Semaphorin Signaling Guides Cranial Neural Crest Cell Migration in Zebrafish

View metadata, citation and similar papers at core.ac.uk brought to you by CORE

provided by Elsevier - Publisher Connector

Developmental Biology 280 (2005) 373–385 www.elsevier.com/locate/ydbio

Semaphorin signaling guides cranial neural crest cell migration in zebrafish

Hung-Hsiang Yu, Cecilia B. Moens*

Howard Hughes Medical Institute and Division of Basic Science, Fred Hutchinson Cancer Research Center, 1100 Fairview Avenue N., Box 19024 Seattle, WA 98109, USA

Received for publication 2 December 2004, revised 19 January 2005, accepted 28 January 2005

Abstract

Cranial neural crest cells (NCCs) migrate into the pharyngeal arches in three primary streams separated by two cranial neural crest (NC)- free zones. Multiple tissues have been implicated in the guidance of cranial NCC migration; however, the signals provided by these tissues have remained elusive. We investigate the function of semaphorins (semas) and their receptors, neuropilins (nrps), in cranial NCC migration in zebrafish. We find that genes of the sema3F and sema3G class are expressed in the cranial NC-free zones, while nrp2a and nrp2b are expressed in the migrating NCCs. sema3F/3G expression is expanded homogeneously in the head periphery through which the cranial NCCs migrate in lzr/pbx4 mutants, in which the cranial NC streams are fused. Antisense morpholino knockdown of Sema3F/3G or Nrp2 suppresses the abnormal cranial NC phenotype of lzr/pbx4 mutants, demonstrating that aberrant Sema3F/3G-Nrp2 signaling is responsible for this phenotype and suggesting that repulsive Sema3F/3G-Npn2 signaling normally contributes to the guidance of migrating cranial NCCs. Furthermore, global over-expression of sema3Gb phenocopies the aberrant cranial NC phenotype of lzr/pbx4 mutants when endogenous Sema3 ligands are knocked down, consistent with a model in which the patterned expression of Sema3 ligands in the head periphery coordinates the migration of Nrp-expressing cranial NCCs. D 2005 Elsevier Inc. All rights reserved.

Keywords: Cranial neural crest migration; Pbx; Neuropilin; Semaphorin

Introduction and to the cartilage, pigment, and connective tissue of the face and neck (Kontges and Lumsden, 1996). When the The formation of the vertebrate head requires a series of migration or differentiation of cranial NCCs is disrupted, patterning events which include the compartmentalization of defects of cranial NCC derived tissues occur that result in anterior neural tube, the definition of skeletal and muscular craniofacial malformations, the most common birth defect in shapes, and the establishment of a precise network of humans (Driscoll et al., 1992; Wilkie and Morriss-Kay, neuromuscular and skeletomuscular connections (Kimmel et 2001; Wilson et al., 1993). Cranial NCCs migrate into the al., 2001; Lumsden and Krumlauf, 1996). Within these, head periphery in three distinct streams that correlate with craniofacial development involves orchestrated interactions the axial level of their origin in the segmented sections of between the hindbrain neuroepithelium, head ectoderm, the hindbrain, called rhombomeres (Birgbauer et al., 1995; head mesoderm, and pharyngeal endoderm (Graham and Kontges and Lumsden, 1996; Lumsden et al., 1991). The Smith, 2001). Cranial neural crest cells (NCCs) play an first (anterior-most) stream of cranial NCCs that emanates essential role in sculpting the craniofacial structures by from rhombomere (r) 1 and r2 populates the first (man- contributing to the neurons and glia of the sensory ganglia dibular) pharyngeal arch, which gives rise to the components of the jaw. The second (middle) stream of cranial NCCs that emanates from r4 populates the second (hyoid) * Corresponding author. Fax: +1 206 667 3308. pharyngeal arch, which gives rise to the components of the E-mail address: [email protected] (C.B. Moens). face. The third (posterior) stream of cranial NCCs that

0012-1606/$ - see front matter D 2005 Elsevier Inc. All rights reserved. doi:10.1016/j.ydbio.2005.01.029 374 H.-H. Yu, C.B. Moens / Developmental Biology 280 (2005) 373–385 emanates from r6 populates the third and more caudal expressed in the striatum, serve as repulsive cues to prevent pharyngeal arches, which give rise to the components of the the Nrp1- and Nrp2-expressing cortical interneurons, neck and, in fish, the gills. These primary cranial neural derived from the basal telencephalon, from migrating to crest (NC) streams are separated by two cranial NC-free the striatum (Marin et al., 2001). In addition, Sema3C can zones, which are located lateral to r3 and r5 in the head promote the migration of cardiac NCCs since the pattern of periphery. cardiac NCCs in the outflow tract of the heart is severely Recent work has shown that signals in the head periphery altered in sema3C mutant mice (Feiner et al., 2001). control the pathway and the regional identity of migrating In this study, we investigate a role for Sema–Nrp cranial NCCs, but the molecular mechanisms underlying signaling in cranial NCC migration in vivo. We have shown these effects have remained elusive. The mechanism is that cranial NCC migration is abnormal in zebrafish likely to be complex since transplantation and ablation embryos in which the lazarus(lzr)/pbx4 gene is mutated experiments have variously implicated the surface ectoderm, (Po¨pperl et al., 2000). The vertebrate pbx genes encode the pharyngeal endoderm, the hindbrain neuroepithelium, TALE homeodomain proteins that act as DNA binding co- and the head mesoderm as influencing cranial NCC factors for Hox proteins (Mann and Chan, 1996). In lzr/ migration and/or differentiation (Couly et al., 2002; Golding pbx4 mutants, cranial NCCs migrate homogeneously, fail- et al., 2000, 2002, 2004; Trainor et al., 2002). The regional ing to avoid the normally cranial NC-free zones (Po¨pperl et identity of cranial NCCs can be specified by signals from al., 2000). lzr/pbx4, which is expressed ubiquitously, the endoderm since heterotopic transplantation of first-arch functions cell autonomously within the hindbrain to specify endoderm induced the differentiation of cartilage structures segment identities, but how it controls cranial NCC characteristic of the first arch (Couly et al., 2002). migration is unclear. Here, we uncover a role for Sema– Rhombomere grafting experiments in chick embryos also Nrp signaling in cranial NCC migration in zebrafish by demonstrated the presence of signal(s) in the head meso- identifying a genetic interaction between sema and lzr/pbx4. derm that control the pathway of migrating NCCs (Trainor We show by genetic mosaic analysis that lzr/pbx4 is non- et al., 2002). In the chick, r2, r4, and r6 produce abundant cell autonomously required for cranial NCC migration, cranial NCCs which migrate into the first, second, and third implying that a signal(s) in the environment of cranial NCCs arches, respectively, while r3 and r5 produce fewer cranial is affected. We have identified zebrafish nrp2a and 2b that NCCs which are diverted anteriorly and posteriorly around are expressed in cranial NCCs, and class 3 semas (sema3Fa, the adjacent cranial NC-free zones (Graham et al., 1993; sema3Ga, and sema3Gb) that are expressed in the cranial Kulesa and Fraser, 1998; Sechrist et al., 1993). When r2 or NC-free zones in the head periphery and are expanded in r4 were grafted adjacent to r3 or r5, their cranial NCCs were lzr/pbx4 mutants. Knockdown of Nrp2a, Nrp2b, or their diverted like r3 or r5 cranial NCCs (Ellies et al., 2002; Farlie ligands, Sema3Fa, Sema3Ga, and Sema3Gb, restores the et al., 1999). Finally, ablation studies in the chick, combined normal migratory pattern of cranial NCCs in lzr/pbx4 with analysis of erbB4 knockout mice, have suggested that mutants, suggesting that lzr/pbx4 functions to limit sema the neuroepithelium and surface ectoderm can also effect the expression to the cranial NC-free zones, and nrp-expressing migration of cranial NCCs, possibly by inducing or cranial NCCs are repelled from these regions. Consistent maintaining the inhibitory signal(s) in the head mesoderm with this model, over-expression of sema3Gb in a back- described above, since the aberrant cranial NCC migration ground in which endogenous class 3 Semas are knocked was observed after the ablation of r3 neuroepithelium and down recapitulates an abnormal cranial NCC migration r3/r5 surface ectoderm (Golding et al., 2000, 2002, 2004). phenotype observed in lzr/pbx4 mutant embryos. We Because cranial NCC migration involves redundant propose a model in which the lzr/pbx4-regulated expression signals from multiple germ layers, the molecules that of class 3 semas in cranial NC-free zones contributes to the control this process have remained elusive. A previous coordination of nrp-expressing cranial NCC migration. study suggested that the semaphorin (Sema)–neuropilin (Nrp) interaction can direct the migration of cranial NCCs since Nrp1-expressing cranial NCCs from explanted chick Materials and methods neural tubes avoided migrating on a Sema3A-containing substratum in an in vitro stripe assay (Eickholt et al., 1999). Fish maintenance Semas comprise a large family of phylogenetically con- served secreted and membrane-bound proteins that function Fish were raised and staged as described (Kimmel et al., in axon guidance and cell migration (Pasterkamp and 1995; Westerfield, 1994). Kolodkin, 2003). Class 3 secreted Semas transduce their signal through a family of transmembrane receptors Nrp1 In situ hybridization and Nrp2 (Chen et al., 1997; He and Tessier-Lavigne, 1997; Kolodkin et al., 1997). Aside from this putative role in RNA in situ hybridizations were performed as described cranial NCC migration, Semas also control cell migration in (Hauptmann and Gerster, 1994). Probes used in this study other developmental contexts: Sema3A and Sema3F, include crestin (Rubinstein et al., 2000), dlx2 (Akimenko H.-H. Yu, C.B. Moens / Developmental Biology 280 (2005) 373–385 375 et al., 1994), engrailed3 (Holland and Williams, 1990), as described (Gritsman et al., 2000; Fig. 2A). 4,5- krox20 (Oxtoby and Jowett, 1993), nrp1a (Yu et al., 2004), dimethyloxy-2-nitrobenzyl ester-caged fluorescein dextran nrp1b (Yu et al., 2004), nrp2a (Yu et al., 2004), nrp2b (Yu dye was injected into pax2.1 GFP-expressing embryos at et al., 2004), sema3Aa (Yee et al., 1999), sema3Ab (Roos the 1–4 cell stage. Uncaging of the caged fluorescein dye et al., 1999), sema3C (GenBank accession no. AY766117), was performed in wild-type and lzr/pbx4 mutants at 10s sema3D (Halloran et al., 1999), sema3Fa (GenBank ac- stage by a short pulse of laser (Photonic Instruments) to cession no. AY766118), sema3Fb (GenBank accession no. label presumptive cranial NCCs at different rhombomere AY766119), sema3Ga (GenBank accession no. AY766120), levels. The positions of uncaged cells were recorded sema3Gb (GenBank accession no. AY766121), and vegf165 immediately after uncaging and again after cranial NCCs (Liang et al., 1998). had migrated into the head periphery. Uncaged cells were identified in fixed embryos using an alkaline phosphatase- RNA and morpholino oligonucleotide injections conjugated anti-fluorescein antibody (Roche).

