Neuropilins define distinct populations of neural crest cells

Background Neural crest cells (NCCs) are a transient embryonic cell type that give rise to a wide spectrum of derivatives, including neurons and glia of the sensory and autonomic nervous system, melanocytes and connective tissues in the head. Lineage-tracing and functional studies have shown that trunk NCCs migrate along two distinct paths that correlate with different developmental fates. Thus, NCCs migrating ventrally through the anterior somite form sympathetic and sensory ganglia, whereas NCCs migrating dorsolaterally form melanocytes. Although the mechanisms promoting migration along the dorsolateral path are well defined, the molecules providing positional identity to sympathetic and sensory-fated NCCs that migrate along the same ventral path are ill defined. Neuropilins (Nrp1 and Nrp2) are transmembrane glycoproteins that are essential for NCC migration. Nrp1 and Nrp2 knockout mice have disparate phenotypes, suggesting that these receptors may play a role in sorting NCCs biased towards sensory and sympathetic fates to appropriate locations. Results Here we have combined in situ hybridisation, immunohistochemistry and lineage-tracing analyses to demonstrate that neuropilins are expressed in a non-overlapping pattern within NCCs. Whereas Nrp1 is expressed in NCCs emigrating from hindbrain rhombomere 4 (r4) and within trunk NCCs giving rise to sympathetic and sensory ganglia, Nrp2 is preferentially expressed in NCCs emigrating from r2 and in trunk NCCs giving rise to sensory ganglia. By generating a tamoxifen-inducible lineage-tracing system, we further demonstrate that Nrp2-expressing NCCs specifically populate sensory ganglia including the trigeminal ganglia (V) in the head and the dorsal root ganglia in the trunk. Conclusions Taken together, our results demonstrate that Nrp1 and Nrp2 are expressed in different populations of NCCs, and that Nrp2-expressing NCCs are strongly biased towards a sensory fate. In the trunk, Nrp2-expressing NCCs specifically give rise to sensory ganglia, whereas Nrp1-expressing NCCs likely give rise to both sensory and sympathetic ganglia. Our findings therefore suggest that neuropilins play an essential role in coordinating NCC migration with fate specification.


Background
Neural crest cells (NCCs) are transient embryonic cells that generate a wide spectrum of derivatives. Based on their origin along the anteroposterior axis, NCCs are broadly divided into different populations that give rise to a restricted set of derivatives. Cranial NCCs arise anterior to the fifth somite, and form derivates such as sensory, sympathetic, and parasympathetic ganglia, connective tissue, bone, cartilage, and muscle tendons of the head [1].
Trunk NCCs arise posterior to the fourth somite, and give rise to cell types such as sympathetic and sensory neurons, adrenal chromaffin cells, and melanocytes [2]. Vagal NCCs arise from the level of somites 1 to 7, and produce enteric NCCs that form the neurons and glia of the enteric nervous system [3], and other cell types such as cardiac NCCs (level of somites 1 to 3) that form vascular smooth muscle lining the great arteries and contribute to the aorticpulmonary septum [4].
Even within these anatomically defined domains, NCCs can be further subdivided by additional properties, including their migration path and their developmental fate [5][6][7]. For example, cranial NCCs migrate in a segmented pattern to populate specific regions of the face and neck. This segmentation is critical for the correct formation and positioning of cranial NCC derivatives, and is under the control of cell intrinsic guidance receptors and signals provided by the environment in which NCCs travel and ultimately differentiate [8,9]. Trunk NCCs also migrate along distinct paths that correlate with their developmental fate [5]. Thus, NCCs migrate ventrally, and either stop within the sclerotome to give rise to sensory neurons and glia within the dorsal root ganglia (DRG), or migrate towards the dorsal aorta to give rise to the neurons and glia of the sympathetic nervous system. A proportion of these ventrally migrating NCCs that initially migrate along axonal pathways also give rise to melanocytes [10]. By contrast, NCCs migrating dorsolaterally between the dermomyotome and epidermis give rise only to melanocytes. A fundamental question for this field is how NCCs navigate their environment to position themselves in appropriate locations.
