Edited by Eric N. Olson, University of Texas Southwestern Medical Center, Dallas, TX, and approved November 14, 2006 (received for review September 21, 2006)
In vertebrates, skeletal myogenesis is initiated by the generation of myoblasts followed by their differentiation to myocytes and the formation of myofibers. The determination of myoblasts and their differentiation are controlled by muscle regulatory factors that are activated at specific stages during myogenesis. During late embryonic and fetal stages a distinct population of resident proliferating progenitor cells is the major source of myogenic cells. How the differentiation of myoblasts and progenitor cells is regulated is not clear. We show that in mouse embryos the Notch ligand Delta1 (Dll1) controls both differentiation of early myoblasts and maintenance of myogenic progenitor cells. Early dermomyotome-derived myoblasts are determined normally in Dll1 mutant embryos, but their differentiation is accelerated, leading to a transient excess of myotomal muscle fibers. Similarly, migratory hypaxial myogenic cells colonize the limb buds and activate muscle regulatory factor expression normally, but muscle differentiation progresses more rapidly. Resident progenitor cells defined by Pax3/Pax7 expression are formed initially, but they are progressively lost and virtually absent at embryonic day 14.5. Muscle growth declines beginning around embryonic day 12, leading to subsequent severe muscle hypotrophy in hypomorphic Dll1 fetuses. We suggest that premature and excessive differentiation leads to depletion of progenitor cells and cessation of muscle growth, and we conclude that Dll1 provides essential signals that are required to prevent uncontrolled differentiation early and ensure sustained muscle differentiation during development.
In vertebrates, all skeletal muscles of the body are derived from the somites, metameric mesodermal structures that form on both sides of the neural tube. Somites differentiate under the influence of signals from the surrounding tissues ventrally to the mesenchymal sclerotome, which will give rise to the axial skeleton, and dorsally to the dermomyotome, which will give rise to all skeletal muscles of the trunk, tail, and limbs (1). Myogenesis is initiated at the dorsomedial quadrant of the somites followed by muscle cell differentiation at the medial and lateral dermomyotomal lips, respectively. Cells originating from the medial lip give rise to the epaxial, cells from the lateral lip to the hypaxial muscles (2, 3).
Muscle development is under the control of the muscle regulatory factors (MRFs) Myf5, MyoD, Myogenin, and MRF4 (4). Activation of Myf5 and MyoD appears to determine the myogenic fate because their forced expression in various cell types is sufficient to induce muscle differentiation (5); and when both genes are disrupted, no muscle precursor cells and consequently no skeletal muscles form (6). In contrast, Myogenin and MRF4 regulate myoblast differentiation and subsequent myotube maturation (7–10). In addition, signals mediated by the evolutionary conserved Notch pathway have been implicated in the regulation of myogenic differentiation in vertebrate embryos and cultured cell lines (11–15). Activation of Notch in C2 myoblast cells can inhibit expression of MRFs and other muscle-specific genes, and it can block cell fusion and myotube formation (11–13). The block of myogenic differentiation appears to be mediated by repression of MRF expression by Hes1 (16, 17) as well as by direct interaction of activated Notch with the myocyte enhancer factor 2C (18). In Xenopus embryos, MyoD directly regulates expression of the Notch ligand Delta1 and thereby activates E(spl)-related Notch target genes (19), suggesting a feedback loop between myogenic regulatory basic helix–loop–helix proteins and Notch signaling during vertebrate myogenesis and a potential role for Notch in myogenic determination similar to Drosophila (20). However, inhibiting or stimulating Notch activity in Xenopus and zebrafish embryos disrupted the segmental arrangement of myogenic cells, but it did not affect their formation or differentiation (21–24); and overexpression of Delta1 in chick embryos did not affect early steps of myogenesis, but it blocked the differentiation of postmitotic myogenic cells (14). This observation suggests that in avian embryos, Notch regulates terminal muscle differentiation, and Notch signaling may have varying roles during myogenesis in different vertebrate species. The physiological role(s) of Notch activity during mammalian embryonic myogenesis has not been defined yet. Here, we show that Delta1 (Dll1), one of the five Notch ligands in mice, has essential functions in regulating mammalian myogenesis in vivo.
Dll1 transcripts were detected in myotomes at embryonic day (E) 10.5 (25) and in skeletal muscles at later stages of development (26). In differentiating somites, Dll1 is expressed at low levels in cells emerging from the dorsomedial lip [supporting information (SI) Fig. 5a ] and subsequently throughout myotomes (SI Fig. 5 b–g and k ). In the limb buds, Dll1 is found in myoblasts (Myf5-positive cells) and myocytes (Myogenin-positive cells; compare SI Fig. 5 h–j ) but not in the progenitor (Pax3- and Lbx1-positive) cells that migrate from the hypaxial somite domain (compare SI Fig. 5 l, q, and v ) and initiate differentiation after they have reached their target (27, 28). Notch2 and Notch3 are expressed in myogenic regions (SI Fig. 5 r–u ), suggesting that Dll1 could regulate myogenic differentiation by interaction with Notch2 and/or 3, potentially in concert with Jag2, whose expression was detected in myotomes but not in myogenic cells in the limb bud (SI Fig. 5 m and n ).
