alone logic. This switch in logic was confirmed when we observed Mymk reporter expression in post-metamorphic muscles, even when combining optimized MRF sites with a disrupted Ebf site (Fig. 5G,H). Interestingly, combining both optimized MRF⁻¹⁵² and MRF⁻¹³⁶ sites resulted in ectopic reporter activation in neurons, in addition to tail muscles, in 44% of larvae (Fig. 5F). This expression might be due to greater affinity for proneural transcription factors that also bind Ebox sequences, such as neurogenin (Kim et al., 2020). This suggests that the exact sequences flanking each site might also be under purifying selection, minimizing ectopic activation of Mymk in tissues where its expression might be detrimental.

In this study, we have investigated the cis-regulatory logic of muscle subtype-specific Mymk expression in Ciona. We have identified two essential transcriptional regulators, MRF and Ebf, that together activate the transcription of Mymk, which encodes a transmembrane protein that drives myoblast fusion and muscle multinucleation in tunicates and vertebrates (Zhang et al., 2022). This is in stark contrast to human MYMK expression, which requires only the activity of MRFs (Zhang et al., 2020). In fact, overexpression of Ebf was detrimental to MYMK expression and to myoblast differentiation and fusion in our human cell culture assay. We believe this may be an indirect effect and not due to direct repression of MYMK by Ebf. Ebf2 is a key activator of adipocyte cell fate, and it is possible that overexpression of Ciona Ebf in MYOD1 CRISPR knockout cells might be converting them to adipocytes. Alternatively, Ebf overexpression might be activating the Delta-Notch pathway, which would indirectly downregulate MYMK expression.

We have also revealed a potentially conserved repressive input into Ciona Mymk transcription, in which direct binding and repression by HES factors might restrict the spatiotemporal window of Mymk expression and, consequently, of myoblast fusion. This repression, likely mediated through Delta-Notch signaling, might pre-date the divergence of tunicates and vertebrates. It was shown that Delta-Notch/HES plays a role in temporarily delaying post-metamorphic Ciona muscle differentiation in subsets of progenitor cells (Razy-Krajka et al., 2014). More specifically, transient Delta-Notch signaling was shown to suppress differentiation in committed atrial siphon and body wall muscle progenitors. Because fully differentiated, multinucleated myofibers cannot proliferate, this might ensure that these post-metamorphic muscles are still allowed to grow while also undergoing differentiation. We propose that Delta-Notch/HES might similarly delay Mymk activation in committed post-metamorphic progenitors, as fusion would only be enabled in cells undergoing differentiation, not during myoblast proliferation. In fact, it has been shown that the Delta-Notch/HES pathway similarly suppresses and/or delays muscle differentiation in vertebrates to maintain a population of muscle stem cells (Mourikis et al., 2012); more specifically it represses Mymk activation during this process (Esteves de Lima et al., 2022). Thus, the role of Delta-Notch/HES in temporarily repressing Mymk expression seems to be part of a deeply conserved mechanism for balancing growth and/or differentiation.

We propose that the difference between ‘MRF+Ebf’ and ‘MRF-alone’ logic is responsible for the difference between the pan-skeletal muscle expression of vertebrate Mymk and the more-selective, post-metamorphic muscle-specific expression in Ciona (Fig. 6A), a difference we were able to experimentally mimic through increasing MRF-binding site affinities in the Ciona Mymk reporter (Fig. 6B). This in turn might underlie the difference between obligate (vertebrate) versus facultative (tunicate) muscle multinucleation. Although Mymk overexpression is not sufficient to drive the fusion of Ciona larval tail muscle cells, Ciona Mymk is sufficient to induce human myoblast fusion (Zhang et al., 2022). Our RNAseq results show that this same MRF-Ebf logic is regulating a larger suite of post-metamorphic muscle-specific genes, some of which might encode additional factors required for myoblast fusion.