There is mounting evidence from chick and mouse studies supporting the notion that NCCs migrating dorsolaterally are specified at the time of delamination [6,[11][12][13], and that lineage-specific expression of cell surface receptors promotes migration along this path [5]. Indeed, several receptor-ligand pairs, including the Slit/ ROBO [14], Eph/ephrin [11], endothelin/EDNR [11], and Steel factor/KIT [15,16] families have been implicated in permitting access to the dorsolateral path. By contrast, there are two competing models to explain the timing at which ventrally migrating NCCs become specified towards the sensory or sympathetic lineages. On the one hand, it is proposed that this population of NCCs are specified prior to or at the time of delamination, and that lineage-specific properties promote migration along restricted pathways. This notion is derived from the findings that neurogenin 2 (Ngn2)-expressing NCCs are strongly biased towards a sensory fate in mice [17], and that nociceptive neurons arise specifically from contralaterally migrating NCCs within the chick neural tube [18]. Specification of pre-migratory NCCs is also supported by lineage-tracing experiments detailing a spatiotemporal fate map of sympathetic-and sensory-fated cells within the chick dorsal neural tube [19]. However, using in vivo optical imaging to analyse the behaviour of single cells, McKinney and colleagues recently found that pre-migratory NCCs in the chick lack positional identity or fate prior to delamination [20], but rather, they migrate from the neural tube stochastically and rapidly change their expression profiles as they migrate. This and other lineage/transplantation studies in chick support an alternative model in which migrating NCCs are proposed to be multipotent, becoming specified by their environment only as they migrate towards or at their target regions [21][22][23].
The molecular cues that cause some ventrally migrating NCCs to remain near the neural tube to form sensory ganglia, while others continue migrating to adopt other fates (e.g. sympathetic ganglia) remain ill defined. In chick, the chemokine receptor CXCR4 is expressed in a specific subset of early migrating NCCs to control their migration towards a source of SDF1 around the anlagen of the sympathetic ganglia [24,25]. Neuropilins (Nrp1 and Nrp2) are transmembrane receptors for guidance molecules of the class 3 semaphorin (SEMA3) family and for heparin-binding isoforms of vascular endothelial growth factor (VEGF) [26]. Nrp1 and Nrp2 are expressed in both cranial NCCs and ventrally migrating trunk NCCs, and are required for proper segmentation of the peripheral nervous system [27][28][29][30]. Notably, phenotypic analysis of single Nrp1 or Nrp2 knockout embryos identified distinct migration defects in different NCC lineages. In the head, NCCs emigrating from r4 are misguided in Nrp1 knockout embryos [31], while NCCs emigrating from r2 are misguided in Nrp2 knockout embryos [31,32]. Accordingly, neurons arising from these populations of cranial NCCs are ectopically positioned within neuropilin knockout embryos: neurons from the facio-acoustic ganglia (VII-VIII) are ectopically positioned in Nrp1 knockout embryos, while neurons from the trigeminal ganglia (V) are ectopically positioned in Nrp2 knockout embryos [31,32]. These lineage-specific NCC defects also extend to the trunk, where Nrp1 knockout embryos have ectopically placed neurons in the sympathetic and sensory nervous systems, and Nrp2 knockout embryos have misplaced neurons in the sensory nervous system [28,33,34]. In combination with expression analyses, these observations suggest that neuropilins may be required for the migration of distinct populations of NCCs, and that this may be controlled by the restricted expression of Nrp1 and Nrp2 in NCCs biased towards sympathetic and/or sensory fates [28]. However, to date, this notion has lacked definitive support from co-expression analyses and fate mapping of neuropilin-expressing NCCs.
Here we have combined in situ hybridisation, immunohistochemistry and lineage-tracing analyses to directly compare the expression of Nrp1 and Nrp2 in migrating NCCs. Our results demonstrate that these receptors are expressed in a non-overlapping pattern within different populations of cranial and trunk NCCs. Whereas Nrp1 is expressed in cranial NCCs migrating out of hindbrain r4 and within trunk NCCs migrating towards the sympathetic and sensory ganglia, Nrp2 is preferentially expressed in cranial NCCs migrating out of r2 and in trunk NCCs migrating towards and positioned within the sensory ganglia. To determine whether Nrp2-expressing NCCs are fate-restricted towards the sensory lineage, we generated a tamoxifen-inducible Cre/LoxP tracing system to genetically label and lineage-trace this population of cells. In combination with our detailed expression analyses and previous phenotypic analyses [33,34], our finding that Nrp2-expressing NCCs give rise to sensory neurons in the DRG and not the sympathetic chain identifies an essential role for neuropilins in coordinating NCC migration with fate specification.

