The delamination of neural crest cells from the dorsal midline of the neural tube and their migration through neighboring structures represents a unique feature of the neuroepithelium, as CNS counterparts migrate and differentiate within the confines of the neural tube. At the same time, this process of epithelialmesenchymal conversion of premigratory neural crest cells, is a common feature during normal development of several embryonic structures and also underlies the formation of metastases during tumor progression (reviewed in Thiery, 2003). Hence, investigating delamination of crest progenitors represents a model system for understanding the underlying molecular basis of epitheliomesenchymal transitions and for evaluating how conserved this process is at various axial levels, across developmental systems and during tumor spreading.
A balance between BMP and its inhibitor noggin regulates neural crest delamination in the trunk
The onset of neural crest cell migration is a complex morphogenetic process which involves the coordinated action of several categories of molecules (cell adhesion molecules, cytoskeletal components, extracellular matrix macromolecules and transcription factors) upon which environmental signals act (Christiansen et al., 2000, Kalcheim, 2000, Nieto, 2001). The identity of these signals was recently elucidated. In initial studies, Delannet and Duband (1992) reported that added TGFβ1 and 2, stimulated emigration of cultured neural crest cells, possibly by increasing the adhesion of the cells to their substrate at the expense of intercellular interactions. The presence of these factors or that of related molecules in the dorsal neural tube was, however, not investigated, neither was their requirement under physiological conditions.
More recently, Sela-Donenfeld and Kalcheim (1999) reported that in avian embryos over a broad age range (15-40 somites), BMP4 mRNA is homogeneously distributed along the longitudinal extent of the dorsal neural tube, whereas its specific inhibitor noggin exists in the dorsal domain of the tube in a gradient of expression that decreases caudorostrally. This rostralward reduction in signal intensity was found to coincide with the onset of emigration of neural crest cells. Hence it was hypothesized that an interplay between noggin and BMP4 in the dorsal tube generates graded concentrations of the latter that in turn triggers the delamination of neural crest progenitors. Consistent with this suggestion, disruption of the gradient by grafting noggin-producing cells dorsal to the neural tube at levels opposite the segmental plate, inhibited emigration of HNK-1-positive crest cells without affecting expression of Slug, either at the mRNA or protein levels, suggesting that BMP4/noggin affect neural crest delamination independently of an earlier effect on cell specification. This notion was further substantiated by the finding that late delamination of crest cells was also inhibited upon noggin treatment of older neural primordia taken from levels opposite epithelial somites, where emigration of crest cells had already started. Hence, specification and delamination of crest cells as induced by BMP signaling are separable processes. In further agreement, Liem et al., (1997) have shown that the competence of neural epithelium to generate neural crest cells in response to BMP is transient. The switch in the ability of BMP4 to generate neural crest cells or to stimulate their delamination may coincide with the downregulation of BMP in the ectoderm and the beginning of its synthesis in the dorsal neural tube. Moreover, a role for BMP2 rather than BMP4 in crest emigration/migration from rhombencephalic areas of mouse embryos was proposed (Kanzler et al., 2000). Notably, a BMP-dependent mechanism for delamination of neural crest cells may not operate along the entire axis of avian embryos since at mesencephalic levels, the sole inhibition of BMP activity did not prevent crest cells from delaminating (Sela-Donenfeld and Kalcheim, Unpublished results).
As pointed out precedently, the activity of BMP4 along the rostrocaudal axis of the neural tube is modulated by changing levels of noggin (Sela-Donenfeld and Kalcheim, 1999). In order to understand the basis underlying regulation of neural crest delamination, it was necessary to clarify what signals help establishing the gradient of production of noggin along the neural tube. The temporal coordination between somite dissociation and the onset of neural crest migration, suggested that factors produced by the paraxial mesoderm might regulate the production of noggin in the dorsal neural tube. In line with this suggestion, experimental manipulations of the paraxial somitic mesoderm altered the pattern of noggin transcription in the dorsal neural tube; deletion of epithelial somites prevented normal downregulation of noggin. Furthermore, partial ablation of either the dorsal half or only the dorsomedial portion of epithelial somites was sufficient to maintain high noggin expression when compared to the normal downregulation of this gene in the control side. These data suggested that the dorsomedial region of dissociating somites produces an inhibitor of noggin transcription in the dorsal neural tube. Consistent with this notion, grafting dissociating somites in the place of the unsegmented mesoderm precociously downregulated expression of noggin and triggered premature emigration of neural crest progenitors from the caudal neural tube, an area never releasing mesenchymal cells under normal conditions. Thus, an inhibitory cross-talk exists between the paraxial mesoderm and the neural primordium that is mediated by regulating levels of noggin transcription. This interaction controls the timing of neural crest delamination to match the development of the somites into a suitable substrate for subsequent crest migration (Sela-Donenfeld and Kalcheim, 2000). Noteworthy, noggin activity is not restricted to the neural primordium. A dynamic expression of this inhibitory protein was also revealed in the somites, where changing levels of transcription were found to be modulated by BMP itself (Sela-Donenfeld and Kalcheim, 2002), suggesting the existence of a feedbak loop by which BMP controls synthesis of its own inhibitor which in turn modulates ligand activity.
