Neural crest is a pluripotent population of cells born in the dorsal aspect of the neural tube that migrates into the periphery to give rise to a wide range of derivatives, including melanocytes, neurons, glial cells of the peripheral nervous system, and skeletal components of the head and neck (Le Douarin and Kalcheim, 1999; Graham, 2003). Within the hindbrain region of the chick and other vertebrates, migration follows the segmental organization of the neuroepithelium into morphologically distinct compartments, called rhombomeres (r), with a restriction into three streams that are separated by crest-free zones (Lumsden et al., 1991; Serbedzija et al., 1992). This stereotypic migration pattern through the mesenchyme is thought to play a key role in the patterning of skeletal derivatives of the head (Graham et al., 1996); however, it also sets critical signals for the coordinated development and integration of the cranial sensory ganglia (Begbie and Graham, 2001). Early migrating neural crest cells move ventrally into specific branchial arches where they differentiate predominantly into cartilage and bone. In contrast, later migrating neural crest cell populations remain dorsally and give rise to neurons and glial cells of the cranial sensory ganglia (Tosney, 1982; Baker et al., 1997). The cranial sensory ganglia also originate from a second spatially discrete migratory cell population; the neurons formed in the epibranchial and trigeminal placodes (D'Amico-Martel and Noden, 1983; Noden, 1993; Begbie and Graham, 2001). It is the late-migrating neuroglial crest cells that recently have been shown to function as guidance tracks for epibranchial placodal neurons emigrating inward from their placodal origin (Begbie and Graham, 2001; Barlow, 2002). Thus, the establishment of organized neural crest cell streams in proximity to the neural tube is of fundamental importance for the positioning and integration of the cranial sensory ganglia.
The formation of distinct neural crest cell streams within the hindbrain region is thought to be due, in part, to an initial phase of apoptotic cell death as the neural crest cells emerge from the dorsal midline of the neural tube (Graham et al., 1993, 1994; Smith and Graham, 2001; Ellies et al., 2002). However, this is unlikely to be the only mechanism responsible for the establishment and maintenance of the mesenchymal crest-free zones around rhombomere (r) 3 and r5. It has been shown that neural crest cells actively avoid invading r3 and r5 mesenchyme pointing toward the existence of inhibitory guidance mechanisms present within these areas (Sechrist et al., 1993; Kulesa and Fraser, 1998, 2000; Farlie et al., 1999; Kulesa et al., 2000). In addition, time-lapse studies suggest that guidance signals for migratory neural crest cells are found within the immediate environment surrounding the neural tube, with neural crest cells that delaminate in r3 and r5, moving rostrally and caudally along the neural tube to join the streams emerging from r2, r4, and r6. Several studies indicate that the segmented neuroepithelium of the hindbrain regulates aspects of the neural crest cell migratory behavior. For example, grafting studies show that r3 placed into r6 position induces a small crest-free zone adjacent to the grafted tissue (Kuratani and Eichele, 1993), and similarly, a respecification of rhombomere identity by the ectopic expression of Hoxa-1 or retinoic acid also results in changes to neural crest cell migration patterns (Lee et al., 1995; Alexandre et al., 1996). The loss of the Neuregulin receptor ErbB4 expressed by the neural tube at the level of r3 and r5 has been shown to induce inappropriate neural crest cell invasion into normally crest-free territory adjacent to r3. Whereas neural crest cells from these mutant mice maintain their intrinsic capacity to respond to environmental cues in wild-type animals, the inappropriate neural crest cell invasion indicates the loss of a repulsive, ErbB4 signaling-dependent factor emitted by r3 (Christiansen et al., 2000; Golding et al., 2000).
Several inhibitory molecules contribute to the repulsive guidance of neural crest cells in the trunk (Le Douarin and Kalcheim, 1999; Krull, 2001); however, the putative guidance signals emitted by specific rhombomeres that establish the neural crest cell streams within the hindbrain region remain elusive. Although Ephrin/Eph signaling prevents intermingling of neural crest cell streams in Xenopus (Smith et al., 1997), there is, at present, no evidence for this interaction in the establishment of the characteristic crest-free zones in mouse or chick (Adams et al., 2001).
