Although all body muscles of amniote embryos derive from the epithelial somites, skeletal muscle development occurs in several different ways (Evans et al.,2006). Muscle progenitor cells either differentiate within the somite to form the segmented myotome which then progressively expands both dorso-medially (epaxially) and ventro-laterally (hypaxially), or they migrate to reach the site where they differentiate, or they adopt a developmental program which is a combination of these two processes (Kalcheim et al.,1999; Hollway and Currie,2005; Evans et al.,2006).
Soon after their formation, somites undergo a compartmentalization giving rise to the sclerotome (origin of vertebrae, ribs, and tendons of back muscles) and the epithelial dermomyotome (origin of skeletal muscle, dermal, some endothelial, smooth muscle, and brown fat precursors; Kalcheim and Ben-Yair,2005; Buckingham,2006; Christ et al.,2007; Seale et al.,2008). Migratory muscle progenitors delaminate from the hypaxial lip of the dermomyotome at specific axial levels and migrate to their final destination (e.g., limbs, diaphragm). Only after reaching their target sites, they enter the myogenic program, activating the expression of the myogenic regulatory factors (MRFs), Myf5, MyoD, MRF4, and myogenin, which orchestrate their commitment and differentiation (Buckingham et al.,2003). Then, myoblasts elongate and fuse to form myotubes, eventually maturing into muscle fibers. In contrast, muscle progenitors which differentiate within the myotomal compartment of the somites are released from the dermomyotomal lips directly into the myotomal space at all axial levels (Ben-Yair and Kalcheim,2005; Hollway and Currie,2005). They enter the myogenic program in situ and form the segmented muscle masses, the myotomes. At a later stage, a new pool of proliferating muscle progenitors expressing Pax3 and Pax7 enter the myotome from the central dermomyotome (Ben-Yair and Kalcheim,2005; Gros et al.,2005; Kassar-Duchossoy et al.,2005; Relaix et al.,2005), contributing to the growth of the myotome. Remarkably, in amniote embryos, the myotome is only a transient structure and does not get innervated (Deries et al.,2008). Rather, the segmented myotomes are transformed: epaxially into the complex deep back muscles and hypaxially into the intercostal and body wall muscles (Christ et al.,1983; Tremblay et al.,1998; Kalcheim et al.,1999; Deries et al.,2010).
All these events occur within the context of a complex environment composed of neighboring cells, the diffusible signaling molecules they produce and the surrounding extracellular matrix (ECM). The ECM is composed of macromolecules, primarily glycoproteins and proteoglycans (Frantz et al.,2010). It is involved in tissue differentiation and development not only as a mechanical support but also by transmitting biomechanical and biochemical signals mediated by transmembrane receptors such as the integrins (van der Flier and Sonnenberg,2001; Barczyk et al.,2010; Rozario and DeSimone,2010).
Although it is well established that the ECM plays an important role in several steps of myogenesis, the picture of its interaction with cells of developing muscles is far from complete (Thorsteinsdóttir et al.,2011). To fully understand the role of the ECM during embryogenesis, it is essential to know the exact organization of this matrix in relation to the developing tissues. No studies have addressed the three-dimensional (3D) organization of the ECM during the different phases of myotome morphogenesis. Most studies so far have analyzed the distribution pattern of ECM components either in whole-mount analysis of low magnification images or with higher detail on tissue-sections with the drawbacks of distorted tissues and limited 3D spatial organization details. The alternative views offered by confocal imaging of whole-mount immunofluorescence-stained embryos followed by 3D image reconstruction techniques (Martins et al.,2007) provide valuable missing details into the complex 3D architecture of tissues and their ECMs.
Using this approach in mouse embryos, we mapped the organization of the fibronectin, tenascin, and laminin matrices around and within the dermomyotome/myotome in 3D, and related that organization with the location and morphology of the dermomyotomal and myotomal cells. We focused exclusively on the epaxial region between the forelimb and hindlimb axial levels because the mechanisms of its development are relatively well known and do not vary within this region.
Fibronectin is the major glycoprotein of the interstitial matrix in the early embryo (Duband et al.,1987; Ostrovsky et al.,1988; Cachaço et al.,2005) and lack of fibronectin leads to defects in somitogenesis and neural tube morphogenesis (George et al.,1993; Georges-Labouesse et al.,1996; Rifes et al.,2007; Martins et al.,2009; Rozario and DeSimone,2010). Moreover, fibronectin has been shown to play a role during myotome development in zebrafish (Snow et al.,2008).