Results and discussion
Nrp1 and Nrp2 are expressed in different populations of cranial NCCs Previous studies, including our own, demonstrated that Nrp1 and Nrp2 knockout mice have non-overlapping phenotypes in different cranial NCC derivatives [31,32]. To address if these receptors may be expressed in distinct populations of cranial NCCs, and thereby explain the distinct migration phenotypes of neuropilin knockout embryos, we completed whole mount in situ hybridisation on wild-type embryos between embryonic day (E) 8.5 to E9.5. By comparing the expression profiles of Nrp1 and Nrp2 with that of the pan-NCC marker Sox10, we found that neuropilins are expressed in different populations of cranial NCCs. At E8.5 and E9.0, the time at which NCCs have delaminated from the neural tube and begun to migrate into the branchial arch tissue, the expression of Nrp1 was restricted to the stream of NCCs emigrating out of r4, while Nrp2 was expressed reciprocally within NCCs emigrating out of r2 ( Figure 1A-F). At E9.5, this restricted expression pattern was maintained; however, low levels of Nrp2 could also be detected in or around the r4 stream of NCCs ( Figure 1G-I). In addition to this presumptive staining within NCCs, Nrp1 expression was identified in the heart, gut, and arteries ( Figure 1B, E, H), while Nrp2 expression was evident in the dorsal regions of the heart ( Figure 1C, F, I).
To investigate whether neuropilins are expressed in NCCs and not other branchial arch tissue such as the surface epithelia or cranial mesoderm, we sectioned E8.5 embryos transversely through the r2 and r4 regions, and counterstained with antibodies specific to the neurotrophin receptor p75. Sox10 and p75 were co-expressed in all migrating cranial NCCs at this stage of development ( Figure 1J-M). Consistent with the results of whole mount staining, Nrp1 expression was identified specifically in NCCs emigrating out of r4 and also within non-NCC tissues such as the cranial mesoderm and the developing heart ( Figure 1N, P). Conversely, Nrp2 expression was restricted to NCCs emigrating out of r2 and additional tissues such as the cranial mesoderm, neural tube, and bulbus cordis region of the developing heart ( Figure 1R, T).
To compare expression of Nrp1 and Nrp2 in NCCs, we immunostained embryos generated by crossing Wnt1Cre [35] mice with Z/EG [36] reporter mice to specifically label all NCCs and their derivatives with green fluorescent protein (GFP). Whole mount immunofluorescence staining clearly demonstrated that Nrp1 and Nrp2 are expressed in a reciprocal pattern, replicating the mRNA expression seen in our previous in situ hybridisation analysis (Figure 2A-D). Immunostaining also identified expression of Nrp1 and Nrp2 in blood vessels around the r2 and r4 migrating NCCs (Figure 2A-D). Longitudinal sections through the head confirmed this restricted expression along the anteroposterior axis, as well as expression within the developing vasculature ( Figure 2E-H). The membrane localisation of neuropilins is consistent with their role as cell surface receptors for guidance molecules of the SEMA3 and VEGF families [30]. Taken together, our in situ hybridisation, immunohistochemical, and lineagetracing analyses confirm that neuropilins are expressed in different populations of cranial NCCs that are defined by their position along the anteroposterior axis. These restricted expression profiles therefore explain the disparate phenotypes of neuropilin knockout mice in which Nrp1 has ectopically placed neurons of the facio-acoustic cranial ganglia (VII-VIII) and Nrp2 has ectopically placed neurons of the trigeminal ganglia (V) [31,32]. Future work addressing how the expression of neuropilins is controlled should provide essential clues to the mechanisms dividing cranial NCCs into distinct populations.

Nrp1 and Nrp2 are expressed in different populations of trunk NCCs
Nrp1 and Nrp2 knockout embryos also displayed disparate phenotypes in different trunk NCC derivatives [33,34]. To explore if neuropilins are also expressed in different populations of ventrally migrating trunk NCCs, we compared their expression with that of the pan-NCC markers Sox10 and p75 over the forelimb region of E9.5 embryos. In situ hybridisation confirmed that Nrp1 was expressed within NCCs migrating in the anterior half of the somite and in non-NCC derivatives such as the dorsal aorta and intersomitic blood vessels ( Figure 3B). Nrp2 was robustly expressed in the anterior half of the somite in a pattern that partially overlapped with Sox10 ( Figure 3C). To determine if neuropilins are expressed within NCCs or the supporting somitic tissue, we next counterstained longitudinal in situ hybridisation sections with anti-p75 antibodies ( Figure 3D-I). In addition to labelling Sox10expressing NCCs, p75 also recognised neuroepithelial cells within the neural tube ( Figure 3D, G). Co-staining of Nrp1 and p75 within the anterior half of the somite indicated that Nrp1 was expressed in trunk NCCs migrating along the ventral path ( Figure 3E, H). In addition, Nrp1 was also expressed in the posterior somite and within intersomitic blood vessels. Nrp2 was broadly expressed throughout the entire anterior half of the somite, with some cells costaining with p75 ( Figure 3F, I).
