The Anterior/Posterior Polarity of Somites Is Disrupted in Paraxis

Developmental Biology 229, 176–187 (2001) doi:10.1006/dbio.2000.9969, available online at http://www.idealibrary.com on

View metadata, citation and similar papers at core.ac.uk brought to you by CORE The Anterior/Posterior Polarity of Somites Is provided by Elsevier - Publisher Connector Disrupted in Paraxis-Deﬁcient Mice

Joshua Johnson,* Jerry Rhee,*,† Sarah M. Parsons,* Doris Brown,‡,1 Eric N. Olson,‡ and Alan Rawls*,†,2 *Department of Biology, Arizona State University, Tempe, Arizona 85287-1501; and †Molecular and Cellular Biology Graduate Program, ‡Department of Molecular Biology, Hamon Center for Basic Cancer Research, University of Texas Southwestern Medical Center, Dallas, Texas

Establishing the anterior/posterior (A/P) boundary of individual somites is important for setting up the segmental body plan of all vertebrates. Resegmentation of adjacent sclerotomes to form the vertebrae and selective migration of neural crest cells during the formation of the dorsal root ganglia and peripheral nerves occur in response to differential expression of genes in the anterior and posterior halves of the somite. Recent evidence indicates that the A/P axis is established at the anterior end of the presomitic mesoderm prior to overt somitogenesis in response to both Mesp2 and Notch signaling. Here, we report that mice deﬁcient for paraxis, a gene required for somite epithelialization, also display defects in the axial skeleton and peripheral nerves that are consistent with a failure in A/P patterning. Expression of Mesp2 and genes in the Notch pathway were not altered in the presomitic mesoderm of paraxis؊/؊ embryos. Furthermore, downstream targets of Notch activation in the presomitic mesoderm, including EphA4, were transcribed normally, indicating that paraxis was not required for Notch signaling. However, genes that were normally restricted to the posterior half of somites were present in a diffuse pattern in the paraxis؊/؊ embryos, suggesting a loss of A/P polarity. Collectively, these data indicate a role for paraxis in maintaining somite polarity that is independent of Notch signaling. © 2001 Academic Press Key Words: somites; paraxis; polarity; ephrin-B2; Notch1, Mesp2.

BACKGROUND crest cells during the development of peripheral nerves is guided by the same A/P polarity. Delineating the genetic The establishment of boundaries between groups of cells pathway that controls the formation and maintenance of is a strategy employed to derive distinct cell lineages from this A/P boundary is fundamentally important to our un- cells of a common embryonic ancestry. During somitogen- derstanding of how the patterning of the embryonic body esis, the primary segmental body plan of vertebrates is plan is achieved. organized, including an anterior/posterior (A/P) axis within Somites form through the segmentation of paraxial me- individual somites. The A/P polarity setup within indi- soderm from the anterior end of the presomitic mesoderm. vidual somites during embryogenesis is an important para- Concurrent with segmentation, the mesenchymal cells digm for the formation of cellular boundaries. The proper undergo a transition to an epithelial-like ball that contains formation of individual vertebrae and ribs, which arise from distinct apical and lumenal surfaces (reviewed in Christ and the somites, is dependent on the specification of cells to Ordahl, 1995). Through a stepwise epithelial to mesenchy- distinct anterior and posterior fates. Migration of neural mal transition, each somite divides into three lineage- specific compartments, the dermomyotome, sclerotome, and myotome, which are the anlagen for the dermis, skel- 1 Current address: Department of Anatomy, University of Cali- etal muscle, and axial skeleton, respectively. Formation of fornia, San Francisco, CA. 2 To whom correspondence should be addressed at Department the vertebral column is initiated by the migration of cells of Biology, Arizona State University, Life Sciences C, Room 544, from the ventral medial region of the sclerotome toward the Tempe, AZ 85287-1501. Fax: (480) 965-2519. E-mail: jrawls@ notochord. These cells then differentiate into cartilage and imap4.asu.edu. form the model for endochondral ossification of the verte-