Unlike fibronectin, tenascin appears to be restricted to very specific areas during embryogenesis (Riou et al.,1992; Jones and Jones,2000). It is first detected after gastrulation and is observed around the somites. Then, it is present during muscle development but declines by late gestation and becomes confined to tendons (Crossin et al.,1986; Saga et al.,1992; Wang et al.,1995).
Laminins are the major glycoproteins of basement membranes. They are formed by the combination of three chains, α, β, and γ (Durbeej,2010). Several laminins are essential for embryogenesis and their assembly is the first step of basement membrane formation and also a prerequisite for the incorporation of the other basement membrane components (Smyth et al., 1999; Miner and Yurchenco, 2004; Yurchenco and Patton, 2009; Durbeej,2010). Laminin 111 and 511 are part of the basement membrane of the epithelial somites, dermomyotome, and myotome (Bajanca et al.,2006; Anderson et al.,2009; Thorsteinsdóttir et al.,2011). Laminin 211 and 221 are expressed later and are the major laminins of the skeletal muscle basement membrane (Cachaço et al.,2005; Durbeej,2010). Their presence in the myotube basement membrane is important for myotube stability during contraction, and lack of laminin α2 chain leads to muscle dystrophy (Guo et al.,2003; Huh et al.,2005).
To precisely analyze cell-ECM relationships during dermomyotome/myotome development, we first defined four developmental stages (DMM stages 1–4), which provide a useful tool for developmental studies of the dermomyotome/myotome. Our detailed 3D analysis shows that fibronectin forms an abundant matrix in all the interstitial spaces and is also localized in the intersomitic border where it may, in combination with tenascin, serve a tendon-like function for the elongated myocytes. In support for this hypothesis, we found that αv-integrins, which can serve as fibronectin and tenascin receptors, are enriched at myocyte tips. Bundles of fibronectin and tenascin also extend along the myocytes of the myotome, thus being in a good position to provide support for cell elongation, alignment, and even fusion and translocation later on. Laminin matrices show a surprisingly dynamic pattern; they are first assembled and then disassembled during myotome development, and produced again by the forming muscle masses. Of interest, the basement membrane matrices lining the dermomyotome and myotome are never completely continuous, most likely due to the transient nature of both of these compartments, but their discontinuous nature may in addition permit a direct communication route with neighboring cells. This detailed 3D analysis provides new insights of the potential roles of fibronectin, tenascin, and laminin matrices during myotome development and epaxial muscle morphogenesis.
RESULTS AND DISCUSSION
Definition of Stages of Dermomyotome/Myotome Development in the Mouse
In the mouse, the formation of the epaxial myotome starts at embryonic day (E) 8.5 in the rostral somites (Venters et al.,1999; Bajanca et al.,2004) and by E12.5, the myotomes have completely disappeared as segmented structures (Deries et al.,2010). During these few days, the myotome develops gradually, and different stages of development can be identified in different embryo ages and in different rostro-caudal levels of the same embryo. Due to differences in maturity of embryos between and within each litter, it has been difficult to rely on a staging of dermomyotomes/myotomes based solely on age and axial level of the embryo. To address this difficulty and to allow accurate descriptions of the changes in ECM organization along epaxial dermomyotome and myotome development, we defined four key stages from E9.5 to E11.5 which we designate dermomyotome/myotome (DMM) stages 1 to 4 (Fig. 1A–D). After DMM stage 4, the dermomyotome fully dissociates and the myotome undergoes a developmentally regulated translocation and splitting (Deries et al.,2010), which culminates in a new organization, visible at E12.5, and is termed the early epaxial muscle masses (Fig. 1E). Very little is known about the formation of cervical and tail epaxial muscle, therefore, this study focuses only on the epaxial muscles from forelimb to hindlimb levels.
DMM stage 1 (as in E9.5 interlimb to E9.5 forelimb level) corresponds to initial stages of myotome formation. The lips of the dermomyotome curl around the forming epaxial myotome and release the myogenic precursor cells (MPCs) into the myotomal space (Venters et al.,1999; Gros et al.,2004; Hollway and Currie,2005). MRF-expressing cells are found in the dorsal (epaxial) part of the myotome and a few myoblasts have differentiated into myosin-positive myocytes which are elongating (Fig. 1A). The dorsal-most part of the dermomyotome is rounded (Fig. 1A).
At DMM stage 2 (as in E10.5 hindlimb level to E10.5 interlimb level), both the dermomyotome and the myotome have expanded dorso-ventrally. Myoblasts and myocytes are numerous and myocytes span the whole width of the segment, except for the epaxial-most ones which are younger and therefore shorter (Venters et al.,1999) (Fig. 1B). The dorsal-most part of the dermomyotome remains rounded (Fig. 1B).