To investigate whether neuropilins are co-expressed or differentially expressed within distinct populations of ventrally migrating NCCs, we performed immunofluorescence analysis on E9.5 Wnt1Cre; Z/EG embryos. Whole mount staining for GFP, Nrp1 and Nrp2 identified mixed populations of NCCs expressing the neuropilin receptors ( Figure 4A-E). Consistent with our in situ hybridisation analysis results, Nrp1 was identified in NCCs in the anterior half of the somite, the intersomitic blood vessels, and the dorsal aorta ( Figure 4D). Nrp2 was also expressed in NCCs in the anterior half of the somite in addition to the intersomitic blood vessels. Notably, Transverse sections confirmed that Nrp1 is specifically expressed in the r4 NCC stream as well as the developing heart and cranial mesoderm (cm).
there was limited overlap of the Nrp1 and Nrp2 expression domains, and NCCs expressing high levels of Nrp1 with low or absent Nrp2 could be identified in the dorsal regions of the somite ( Figure 4A-E arrow).
To define the expression of neuropilins within individual trunk NCCs, we next analysed transverse sections of E9.5 embryos taken from the anterior half of the somite at the level of the developing forelimb (somites 13-14, Figure 4F). Consistent with the whole mount staining results, we identified Nrp1 within migrating NCCs, the dorsal aorta, and the periaortic mesenchyme. Nrp2 was also identified in NCCs and NCC precursors within the neural tube, the floor plate, somitic mesenchyme, and anterior cardinal veins. Notably, NCCs located around the dorsal aorta that had sympathetic fate preferentially expressed Nrp1 (Figure 4Fii, white arrow). Moreover, NCCs with presumptive sympathetic fate migrating towards the dorsal aorta also preferentially expressed high levels of Nrp1 (Figure 4Fii and inset). By contrast, NCCs migrating within the somite and/or arresting within the anlagen of the DRG expressed various combinations of the neuropilin receptors. Although the majority of the NCCs within the area of the DRG co-expressed Nrp1 and Nrp2, we also identified NCCs within and around the DRG that preferentially expressed either Nrp1 (Figure 4Fi, arrow) or Nrp2 alone (Figure 4Fi and inset). NCCs expressing only Nrp1 were mostly located at the ventral side of the DRG in close association with the neural tube, while NCCs expressing only Nrp2 were located at the prospective dorsal root entry zone.
We tested the notion that NCCs express combinations of the neuropilin receptors by purifying GFP-positive trunk NCCs from E9.5 Wnt1Cre; Z/EG embryos, using fluorescence-activated cell sorting (FACS). Consistent with our in vivo expression analysis, counterstaining of Nrp1 and Nrp2 identified three populations of trunk NCCs based on their neuropilin expression profiles: 1) Nrp1 high/Nrp2 high, 2) Nrp1 high/Nrp2 low, and 3) Nrp2 high/Nrp1 low ( Figure 5). Given the different locations of these cells within the embryo, our expression profiling also suggests that each of these NCC types may give rise to different derivatives. We therefore propose that the majority of Nrp1 high/Nrp2 high NCCs form neurons and glia of the DRG, Nrp2 high/Nrp1 low NCCs form boundary cap cells at the dorsal root entry zone, and Nrp1 high/ Nrp2 low NCCs form neurons and glia of the sympathetic nervous system.
Upon reaching the dorsal aorta, NCCs commence differentiation into bona fide sympathetic neurons via the elevated expression of the basic helix-loop-helix transcription factor MASH1. NCC derivatives that have initiated this differentiation program also begin to express Nrp2 [37]. Consistent with this developmental progression, our expression analysis in sections from more developmentally advanced regions of E9.5 embryos (somites 5-6 that are anterior to the forelimb) identified a small number of GFP-positive cells at the dorsal aorta co-expressing Nrp1 and Nrp2 (not shown). Previous phenotypic analysis of neuropilin knockout embryos found that Nrp1 is required for correct positioning of both sensory and sympathetic neurons, while Nrp2 is required for positioning of sensory neurons [28,33,34]. Our expression analysis therefore explains the disparate phenotypes between the two types of knockout mice, and further suggests that neuropilins may play essential roles in coordinating NCC migration with fate specification.

