Neural crest and somitic mesoderm as paradigms to investigate cell fate decisions during development
Department of Medical Neurobiology, Institute for Medical Research Israel-Canada, and Edmond and Lily Safra Center for Brain Sciences - Hebrew University-Hadassah Medical School, Jerusalem 91120, Israel
Department of Medical Neurobiology, Institute for Medical Research Israel-Canada, and Edmond and Lily Safra Center for Brain Sciences - Hebrew University-Hadassah Medical School, Jerusalem 91120, Israel
The dorsal domains of the neural tube and somites are transient embryonic epithelia; they constitute the source of neural crest progenitors that generate the peripheral nervous system, pigment cells and ectomesenchyme, and of the dermomyotome that develops into myocytes, dermis and vascular cells, respectively. Based on the variety of derivatives produced by each type of epithelium, a classical yet still highly relevant question is whether these embryonic epithelia are composed of homogeneous multipotent progenitors or, alternatively, of subsets of fate-restricted cells. Growing evidence substantiates the notion that both the dorsal tube and the dermomyotome are heterogeneous epithelia composed of multipotent as well as fate-restricted precursors that emerge as such in a spatio-temporally regulated manner. Elucidation of the state of commitment of the precedent progenitors is of utmost significance for deciphering the mechanisms that regulate fate segregation during embryogenesis. In addition, it will contribute to understanding the nature of well documented neural crest-somite interactions shown to modulate the timing of neural crest cell emigration, their segmental migration, and myogenesis.
The dorsal neural tube (NT) that generates neural crest (NC) cells and the paraxial mesoderm-derived somites are pseudostratified epithelia in which progenitor cells undergo interkinetic nuclear migration and typical patterns of cell proliferation (Burstyn-Cohen & Kalcheim 2002; Ben-Yair et al. 2003; Spear & Erickson 2012). These epithelia are transient as progenitor cells either delaminate progressively or fully dissociate to generate mesenchymal cells. In the dorsal NT of avians at trunk levels, these epithelial-to-mesenchymal transitions (EMT) and cell delamination are progressive events lasting for almost two consecutive days during which the dorsal NT preserves its epithelial structure (Krispin et al. 2010b). In contrast, cranial NC cells delaminate en masse and the regulatory networks controlling EMT at each level clearly differ (reviewed in [Strobl-Mazzulla & Bronner 2012]). Comparable processes take place in the dermomyotome (DM), the dorsal remnant of the epithelial somite, which is composed of four contiguous lips and a central sheet. Whereas the lips produce myocytes over a few days while remaining epithelial, the central sheet completely dissociates to produce dermis and mitotic progenitors (Cinnamon et al. 2001; Ordahl et al. 2001; Ben-Yair & Kalcheim 2005; Delfini et al. 2009).
An additional feature shared between both systems is the existence of long and short-range cellular migrations that follow initial EMT. For instance, studies performed in avian embryos revealed that NC progenitors that form sympathetic ganglia migrate a long distance from the dorsal NT toward the dorsal aorta, and then along the aorta for a length of three consecutive somites rostral and caudal to their exit point from the NT (Fig. 1) (Yip 1986). In contrast, sensory progenitors of the dorsal root ganglia (DRG) migrate dorsoventrally a short distance until reaching the DRG primordium and also move longitudinally for the length of about one and a half segments to populate two consecutive DRG in a stereotypic pattern (Fig. 1B) (Teillet et al. 1987). Similarly, the ventrolateral lip of the DM at brachial and lumbosacral levels of a variety of species undergoes dissociation and Pax3/7-positive progenitors migrate a long way into the limb primordia to generate the appendicular muscles (Buckingham et al. 2003). In addition, endothelial cells derived from the lateral DM of a given somite colonize blood vessels along at least two segments (Ben-Yair & Kalcheim 2008). Different from the precedent cases, progenitors of myotomal muscles either translocate directly from the DM into the adjacent myotome and then differentiate into mononucleated, unit-length myofibers (Fig. 1B) (Kalcheim et al. 1999; Gros et al. 2004; Kalcheim & Ben-Yair 2005; Ben-Yair et al. 2011) or, as in the case of avian pioneer myoblasts, undergo a Slit1/Robo2-dependent caudal to rostral migration within a single segment followed by differentiation in the opposite direction (Halperin-Barlev & Kalcheim 2011). Likewise, in zebrafish, slow twitch muscle cells derived from adaxial progenitors were shown to undergo a N- and M-cadherin-dependent migration from the medial notochord to the lateral aspect of the myotome within the same segment (Cortes et al. 2003).
Altogether, NC and dorsal somite progenitors share many similar features during both epithelial and mesenchymal states. Nevertheless, heterogeneity in EMT patterns and migratory behavior are apparent within each system along the axis and at various stages. Notably, the most important common characteristic of the above ectodermal and mesodermal progenitors, respectively, is the rich variety of derivatives that each system produces. The NC is the source of neurons, glia and Schwann cells of the peripheral nervous system (PNS) (sensory, sympathetic, parasympathetic and enteric ganglia). It also generates pigment cells of the skin, chromaffin cells in the adrenal gland, ectomesenchyme in the head and the cardiac outflow tract (Bronner-Fraser 1993; Le Douarin & Kalcheim 1999). The dorsal somite is the source of the DM epithelium, which in turn produces epaxial and hypaxial muscles of the body, appendicular muscles, epaxial dermis, smooth muscles and endothelium lining blood vessels (Christ & Scaal 2008) (Fig. 1). These multiple fates raise the obvious question whether the epithelia of origin are composed of naïve and homogeneous precursors or, alternatively, of already fate-restricted cells. In the former case, one would postulate that cell fate decisions are taken during migration or at the homing sites thus imposing the search for responsible mechanisms operating after EMT. In the latter event, research should focus on mechanisms operating within the epithelium itself that would account for the emergence of an early heterogeneity. The present review analyzes the available evidence suggesting heterogeneity in the state of commitment of progenitors within each type of epithelium and briefly overviews some known mechanisms. This knowledge is particularly important in view of well documented NC-somite interactions, the nature of which depends on the precise identity and properties of the interacting progenitors at different times and locations.
