Differential effects of N-cadherin-mediated adhesion on the development of myotomal waves

The embryonic myotome is the source of all epaxial and hypaxial skeletal muscles. The mechanisms underlying myotome development were recently reevaluated (Brent and Tabin,2002; Buckingham,2001; Hollway and Currie,2003; Kalcheim and Ben-Yair,2005). Studies in avian embryos showed that the dorsomedial lip(DML) of the dermomyotome (DM) and, later, the ventrolateral lip (VLL)contribute to myotome growth in medial and lateral orientations, respectively(Denetclaw et al., 2001; Denetclaw et al., 1997; Denetclaw and Ordahl, 2000; Ordahl et al., 2001). Our studies distinguished at least two separable waves that account for the formation of the postmitotic myotome. A first wave of pioneer myoblasts constitutes the medial region of the epithelial somite with cells expressing Myod and Myf5. Upon sclerotome dissociation, pioneers delaminate, migrate towards the rostral somitic domain, and, from this region,generate myofibers in rostrocaudal and mediolateral directions(Kahane et al., 1998b; Kahane et al., 2002). Following the establishment of this primary structure, the advent of a second wave of myoblasts was shown, which emanates from all four lips of the DM, the contribution of the DML and VLL being restricted to the epaxial and hypaxial domains, respectively, and that of the rostral and caudal lips spreading along the entire mediolateral myotome. This unique mechanism leads to an overall intercalatory mode of myotome expansion(Cinnamon et al., 2001; Cinnamon et al., 1999; Kahane et al., 1998a; Kahane et al., 2002). The contributions of the four DM lips were further confirmed by quail-chick analysis (Huang and Christ,2000) and by GFP electroporation(Gros et al., 2004).

Subsequent growth of the myotome occurs by addition of cells that remain transiently proliferative. We reported that the first mitotically active muscle progenitors, termed the `third wave', originate in the rostral and caudal lips of the DM (Kahane et al.,2001). Somewhat later, the central DM sheet dissociates and its progeny, composed of at least bipotent cells, colonizes both the subectodermal space to generate dermis, and the myotome to give rise to the majority of mitotic muscle progenitors, a population that maintains the expression of DM markers, such as PAX3, PAX7 and FREK(Ben-Yair and Kalcheim, 2005; Ben-Yair et al., 2003). The final fate of these progenitors is varied. Most are likely to withdraw from the cell cycle at progressive stages and undergo myogenesis, some may also contribute to muscle fibroblasts, endothelial cells and muscle satellite cells(Gros et al., 2005; Kassar-Duchossoy et al., 2005; Relaix et al., 2005; Scaal and Christ, 2004). Altogether, an emerging fate map suggests that the entire DM epithelium has myogenic potential that is, nevertheless, regionally biased to the immediate production of myofibers and/or of mitotic muscle progenitors.

The existence of these discrete somitic microenvironments endowed with different and multiple fates raises fundamental questions about the mechanisms of lineage segregation in the various subdomains. Based on these findings, a functional analysis of genes expressed in somites can now be undertaken in the perspective of a clearer cellular picture underlying morphogenesis. In the present study, we investigate the precise expression and role of N-cadherin in each of the described waves of somitic myogenesis.

N-cadherin belongs to a family of Ca²⁺-dependent cell adhesion molecules (Hatta et al., 1988; Hatta et al., 1987; Hatta and Takeichi, 1986; Volk and Geiger, 1984; Volk and Geiger, 1986a; Volk and Geiger, 1986b) and is crucial for a diversity of steps during embryonic development(Gumbiner, 2000; Nelson and Nusse, 2004). It is characterized by five extracellular cadherin-binding domains that are separated by Ca²⁺-binding pockets, a transmembrane domain and an intracellular β-catenin-binding domain(Tepass et al., 2000). Among numerous other systems, N-cadherin is also dynamically expressed in the paraxial mesoderm (Duband et al.,1987; Duband et al.,1988). In vitro studies implicated N-cadherin in the regulation of the specification and differentiation of muscle progenitors(George-Weinstein et al.,1997; Goichberg and Geiger,1998; Holt et al.,1994; Knudsen et al.,1990), but knowledge of its role(s) in vivo is still very limited. A pivotal role for N-cadherin in somite morphogenesis was demonstrated(Horikawa et al., 1999; Linask et al., 1998; Radice et al., 1997). However,this early requirement during somite formation, and the premature death of embryos carrying a mutation in N-cadherin, precluded phenotypic analysis at muscle-forming stages, thus, genetic evidence awaits the analysis of conditional knockouts. In avians, adenovirus-mediated missexpression of N-cadherin mutants revealed a role in myofiber arrangement, yet this technique allowed analysis of only an early subset of myofibers(Horikawa and Takeichi, 2001). In zebrafish, an interplay between N- and M-cadherins was reported to control the migration of slow muscle cells from their paranotochordal region of origin to the most lateral surface of the myotome(Cortes et al., 2003).

