Scleraxis-lineage cells are required for correct muscle patterning

The musculoskeletal system is a complex multi-tissue system that consists of muscle, tendon and bone, as well as associated connective tissues. Elaborate body movements of vertebrates are supported by the precise shape, position, and functional connections of these components. Although the differentiation process of each component has been well studied (Asahara et al., 2017; Buckingham and Rigby, 2014; Kozhemyakina et al., 2015), their tissue integration process remain largely unexplored. Most skeletal muscles that exist in the mammalian body are derived from somites. Myogenic progenitor cells migrate long distances from somites to destinations such as the limbs, form a precise bundle shape, and attach with appropriate tendons and bones during embryonic development (Buckingham et al., 2003; Comai and Tajbakhsh, 2014). Although the guidance cues for the migration of myogenic progenitor cells from somite to limb have been identified as SF/HGF (also known as HGF) and SDF-1 (CXCL12) (Dietrich et al., 1999; Griffin et al., 2010), mechanisms that regulate muscle shape and tissue integration between post-migration muscle cells and tendons and bones are not fully understood (Kardon, 2011; Schweitzer et al., 2010). Considering the distinct cellular origins (i.e. muscles are from the somites whereas tendons, bones and connective tissues are from the lateral plate mesoderm) and large number of integrations to be formed in limited time and space, it is reasonable to assume that the local tissue–tissue interactions between myogenic progenitor cells and the surrounding cells/tissues are important during these processes. Indeed, several factors, including transcription factors in limb mesenchyme, extracellular matrix (ECM) and muscle connective tissue (MCT) cells, have been reported to regulate muscle patterning in vertebrate limbs (Besse et al., 2020; Hasson et al., 2010; Helmbacher and Stricker, 2020; Kardon et al., 2003; Kutchuk et al., 2015; Rodriguez-Guzman et al., 2007; Swinehart et al., 2013). However, whether tendon cells, the intrinsic partner of skeletal muscles cells, have any role in instructing the muscle shape and patterning in mammals is unclear.

Scleraxis (Scx), which encodes a basic helix-loop-helix transcription factor, is an early marker gene expressed in tendon progenitor cells (Brent et al., 2003; Schweitzer et al., 2001). Scx is expressed in virtually all tendon/ligament cells in the vertebrate embryo and is necessary for the development of force-transmitting tendons (Murchison et al., 2007; Schweitzer et al., 2001; Shukunami et al., 2018). As such, the Scx promoter/enhancer is the most commonly used Cre driver to induce recombination in tendon/ligament tissue (Best et al., 2021; Schlesinger et al., 2021; Sugimoto et al., 2013a,b; Yoshimoto et al., 2017). Because Scx is also expressed in non-tendon/ligament tissues, including chondrocytes of the rib cage, interstitial cells of muscles, and transiently in tendon–bone interfaces (Pryce et al., 2007; Schlesinger et al., 2021; Sugimoto et al., 2013a), we designate cells affected by ScxCre as ‘Scx-lineage cells’. In this study, to elucidate the role of Scx-lineage cells in muscle patterning, we performed tissue-ablation experiments using the ScxCre-L Tg (Sugimoto et al., 2013a) mouse strain. As a result of Scx-lineage cell ablation, muscle bundle shapes and attachment patterns were significantly altered in the embryo. Muscles in the forelimb showed impaired bundle separation and limb girdle muscle attachment sites were mislocated. We also examined which cell population of Scx-lineage cells was mainly reduced by the ablation, and reduction of which cell population could affect the muscle patterning. After careful examination of lineage-tracing and immunostaining data, we conclude that the muscle-patterning defect seen in embryos with Scx-lineage cell ablation is primarily attributed to the reduction of tendon/ligament cells. Our results indicate that Scx-lineage cells have an important instructive role in spatially precise muscle attachment patterns, and are, in turn, essential for reproducible and robust musculoskeletal morphogenesis.