DMM stage 3 (as in E10.5 forelimb level to E11.5 hindlimb level) is characterized by the expansion of the myotome and the dermomyotome rostro-caudally and myocytes have become more elongated (Fig. 1C). The myotome keeps growing dorso-ventrally, and is now thicker in the medio-lateral direction (Fig. 1C). Furthermore, the dorsal-most region of the dermomyotome has changed shape and become oblong (Fig. 1C). It is during this stage that the first MPCs expressing Pax3/Pax7 are released from the central part of the dermomyotome into the myotome (Kassar-Duchossoy et al.,2005; Relaix et al.,2005) some of which enter myogenesis. Thus, non-elongated myoblasts can be seen between the dermomyotome and myotome (Fig. 1C).
At DMM stage 4 (as in E11.5 interlimb to E11.5 forelimb), the cells of the central region of the dermomyotome have undergone an asymmetric cell division thus segregating the muscle and the dermis precursor cells (Ben-Yair and Kalcheim,2005) leaving only the epaxial lip (as well as the hypaxial lip; Tajbakhsh and Buckingham,2000) as epithelial remnants. The myotome has continued to grow and some myocytes of the dorsal-most part of the segment have changed orientation and are no longer parallel to the rostro-caudal axis of the embryo (Fig. 1D). Moreover, it is at this stage that myotomal myocytes are becoming multinucleated (Deries et al.,2010). From this point onward, the segmented shape of the myotome progressively transforms into the complex organization of the axial muscle masses (Deries et al.,2010). An early epaxial muscle mass stage (from E12.5 onward) is represented in Figure 1E to illustrate the disappearance of the dermomyotome/myotome in the mammalian embryo. The most dorsal muscle masses at this stage are the anlagen of the transversospinalis muscles, some of which stay unisegmental. Ventro-lateral to them, the longissimus is formed, followed by the iliocostalis (Fig. 1E), both of which span several segments (Vallois,1922).
In this study, we focus on the relationship between the ECM and the dermomyotome/myotome along the four DMM stages. We also describe the relationship between the ECM and the translocating myocytes of the presumptive unisegmental transversospinalis muscles, which are the first to translocate (Deries et al.,2010).
The Interstitial Matrix Around the Dermomyotome and Myotome Is Rich in Fibronectin
To better understand the relationship of the dermomyotomal and myotomal cells with their surrounding interstitial ECM, the precise 3D organization of fibronectin and tenascin around and within the dermomyotome/myotome was analyzed from DMM stages 1 to 4 and after the dissociation of the dermomyotome and translocation of the myotome. Whole-mount embryos were co-immunolabeled for myogenin/myosin heavy chain (MHC) and fibronectin or tenascin and confocal images were rendered to obtain a 3D reconstruction of these matrices with the myogenic cells (see the Experimental Procedures section).
During all DMM stages, a fibronectin matrix (Fig. 2; Supp. Movies S1–S4, which are available online) fills the interstitial space on the lateral side of the dermomyotome (between the dermomyotome and ectoderm; Fig. 2A,C,D,F,G,I,J,N,O; Supp. Movies). Fibronectin is highly enriched at intersegmental boundaries at all DMM stages (shown for DMM stage 3; Fig. 2I,K) and is also found on the medial side of the myotome (Fig. 2B,C,E,F,H,J,L,N,O).
Tenascin has a much more restricted distribution than fibronectin (Fig. 3; Supp. Movies S5–S7). Consistent to what has been reported for the chick embryo (Crossin et al.,1986), at DMM stage 1, tenascin is detected at intersegmental borders, although the levels of the immunostaining were not sufficiently above background to allow a reliable reconstruction. From DMM stage 2 onward, tenascin progressively becomes highly enriched in the intersegmental space (Fig. 3A,B,D,E). Some tenascin matrix is present on the lateral side of the dermomyotome, but the immunostaining was again weak and could not be reconstructed.
The epithelial and segmented organization of the somite is dependent on the surrounding fibronectin matrix as shown by blocking fibronectin matrix assembly in cultured chick embryo explants (Rifes et al.,2007; Martins et al.,2009). In these explants, not only is somitogenesis impaired, but also already formed somites start to de-epithelialize. The fact that the dermomyotome retains a fibronectin matrix after somite compartmentalization suggests that the support of a fibronectin matrix is still needed for its epithelial and segmented organization. During early embryogenesis, tenascin has been reported to be deposited after a fibronectin matrix is in place (Crossin et al.,1986; Saga et al.,1992), which is consistent with our observations.