Lineage analysis of NC progenitors
It has long been appreciated that at different axial levels, NC cells adopt various fates and migrate through unique pathways (Le Douarin & Kalcheim 1999). For instance, at rhombencephalic levels, NC cells primarily migrate along discrete dorsolateral pathways ventral to the ectoderm. These progenitors generate specific sensory ganglia and pigment and also colonize the pharyngeal arches to give rise to bones and cartilage (Kulesa & Gammill 2010; Minoux & Rijli 2010). In contrast, at trunk levels, NC cells first migrate ventrally to give rise to chromaffin cells, neurons and glia of the PNS, and later migrate dorsolaterally to give rise exclusively to melanocytes.
Notably, at vagal levels (somite levels 1–7) differences were detected in the migratory pathways between the rostral vagal region (somite levels 1–4) and the caudal vagal domain of avians (somite levels 5–7). In the former, three sequential waves of cell migration were reported: an initial wave is subectodermal resembling the behavior of cranial NC. Progenitors moving along this route mainly colonize the pharyngeal arches and the heart. A later stream is dorsoventral and likely contributes to neural derivatives including enteric innervation. The final wave is a distinct subectodermal stream that generates melanocytes. However, at somite levels 5–7 the temporal segregation of migratory pathways resembles the trunk with a first general pathway in the dorsoventral direction yielding neural derivatives and a later dorsolateral path that produces melanocytes. Hence, the vagal level of the axis was suggested to behave as a transition zone between NC of the head and body (Kuo & Erickson 2010, 2011), a phenomenon likely to bear evolutionary significance.
A more refined map relating the pathways of NC migration with the final fates adopted by the cells was described in the trunk of both avian and mouse embryos (Serbedzija et al. 1989, 1990; Krispin et al. 2010b). In the chick, it has been shown that NC cells colonize their definitive domains in a stereotypic ventral to dorsal order. The first cells to delaminate migrate farthest to colonize the sympathetic primordium, followed by Schwann cells that line along the spinal nerves. Later, the DRG domain is settled, and finally melanocyte progenitors populate the skin (Serbedzija et al. 1989; Krispin et al. 2010b). Thus, the time of NC migration is associated with the production of specific derivatives.
This raises the question whether the fates of NC cells can be predicted earlier than at migratory stages, that is, already within the premigratory domain of the dorsal NT. To address this question, Krispin et al. (2010b) labeled the dorsal midline of the NT at sequential times. Discrete cell subsets were labeled by focal electroporation of GFP-DNA and single cells by direct plasmid injection (Ben-Yair & Kalcheim 2010) (Fig. 2A). Consistent with results of more massive cell labeling, the earliest delaminating cells generated SG and the latter gave rise to melanocytes, according to the previously established order. To complement this approach, the authors discretely labeled epithelial progenitors at various dorsoventral locations within the dorsal NT at a given stage. In this paradigm, dorsal progenitors of the NT formed ventral derivatives and, conversely, ventral epithelial progenitors generated melanocytes. Hence, the dorsoventral localization of a premigratory NC cell and its final position after homing are inversely related. Furthermore, using time lapse analysis, it was shown that concomitant with successive cell emigration, presumptive NC cells undergo a ventral-to-dorsal relocation, suggesting that the midline region of the NT acts as a sink for the ordered ingression and departure of progenitors. Altogether, both approaches showed that the final localization of NC-derived cells and hence their fates can be predicted from their relative position within the premigratory domain and/or by their time of delamination and that these two parameters are fully equivalent. Therefore, a dynamic spatio-temporal fate map of NC derivatives of the trunk is already apparent in the dorsal NT (Krispin et al. 2010b). Furthermore, since the domain of expression of the premigratory NC markers Snail2, Sox9 and Foxd3 was shown to be initially broad in the dorsal tube and to become progressively restricted to the dorsal midline (Krispin et al. 2010b). A parallel study in which NC behavior was tracked by time-lapse imaging in slice cultures did not fully appreciate this phenomenon and claimed instead that trunk NC cells emigrate from any region of the dorsal neural tube, and not solely from a discrete dorsal region that would correspond to the Foxd3+ domain at the relatively late time at which they performed their experiments (Ahlstrom & Erickson 2009). In this study, slice cultures may have lost the normal tension present in the embryo that is putatively required for accurately monitoring such morphological events. In addition, the relatively short duration of the imaging sessions may not have been long enough to track the dynamic changes in the pattern of cell emigration of the entire progeny.
While the above described fate map is relevant to the flank level of the neuraxis, additional research is needed to unravel whether analogous premigratory NC fate maps exist at additional axial levels and when, during NC ontogeny, can such fate maps be first detected. Since unlike trunk NC progenitors, cephalic NC cells delaminate en masse during a short period rather than as individuals during a prolonged time, creating such a fate map is a difficult task. Despite these differences, it was shown that in cephalic regions of the zebrafish embryo, premigratory lateral progenitors emigrate first and generate neurons, more medial cells migrate later and generate Schwann and pigment cells, while the most medial progenitors form ectomesenchyme (Schilling & Kimmel 1994).
The results described above uncovered an unprecedented order linking the localization of NC progenitors in the dorsal NT and the time of delamination and migration with fate. Additionally, it was found that in at least 80% and 90% of cases, the progeny of presumptive trunk NC cells labeled right before emigration was confined to a single derivative in both zebrafish (Raible & Eisen 1994) and chick (Krispin et al. 2010b) (Fig. 2A), respectively. This suggests that most premigratory NC cells are lineally restricted. This does not necessarily reflect full commitment as, once arrived to the ganglionic primordia, NC progenitors may still chose between a neuronal or satellite cell fate (Wakamatsu et al. 2000). In apparent contrast, when lysinated rhodamine dextran (LRD) was injected into single dorsal NT progenitors, most clones contained labeled progeny in multiple NC sites, not always respecting the sequential order of colonization discussed above. Other clones were composed of both NT and NC or only of NT cells (Bronner-Fraser & Fraser 1988, 1989) favoring the notion that premigratory NC cells are multipotent. Possible differences between these studies may reside in the accuracy of the stages and axial levels analyzed. In the analyses by Bronner-Fraser and Fraser, embryos ranged considerably in stages and somitic levels making it possible that many injections labeled primitive neuroectodermal progenitors, in which fate segregation between NC and NT or between various NC fates may not have occurred yet. To solve these apparent discrepancies and to accurately determine the onset of NC restrictions in the trunk, a systematic clonal analysis should be performed beginning at neural plate stages until the onset of NC EMT. Based on recent evidence, such a study should independently consider precise axial levels as well as dorsoventral locations of presumptive NC progenitors within the NT (or equivalent mediolateral positions in the neural plate).