We have shown previously that a secreted member of the Semaphorin family of axon guidance molecules, Sema 3A is expressed within the chick hindbrain (Eickholt et al., 1999). In this study, we examine the putative function of Sema 3A to confine neural crest cell streams in the developing hindbrain region and extend the study to a second Class III Semaphorin, Sema 3F. Our results suggest that Class III Semaphorin/Neuropilin-dependent signaling plays a key role in the guidance and positioning of neural crest cells in proximity to the neuroepithelium within the hindbrain.
Expression of Sema 3A and Sema 3F Within the Neuroepithelium Correlates With the Inhibitory Environment for Neural Crest Cell Migration Within the Neighboring Mesenchyme
We have shown previously that Sema 3A, a member of the secreted Class III Semaphorins, is present in odd-numbered rhombomeres within the hindbrain. In addition, neural crest cells express the Sema 3A binding receptor Neuropilin-1 (NP-1) and respond to Sema 3A in vitro (Eickholt et al., 1999). In this study, we describe the expression of another secreted Semaphorin, Sema 3F, within the hindbrain and investigate temporal and spatial changes in the expression of both ligands.
As described previously, Sema 3A is first detectable from stage 10 (Shepherd et al., 1996), and by stage 11, Sema 3A is expressed in r1, r3, and r5, with highest levels within r5 (Fig. 1A; Shepherd et al., 1996; Eickholt et al., 1999). Sema 3F expression within the hindbrain is first detected at stage 9 (eight somites, Fig. 1E) at the level of r3 with very low levels of expression seen additionally in r1 and r5. By stage 11, Sema 3F is expressed throughout the dorsal–ventral aspect of the neural tube at the levels of r1, r3, and r5, correlating with crest-free zones adjacent to the neural tube at the levels of r3 and r5 (Fig. 1F). At stage 13, the expression of both Semaphorins remains highly expressed at the levels of r1, r3, and r5 (Fig. 1B,G). By stage 15, we observe that Sema 3A and Sema 3F are expressed continuously throughout the basal plate of the neural tube, whereas expression is maintained dorsal–ventrally in the odd-numbered rhombomeres (Fig. 1C,H). As development continues, this pattern of expression further refines such that, by stage 17, Sema 3A expression is absent dorsally at the level of r3 (Fig. 1D), whereas Sema 3F is absent dorsally at the level of r5 (Fig. 1I). Class III Semaphorins bind to Neuropilin receptors that associate with members of the Plexin-A family, to form a functional signaling complex (Takahashi et al., 1999; Tamagnone et al., 1999). Neuropilin-1 is present in cranial neural crest cells (Eickholt et al., 1999; Gammill and Bronner-Fraser, 2002; Lee et al., 2002) and in situ hybridizations performed with a Neuropilin-2–specific probe highlights expression in neural crest cell streams emerging from the hindbrain at stages 11 and 13 (Figs. 2A and B, respectively) and invading the branchial arches at stage 15 (Fig. 2C). We also analyzed the expression of a Neuropilin signaling coreceptor, Plexin-A1, and observe it to be restricted at stage 11 to neural crest cells migrating at the level of r4 (Fig. 2D, arrow). In conjunction with the fact that the secreted Class III Semaphorins function predominantly as repulsive guidance molecules, the expression patterns suggest that Sema 3A and Sema 3F signaling may create nonpermissive environments for neural crest cell migration in proximity to the neural tube.