0012-1606/01 $35.00 Copyright © 2001 by Academic Press 176 All rights of reproduction in any form reserved. Paraxis Regulation of Somite A/P Polarity 177 bral body. Similarly, cells from the dorsal medial region of Paraxis is a member of the bHLH transcription factor the sclerotome migrate medially to form the neural arches. superfamily (Burgess et al., 1996) and is necessary for newly The vertebrae and intervertebral muscles maintain the formed somites to undergo a morphological transition from segmental pattern established during somite formation. mesenchyme to epithelium. Segmentation was not dis- However, the vertebral body is offset by a half-segment turbed in paraxisϪ/Ϫ embryos, indicating that epithelializa- compared to muscle. This topographical shift is the result tion is controlled by a distinct regulatory pathway (Burgess of resegmentation of the sclerotome along the A/P bound- et al., 1996). Paraxis is expressed in the anterior two-thirds ary (Fig. 1; reviewed by Brand-Saberi and Christ, 2000). The of the presomitic mesoderm and throughout the epithelial posterior and anterior halves of adjacent somites reassociate somite (Burgess et al., 1995; Blanar et al., 1995; Sosic et al., to form the vertebral body (Bagnall et al., 1988, 1992). In 1997; Barnes et al., 1997; Shanmugalingam and Wilson, contrast, the neural arches that give rise to the pedicles and 1998). The axial skeleton of neonatal paraxisϪ/Ϫ mice ex- laminae are derived only from cells of the posterior scle- hibits caudal agenesis, random A/P fusions along the spinal rotome and therefore do not resegment (Goldstein and column, and fused ribs that fail to attach to the vertebral Kalcheim, 1992; Huang et al., 1996). The peripheral nervous column. This suggests that paraxis plays a role in chondro- system also possesses a metameric pattern that is imposed genesis, either in specification of cells fated to enter the by the A/P polarity of individual somites. The sensory and chondrogenic lineage or in the establishment of the pattern sympathetic ganglia are derived from neural crest cells that of cartilage formation. migrate ventrally from the dorsal neural tube into the Here, we examine the role of paraxis in regulating A/P anterior half of the sclerotome (Rickman et al., 1985; patterning of somites during embryogenesis. In paraxis-null Bronner-Fraser, 1986; Loring and Erickson, 1987; Teillet et embryos, both cartilaginous precursors of the axial skeleton al., 1987). and peripheral nerves fail to segment. This indicated a role A/P polarity is established in the presomitic mesoderm for paraxis in regulating A/P polarity of the newly formed prior to overt segmentation, through the specification of somite. Analysis of paraxisϪ/Ϫ embryos revealed normal cells to an anterior or posterior fate (Bronner-Fraser and expression and activity of genes in the Notch signaling Stern, 1991). Genetically, specification induces a stable pathway in the presomitic mesoderm. In contrast, polarized change in gene expression in domains of the presomitic gene expression of Notch1, Dll1, and ephrin-B2 was reduced mesoderm that prepattern the anterior and posterior halves and diffuse in the paraxis-null somites. This predicts a role of the next two somites. This A/P polarity is maintained in for paraxis in establishing or maintaining A/P boundary the somites and then the sclerotome through anterior- and within somites that is independent of Notch1 activity. posterior-specific expression of signaling factors (e.g., Notch1, Dll1, and FGFR1), transcription factors (e.g., Uncx4.1), and extracellular matrix proteins (e.g., collagen MATERIALS AND METHODS IX and chondroitin sulfate proteoglycan; reviewed in Rawls et al., 2000). Regulation of specification is intimately re- Intercrosses and Genotyping lated to segmentation during somitogenesis. Inactivation of components of the Notch signaling pathway or the bHLH Mice (bl6/C129 hybrid) carrying a targeted null allele for the paraxis gene were described previously (Burgess et al., 1996). The transcription factor, Mesp2, which disrupt normal segmen- mutation was maintained in a C57/bl6 background. Mice carrying tation, resulted in the failure to specify the posterior and mutations in this gene were identified by Southern blot analysis by anterior fate of the somite, respectively (Swiatek et al., probing SacI-digested genomic DNA with a 300-bp EcoRI-SacI 1994; Conlon et al., 1995; Hrabe de Angelis et al., 1997; fragment. A transgenic mouse that expresses ␤-galactosidase (LacZ) Shen et al., 1997; Wong et al., 1997; Kusumi et al., 1998; in the dorsal root ganglia between Embryonic Day (E) 11.5 and Evrard et al., 1998; Zhang and Gridley, 1998). E14.5 embryos was generated by a random insertion of a construct

FIG. 1. A schematic diagram of sclerotome resegmentation during the development of the vertebral column. Individual sclerotome compartments are subdivided into and an anterior (A; red) and posterior (P; blue) compartment. Black arrows indicate the direction of sclerotome cell movement during the reassociation of the anterior and posterior halves of the sclerotome to form the vertebral bodies. The ventromedial path of neural crest cells into the posterior sclerotome is demarcated with blue dashed arrows. FIG. 2. The ventral aspect of the vertebral column fails to segment in the absence of paraxis. A ventral view of the vertebral column of wild-type (A) and paraxisϪ/Ϫ (B) neonatal mice stained for cartilage (Alcian blue) and bone (Alizarin Red). The vertebral bodies of the paraxisϪ/Ϫ mutants possess two ossification sites and fuse along the A/P axis (yellow arrowhead). Alcian blue selectively binds to proteoglycans that are prevalent in hyaline cartilage, but absent in fibrocartilage. Thus hyaline (vertebrae) and fibrous (intervertebral disc) cartilage can be distinguished by the preferential binding of Alcian blue. The hyaline cartilage in the paraxisϪ/Ϫ mutants is not segmented, suggesting the absence of the intervertebral discs. Lateral (C and D) and dorsal (E and F) views of a wild-type (C and E) and paraxisϪ/Ϫ (E and F) E14.5 embryo stained for cartilage and bone. The vertebral bodies have clearly coalesced into distinct cartilaginous segments in the wild-type but not the mutant embryos. Note that the dual ossification centers of the paraxisϪ/Ϫ vertebral bodies are prepatterned by a failure of the cartilage to fuse at the ventral midline. ivd, intervertebral disc; na, neural arch; pr, proximal rib; vb, vertebral body.