The coding region of zebrafish sema3Gb was cloned into Genetic mosaic analysis the pCS2+-GFP-LT-XLT vector. Capped sema3Gb mRNA and constitutively active variant of TARAM-A (Tar*) mRNA Genetic mosaic experiments were performed as described (Peyrieras et al., 1998) were synthesized using the mMes- (Moens and Fritz, 1999; Fig. 3A). Donor embryos were sage mMachine kit (Ambion). 150–200 pg of sema3Gb labeled at the 1–2 cell stage by injecting with a mixture of mRNA or 30–50 pg of Tar* mRNA was injected into one to rhodamine/biotin-dextran dye (Molecular Probes). Labeled two-cell stage zebrafish embryos. Morpholino oligonucleo- donor cells were transplanted into the region of shield stage tides (MOs) were purchased from Gene Tools (Philomath, host embryos that is fated to give rise to dorsal hindbrain OR), and sequences are listed in Table 1. MOs were made in and cranial NCCs (Kimmel et al., 1990; Woo and Fraser, 1Â Danieau buffer. 0.5–2nl of these MOs was injected into 1995). After transplantation, both donor and host embryos one to four-cell stage zebrafish embryos. To determine the were allowed to develop until 22 h post fertilization (hpf), efficacy of translation-blocking MOs, capped mRNAs of and the donor genotype was determined by RNA in situ GFP fusion constructs of nrp1a, nrp2a, nrp2b, sema3Aa, hybridization with krox20 (Po¨pperl et al., 2000). Biotin- sema3Ab, sema3D, sema3Fa, sema3Ga, and vegf165, con- labeled donor-derived cells were detected after in situ taining target sequences for the corresponding MO, were hybridization using the ABC kit (Vector Laboratories). injected into wild-type embryos. The effect of these MOs (3–5 ng) was observed by blocking the expression of GFP in co-injected embryos (data not shown). The effect of Results sema3Gb splice blocker MO (3 ng), which interrupts splicing of endogenous transcripts in MO-injected wild- Aberrant cranial NCC migratory pattern in the head type embryos, was detected by RTPCR (data not shown). periphery of lzr/pbx4 mutants

Mapping the cranial NC fate Instead of migrating as three distinct streams, cranial NCCs in lzr/pbx4 mutants migrate as a homogenous sheet pax2.1-GFP transgenic fish expressing GFP in r3 and r5 (Po¨pperl et al., 2000). To start to understand mechanisms beginning at 10 somite (10s) stage were used to accurately underlying the segmental migration of cranial NCCs, we label cranial NCCs derived from different rhombomeres determined whether the abnormal migration of cranial (Picker et al., 2002). Uncaging experiments were performed NCCs in lzr/pbx4 mutants was due to inappropriate specification or migration. Cranial NCCs, revealed by an NC marker crestin (Rubinstein et al., 2000), exist as Table 1 Sequence of MOs bilateral homogenous populations at 10 somite (10s) stage both in wild-type and lzr/pbx4 mutant embryos (arrowheads MO Sequence (5V to 3V) in Figs. 1A and 1B). Cranial NCCs were observed to sema3Aa GAAAAATCCCAACAAGGTAATCCAT segregate into three populations in r2, r4, and r6 at 13s sema3Ab GTACAATCCACCACAAGTAGTCCAT sema3D CTCCCCTGCAGTCTTCATGATGGAC stage. The separation between the second and third cranial sema3Fa TATCCATAAAACCCACAAGAGATTT NC streams is the most dramatic, and this separation was sema3Ga ACTCTCATAGTTCCAGCCATTCGCA observed both in wild-type and lzr/pbx4 mutant embryos a sema3Gb ATCGCAACATTTCTCACCTTTGTAT (asterisks in Figs. 1C and 1D). The fusion of cranial NC nrp1a GAGGATCAACACTAATCCACAATGC streams in the periphery is not observed in lzr/pbx4 mutants nrp2a CCAGAAATCCATCTTTCCGAAATGT nrp2b AAGATCCACAGAGCTTTGCGAATAA until cranial NCCs migrate beyond the otic vesicle (OV) at vegf165 GTATCAAATAAACAACCAAGTTCAT 18s stage (asterisks in Fig. 1F). Although the size of the OV a sema3Gb MO is a splice blocker targeting a splice donor. All other MOs is reduced in lzr/pbx4 mutants, this is unlikely to be the are translation blockers. cause of the fusion of cranial NCCs in lzr/pbx4 mutants 376 H.-H. Yu, C.B. Moens / Developmental Biology 280 (2005) 373–385

not altered in lzr/pbx4 mutants. This analysis of crestin staining suggests that the abnormal cranial NC fusion phenotype occurs in lzr/pbx4 mutants when cranial NCCs migrate into the normally cranial NC-free zones.

Mis-migrated cranial NCCs in lzr/pbx4 mutants derived from r4

To reveal the origin of cranial NCCs occupying the normally cranial NC-free zones in lzr/pbx4 mutants, we used a laser to activate caged fluorescein dextran dye at specific axial levels of the hindbrain before the onset of cranial NCC migration. In order to accurately label cranial NCCs from specific rhombomeres, we used pax2.1-GFP transgenic fish in which GFP is expressed in only r3 and r5 in the hindbrain (Picker et al., 2002). In wild-type embryos, labeled r4 and r5–r6 cranial NCCs migrate into the second and third cranial NC streams, respectively (arrowheads in Figs. 2D and 2H), consistent with the previous reports in chick, mice, and zebrafish (Kontges and Lumsden, 1996; Schilling et al.,