Exposure to Sema 3A Affects Migratory Neural Crest Cells in the Developing Hindbrain Region
Neural crest cells that do not invade the branchial arches but remain dorsally generate the neurons and glial cells of the cranial sensory ganglia (Tosney, 1982; Baker et al., 1997). At stage 16, this neural crest cell population remains lateral to r2, r4, and r6 and can be visualized by expression of the HMG transcription factor Sox-10 (Southard-Smith et al., 1998; Britsch et al., 2001; Stolt et al., 2002; Fig. 3A) and the Retinoid X Receptor Gamma (RXRγ, Fig. 3B; Rowe et al., 1991; Rowe and Brickell, 1995). Similarly, the cell adhesion molecule Cadherin-7 (Cad-7) labels the neural crest cell population that closely associates with the surface of the neural tube at the level of r2, r4, and r6, delineating the cranial nerve entry/exit points (Fig. 3C; Nakagawa and Takeichi, 1995; Niederlander and Lumsden, 1996). To examine whether the presence of exogenous Sema 3A protein within the neural tube could affect neural crest cell migration within the adjacent mesenchyme, we implanted beads soaked with excess purified Sema 3A protein into the hindbrain at stage 10 and incubated to stage 15. In the presence of Sema 3A–soaked beads, we failed to observed Cad-7 expression (Fig. 3D,E, n = 7/7), whereas implantation of control beads (soaked in Ringer's solution) showed normal Cad-7 distribution (n = 7/7), suggesting that Sema 3A protein, applied in the neural tube, may indeed affect neural crest cells in proximity to the neural tube. Thus, we used a second strategy to examine the effect of Sema 3A on Cad-7–positive neural crest cells and coelectroporated a Sema 3A expression construct together with a green fluorescent protein (GFP) expression vector. We positioned the electrodes such that expression of the constructs was established within the hindbrain neuroepithelium only (Fig. 4A) and subsequent Sema 3A in situ hybridizations confirmed the overexpression of the Sema 3A transcript unilaterally within the neural tube (Fig. 4C). Analysis of Sema 3A/GFP electroporated embryos by Cad-7 in situ hybridization revealed a reduction in Cad-7–positive neural crest cells at the presumptive entry/exit points on the electroporated site of the neural tube (Fig. 4D,D′; n = 6/10). In comparison, Cad-7 expression was unchanged within control GFP electroporated embryos (Fig. 4B,B′; n = 11/11). In summary, these results suggest that Sema 3A, applied or misexpressed within the neural tube, can affect Cad-7–positive neural crest cells residing in close proximity to the neural tube.
Ectopic Expression of Sema 3A or Sema 3F Within the Hindbrain Neural Tube Affects Neuroglial Crest Cells Within the Neighboring Mesenchyme
Sema 3A signaling requires the interaction with the Neuropilin-1 receptor, whereas the biological activity of Sema 3F depends on binding to Neuropilin-2. Given that Neuropilin-2 is expressed by cranial neural crest cells, we wished to test whether both ligands, Sema 3A and Sema 3F, could affect neural crest cells and whether neural crest cells use Semaphorins more “generally” as guidance signals by labeling neural crest cells of treated embryos with HNK-1 antibody. A schematic of migratory cranial neural crest cells is presented in Figure 5A, and we find that expression of Sema 3A/GFP or Sema 3F/GFP resulted in the disruption or loss of HNK-1–positive neural crest cell streams (Fig. 5C,D, n = 9/11), when compared with control GFP electroporated embryos (Fig. 5B, asterisk, n = 0/6). To elucidate whether the loss of neural crest cell streams reflects changes in the production and/or the initial migration of neural crest cells, we 1,1′, di-octadecyl-3,3,3′,3′,-tetramethylindo-carbocyanine perchlorate (DiI) -labeled neural crest cells at stage 9–10 and subsequently electroporated as previously described. Electroporation of Sema 3A or Sema 3F resulted in the loss of DiI- labeled crest cells on the electroporated side (Fig. 5F, n = 18/25) when compared with the control side of the same embryo, or embryos electroporated with GFP alone (Fig. 5E, n = 0/12). Confocal images of transverse sections taken through a Sema 3A/GFP electroporated embryo at the level of r6 clearly demonstrates the absence of DiI-labeled neural crest cells migrating on the electroporated side of the embryos in comparison to the control side of the same embryo. However, Sema 3A/GFP expression does not prevent the production of neural crest cells at the dorsal midline of the neural tube (Fig. 5G, arrows).
We also performed electroporations as described above and incubated embryos for 24 hr, followed by the analysis of RXRγ and Sox-10 expression. Unilateral electroporation of Sema 3A and Sema 3F into r2–r6 (as assessed by GFP expression before in situ hybridization) results in the depletion of RXRγ-expressing neural crest cell streams clearly present in the control side of the same embryo (Fig. 6B′,C′, n = 10/10). More localized electroporation into r2/3 diminishes the neuroglial crest cell contribution to the trigeminal ganglion without affecting post-otic neural crest cells, as visualized here by Sox-10 expression (Fig. 6D′,E′, n = 11/15). Introduction of the GFP expression vector alone did not affect this neural crest cell population as detected with RXRγ or Sox-10 (Fig. 6A′, n = 8/8). These findings suggest that ectopic expression of Class III Semaphorins within the hindbrain neural tube can affect the neuroglial crest cell population residing in the adjacent mesenchyme.