Fibronectin and Tenascin Are Enriched at Intersegmental Borders, Possibly Serving a Tendon-Like Function for Myotomal Myocytes
When myotomal myocytes elongate across the full length of the segment, their tips insert into the fibronectin and tenascin matrices at intersegmental borders (Figs. 2K, insert; 3A–F, arrows). Thus, the fibronectin and tenascin matrices may serve as a tendon-like structure providing an anchor for the elongated, differentiated myocytes. This intersegmental structure has been shown to be essential for myotome organization in zebrafish embryos because the disruption of fibronectin leads to a disorganization of myotomal cells (Snow et al.,2008). It would be interesting to verify whether the disruption of this matrix in the mouse leads to the same phenotype.
The α4β1 integrin, a fibronectin and VCAM1 receptor, is enriched at the tips of elongated myocytes (Bajanca et al.,2004), thus being in a position to mediate their attachment to the intersegmental fibronectin matrix. Furthermore, mRNA for Itga5, the α chain of the α5β1 fibronectin receptor, is present in the myotome (Bajanca et al.,2004; Cachaço et al.,2005). We have extended those observations to other fibronectin receptors as well as tenascin receptors on the myocytes. The integrins αvβ3 and α9β1 are two receptors for tenascin and both αvβ3 and αvβ1 bind to fibronectin (Thorsteinsdóttir et al.,2011). Immunohistochemistry on frontal cryosections for αv and α9, combined with MHC staining, shows that both these subunits are present in the mature somite already at DMM stage 2 onward (Fig. 4), earlier than previously described using immunohistochemistry with an enzymatic reaction (Hirsch et al.,1994; Wang et al.,1995). The αv subunit is present in all elongated myocytes from the formation of the myotome to the morphogenesis of epaxial muscle masses (Fig. 4A–C and data not shown). This subunit is enriched at the myocyte tips (Fig. 4A–C, arrows), making it a good candidate as a receptor for both fibronectin and tenascin. In contrast, the α9 subunit is not present on myocytes, but is localized on the basal and apical sides of the dermomyotome as well as in the sclerotome (Fig. 4D–F). Thus, although it is present at intersegmental borders, where tenascin is localized, α9β1 appears to be restricted to the dermomyotome and is not present on myocyte tips. The α9β1 also binds laminin (Durbeej,2010), which could be its major ligand at the basal side of the dermomyotome (see below).
We conclude that myotomal myocytes are in a position to attach to the fibronectin matrix through α4β1, α5β1 and αv-containing integrin(s) and may also attach to tenascin through αvβ3. These integrins could be playing an important role in anchoring myogenic cells to the ECM at intersegmental borders, providing mechanical support and signals which would promote their differentiation and/or maturation.
Fibronectin and Tenascin Bundles Penetrate the Space Between the Dermomyotome and Myotome as Well as Between Myocytes
By using immunolabeling in cryosections, we have previously shown that a patchy immunoreactivity for fibronectin is present between the dermomyotome and myotome as well as within the myotome (Cachaço et al.,2005). The 3D reconstruction of the organization of the fibronectin matrix reveals that the patches seen in sections represent fibronectin bundles which are spread on the surface of and within the myotome, generally running parallel to the myocytes (Fig. 2J,K). At DMM stage 1, these bundles are present on the medial side of the myotome (Fig. 2B,C). From DMM stage 2, they also appear in the space between the myotome and dermomyotome (Fig. 2F, arrow), and become particularly evident in this location from DMM stage 3 onward (Fig. 2J, arrow). They are also seen within the muscle mass (Fig. 2J, arrowhead). Fibronectin bundles remain abundant on the medial side of the myotome at all DMM stages (Fig. 2B,E,H). At early stages, the bundles are thin (Fig. 2B,E) but as the myotome grows, they become thicker (Fig. 2H,K). 3D reconstruction of the tenascin matrix at DMM stage 2 shows that fibrils of tenascin, originating in the dense intersegmental matrix, are also running along the surface of the myotome, on both the medial and the lateral sides (Fig. 3A,B, arrowheads; Supp. Movie S5). By DMM stage 4, thick tenascin bundles spread over the myotome both laterally and medially (Fig. 3D,E, arrowheads; Supp. Movie S6) and tenascin bundles are also detected within the muscle mass (Fig. 3G, arrowhead; Supp. Movie S6). These fibrils are distinct from the ones of fibronectin and their shape is different (compare Fig. 2H,K and 3D,E). As in the case of the fibronectin fibrils, the tenascin fibrils grow in size, extend from the rostral and caudal intersegmental borders and join in the middle of the myotome (Fig. 3D,E).