Fig. 1. Migrating cranial NCCs in wild-type (A, C, E, G, I) and lzr/pbx4 mutant (B, D, F, H, J) embryos revealed by crestin RNA in situ hybridization (blue). Flat-mount embryos are arranged with anterior to the left and krox20 (pink) indicates r3 and r5. Arrows indicate the second cranial NC stream in C–E. (A–B) Cranial NCCs (arrowheads) were observed as bilateral homogenous populations at the 10s stage both in the wild-type and lzr/pbx4 mutant embryos. (C–D) The separated cranial NCC streams and the cranial NC-free zone (asterisks) were observed at 13s stage both in wild-type and lzr/pbx4 mutant embryos. (E–F) The fusion of cranial NC streams in the periphery was observed at 18s stage in lzr/pbx4 mutants (asterisks in F), while the cranial NC-free zone was observed at the same stage of wild-type embryos (asterisk in E). The top and bottom panels in A, B, and E are different focal planes on both sides of the same embryo. (G–J) The distribution of cranial NCCs around the anterior OV (arrowheads) is shown in the two sequential cross sections of both the wild-type and lzr/ pbx4 mutant embryos. G and H are the cross sections anterior to I and J, respectively. Scale bar: 20 Am in A–J. Fig. 2. Mis-migrated lzr/pbx4 cranial NCCs are derived from r4. (A) Schematic figure of the uncaging experiment for tracing the migration of cranial NCCs. The cells labeled by laser uncaging are shown in blue and since the cranial NCC pattern is normal in ace/fgf8 mutants krox20 and dlx2 (for cranial NCCs) are shown in orange. B, C, F, and G are in which the OV is reduced or absent (Reifers et al., 1998; dorsal images of D, E, H, and I, respectively, to show the labeled unpublished observation). The continuously distributed rhombomeres. In wild-type embryos (B, D, F, H), labeled r4 and r5–r6 cranial NCCs can also be observed around the anterior cranial NCCs (arrowhead) migrate into the second and third cranial NC streams, respectively (D and H). In lzr/pbx4 mutants (C, E, G, I), cells OV in cross sections of lzr/pbx4 mutants, but not in wild- marked in r4 but not in r5–r6 give rise to cranial NCCs in the normally type embryos (Figs. 1G–1J). These cross sections also cranial NC-free zone. Asterisks mark the region with the mis-migrated lzr/ demonstrate that gross morphology in the cranial region is pbx4 cranial NCCs in E and I. Scale bar: 40 Am in B–I. H.-H. Yu, C.B. Moens / Developmental Biology 280 (2005) 373–385 377

2001; Trainor et al., 2002). In lzr/pbx4 mutants, cranial shown), suggesting that hindbrain neuroepithelium alone is NCCs occupying the normally cranial NC-free zone were not sufficient for cranial NCC patterning. largely derived from r4 (n = 17; arrowhead in Fig. 2E) while Since the pharyngeal endoderm can influence cranial r5- to r6-derived cranial NCCs migrated to their normal NCC migration and/or differentiation of cartilage structures position (n = 14; arrowhead in Fig. 2I). This suggests that (Couly et al., 2002), we targeted wild-type cells to the signals that normally control the migration of r4-derived pharyngeal endoderm of lzr/pbx4 mutant hosts by trans- cranial NCCs may be affected in lzr/pbx4 mutants. planting wild-type donor cells expressing the constitutively active variant of TGFh related receptor TARAM-A (Tar*), The requirement for lzr/pbx4 in cranial NCC migration is which causes cells to take on the endodermal fate (Peyrieras non-cell autonomous et al., 1998). We did observe the Tar*-expressing wild-type donor cells targeted specifically to the pharyngeal endo- We next used genetic mosaics to determine whether lzr/ derm, but the aberrant cranial NCC migration in lzr/pbx4 pbx4 is cell autonomously or non-cell autonomously required mutants was not rescued (data not shown). Although these for cranial NCC migration. When wild-type donor cells were experiments rule out a sole contribution of either hindbrain transplanted into wild-type hosts, many donor cells were neuroepithelium or pharyngeal endoderm in guiding cranial detected in the cranial NC streams (arrows in Fig. 3B; in this NCC migration, they do not rule out the possibility that example, donor cells are contributing primarily to the second either one may contribute in a more complex process and third arches) but not in the NC-free zones (arrowhead in involving more than one tissue. Fig. 3B). In contrast, when wild-type donor cells were transplanted into lzr/pbx4 hosts, wild-type cranial NCCs Nrp2a or Nrp2b knockdown suppresses the cranial NCC migrated aberrantly like the surrounding mutant host cells, migration defect in lzr/pbx4 mutants taking up positions in normally cranial NC-free zones (n = 19; arrowhead in Fig. 3C), suggesting that lzr/pbx4 is Our genetic mosaic analysis, as well as previous reports required in the environment of the migrating cranial NCCs. (Farlie et al., 1999; Golding et al., 2000, 2002, 2004; In addition, transplanting wild-type donor cells to the Trainor et al., 2002), suggests that a factor(s) outside the hindbrain neuroepithelium, which is mis-patterned in lzr/ cranial NCCs controls their migration, but the identity of pbx4 mutants (Po¨pperl et al., 2000), was not sufficient to this factor(s) is unknown. Sema3A-Nrp1 signaling has been restore the normal cranial NCC migratory pattern (data not shown to influence cranial NCC migration in tissue explant experiments (Eickholt et al., 1999). To test whether Sema– Nrp signaling is involved in cranial NCC migration in vivo, we cloned zebrafish nrp genes (nrp1a, nrp1b, nrp2a, and nrp2b; Yu et al., 2004). The developmental expression patterns of these four zebrafish nrp genes have been previously described (Bovenkamp et al., 2004; Lee et al., 2002; Martyn and Schulte-Merker, 2004; Yu et al., 2004). Here, we focus on the expression patterns of these genes during cranial NCC migration in wild-type and lzr/pbx4 mutants. In wild-type embryos, zebrafish nrp1a is expressed in the first cranial NC stream (arrowhead in Fig. 4A) and nrp1b is expressed in a subset of proximal cranial NCCs contributing to the cranial sensory ganglia which do not overlap with the more distal dlx2-expressing cranial NCCs (arrowheads in Fig. 4C). nrp2a and nrp2b are also expressed in cranial NCCs in wild-type embryos (Figs. 4E and 4G). While nrp2a is expressed strongly in the first and third cranial NC streams and weakly in the second cranial NC stream, nrp2b is expressed in all three cranial NC streams. In lzr/pbx4 mutants, expression of all four nrp Fig. 3. lzr/pbx4 functions non-cell autonomously to control cranial NCC migration. (A) Schematic of the approach for generating cranial NCC genes is maintained, although the organization of expressing mosaics. Donor-derived cells are brown, dlx2 and krox20 expression are in cells is altered in a manner corresponding to the aberrant blue. Red dash-lines circle the position of the OV in B-C. (B) In control pattern of cranial NCC migration in these mutants. nrp1a experiments, wild-type donor cells migrate normally into the second and expression in the first arch is expanded posteriorly, and third cranial NC streams of the pharyngeal arches (arrows) without nrp1b expression is more diffuse and extends between the occupying the cranial NC-free zone (arrowhead) in a wild-type host embryo. (C) In contrast, wild-type donor cells are seen in the normally first and second arch along the edge of the neuroepithelium cranial NC-free zone (arrowhead) after transplantation into a lzr/pbx4 (Figs. 4B and 4D), while nrp2a and nrp2b expression loses mutant host. Scale bar: 40 Am in B–C. its segmental organization (Figs. 4F and 4H). 378 H.-H. Yu, C.B. Moens / Developmental Biology 280 (2005) 373–385

NCC migration in wild-type embryos or enhanced or suppressed the abnormal cranial NCC migration phenotype of lzr/pbx4 mutants. When wild-type embryos were injected with these MOs, no cranial NCC migration phenotype was observed (Table 2), indicating that these genes are not strictly required for normal cranial NCC migration. However, when we injected nrp2a or nrp2b MO into lzr/pbx4 mutants, we observed a suppression of the cranial NC fusion phenotype (Figs. 4K and 4L; Table 2), so that gaps between cranial NC streams were restored. This suppression was stronger when nrp2a and nrp2b MOs were injected together (Table 2), suggesting nrp2a and nrp2b may act redundantly to regulate the migration of cranial NCCs. No such rescue was observed when we injected nrp1a MOs into lzr/pbx4 mutants (Table 2).

Sema3F and 3G are ligands for Nrp2a and Nrp2b in controlling cranial NCC migration