To induce cell death in Scx-lineage cells, we mated a ScxCre-L Tg mouse (Sugimoto et al., 2013a) with a Rosa26-LSL-DTA mouse (Voehringer et al., 2008) (hereafter ‘Scx-DTA’ mice). We examined the tendon tissue of Scx-DTA embryos using the Mkx-Venus knock-in allele (Ito et al., 2010). As shown in Fig. 1A,E, the long tendons in the zeugopod were reduced in embryonic day (E) 17.5 Scx-DTA embryos. Limb sections showed that the flexor digitorum profundus (FDP) tendon (Fig. 1B,F) and flexor digitorum sublimis (FDS) tendon (Fig. 1C,G) in the autopod, and extensor carpi radialis tendons, extensor digitorum communis (EDC) tendons, and the palmaris longus tendon (Fig. 1D,H) in the zeugopod were greatly reduced in Scx-DTA embryos. Reduction of tendon tissue in other parts of the body, such as the Achilles tendon (Fig. S1D,H) and tail tendons (Fig. S1E,I), was also apparent. Furthermore, ligamentous tissues, such as the cruciate and patella ligaments, were diminished (Fig. S1F,J) and the outer annulus fibrosus of the intervertebral disc was also reduced (Fig. S1G,K). To gain further insight into quantitative and temporal aspects of tendon cell reduction in Scx-DTA embryos, we examined Scx⁺ cell number by immunofluorescence and cell death by cleaved caspase 3 immunofluorescence and terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL) assay in tendon cells. We found that Scx⁺ cells were significantly reduced in E12.5 limbs (Fig. 1K-L′,O, 45% reduction, P<0.05), E14.5 limbs (Fig. 1R-S′,V, 40% reduction, P<0.05) and diaphragm (Fig. S4J-K′,Q, 58% reduction, P<0.01), but not in E11.5 limbs (Fig. 1I-J′,M). Although not statistically significant, Mkx⁺ cells were also reduced in E14.5 limbs (Fig. 1P-Q′,U, 26% reduction, P=0.074). Consistent with this, significant increases in cell death were observed in Scx⁺ cells in E12.5 limbs (Fig. 1K-L′,N, 896% increase in Scx⁺/TUNEL⁺ cells, P<0.05; Fig. S1A-C, 742% increase in TUNEL⁺ cells in Scx⁺ region, P<0.05), Mkx⁺ cells in E14.5 limbs (Fig. 1P-Q′,T, 275% increase in Mkx⁺/cleaved-caspase3⁺ cells, P<0.05) and E14.5 diaphragm (Fig. S4N-P, 480% increase in Mkx⁺/cleaved caspase 3⁺ cells, P<0.05). Together, these results illustrated that our approach was able to successfully induce cell death in tendon cells and in turn reduce the tendon and ligament tissues from the developing embryo, albeit not completely.

We found that Scx-DTA pups died soon after birth (Fig. S2A,B). Loss of the rib cage or defects in the diaphragm (see below) could have caused insufficient respiratory function in newborn Scx-DTA pups. Thus, in the present study, we focused on developmental aspects of Scx-lineage depletion.

Because Scx is also expressed in non-tendon/ligament cells such as chondrocytes, muscle interstitial cells, and limb mesenchyme cells that transdifferentiate into myofibers (Lima et al., 2021; Sugimoto et al., 2013a,b; Yoshimoto et al., 2017), we next wanted to examine the range of cell lineages that have ScxCre expression history and determine whether those cell types were also reduced in Scx-DTA mice. In ScxCre-L;Ai14;Scx-GFP forelimbs, tdtomato signal (i.e. Scx expression history) was detected in GFP-positive (i.e. currently expressing Scx) tendon/ligament cells (Fig. 2A-C″, white arrowheads) as well as other surrounding cell types at E14.5 (Fig. 2A-C″, yellow arrowheads). In particular, Sox9⁺ chondrocytes and Tcf4⁺ MCT cells were partially tdtomato⁺ (Fig. 2D-E′). A small fraction of myosin heavy chain (MHC)⁺ myofibers also showed weak tdtomato signal (Fig. 2F). Thus, we sought to examine whether each of these cell types was affected in Scx-DTA embryos. For Sox9⁺/tdtomato⁺ chondrocytes, our observation probably reflected the Sox9⁺/Scx⁺ double-positive progenitor cells previously reported (Blitz et al., 2013; Sugimoto et al., 2013b). Skeletal elements, such as the rib cage or deltoid tuberosity, were reduced in Scx-DTA embryos, likely as a result of Scx expression in chondrocytes (Fig. S2C-F). For Tcf4⁺/tdtomato⁺ MCT cells, as the relative expression pattern of Tcf4 and Scx was not clear, we examined the expression of Scx and Tcf4 in embryonic limbs. We found that Scx and Tcf4 expressions were partially overlapping at E12.5, but were mutually exclusive at E14.5 (Fig. 2G-H″), likely indicating that the early bipotential progenitor cells differentiated into either Scx⁺ tendon or Tcf4⁺ MCT cells. The Tcf4⁺/tdtomato⁺ cells observed at E14.5 were likely the descendants of these Scx⁺/Tcf4⁺ progenitor cells. Consistent with this, cleaved caspase 3⁺ apoptotic cells were seen in Scx⁺/Tcf4⁻ or Scx⁺/Tcf4⁺ domains, but not in the Scx⁻/Tcf4⁺ domain in E12.5 Scx-DTA embryonic limbs (Fig. S3A-L′). Importantly, this early cell death in Scx⁺/Tcf4⁺ cells had little impact on Tcf4⁺ MCT cells at later stages. Tcf4⁺ cells were detected at a similar level in Scx-DTA limbs and diaphragms compared with control embryos at E14.5 (Fig. S4E,F,L-M′), whereas Scx⁺ tendon cells were reduced (Fig. S4C,D,J-K′,Q). Because MCT cells were heterogenic and Tcf4⁺ cells represent only a subpopulation, mRNA levels of MCT marker genes expressed in different populations (i.e. Tcf4, Osr1 and Osr2) were examined. We found that all the MCT marker genes were expressed at comparable levels between Scx-DTA and control limbs, whereas tendon cell markers (i.e. Scx, Mkx and Tnmd) were significantly reduced (Fig. S4G). MHC⁺/tdtomato⁺ myofibers were probably descendants of Scx⁺ limb mesenchymal cells that fused into myofibers during embryonic development (Lima et al., 2021). However, cell death in the myogenic lineage of Scx-DTA embryos [i.e. MyoD (Myod1)⁺/cleaved caspase 3⁺ myocytes in E12.5 limbs and MHC⁺/cleaved caspase 3⁺ myofibers in E14.5 diaphragms] showed no significant increase compared with control (Figs S5A-C, S4H-I′,R). Consistent with this, the number of MyoD⁺ myocytes in E12.5 limbs was not significantly altered (Fig. S5A-B′,D). In summary, cells with Scx expression history include tendon/ligament cells, a subpopulation of chondrocytes, MCT cells and myofibers. Tendon/ligament cells and chondrocytes were the cell types mainly reduced by ScxCreL-Tg-dependent tissue ablation. Because a subpopulation of MCT cells was Scx positive, it is possible that MCT cells could have been reduced in early stages due to a depletion of this subpopulation. However, because we saw no difference in the MCT population at later stages, if there were an early reduction, it was likely compensated for by an increase in Scx-negative subpopulations during later development.