The fact that the bundles of fibronectin run mainly parallel to the myocytes (Fig. 2H,K) suggests that the myocytes may have a role in their assembly and/or that these bundles aid myocyte alignment along the axis. The tenascin matrix may give structural support to the fibronectin fibrils as reported in other systems (Jones and Jones,2000). It is tempting to propose that when muscle cells become multinucleated, cell alignment for myocyte fusion is guided by the fibronectin matrix. Although α5β1, α4β1 and αv integrins are not essential for myogenic cell fusion itself in vitro (Yang et al.,1996; Blaschuk et al.,1997; Taverna et al.,1998), cell engagement with the fibronectin matrix present within the myotome in vivo may serve to promote the approximation of myogenic cells, increasing the opportunities for fusion. The fact that β1D integrin knock-in and conditional β1 integrin knock-out mouse embryos show defects in myotube formation (Cachaço et al.,2003; Schwander et al.,2003) supports this hypothesis.
It is interesting to note that the fibronectin fibrils on the lateral side of the myotome, i.e., between the myotome and the dermomyotome (Fig. 2J, arrow, K), are present at the time when the central part of the epithelial dermomyotome gives rise to the muscle progenitor cells expressing Pax3 and Pax7 (Kassar-Duchossoy et al.,2005; Relaix et al.,2005). It has been shown by live imaging that epithelial somite cells are very dynamic, being able to extend protrusions, reaching for the fibronectin matrix and shuttling between the somitic epithelium and the somitocoel (Martins et al.,2009). It is therefore tempting to suggest that the Pax3/Pax7 progenitor cells may use the fibronectin matrix to translocate from the dermomyotome into the myotome in a similar way as the cells of epithelial somites enter and exit the somitocoel.
When the Dermomyotome Dissociates, a Fibronectin Matrix Ensheaths and Penetrates the Myotome, While the Tenascin Matrix Remains at Myocyte Ends
After the dissociation of the central dermomyotome at DMM stage 4, fibronectin has invaded all the tissues surrounding the myotome, including the forming dermis (Fig. 2L–O, arrowheads in N,O; Supp. Movie S4). This fibronectin matrix is dense and is tightly associated with the myotome. While the epaxial dermomyotome lip remains epithelial, the area occupied by it is free of fibronectin (Fig. 2N, arrow; transverse slab). However, as soon as the lip de-epithelializes, the fibronectin matrix invades the space and becomes tightly associated with the muscle cells (Fig. 2O, arrow; transverse slab, showing a more mature myotome, two segments rostral to the one depicted in Figure 2N). The interstitial fibronectin matrix remains present and is very dense during the whole transformation of the myotome. The fibronectin receptors α4β1, α5β1 and αv-containing integrins are expressed on the myocytes during the whole process (Cachaço et al.,2005; and data not shown). Fibronectin is thus in a position to guide the reorganization of the myogenic cells of the myotome by promoting directional fusion or by serving as a substrate for their translocation (Deries et al.,2010), or both.
In contrast to the almost ubiquitous presence of fibronectin around the DMM stage 4 myotome and early dorsal-most muscle masses, the tenascin matrix has a specific 3D organization, being enriched at the ends of and along myotomal myocytes (Fig. 3H,I; Supp. Movie S7). Tenascin is known to play a crucial role during tissue repair and tumor development (Jones and Jones,2000; Chiquet-Ehrismann and Tucker,2004; Calve et al.,2010), but tenascin-null mice do not show any dramatic phenotype during development (Saga et al.,1992; Forsberg et al.,1996; Garcion et al.,2001). It is possible that other matrices compensate for its absence or that the early defects are compensated later in development. It would be interesting to investigate early stages of development and verify the phenotype of early myogenesis in these mutants. Tenascin and fibronectin matrices are both localized at the intersegmental border and along the myocytes, so fibronectin may also compensate for the absence of tenascin. However, when the myotome transforms into the dorsal-most epaxial muscle masses, their pattern becomes different. A fibronectin matrix fills the space around and within the myotome (as shown for DMM stage 4; Fig. 2L–O; Supp. Movie S4), whereas tenascin is primarily found at the tips of elongated myocytes/early myotubes and between cleaving muscle masses (Fig. 3H,I, arrows; Supp. Movie S7). This indicates that, at this stage, their roles diverge. We suggest that fibronectin may guide the reorientation of the myotomal myocytes whereas tenascin is in a position to play a tendon- and muscle connective tissue-like role (Deries et al.,2010; Mathew et al.,2011).