Our results suggested that cranial NCCs may respond to Nrp ligand(s) that are incorrectly distributed in lzr/pbx4 mutants. From studies in other vertebrates, the potential ligands for nrp2 include sema3C, sema3F, and vascular endothelial growth factor 165 (vegf165; reviewed in Neufeld et al., 2002). We have analyzed the expression patterns of class 3 secreted semas and vegf165 around 18s stage in both wild-type and lzr/pbx4 mutant embryos. We examined three previously cloned sema3 genes, sema3Aa (Yee et al., 1999), sema3Ab (Roos et al., 1999), and sema3D (Halloran et al., 1999) which are thought to be ligands for nrp1a and nrp1b, as well as five new sema3 genes-sema3C, sema3Fa, sema3Fb, sema3Ga, and sema3Gb that we cloned. Analysis of synteny between these zebrafish genes and their human homologs indicates that zebrafish sema3C is the ortholog of human sema3C, and that zebrafish sema3Fa and sema3Fb are duplicated orthologs of human sema3F. Although Fig. 4. Zebrafish nrps interact with lzr/pbx4 during cranial NCC migration. zebrafish sema3Ga and sema3Gb are very similar to human nrp expression is shown in blue in A–H; krox20 and dlx2 are in pink or sema3F, they map to a region of the zebrafish genome that orange in A–H and blue in I–L. Arrows indicate the second cranial NC is syntenic with a region on the human X chromosome but stream. (A–B) nrp1a is expressed on the first cranial NC stream that does not contain sema genes, suggesting that (arrowhead). (C–D) nrp1b is expressed in a subset of proximal cranial the sema3G ortholog was lost during tetrapod evolution NCCs contributing to the cranial sensory ganglia (arrowheads). (E–H) nrp2a in E-F and nrp2b in G–H are expressed in cranial NCCs. nrp2a is (H.-H.Y. and C.B.M. unpublished). expressed strongly on the first and third cranial NC streams and weakly on Here, we focus on the expression of semas in the head the second cranial NC stream, while nrp2b is expressed in all three streams. periphery during cranial NCC migration. Zebrafish sema3Aa Arrowheads indicate the third cranial NC stream in E and G. In lzr/pbx4 is expressed in the head periphery between the second and mutants, nrp expression in the cranial NC is disorganized in a manner third cranial NC streams in wild-type embryos (arrowhead in corresponding to the disorganization of cranial NCCs. (I–L) MO knockdown of Nrp2a (K) or Nrp2b (L) results in the restoration of the normal Fig. 5A) and this expression is maintained in lzr/pbx4 cranial NCC migratory pattern. Asterisks mark the cranial NC-free zone in mutants (arrowhead in Fig. 5B). Zebrafish sema3Ab is not I. K and L and cranial NCCs in this normally cranial NC-free zone in lzr/ expressed in the head periphery of wild-type embryos (Fig. pbx4 mutants (J). Scale bar: 40 Am in A–L. 5C), but its expression is detected in lzr/pbx4 mutants (arrowhead in Fig. 5D). Zebrafish sema3D is strongly We investigated the role of nrp2a and nrp2b in cranial expressed in r4 and in the second cranial NC steam in wild- NCC migration, with particular attention to the separation of type embryos (arrow in Fig. 5G), and interestingly this second and third cranial NC streams since these streams are expression is gone in lzr/pbx4 mutants (Fig. 5H). This is the most dramatically separated in wild-type zebrafish similar to the behavior of hoxa2, which is also normally embryos. We asked whether knocking down nrp2a or nrp2b expressed in r4 and in the second cranial NC stream, and is using antisense morpholinos (MOs) either disrupted cranial lost in lzr/pbx4 mutants (Po¨pperl et al., 2000). H.-H. Yu, C.B. Moens / Developmental Biology 280 (2005) 373–385 379

Table 2 Effect of MOs in lzr/pbx4 mutants MO Total Severea Mild Normal wt – 97 0 (0%) 0 (0%) 97 (100%) lzr uninjected 122 122 (100%) 0 (0%) 0 (0%) lzr nrp1a, 6 ng 54 53 (99%) 1 (1%) 0 (0%) lzr nrp2a, 3 ng 72 14 (20%) 22 (31%) 36 (49%) lzr nrp2b, 3 ng 48 12 (25%) 20 (42%) 16 (33%) lzr nrp2a and 2b, 3 ng 80 4 (5%) 24 (30%) 52 (65%) lzr sema3Aa, 7.5 ng 42 42 (100%) 0 (0%) 0 (0%) lzr sema3Ab, 7.5 ng 16 16 (100%) 0 (0%) 0 (0%) lzr sema3D, 7.5 ng 68 68 (100%) 0 (0%) 0 (0%) lzr sema3Fa, 7.5 ng 90 52 (58%) 27 (30%) 11 (12%) lzr sema3Ga, 7.5 ng 82 48 (59%) 24 (29%) 10 (12%) lzr sema3Gb, 7.5 ng 78 36 (46%) 26 (33%) 16 (21%) lzr sema3Fa and Ga, 6.7 ng 42 22 (52%) 12 (27%) 8 (20%) lzr vegf165, 7.5 ng 78 74 (95%) 4 (5%) 0 (0%) No cranial NC defect was observed in wild-type embryos injected with 3–7.5 ng of MOs used. a bSevereQ is equivalent to the cranial NC fusion phenotype of lzr/pbx4 mutants.

Zebrafish sema3C is expressed in part of the otic ing that these semas function as nrp2 ligands to influence epithelium in both wild-type and lzr/pbx4 mutant embryos cranial NCC migration into the head periphery (Table 2). (arrowhead in Figs. 5E and 5F). sema3Fa and sema3Ga are When a sema3Gb MO was injected into lzr/pbx4 mutants, we expressed in head periphery corresponding to the cranial observed suppression of the lzr/pbx4 cranial NC fusion NC-free zones in wild-type embryos (asterisks in Figs. 5I phenotype (Fig. 5X; Table 2). Similarly, MO knockdown of and 5M). Cross-sections through the NC-free zone between Sema3Fa and Sema3Ga also suppress the lzr/pbx4 cranial the second and third arches (red asterisks in Figs. 5I and 5J) NCC phenotype, albeit less effectively (Figs. 5V and 5W; shows that this expression is both endodermal and Table 2). On the contrary, injection of MOs targeting mesodermal (arrows and arrowheads, respectively, in Figs. Sema3Aa and Sema3Ab did not suppress the cranial NC 5T and 5U). sema3Fa and sema3Ga expression is expanded fusion of lzr/pbx4 mutants, consistent with previous evidence between the NC-free domains in the head periphery of lzr/ suggesting that Sema3A are ligands for Nrp1 (Chen et al., pbx4 mutants (arrowheads in Figs. 5J and 5N). Zebrafish 1997; He and Tessier-Lavigne, 1997; Kolodkin et al., 1997). sema3Fb is not detectable in the head periphery of either MO knockdown of Sema3D also did not affect the cranial NC wild-type or lzr/pbx4 mutant embryos (Figs. 5K and 5L). fusion of lzr/pbx4 mutants, as we expected since sema3D Zebrafish sema3Gb is weakly expressed in the tissues expression is already extinguished in lzr/pbx4 mutants (Figs. surrounding the cranial NCCs in wild-type embryos (white 5G–5H). Finally, knocking down VEGF165 in lzr/pbx4 arrows in Fig. 5O) and its expression is broadly expanded in mutants did not suppress the lzr/pbx4 cranial NC phenotype, the region through which cranial NCCs are migrating in lzr/ suggesting that zebrafish vegf165 does not contribute to pbx4 mutants (arrowhead in Fig. 5P). Like sema3Fa, controlling cranial NCC migration. Taken together, the vegf165 is expressed in the cranial NC-free zones in wild- expression of nrp2 and sema3 genes and the effects of type embryos (asterisks in Fig. 5Q; also see in Liang et al., knocking them down in the lzr/pbx4 background suggest a 1998) and its expression is expanded in the head periphery model (Fig. 7) in which sema3Fa, sema3Ga, and sema3Gb in lzr/pbx4 mutants (arrowhead in Fig. 5S). are all expressed in the cranial NC-free zones of the head If a given sema3 gene or vegf165 functions as a ligand for periphery (endoderm and/or mesoderm), where they may nrp2a or nrp2b during cranial NCC migration, we predict function with other cues to repel r4-derived cranial NCCs, that knocking it down will suppress the lzr/pbx4 cranial NC resulting in three distinct cranial NC streams. In lzr/pbx4 fusion phenotype in a manner similar to knockdown of Nrp2a mutants, expression of these ligands is expanded homoge- and Nrp2b. We designed antisense MOs to each of sema3Aa, neously throughout the head periphery, causing cranial NCCs sema3Ab, sema3D, sema3Fa, sema3Ga, sema3Gb, and to migrate aberrantly. MO knockdown of individual ligands vegf165. Since sema3C and sema3Fb are not expressed at or receptors partially restores normal cranial NCC migratory the right time or place to affect cranial NCC migration, we did pattern, revealing the presence of an independent redundant not consider them further. We found that injection of any mechanism controlling cranial NCC migration. sema MO alone or in combination did not cause a cranial NC fusion phenotype in wild-type embryos, again suggesting Over-expression of sema3Gb results in an abnormal cranial that these genes are not strictly required to sculpt patterns of NCC migration phenotype cranial NCC migration. However, knocking down some class 3 sema genes (sema3Fa, sema3Ga, and sema3Gb) did If the cranial NCC migration defect in lzr/pbx4 mutants is suppress the lzr/pbx4 cranial NC fusion phenotype, indicat- in part due to the homogeneous expression of sema3F and 380 H.-H. Yu, C.B. Moens / Developmental Biology 280 (2005) 373–385