Next, we examined whether the patterning (i.e. shape or attachment) of skeletal muscles was altered in Scx-DTA embryos. First, we performed whole-mount immunohistochemistry with an MHC antibody to analyze postnatal day (P) 0 pup forelimbs. In control limbs, the muscles were located between their regular attachment sites (origin and insertion); for example, the deltoid muscle originated from the spine of the scapula and inserted into the deltoid tuberosity (Fig. 3B,C, blue arrowhead). In contrast, muscles in Scx-DTA mice showed changes in their attachment sites; for example, the insertion site of the deltoid muscle changed to the shoulder joint (Fig. 3F,G, red arrowheads). The shapes/attachments of muscles in the zeugopod also changed; for example, the extensor carpi ulnaris (ECU) and extensor digiti quarti/quinti (EDQ) muscles were clearly separated in control limbs (Fig. 3D, blue arrowheads), but they were fused into a single muscle bundle in Scx-DTA limbs (Fig. 3H, red arrowhead). The extensor pollicis muscles, which are normally covered by superficial muscles, became visible (Fig. 3D,H, arrows). Of note, despite their morphological change, most of the muscles in the zeugopod attached with tendons at their distal ends.

In sections of E18.5 forelimbs, muscles in the metacarpal position were grossly normal (Fig. 3I,L,O). However, FDS muscles were found in the wrist position of Scx-DTA limbs, where muscle was not observed in control embryos (Fig. 3J-M′,O). Conversely, FDS muscles were missing from their normal position in the zeugopods of Scx-DTA mice (Fig. 3K,N,O, arrowheads). FDS muscles uniquely undergo post-fusion migration between E14.5 and E16.5, and, interestingly, similar distal mislocation of the FDS muscles has been reported in Scx-knockout mice due to failure of the FDS tendons (Huang et al., 2013). It is likely that the change in FDS muscle position seen in Scx-DTA mice may also be due to a reduction of the FDS tendons. In addition, the arrangements of the extensor carpi (longus/brevis) and FDP muscles were changed (Fig. 3K,N, arrows). By contrast, muscles in the hindlimb generally showed normal attachment patterns (Fig. S6).

Altered attachment sites were also observed in other muscles, especially in muscles connecting to the body trunk and limbs. For example, the pectoral muscles originated from the sternum and inserted into the deltoid tuberosity in control embryos (Fig. 3P,P′, arrowhead); however, the insertion sites of these muscles were distally dislocated toward the elbow joint in Scx-DTA mice (Fig.  3R,R′, arrowhead). In a severe case, the pectoral minor muscle separated into two bundles (Fig. 3R,R′, arrows). Similarly, in Scx-DTA mice, the insertion site of the gluteus maximus muscle, which normally originates from the ilium and inserts into the gluteus tuberosity (Fig. 3Q,Q′, arrowhead) changed distally toward the knee joint (Fig. 3S,S′, arrowhead). The organization of muscles in the trunk, such as the rectus abdominis, was also changed in Scx-DTA mice (Fig. 3T,V). In the diaphragm, the width of the central tendon was greatly reduced (Fig. 3U,W, arrows) and the orientation of associated muscle fibers was changed (Fig. 3U,W, arrowheads). Although we observed patterning defects consistently in mutants, the phenotypes observed (e.g. differences in altered attachment sites) varied between embryos. In summary, the organization of skeletal muscles in many body parts was changed in Scx-DTA mice, suggesting that Scx-lineage cells are necessary for skeletal muscles to form precise shapes and attachments in the correct positions.

Tendons and ligaments were clearly reduced in size and cell number in the Scx-DTA mutant mice; however, we also observed reductions in skeletal elements, such as in the ribcage and deltoid tuberosity (Fig. S2C-F). We therefore wanted to examine whether the muscle patterning defects could be due to skeletal malformation. To test this hypothesis, we examined Myf5-deficient (Myf5^Cre/Cre) embryos and Sox9 heterozygous (Sox9^fl/+;Meox2Cre) embryos in which loss of the rib cage and deltoid tuberosity, respectively, were previously reported (Braun et al., 1992; Kist et al., 2002). We found that the embryos exhibited the previously reported skeletal defects, but, importantly, the insertion sites of the pectoralis major muscles were not altered in either mutant embryo (Fig. S7A-H). In addition, previous studies have also reported that changes introduced in skeletal elements did not affect the surrounding muscle patterning (Li et al., 2010; Tokita and Schneider, 2009). These results suggest that the loss of skeletal elements is not the primary cause of the muscle patterning defect seen in Scx-DTA embryos.