The Laminin Matrices of the Dermomyotome and Myotome Increase in Complexity From DMM Stages 1 to 3, Without Ever Forming Fully Continuous Basement Membranes
Two basement membranes are essential during the first stages of myogenesis in the mouse: one lines the basal side of the dermomyotome and plays a significant role in preventing muscle progenitors from entering the myogenic program too early (Bajanca et al.,2006). The other is the basement membrane covering the medio-ventral surface of the myotome, which separates it from the sclerotome and keeps myogenic cells from getting dispersed into the sclerotome (Tosney et al.,1994; Tajbakhsh et al.,1996; Bajanca et al.,2006; Anderson et al.,2009). It should also be noted that both are distinct from the basement membrane that later surrounds the maturing muscle fibers (Cachaço et al.,2005).
Earlier analyses of laminin distribution in mouse embryo sections, using an antibody that recognizes all laminins present (except the epidermal laminin 332), reveals a continuous line of immunoreactivity on the dermomyotome and a line of matrix separating the myotome from the sclerotome (Bajanca et al.,2004; Cachaço et al.,2005; Anderson et al.,2009). The other basement membrane molecules (collagen type IV, nidogen, and perlecan) have an identical pattern (Bajanca et al.,2006; Anderson et al.,2009). Our 3D analysis of laminin distribution with the same antibody (see the Experimental Procedures section) confirms that laminin is organized into a basement membrane sheet on the basal side of the dermomyotome and between the myotome and sclerotome. However, it also provides new information, revealing that this basement membrane sheet changes significantly along the DMM stages defined here. At DMM stage 1 (Fig. 5A–D; Supp. Movie S8), the basal side of the dermomyotome is lined by a loose basement membrane matrix, containing several holes (Fig. 5A,C), but this matrix becomes slightly more dense at the rostral and caudal segmental boundaries of the dermomyotome (Fig. 5D, arrowheads). The myotomal laminin matrix, which is progressively assembled by the Myf5-positive myoblasts (Bajanca et al.,2006) is visible as a discontinuous sheet on the medial side of the myotome at DMM stage 1 (Fig. 5B,C, arrowhead; Supp. Movie S8). It is present only in association with myoblasts.
As the dermomyotome/myotome develops, the laminin basement membrane of the dermomyotome becomes denser and the laminin matrix of the medial side of the myotome grows concomitantly with the addition of new cells to the myotome. The central part of this matrix is denser than the dorsal and the ventral part (shown at DMM stage 3, Fig. 5F, asterisk, Supp. Movie S9), probably because it is the most mature part of the myotome (Venters et al.,1999). At DMM stage 3 (Fig. 5E–H; Supp. Movie S9), both of these laminin matrices have reached their highest density. Strikingly, these laminin matrices are never completely continuous basement membrane sheets (see holes, Fig. 5E–H, arrows; Supp. Movie S9) such as those of the ectoderm and the neural tube (data not shown). These discontinuities in the matrix are not visible on sections (Bajanca et al.,2006; Anderson et al.,2009).
The observation that these two laminin matrices never form a complete basement membrane is surprising. Nonetheless, they maintain the epithelial dermomyotome and contain the myogenic cells within the myotomal compartment. Furthermore, laminin engagement by dermomyotomal cells has been shown to promote the undifferentiated state of the dermomyotome (Bajanca et al.,2006) while the myotomal laminin matrix aids the alignment of myotomal myocytes (Wilson-Rawls et al.,1999; Bajanca et al.,2006; Seo et al.,2006). It is possible that, due to the transient nature and rapid growth of both the dermomyotome and the myotome, the production of a complete basement membrane is an unnecessary investment and an incomplete basement membrane is sufficient in this context. Furthermore, the holes in the basement membrane give the cells of both the dermomyotome and myotome access to the surrounding interstitial matrix which could provide structural support for both of these tissues. Another hypothesis to explain the incomplete dermomyotomal basement membrane could be that this epithelium is very dynamic with protrusions of cells sent out and retracted again (Gros et al.,2009; Martins et al.,2009), not permitting the formation of a continuous laminin sheet. The holes in the laminin matrix may also facilitate juxtacrine cell–cell interactions. In this context, it is interesting to note that passing migrating neural crest cells have been shown to touch epaxial dermomyotome cells and thus induce notch activation in these dermomyotomal cells which triggers their entry in the myogenic program (Rios et al.,2011).
DMM Stage 4 Is Characterized by a Progressive Disassembly of the Laminin Matrices
At DMM stage 4, the dermomyotomal laminin matrix has largely disappeared (Fig. 5I–L; Supp. Movie S10), possibly through the action of matrix metalloproteases (MMPs; Duong and Erickson,2004), and most of the dermomyotome has dispersed to release the mesenchymal dermis and muscle precursors (Ben-Yair and Kalcheim,2005). The laminin matrix has also been disassembled on the medial side of the myotome. Only the dorsal-most part of the laminin matrix remains (Fig. 5I–L). Intriguingly, the 3D reconstructions reveal that this matrix forms a pouch-like structure on the medial side of the myotome (Fig. 5I–L, arrows; Supp. Movie S10). However, the dorsal-most myocytes, which at this stage are acquiring a new orientation, cross the laminin matrix at the rostral and caudal segment borders, and extend through the matrix (Fig. 5L, arrowhead; Supp. Movie S10). This indicates that the basement membrane lining this pouch-like refuge of epithelial cells is no longer capable of containing myocytes.