Based on these functional data, it was important to assess whether BMP receptors were correspondingly expressed at the right sites and time. The responses to BMP family members are thought to be mediated by heterotetrameric complexes composed of type II receptors in concert with type I receptors of either class A or B, which transduce preferentially signaling by BMP2/4 or BMP7, respectively (see for example Massague and Chen, 2000). In the developing nervous system, BMP receptors of type IA were already visible in the dorsal neural folds at caudal levels of the neuraxis and later in the dorsal midline of the neural tube where premigratory crest cells reside (Sela-Donenfeld and Kalcheim, 2002). Notably, receptor mRNA signal was still detectable in the delaminating crest cells but was rapidly downregulated to undetectable levels in the migrating progenitors as they moved farther from the tube. At variance, expression of type IB receptors was largely restricted to the mesoderm. Taken together, the observed expression patterns are consistent with a possible role for BMP receptors of type IA, but presumably not for type IB, in mediating the effects of BMP4 on dorsal tube and crest development in avian embryos (Sela-Donenfeld and Kalcheim, 2002).
BMP-dependent genes and neural crest delamination
Genes such as Slug, FoxD3, PAX3, rhoB, Cad- 6, Msx1 and 2, Wnt 1 and 3a, etc, are either specifically expressed or become restricted to the dorsal tube from early stages onward, making it in some instances difficult to discriminate between possible roles in specification of the neural crest, subsequent delamination, or both. Experiments had to be designed to inhibit delamination without affecting initial specification of crest cells, hence BMP activity was abrogated following initial expression of these genes. The inhibition of neural crest emigration observed in vivo following noggin treatment was preceded by a partial or total reduction in the expression of cadherin 6B, rhoB, PAX3, Msx1,2 and Wnt1, but not that of Slug (see above, Sela-Donenfeld and Kalcheim, 1999, Burstyn-Cohen et al., 2004). Their local downregulation suggests these genes may be part of a molecular cascade triggered by BMP4, that leads to the separation of neural crest cells from the neural tube. This hypothesis requires that the effect of each factor be tested in experimental contexts in which it is possible to dissociate between specification and delamination events.
Wnt proteins play significant roles in neural crest cell development at different developmental times and in several species (reviewed in Wu et al., 2003). In avian embryos, Wnt 6 is synthesized in the epidermal ectoderm and might mediate crest specification (Garcia-Castro et al., 2002) although a direct link between the two is still lacking. Slightly later, Wnt1 and Wnt3a are present in the dorsal neural tube following initial specification of crest cells (Dickinson et al., 1995). Yet, Wnt3a is intense already opposite the segmental plate while Wnt1 becomes apparent slightly later, opposite epithelial somites and concomitant with BMP relief from noggin inhibition (Burstyn-Cohen et al., 2004). Whereas Wnt1 is likely to be directly regulated by BMPs (Marcelle et al., 1997, Sela-Donenfeld and Kalcheim, 2002, Burstyn-Cohen et al., 2004), the transcription of Wnt3a is not; suggesting that Wnt1 better fits to be a putative candidate in crest delamination (Burstyn-Cohen et al., 2004). Significantly, Wnt signaling through the transmembrane receptor Frizzled is required to modulate the distribution and function of β-catenin (Miller and Moon, 1997). β- catenin, as well as plakoglobin (γ-catenin) associate directly with the highly conserved cytoplasmic domain of cadherins. The so formed cadherin-catenin complex links to the actin filament network via actinin or vinculin (Ozawa et al., 1989; Hinck et al., 1994; Knudsen et al., 1995; Weiss et al., 1998). BMP/Wnt-mediated signals could induce changes in the actin cytoskeleton via rhoB and possible relations between rhoB and cadherin pathways remain to be clarified. A role for rhoB in crest delamination has been already suggested based on inhibition experiments in culture (Liu and Jessell, 1998). A molecular pathway for the activation of Rho by Wnt/frizzled was suggested, which involves the formation of a complex between Rho, dishevelled and Daam1 in the plasma membrane, resulting in the generation of a polarized cytoskeleton (Habas et al., 2001). Thus, the dynamic association of the catenincadherin complex and that of rhoB with the cytoskeleton may be essential for regulating cell-cell interactions leading to neural crest delamination. On the other hand, Ikeya et al., (1997) proposed that Wnt signaling might be required for the expansion of a pool of neural crest cells, a process that could also affect neural crest delamination (see next section). Notably, Rho GTPases could also be effectors of Wnt signals in this pathway as they were shown to affect morphogenesis by interfering with cell proliferation (Wei et al., 2002).