Copyright © 2001 by Academic Press. All rights of reproduction in any form reserved. aai euaino oieAPPolarity A/P Somite of Regulation Paraxis oyih 01b cdmcPes l ihso erdcini n omreserved. form any in reproduction of rights All Press. Academic by 2001 © Copyright

FIG. 3. The dorsal root ganglia are fused in the absence of paraxis. A transgenic mouse line expressing ␤-galactosidase in the dorsal root ganglia was bred into the paraxisϪ/Ϫ background. (A) E14.5 embryos of the specified genotype were stained for ␤-galactosidase activity and cleared with BABB. The dorsal root ganglia were fused along the axis of the paraxisϪ/Ϫ embryo. (B) Histological sections of E14.5 paraxisϩ/Ϫ/lacZϩve and paraxisϪ/Ϫ/lacZϩve embryos. E14.5 embryos of the specified genotype were stained for ␤-galactosidase activity, fixed, and embedded in paraffin. Twelve-micrometer sections were generated and stained with eosin. Note that expression of lacZ was not uniform throughout the DRG. No clear boundaries can be distinguished between the DRG in the absence of paraxis. The sensory neurons in the mutant embryos fail to form proper bundles. drg, dorsal root ganglia; tg, trigeminal ganglia. FIG. 4. Transcription of genes in the Notch signaling pathway in the absence of paraxis. In situ hybridizations were performed on wild-type and paraxisϪ/Ϫ littermates using antisense RNA probes specific for Notch1 (A and B), Dll1 (C and D), L-fng (E and F), and Mesp2 (G and H). In E9.5 embryos, transcription of Notch1 and Dll1 in the presomitic mesoderm was not altered in the absence of paraxis. Note that the domain of transcription of both of the genes was diffuse Ϫ Ϫ in the somites of paraxis / embryos. Black arrowheads denote expression of Notch1 and Dll1 in the posterior half of the somite. The pattern of L-fng (white 179 arrows) at E9.0 and Mesp2 (purple arrowhead) at E10.5 in the presomitic mesoderm was not altered in the mutant littermates. 180 Johnson et al.

10% fetal bovine serum. The somites and presomitic mesoderm were teased away from the surrounding tissue and transferred to RNase-free 1.5-ml centrifuge tubes. Total RNA was isolated using a modiﬁed version of the TRIzol method (Gibco-BRL), as described in Rawls et al. (1995).