Fig. 5. Class 3 secreted semas are expressed in the cranial NC-free zones. In A–U, sema expression is in blue, and engrailed3, krox20, and dlx2 are in pink or orange. In V–Y, krox20 and dlx2 are in blue. Arrows indicate the second cranial NC stream. (A–B) sema3Aa is expressed in the head periphery between the second and third cranial NC streams (arrowhead), and its expression appears unaffected in lzr/pbx4 mutants. (C–D) sema3Ab is not expressed in the head periphery of wild-type embryos, but its expression is expanded in the head periphery between the second and third cranial NC streams of lzr/pbx4 mutants (arrowhead). (E–F) sema3C is expressed in the OV (arrowhead) in wild-type embryos, and its expression is also unaffected in lzr/pbx4 mutants (G–H) sema3D is expressed in the second cranial NC stream (arrow) in wild-type embryos, but the expression is gone in lzr/pbx4 mutants. sema3Fa (I–J), sema3Ga (M–N), and vegf165 (Q–S) are expressed in the cranial NC-free zones (asterisks) in wild-type embryos and their expression is expanded in the head periphery of the lzr/ pbx4 mutants (arrowhead). (K–L) sema3Fb is not expressed in the head periphery. (O–P) sema3Gb is expressed in cells surrounding the cranial NCCs in the wild-type embryo (white arrows) and its expression is expanded in the head periphery of lzr/pbx4 mutants (arrowhead). (T–U) The expression of sema3Fa (T) and sema3Ga (U) in the endoderm (arrows) and/or mesoderm (arrowheads) of the cranial NC-free zones in wild-type embryos is examined in cross sections, whose positions are indicated as red asterisks in I and M. (V–Y) MO knockdown of Sema3Fa (V), Sema3Ga (W), Sema3Gb (X), or Sema3Fa/3Ga (Y) in lzr/ pbx4 mutants partially suppresses the cranial neural crest fusion phenotype, resulting in the re-appearance of a cranial NC-free zone (asterisks) in V–Y. Scale bar: 40 Am in A–S and V–Y; 20 Am in T–U. sema3G genes in the head periphery, then over-expression of Therefore, the sema3Gb mRNA and MOs against zebrafish these genes in wild-type embryos should recapitulate the lzr/ sema3Fa, sema3Ga, and sema3Gb were injected into wild- pbx4 cranial NCC migration defect. A mild cranial NCC type embryos to generate a homogeneous expression of migration defect was occasionally observed when sema3Gb sema3Gb in the absence of endogenous Sema3F/3G in the was over-expressed in wild-type embryos (arrowhead in Fig. head periphery. We note that the injected sema3Gb mRNA is 6B; Table 3). We reasoned that homogeneous sema3Gb not knocked down by the splice blocking MO we used to expression might not be sufficient to override the inhibitory knockdown endogenous Sema3Gb. Surprisingly, we did not effects of endogenous expression of sema3Fa, sema3Ga, and observe severe cranial NCC migration defects in embryos sema3Gb, all of which are affected in lzr/pbx4 mutants. manipulated in this manner (Fig. 6C; Table 3). We then H.-H. Yu, C.B. Moens / Developmental Biology 280 (2005) 373–385 381

could suppress the migration defect caused by over-expression of sema3Gb and knockdown of endogenous Sema3D, Sema3Fa, Sema3Ga, and Sema3Gb (Fig. 6H; Table 3). This result indicates that the effect of injecting sema3Gb mRNA in a cocktail with four different MOs is not a non-specific defect caused by injecting large amounts of mRNA and MOs.

Discussion

Segregation of cranial NCCs into distinct streams within the head periphery is one of the most striking patterning events in vertebrate head morphogenesis. Although a wealth of information exists concerning the cellular interactions and molecules that regulate cell migration and segmental patterning within the hindbrain neuroepithelium, relatively little is known about the mechanisms that pattern the migration of cranial NCCs within the head periphery (Trainor and Krumlauf, 2000). Previous studies have suggested that unidentified inhibitory signals in cranial NC-free zones of the head mesoderm are essential for controlling cranial NCC migration (Farlie et al., 1999; Trainor et al., 2002). By analyzing the zebrafish lzr/pbx4 Fig. 6. Over-expression of sema3Gb in wild-type embryos results in aberrant mutant, in which cranial NCC migration is mis-regulated, cranial NCC migration. Manipulation in each panel is under the wild-type we have identified class 3 sema genes as inhibitory signals genetic background. krox20 and dlx2 are in blue in all panels. Arrows for directing the migration of nrp-expressing cranial NCCs. indicate the second cranial NC stream, while arrowheads indicate cranial NCCs in normally cranial NC-free zones. (A) An uninjection control of a wild-type embryo. (B) A mild cranial NCC migration defect (arrowhead), in A model for the role of sema3 signaling in cranial NCC which a few cranial NCCs were detected in the normally cranial NC-free migration zone, was occasionally observed in sema3Gb over-expressing embryos. (C) Over-expression of sema3Gb combined with knockdown of endogenous We have shown that in wild-type zebrafish embryos Sema3Fa, Sema3Ga, and Sema3Gb causes a similar mild defect. (D) Knockdown of Sema3D in the presence of over-expression of sema3Gb had a nrp2a and nrp2b are expressed in migrating cranial NCCs mild defect. (E-F) Over-expression of sema3Gb in wild-type embryos and that their ligands, sema3Fa, sema3Ga, and sema3Gb injected with Sema3D, Sema3Fa, Sema3Ga, and Sema3Gb MOs caused a are expressed in regions of the head periphery normally free cranial NCC migration phenotype resembling that of lzr/pbx4 mutants. One of cranial NCCs. In lzr/pbx4 mutants, aberrant migration of (arrowhead in F) or both (arrowheads in E) sides of the manipulated embryos cranial NCCs into the normally cranial NC-free zones showed the cranial NC fusion phenotype. (G) Knockdown of endogenous Sema3D, Sema3Fa, Sema3Ga, and Sema3Gb also had a similar mild defect. correlates with a homogeneous spreading of sema3F/G (H) Knockdown of Nrp2b suppresses the cranial NC fusion phenotype expression. MO knockdown of Nrp2a and Nrp2b or of observed in the manipulated embryos of E and F. An aberrantly migratory Sema3Fa, Sema3Ga, and Sema3Gb partially restores the cranial NCC is indicated by arrowhead. Scale bar: 40 Am in A–H. separation of cranial NC streams in lzr/pbx4 mutants. This genetic suppression leads us to a model for the role of speculated that the loss of sema3D expression in the second Sema–Nrp signaling in cranial NCC migration (Fig. 7). In cranial NC stream in lzr/pbx4 mutants is also important for wild-type embryos (Fig. 7A), lzr/pbx4 functions to suppress the cranial NC fusion phenotype. No cranial NC fusion was sema3F/3G expression in the territories that are permissive observed in embryos injected with sema3Gb mRNA and a for cranial NCC migration, resulting in sema3F/3G expres- sema3D MO (Fig. 6D; Table 3). However, when we sion being restricted to the cranial NC-free zones. nrp2- generated embryos over-expressing sema3Gb in a back- expressing cranial NCCs are then repelled from the cranial ground in which the endogenous Sema3D, Sema3Fa, NC-free zones by sema3F/3G expression, and as a result Sema3Ga, and Sema3Gb were all knocked down, a severe migrate in three separate streams into the pharyngeal arches. cranial NC fusion phenotype was observed in 16% (n = 25) of In lzr/pbx4 mutants (Fig. 7B), sema3F/3G expression is the injected embryos and a milder cranial NCC migration expanded throughout the head periphery. Since cranial phenotype was observed in a further 39% (n = 61) of the NCCs are intrinsically migratory, they persist in migrating injected embryos (arrowheads in Figs. 6E and 6F; Table 3). into the pharyngeal arches even though they continue to No cranial NCC migration fusion phenotype was observed in express nrp2 genes, but they migrate homogeneously embryos injected with the four MOs alone (Fig. 6G; Table 3). because they lack a preferred sema3F/3G-negative path. We further found that co-injection of nrp2a or nrp2b MO Knockdown of Nrp2 (Fig. 7C) or Sema3F/3G (Fig. 7D) in 382 H.-H. Yu, C.B. Moens / Developmental Biology 280 (2005) 373–385