Pax-3 is expressed in both the dorsal neural tube and the adjacent somites (Goulding et al., 1991). The mouse mutation Splotch (Russell, 1947) represents a deletion in the gene coding for Pax-3 (Kessel and Gruss, 1990; Epstein et al., 1991). Splotch mutants are characterized by defects in neural tube closure and severe reduction or even absence of certain neural crest derivatives including pigment cells, sympathetic and spinal ganglia, enteric neurons and cardiac structures. These defects were suggested to result from a delay in the onset of neural crest emigration from the neural tube (Moase and Trasler, 1990). Another study found that crest cell emigration (or formation) was severely affected in the vagal and rostral thoracic areas, while virtually no cells emigrated from the tube more caudally, perhaps as a result of aberrant interactions among adjacent neural tube progenitors or between neural crest and somitic cells (Serbedzija and McMahon, 1997). A possible role for Pax-3 in mediating epithelial-mesenchymal interactions was suggested in other systems (Wiggan et al., 2002) as well as the possibility that Pax-3 triggers a non-canonical Wnt signaling cascade entailing JNK activation (Wiggan and Hamel, 2002).
The role of FoxD3 in formation of the neural crest was documented (see section III), yet its possible function in cell delamination remains unclear. Dottori et al., (2001) reported that forced expression of FoxD3 induced ectopic HNK-1 expression in the lateral part of the neuroepithelium and this event was followed by significant cell delamination. At variance, ectopically induced HNK-1-positive progenitors failed to reveal dispersive behavior according to Kos et al., (2001).
As previously discussed, BMP4 had no effect on the maintenance of Slug expression either at the mRNA or protein levels and yet inhibiting BMP prevented crest emigration in the trunk (Sela-Donenfeld and Kalcheim, 1999). This result would indicate that Slug activity is not sufficient for emergence of neural crest cells at least in the trunk region. In support of this notion, it was reported that neural crest cells still leave the neural primordium by stages 18-20 of development at trunk levels of the axis (Erickson et al., 1992, Reedy et al., 1998) a time when Slug is not transcribed any longer (Sela-Donenfeld and Kalcheim, 1999, Liu and Jessell, 1998). Furthermore, forced expression of Slug enhanced the production and migration of neural crest cells in the head but not in the trunk where Slug is exclusively expressed in the pre-migratory population (Del Barrio and Nieto, 2001). Altogether, these results suggest that Slug expression in the trunk neuroepithelium may be a hallmark of early forming neural crest but not be instrumental for subsequent cell delamination. A different situation holds for cranial areas, where Slug is expressed both in premigratory as well as in the migrating cells (Nieto al, 1994). At this level, Slug activity might affect the progression of crest migration, as shown in Xenopus embryos (Carl et al., 1999), but a direct effect of the Slug protein on delamination of cranial crest cells is still lacking in the chick as neither loss or gain of function experiments discriminated between specification versus epithelio-mesenchymal conversion (Nieto et al., 1994, Del Barrio and Nieto, 2001).