The in vivo tracing of cell lineages described above reflects the state of commitment of a progenitor at a given site and time in development within a normal embryonic context. It is widely accepted that the state of specification of a progenitor cell is narrower than the potential fates it can generate upon exposure to external or intrinsic conditions other than those imposed by the embryonic environment. In vitro clonal analyses of NC cells performed by several authors adequately exemplify this concept and this is briefly elaborated below (see section on Challenging the state of commitment of NC progenitors). An example of an intermediate situation is provided by the analysis of cell lineage segregation within a cohort of NC cells emigrating from explanted neural primordial (Henion & Weston 1997). In this study, LRD was injected into single avian NC progenitors soon after they delaminated onto a planar substrate on which the general order of cell emigration (neural progenitors first and then melanocytes) as well as intercellular interactions were maintained (Reedy et al. 1998). They found that nearly half of the NC population, labeled 1–6 h after initial cell emigration from the NT, already comprises fate-restricted precursors which generated a single cell type; either neurons, glia or melanocytes. This proportion increased to almost 90% when analyzing the fates of late emigrating progenitors, labeled 30–36 h after the onset of NC emigration. They concluded that NC progenitors segregate early by asynchronous restriction of distinct cell fates.
Together, growing evidence suggests that the premigratory pool of NC progenitors is not homogenous as initially hypothesized. Instead, while not revoking the existence of multipotent cells even at later stages, it is clear that fate-restricted progenitors are present already in the dorsal NT. Future studies are needed to understand the precise timing of cell fate acquisition in the dorsal neurepithelium and the underlying cellular and molecular mechanisms (Krispin et al. 2010a).
Challenging the state of commitment of NC progenitors
The lineage tracing studies discussed above suggested the existence of early fate restrictions of NC cells. This notion is further supported by in vivo and in vitro experiments in which the migratory environment of these progenitors was altered, thereby challenging their potential to adopt new identities.
The ultimate test for examining the potential of NC progenitors was to isolate single cells from delaminating or migratory stages stemming from various axial levels and culture them in the presence of defined growth and survival factors and/or onto cellular substrates such as growth-arrested 3T3 fibroblasts (see below). This approach revealed that early NC cells are a heterogenous population, where both multipotent, intermediate and fate-restricted precursors can be found. Notably, cephalic quail NC clones contained, on the one hand, a pluripotent NC-like stem cell at a frequency of 1/305 clones, while, on the other hand, 20% of clones contained progeny bearing a unique phenotype (either neurons, cartilage, or Schwann cells) even under conditions that were permissive for proliferation and differentiation along multiple lineages (Baroffio et al. 1990, 1991; Le Douarin & Dupin 2003; Trentin et al. 2004; Le Douarin 2008). Likewise, it was found that while some NC cells are pluripotent (Sieber-Blum & Cohen 1980), at least three lineages are committed prior to migration – sensory neurons (Sieber-Blum 1989), most pigment cells of the trunk (Sieber-Blum & Cohen 1980) and smooth muscle cells at cranial levels of the axis (Sieber-Blum & Zhang 1997).
To challenge the potential of NC cells in vivo, heterotopic and heterochronic transplantations were performed. In avians, this technique was pioneered by Le Douarin and colleagues using quail-chick chimeras, which demonstrated that, at a population level, the potential of many but not all NC progenitors is by far larger than their actual lineage (reviewed in [Le Douarin 1982]). Since grafts were performed at early stages (E2) and embryos were analyzed following full differentiation (E7 or later), the possibility of selective mechanisms operating in the chimeras, such as death of “inappropriate” cells including that of committed progenitors that were unable to develop in ectopic locations, could not be excluded. In fact, the existence of such a mechanism has already been described (Wakamatsu et al. 1998). In the above study, neurogenic neural crest precursors were transplanted between ectoderm and dorsal somites, and a transient wave of death of the grafted cells was observed following a brief period of expression of neuronal traits within this pathway.
In zebrafish, when early neural cells were transplanted into a late migratory environment, the cells maintained their neural fate and gave rise to derivatives in DRG. Consistently, when late emigrating pigment cells were introduced into an early migratory environment, they failed to produce neuronal fates (Raible & Eisen 1994, 1996). Experiments in avian embryos allowed to develop for shorter times compared to classical quail-chick chimeras, produced a similar result. Neural progenitors were introduced into a melanocytic environment by missexpressing the guidance receptor Ednrb2 (Harris et al. 2008), by heterochronic transplantations or combining both methods. Diverted neuronal cells did not alter their fate to adapt to the new environment. Instead, they maintained their own identity by upregulating an array of neuronal markers in an ectopic time and location. As expected, these neurons did not survive in the ectopic sites beyond a few days (Krispin et al. 2010b). Furthermore, transient inhibition of NC migration for 24 h was performed by transfecting a tetracycline-dependent dominant negative form of Rac1. Upon release of the inhibition at a late stage corresponding to the onset of lateral migration, NC progenitors nevertheless migrated dorsoventrally toward the DRG where they generated neurons and glia despite being exposed to a melanocytic environment (Shoval & Kalcheim 2012). Along the same line, ectopic migration of early NC progenitors was observed in mice lacking neuropilin-1 gene activity. These ectopically migrating cells, however, maintained sympathetic or sensory fates, further suggesting early restrictions of the neural progenitors of the NC (Schwarz et al. 2009). In further support of this notion, several markers were reported to be differentially expressed in crest subsets and specific crest-derived lineages. An example is provided by the Seraf gene, a secreted protein with EGF-like repeats, that was found to be expressed from early stages onward in a subset of ventrally migrating crest cells followed by expression in the Schwann cell lineage along with P0 (Wakamatsu et al. 2004).
Mechanisms underlying early fate specification of NC progenitors
An increasing body of experimental evidence addresses possible mechanisms of segregation of NC progenitors into its various sublineages (see for example [Harris & Erickson 2007; Lallemend & Ernfors 2012; Pavan & Raible 2012]). Here, we will focus on some molecular mechanisms likely to operate either in the dorsal NT or in early migrating cells. The existence of a premigratory fate map of NC derivatives in the avian trunk along with the demonstration of early fate-restricted cells in the NC of various species and axial levels (see above) indicates that some fate decisions are taken already at the premigratory stage. Along this line, epithelial NC progenitors could be exposed to graded effective concentrations of a dorsal NT signal prior and/or during the period of NC delamination. Such a gradient would operate along the dorsoventral extent of the premigratory pool at a given time. Since premigratory epithelial progenitors progressively relocate dorsalward before exiting the tube (Krispin et al. 2010b) it could be envisaged that presumptive melanoblasts are exposed for longer durations to such a factor while transiting from low to high factor concentrations. Reciprocally, putative sympathoblast progenitors, localized closer to the midline, would be exposed to high concentrations for a shorter time as they emigrate first (Krispin et al. 2010a).