Programmed cell death has been reported to affect limb morphogenesis and muscle belly segregation (Guha et al., 2002; Rodriguez-Guzman et al., 2007). Also, although we did not detect a significant increase of myogenic cell death in Scx-DTA embryos, it is possible Scx⁺ cells that fuse into myofibers (Lima et al., 2021) could induce cell death in muscle cells. Therefore, we wanted to examine whether cell death in muscle could cause the muscle pattern alterations seen in Scx-DTA mutants. To explore this possibility, we examined the muscle patterning of Myf5^Cre/+:Rosa26-LSL-DTA embryos, hereafter ‘Myf5-DTA’, in which myocytes are ablated in early development. Myf5-DTA embryos showed a significant reduction of MyoD⁺ myocytes (60%, P<0.05) with an increase of MyoD⁺/cleaved caspase 3⁺ cells (529%, P<0.01) at E12.5 (Fig. S7I-L). However, the pectoralis major muscle was correctly inserted into the deltoid tuberosity (Fig. S7M,N, arrowheads). Considering that there was neither significant reduction of myocytes nor increase in cell death in the myogenic lineage of Scx-DTA mutants (Figs S5A-C, S4H-I′,R), it is unlikely that cell death in the myogenic cells is causing the muscle patterning phenotype seen in the Scx-DTA mutant. Although these experiments cannot completely rule out the possibility of myogenic or chondrogenic contributions to the Scx-DTA patterning phenotype, neither a reduction in skeletal elements nor cell death in muscle itself showed the muscle patterning phenotype seen in the Scx-DTA mutant. Together, these results support the hypothesis that it is the reduction of tendon/ligament cells that is the primary cause of the muscle patterning defects in Scx-DTA mice.

Next, to elucidate the stage at which muscle interacts with Scx-lineage cells, we monitored the location and shape of the myogenic cell lineage in a step-by-step manner. First, we found that Pax3-positive myogenic progenitor cells migrated normally into the limb buds at E11.5 (Fig. 4A,D, arrowheads). Myogenin-positive myoblast positions were grossly normal at E12.5 and E13.5 (Fig. 4B,C,E,F), suggesting that myoblast segregation occurred in a similar manner in control and Scx-DTA mice. Furthermore, Myh3-positive myofibers were observed in the same position in Scx-DTA and control embryos at E12.5 (Fig. 4G,J), indicating that myofiber differentiation is not dependent on Scx-lineage cells. Of note, we found that the Myh3 signal became significantly elevated at the distal/proximal ends of myofibers from E13.5 (Fig. 4H,I arrowheads) (Tanji et al., 2023). By comparing the expression of Myh3 and a mature tendon marker, tenomodulin (Tnmd) (Shukunami et al., 2001, 2006), we noted that this high Myh3 expression marks the boundary of muscle and tendon (Fig. S8). We therefore interpreted the region of high Myh3 expression as a surrogate of the muscle–tendon boundary. The shape of Myh3-positive muscle bundles clearly differed between Scx-DTA and control embryos at E13.5 (Fig. 4H,K). Whereas the EDC muscle in control embryos formed a sharp boundary at its distal tip (Fig. 4H, arrowheads), the distal tip of the EDC in Scx-DTA embryos remained a loose shape without a clear boundary formation (Fig. 4K, arrowheads). Moreover, whereas the ECU and EDQ muscle bundles in control embryos were clearly separated and formed two distal boundaries (Fig. 4H, blue arrowheads), those muscles did not separate or form normal distal boundaries in Scx-DTA embryos (Fig. 4K, red arrowheads). The muscle patterning defects became more apparent at E14.5, when EDC, EDQ and ECU muscles could not be distinguished (Fig. 4I,L, arrowhead). Immunohistochemistry with MHC antibodies confirmed that the shapes of MHC-positive muscle bundles were loose in Scx-DTA embryos (Fig. 4M-P) and the insertion sites of several muscles, including the deltoid muscles (Fig. 4M-P, arrowheads), were dislocated. Because the reduction of Scx⁺ cells was evident at E12.5 and later, these results indicate that Scx-lineage cells are dispensable for the initial segregation of myoblasts in the limb bud and differentiation of myofibers, but are required for the formation of proper muscle bundle morphologies and the precise location of the attachment sites. Conversely, as Scx⁺ cells were not reduced at E11.5 (Fig. 1I-J′,M), the requirement of Scx-lineage cells for the migration of myogenic progenitor cells to the limb would need to be examined using a different approach.

The results shown above implied that skeletal muscles define their attachment sites after myofiber formation (e.g. Fig. 4G-L). We then investigated whether the position of muscle could be altered after formation of the insertion, which may confer further plasticity and robustness to skeletal muscle patterning. To examine this, we observed the position and insertion of the gluteus maximus muscle at two time points: E14.5 and E17.5. As shown in Fig. 5A,D, the distal tip of the gluteus maximus was located at the middle of the femur (i.e. the gluteus tuberosity) at E14.5 in both control and Scx-DTA mice (Fig. 5A,D). A detailed section analysis confirmed that the gluteus maximus muscles formed insertions with the gluteus tuberosity through tendons at this stage, although the tendons were reduced in size in Scx-DTA mice (Fig. 5B,C,E,F). As embryonic development proceeded, the junction of the gluteus maximus muscle/tendon matured and was firmly inserted into the gluteus tuberosity at E17.5 in control mice (Fig. 5G-I). By contrast, the tip of gluteus maximus muscle of Scx-DTA mice was distally dislocated and most of the tendon and insertion into the gluteus tuberosity was lost by this stage (Fig. 5J-L). These results imply that the skeletal muscle is able to reposition its attachment site according to environmental changes, even after myofiber differentiation or the formation of an initial attachment.