The above results highlight the existence of significant differences in the mechanisms leading to delamination of neural crest cells in cranial as compared to trunk levels of the axis. These are exemplified not only by differences in Slug function and in the duration and intensity of cell delamination between the two areas, but also in differential regional expression of other relevant genes, such as noggin (our unpublished results), AP2 (Schorle et al., 1996, Zhang et al., 1996), etc, in the hierarchical relationship between BMP and Wnt signals in the two areas (Ellies et al., 2000, Marcelle et al., 1997) and in the cell cycle characteristics of delaminating cells (see next section). Interestingly, differences in the role of specific factors in cranial as compared to adjacent vagal levels of the axis were also documented. For instance, deleting the–zfhx1b gene, a zinc finger and homeodomaincontaining transcription factor that encodes Smad-interacting protein-1, caused arrest of delamination of cranial neural crest cells without impairing their specification and yet resulted in a failure of the actual formation of vagal-level progenitors. The latter led to a phenotype partially resembling the aganglionic megacolon syndome observed in humans carrying a mutation in this gene (Van de Putte et al., 2003). Hence, the upstream trigger/s of the massive and rapid delamination of crest cells in the head remain to be elucidated and the intracellular mechanisms remain to be worked out in all regions.
The role of the cell cycle in neural crest delamination
Neural crest cells are mitotically active progenitors while residing in the dorsal neural tube and throughout migration. This initially discrete population must expand to reach the final number of cells that populates peripheral ganglia and other derivatives. The first post-mitotic cells appear by the time of gangliogenesis (Kahane and Kalcheim, 1998). Prior to emigration, prospective neural crest progenitors are an integral part of the neuroepithelium and, as such, they undergo interkinetic nuclear migration whereby the position of the cell soma with its nucleus changes upon the phase of the cell cycle (Martin and Langman, 1965, Langman et al., 1966 and refs. therein). Moreover, they reveal similar cell cycle characteristics to laterally located progenitors with a mean generation time of about 8 hours in avian embryos (Langman et al., 1966, Smith and Schoenwolf, 1987,1988 and refs. therein). Nevertheless, the dorsal area of the neural tube becomes highly distinct from the remaining neuroepithelium when noggin is downregulated and consequently, BMP becomes activated at high levels thus triggering crest cells to delaminate. The question was then asked whether neural crest cells randomly emigrate at any phase of the cycle or alternatively, whether there is a preferred phase for delamination. Trunk-level avian neural crest cells were found to emigrate synchronously in the S-phase of the cell cycle (Burstyn-Cohen and Kalcheim, 2002), in support of the latter possibility. The functional significance of the observed synchronous delamination was examined by inhibiting G1-S transition with olomoucine, AG555 or mimosine. All treatments prevented initial delamination of neural crest cells that could be rescued upon drug removal. In contrast, aphidicolin or VM-26, which inhibit the cycle at S and G2 phases, respectively, had no effect. Furthermore, in ovo overexpression of the 15 amino-acid domain of MyoD, which specifically binds to cdk-4/6 and thus prevents G1-S transition, inhibited both Brdu incorporation and NC delamination, but affected neither specification nor survival of the neural progenitors. Likewise, overexpression of the cyclin-dependent kinase (cdk) inhibitor p27 and of a dominant-negative form of the retinoblastoma- binding E2F-1 transcription factor, prevented both entry into S phase as well as neural crest delamination. These results showed for the first time that the transition between G1 to S is a necessary event for the epithelial-to-mesenchymal conversion of premigratory neural crest cells (Burstyn-Cohen and Kalcheim, 2002).
Roles of the cell cycle in morphogenesis and in the generation of cell movement movement
Previous studies have already highlighted possible links between specific phases of the cell cycle and generation of cell movement. Short-range interactions between the ganglionic eminence and neocortical epithelium influence interkinetic nuclear migration and cell exit from the primary epithelium (Miyama et al., 2001). Studies on the mechanisms of cell division in Drosophila provided a paradigm for understanding how information that controls stereotypic mitoses is translated into cell movement (Follette and O’Farrell, 1997). Fibronectin substrates induce shortening of the G1 period in migratory neural crest cells (Paglini and Rovasio, 1999). A role for the cell cycle in patterning and morphogenesis during neural development has also been suggested. For instance, lengthening of specific phases of the cycle was found to be associated with bending of the neural plate (Smith and Schoenwolf, 1987, 1988). The laminar fate of cortical neurons was shown to be determined during the S or G2-phases of the final cell cycle (McConnell and Kaznowski, 1991, Ohnuma et al., 2001). The control of cell proliferation is also crucial for the establishment of the correct number of daughter cells and could influence cell fate. The choice between cell cycle progression or exit followed by differentiation is influenced by extrinsic signals operating during G1 (Elshamy et al., 1998). At this point, mitogens stimulate activation and synthesis of pro-cell cycle proteins of the D- and E-type cyclins and their partners, cdk4/6 and cdk2, respectively, the key regulators of the G1 restriction point and the G1-to-S phase transition. In contrast, differentiation signals upregulate cell cycle inhibitory proteins such as p21, p27 or p57 (Ohnuma et al., 2001, Zhang, 1999). The pigmented epithelium modifies the plane of cell division in adjacent retinal progenitors, an event with possible significance in determining cell fate (Cayouette et al., 2001). Regulation of the cell cycle is also intimately linked to cell death. In neuronal cells, apoptosis caused by deprivation of trophic support can be prevented by agents that block G1/S transition (Farinelli and Greene, 1996, Park et al., 1997).