BMP4 fulfills the requirements to be such a factor as it is expressed in the dorsal NT both early during NC ontogeny and also later when the definitive roof plate is established. Hence, the premigratory pool of NC cells could respond to BMP4 in a spatiotemporally regulated manner. In fact, BMP4 is produced by the NC cells themselves, which in turn respond to it, likely by a paracrine mechanism, since they express the BMP Receptor type 1A (but not type 1B) during their stay in the dorsal neural primordium and progressively downregulate receptor transcripts following emigration (Sela-Donenfeld & Kalcheim 2002). Local BMP signaling was shown in the dorsal NT to stimulate Wnt1 production (Sela-Donenfeld & Kalcheim 1999, 2002; Burstyn-Cohen et al. 2004; Chesnutt et al. 2004) thus inducing EMT of NC cells (Burstyn-Cohen et al. 2004). In addition, both BMP and Wnt signaling systems were associated with differential fate acquisition by NC progenitors (see below), hence implicating BMP/Wnt as factors with a dual activity on the generation of cellular movement as well as on differential fate specification. BMP and Wnt also play a role during secondary neurulation in the caudalmost neural tube. In this region, only melanocytes and glia are produced by the NC with virtually no neurons. This defect was shown to result from the persistence of the BMP inhibitor noggin, which also impaired Wnt1 transcription. Missexpression of BMP restored the neurogenic potential of the caudal NC while preventing cell apoptosis otherwise observed during normal development in this region (Osorio et al. 2009). Hence, regulated levels of BMP/Wnt1 activity are likely to play a role in cell survival and differential lineage segregation.
It would be significant to understand whether and how does the epithelial versus mesenchymal conformation of a NC cell affect its state of specification. An example along this line is provided by the expression of N-cadherin in the premigratory progenitors, which not only serves to maintain epitheliality but also to inhibit canonical Wnt signaling (Shoval et al. 2007) likely by recruiting β-catenin to the apical junctions at the expense of the pool of β-catenin that translocates into the cell nucleus where it acts as a transcriptional activator.
The involvement of BMP signaling in determining the fate of NC progenitors is exemplified by a recent study in zebrafish embryos. In cephalic regions, Twist1 was found to play an essential role in promoting ectomesenchyme at the expense of non-ectomesenchymal fates such as PNS and pigment derivatives. Twist1 biases development of the ectomesenchyme lineage by inducing fibroblast growth factor (FGF) signaling and by activating fli1a through a conserved ectomesenchyme-specific enhancer. BMP stemming from the dorsal neuroepithelium signals through Id2a to inhibit Twist1 activity. Thus, cranial NC cells migrating ventrally and away from the dorsal source of BMP are relieved from repression by Id2a, favoring ectomesenchyme specification, while cells that remain close to the source of BMP become specified to a neural fate (Das & Crump 2012). Therefore, transient exposure to BMP signaling can influence the fates adopted by early NC cells.
Wnt1 was found to act downstream of BMP in the dorsal neural tube. Consistent with the notion that Wnt may be part of a signaling network that operates in the neuroepithelium to regulate NC specification, it was found that at trunk levels of mouse embryos, sensory neurons and melanocytes are successively generated by sequential β-catenin activity. Excess β-catenin at premigratory stages promoted sensory neurogenesis at the expense of virtually all other NC derivatives, while excess β-catenin at migratory stages stimulated ectopic melanogenesis. Thus, since the same signaling module underlies development of distinct phenotypes as a function of time, the state of commitment of the target progenitors might already differ between the above stages (Lee et al. 2004; Hari et al. 2012).
Initial evidence points to a molecular difference between putative progenitors within the dorsal neural tube. Foxd3, Snail2 and Sox9 were found to be expressed in neural but not melanogenic avian progenitors already at premigratory stages (Krispin et al. 2010b). These genes may serve as fate determinants to drive neural and/or inhibit melanocyte development, respectively ([Thomas & Erickson 2009; Krispin et al. 2010b], Nitzan et al., unpubl. data, 2012). Transcription of all three genes is positively regulated by BMP4 and Wnt1 (Sela-Donenfeld & Kalcheim 1999; Burstyn-Cohen et al. 2004), suggesting the existence of a premigratory gene regulatory network that comprises signaling and transcription factors responsible for early fate decisions. Whereas the precise role(s) of each of the above factors and the nature of their interactions remain to be elucidated, additional genes that exhibit differential spatial and/or temporal expression in the dorsal neural tube should be identified to further validate the above hypothesis.
Working in the zebrafish, the Eisen group provided an elegant model for a fate switch between Rohon-Beard (RB) cells and sensory neurons in the DRG. DeltaA mutants showed excess of RB cells in the dorsal spinal cord but lacked trunk NC, although cephalic NC development remained intact. In addition, Neurogenin 1 (Ngn1) was first required for specification of RB cells and only later required in DRG cells. Furthermore, Delta/Notch signaling was found to repress Ngn1 expression, thereby promoting the NC fate in the lateral neural plate (Cornell & Eisen 2000, 2002). Perez and colleagues (1999) found that in the chick embryo, a subset of NC cells expresses Ngn1/2 early after emigration and that these cells are biased to a sensory fate. Overexpression of these transcription factors localizes cells to the DRG and induces the expression of sensory neuron markers. However, mouse cells expressing a reporter for Ngn2 were found to equally contribute to sensory neurons and glia (Zirlinger et al. 2002), a fate choice that probably takes place after arrival to the target site (see below). In addition, tracing the fate of a subset of contralaterally migrating NC cells revealed that they generate specifically nociceptive neurons in the DRG (George et al. 2007), further supporting the idea of an early specification of some NC cells to the sensory lineage.
As mentioned above, some fate choices adopted by NC precursors take place after cell homing. For example, the choice whether to become a neuron or a glial cell takes place within the ganglion primordium. After populating the DRG, progenitors were shown to undergo asymmetric cell division, upon which Numb- and Delta-expressing cells become the neuronal core of the ganglion, while Notch-expressing cells form glia mainly localized to the periphery. Experimental manipulations confirmed that this choice depends on a mechanism of Notch-mediated lateral inhibition (Morrison et al. 2000; Wakamatsu et al. 2000).