A null mutation of Dll1 (Dll1lacZ ) resulted in embryonic lethality around E11 (29), whereas embryos heteroallelic for the null and a hypomorphic allele (Dll1Dll1ki ) that we generated (SI Fig. 6), survived until birth, despite significantly reduced Notch activity (SI Fig. 7). Heteroallelic (Dll1lacZ/Dll1ki ) E18.5 fetuses were motionless, and they showed a severe reduction of skeletal muscle (Fig. 1 a–f). In E12.5 Dll1lacZ/Dll1ki embryos, anlagen of the major muscle groups were present (Fig. 1 g and h). However, from E13 onward, skeletal muscles in mutant embryos were clearly hypotrophic (Fig. 1 i–n). Beginning at E14.5, secondary myotubes form in tight association with primary myotubes, and they account for much of the muscle growth during fetal development (28). Secondary myotubes were severely reduced in mutant muscles (Fig. 1 p and r), but they were frequently observed in wild-type (arrowheads in Fig. 1 o and q). Collectively, these data suggested that primary myotubes formed in their normal location, but subsequent muscle growth is not sustained in Dll1 mutants.
Muscle hypotrophy in Dll1 mutant embryos. Sections of paraffin-embedded embryos were stained for myosin heavy chain (MHC; a–n, q, and r) or for MHC and laminin (o and p). (a–f) Representative sections of shoulder (a and b), intercostal (c and d), and body wall (e and f) muscles of E18.5 fetuses. (g–n) Muscles in the cervical region (g and h), body wall (i and j), diaphragm (k and l), and limbs (m and n) at the indicated stages. Arrowheads point to skeletal muscles in mutant embryos. (o–r) Secondary myotubes (white arrowheads) in epaxial (o and p) and hypaxial (q and r) muscles. he, heart; li, liver.
Embryonic and fetal muscle growth depends on a population of proliferating muscle progenitor cells that are characterized by expression of the paired-box transcription factors Pax3 and Pax7 (30, 31). Pax3- and Pax7-positive cells, respectively, were detected by immunohistochemistry in the myotomes of E9.5 and E10.5 wild-type embryos (Fig. 2 a, c, e, and g) and subsequently in skeletal muscles (Fig. 2 i, k, m, o, q, and s), with declining Pax3 expression levels beginning at E11.5 as described in refs. 30–32. Pax3- and Pax7-positive cells were present in the myotomes of mutant E9.5 and E10.5 embryos (Fig. 2 b, d, f, and h), indicating that these skeletal muscle progenitor cells were generated. However, progressively fewer Pax3- and Pax7-expressing cells were found in mutant skeletal muscles at subsequent stages (Fig. 2 j, l, n, and p), and at E14.5 these cells were virtually absent (see Fig. 4 r and t). We observed no evidence for increased apoptosis in mutant myotomes or skeletal muscles during these stages (data not shown), which suggests depletion of progenitor cells by uncontrolled premature differentiation as a major reason for reduced muscle growth, although we cannot exclude that a low-level increase of programmed cell death contributes to reduced muscle mass.
Progressive loss of myogenic progenitor cells in mutant muscles. Sections of paraffin-embedded embryos were stained for Pax3 and Pax7 protein. Pax3- and Pax7-expressing cells (red arrowheads), respectively, are detected in wild-type and mutant myotomes at the forelimb level at E9.5 (a–d) and E10.5 (e–h). (i–t) Sections of epaxial muscles of the cervical region. With increasing age, progressively fewer Pax3- and Pax7-expressing cells are detected.
To study early myogenic differentiation we first analyzed expression of MHC, a marker for fully differentiated myocytes. At somite stage (ss) 13, heteroallelic as well as Dll1-null mutant embryos had MHC-positive cells in the anterior-most myotome (data not shown), and the cervical myotomes of 16ss mutant embryos contained differentiated myocytes (Fig. 3 b and c), whereas MHC-positive cells were not detected in wild-type embryos of the same stage (Fig. 3 a). Subsequently, at the ss 25 and ss 27, more differentiated myocytes were detected in mutant compared with wild-type embryos both by staining of MHC and desmin (Fig. 3 d–i). Thus, in contrast to the late muscle hypotrophy, loss of Delta1 leads to premature and excessive muscle differentiation early, suggesting that Dll1 also regulates the differentiation of dermomyotome-derived early myoblasts that develop independently from Pax3/Pax7 (30).