In a previous study, we determined that the dual generation of dermal and mitotic muscle progenitors from the central DM is associated with a sharp change in the plane of cell division from the young epithelium, in which symmetrical divisions occur parallel to the mediolateral plane of the DM, to the dissociating DM, in which cell divisions become mostly perpendicular with one daughter cell facing the dermis and the other positioned adjacent to the myotome (Ben-Yair and Kalcheim,2005). Here, we report that during the growth phase of the DM sheet, when the orientation of mitotic spindles is parallel to the mediolateral extent of the epithelium, N-cadherin is inherited by both daughter cells. Prior to DM dissociation into dermis and muscle progenitors,when mitoses become perpendicularly oriented, N-cadherin remains associated with only the apical cell located in apposition to the myotome, and is absent from the basally oriented cell, thus generating molecular asymmetry between basal and apical progeny. In line with this differential segregation, local gene misexpression confirms that N-cadherin-mediated adhesion is sufficient for promoting myotome colonization, whereas interfering with the above process prematurely drives the cells towards the dermal domain with no PAX-positive progenitors colonizing the myotome. Hence, N-cadherin-mediated adhesion is both necessary and sufficient for the differential allocation of DM progenitors into the myotome. Notably, cells overexpressing N-cadherin that translocate into the myotome differentiate into myofibers, rather than remaining as mitotically active PAX⁺ progenitors, as has been reported for their normal counterparts, suggesting that N-cadherin is also involved in muscle specification. Likewise, interference with the activity of N-cadherin-mediated cell adhesion in the bordering lips (DML, rostral and caudal) caused the rapid dissociation of the DM epithelium. Consequently, no cells entered the myotome and no fibers differentiated. Notably, in the rostral and caudal lips of the DM, early electroporation prevented the onset of Myod transcription and myofiber generation, further implying that N-cadherin-mediated adhesion is required for proper myogenic specification. In contrast to DM lip-derived fibers, differentiation of the pioneer fibers was not affected. This finding provides the first molecular evidence that the pioneer population is unique and distinct from the later myofibers deriving from the DM and, in particular, from the DML.

Fertile quail (Coturnix coturnix Japonica) eggs from commercial sources were used.

Four different expression vectors were employed: pCAGGS-AFP, which served as control (Momose et al.,1999); full-length chicken N-cadherin; a dominant-negative version of N-cadherin lacking part of the extracellular domain (cN390Δ); and N-cadherin lacking its intracellular β-catenin-binding domain (CBR-)(Fujimori and Takeichi, 1993; Nakagawa and Takeichi, 1998). The latter three were subcloned into the pCAGGS vector and fused in frame to a GFP-encoding sequence.

Electroporations were performed under a dissecting microscope. DNA (4μg/μl) was microinjected into the center of flank-level epithelial or dissociating somites at specific stages as depicted in the figure legends. Five different types of electroporation were performed: to the medial epithelial somite, to the DM sheet, to the DML and to the rostral and caudal lips (Fig. 1). To this end,tungsten electrodes were mounted on two separate micromanipulators and placed to reach each somitic domain as described in Fig. 1. The direction of the electric field is indicated by the black arrow. A four parameter PulseAgile square wave electroporator (PA-4000, Cyto Pulse Sciences)was used to deliver three groups of sequential pulses as follows: 3×30 V, 20 mseconds each; 1×38 V, 5 mseconds; 3×30 V, 20 mseconds each.

The number of embryos showing a given phenotype out of the total number of transfected embryos for each treatment is presented in the corresponding figure legends.