Consequences of Neuroglial Crest Cell Disruption
The cranial sensory ganglia of the vertebrate head are derived from two diverse migratory populations of cells that originate in the neural crest and the neurogenic placodes (D'Amico-Martel and Noden, 1983). Recently, it has been shown that the patterning of neural crest cell migration within the hindbrain region is essential to direct the epibranchial neuronal cells inward to establish their central connections (Begbie and Graham, 2001). In this previous study, the removal of neural crest cells led to the impeded inward migration of the epibranchial neuronal cells, which remained in their subectodermal position. To test whether the loss of the neuroglial crest cells after Sema 3A, or Sema 3F overexpression within the hindbrain neural tube (Fig. 7A) results in a failure to mediate these known guidance properties of neural crest cells, we electroporated as above and incubated for 36 hr before analyzing by staining for neurofilament protein to visualize the migratory placodal cells (Fig. 7B).
Ectopic expression of Sema 3A or Sema 3F in the neural tube at the level of r1–r3 resulted in two distinct phenotypes in comparison to the control side of the embryos and to untreated control embryos. First, within the trigeminal ganglion the ophthalmic lobe appears disorganized and less tightly packed than in controls (Fig. 7C′,D′, n = 6/7). The second phenotype observed was either the failure of (Fig. 7C′, arrowhead) or reduced inward movement of the geniculate placodal neurons (Fig. 7D′, asterisk). This finding contrasts with the control sides of the treated embryos, where the characteristic stream of neuronal migration is clearly visible (Fig. 7C,D). We also expressed Sema 3A or Sema 3F in post-otic regions of the neural tube (r5–8), which resulted in a diminished or absent inward migration of neuronal placodal cells of the petrosal and nodose placodes (Fig. 7E′,F′, arrowheads, n = 5/6). These observations indicate that the loss (or reduction) of the neuroglial crest cells after Sema 3A and Sema 3F treatment results secondarily in the failed inward movement of placodal neuronal cells and, thus, reinforces the concept that neural crest cells are fundamental for the integration of the cranial sensory ganglia.
Ectopic Expression of Soluble Neuropilin-1/Neuropilin-2 Affects Neuroglial Crest Cell Migration
If Semaphorin-dependent signaling contributes to the maintenance of neural crest cell streams within the hindbrain, we would expect that interfering with Semaphorin signaling results in the aberrant entry of neural crest cells into the normally crest-free mesenchyme. We tested this hypothesis by electroporating constructs expressing soluble Fc chimeras consisting of the ectodomains of Neuropilin-1 or Neuropilin-2 at 10–12 somites (stages 10/11) and performed Sox-10 in situ hybridization. After these treatments, large numbers of Sox-10– positive neural crest cells are clearly found inappropriately positioned in r3 mesenchyme (Fig. 8B′,C′, n = 17/24) when compared with control GFP electroporated embryos (Fig. 8A′, n = 0/6). Serial sections taken through Neuropilin-Fc electroporated embryos confirms the presence of Sox-10–positive cells adjacent to r3 on the electroporated side (Fig. 9B,C, asterisks), whereas the crest-free zone on the control untreated side of the same embryo and within GFP electroporated embryos is preserved (Fig. 9A). This result indicates that the loss of Class III Semaphorin signaling enables neural crest cells to invade areas normally inhibitory to their migration.
The stereotypic migration pattern of neural crest cells through the mesenchyme of the hindbrain region is thought to play a key role in patterning the skeletal derivatives of the head (Graham et al., 1996), in addition to establishing critical signals for the coordinated development and integration of the cranial sensory ganglia (Begbie and Graham, 2001). Studies in chick and mouse have indicated that one of the dominant signaling events in the establishment of the three characteristic neural crest cell streams is derived from the segmented organization of the hindbrain neuroepithelium (Golding et al., 2000, 2002; Trainor et al., 2002).