To gain insights into the molecular aspects of Scx-lineage cell ablation and their interactions with muscle cells, RNA-sequencing (RNAseq) analysis was conducted using E13.5 forelimbs. A total of 666 genes were significantly upregulated and 151 genes were significantly downregulated [fold change >1.5, false discovery rate (FDR)<0.01]. Because we wanted to investigate the muscle and Scx-lineage interaction that is reduced in Scx-DTA embryos, we focused on genes downregulated in Scx-DTA limbs. As expected, marker genes of tendon/ligament cells, such as Fmod (Svensson et al., 1999), Tnmd (Shukunami et al., 2001), Htra3 (Havis et al., 2014) and Ssc5d (Liu et al., 2021), or of chondrocytes, such as Col2a1 (Li et al., 1995a,b), Col9a1 (Fässler et al., 1994), Col11a1 (Li et al., 1995a,b) and Wwp2 (Inui et al., 2018) (Table 1, Table S1), were included in the downregulated genes. In contrast, genes related to myogenic or MCT genes were not identified (Table 1, Table S1). Gene Ontology (GO) enrichment and KEGG pathway analyses using g:Profiler (Raudvere et al., 2019) indicated significant enrichment (P<0.05) of 30 biological processes and five pathways (Tables 2 and 3). The enriched biological processes included ‘cartilage development’, ‘tendon development’ and ‘connective tissue development’, as expected from Scx-lineage ablation. In addition, pathways such as ‘ECM–receptor interaction’ and ‘focal adhesion’ were identified in the KEGG pathway analysis, implying that genes related to cell attachment to the ECM are enriched in Scx-lineage cells, and/or those genes have roles in muscle patterning. Interestingly, ‘negative regulation of BMP signaling pathway’ and ‘response to BMP’ were identified in the GO enrichment analysis. BMP signaling was reported to be active at the tendon–muscle interface (Lima et al., 2021; Wang et al., 2010) and thus may have a role in tendon–muscle interaction. Further investigation is required to fully reveal the interaction mechanisms between muscle and Scx-lineage cells.

In this study, the role of Scx-lineage cells in muscle patterning of mouse embryos was revealed by a lineage-ablation approach. The role of Scx-lineage cells or tendon cells on muscle patterning has been shown indirectly in other vertebrate species; for example, surgical removal of the tendon primordia resulted in ectopic extension of the muscle into the knee joint in chicken embryos (Kardon, 1998). In addition, Scx-knockout zebrafish embryos showed abnormal muscle patterning (Kague et al., 2019) and Scx-lineage cell ablation in postnatal zebrafish resulted in muscle patterning defects (Niu et al., 2020). In mammals, muscle–tendon interaction has been mostly studied in the muscle-to-tendon direction, but the effect of tendon cells on muscle patterning has been less studied (Brent et al., 2005; Pryce et al., 2009). Postnatal ablation of Scx-lineage cells has been shown to alter tendon collagen fibril diameter and density, but no muscle patterning defect has been reported (Best et al., 2021). Our results represent the first evidence of the importance of Scx-lineage cells in muscle patterning in mammalian embryos.

The patterning of limb muscles is regulated by surrounding lateral plate mesoderm-derived limb mesenchyme at various levels (Helmbacher and Stricker, 2020). Transcription factors that are broadly expressed in limb mesenchyme, such as Tbx5, Hox11 genes, Shox2 and Lmx1b, regulate the patterning of both tendon and muscle (Chen et al., 1998; Hasson et al., 2010; Swinehart et al., 2013; Vickerman et al., 2011). MCT cells expressing Osr1, Osr2 and Tcf4 affect myoblast clustering and muscle splitting (Besse et al., 2020; Kardon et al., 2003; Vallecillo-García et al., 2017). Loxl3-positive mesenchymal cells that fuse into myofibers are involved in correct myotendinous junction formation (Kraft-Sheleg et al., 2016; Yaseen et al., 2021). Our results imply that Scx-lineage cells play their role to finalize the muscle bundle position and attachment after muscle segregation is initiated by MCT. Of note, Scx is expressed in cell types other than tendon/ligament cells and therefore it is possible that ablation of those non-tendon/ligament cell types may be causing the muscle patterning defects in Scx-DTA mice. In particular, expression in MCT cells (Besse et al., 2020; Kardon et al., 2003; Vallecillo-García et al., 2017) or a subpopulation of Scx-lineage cells that fuses into myofibers (Grimaldi et al., 2022; Lima et al., 2021; Yaseen et al., 2021) could induce muscle cell death and should be carefully considered. In our experiments, we did not observe a reduction of MCT cells in Scx-DTA mice (Fig. S4). We also did not find a significant reduction of myogenic cells in Scx-DTA mice (Fig. S5) and no pectoral muscle patterning defects were seen even when myogenic cells were ablated in Myf5-DTA mutant mice (Fig. S7). Therefore, although the possible contribution of these cell types to the Scx-DTA mice phenotype cannot be completely excluded, our results support the idea that the phenotypes in Scx-DTA mice, at least those of limb girdle muscles and limbs, are caused by cell ablation in tendon/ligament tissues specifically. Further study would be necessary to elucidate fully the contribution of each cell type to muscle patterning, possibly by using Cre drivers with additional temporal and cell-type specificity. We assume that multiple inputs from surrounding cells regulate limb muscle patterning sequentially and in an overlapping manner, which provides musculoskeletal morphogenesis reproducibility and robustness against environmental or genetic perturbations. Limb girdle muscles, such as deltoid, pectoral and gluteus muscles, were significantly affected by loss of the Scx lineage, whereas axial muscles and hindlimb muscles were less affected. Moreover, limb girdle muscles in Scx-DTA embryo exhibited variability in altered morphology, whereas ECU/EDQ muscles in the Scx-DTA mutant zeugopod showed reproducible morphology, albeit different from control. Currently, the reasons underlying these differences are not clear, but could include technical reasons (such as differences in Scx expression level/Cre activity, degree of tendon cell reduction, etc.), and/or intrinsic differences between individual tendons (such as long versus short anchoring tendons, etc.) and muscles (long migrating muscles versus axial muscles, etc.). Characterization of individual tendons/muscles from a morphogenetic point of view will be an interesting issue to be explored in future studies.