Accelerated progression of myogenesis in Dll1 mutant embryos. (a–i) Premature and excessive generation of differentiated myotomal myocytes were visualized by whole-mount immunohistochemistry using anti-MHC (a–f) and anti-desmin (g–i) antibodies. Fully differentiated myocytes appear earlier (arrowheads in b and c), and they are more abundant in early myotomes (brackets in d–f, arrowheads in g–i). (j–zs) MRF expression in myotomes (j–y) and limb buds (z–zs) at the indicated stages detected by whole-mount in situ hybridization (j–u and x–zs) in whole embryos (j, k, p–s, x–zc, zf–zm, zp–zs) or vibratome sections (l–o, t, u, zd, ze, zn, zo) or by immunohistochemistry on sections (v and w). MyoD expression in somites of the cervical (j–m) and upper trunk (n and o) region is shown. Black arrowheads point to the regions of low MyoD expression in wild-type embryos. (p and q) Similar initiation of Myogenin expression (black arrowheads) and subsequent increase in myogenin-expressing cells (r–w). (x and y) Accelerated up-regulation of Myf6 (red arrowheads) in cervical myotomes. Accelerated up-regulation of MyoD (z–zi) and Myogenin (zj–zs) in limb buds is shown. Red arrowheads point to expression domains in mutant fore (fl) and hind (hl) limb buds.
To define the progression of myogenesis in mutant embryos more precisely, we analyzed the expression of MRF genes. Myf5 is the earliest MRF to be activated, and it marks the emergence of the first myoblasts from the dermomyotome. Subsequently, Myogenin expression indicates the differentiation to myocytes (4). In the anterior-most somites of both wild-type and mutant embryos, Myf5 expression was detected at ss 2–3 (data not shown), and Myogenin at ss 10–11 (Fig. 3 p and q). Likewise, expression of MyoD (activated after Myf5 in myoblasts) at ss 13–14, and Myf6 (activated in myocytes) at ss 15–16 was initiated normally (data not shown), indicating that the determination of early myoblasts and the initiation of myogenic differentiation were normal. Myf5 expression in myotomes progressed virtually identically in wild-type and heteroallelic mutant embryos up to E10.5 (SI Fig. 8 a–d ). However, after their initial activation transcription of MyoD (Fig. 3 l–o and SI Fig. 8 k and l ), Myogenin (Fig. 3 r–w and SI Fig. 8 e and f ) and Myf6 (Figs. 3 x and y and 4 c–f) was up-regulated in the myotomes of mutants. Consistent with the early excess of differentiated MHC-positive myocytes (Fig. 3 e and f), up-regulated MRF expression reflected an increase in differentiating muscle cells as indicated by enlarged myotomes and an increase of Myogenin-positive cells detected by immunohistochemistry (Fig. 3 u and w). The massive up-regulation of MyoD, Myogenin, and Myf6 in myotomes was transient, and subsequently more similar expression levels were detected in heteroallelic and wild-type embryos (SI Fig. 8 g–j and m–p and Fig. 4 e and f). Myf6, which is normally expressed exclusively in myotomal cells in a short transient wave between E9 and E11.5 (33), and Myf5, whose expression is normally down-regulated beginning on E11 (34), were down-regulated prematurely (Fig. 4 g–j, and a and b, respectively). Taken together, these analyses indicated that dermomyotome-derived myoblasts differentiated more rapidly, leading to a transient excess of differentiated myocytes.
Premature cessation of myogenesis in Dll1 mutant embryos and proposed functions of Dll1. (A) Expression of Myf5 (a and b) and Myf6 (c–j) detected by whole-mount in situ hybridization showing premature down-regulation of Myf5 beginning at E11.75 (a and b), transient up-regulation of Myf6 expression between E9.0 and E11.5 (indicated by red lines), and the premature down-regulation of Myf6 in the trunk, whereas expression in the tail (white line) is still comparable with wild-type. (B) Working model of proposed Delta1 functions. Early, Delta inhibits differentiation of myoblasts that are derived from the medial dermomyotomal lip and are independent of Pax3/Pax7 function. Later, Delta also regulates progenitor cell differentiation.
The progenitors of limb muscles were present, migrated, and initiated differentiation normally in Dll1 mutants (SI Fig. 5 and data not shown). However, after the initiation of differentiation at E10.5, MyoD, which is required for normal progression of limb myogenesis (35), and Myogenin were up-regulated in the forelimb buds (Fig. 3 z–ze and zj–zo) and at E11.5 also in the hindlimb buds (Fig. 3 zf–zi and zp–zs), whereas the Myf5 levels in limb buds declined (Fig. 4 a and b). This finding indicated that Dll1 also controls the differentiation of migratory hypaxial progenitors to myoblasts and myocytes. Additional premature differentiation was detected in the hypaxial myotome-derived hypoglossal chord and in muscles of the branchial bars and eye (SI Fig. 8 g–j and m–p , respectively), suggesting that Dll1 controls myogenic differentiation in all skeletal muscles.