Although we have largely revealed the basis of Mymk regulation in Ciona, there are still details that have yet to be elucidated. For example, what is the mechanism of MRF-Ebf cooperative activity? MRF-Ebf synergy in transcription of muscle subtype-specific genes has been reported in mammals, e.g. the activation of the Atp2a1 gene by MyoD1 and Ebf3 specifically in mouse diaphragm muscles (Jin et al., 2014). Myod1 and Ebf3 and its homologs alone have the ability to activate Atp2a1, but this expression was substantially higher when both MyoD1 and Ebf3 were present. However, there was no evidence of the two transcription factors directly contacting one another to drive cooperative binding. This may be similar to the mechanism of activation of Mymk by MRF and Ebf in Ciona. Optimization of the predicted Ebf site in the Mymk promoter did not significantly increase reporter plasmid expression in Ebf⁺ larval neurons. This suggests that MRF and Ebf might act on different steps of Mymk activation in Ciona (i.e. chromatin accessibility versus RNA polymerase recruitment). For example, it has been shown that EBF1 acts a ‘pioneer’ factor in human pro-B cell lineages (Li et al., 2018), and is required continuously for stable chromatin accessibility at key B-cell lineage genes (Zolotarev et al., 2022). In other words, the MRF-Ebf combinatorial logic we have revealed might not be dependent on cooperative binding, as is the case for other examples of cooperativity (Zeitlinger, 2020).

Given the difference between vertebrates (obligate) and tunicate (facultative) activation of Mymk, we present two hypotheses for how regulation of Mymk might have been controlled in the last common ancestor of tunicates and vertebrates (Fig. 6C). In the first scenario, the ancestor would have been more like tunicates, in which the combination of MRF and Ebf would have cooperatively activated Mymk in only a subset of muscles. In the second scenario, the ancestor would have been more like vertebrates, in which MRF alone would have activated Mymk in all muscles.

In the first scenario, the last common ancestor would have had both mononucleated and multinucleated muscles, as we see in most tunicates. It is unclear whether the common ancestor had a biphasic life cycle or not, but it is likely they had separate lineages for the trunk and pharyngeal muscles, as seen in both vertebrates and tunicates (Razy-Krajka et al., 2014). The ancestor may have had specialized pharyngeal muscles homologous to the siphon muscles of tunicates. One key feature of tunicate siphon muscles is that they are formed by a series of concentric circular myotubes. It is possible that the ancestor had a similar set of circular muscles around the openings of a pharyngeal atrium, and the process of Mymk-driven myoblast fusion might have evolved to allow the formation of such muscles. After splitting from tunicates, vertebrates would have lost the requirement of Ebf for Mymk expression, and MRF would have become the sole activator of Mymk, allowing all muscle cells to become multinucleated. This may have been advantageous for their survival, perhaps permitting larger myofibers throughout the body and advanced movement capabilities.

Alternatively, the last common ancestor might have only had multinucleated muscles under the regulation of MRF alone, as in extant vertebrates. Later, vertebrates would have kept this mode of regulation, whereas tunicates would have recruited Ebf to activate Mymk only in post-metamorphic muscles, as an adaptation specifically tied to their biphasic life cycle. As it stands, we do not have enough evidence to conclusively favor one evolutionary scenario over the other. On the vertebrate side, there are no reports of muscle subtype-specific fusion as far as we can tell. On the tunicate side, with the exception of groups that have generally lost the larval phase (e.g. salps and pyrosomes), there are no reports of obligate myoblast fusion. However, we have shown that a switch to pan-muscle expression of Mymk is possible through ‘optimization’ of putative MRF-binding sites in its promoter, or by creating an additional high-scoring predicted MRF site (Fig. 6B). Whether this is actually a recapitulation of what happened in evolution or not, we may never know.

Ciona robusta (intestinalis Type A) specimens were obtained and shipped from San Diego, California, USA (M-REP). The eggs were fertilized, dechorionated and subjected to electroporation using established methods as detailed in previously published protocols (Christiaen et al., 2009a,b). The embryos were then raised at a temperature of 20°C. At various stages, including embryonic, larval and juvenile, the specimens were fixed using MEM-FA solution (composed of 3.7% formaldehyde, 0.1 M MOPS at pH 7.4, 0.5 M NaCl, 1 mM EGTA, 2 mM MgSO₄ and 0.1% Triton-X100), followed by rinsing in 1×PBS with 0.4% Triton-X100 and 50 mM NH₄Cl to quench autofluorescence, and one final wash in 1×PBS with 0.1% Triton-X100.