Although tendon/ligament tissue reduction in Scx-DTA was incomplete (approximately 40-60%; Fig. 1, Figs S1, S4), the reduction was as severe as or more severe than that of previously reported mice with knockout of tendon transcription factors (Guerquin et al., 2013; Ito et al., 2010; Murchison et al., 2007). In particular, the reduction of Scx⁺ cells was observed as early as E12.5 in Scx-DTA embryos, which is earlier than the reported tendon reductions in TF mutants (Guerquin et al., 2013; Ito et al., 2010; Murchison et al., 2007). Incomplete loss of tendon tissue could be due to the continuous recruitment of Scx-positive cells from a Scx-negative cell population (Huang et al., 2019; Shwartz et al., 2016) or the penetrance of Cre activity (Comai et al., 2014; Tan et al., 2020, 2021). The fact that the degree of tendon cell reduction remained similar between E12.5 and E14.5 whereas increased tendon cell death in Scx-DTA was continuously observed implies the existence of continuous recruitment. However, the low percentage of cell death marker-positive tendon cells observed at each time point could indicate limited Cre efficacy. The population dynamics of tendon cells would be an interesting topic to study in the future.

The molecular mechanisms underlying this muscle–tendon interaction remain to be explored. In Scx-DTA mice, muscles in the zeugopod (i.e. the EDC/EDQ/ECU muscles) fused and attached to the remaining tendons at their distal ends (Fig. 3H). The insertion sites of the muscles in the girdle (i.e. the pectoral and gluteus muscles) dislocated and reattached toward the joint regions, such as the shoulder, elbow and knee (Figs 3 and 5, Fig. S9B). The remaining tendon cells were relatively abundant in the joint areas of Scx-DTA embryos (Fig. 1A,E, Fig. S9A), probably owing to the initial amount of cells and continuous cell recruitment/differentiation from Scx-negative cell populations (Huang et al., 2019; Shwartz et al., 2016). The fact that muscle did not attach randomly to nearby bone but instead changed its morphology toward the distal tendon tissue supports the hypothesis that diffusible molecule(s) secreted from tendon cells attract myofibers. Indeed, recent studies reported the involvement of retinoic acid in the formation of the extra-ocular functional unit (Comai et al., 2020), and FGF and BMP signaling has been shown to be active at the interface of the embryonic tendon and muscle (Eloy-Trinquet et al., 2009; Wang et al., 2010). Our RNAseq results also suggest that BMP signaling may have a role in Scx-lineage cells. However, cell adhesion molecules (Hasson et al., 2010) and ECM (Kutchuk et al., 2015) can also play parallel roles. Our RNAseq data also pointed to the importance of cell adhesion and ECM in Scx-lineage cells. Clearly, more studies are required to understand fully the molecular and cellular mechanisms of precise skeletal muscle patterning. We believe that revealing the molecular mechanisms behind this process will shed light on broad areas of biology, such as the diversity of muscle patterning among species and regenerative medicine.

Scx-GFP and ScxCre-L transgenic (Tg) mouse strains have been described previously (Sugimoto et al., 2013a). The Rosa26-LSL-DTA mouse strain was kindly provided by Dr Ohteki (Tokyo Medical and Dental University, Japan). The mouse was originally purchased from The Jackson Laboratory [B6.129P2-Gt(ROSA)26Sortm1(DTA)Lky/J, strain#009669] by Dr Ohteki and transferred under the permission of The Jackson Laboratory. A Mohawk-Venus knock-in mouse was generated and described in our previous study (Ito et al., 2010). Myf5Cre [B6.129S4-Myf5^tm3(cre)Sor/J, strain #007893] and Meox2Cre mice [B6.129S4-Meox2^tm1(cre)Sor/J, strain #003755] (Tallquist and Soriano, 2000; Tallquist et al., 2000) were purchased from The Jackson Laboratory. The Sox9-flox mouse strain was kindly provided by Dr Scherer and Dr Kist (Kist et al., 2002). ICR mice were purchased from Sankyo Lab-Service. The mice were housed under controlled environmental conditions (22-24°C, 40-60% humidity) with free access to water and food. All animal experiments were approved by the animal experiment committees of Meiji University (approval number IACUC17-0007, MUIACUC2022-04) and the National Research Institute for Child Health and Development (approval number A2004-003).