An additional target site where late fate decisions are taken is along peripheral nerves exiting the spinal cord, which are lined by NC-derived Schwann cell progenitors. The latter cells were reported to be an additional source of melanocytes (Adameyko et al. 2009). Recently, a mechanism was proposed by which reciprocal inhibition between Sox2 and Mitf dictates the decision to become either Schwann cells or melanocytes, and that downregulation of Sox2 in nerve-derived Schwann cell progenitors is necessary for upregulation of Mitf and consequent specification to the melanocytic lineage (Adameyko et al. 2011).
A spatio-temporal fate map of DM derivatives
The DM is composed of a central sheet limited by four lips. In spite of initial controversy stemming from various studies in avian embryos, it is now well accepted that the four DM lips generate fully elongated myocytes (Fig. 1) (Kahane et al. 1998, 2002; Cinnamon et al. 1999, 2006; Denetclaw & Ordahl 2000; Huang & Christ 2000; Gros et al. 2004). More recently, the young, still epithelial, central sheet was also shown to produce myocytes by a mechanism that remains to be fully elucidated (Ben-Yair et al. 2011). Current knowledge reveals that the above progenitors delaminate into the underlying myotome where they transiently span its entire thickness and keep apico-basal polarity before differentiating into unit-length fibers. Hence, the entire DM has the capacity to generate myotomal myocytes (Fig. 1B).
At a later stage, the central sheet of the DM dissociates producing dermis and also Pax3/7-positive muscle progenitors that remain mitotically active within the myotome (Kahane et al. 2001; Ben-Yair & Kalcheim 2005). These progenitors were shown in both mouse and chick to subsequently develop into fibers or satellite cells (Gros et al. 2005; Kassar-Duchosoy et al. 2005; Relaix et al. 2005). Importantly, both mitotic myoblasts and dermis originate from single DM cells that can therefore be considered to be at least bipotent under normal conditions (Ben-Yair et al. 2003; Ben-Yair & Kalcheim 2005).
The DM is also a source of vascular cell types in the avian embryo. Lineage analysis performed at the single cell level revealed that of all epithelial domains, the lateral region is the most prolific producer of smooth muscle and endothelium. Typical locations of endothelial and smooth muscle phenotypes produced by the DM were the cardinal veins and vitelline arteries, mesonephric, dermal and somatopleural vessels. Notably, production of endothelial cells was maximal in the lateral epithelial somite (E2) and progressively diminished with development. The proportion of smooth muscle cells was highest both at E2 and E2.5 when compared to E3 when myotomal cells were the major derivative. These results suggest an ordered time course of lineage segregation from the lateral portions of the somite and subsequent DM with an overlap in the time of generation of smooth and striated muscle sublineages (Ben-Yair & Kalcheim 2008). This is particularly significant, as retrospective lineage analysis in transgenic mice using the nlaacZ reporter showed that endothelial, smooth muscle and striated muscle cells share a common early progenitor present before somitogenesis (Esner et al. 2006).
Unlike at flank levels of the axis, fate analysis of lateral somite cells performed at hindlimb levels proved that a significant proportion of single progenitors produce both endothelial and striated muscle cells (Kardon et al. 2002a) suggesting that fate segregation at hindlimb levels is a later process as lateral progenitors delaminate and migrate extensively into the limb prior to overt differentiation.
Altogether, avian embryos proved to be extremely useful for lineage tracing of discrete DM subdomains owing to the feasibility of performing spatio-temporally controlled focal electroporations and single cell injections of lineage tracers (Fig. 2B) (Ben-Yair & Kalcheim 2010). These studies highlight the heterogeneity of the epithelial somite and the DM raising the question of the mechanisms responsible for fate segregation (see below).
The state of specification of DM progenitors
Lineage analysis of DM progenitors showed that segregation into muscle, vascular and dermal lineages occurs in a spatio-temporally regulated fashion (see above and [Scaal & Christ 2004; Kalcheim & Ben-Yair 2005; Kalcheim et al. 2006; Ben-Yair & Kalcheim 2008; Buckingham & Montarras 2008; Buckingham & Vincent 2009]). Therefore, the DM is an attractive model to investigate the molecular mechanisms underlying cell-fate decisions operating in an epithelium.
For example, maturation of the central DM sheet involves a striking shift in the plane of epithelial cell division from an initial planar orientation in which the mitotic spindle is oriented parallel to the mediolateral extent of the DM sheet into a perpendicular orientation prior to cell dissociation that generates one apical and one basal daughter cell and this shift was shown to depend upon the function of LGN. LGN is the vertebrate homologue of Drosophila Partner-of-Insc (Pins), which is essential for spindle positioning by linking the cell cortex with the mitotic spindle (Du et al. 2001; Gotta et al. 2003; Du & Macara 2004; Sanada & Tsai 2005; Siller et al. 2006). LGN-dependent planar cell divisions in the early DM sheet were shown to be required for maintenance of symmetric divisions that allocate progenitors to either DM (self-renewing progenitors) or to the myotome as myocytes. Furthermore, the normal 90° shift in the plane of cell division prior to epithelial dissociation was shown to be essential for generating a balance between muscle versus dermal fates. Hence, the central epithelium generates sequentially myocytes and then mitotic myoblasts and dermis, and LGN-dependent orientation of cell divisions is critical for fate segregation at both stages (Ben-Yair et al. 2011). Whether this process involves the asymmetric allocation of cell fate determinants to daughter cells, or merely results in the differential translocation of multipotent cells to myotome and dermis where actual specification occurs, remains to be clarified. Along this line, N-cadherin was shown to be equally inherited by both daughter cells during planar mitoses in the DM and only by the apical daughter cells during perpendicular cell divisions. Furthermore, gain and loss of N-cadherin activity resulted in DM progenitors colonizing the myotome and dermis, respectively, (Cinnamon et al. 2006). It is not clear, however, whether N-cadherin acts as a fate determinant in the DM or merely as a cell adhesion molecule that, via homophilic interactions with myotomal cells that also express the protein, drives cell homing to the myotomal primordium at the expense of dermis. A concentration of perpendicular mitoses was reported to predominate in the DM dorsomedial lip (Venters & Ordahl 2005). The occurrence of such divisions was closely associated with asymmetric localization of the Notch pathway factor Numb, defining such divisions as asymmetric. In contrast, planar divisions were devoid of Numb and were therefore termed symmetric. It is tempting to speculate that in the above perpendicular mitoses, the Numb-expressing daughter cells translocate into the myotome, whereas the Numb-negative cells remain in the DM epithelium as proliferating precursors, yet no experimental data are available to support this notion.