After the reorganization of the epaxial muscles, the dermomyotomal and myotomal laminin matrices have disappeared and laminin is observed in a “spotty” form around and within the early muscle masses, indicating de novo production (Fig. 5M–P; Supp. Movie S11). It is tempting to suggest that this laminin may, at least in part, contribute to the basement membrane seen surrounding each muscle fiber from E13.5 onward (Cachaço et al.,2005).
The Last Assembled Laminin Matrix Correlates With the Epithelial State of the Dermomyotomal Epaxial Lip
Because the epaxial pouch-like laminin matrix is no longer present as a barrier for myocytes, it may represent the basement membrane of the remaining epithelial lip of the epaxial dermomyotome, thus containing a last refuge of dermomyotomal epithelial cells, in an otherwise fully mesenchymal milieu (Fig. 5I–L, arrows).
Whole-mount co-immunofluorescence staining for Pax3/Pax7 (marking dermomyotomal cells and MPCs) and laminin showed that the persistent laminin structure lines the epithelial Pax3/Pax7-positive cells (Fig. 6A,D,G). Thus, we conclude that it is part of the basement membrane of the epaxial dermomyotomal lip. To visualize the laminin matrix during the de-epithelialization of the epaxial lip, reconstruction of this region and its laminin matrix was done every two segments. Figure 6 represents three segments of one embryo, from the epithelial state of the lip through to when the Pax3/Pax7-positive cells have all become mesenchymal (caudal to rostral). As the epaxial lip undergoes an epithelio-mesenchymal transition, the laminin matrix disassembles.
The epaxial lip of the dermomyotome, source of myogenic cells (Ben-Yair and Kalcheim,2005), is kept epithelial with an intact basement membrane longer than the rest of the dermomyotome. Our data strongly suggest that this region is maintained as a refuge for the mitotic Pax3/Pax7-positive muscle progenitors. This region corresponds to the future segmented transversospinalis muscles (Vallois,1922; Deries et al.,2010). One can hypothesize that having a segmented organization of proliferative muscle progenitors when the rest of the muscle is starting to re-organize and the borders of the myotomes are blending, would be a way to sustain this segmentation to form the segmented transversospinalis muscle.
Down-regulation of Two Major Laminin Receptors on Myocytes Correlates With the Transformation of the Myotome into the Early Epaxial Muscle Masses
Myocytes have been reported to engage the myotomal laminin matrix through α6β1 integrin (Bajanca et al.,2006) and probably also, later on, α7β1 integrin (Bajanca et al.,2004). Because myotomal myocytes are not contained by the laminin matrix at the epaxial lip (Fig. 5L, arrowheads) and these two laminin receptors are not detected at E12.5 (Cachaço et al.,2005), we examined the exact distribution of α6 and α7 integrin subunits by immunohistochemistry on serial sections before, during and after the myotome transforms (Fig. 7). Both α6 and α7 become absent on the dorsal-most part of the segment at DMM stage 4, when these myocytes are changing their orientation (Fig. 7A,E,I arrows). Then during the reorganization of the whole muscle mass, α6 disappears completely from the muscle (Fig. 7B,C) and eventually so does α7 (Fig. 7K). These proteins are only detected again when the muscle masses are completely reorganized (Fig. 7D,H,L).
Our results suggest that immediately before myotome translocation, differentiated myocytes cease to interact with the laminin matrix first through α6β1 and then through α7β1, although they could still be interacting with laminin through other receptors, such as dystroglycan. Nevertheless, soon after, both α6β1 and α7β1 are down-regulated, all the laminin matrix around the myocytes is disassembled or degraded. We propose that myocytes need to detach from, or at least weaken, their binding to laminin to be able to reorient and grow in different directions to form the adult epaxial muscles. We conclude that the laminin matrix is essential not only for early stages of dermomyotome/myotome development, but its remodeling in later stages most likely plays a crucial role in muscle morphogenesis.