Generation of an inducible lineage-tracing Nrp2 transgenic mouse model
To identify the fate of neuropilin-expressing NCCs and to determine if Nrp2-expressing trunk NCCs are biased towards the sensory neuronal lineage, we generated a tamoxifen-inducible Nrp2 lineage-tracing mouse model. A Nrp2-CreERT2/Kikume transgenic construct was generated by replacing the ATG start codon of the Nrp2 gene with the entire CreERT2 IRES Kikume sequence followed by a polyadenylation signal sequence, thus placing CreERT2 IRES Kikume under control of the endogenous regulatory elements contained in the Nrp2 gene locus ( Figure 6A). The engineered construct was used to generate a transgenic founder mouse that was identified by PCR using primers specific for CreERT2.
The offspring from the founder line appeared grossly normal, and the transgene was transmitted as predicted by Mendelian ratios.
To permanently label and follow the fate of Nrp2 expressing cells throughout embryogenesis, Nrp2-CreERT2/ Kikume transgenic mice were crossed with R26R LacZ reporter mice to create Nrp2-CreERT2/Kikume; R26R LacZ embryos. Upon addition of tamoxifen to these embryos, CreERT2 translocated to the nucleus to promote stable expression of β-galactosidase (lacZ) in cells and descendants of cells endogenously expressing Nrp2 at the time of injection. No lacZ labelling was detected without tamoxifen injection (data not shown).
We next determined that intraperitoneal injection of tamoxifen at 2-4 mg per 40 g body weight to pregnant dams was sufficient to label Nrp2-expressing cells without resulting in embryonic lethality. As expected, higher doses of tamoxifen injected at E7.5 to E8.5 had detrimental effects on embryos.
Having optimised the dosage of tamoxifen, we first validated that the transgene recapitulated the Nrp2 expression profile in NCCs. Tamoxifen was injected at E9.0 and E9.5, and embryos collected at E10.0. Analysis of Xgal staining in whole mount embryos demonstrated that the transgene is expressed in a highly overlapping pattern to endogenous Nrp2 (n = 12; compare Figure 6B  with Figures 1I, 2D, 3C, and 4E). Indeed, the transgene expression pattern replicated the expression of Nrp2 within a distinct striped pattern in the trunk and within NCCs migrating out of r2 ( Figure 6B-D). To confirm that the expression was occurring within NCCs, we sectioned E10.0 embryos and counterstained with the NCC marker p75 ( Figure 6E-H populations and other cell types in which Nrp2 is expressed.

Nrp2-expressing NCCs give rise to neurons and glia in the trigeminal ganglia
Using this inducible lineage-tracing mouse model, we first determined the derivatives of Nrp2-expressing NCCs in the head. Tamoxifen was administered to pregnant dams when embryos were at E9.5 and E10.0, and later collected at E11.5. Consistent with Nrp2 being restricted to specific cranial NCC subpopulations, analysis of whole mount embryos revealed distinct X-gal staining in the trigeminal (V) ganglia that are derived from NCCs migrating out of r2 (n = 8, Figure 7A). By contrast, minimal staining was detected in r4 NCC derived structures such as the facio-acoustic ganglia (VII-VIII). X-gal staining also identified additional Nrp2-expressing cells in the forebrain region. Longitudinal sections counterstained with eosin confirmed the restriction of Nrp2-expressing NCCs to the trigeminal ganglia (V) (Figure 7B-C).