An additional paradigm to study lineage segregations is the lateral somite/DM of avians where clones derived from individual precursors gave rise to either endothelial or smooth muscle cells but not both, indicating an early segregation of the vascular lineages. In contrast, mixed clones containing both smooth and striated muscle cells were apparent (Fig. 2B lateral). Consistently, BMP signaling was required for endothelial cell differentiation and/or migration but inhibited striated muscle differentiation (Kahane et al. 2007; Ben-Yair & Kalcheim 2008). In addition, Notch activity was reported to be necessary for smooth muscle production while inhibiting striated muscle differentiation. Hence, the choice to become smooth versus striated muscle depends upon Notch signaling (Ben-Yair & Kalcheim 2008). In the mouse, Buckingham and colleagues (Lagha et al. 2009) identified the transcription factor Foxc2 as a gene that is negatively regulated by Pax3/7 and showed that they inhibit each other's expression. Compound mutant analyses and manipulation of somite explants indicate that the Pax3/Foxc2 ratio affects myogenic versus vascular cell fate choices, respectively. In contrast, in the limb, Pax3 does not appear to commit lateral dermomyotomal cells to a muscle cell fate (Kardon et al. 2002b). Future studies should unravel the relationship between signaling systems such as Notch and BMP and these downstream transcription factors in segregating between the above lineages.
Interactions between NC and somites control multiple events
The actual induction of the NC is influenced by interactions with mesodermal tissue (see [Basch & Bronner-Fraser 2006; Basch et al. 2006; Stuhlmiller & Garcia-Castro 2012] for comprehensive reviews). The dorsolateral marginal zone (DLMZ) of the Xenopus gastrula that lies ventral to the prospective NC and generates paraxial and intermediate mesoderm is the source of NC-inducing signals (Fig. 3A). Recombination experiments with the DLMZ and animal caps or grafts of paraxial mesoderm into ventral epidermis stimulated expression of NC markers (Bonstein et al. 1998; Monsoro-Burq et al. 2003; Steventon et al. 2009). Furthermore, explants of the NC at the neurula stage did not maintain expression of NC markers unless co-cultured with paraxial mesoderm, implicating the mesoderm in the maintenance phase of NC identity (Bonano et al. 2008; Steventon et al. 2009). The DLMZ expresses multiple Wnt and FGF ligands and the BMP antagonist Chordin (Mayor et al. 1995; Monsoro-Burq et al. 2003; Hong et al. 2008; Steventon et al. 2009) (Fig. 3A), molecules known to be involved in NC induction. Likewise, in avians, recombination between nascent neural tissue and somites or lateral mesoderm also generated NC-derived melanocytes (Selleck & Bronner-Fraser 1995).
The timing of NC EMT
A significant body of evidence, primarily stemming from avian embryos, relates the onset of NC migration in the trunk with somitogenesis and subsequent somite dissociation (Loring & Erickson 1987; Teillet et al. 1987). At the level of the segmental plate, presumptive NC progenitors are still confined to the dorsal NT (Teillet et al. 1987). Emigration of the first NC cells becomes apparent at levels opposite epithelial somites. Furthermore, upon somite dissociation, they continue exiting the neuroepithelium and begin simultaneously invading the somite in a segmental fashion (Fig. 1B) (Loring & Erickson 1987; Teillet et al. 1987; Kalcheim & Teillet 1989; Debby-Brafman et al. 1999). This suggested that the paraxial mesoderm regulates aspects of NC EMT and emigration. However, in the rostral trunk, signals triggering NC EMT may not be under the strict control of the somites as at early stages, there is an asynchrony between somitogenesis and neural crest departure (Newgreen & Erickson 1986).
An interplay between noggin and BMP4 in the dorsal NT was found to generate a graded activity of the latter that, via regulation of Wnt1 transcription and Wnt-dependent canonical signaling, triggers delamination of NC progenitors and the consequent onset of cell migration (Sela-Donenfeld & Kalcheim 1999; Burstyn-Cohen et al. 2004) (Fig. 3B). This rostral-to-caudal gradient of BMP4 activity is generated in spite of a constant level of BMP4 mRNA production along the dorsal NT by virtue of an opposing, decreasing gradient of noggin transcription and activity that also correlates with somite development. Downregulation of noggin progressively relieves inhibition of BMP and allows NC EMT. How is this graded expression of noggin generated in the dorsal NT remained unclear. Due to the temporal correlation between noggin levels and somite development, somitic factors were suggested to influence the levels of noggin mRNA in the NT. Consistent with this notion, grafting experiments revealed that dissociating, but not younger somites, emit an inhibitor of noggin production in the dorsal NT thereby coupling the time of EMT with the development of the somites as suitable substrates for subsequent NC migration (Sela-Donenfeld & Kalcheim 2000). Nevertheless, the precise identity of this factor(s) remained unknown for years. Recently, opposing gradients of FGF and retinoic acid, apparent in the paraxial mesoderm, were reported to control the timing of NC EMT, in addition to affecting specific aspects of NC induction. A decrease in FGF at somitic levels was suggested to be required for noggin downregulation. In contrast, retinoic acid was not necessary for modulating noggin transcription. In addition, FGF signaling prevented the premature expression of Wnt1, whereas retinoic acid triggered Wnt1 transcription at axial levels that contain specified NC progenitors. Hence, counter-gradients of FGF and retinoic acid affect NC EMT in part through the modulation of specific aspects of the BMP and Wnt signaling pathways (Fig. 3B) (Martinez-Morales et al. 2011), previously shown to set in motion the process (see above).
In this context, it is relevant to mention that ectodermal-derived Wnt6 promotes initial epithelialization of the somite, yet Wnt6 is downregulated upon DM formation (Schmidt et al. 2004). At a later stage, FGF8 from the myotome was suggested to promote dissociation of the overlying central DM via a Snail1-dependent mechanism (Delfini et al. 2009). Hence, early ectodermal Wnt followed by FGF signaling from the paraxial mesoderm and its derivatives play a direct or indirect role in EMT of both NC and DM precursors. Whereas downstream mechanisms leading to NC EMT have been extensively studied, we still lack a molecular understanding of the processes responsible for dissociation of the DM sheet, or for progressive cell delamination from the DM lips, except for the ventrolateral edge at brachial and lumbosacral levels of the axis where hepatocyte growth factor/scatter factor through the Met receptor triggers EMT and migration into the limb primordium (Dietrich et al. 1999; Vasyutina & Birchmeier 2006).