Through 3D imaging methods, we defined four stages of dermomyotome/myotome development (DMM stages 1–4) and present a new perspective of cell–ECM interactions during dermomyotome and myotome development in the mouse. This study showed that the ECM is a very dynamic entity which is extensively remodeled throughout the various stages of dermomyotome/myotome development (Fig. 8), suggesting that the regulation of the ECM may play a significant role in the successful development of the axial muscles. The three different components analyzed here have very specific patterns, which change during the stages studied. While fibronectin is abundant in all interstitial tissues, including intersegmental borders which are transient myotome tendon-like structures, tenascin is restricted to these sites. Fibronectin and tenascin bundles are also in a good position to promote alignment of myocytes and guide the entry of muscle precursors in the myotome. Laminin basement membranes undergo organized dynamic changes during dermomyotome and myotome development without ever becoming continuous, possibly because of the transient nature of these embryonic tissues. Our results suggest that during early epaxial morphogenesis, myocytes weaken the binding to laminin and might use fibronectin matrix as a substrate to acquire their final orientation.
Unraveling the exact architecture of the ECM environment during myotome development, raises new hypotheses regarding the developmental roles for both the interstitial ECM components fibronectin and tenascin, as well as the laminins of basement membranes. With the knowledge gained in this study, we are now in a position to design experiments to test these hypotheses.
Dated pregnancies were obtained from Charles River (CD1) mice (Harlan Iberica) and the day of the plug was E0.5. Embryos were collected from E9.5 to E13.5 after cervical dislocation of the female (project approved by Direcção Geral de Veterinária, Portugal).
Whole-mount immunohistochemistry was performed as described in Deries et al.,2008. Briefly, embryos were fixed in 2% buffered paraformaldehyde, then washed and dehydrated in methanol. They were permeabilized with a solution of dimethyl sulfoxide (DMSO) and methanol (1:4) for 3 weeks. Then they were stored in methanol 100% at −20°C until use. After rehydration, embryos were dissected as appropriate and processed for immunohistochemical staining. Samples were incubated with primary antibodies (Table 1) in phosphate buffered saline (PBS) containing 1% bovine serum albumin (BSA; Sigma) and 20% DMSO overnight at 4°C and were extensively washed in PBS during 2 days. Co-immunohistochemistry experiments were performed to stain fibronectin, tenascin or laminins together with myogenic cells. At early stages myogenic cells were detected by a “cocktail” of two mouse monoclonal antibodies: anti-myogenin (marking nuclei of differentiating cells) and anti-embryonic myosin heavy chain (MHC, marking cytoplasm of fully differentiated myocytes). At later stages, only the anti-MHC antibody was used to detect myogenic cells. Application of the secondary antibodies (Table 1) was done as for the primary antibodies. After immunolabeling, samples were dehydrated in methanol, cleared in methyl salicylate (Martins et al.,2009) and mounted between two coverslips sealed with paraffin. This mounting allowed us to view the thick specimens from two sides. Samples were imaged in a Leica SPE confocal microscope (see below).
Fluorescent immunohistochemistry on 12- or 30-μm cryosections was performed as in Bajanca et al. (2004). Embryos were fixed in 0.2% buffered paraformaldehyde, gradually embedded in sucrose, followed by embedding in gelatine and stored at −80°C. After sectioning in a Bright Clinicut cryostat, the samples were stained in primary antibody overnight at 4°C and secondary antibody for 2 hr at room temperature. Antibodies (see Table 1) were diluted in PBS containing 1% BSA. Sections were mounted in propylgallate and viewed in a Leica SPE confocal microscope.
Imaging and 3D Reconstruction
To visualize the 3D matrix organization, confocal stacks were acquired on a Leica SPE confocal microscope with a ×40 ACS APO 1.15 oil-immersion lens. From DMM stage 3 onward, the segment spanned beyond the field of the ×40 lens, therefore, the hypaxial region of the myotome is not visible in the images. The confocal stacks were 3D reconstructed and analyzed using the Amira v5.3 software (Visage Inc). When appropriate, one individual segment and its associated matrix were isolated by manual contouring of confocal sections (i.e., other cells and the matrix from surrounding tissues were “digitally erased”). The segments are depicted in lateral, medial and transverse “3D views.”
We thank members of our group, in particular Pedro Rifes, for helpful discussions and Arnoud Sonnenberg and Ann Sutherland for their kind gift of antibodies (LAT-2/GoH3 and CA5). The F59, Pax3, and Pax7 antibodies were developed by F.E. Stockdale, C.P. Ordahl, and A. Kawakami, respectively, and were obtained from the Developmental Studies Hybridoma Bank, developed under the auspices of the NICHD and maintained by The University of Iowa, Department of Biology, Iowa City, IA52242, USA. This work, R.V., and M.D. were supported by Fundação para a Ciência e a Tecnologia (Portugal). A.B.G. is a student of the Masters in Evolutionary and Developmental Biology (FCUL).