Table 3 Effect of over-expression of sema3Gb mRNA mRNAa MOb Total Severe Mild Normal wt – – 97 0 (0%) 0 (0%) 97 (100%) wt sema3Gb – 45 0 (0%) 2 (4%) 43 (96%) wt sema3Gb 3D 51 0 (0%) 4 (8%) 47 (92%) wt sema3Gb 3Fa, 3Ga, 3Gb 119 1 (1%) 14 (12%) 104 (87%) wt sema3Gb 3D, 3Fa, 3Ga, 3Gb 157 25 (16%) 61 (39%) 71 (45%) wt sema3Gb 3D, 3Fa, 3Ga, 3Gb, nrp2a 19 0 (0%) 6 (32%) 13 (68%) wt sema3Gb 3D, 3Fa, 3Ga, 3Gb, nrp2b 27 0 (0%) 9 (30%) 18 (70%) wt sema3Gb 3D, 3Fa, 3Ga 49 0 (0%) 9 (18%) 40 (82%) wt sema3Gb 3D, 3Fa, 3Gb 61 1 (2%) 8 (13%) 52 (85%) wt sema3Gb 3D, 3Ga, 3Gb 68 0 (0%) 10 (15%) 58 (85%) wt – 3D, 3Fa, 3Ga, 3Gb 56 0 (0%) 7 (13%) 49 (87%) a The injection of sema3Gb mRNA is 150–200 pg. b 3D, 3Fa, 3Ga, 3Gb, nrp2a, and nrp2b are 5ng MO of sema3D, sema3Fa, sema3Ga, sema3Gb, nrp2a, and nrp2b respectively. lzr/pbx4 mutants partially restores normal cranial NCC mechanisms controlling the process. Although the suppres- migratory pattern, confirming a role for these genes in this sion of the lzr/pbx4 cranial NC phenotype by knockdown of process, but also uncovering redundant pathways for Nrp2 or Sema3F/3G clearly implicates this pathway in patterning cranial NCC migration. patterning cranial NCC migration, MO knockdown of both If Sema3F/3G signals are repulsive for migrating cranial Nrp2a and Nrp2b or of Sema3Fa, Sema3Ga, and Sema3Gb NCCs, a prediction is that when their expression is does not itself cause a cranial NCC migration phenotype, expanded homogeneously as we observed in lzr/pbx4 indicating that other mechanisms compensate for the mutants, cranial NCC migration would be prevented absence of Sema3F/3G-Nrp2 signaling. Furthermore, altogether. However the loss of directional preference in knockdown of Nrp2 or Sema3F/3G in lzr/pbx4 mutants migratory behavior due to the global expression of Sema is restores the separation of cranial NC streams, revealing an not unprecedented. When rat embryonic cortical neurons are underlying mechanism that can pattern cranial NCC plated onto cortical slices, they extend axons directionally migration when Sema3F/3G-Nrp2 signaling is reduced. toward the ventricular surface in response to a repulsive This underlying patterning mechanism may be established Sema3A signal in the mantle layer. However when the independent of lzr/pbx4, but this is not necessarily the case, cortical slices are pre-incubated with recombinant Sema3A, since we have shown that both maternal lzr/pbx4 and plated neurons do not fail to extend axons, but rather extend another pbx, pbx2, are still active in lzr/pbx4 mutants them randomly (Polleux et al., 1998). (Waskiewicz et al., 2002). It is possible that both the sema3F/3G pattern and the other redundant pattern are both Redundant mechanisms underlying the segregation of dependent on pbx genes, but that the sema3F/3G pattern is cranial NCCs more sensitive to reduced pbx function. Our model specifically predicts that over-expression of Our model for the role of Sema3F/3G-Nrp2 signaling in sema3F and/or sema3G will recapitulate the cranial NCC cranial NCC migration specifically invokes redundant migration phenotype of lzr/pbx4 mutants. We found this to