Segmental migration of NC and patterning of the peripheral nervous system
The PNS of higher vertebrates is segmented to align the peripheral ganglia and nerves with the vertebrae. This pattern is established during embryogenesis, when vertebrae develop from the somite-derived sclerotome and NC cells preferentially migrate into the rostral sclerotome halves (Fig. 1). Grafting experiments in avian embryos showed that the metameric arrangement of the PNS depends upon the rostro-caudal alternation of sclerotomal properties (Keynes & Stern 1988; Keynes et al. 1990; Kalcheim & Goldstein 1991; Fraser 1993; Kalcheim 2000), and so is the development of specific components of the vertebrae (Goldstein & Kalcheim 1992; Bruggeman et al. 2012). Several molecular families were shown to mediate segmental NC migration primarily through repulsive interactions between the caudal sclerotome and the NC cells; these include Eph and Ephrins, F-spondin, Neuropilins and Semaphorins, T-cadherin, etc. (Krull et al. 1997; Debby-Brafman et al. 1999; Krull 2001; Gammill et al. 2006; Kuriyama & Mayor 2008; Roffers-Agarwal & Gammill 2009; Schwarz et al. 2009). For instance, the repulsive guidance cue SEMA3A from the caudal sclerotome and DM and its receptor neuropilin 1 (NRP1) are essential to direct the migration of NC cells through the rostral sclerotome from a default pathway alongside intersomitic blood vessels. Loss of function for either gene caused excessive intersomitic NC migration, loss of segmentation and ectopic neuronal differentiation. Hence, switch from the intersomitic to the sclerotome path is a pre-requisite for the proper pattering of the PNS and it is controlled by SEMA3A/NRP1 interactions between sclerotome and responsive NC progenitors (Schwarz et al. 2009).
Whereas some activities are likely to operate sequentially (Debby-Brafman et al. 1999; Krull 2001; Gammill et al. 2006; Kuriyama & Mayor 2008; Roffers-Agarwal & Gammill 2009; Schwarz et al. 2009), the fact that multiple signaling systems participate in channeling migration of NC progenitors through ventral pathways suggest cooperative interactions between them and hence the existence of a regulatory network that ensures proper segmental patterning. Notably, to date, no experimental data are available to support this notion.
It is worth mentioning in this context that additional interactions take place within individual segments that are accounted for by a rostrocaudal polarity of sclerotomal properties. Such are interactions between the caudal sclerotome and nascent myotome, shown to pattern the formation of avian pioneer myoblasts. Pioneer myoblasts generate the first myotomal fibers and act as a scaffold to pattern further myotome development. From their origin along the entire medial epithelial somite (Fig. 1, E2 green), they dissociate and migrate towards the rostral edge of each somite, from which differentiation proceeds in both rostral-to-caudal and medial-to-lateral directions. This directional behavior that occurs within individual segments is reminiscent of the polarity of NC migration. Grafting experiments showed that regulation of pioneer patterning is somite-intrinsic. Furthermore, pioneer myoblasts express Robo2, whereas the DM and caudal sclerotome express Slit1. Loss of Robo2 or of sclerotome-derived Slit1 (but not of DM-derived Slit1) function perturbed both directional cell migration and fiber formation, and their effects were mediated through RhoA (Halperin-Barlev & Kalcheim 2011). Hence, signals emanating from the caudal sclerotomal domain guide both neural as well as skeletal progenitors highlighting the significance of intrasomitic heterogeneity in patterning several aspects of the body plan that are functionally associated at later stages.
Melanoblast migration along the dermomyotome-derived dermis
As elaborated above, NC cells migrate along several discrete pathways within the trunk of developing embryos. In the chick, early migrating crest cells that generate neural derivatives are confined to ventral pathways medial to the DM and between somites while later cells that produce melanocytes migrate along a dorsolateral pathway underneath the ectoderm. This pathway becomes normally “available” upon dissociation of the epithelial DM that generates the dorsal dermis, a mesenchyme that serves as substrate for melanocyte migration (Erickson et al. 1992; Tosney 1992). Thus, also in the case of melanoblast migration, a close interaction between the NC progenitors and somite derivatives is apparent.
Slit ligands and Robo receptors are involved in different aspects of trunk NC migration. Slit2 is expressed in the DM and early migrating NC cells express Robo1 and Robo2. Furthermore, Slit2 was shown to repel NC migration in an in vitro assay, and misexpression of a dominant-negative Robo1 receptor induced a significant fraction of early NC cells to migrate ectopically along the dorso-lateral pathway normally invaded by melanocytes. These findings suggest that DM-derived Slit2 represses the entry of the NC into the dorsolateral pathway thus confining the migration of early NC cells to the ventral pathway (Fig. 3C) (Jia et al. 2005). Additional repulsive cues present in both the caudal sclerotome and also the dorsolateral pathway that are known to restrict the migration of neural precursors include the ephrins, F-spondin, chondroitin sulfate proteoglycans and PNA-binding molecules (see above and [Oakley & Tosney 1991; Oakley et al. 1994; Debby-Brafman et al. 1999; Santiago & Erickson 2002]). Some of these inhibitory molecules remain in the dorsolateral pathway by the time of melanoblast invasion of this domain and their migration in avian embryos seems nevertheless to be possible thanks to expression of positive chemotactic guidance molecules such as the Ednrb2 and EphB2 receptors (Harris et al. 2008). For example, Ednrb2 is upregulated in melanoblasts prior to entering the dorsolateral domain, and endothelin3 (ET3), its ligand, is expressed by cells of the ectoderm, DM and the mesenchymal dermis (Lecoin et al. 1998; Nataf et al. 1998; Nagy & Goldstein 2006; Krispin et al. 2010b). Harris and Erickson found that Ednrb2 and EphB2 drive melanoblast invasion of the dorsolateral pathway and suggested that signaling from these receptors is additive because the overexpression of one receptor can rescue the loss of the other (Harris et al. 2008). Of interest is that even if initiation of dorsolateral migration concurs with appearance of the dorsal dermis, misexpression of Ednrb2 in NC progenitors at earlier stages is sufficient for driving cell migration prematurely between ectoderm and epithelial DM (Krispin et al. 2010b). Possibly, high levels of the Ednrb2 receptor are dominant over inhibitory cues present in the superficial pathway at early stages (Fig. 3C). Thus, dermis-derived factors attract NC-derived melanoblasts that will later invade the skin to confer it with characteristic patterns of pigmentation.