Embryos were fixed with 4% formaldehyde in PBS, embedded in paraffin wax and sectioned at 8 μm. Immunostaining for desmin was as described(Kahane et al., 2001). Rabbit anti-GFP (Molecular Probes) was used at a dilution of 1:500, alone, in combination with desmin immunolabeling, or with in situ hybridization for Myod, Alx4 or Pax7(Cinnamon et al., 2001). Monoclonal anti-N-cadherin antibodies were from Zymed and were used at a 1:100 dilution in PBS containing 5% fetal calf serum and 0.1% Triton X-100. PAX7 immunolabeling, staining of centrosomes with γ-tubulin antibodies and Hoechst nuclear staining were as described by Ben-Yair and Kalcheim(Ben-Yair and Kalcheim, 2005). Whole-mount embryo preparations and sections were photographed using a DP70(Olympus) cooled CCD digital camera mounted on a BX51 microscope(Olympus).

The progenitors of the first wave arise along the medial surface of epithelial somites. In embryos aged 22-25ss, these progenitors express Myod and Myf5 and reveal a low level of DNA synthesis relative to the rest of the somite (Kahane et al., 1998a; Kahane et al., 1998b). N-cadherin is homogeneously distributed throughout the epithelium, including the medial domain, with predominant staining in the apical surfaces and weaker expression in the somitocoele (Fig. 2A). Upon somite dissociation, pioneer myoblasts bend underneath the forming DM, dissociate into mesenchymal cells and engage in caudorostral migration (Kahane et al.,1998b; Kahane et al.,2002) (see also Fig. 2F and Fig. 8C). Slightly before this process begins, pioneer myoblasts that express desmin downregulate N-cadherin to a basal level (arrowheads in Fig. 2B-G), similar to the behavior of dissociating sclerotomal cells(Fig. 2B,C)(Duband et al., 1987). The young DM retains strong apical N-cadherin expression(Fig. 2B-I). Following polarized migration, the first myotomal fibers are generated in rostrocaudal and mediolateral directions (Kahane et al., 1998a; Kahane et al.,1998b; Kahane et al.,2002), a stage at which they regain N-cadherin protein on their surface (Fig. 2H,I). Hence,pioneer myoblasts are positive for N-cadherin at the epithelial stage, lose the protein during dissociation and migration, and re-express it during differentiation into myofibers.

The second wave of myogenesis generates fibers and emanates from all four lips of the DM. Rostral and caudal lips that express Myod in a scattered pattern directly generate fibers into the myotome(Kahane et al., 2001; Kahane et al., 1998a) (see also Fig. 7), and express N-cadherin primarily in the apical aspect of the epithelial cell, as revealed by N-cadherin staining of cells transfected with a GFP-encoding DNA(Fig. 3D-G).

The DML and VLL are also N-cadherin positive(Fig. 3H-M). We have previously shown that these lips contribute indirectly to myofiber generation, by releasing intermediate progenitors that localize underneath these lips in a region we termed the sub-lip domain (SLD). Similar to the DML and VLL, the SLD is also negative for desmin but, unlike the DML and VLL, it is positive for Myod, Myf5 and FREK(Cinnamon et al., 2001), and also for N-cadherin, even though these cells are no longer epithelial(Fig. 3H-J).

Finally, the DM sheet is composed of cells that generate both PAX-positive mitotic muscle progenitors and dermis(Ben-Yair and Kalcheim, 2005). This part of the epithelium also reveals N-cadherin immunoreactivity enriched in the apical pole of the cells (arrowheads in Fig. 3A-C). N-cadherin protein is, however, downregulated in the emerging dermis(Fig. 3K-M), but is stably maintained in desmin-expressing myofibers that, contrary to the DM, are decorated in a homogeneous pericellular pattern (Figs 3, 4).

Following dissociation, many individual DM sheet progenitors generate both mitotic myotomal progenitors as well as dermis. This is associated with a shift in the orientation of cell divisions which, in the young DM epithelium,are parallel to the mediolateral axis of the DM and prior to DM dissociation progressively change to become perpendicularly oriented(Ben-Yair and Kalcheim, 2005). To begin approaching the relationship between mitotic orientations,dissociation of the DM epithelium and fate segregation, we determined the localization of N-cadherin protein during mitosis(Fig. 4). As in other types of epithelia, the mitotic phase of the cell cycle in the DM occurs at the apical pole where N-cadherin expression is predominant(Fig. 4A). In all parallel cell divisions, daughter cells remain side by side, associated with the apical surface of the epithelium, and N-cadherin is symmetrically inherited by both cells (Fig. 4B). In perpendicular mitoses, one daughter cell points towards the dermis and the other towards the myotome. In all such cases, N-cadherin remains associated only with the apical cell located in apposition to the myotome and is lost from the basally positioned cell (Fig. 4C,D), thus generating a molecular asymmetry between the basal and apical progeny. Notably, N-cadherin in oblique mitoses was similarly restricted to the apical progeny, similar to what was observed in the perpendicular cell divisions (Fig. 4E), further validating by molecular means the classification previously employed to distinguish between parallel and perpendicular mitoses(Ben-Yair and Kalcheim, 2005; Cayouette et al., 2001).