For paraffin-embedded sections, embryos were fixed with 4% paraformaldehyde (PFA; Wako, 162-16055) at 4°C overnight, dehydrated with methanol, cleared with xylene, and embedded in paraffin wax. For cryosections, embryos were fixed with 4% PFA at 4°C overnight, washed with PBS, followed by a sucrose gradient, and embedded with OCT compound (Sakura Finetek, 4583). The embryos were sectioned at 7 µm for paraffin-embedded sections used for Hematoxylin and Eosin (H&E) staining, and at 10 µm for OCT-embedded cryosections used for TUNEL staining and immunofluorescence analyses. The TUNEL analysis was performed using the In-situ Cell Death Detection Kit (Roche, 11684795910) according to the manufacturer's instructions. The names of muscles and tendons appearing in sections were judged according to Watson et al. (2009).

Embryos were fixed with 4% PFA at 4°C overnight, dehydrated with methanol, and rehydrated with PBS supplemented with 0.1% Triton X-100 (PBSTx). Embryos were digested with 10 μg/ml Protease K at 37°C for 60 min, re-fixed with 4% PFA, blocked with 2% bovine serum albumin (BSA) in PBSTx, and incubated with anti-MHC antibody (Sigma-Aldrich, M4276, My-32, 1:1000 in 2% BSA/PBSTx) at 4°C overnight. Then embryos were washed ten times with PBSTx, incubated with AP-conjugated mouse IgG (1:2000 in 2% BSA/PBSTx) (Abcam, ab5880) at 4°C overnight, and washed ten times with PBSTx. The signal was developed in NBT/BCIP solution (Roche, 11681451001).

Paraffin sections and cryosections were prepared as described in the ‘Histological analyses’ section. Cryosectioned slides were washed in PBS, followed by fixation for 5 min in 4% paraformaldehyde. Slides were washed again in PBS and then blocked for 10 min in 2% skim milk in PBS. Primary antibody was diluted in 2% skim milk in PBS and slides were incubated overnight at 4°C in a humidified chamber followed by a PBS wash. Secondary antibody was diluted in 2% skim milk in PBS and slides were incubated for 2 h at room temperature in a humidified chamber. Slides were washed in PBS and then stained with 1:1000 DAPI in PBS for 10 min at room temperature. Slides were washed in PBS and mounted using Fluoroshield mounting media (ImmunoBioScience Corporation, AR-6500-01) for imaging.

Antibodies used were: anti-MHC (Sigma-Aldrich My-32, 1:1000, or Developmental Studies Hybridoma Bank, MF20, 1:100), anti-MyoD (Dako, M3512, 1:100), anti-GFP (Abcam, ab13970, 1:1000), anti-TCF4 (Cell Signaling Technology, 2569, 1:500), anti-cleaved-caspase3 (Cell Signaling Technology, 9664, 1:500), anti-Scx (Kumagai et al., 2012), Cy3-conjugated anti-mouse IgG (Jackson ImmunoResearch, 715-165-150, 1:1000), Alexa Fluor 488-conjugated anti-mouse IgG (Thermo Fisher Scientific, A21202, 1:1000), Alexa Fluor 488-conjugated anti-rabbit IgG (Jackson ImmunoResearch, 111-545-003, 1:1000) and Alexa Fluor 488-conjugated anti-chicken IgY (Jackson ImmunoResearch, 703-545-155, 1:1000).

Fluorescent images of sections were captured using a Zeiss LSM 880 confocal microscope (Carl Zeiss Microscopy). Sections from control and mutant embryos were region matched for analysis. A minimum of three sections per embryo were counted and averaged to provide a single value per mouse.

Cells were manually counted. Positive cells were determined by immunofluorescent cell markers. For most cell death analyses, cell death (TUNEL⁺ or cleaved caspase 3⁺) was determined by an overlap of the cell marker and the cell death marker. For Scx-DTA Scx⁺ regional TUNEL⁺ analysis, a tight boundary was drawn surrounding the region of Scx⁺ cells. All TUNEL⁺ cells in this region were counted, regardless of overlap with Scx⁺.

All experiments were performed in triplicate as a minimum. Paired two-tailed Student's t-tests were performed to determine significance. Significant differences were detected at P<0.05. Results are presented as beeswarm boxplots. The top and bottom of the box are the 3rd and 1st quartile. The upper and lower whiskers indicate the largest and smallest data points in the range from [1st quartile-1.5×(3rd quartile-1st quartile)] to [3rd quartile+1.5×(3rd quartile-1st quartile)], respectively. Graphs were created using R Software (v4.2.2, R Core Team).