Roles of cell cycle genes in morphogenetic processes
Cell cycle progression is regulated by cdk’s that are activated upon interaction and binding to cyclins and inhibited by cdk inhibitors. Cdk’s regulate diverse biochemichal pathways while integrating extracellular and intracellular signals, the nature of which can be either mitogenic or growth-inhibitory. Integration of these signals is interpreted by means of cell cycle transitions. The G1/S transition is governed by cdk’s coupled to Cyclin D, A and E; while cyclin B-associated cdk’s regulate transition between G2 and M phases. Two families of cdk inhibitors negatively regulate the cell cycle: the INK4 family (p15INK4B, p16INK4A, p18 and p19) bind to and inhibit cdk4/6 and the CIP/KIP family of proteins (p21CIP1, p27Kip1 and p57Kip2) which binds primarily to cyclin E- and cyclin A-bound cdk2 and to cyclin D-bound cdk4/6 with lower affinity (reviewed in Singerland and Pagano, 2000; Coqueret, 2002).
The key regulators of the cell cycle, cdk’s and cyclins, are now being “rediscovered” with novel roles that are independent of their classical functions in controlling the cell cycle. Such functions include centrosome formation and DNA replication by cyclin E (reviewed in Winey, 1999), transcriptional control of various genes by cyclin D (reviewed in Coqueret, 2002), muscle and neuronal differentiation as well as acquisition of cell motility by p27 (McAllister et al., 2003; Vernon and Philpott, 2003; Vernon et al., 2003), etc. This association between specific cell cycle genes and cellular functions might reflect the need of the cell to coordinate important events in a timely rather than premature fashion, to avoid malformations or even lethality. Several novel roles of cell cycle proteins will be briefly discussed below, together with their possible relevance to neural crest migration.
The D-type cyclins are the first cyclins to be activated during the G1 phase. Following mitogen stimulation, cyclins of type D bind cdk4 or cdk6 and activate their kinase activity to phosphorylate target proteins, including pRB. Hyperphosphorylation of pRB disrupts its interaction with histone deacetylase and histone methylase, facilitating chromatin accessibility to transcription. Phosphorylated pRb also releases the bound transcription factor E2F-1 from repression, thus enabling E2F-1– dependent transcription of genes that are required for transition into S phase, including that of cyclin E (Coqueret, 2002). Unlike the transcription of cyclins E, A and B, that of D-type cyclins, as well their accumulation at a protein level and their cellular localization are largely dependent on extracellular signals such as mitogens and nutrient stimulation (Matsushime et al., 1991). This places D-type cyclins as a putative link between growth inducers and the cell cycle machinery. In addition to this central role, Cyclin D has recently been suggested to control transcription of DNA-binding proteins, that in turn regulate specific target genes. This transcriptional control of cyclin D is probably independent of its cell cycle role as it does not involve cdk4 activation. Furthermore, cyclin D1 is also able to affect the differentiation state of myoblasts through inactivation of MyoD transcription and restricts premature differentiation of intestinal epithelial cells through inhibition of the specific Beta2/NeuroD transcription factor (reviewed in Coqueret, 2002).
In light of the precedent findings, it is tempting to speculate that cyclin D1 might also play a role in the ontogeny of neural crest cells. Cyclin D1 transcription is weak in the neural tube opposite segmental plate levels and becomes gradually prominent at axial levels corresponding to neural crest emigration (Burstyn-Cohen et al., 2004). Along this line, Wnt-dependent transcription of Cyclin D1 in the dorsal tube (Megason and McMahon, 2002) mediates delamination of crest cells by affecting transcription of genes involved in cell adhesion and in the generation of cell movement (Burstyn-Cohen et al., 2004). Furthermore, the central role of cyclin D1 in the cell cycle, together with its enrichment in the dorsal neural tube, suggests it might be involved in maintaining the balance between neural crest proliferation (G1/S transition) and delamination. Cyclin D1 could affect the continuous recruitment of progenitors to the midline thus ensuring the dorsal tube is not depleted from cells due to extensive crest delamination.