Imaging of the specimens was carried out using either a Leica DMI8 or DMIL LED inverted epifluorescence microscope. Scoring was carried out only on mCherry⁺ individuals so as to exclude potentially unelectroporated animals, unless otherwise noted in the figure legends. To carry out CRISPR/Cas9-mediated mutagenesis of MRF in the B7.5 lineage, we used Mesp>Cas9 to restrict Cas9 expression to this lineage (Stolfi et al., 2014), together with previously validated MRF-targeting sgRNA plasmids U6>MRF.2 and U6>MRF.3 (Gandhi et al., 2017). For the negative control, a previously published U6>Control sgRNA vector was used that expresses an sgRNA predicted not to target any sequence in the C. robusta genome (Stolfi et al., 2014). The sgRNAs are expressed in vivo from plasmids using the ubiquitous RNA polymerase III-transcribed U6 small RNA promoter (Nishiyama and Fujiwara, 2008). Mutations to disrupt or optimize putative binding sites were all generated through de novo synthesis and custom cloning by Twist Bioscience. All GFP or mCherry sequences fused to the N-terminal Unc-76 extranuclear localization tag (Dynes and Ngai, 1998), unless otherwise specified. All plasmid, protein and sgRNA sequences, and electroporation mixes can be found in the supplementary Materials and Methods. All statistical tests are summarized in Table S1.

Human MYOD1-knockout myoblasts were generated by CRISPR-Cas9-mediated gene editing and cultured as described previously (Zhang et al., 2020). Retroviral expression vector pMXs-Puro (Cell Biolabs, RTV-012) was used for cloning and the expression of human MYOD1, Ciona MRF (Transcript Variant 2) and Ciona Ebf. The DNA sequences were verified by Sanger sequencing. For the myogenic rescue experiments, the sgRNA-insensitive version of human MYOD1 open reading frame was used. Retrovirus was produced through transfection of HEK293 cells using FuGENE 6 (Promega, E2692). Two days after transfection, virus medium was collected, filtered and used to infect human myoblasts assisted by polybrene (Sigma-Aldrich, TR-1003-G). When the culture reached 80-90% confluency, cells were induced for myogenic differentiation by switching to myoblast differentiation medium (2% horse serum in DMEM with 1% penicillin/streptomycin). Human myoblasts were differentiated for 3 days and used for immunostaining and RNA extraction. For immunostaining, the primary antibody for myosin (Developmental Studies Hybridoma Bank, MF20) and the primary antibody for myogenin (Developmental Studies Hybridoma Bank, F5D) were used. The qPCR primers for measurements of human MYMK and 18S expression are provided in the supplementary Materials and Methods.

Total RNA was extracted at 11 h post-fertilization (Stage 23, late tailbud) from two independent replicates each of electroporated larvae that were transfected either with 50 g MRF>CD4::GFP (negative control) or 50 g MRF>Ebf transcript variant 1 (Ebf overexpression). Library preparation was carried out at the Georgia Tech Molecular Evolution Core Facility, as previously described (Johnson et al., 2024), and sequencing was carried out on an Illumina NovaSeq 6000 with a SP PE100bp run. Reads were processed and differential gene expression analysis was performed using DESeq2 in Galaxy, as previously described (Johnson et al., 2023 preprint). KY21 gene model ID numbers (Satou et al., 2022) were matched to KH gene model ID numbers (Satou et al., 2008) using the Ciona Gene Model Converter application: https://github.com/katarzynampiekarz/ciona_gene_model_converter (Johnson et al., 2023 preprint). Our RNAseq analysis was also compared with published microarray analysis of Ebf perturbations in FACS-isolated cardiopharyngeal lineage cells (Razy-Krajka et al., 2014). Volcano plots and comparative transcriptome plots were constructed using R studio and Bioconductor (Huber et al., 2015) with packages EnhancedVolcano (Blighe et al., 2018) and ggplot2 (Wickham, 2016). Raw sequencing reads have been deposited in NCBI BioProject under accession number PRJNA1068599.

The authors are indebted to Dr Florian Razy-Krajka for thoughtful discussion on Ebf function in specifying post-metamorphic muscle identity in Ciona. We thank Lindsey Cohen for technical assistance and Shweta Biliya for help with RNAseq library preparation and sequencing in the Molecular Evolution Core Facility at Georgia Institute of Technology. We thank Brian Hammer, Annalise Paaby and Will Ratcliff for critical reading and helpful suggestions.

The peer review history is available online at https://journals.biologists.com/dev/lookup/doi/10.1242/dev.202968.reviewer-comments.pdf