The observation that myotomal cells are N-cadherin positive whereas dermal progenitors downregulate the protein (Figs 3, 4), together with the finding that N-cadherin becomes differentially segregated to the apical daughter cells during mitosis (Fig. 4),suggest that, of the single cells previously shown to generate both muscle and dermal progenitors, the N-cadherin-positive DM daughter cell translocates into the N-cadherin-expressing myotome as a result of homophilic interactions,whereas the cell lacking N-cadherin generates dermis. To examine this possibility, the nascent DM sheet was focally electroporated (see Fig. 1B), with control GFP,wtN-cadherin-GFP or N-cadherin bearing a deletion in the extracellular domain(cN390Δ-GFP) (Fig. 5). This mutation was documented to interfere with activity of endogenous cadherin by competing for interactions with cytoskeletal components, resulting in the abrogation of cadherin-mediated cell adhesion(Fujimori and Takeichi, 1993). The transfection method used did not reach the bordering lips and was confined to the center of the DM sheet. Up to 18 hours after transfection, the GFP-labeled DM was still epithelial in control-treated segments, but in cells expressing cN390Δ-GFP the beginning of de-epithelialization could be detected as early as 5 hours following electroporation; this was also revealed by a gradual loss of basement membrane-associated laminin immunoreactivity,which is observed in the basal surface of the DM (see Fig. S1 in the supplementary material, data not shown). By 30 hours, the DM sheet had already dissociated in control segments, and GFP-labeled cells localized in both dermis and myotome (Fig. 5A)(Ben-Yair and Kalcheim, 2005). In clear contrast, the cells expressing cN390Δ-GFP failed to colonize the myotome and localized instead to the prospective dermal domain. There,they sorted out from the unlabeled progenitors to become aggregated subectodermally (Fig. 5B). These data suggest that N-cadherin-mediated adhesion is necessary for myotome colonization. To examine whether it is also sufficient, wtN-cadherin-GFP was similarly delivered. In contrast to control GFP and to cN390Δ-GFP, all cells overexpressing N-cadherin translocated into the myotome with no contribution to dermis (Fig. 5C). This effect was mimicked by overexpression of a mutant of N-cadherin in which the intracellular β-catenin-binding domain (CBR-) was deleted (Fig. 5D). Altogether,these results suggest that the extracellular domain of N-cadherin that mediates homophilic cell adhesion is both necessary and sufficient `in vivo'to drive cell translocation into the myotome.

Next, we analyzed the fates of cells overexpressing either wtN-Cadherin-GFP or cN390Δ-GFP. As previously documented, control GFP-treated cells remained as mesenchymal Pax7-positive muscle progenitors in the myotome(Fig. 5E)(Ben-Yair and Kalcheim, 2005). By contrast, N-cadherin (wt or CBR-)-treated cells entered the myotome as Pax-positive progenitors, gradually downregulated Pax by 24-30 hours (data not shown) and then differentiated into fibers by 48 hours after transfection(Fig. 5F). Thus, N-cadherin is not only necessary and sufficient for cell translocation into the myotome, it also triggers myogenic specification followed by differentiation in cells that otherwise remain as proliferative myotomal progenitors.

We then examined the phenotype of cells that received cN390Δ-GFP and localized subectodermally to where dermis develops(Fig. 5B). Similar to control GFP-treated cells, these cells remained mitotically active (see Fig. S2 in the supplementary material). Like normal dermis progenitors, they maintained expression of Alx4, a marker of the central DM region that is later expressed in dermis (Cheng et al.,2004) (Fig. 5G,H),and they rapidly downregulated Pax7 mRNA expression(Fig. 5I), while still exhibiting a low level of Pax7 protein; the protein had completely disappeared by 20 hours after transfection (see Fig. S2 in the supplementary material,data not shown). Thus, cN390Δ-GFP-treated cells display features of early dermal progenitors (maintenance of Alx4 and downregulation of Pax genes). Because of a relatively rapid loss of the transfected signal, we were unable to trace them after E5, when more mature markers of dermis development, such as Dermo1, or differentiation features of mature dermis become apparent.