Fig. 7. Model of Sema3F/3G-Nrp2 function in cranial NCC migration. In wild-type embryos (A), nrp2-expressing cranial NCCs (blue; I, II, and III indicate cranial NC streams migrating into the first, second, and third arches, respectively) avoid sema3F and sema3G-expressing cells (indicated by x’s) as they migrate into the head periphery, resulting in three distinct cranial NC streams separated by NC-free zones. (B) In lzr/pbx4 mutants, expansion of sema3F and sema3G-expression in the head periphery eliminates a bpreferred migratory pathQ, resulting in fusion of cranial NC streams. Knockdown of Nrp2 (C) or of Sema3F and Sema3G (D) restores normal cranial NCC migration in lzr/pbx4 mutants due to the predominating effect of an unidentified redundant mechanism(s) which is unaffected or less affected in lzr/pbx4 mutants. H.-H. Yu, C.B. Moens / Developmental Biology 280 (2005) 373–385 383 be true, but only when sema3D and the other sema3F/3G in the surface ectoderm and mesodermal core containing genes that are normally expressed in the cranial NC-free muscle precursor cells, respectively (Chilton and Guthrie, zones were knocked down. Even then, the severe cranial NC 2003). sema3F and nrp2 knockout mice have been fusion phenotype similar to that of lzr/pbx4 mutants was generated, but the phenotypic analysis of these mice focused only observed in 16% of injected embryos. This may be due on the nervous system (Chen et al., 2000; Cloutier et al., to poor perdurance of the injected mRNA or its protein 2000; Giger et al., 2002; Sahay et al., 2003). Sema3F-Nrp2 product, or because of partially redundant pathways that signaling is essential for the proper organization of specific contribute to patterning cranial NCC migration are affected cranial nerve projections, for wiring and patterning of the in lzr/pbx4 mutants (see below) but not in wild-type accessory olfactory system and for the establishment of embryos injected with sema3Gb mRNA. For example, limbic tracts in the rostral forebrain, midbrain, and hippo- previous work has implicated ephrinB2a as a repulsive campus. Since cranial NCC migration has not been carefully signal that prevents the mixing of cranial NC streams (Smith analyzed in these mutant mice, we do not know the role of et al., 1997). Expression of ephrinB2a, which is normally Sema3F-Nrp2 signaling in cranial NCC migration in mice. restricted to second arch cranial NCCs, is lost in lzr/pbx4 Moreover, we expect such effects to be subtle, given that in mutants (H.-H.Y and C.B.M., unpublished) and this loss zebrafish we uncovered a role for this pathway in the lzr/ may contribute to the very robust cranial NC fusion pbx4 mutant background and not in Sema3F/3G or Nrp2 phenotype in lzr/pbx4 mutants. morphants themselves. The role of sema3D has not been reported in cranial NCC migration even though sema3D is strongly expressed Regulation of inhibitory signals in cranial NC-free zones in cranial NCCs of the second pharyngeal arch. Our fate- mapping experiment showed that this is the cranial NC An important question is how lzr/pbx4 regulates sema3 stream that is most clearly affected in lzr/pbx4 mutants (Fig. expression in the head periphery. Our genetic data do not 2E), and sema3D expression in the second stream, like allow us to infer whether this regulation is direct or ephrinB2a expression, is strongly reduced in lzr/pbx4 indirect. However our genetic mosaic analysis indicates mutants (Figs. 5G–5H). As mentioned above, knockdown that lzr/pbx4 function is non-autonomous with the respect of Sema3D is required to generate the severe cranial NCC to cranial NCC migration since wild-type donor cells migration defect caused by sema3Gb over-expression (Figs. migrate aberrantly in lzr/pbx4 mutant hosts. Lzr/Pbx4 is a 6E–6F; Table 3). Although these data suggest that sema3D Hox co-factor that functions cell-autonomously within the may function to maintain the integrity of the second cranial hindbrain to specify rhombomere identities (Po¨pperl et al., NC stream, no cranial NCC migration defect was observed 2000; Waskiewicz et al., 2002). Studies have suggested in sema3D MO-injected embryos. It is nevertheless possible that factors from the neuroepithelium and/or the surface that sema3D functions redundantly with the sema3Fa/3Ga/ ectoderm are important for the production of inhibitory 3Gb mechanism in the head periphery, either as an signals in cranial NC-free zones. Aberrant cranial NCC inhibitory signal to repel nrp2-expressing cranial NCCs of migration and fusion of cranial NC streams were observed the third cranial NC stream, or as an attractant to maintain after the ablation of r3 neuroepithelium and/or r3/r5 the cohesion of cranial NCCs migrating into the second surface ectoderm (Golding et al., 2002, 2004). The arch. Previous work has shown that Sema3D can bind to receptor tyrosine kinase erbB4, expressed in r3 neuro- Nrp1 receptors (Feiner et al., 1997); however, nrp1 genes epithelium, is implicated in the regulation of inhibitory are not expressed in the second and third cranial NC streams signals in the r3-adjacent cranial NC-free zone since a during cranial NCC migration. In addition, Sema3D has fusion of cranial NC streams, a phenotype similar to the been shown to attract anterior commissural axons of ablation of r3 neuroepithelium, was observed in erbB4 zebrafish telencephalic neurons through Nrp1a and Nrp2b deficient mice (Golding et al., 2000, 2002, 2004). (Wolman et al., 2004). However, it remains to be determined Furthermore, sema3A is expressed in r3 and r5 and has whether Nrp2 receptors can transduce Sema3D signals in been implicated in the regulation of cranial NC migration cranial NCC migration. (Eickholt et al., 1999), and we observed r3/r5-restricted expression of sema3Ab and sema3Fb in the zebrafish that Sema3F-Nrp2 signaling in cranial NCC migration in other is disrupted in lzr/pbx4 mutants. One possibility is that lzr/ vertebrates pbx4 functions indirectly through erbB4, sema3A, 3F or other factors normally expressed within the hindbrain to Although Sema3A-Nrp1 signaling has been implicated in establish cranial NC-free zones in the head periphery. We cranial NCC migration in chick (Eickholt et al., 1999), a role note, however, that knockdown of Sema3Ab does not for Sema3F-Nrp2 signaling in cranial NCC migration has cause a cranial NCC migration phenotype (Table 2)as not previously been demonstrated. While the expression of would be predicted if it alone was responsible for sema3F and nrp2 in tetrapod vertebrates during cranial NC preventing NCC migration lateral to r3 and r5. Further- migration has not been described, sema3F and nrp2 are more, a model in which sema3A, sema3F, or other factors later expressed in pharyngeal arches in the chick, including expressed within the CNS create cranial NC-free zones 384 H.-H. Yu, C.B. Moens / Developmental Biology 280 (2005) 373–385 directly, by repelling cranial NCCs from r3 and/or r5 prior References to their emigration from the hindbrain neuroepithelium (Eickholt et al., 1999), is unlikely since (1) in lzr/pbx4 Akimenko, M.A., Ekker, M., Wegner, J., Lin, W., Westerfield, M., 1994. Combinatorial expression of three zebrafish genes related to distal- mutants the initial separation of cranial NC streams is less: part of a homeobox gene code for the head. J. Neurosci. 14, relatively normal (Fig. 1D), but they fuse after migrating 3475–3486. into the head periphery (Fig. 1F); (2) our genetic mosaic Birgbauer, E., Sechrist, J., Bronner-Fraser, M., Fraser, S., 1995. Rhombo- analysis showed that the transplantation of significant meric origin and rostrocaudal reassortment of neural crest cells revealed amount of wild-type donor cells into the dorsal hindbrain by intravital microscopy. Development 121, 935–945. Bovenkamp, D.E., Goishi, K., Bahary, N., Davidson, A.J., Zhou, Y., of lzr/pbx4 mutants is unable to rescue the cranial NCC Becker, T., Becker, C.G., Zon, L.I., Klagsbrun, M., 2004. Expression migration defect (data not shown). and mapping of duplicate neuropilin-1 and neuropilin-2 genes in A second possibility is that lzr/pbx4 functions in the head developing zebrafish. Gene Expr. Patterns 4, 361–370. periphery to control cranial NCC migration. Hox genes have Chen, H., Chedotal, A., He, Z., Goodman, C.S., Tessier-Lavigne, M., 1997. been reported to be expressed in the head mesoderm, Neuropilin-2, a novel member of the neuropilin family, is a high affinity receptor for the semaphorins Sema E and Sema IV but not Sema III. particularly Hox paralog group 1 genes which are estab- Neuron 19, 547–559. lished early in development with an anterior limit at the Chen, H., Bagri, A., Zupicich, J.A., Zou, Y., Stoeckli, E., Pleasure, S.J., level of r4 (Frohman et al., 1990). It is possible that Lzr/ Lowenstein, D.H., Skarnes, W.C., Chedotal, A., Tessier-Lavigne, M., Pbx4 interacts with Hox-1 proteins in the head mesoderm to 2000. Neuropilin-2 regulates the development of selective cranial and establish a boundary(s) that is later interpreted in the sensory nerves and hippocampal mossy fiber projections. Neuron 25, 43–56. regulation of sema3 genes. Alternatively, Lzr/Pbx4 could Chilton, J.K., Guthrie, S., 2003. Cranial expression of class 3 secreted interact with other, non-Hox homeodomain proteins to limit semaphorins and their neuropilin receptors. Dev. Dyn. 228, 726–733. sema3 expression to the cranial NC-free zones. We note that Cloutier, J.F., Giger, R.J., Koentges, G., Dulac, C., Kolodkin, A.L., Ginty, we have been unable to address the singular importance of D.D., 2000. Neuropilin-2 mediates axonal fasciculation, zonal segre- head mesoderm in the genesis of the lzr/pbx4 cranial NCC gation, but not axonal convergence, of primary accessory olfactory neurons. Neuron 33, 877–892. migration phenotype by genetic mosaic analysis due to the Couly, G., Creuzet, S., Bennaceur, S., Vincent, C., Le Douarin, N.M., 2002. difficulty of transplanting sufficient head mesoderm pre- Interactions between Hox-negative cephalic neural crest cells and the cursors to effect rescue. foregut endoderm in patterning the facial skeleton in the vertebrate head. Development 129, 1061–1073. Driscoll, D.A., Spinner, N.B., Budarf, M.L., McDonald-McGinn, D.M., Zackai, E.H., Goldberg, R.B., Shprintzen, R.J., Saal, H.M., Zonana, J., Conclusion Jones, M.C., 1992. Deletions and microdeletions of 22q11.2 in velo- cardio-facial syndrome. Am. J. Med. Genet. 44, 261–268. The stereotyped migration of cranial NCCs into the Eickholt, B.J., Mackenzie, S.L., Graham, A., Walsh, F.S., Doherty, P., 1999. segmented pharyngeal arches involves both intrinsic pat- Evidence for collapsin-1 functioning in the control of neural crest terning of the cranial NCCs themselves and external signals migration in both trunk and hindbrain regions. Development 126, 2181–2189. that direct their migration. By looking for genes that interact Ellies, D.L., Tucker, A.S., Lumsden, A., 2002. Apoptosis of premigratory with lzr/pbx4 to control cranial NCC migration in zebrafish, neural crest cells in rhombomeres 3 and 5: consequences for patterning we have identified molecules of the Sema3F and a new of the branchial region. Dev. Biol. 251, 118–128. Sema3G subclass that are expressed in the head periphery in Farlie, P.G., Kerr, R., Thomas, P., Symes, T., Minichiello, J., Hearn, C.J., cranial NC-free zones and which repel cranial NCCs Newgreen, D., 1999. A paraxial exclusion zone creates patterned cranial neural crest cell outgrowth adjacent to rhombomeres 3 and 5. Dev. Biol. expressing their receptors Nrp2a and Nrp2b. Like other 213, 70–84. genes identified in suppressor and enhancer screens, these Feiner, L., Koppel, A.M., Kobayashi, H., Raper, J.A., 1997. Secreted chick genes would have not been identified through normal semaphorins bind recombinant neuropilin with similar affinities but screens for mutations affecting cranial NCC migration since bind different subsets of neurons in situ. Neuron 19, 539–545. their loss is compensated for by as-yet unidentified re- Feiner, L., Webber, A.L., Brown, C.B., Lu, M.M., Jia, L., Feinstein, P., Mombaerts, P., Epstein, J.A., Raper, J.A., 2001. Targeted disruption of dundant mechanisms. semaphorin 3C leads to persistent truncus arteriosus and aortic arch interruption. Development 128, 3061–3070. Frohman, M.A., Boyle, M., Martin, G.R., 1990. Isolation of the mouse Acknowledgments Hox-2.9 gene; analysis of embryonic expression suggests that positional information along the anterior-posterior axis is specified by mesoderm. Development 110, 589–607. We thank Ruowen Ge for and Jeffrey Miller for vegf165 Giger, R.J., Cloutier, J.F., Sahay, A., Prinjha, R.K., Levengood, D.V., and pCS2+-GFP-LT-XLT cDNAs, respectively. We thank Moore, S.E., Pickering, S., Simmons, D., Rastan, S., Walsh, F.S., the members of the Moens lab for their thoughtful Kolodkin, A.L., Ginty, D.D., Geppert, M., 2002. Neuropilin-2 is comments on the manuscript. This work was supported by required in vivo for selective axon guidance responses to secreted NIH grant HD37909 and NSF grant 1BN-9816805, NRSA semaphorins. Neuron 25, 29–41. Golding, J.P., Trainor, P., Krumlauf, R., Gassmann, M., 2000. Defects in fellowship 5 F32 HD043534-03 (H.-H.Y.). C.B.M. is an pathfinding by cranial neural crest cells in mice lacking the neuregulin assistant investigator with the Howard Hughes Medical receptor ErbB4. Nat. Cell Biol. 2, 103–109. Institute. Golding, J.P., Dixon, M., Gassmann, M., 2002. Cues from neuroepithelium H.-H. Yu, C.B. Moens / Developmental Biology 280 (2005) 373–385 385