NC- mesoderm interactions in the regulation of myogenesis
An extensive cross-talk between NC cells and the adjacent mesoderm was recognized at all levels of the neuraxis. In the head, NC cells are known to contribute to the formation of skeletal elements and connective tissue. Progressive interactions between NC and the cranial mesoderm shape craniofacial morphogenesis and aspects of mutual cell differentiation (Noden & Trainor 2005). NC cells are also a source of autonomous molecular information that generates morphology of the avian beak as grafted quail NC cells produced quail beaks in duck hosts and duck NC produced duck bills in quail hosts. These transformations involved regulation of gene expression in the adjacent host tissues as well as morphological changes to host beak tissues that included ectoderm and also mesoderm-derived muscles (Schneider & Helms 2003; Le Douarin et al. 2004). Cranial NC cells also populate the pharyngeal arches, which have at their center a mass of mesodermal cells. These mesoderm-derived cells are surrounded by the NC, externally covered by ectoderm and internally by endoderm (Le Douarin & Kalcheim 1999; Graham 2003). The mesodermal cells fuse together to form myofibers, that attach to specific NC-derived skeletal components in a highly coordinated manner (Noden 1983; Couly et al. 1992; Köntges & Lumsden 1996; Schilling & Kimmel 1997; Grammatopoulos et al. 2000; Cerny et al. 2004; Grenier et al. 2009). This arrangement has great significance for craniofacial development because disruptions in the interactions between mesoderm, NC and epithelium have profound effects on craniofacial development. A functional mouth, for example, depends upon the coordinated development of the NC-derived facial skeleton and its associated musculature, which is derived from mesoderm. Furthermore, signals emanating from NC cells instruct mesodermal progenitors to differentiate into myoblast precursors, and then to organize themselves around the developing skeletal elements. Using both mice and chick embryos, the Tzahor team showed that early myogenesis is independent of interactions with the cranial NC, yet the migration, patterning, proliferation and differentiation of muscle precursors are regulated by the crest progenitors (Rinon et al. 2007).
In the trunk, growing evidence documents that NC-somite interactions play a fundamental role in the development of somite derivatives. For instance, the dorsal NT that contains premigratory NC regulates specific aspects of DM development through BMP4 activity (Sela-Donenfeld & Kalcheim 2002) as well as subsequent formation of the dorsal dermis through Neurotrophin 3 (Brill et al. 1995). In addition, Wnt signaling patterns the medial DM (Spence et al. 1996; Capdevila et al. 1998; Ikeya & Takada 1998; Schmidt et al. 2000; Olivera-Martinez et al. 2001). For instance, Wnt1 from the dorsal NT acting through the beta-catenin-dependent pathway, controls expression of Wnt11 in the medial DM (Fig. 3D) (Marcelle et al. 1997), which in turn orients myocyte elongation. Wnt11 mediates this effect through the planar cell polarity pathway (Gros et al. 2009).
Once engaged in migration, NC cells were recently shown to provide pro-myogenic cues via activation of the Notch pathway in medial DM progenitors of the chick embryo. Upon interaction with en passant NC progenitors, cells of the medial DM lip that express Pax7 transiently activate Notch signaling as a consequence of which they translocate into the myotome where they become Myf5-positive and further differentiate into MyHC-expressing myocytes (Fig. 3E) (Rios et al. 2011). It would be important to clarify whether this event results from a local interaction between ventrally migrating NC cells that contact the DM lip or a more general feature of migrating NC cells that contact additional domains of the DM that are also endowed with myogenic potential. Moreover, possible interactions between transient Notch activation and Wnt11 in the medial DM lip remain to be examined in the context of myogenic induction and proper patterning, respectively.
Another recent study highlighted the significance of NC-derived Neuregulin1 that, acting primarily through the ErbB3 receptor, was shown to regulate murine muscle development by maintaining the Pax7 progenitor pool and preventing premature myogenic differentiation (Fig. 3E) (Van Ho et al. 2011). First, the authors specifically targeted a dominant-negative Pax3 construct to trunk NC cells, a procedure that resulted in a 30% reduction of ventrally migrating NC cells. Consequently, the number of Pax7-positive progenitors was reduced in both the central DM and the hypaxial muscle, with an early upregulation of specification genes but subsequent failure of terminal muscle differentiation. Hence, the lack of NC progenitors caused an exhaustion of the progenitor pool, finally leading to smaller muscles. Further to this initial observation, they characterized NC-derived Neuregulin1 as the factor responsible for maintaining the Pax7-positive progenitor pool at the expense of differentiated muscle. It would be interesting to compare whether Pax7-positive progenitors in the DM and those resident in muscle similarly respond to Neuregulin1 and/or to additional NC-derived signals (Kalcheim 2011).
The early heterogeneity already apparent in the dorsal NT and the DM imposes the need for searching local cues that operate within the epithelia themselves such as cell adhesion-dependent mechanisms, oriented cell divisions, morphogen gradients etc., and for local cues in the vicinity that may impact upon epithelial behavior. In addition, these epithelia are highly dynamic as different populations transit, for example, through the dorsal NT and the lateral DM while epithelial integrity is maintained. How is the knowledge for establishing successive waves of progenitors acquired remains a central issue of research.
The existence of intimate interactions between NC cells and the various mesoderm-derived populations has already been firmly established to provide a basis for correct patterning of the segmental body plan (both neural and skeletal components), proper territory colonization by pigment cells and craniofacial morphology; and the repertoire of underlying molecular mechanisms is steadily increasing. An important aspect of these interactions is that these transient tissues continuously feed-back on each other to affect distinct events in a sequential manner. Future challenges are to elucidate whether all NC cells are equivalent in their ability to influence somitic processes, or, alternatively, is NC-somite signaling compartmentalized in space and time to distinct NC subsets and, if so, to pinpoint the specific populations and the responsible genes. Reciprocally, since the timing of NC EMT, their migration and differentiation are influenced by distinct somite cell subsets, elucidating the responsible gene regulatory network is the next essential requirement.
We thank the members of our lab for critical reading of the manuscript and Raz Ben-Yair for assistance with figure preparation. This work was supported by grants from the Israel Science Foundation (ISF), Association Francaise contre les Myopathies (AFM), SFB 488 and DFG to CK.