The dynamic expression of N-cadherin to pioneer muscle progenitors and to DM lips, as well as its effects on cells originating in the DM sheet, prompted us to further examine its involvement in the development of the first two myogenic waves that generate myofibers.

In previous studies, we subdivided the contribution of the medial somite into early pioneer myoblasts and later DML-derived fibers. Here, we wished to determine whether these two populations can be further distinguished by a differential sensitivity to N-cadherin (see Figs 6, 8). To examine its effect on DML progenitors, the nascent DML was electroporated with control GFP or with cN390Δ-GFP (Fig. 1C). In control-treated segments, the DML remained both epithelial and labeled for the entire duration of the experiments (Fig. 6A-F), further suggesting that this domain acts as a continuous source for myofibers (Ordahl et al.,2001); this is different from pioneers, which are a limited cell subset leaving no residual labeled cells after myogenesis (see Fig. 8). Approximately 20 hours after electroporation, no DML-derived fibers were yet apparent (data not shown); this was again different from pioneers, many of which by this time had already differentiated into myofibers (see Fig. 8D). Thirty hours following electroporation, the first labeled fibers became apparent in the desmin-positive myotome, being restricted to its medial region(Fig. 6E). In striking contrast, shortly after electroporation of the mutant cadherin, some labeled cells had already adopted a round morphology and become localized subectodermally (Fig. 6G-J). This phenotype became stronger by 30 hours, when, in addition, the labeled cells completely failed to enter the myotome(Fig. 6K). This effect does not reflect a delay in myotome colonization because a similar picture was observed 42 hours after transfection, at stages following DM dissociation and dermis formation (Fig. 6L, compare with 6F).

When dissociating somites were electroporated with control GFP at their rostral and caudal edges (Fig. 1D), expression of the transgene appeared shortly afterwards in the rostral and caudal lips of the DM, respectively (not shown). A day later,fibers emanating from all along the corresponding lips elongated toward the opposite direction (Fig. 7A,C,see also Fig. S2A in the supplementary material), and Pax7/GFP-immunoreactive cells were also apparent in the myotome (Fig. S2 in the supplementary material, arrows in A and B). By contrast, overexpression of cN390Δ-GFP at this stage caused the dissociation of the lip cells into round progenitors apparent in the intersomitic region, and no myofibers or Pax7-positive progenitors colonized the myotome (Fig. 7B,D, and Fig. S2D-F in the supplementary material). By contrast,overexpression of wtN-cadherin robustly generated myofibers(Fig. 7E).

To begin to understand the mechanism underlying the failure of myofiber differentiation observed in cN390Δ-GFP-treated lips, control and experimental embryos were in situ hybridized to detect Myod. In previous studies, we have shown that Myod mRNA appears in the extreme DM lips in a scattered fashion following establishment of the pioneer myotome,about a day after epithelial somites had formed(Kahane et al., 2001; Kahane et al., 1998a). By this time, control GFP-positive epithelial cells in both lips co-expressed Myod (Fig. 7F,H,J). However, cN390Δ-GFP-treated cells were Myod negative(Fig. 7G,I,K). These results suggest that the cN390Δ-induced early dissociation of the somite epithelium prevented the onset of Myod transcription and the subsequent formation of myotomal fibers. Together with our finding that forced expression of wtN-cadherin or N-cadherin CBR-stimulate myofiber differentiation in otherwise mitotic progenitors, these data support the notion that, in vivo, N-cadherin triggers a cascade of events leading to muscle specification and differentiation.