To define the fate of Nrp2-expressing NCCs, we next counterstained longitudinal X-gal-stained sections with antibodies specific for neurons (anti-Tuj1 antibodies) and glia (anti-Sox9 antibodies) ( Figure 7D-G). Under high magnification, Nrp2-expressing NCCs were found to give rise to both Tuj1-positive neurons and Sox9-positive glia within the trigeminal ganglia (V) (Figure 7Gi, Gii, respectively). Taken together with our expression profiling results, we conclude that Nrp2 expression is restricted to NCCs emigrating out of hindbrain r2. As we were unable to detect X-gal staining in other derivatives of the r2 stream such as bone and cartilage of the mandible, our results suggest that Nrp2-expressing NCCs preferentially give rise to neuroglial lineages. In addition, both whole mount staining and staining of transverse sections (not shown), suggest that Nrp2-expressing NCCs give rise to

Nrp2-expressing trunk NCCs are biased towards sensory ganglia
Our expression analysis raised the hypothesis that ventrally migrating trunk NCCs expressing Nrp2 are biased towards neurons and glia of the sensory ganglia. To test this notion, we completed fate-mapping studies with the Nrp2-inducible lineage-tracing mouse model. Tamoxifen was administered to pregnant dams when embryos were E9.5 and E10.0, and these were later collected at E11.5, a stage at which the sensory and sympathetic ganglia have started to condense. Consistent with the staining of Nrp2 in NCCs and paraxial mesoderm at E9.5, X-gal staining identified derivatives of Nrp2-expressing cells within the DRG and skeletal muscle ( Figure 8A). Eosin staining of longitudinal sections anterior to the hind limb clearly demonstrated X-gal staining within the DRG ( Figure 8B-C). Transverse sections through the same axial level were also counterstained with the pan neuronal marker Tuj1 and the sympathetic specific neuronal marker tyrosine hydroxylase (TH) ( Figure 8D). In all cases examined (n = 3), X-gal staining was robustly observed in the DRG ( Figure 8D-F). By contrast, we were unable to observe X-gal staining in sympathetic ganglia ( Figure 8G-I). This finding demonstrates that Nrp2-expressing NCCs are strongly biased towards neurons and glia of the DRG. Our fate-mapping results of Nrp2-expressing NCCs are strikingly similar to that presented for lineage tracing of Ngn2-expressing NCCs in mice [17]. In the latter, Ngn2-positive cells were also biased towards sensory ganglia. As Ngn2 is a basic helix-loop-helix transcription factor that plays an essential role in sensory neuronal differentiation [38], it will be of interest to determine if Nrp2 is also involved in the same molecular pathway within NCCs.

Conclusions
The work presented here identified molecularly distinct populations of cranial and trunk NCCs based on their expression profiles of the cell surface receptors Nrp1 and Nrp2. In combination with the recent analyses of Nrp1 and Nrp2 knockout mice [31][32][33][34], this work further identifies essential roles for neuropilins in sorting specific populations of NCCs to their final destinations within the embryo ( Figure 9A). In the head, Nrp1 is expressed in NCCs emigrating from r4, and is required for the migration and condensation of neurons and glia of the facio-acoustic ganglion (VII-VIII) [31]. By contrast, Nrp2 is expressed in NCCs emigrating from r2, and is required for the migration and condensation of neurons and glia of the trigeminal ganglion (V) [32].
In addition to their differential expression in restricted populations of NCCs along the anteroposterior axis, neuropilins also label divergent populations of NCCs along the dorsoventral axis ( Figure 9A). Trunk NCCs migrating along the ventral path through the anterior somite consist of a mixed population of sympathetic and sensory progenitors. Our finding that Nrp1 and Nrp2 expression correlates with sympathetic and sensory progenitors, and that Nrp2-expressing NCCs are biased towards neurons and glia of the sensory nervous system therefore suggests that neuropilins, possibly in combination with other guidance molecules such as CXCR4 [25], provide part of the molecular machinery to guide ventrally migrating NCC precursors to correct target regions of the embryo ( Figure 9B). Indeed, phenotypic analysis of neuropilin knockout mice is in strong agreement with this notion [33,34].
Our study identified three distinct NCC sub-types within and around the DRG that can be segregated based on their neuropilin expression profiles. While Nrp2expressing NCCs (Nrp2-high/Nrp1-high and Nrp2-high/ Nrp1-low) were traced to the DRG it will be important to complete the inverse experiment with lineage tracing of Nrp1-expressing NCCs to determine if this population is also fate-restricted.