and surface ectoderm maintain neural crest-free regions within cranial that modulate axon guidance and angiogenesis. Trends Cardiovasc. mesenchyme of the developing chick. Development 129, 1095–1105. Med. 12, 13–19. Golding, J.P., Sobieszczuk, D., Dixon, M., Coles, E., Christiansen, J., Oxtoby, E., Jowett, T., 1993. Cloning of the zebrafish krox-20 gene Wilkinson, D., Gassmann, M., 2004. Roles of erbB4, rhombomere- (krx-20) and its expression during hindbrain development. Nucleic specific, and rhombomere-independent cues in maintaining neural crest- Acids Res. 21, 1087–1095. free zones in the embryonic head. Dev Biol. 266, 361–372. Pasterkamp, R.J., Kolodkin, A.L., 2003. Semaphorin junction: making Graham, A., Smith, A., 2001. Patterning the pharyngeal arches. BioEssays tracks toward neural connectivity. Curr. Opin. Neurobiol. 13, 79–89. 23, 54–61. Peyrieras, N., Strahle, U., Rosa, F., 1998. Conversion of zebrafish Graham, A., Heyman, I., Lumsden, A., 1993. Even-numbered rhombo- blastomeres to an endodermal fate by TGF-beta-related signaling. Curr. meres control the apoptotic elimination of neural crest cells odd- Biol. 8, 783–786. numbered rhombomeres in the chick hindbrain. Development 119, Picker, A., Scholpp, S., Bohli, H., Takeda, H., Brand, M., 2002. A novel 233–245. positive transcriptional feedback loop in midbrain-hindbrain boundary Gritsman, K., Talbot, W.S., Schier, A.F., 2000. Nodal signaling patterns the development is revealed through analysis of the zebrafish pax2.1 organizer. Development 127, 921–932. promoter in transgenic lines. Development 129, 3227–3239. Halloran, M.C., Severance, S.M., Yee, C.S., Gemza, D.L., Raper, J.A., Polleux, F., Giger, R.J., Ginty, D.D., Kolodkin, A.L., Ghosh, A., 1998. Kuwada, J.Y., 1999. Analysis of a Zebrafish semaphorin reveals Patterning of cortical efferent projections by semaphorin-neuropilin potential functions in vivo. Dev. Dyn. 214, 13–25. interactions. Science 282, 1904–1906. Hauptmann, G., Gerster, T., 1994. Two-color whole-mount in situ hybri- Pfpperl, H., Rikhof, H., Cheng, H., Haffter, P., Kimmel, C.B., Moens, C.B., dization to vertebrate and Drosophila embryos. Trends Genet. 10, 266. 2000. Lazarus is a novel pbx gene that globally mediates hox gene He, Z., Tessier-Lavigne, M., 1997. Neuropilin is a receptor for the axonal function in zebrafish. Mol. Cell 6, 255–267. chemorepellent semaphorin III. Cell 90, 739–751. Reifers, F., Bohli, H., Walsh, E.C., Crossley, P.H., Stainier, D.Y., Brand, M., Holland, P.W., Williams, N.A., 1990. Conservation of engrailed-like 1998. Fgf8 is mutated in zebrafish acerebellar (ace) mutants and is homeobox sequences during vertebrate evolution. FEBS Lett. 277, required for maintenance of midbrain-hindbrain boundary development 250–252. and somitogenesis. Development 125, 238123–238195. Kimmel, C.B., Warga, R.M., Schilling, T.F., 1990. Origin and organization Roos, M., Schachner, M., Bernhardt, R.R., 1999. Zebrafish semaphorin Z1b of the zebrafish fate map. Development 108, 581–594. inhibits growing motor axons in vivo. Mech. Dev. 87, 103–117. Kimmel, C.B., Ballard, W.W., Kimmel, S.R., Ullmann, B., Schilling, T.F., Rubinstein, A.L., Lee, D., Luo, R., Henion, P.D., Halpern, M.E., 2000. 1995. Stages of embryonic development of the zebrafish. Dev. Dyn. Genes dependent on zebrafish cyclops function identified by AFLP 203, 253–310. differential gene expression screen. Genesis 26, 86–97. Kimmel, C.B., Miller, C.T., Moens, C.B., 2001. Specification and Sahay, A., Molliver, M.E., Ginty, D.D., Kolodkin, A.L., 2003. Semaphorin morphogenesis of the zebrafish larval head skeleton. Dev. Biol. 233, 3F is critical for development of limbic system circuitry and is required 239–257. in neurons for selective CNS axon guidance events. J. Neurosci. 23, Kolodkin, A.L., Levengood, D.V., Rowe, E.G., Tai, U.-T., Giger, R.J., 6671–6680. Ginty, D.D., 1997. Neuropilin is a semaphorin III receptor. Cell 90, Schilling, T.F, Prince, V., Ingham, P.W., 2001. Plasticity in zebrafish hox 753–762. expression in the hindbrain and cranial neural crest. Dev. Biol. 231, Kontges, G., Lumsden, A., 1996. Rhombencephalic neural crest segmenta- 201–216. tion is preserved throughout craniofacial ontogeny. Development 122, Sechrist, J., Serbedzija, G.N., Scherson, T., Fraser, S.E., Bronner-Fraser, 3229–3242. M., 1993. Segmental migration of the hindbrain neural crest does not Kulesa, P.M., Fraser, S.E., 1998. Neural crest cell dynamics revealed by arise from its segmental generation. Development 118, 693–701. time-lapse video microscopy of whole embryo chick explant cultures. Smith, A., Robinson, V., Patel, K., Wilkinson, D.G., 1997. The EphA4 and Dev. Biol. 204, 327–344. EphB1 receptor tyrosine kinases and ephrin-B2 ligand regulate targeted Lee, P., Goishi, K., Davidson, A.J., Mannix, R., Zon, L., Klagsbrun, M., migration of branchial neural crest cells. Curr. Biol. 7, 561–570. 2002. Neuropilin-1 is required for vascular development and is a Trainor, P., Krumlauf, R., 2000. Plasticity in mouse neural crest cells mediator of VEGF-dependent angiogenesis in zebrafish. Proc. Natl. reveals a new patterning role for cranial mesoderm. Nat. Cell Biol. 2, Acad. Sci. U. S. A. 99, 10470–10475. 96–102. Liang, D., Xu, X., Chin, A.J., Balasubramaniyan, N.V., Teo, M.A., Lam, Trainor, P.A., Sobieszczuk, D., Wilkinson, D., Krumlauf, R., 2002. T.J., Weinberg, E.S., Ge, R., 1998. Cloning and characterization of Signaling between the hindbrain and paraxial tissues dictates neural vascular endothelial growth factor (VEGF) from zebrafish, Danio rerio. crest migration pathways. Development 129, 433–442. Biochim. Biophys. Acta 1397, 14–20. Waskiewicz, A.J., Rikhof, H.A., Moens, C.B., 2002. Eliminating zebrafish Lumsden, A., Krumlauf, R., 1996. Patterning the vertebrate neuraxis. pbx proteins reveals a hindbrain ground state. Dev. Cell 3, 723–733. Science 274, 1109–1115. Westerfield, M., 1994. The Zebrafish Book, 2.1 ed. University of Oregon Lumsden, A., Sprawson, N., Graham, A., 1991. Segmental origin and Press, Oregon. migration of neural crest cells in the hindbrain region of the chick Wilkie, A.O., Morriss-Kay, G.M., 2001. Genetics of craniofacial develop- embryo. Development 113, 1281–1291. ment and malformation. Nat. Rev. Genet. 2, 458–468. Mann, R.S., Chan, S.K., 1996. Extra specificity from extradenticle: the Wilson, D.I., Burn, J., Scambler, P., Goodship, J., 1993. DiGeorge partnership between HOX and PBX/EXD homeodomain proteins. syndrome: part of CATCH 22. J. Med. Genet. 30, 852–856. Trends Genet. 12, 258–262. Wolman, M.A., Liu, Y., Tawarayama, H., Shoji, W., Halloran, M.C., 2004. Marin, O., Yaron, A., Bagri, A., Tessier-Lavigne, M., Rubenstein, J.L., Repulsion and attraction of axons by semaphorin3D are mediated by 2001. Sorting of striatal and cortical interneurons regulated by different neuropilins in vivo. J. Neurosci. 24, 8423–8435. semaphorin-neuropilin interactions. Science 293, 872–875. Woo, K., Fraser, S.E., 1995. Order and coherence in the fate map of the Martyn, U., Schulte-Merker, S., 2004. Zebrafish neuropilins are differ- zebrafish nervous system. Development 121, 2595–2609. entially expressed and interact with vascular endothelial growth factor Yee, C.S., Chandrasekhar, A., Halloran, M.C., Shoji, W., Warren, J.T., during embryonic vascular development. Dev. Dyn. 231, 33–42. Kuwada, J.Y., 1999. Molecular cloning, expression, and activity of Moens, C.B., Fritz, A., 1999. Techniques in neural development. Methods zebrafish semaphorin Z1a. Brain Res. Bull. 48, 581–593. Cell Biol. 59, 253–272. Yu, H.H., Houart, C., Moens, C.B., 2004. Cloning and embryonic Neufeld, G., Cohen, T., Shraga, N., Lange, T., Kessler, O., Herzog, Y., expression of zebrafish neuropilin genes. Gene Expr. Patterns 4, 2002. The neuropilins: multifunctional semaphorin and VEGF receptors 371–378.