The origin, migration and pattern of differentiation of pioneer myoblasts was previously documented using lineage tracing with the lipophilic dye DiI,pulse-chase experiments with tritiated thymidine, and expression of Myod,Myf5 and desmin (Kahane et al.,2002). To interfere with N-cadherin activity in pioneers, we first determined conditions to specifically transfect DNA into this cell population. A GFP-encoding DNA was electroporated into the medial somite as shown in Fig. 1A; this method differentiates between the pioneer population and the prospective DML and DM sheet (Fig. 1B,C). The position of fluorescent cells observed 5 hours after transfection(Fig. 8A,B, see also 8E)confirms the specificity of labeling. Consistent with previous results, by 16 hours after transfection, labeled cells have mesenchymalized and are predominantly localized to the rostral half of the segment, revealing a general triangular pattern (Fig. 8C) that, as directly demonstrated elsewhere, reflects caudorostral cell movement (Kahane et al.,1998b; Kahane et al.,2002). In addition, the first fibers to arise are located adjacent to the medial edge and are anchored to the rostral lip of the segment(Fig. 8C, arrow), further substantiating the medial-to-lateral and rostral-to-caudal directions of fiber generation (Kaehn et al.,1988) [see also figure 1B,C in Kahane et al.(Kahane et al., 2002)]. By 24 hours, full-length myofibers are already present that span a significant fraction of the mediolateral extent of the somite(Fig. 8D).

Next, we examined the function of N-cadherin on the formation of the pioneer fibers by electroporating the medial epithelial somite with cN390Δ-GFP (Fig. 8E,F). Electroporation of 23-25ss embryos revealed the formation of a normal myotome(Fig. 8G), similar to control GFP-transfected somites. Similar electroporations performed on younger embryos aged 15-16ss also resulted in the formation of normal myofibers. which were already apparent by 20 hours (Fig. 8H,I). To further test whether preventing the normal downregulation of N-cadherin that takes place upon cell dissociation (see Fig. 2B-E) has any effect on the development of the pioneer fibers, full-length N-cadherin-GFP was similarly delivered. N-cadherin overexpression did not prevent the dissociation of medial epithelial progenitors (data not shown). Furthermore,no effect on the differentiation of pioneer fibers was detected, with the earliest full-length fibers being generated medially and the lateral somite still containing mesenchymal cells 20 hours after transfection(Fig. 8J,K). Thus,N-cadherin-mediated adhesion is not necessary for the development of the first wave of myotomal fibers. Moreover, forced expression of N-cadherin does not impair either their dissociation or their ability to generate fibers.

We addressed the effects of N-cadherin on the development of the three waves that constitute the avian myotome. Although the first wave of pioneer myoblasts is not affected, specification and differentiation of the second wave emanating from the DM lips is severely harmed by a lack of N-cadherin-mediated intercellular interactions (see Table S1, in the supplementary material). Hence, the present results show for the first time a differential role for N-cadherin in myogenic specification and fiber differentiation.

Likewise, the DM sheet, which generates a large proportion of mitotic myotomal progenitors (third wave), as well as dermis, dissociates in the absence of cadherin-mediated adhesion, and cells populate the dermal domain but fail to colonize the muscle. Conversely, full-length N-cadherin or N-cadherin lacking its β-catenin-binding domain (CBR-) trigger translocation of DM progenitors into the myotome at the expense of dermis,followed by muscle differentiation. This process is associated with asymmetric localization of N-cadherin to the apical daughter cells during the apical-basal type cell divisions in the mature DM, suggesting that cadherin is involved in coupling asymmetric cell divisions with cell fate. Altogether,N-cadherin-mediated adhesion is both necessary and sufficient for myotome colonization by both second and third wave myoblasts, thus demonstrating substantial, yet distinct, spatiotemporal effects of N-cadherin on somitic myogenesis in vivo.

An important observation is that N-cadherin differentially segregates to the apical, but not to the basal daughter cells during mitosis in the mature DM. This occurs at a time when a significant proportion of mitotic figures become perpendicularly oriented with respect to the mediolateral axis of the epithelium. In addition, loss of cadherin-mediated adhesion drives cells to the dermal domain where they exhibit features of early dermal cells. Conversely, overexpression of wtN-cadherin or of N-cadherin (CBR-) triggers DM cells to translocate into muscle. Taken together, these lines of evidence suggest that during normal development of the DM sheet, apical daughter cells maintaining N-cadherin translocate into the N-cadherin-positive myotome via homophilic interactions, whereas basal daughter cells that lose N-cadherin become dermis. This supports a role for N-cadherin as a cell surface determinant required for the fate segregation of DM progenitors, a function accomplished by directing cell translocation into myogenic versus dermogenic primordia. These results also suggest that the shift in the orientation of cell divisions in the DM (Ben-Yair and Kalcheim, 2005) reflects a change to an asymmetric mode of cell division, and is therefore of functional significance for fate segregation, as shown during the development of several invertebrate and vertebrate systems(Betschinger and Knoblich,2004).