Our expression analysis also identified Nrp2 within presumptive NCC precursors in the dorsal neural tube ( Figure 4F). Given that Nrp2 knockout embryos have defects restricted to sensory NCC derivatives, and that Nrp2-expressing NCCs are fate-restricted towards a  Figure 9 Neuropilins define molecularly distinct populations of ventrally migrating trunk NCCs. (A) Schematic detailing the neuropilin expression profiles in migrating trunk NCCs. NCCs migrating to the dorsal aorta (da) give rise to the sympathetic ganglia, while NCCs that stall within the somite alongside the neural tube (nt) give rise to sensory neurons of the DRG. Trunk NCCs at the level of the forelimb begin to delaminate from the neural tube at E8.5, and by E9.0, have started to migrate within the somite. At E9.5, some NCCs have already reached the dorsal aorta, while others have stalled in the anlagen of the DRG. At E10.5, NCCs with sympathetic fate have condensed, while NCCs with sensory fate have started differentiating in the DRG. Expression profiling identified distinct populations of NCCs with Nrp1 (red), Nrp2 (blue), and Nrp1/Nrp2 (purple). Nrp1-expressing cells preferentially migrate towards the dorsal aorta while Nrp2-expressing cells stall within the area of the DRG. Nrp2 was also expressed in presumptive NCC precursors within the dorsal neural tube. (B) Schematic diagram detailing the fate restriction of trunk NCCs. NCCs delaminate from the neural tube to migrate along two separate paths. Ventrally migrating NCCs initially travel through the intersomitic space to seed the sympathetic ganglia, and then switch to travel ventrally through the anterior half of the somite to give rise to sensory ganglia, sympathetic ganglia, and melanocytes. NCCs migrating dorsolaterally also give rise to melanocytes. Expression, fate-mapping, and phenotypic studies suggest that Nrp2 (Nrp2 alone (blue) and Nrp1/Nrp2 (purple)) is a marker of NCCs biased towards sensory ganglia. In addition, expression and phenotypic studies suggest that Nrp1 (red) is a marker of ventrally migrating NCCs that give rise to sympathetic ganglia. sensory phenotype, our data could also be taken to support the notion that at least some pre-migratory NCCs are fate-biased prior to delamination. However, as clonal analysis or transplantation of Nrp2-expressing premigratory/migrating NCCs into ectopic environments has not yet been achieved, our studies are currently unable to decipher between the two models of when ventrally migrating NCCs are segregated into specified lineages. For example, although the Nrp2-expressing cells may be biased towards a sensory phenotype, they may only commit to this lineage upon reaching their final resting place within the sclerotome.
Taken together with previous reports, our studies demonstrate that neuropilins provide part of the molecular machinery to sort heterogeneous populations of ventrally migrating NCCs towards correct target regions.

Mice
All experimentation was approved by and conducted in accordance with the guidelines of the Animal Ethics Committee of the SA Pathology/Central Adelaide Local Health Network, and followed the Australian code of practice for the care and use of animals for scientific purposes. To obtain embryos of defined gestational ages, animals were mated in the evening, and the morning of vaginal plug formation was counted as embryonic day (E) 0.5. To lineage-trace NCCs and their derivatives throughout development, we crossed Wnt1Cre [35] with Z/EG [36] mice.

Generation of Nrp2-CreERT2/Kikume transgenic mice
The BAC clone RP24-250G22, containing 62 kb of DNA upstream of the first exon to 4 kb past the final coding exon of the Nrp2 locus, was obtained from the BAC resource at Children's Hospital Oakland Research Institute (Oakland, California, USA). A CreERT2 IRES FLAG-Kikume cassette was inserted at the initiation codon of the Nrp2 coding sequence, which is located within exon 1, by homologous recombination using a targeting vector designed for BAC recombineering [39,40]. A 5′ homology arm was amplified by PCR with a forward primer 5′-taactagtctcgagctctgggaacacagagctgag-3′ and reverse primer 5′-taactagtagagagcgatccgattacg-3′ and inserted upstream of the CreERT2 IRES FLAG-Kikume sequence in plasmid pBS-CreERT2 IRES FLAG-Kikume (Schwarz laboratory). The 5′ homology arm and CreERT2 IRES FLAG-Kikume fragment was then sub-cloned to the recombineering vector PL451 using XhoI restriction sites. A 3′ homology arm was amplified by PCR with a forward primer 5′-taggatccgtaagcccttcaaagtttttc-3′ and reverse primer 5′-tagcggccgcaaagaatccacacatgtgaaaag-3′, and inserted downstream of CreERT2 IRES FLAG-Kikume Neo/Kan in the PL451 vector using unique BamHI/NotI sites. This created the Nrp2-CreERT2/Kikume PL451 recombineering vector.