Based on our results we propose that Dll1 controls in vivo two aspects of mammalian myogenesis after the determination of myoblasts and initiation of myogenic differentiation (Fig. 4 B). First, Dll1 inhibits the differentiation of myoblasts to fully differentiated myocytes, as indicated by the premature generation of primary myocytes in myotomes and limb buds. The rapid up-regulation of Myogenin before overt MyoD expression in epaxial cells implies that Notch activity can also act independently of MyoD in Myf5-expressing myoblasts. Second, Dll1 signals prevent resident progenitor cells from progressing along the myogenic program, as indicated by the depletion of the Pax3/Pax7-expressing cell population in differentiating muscles. In addition, it is conceivable that Dll1 signals contribute positively to the maintenance of muscle precursor cells independently of inhibition of terminal differentiation, and Dll1 activity might also be required to maintain the myogenic fate of precursor cells. Because Dll1 is expressed in myoblasts as well as in differentiated myocytes (SI Fig. 5 h and i , Fig. 3 m, and ref. 26), both cell types may provide signals that regulate differentiation and maintain proliferating skeletal muscle progenitor cells that sustain embryonic and fetal myogenesis (30, 31). Thus, Dll1 is of prime importance for skeletal muscle formation during mammalian development, similar to Drosophila. However, the precise role of Notch activity appears to have diverged because in flies Notch regulates the formation of muscle progenitors from somatic mesoderm (20, 36). Delta1 has been implicated in the regulation of satellite cell proliferation and regeneration of adult muscle (37, 38). Our results establish that Delta1-mediated Notch activity is equally important to regulate myogenic progenitor cell proliferation and differentiation in prenatal stages of myogenesis.
Generation and genotyping of Dll1lacZ mice has been described in ref. 29. The Dll1Dll1ki allele was generated as described in SI Fig. 5a . Genotyping was performed by PCR using primers TGGATGTGGAATGTGTGCGAG and AAGGGGAGAAGATGCTTGATAACC.
Histology, immunohistochemistry, and in situ hybridization were done by standard procedures. Mutant and wild-type embryos were analyzed in parallel with a given probe or antibody under identical conditions. More than two embryos were analyzed with each marker at a given developmental stage; and in the case of immunohistochemical analyses on sections, >15 consecutive sections were evaluated. Antibodies used were as follows: myogenin, mouse monoclonal antibody [F5D, 1:100; Developmental Studies Hybridoma Bank (DSHB), Iowa City, IA; or MY-32, 1:250; Sigma, St. Louis, MO]; myosin, mouse monoclonal antibody (MF20, 1:50; DSHB); desmin, mouse monoclonal antibody (1:100; DAKO, Carpinteria, CA); Pax3, mouse monoclonal antibody (1:20; DSHB); Pax7 mouse monoclonal antibody (1:20; DSHB); caspase3 rabbit polyclonal antibody (1:200; BD Biosciences, San Jose, CA); laminin, rabbit polyclonal antibodies (L9393, 1:60; Sigma). Secondary antibodies were coupled to FITC or CY3 or to peroxidase. Signal amplification was done by using the ABC kit (Vectastain; Vector Laboratories, Burlingame, CA).
Images of sections were obtained with a DM5000B microscope, a DFC300FX camera, and FireCam software (Leica, Wetzlar, Germany), of whole embryos with a FLIII microscope, a DFC300 camera and FireCam software (Leica), or with a M420 microscope (Leica) and HC300Z camera (Fuji, Tokyo, Japan) and FujixPhotograb300Z software (Fuji Foto Film Co.). Images were processed by using Photoshop CS (Adobe, San Jose, CA).
We thank H.-H. Arnold, T. Braun, and M. Gessler for probes; B. Herrmann, M. Kessel, A. Kispert, and T. Braun for discussion; H. Burckhardt and A. Heiser for assistance; and the Developmental Studies Hybridoma Bank developed under the auspices of the National Institute of Child Health and Human Development and maintained by the University of Iowa, Department of Biological Sciences, Iowa City, IA, for providing antibodies. This work was supported by Deutsche Forschungsgemeinschaft Grant Go449/9-1 (to A.G.).
Author contributions: K.S.-G. and R.C. contributed equally to this work; K.S.-G., R.C., and A.G. designed research; K.S.-G. and R.C. performed research; K.S.-G., R.C., and A.G. analyzed data; and A.G. wrote the paper.
↵*Present address: Brühlstrasse 2, 32423 Minden, Germany.
The authors declare no conflict of interest.
This article is a PNAS direct submission.
This article contains supporting information online at www.pnas.org/cgi/content/full/0608281104/DC1.