Previous studies postulated that the contribution of the medial somite to the formation of the myotome occurs via a single mechanism whereby stem cells constituting the DML directly translocate into the myotome and differentiate without migration (Denetclaw et al.,2001; Gros et al.,2004). We found instead that the medial contribution is subdivided into two components: first, the pioneer wave; and, later, fibers from the DML. This distinction is based on several findings. Pioneer myoblasts arise in the medial wall of the still epithelial somite, defining a region different from the future DML. This region expresses Myod and Myf5 and is characterized by a significantly lower rate of cell proliferation, including the presence of post-mitotic cells (Kahane et al., 1998b), suggesting that this is a limited myogenic subpopulation. Hence, it differs from the later DML, which in avians lacks expression of Myod or Myf5, and reveals a continuously high rate of cell proliferation (Ben-Yair et al., 2003). This is consistent with the suggestion that DML cells behave as a continuous source of myoblasts(Ordahl et al., 2001; Venters and Ordahl, 2002). The distinction between fibers derived from pioneers relative to DML was further confirmed in this study, in which GFP-DNA was differentially electroporated into the medial versus the dorsomedial regions of the somite. In the first case, following dissociation of the epithelial pioneers and fiber formation,no residual GFP-labeled cells remained in the epithelium (indicative of a finite cell subset), whereas a similar transfection to the DML always revealed residual labeling of the lip (suggesting a stem-like mechanism). Moreover,pioneer myoblasts generated a myotome with a significant amount of full-length fibers present already by 20 hours after GFP transfection, whereas DML-derived fibers were still absent after 24 hours.

Another unique feature of pioneer myoblasts is that they undergo a polarized migration towards the rostral edge of the DM, from which initial myofiber generation proceeds in a rostral to caudal direction(Kahane et al., 2002). This particular migratory behavior was further confirmed in this study by tracing the cells after focal GFP electroporation (see Fig. 8). The results of our lineage analyses are consistent with the observed expression patterns of Myod, Myf5, desmin (Kahane et al., 2002; Kalcheim et al.,1999), and the flamingo homologue c-fmi(Formstone and Mason, 2005) in the developing myotome.

This mode of initial myotome formation clearly differs from the contribution of the DML that generates fibers through an intermediate SLD(Cinnamon et al., 2001) (see also Denetclaw et al., 2001; Gros et al., 2004). The medial cells involved in generation of the second myogenic wave (the DML itself and the SLD) and the resulting fibers, express N-cadherin throughout the entire process. Consistent with this continuous expression, cN390Δ-GFP caused the rapid dissociation of the DML, as well as that of the DM sheet, with cells that sorted out to become localized subectodermally, and consequent failure of myotome colonization. These data suggest that, in addition to keeping the epithelial integrity of the somite, N-cadherin-mediated homophilic attraction between DM progenitors and the underlying pre-existing myofibers is necessary for progenitor translocation into the myotomal domain.

By contrast, we report that N-cadherin-mediated adhesion is not necessary for establishment of the first wave of pioneer fibers. This might be related to the early downregulation of the protein that is apparent immediately before the dissociation and rostralward migration of the pioneers. It is, however,unlikely to depend upon the state of specification of these progenitors, which already express Myod and Myf5 at epithelial stages, as a similar electroporation of cN390Δ-GFP to epithelial somites of younger embryos (aged 15ss), prior to significant medial expression of the myogenic genes and to their withdrawal from the cell cycle (N. Kahane and C.K.,unpublished), also had no effect on myofiber development.

Overexpression of wtN-cadherin at this stage was also without effect,strongly suggesting that N-cadherin is neither necessary nor sufficient for the earliest progenitors to migrate and generate a myotome. This further highlights the molecular differences between pioneers and second wave progenitors, and suggests that mechanisms other than, or additional to,N-cadherin operate to regulate the formation of this cell population. Unlike the avian pioneers, differential cell adhesion mediated by N- and M-cadherins in the zebrafish myotome drives the migration of adaxial cells(Cortes et al., 2003). Adaxial cells, like the avian pioneers, originate in a medial position in the somite,where they express Myod and Myf5, and migrate laterally to localize superficially in the myotome(Devoto et al., 1996). Furthermore, on their way, zebrafish adaxial cells also exhibit an intermediate phase of rostralward migration(Cortes et al., 2003). Nevertheless, whereas zebrafish adaxials continuously express N- and M-cadherins from their site of origin and throughout migration(Cortes et al., 2003), avian pioneers lose N-cadherin expression prior to or concomitant with the onset of delamination from the medial somite, migrating as cadherin-negative progenitors. Hence, the molecular mechanisms underlying the migration of the avian pioneers seem to differ from those of the zebrafish adaxial cells.