The RP24-250G22 BAC was transformed into the recombinogenic Escherichia coli strain EL250 and maintained with chloramphenicol resistance. EL250 cells carrying RP24-250G22 BAC were electroporated with Nrp2-CreERT2/Kikume PL451, and homologous recombined positive clones were selected with kanamycin and chloramphenicol. The Neo/Kan cassette was removed by L-arabinose induction of flp recombinase, resulting in a Nrp2-CreERT2/Kikume BAC. Homologous recombination was confirmed by PCR and sequencing of the BAC using primers that flanked upstream of the Nrp2 5′ homology arm and downstream of the 3′ homology arm. High-quality BAC DNA was prepared using the Large Construct Kit (Qiagen, Chadstone, VIC, Australia) and analysed by pulsed field gel electrophoresis, restriction analysis, and BAC end sequencing. Prior to microinjection, the DNA elution buffer was replaced with fresh microinjection buffer (10 mM Tris-HCl, pH 8.0 0.25 mM EDTA, pH 8.0) by microdialysis. The Nrp2-CreERT2/Kikume BAC was microinjected into fertilized embryos by standard pronuclear injection techniques at the Transgenic Animal Services Queensland (TASQ, Brisbane, GLD, Australia). Genomic DNA was isolated from tail samples [41], and founder mice carrying the BAC transgene were identified by PCR with primer pairs recognising CreERT2: iCre forward 5′-gagagatggatctctgtgtc-3′ and iCre reverse 5′-gacttcatcagaggtggcatc-3′, yielding a 580 bp product. Founders and offspring were subsequently genotyped with the same primer pair. One founder line was produced that transmitted the transgene in a normal mendelian inheritance pattern, with all offspring appearing grossly normal.

In situ hybridisation
Whole mount and section in situ hybridisation was performed as described previously [42]. Riboprobes were transcribed from plasmids containing the cDNA sequence for Sox10, Nrp1, and Nrp2 [33,34].

Tamoxifen
Tamoxifen 1 g (Sigma-Aldrich, Sydney, NSW, Australia) was suspended in 5 ml of ethanol, and then dissolved in 45 ml of sunflower oil to produce a final stock at 20 mg/ml concentration. To dissolve the tamoxifen completely, the stock was sonicated in an ice bath for 5 min (with 30 second intervals). Pregnant dams were injected intraperitoneally with 2 mg tamoxifen per 40 g body weight at the indicated time points.

FACS sorting of primary neural crest cells
Primary NCCs were isolated from E9.5 Wnt1Cre; Z/EG embryos as previously described [43]. Trunk regions posterior to somite 10 were dissociated using Tryple Express (Invitrogen, Mulgrave, VIC, Australia) for 10 minutes at room temperature. Dissociated cells were washed twice with Dulbecco's modified Eagle's media (DMEM) containing 10% fetal calf serum (FCS), and resuspended in DMEM with 1% FCS for cell sorting. Cell sorting was performed on a Beckman Coulter Epics Altra HyperSort using Expo MultiComp Software (v1.2B; Beckman Coulter, Lane Cove, NSW, Australia) equipped with an Innova 300C water-cooled 488 nm argon laser at 100 mW. Sorting was conducted at room temperature, with the instrument pressurised to 12 psi and equipped with a 100 μm nozzle. Linear forward scatter (FSC) peak signal (pk), width (time of flight (TOF)) and internal signal area (INT) signals were collected to allow for standard scatter and doublet discrimination. Linear side scatter (SSC) and INT signal were collected with a 488/10 band pass filter in the photomultiplier tube (PMT1). Log GFP signal was collected in PMT2 with a 525/25 band pass filter behind a 488 nm long pass dichroic mirror. A gate was drawn on FSC versus SSC plot to exclude debris and dead cells, as discriminated by scatter properties alone. Following this, an FSC pk versus FSC INT plot was examined to allow distinction of single cells. Linearly related cells were gated for further analysis on a GFP versus SSC plot. Cells were collected into DMEM with 10% FCS. After collection, primary NCCs were centrifuged, resuspended in NCC growth media (DMEM F12, 5% chicken embryo extract, 10 mM HEPES, 2% B27, 1% N2, 20 ng/ml insulin-like growth factor (IGF, R&D, Minneapolis, MN, USA), and 100 U/ml Penicillin/Streptomycin (Invitrogen, Mulgrave, VIC, Australia). Cells were plated in this media in Ibidi μ-slide eight-well dishes coated with 50 mg/ml fibronectin (Roche, Basel, Switzerland) at a concentration of 5 × 10 4 cells/well. After 4 hours, cells were fixed with 4% PFA for 10 minutes, and immunolabelled with anti-Nrp1 and anti-Nrp2 antibodies. Images were acquired on an Olympus IX81 microscope using Cell X-cellence software (Olympus, Edwardstown, SA, Australia).