Whole-mount in situ hybridization was performed according to the methods described by Yokoyama et al. (2009). Briefly, embryos were fixed with 4% PFA at 4°C overnight, dehydrated with methanol, and rehydrated with PBS supplemented with 0.1% Tween 20 (PBST). Embryos were digested with 10 μg/ml Protease K at 37°C for 20-60 min (depending on the stage), re-fixed with 4% PFA, and hybridized with an anti-sense probe labeled with digoxigenin (DIG) or fluorescein in hybridization buffer at 65°C overnight. Embryos were washed with wash buffer at 65°C, blocked with 10% fetal bovine serum (FBS) in PBST for 2 h, and incubated with anti-DIG or anti-fluorescein antibody conjugated with alkaline phosphatase (Roche, 11093274910, 11426338910, 1:2000 in 10% FBS/PBST) at 4°C overnight. Then embryos were washed ten times with PBST and the signal was developed in NBT/BCIP solution. For double in situ hybridization, DIG-labeled Tnmd probe and fluorescein-labeled Myh3 probe were simultaneously hybridized. Tnmd was stained with anti-DIG-AP antibody and NBT/BCIP substrate, post-fixed with 4% PFA, dehydrated with methanol, and rehydrated with PBST. Myh3 was stained with anti-fluorescein-AP antibody and INT (Tokyo Chemical Industry, TK-B0280) and BCIP (Roche, 11383221001) substrates. Tnmd probe was generated as described before (Shukunami et al., 2001). Myh3 probe was generated as described before (Tanji et al., 2023). Note that the Myh3 probe used in this study may hybridize also with Myh7 and Myh8 mRNAs, which essentially have identical expression patterns as Myh3 (Tanji et al., 2023). The sequences of the primers used for the amplification of probes were as follows: myogenin fw: 5′-ACCTGATGGAGCTGTATGAGACATC-3′, myogenin rev: 5′-CATTTAGGTGACACTATAGCAGATGTGCACACTTGTCCAGG-3′; myh3 fw: 5′-CGTTTTGGACATTGCGGGTT-3′, myh3 rev: 5′-ATGGACTCCCTCCTCTGCAT-3′.

Skin and internal organs were removed from embryos. Embryos were then fixed with 100% ethanol and serially stained with 0.03% Alcian Blue (Sigma-Aldrich, A5268) solution and 0.01% Alizarin Red (Sigma-Aldrich, A5533) solution. When embryos were used after whole-mount immunohistochemistry, the embryos were post-fixed with 4% PFA, dehydrated with ethanol, and stained as described above.

Total RNA was isolated from E13.5 limbs using ISOGEN-II (NIPPON GENE, 311-07361) according to the manufacturer's instructions. cDNA was synthesized using SuperScript II Reverse Transcription enzyme (Thermo Fisher Scientific, 18064071). Quantitative PCR was performed using Power SYBR Green Master Mix (Thermo Fisher Scientific, 4367659). The sequences of the primers used for the amplification were as follows: Scx fw: 5′-GCAGCTGTAGGGCTTGATGT-3′, Scx rev: 5′-GCAGCTGTAGGGCTTGATGT-3′; Mkx fw: 5′-AGTAAAGACAGTCAAGCTGCCACTG-3′, Mkx rev: 5′-TCCTGGCCACTCTAGAAGCG-3′; Tnmd fw: 5′-AACACTTCTGGCCCGAGGTAT-3′, Tnmd rev: 5′-AAGTGTGCTCCATGTCATAGGTTTT-3′; Tcf4 fw: 5′-AAGCCTCCAGAGCAGACAAA-3′, Tcf4 rev: 5′-TAAGTGCGGAGGTGGATTTC-3′; Osr1 fw: 5′-GCACACTGATGAGCGACCT-3′, Osr1 rev: 5′-TGTAGCGTCTTGTGGACAGC-3′; Osr2 fw: 5′-CACACAGACGAGAGGCCATA-3′, Osr2 rev: 5′-GCAGCTGTAGGGCTTGATGT-3′.

Poly(A)+ RNAs were purified from 2 μg total RNA using the NEBNext Poly(A) mRNA Magnetic Isolation Module (New England Biolabs, E7490). RNAseq libraries were constructed using the NEBNext Ultra RNA Library Prep Kit for Illumina (New England Biolabs, 7530) according to the manufacturer's protocol. The libraries were sequenced with 150-bp paired-end reads for each sample and biological triplicates per sample using an Illumina Novoseq. Low-quality reads were removed using prinseq-lite software (Schmieder and Edwards, 2011). Trimmed reads were aligned to the mouse reference genome (mm10) using HISAT2 (Kim et al., 2019). Aligned reads were converted and sorted to Bam files using SAMtools software (Li et al., 2009) and counted with featurecounts software (Liao et al., 2014). Statistical analysis of differentially expressed genes was performed using EdgeR software (Robinson et al., 2010) with exactTest. GO analysis was performed with g:profiler (Raudvere et al., 2019).

We thank Dr Scherer (University of Freiburg), Dr Kist (Newcastle University), Dr Ohteki (TMDU) and Dr Kanagawa (Ehime University) for providing the Sox9 flox, Rosa26-LSL-DTA and Myf5Cre mice.

The peer review history is available online at https://journals.biologists.com/dev/lookup/doi/10.1242/dev.201101.reviewer-comments.pdf.