Our findings are consistent with results in which the infection of the chick segmental plate with adenoviruses expressing either full-length N-cadherin or cN390Δ had no effect on early myofiber generation(Horikawa and Takeichi, 2001). Collectively, a wealth of independent evidence, including lineage tracing with DiI and GFP (this study), patterns of marker expression, distinct proliferative behavior, and also functional data on the differential sensitivity to N-cadherin-mediated adhesion (this study), fully validates the notion that pioneers and DML cells are distinct myogenic populations.

During myogenesis, the onset of the second wave of muscle colonization generally follows the formation of the pioneer myotome(Kahane et al., 1998a; Kahane et al., 1998b). The production of myofibers from the extreme lips is associated with the local upregulation of Myod, with transcripts appearing scattered to subpopulations of lip cells (Kahane et al., 2001; Kahane et al.,1998b). The factors inducing the transcription of Myod in these lips had remained unknown. Here, we show that the disruption of N-cadherin-mediated adhesion in the nascent lips prevents the synthesis of Myod and, consequently, the generation of myofibers. Hence,N-cadherin acts by mediating specific adhesive interactions among neighboring lip cells, and/or between lips cells and the pre-existing pioneer myofibers that attach to the lips. These interactions in turn promote signaling events that trigger myogenic specification.

This notion is further supported by the finding that, in the DM sheet,overexpression of N-cadherin not only drives cell entry into the myotome at the expense of dermis, but also causes myofiber formation rather than maintenance of mitotic myotomal progenitors, as is observed under normal conditions. The reason for this phenotypic shift is unclear. One possibility is that in normal development, Pax3/Pax7-positive mitotic cells that enter the myotome as N-cadherin-expressing cells subsequently lose the protein in order to remain in a progenitor state. An additional possibility is that the relatively high amount of protein in N-cadherin-transfected cells is not compatible with maintenance of a mitotic state, whereas physiological amounts are. As this phenotype was mimicked by N-cadherin (CBR-), we favor the interpretation that N-cadherin-mediated adhesion rather than N-cadherin-mediated signaling is responsible both for cell translocation and differentiation. This is further strengthened by the observations that N-cadherin delivered to the central DM sheet, which is normally devoid of Myod, was not sufficient to upregulate Myod in the epithelium (Y.C.,unpublished), and that the N-cadherin-overexpressing cells entered the myotome as Pax3/Pax7-positive progenitors, which generated fibers only 24 hours after transfection into the DM sheet. Therefore, N-cadherin is likely to indirectly affect muscle-specific gene transcription and myogenesis by mediating homophilic cell interactions that enable cell translocation into an appropriate environment. Our results are consistent with the notion proposed on the basis of in vitro paradigms, that local cell-cell interactions are required for myogenesis (Cossu et al.,1995; Gurdon et al.,1993), and that N-cadherin is capable of mediating such interactions by inducing first the assembly of adherens-type junctions and then the expression of muscle-specific markers(Goichberg and Geiger, 1998; Holt et al., 1994). Also, in vitro, activation of N-cadherin was found to stimulate Rho GTPase activity,which in turn activates the serum response factor that enhances muscle-specific transcription and promotes myogenesis (reviewed by Krauss et al., 2005). This molecular cascade couples changes in cell adhesion and morphology with patterns of gene expression and cell differentiation.

We thank all members of our group for discussions and, in particular, Irit Tamir for subcloning full-length N-cadherin-GFP, cN390Δ-GFP and N-cadherin (CBR-)-GFP. We also thank J. Yisraeli for critical reading of the manuscript. We are indebted to M. Takeichi for the N-cadherin DNAs. This work was supported by grants from the Israel Science Foundation (ISF), the EEU 6th Framework program Network of Excellence MYORES, the March of Dimes, ICRF and DFG (SFB 488) to C.K.