As mentioned above, p27 is a member of the Cip/Kip family of cell cycle inhibitors that negatively regulates cell cycle progression at G1 (Singerland and Pagano, 2000). Recent work by McAllister et al. (2003) suggests a novel cell scattering activity for p27 which is mediated by a motility domain localized to the C-terminus of the molecule. It has been proposed that following HGF/SF stimulation of human hepatocellular carcinoma cells, mediated by the Met receptor, p27 is phosphorylated and exported from the nucleus to the cytoplasm, where it binds F-actin and modulates cytoskeletal rearrangements leading to cell migration. In line with this is the fact that p27-deficient primary fibroblasts failed to migrate, a motility defect that was rescued by introducing into the cells either wild-type p27 or the C-terminal motility domain. Notably, this activity of p27 was reported to be independent of its function in cell cycle inhibition, as the cdk-cyclin binding domain resides in its Nterminus. This scattering ability of p27 may be context-dependent, as mimosine, a drug that stimulates induction of p27 and its translocation into the nuclei of neural crest cells, prevented the emigration of crest progenitors from neural primordia (Burstyn- Cohen and Kalcheim, 2002).
Possible interactions between the BMP cascade and cell cycle genes in controlling neural crest delamination
Having shown that the epithelial to mesenchymal conversion of premigratory neural crest is triggered by a local balance between BMP4 and its inhibitor noggin and that these neural tubespecific events are temporally modulated by an inhibitor of noggin transcription produced in the dorsomedial somites (Sela-Donenfeld and Kalcheim, 1999, 2000), an essential question was what is the relationship between the above environmental signals and cellintrinsic mechanisms such as the requirement for G1/S transition for cell delamination (Burstyn-Cohen and Kalcheim, 2002). One possible link between the two is that BMP4 induces a cascade of secondary signals that influence G1/S transition via activation of cyclin-cdk complexes in dorsal tube progenitors, a process which is in turn translated into parameters of the delamination machinery. In favor of such a content, Panchinsion et al., (2001) have demonstrated that BMP receptors of type 1A, but not 1B, transduce a mitogenic signal in mouse neuroepithelial cells. Notably, the dorsal midline of the avian tube also expresses type 1A receptors at a comparable time in avians (Sela-Donenfeld and Kalcheim, 2002). In addition, Msx1, a downstream transcription factor induced by BMP in the dorsal neural tube upregulates cyclin D1 and cdk4 activity (Hu et al., 2003). Dorsal neural tube-derived BMP4 also stimulates transcription of Wnt1 (Marcelle et al., 1997, Sela- Donenfeld and Kalcheim, 2002) and the Wnt1-dependent β- catenin/LEF-1 pathway regulates transcription of cyclin D1 and cell proliferation in a variety of cells (Kioussi et al., 2002; Shtutman et al., 1999; Tetsu and McCormick, 1999) including the avian neuroepithelium (Kubo et al., 2003; Megason and McMahon, 2002 but see Hari et al., 2002). Hence, it is likely that the roles played by environmental signaling such as BMP/noggin and by cell autonomous events such as G1/S transition in delamination of neural crest cells, are part of a single pathway which operates through an intermediate stage that requires Wnt activity. Recent evidence is accumulating in support of this view. Recently, BMP was found to regulate G1/S transition via the canonical pathway of Wnt signaling and inhibition of the latter prevented emigration of neural crest progenitors while downregulating cyclin D1 (Burstyn- Cohen et al., 2004). Whereas the molecular backbone leading to neural crest delamination at trunk levels begins to be clarified, the downstream signals that translate cell cycle parameters into the generation of crest cell movement remain to be investigated.
Research in our laboratory is supported by grants from the Israel Science Foundation, the Israel Cancer Research Foundation (ICRF), the March of Dimes Birth Defects Foundation and the Deutcheforschungsgemeinschaft (SFB 488) to C. K. T.B-C is partially supported by a fellowship from the Lady Davis Foundation.
Note added in proof
The present review covers studies published until 2003. Many exciting results appeared during 2004 and subsequently, but only few could be added during the final stages of editing this volume. Our apologies to those authors whose work could not be properly addressed.