Here, we show that two Class III Semaphorins, Sema 3A and Sema 3F, are expressed in the chick by the odd-numbered rhombomeres, at developmental stages that coincide precisely with the movement of neural crest cells though the mesenchyme flanking even-numbered rhombomeres. Both Semaphorins are known to mediate their biological activity through their binding receptor Neuropilin-1 and Neuropilin-2 that are expressed by cranial neural crest cells (Eickholt et al., 1999; Gammill and Bronner-Fraser, 2002; and this study). Thus, we demonstrate an involvement of these molecules, acting as “chemorepellents,” in the definition of cranial neural crest cell streams during chick hindbrain development. We find that attenuation of Semaphorin function by in ovo expression of soluble Neuropilin-Fc resulted in neural crest cells invading adjacent mesenchymal territories normally devoid of neural crest cells. Thus, the Class III Semaphorins may be one of the long hypothesized signals expressed and released by the hindbrain neuroepithelium to contribute to the establishment of neural crest cell streams. In the mouse, crest-free zones are also established by interactions between the neuroepithelium and the mesenchymal environment (Trainor et al., 2002). However, analysis of Neuropilin-1 or Neuropilin-2 mutant mice failed to reveal obvious phenotypes indicative of defects in cranial neural crest cell migration (Kitsukawa et al., 1997; Giger et al., 2000; Gu et al., 2003), and this might reflect the level of redundancy within this system. At present, it is not known whether Semaphorin expression in the early developing mouse hindbrain follows the segmental organization of the neural tube, nevertheless, several lines of evidence support a function for Neuropilin-dependent signaling in different neural crest cell populations. First, Sema 3A/Neuropilin-1 signaling has been shown to regulate sympathetic chain assembly by regulating the arrest and aggregation of their neural crest precursor cells in mice (Kawasaki et al., 2002). Second, Mouse Sema 3A (previously M-SemD) is expressed within the posterior half somite (Adams et al., 1996), suggesting that it might contribute to the segmental migration of trunk neural crest cells. Moreover, targeted disruption of Sema 3C in mice leads to a phenotype similar to that reported after ablation of the cardiac neural crest cells in chick embryos (Feiner et al., 2001).
Neural crest cell invasion into inappropriate hindbrain mesenchyme has also been described in mice carrying mutations in the Neuregulin receptor ErbB4 and the basic helix-loop-helix transcription factor Twist (Golding et al., 2000; Soo et al., 2002). ErbB4 signaling in r3 and r5 is thought to provide positional information that governs the patterning of neural crest cell migration through the paraxial mesenchyme and it would feasible that altered ErbB4 function in the mutant mice affects expression of Class III Semaphorins. Similarly, Twist signaling might be required for appropriate Neuropilin function and in this context, it is intriguing that neural crest cells in these mutant mice also invade abnormally more ventral domains of the paraxial mesenchyme (Soo et al., 2002), a phenotype that we observed with lower penetrance in our Neuropilin-Fc electroporated embryos (Fig. 9B, arrow).
We have shown previously that neural crest cells respond to Sema 3A in vitro by reducing their cell area and in the stripe assay by avoiding immobilized Sema 3A stripes (Eickholt et al., 1999). Here, we demonstrate that the overexpression of Sema 3A or Sema 3F in vivo leads to a disruption of normally tightly packed neural crest cell streams. This altered migratory behavior may be a result of several molecular mechanisms that have been linked to Semaphorin function in other cell types. Both Sema 3A and Sema 3F have been shown to induce the growth cone and cell collapse responses, processes that may in turn lead to a general inhibition of motility in areas of high ligand concentration. In addition, an attractive parallel can be drawn with the effect of Sema 3A on endothelial cell migration. In this cell type, it was recently shown that Class III Semaphorins can antagonize integrin function (Serini et al., 2003). Because integrin-mediated adhesion plays a central role in neural crest cell migration (Perris and Perissinotto, 2000), our results are consistent with a model whereby Class III Semaphorins, expressed and secreted by odd-numbered rhombomeres, may inhibit neural crest cell migration into inappropriate regions through the modulation of integrin function. Sema 3A has also been proposed as a selective neuronal cell death factor, with prolonged exposure shown to induce apoptosis in a neuroectodermal progenitor cell line (Gagliardini and Fankhauser, 1999; Shirvan et al., 1999, 2000; Bagnard et al., 2001). It is possible, therefore, that Semaphorins may also function to eliminate neural crest cells that persist in inappropriate regions. Indeed, at later stages of development, we find that after the overexpression of both Sema 3A and Sema 3F dorsally located neuroglial crest cells are decreased or completely absent. Further work will elucidate the molecular mechanisms by which these affects are mediated.
In summary, we provide functional evidence that Class III Semaphorin/Neuropilin signaling contributes to the rostral–caudal positioning of neural crest cells by creating a repellent environment for migration. Thus, Sema 3A/Sema 3F-Neuropilin–dependent signaling events act in concert to ensure appropriate positioning of neural crest cells in proximity to the neural tube, a process absolutely vital for the subsequent establishment of neuronal connectivity within the hindbrain region.
Bovans Goldline hen's eggs from Norfolk flocks (Henry Stewart & Co., Ltd.) were incubated at 37°C in a humidified environment to stages required (Hamburger and Hamilton, 1992).
Beads (Affi-gel, Bio-Rad, UK) were prepared for implantation by washing in phosphate buffered saline, and incubated at room temperature in either chick Ringer's solution (control) or purified Sema 3A-Fc for at least 30 min. Embryos were incubated until stage 10 when 1 ml of albumin was removed and the eggs windowed. To visualize the embryos, ink (1:10 in chick Ringer's solution) was injected into the yolk. The vitelline membrane was removed by using a flame-sharpened needle, and an incision was made into the neural tube at the level of r1. Beads were placed onto the neural tube and pushed into position by using a blunt wire. Eggs were sealed and incubated up to 24 hr before collection and fixation.
In Ovo Electroporation
Full-length mouse Sema 3F and chick Sema 3A were generated by reverse transcriptase-polymerase chain reaction (RT-PCR) using the following primers: Sema 3F, 5′-GGTGTCCACAGAACTGGA, 3′-CTAGGCTGGTCCTATGCAG; Sema 3A, 5′-CAGTGGATCCTGCAGCATGGGCTGG, 3′-GCTTGCGCTCGAGTCACCAAACCTGCTCACAGAATTC. Sema 3F was then cloned into a pcDNA1.1 vector using XhoI and XbaI sites, whereas Sema 3A was cloned into pCRII (Invitrogen) for direct sequencing then subcloned into the pCAβlinker vector using XhoI. To verify the activity of the constructs used for electroporation, Cos-7 cells were transiently transfected with the Sema 3A or Sema 3F expression vectors. Collected supernatants induced the collapse of growth cones extending from dorsal root ganglion (Sema 3A) and hippocampal (Sema 3F) explants.
Embryos were prepared as described above, and then silver and tungsten electrodes were positioned on either side of the neural tube such that electroporation always occurred into the left side of the neuroepithelium. A solution containing fast green diluted 1:5 in either pCAβ EGFPm5 alone, a mix of pCAβ linker- Sema 3A/pCAβ EGFPm5, or pcDNA1.1- Sema 3F/pCAβ EGFPm5, pIGplus-ECD Neuropilin-1/pCAβ EGFPm5, or pIGplus-ECD Neuropilin-2/pCAβ EGFPm5 were then injected into the lumen of the neural tube. The electroporation was carried out using an Intracell dual pulse isolated stimulator using four, 50-msec, 20-V pulses. The eggs were then sealed, and embryos were allowed to develop for a further 12, 24, or 36 hr, before they were collected and fixed. After fixation, embryos were subjected to in situ hybridization or immunohistochemistry. For DiI labeling of neural crest cells, DiI was injected into the lumen of the neural tube as previously described (Begbie and Graham, 2001) 4 hr before electroporation.
In Situ Hybridization and Wholemount Immunohistochemistry
Chicken Plexin-A1 was cloned by RT-PCR using the following degenerate primers: 5′-CCNGTNYTNAARGARATGGA, 3′-YTTDATNACRTTNACCCA, generating an approximately 400-base pair fragment. This fragment was confirmed by direct sequencing. Whole-mount in situ hybridization was carried out according to Henrique et al. (Henrique et al., 1995), using the chick Sema 3A, Sema 3F, RXRγ, Sox-10, Neuropilin-2, Plexin-A1, and Cad-7 probes. Embryos were photographed and embedded before cutting 50-μm vibratome sections. The whole-mount HNK-1 (1:200, Zymed) and neurofilament (1:10,000, Zymed) stainings used the protocol previously described by Eickholt et al. (2001). Embryos were bisected and analyzed using section confocal microscopy.
We thank Anthony Graham for the RXRγ probe, Paul Scotting for the Sox-10 probe, S. Nakagawa and Masatoshi Takeichi for the Cad-7 probe, and Yuji Watanabe for the Neuropilin-2 probe. The Neuropilin-Fc expression constructs were kindly provided by Rabinder Prinjha. We thank John Gilthorpe for the pCAβ EGFPm5 and the pCAβ linker electroporation vectors. We also thank Christine Ferguson for training and advice with the electroporation, and Anthony Graham and Sarah Mackenzie for critical suggestions and helpful discussion.