Reverse Transcription-PCR The detection of ephrin-B2, EphA4, Notch1, Dll1, and L7 mRNA in the three newest epithelial somites was performed as described by Mu¨nsterberg et al. (1995) and as modified by Rawls et al. (1995). Total RNA isolated from the three newest somites FIG. 5. Analysis of gene transcription in the three newest was reverse-transcribed using a random hexamer primer and somites. Semiquantitative RT/PCR was performed on total RNA then amplified using gene-specific primers. PCR was performed isolated from the three newest somites. The level of transcription under semiquantitative conditions that allow for a direct com- of Notch1, Dll1, EphA4, and ephrin-B2 was compared between E9.5 parison of gene transcription levels in different genetic back- Ϫ/Ϫ wild-type and paraxis littermates. PCR amplifications were grounds. The amount of input RNA into each reaction was performed in duplicate with a twofold change in the amount of controlled by parallel amplification reactions with L7, a ribopro- input cDNA. PCR amplifications were considered valid if a twofold tein that is transcribed uniformly in all cells. In an attempt to difference was observed in the product. L7, a ribosomal RNA, was maintain amplification in the linear range, control reactions used as a RNA loading control between samples. A mock RT (M) were performed with each primer set to assess the appropriate PCR amplification was performed for each primer pair using RNA range of cycles for the input cDNA. In addition, duplicate isolated from wild-type somites. A representative comparison of reactions were set up with half the amount of input cDNA. The Ϫ/Ϫ wild-type and paraxis amplifications at a single concentration in amplification products of the duplicate reactions must be 1:2 for the linear range of amplification is shown. the experiment to be valid. Primers for detecting gene-specific transcripts were Notch1 forward primer 5Ј-CCTGTTGGA AG- TCCTTTCCA-3Ј and reverse primer 5Ј-TGGAACAGATATGT- TTTTCCCC-3Ј; Dll1 forward primer 5Ј-CGCCTTCAGCAA- CCCCATCC-3Ј and reverse primer 5Ј-GCAACCTTCTCCGTAG- containing a minimal TK promoter upstream of the LacZ gene. The TAGT-3Ј; ephrin-B2 forward primer 5Ј-TTCAGCCCTAACCTC-3Ј insertion site has not been characterized. and reverse primer 5Ј-GCCTCAGACCTTGTA-3Ј; EphA4 forward primer 5Ј-GGTATGATGTCCGCTCTGAT-3Ј and reverse primer 5Ј- Histology and ␤-Galactosidase Staining GTTGTGATTGGTGTTTTGCTG-3Ј; and L7 forward primer 5Ј- GGAGCTCATCTATGAGAAGGC-3Ј and reverse primer 5Ј- The expression of the dorsal root ganglia-specific lacZ transgene AAGACGAAGGAGCTGCAGAAC-3Ј. was detected by ␤-galactosidase activity, as described previously (Cheng et al., 1995). Briefly, mothers were sacrificed and the embryos dissected away from the uterus. The extraembryonic In Situ Hybridization membranes were removed and saved as a source of genomic DNA Whole-mount in situ hybridization was performed using anti- for Southern blot analysis. Embryos were fixed in 2% paraformal- sense RNA probe labeled with digoxygenin-11-UTP as described by dehyde, 0.2% glutaraldehyde, in phosphate-buffered saline (PBS), Wilson-Rawls et al. (1999). E9.5 and E10.5 embryos were hybrid- and then stained in 5 mM potassium ferricyanide, 5 mM potassium ized with antisense probes specific to Dll1 (Bettenhausen et al., ferrocyanide, 2 mM MgCl2, 1 mg/ml 5-bromo-4-chloro-3-indolyl- 1995), Notch1 (Williams et al., 1995), Lunatic Fringe (L-fng) (Evrard ␤ -D-galactopyranoside (X-gal; Sigma) in PBS, overnight at room et al., 1998), Mesp2 (Saga et al., 1997), ephrin-B2 (Bergemann et al., temperature. Embryos were washed in PBS and postfixed in 4% 1995), and EphA4 (Nieto et al., 1992). paraformaldehyde for an additional 24 h. The preparation of embryonic tissue for histological analysis was performed as described previously (Rawls et al., 1998). The tissue was embedded in paraffin Bone and Cartilage Staining after a stepwise dehydration in methanol and xylene. Twelve- The cartilage and bone development in neonatal and embryonic micrometer transverse sections were generated, deparaffinized in mice were stained as described by Martin et al. (1995). The xylene, and stained with eosin. neonates or E14.5 embryos were eviscerated, and the skin was removed and fixed in 95% ethanol overnight. Cartilage was visu- Isolation of RNA from Somitic and Presomitic alized by staining in 0.15% Alcian blue in ethanol overnight. Tissue Alcian blue specifically binds to glycosaminoglycans, a component of the extracellular matrix of hyaline cartilage. After staining, the The three newest somites and the anterior presomitic mesoderm tissue was destained overnight in several changes of 95% ethanol. were isolated from E9.5 embryos from a paraxisϩ/Ϫ cross. Extraem- Neonates and embryos were then treated with 2% KOH for 1 to 3 h bryonic membrane was removed in PBS and used for genotyping. at room temperature and then stained with 0.005% Alizarin Red S The anterior two-thirds of the embryo was cut away and the in 2% KOH, overnight. Alizarin Red will stain bone by binding remaining tissue was transferred to Tyrode’s solution. The tissue specifically to calcium. The tissue was destained in 2% KOH for 24 was mildly digested in dispase for 10 min at room temperature. The to 48 h. To maintain the skeleton intact, the buffer was changed to enzymatic reaction was stopped in DMEM supplemented with 1:1 2% KOH:glycerol.

ϩve RESULTS Ϫ/lacZ cross were harvested at E14.5 and stained for lacZ expression. Segmented pairs of DRG on either side of the Defects in the Segmentation of the Ventral neural tube were easily distinguishable along the entire Structures of the Developing Vertebrae of Paraxis- length of the wild-type embryo (Fig. 3A). In contrast, Deficient Mice boundaries were not discernible between DRG in paraxisϪ/Ϫ A hallmark of A/P polarity is the resegmentation of the littermates. The segmental nature of DRG was not com- sclerotomal compartments to form the vertebral body and pletely lost, as it can be seen at the ventralmost aspect, neural arches of individual vertebrae. Previously, we de- where the sensory nerves project into the periphery. Histo- paraxisϪ/Ϫ logical thin sections were generated to examine the DRG scribed defects in the vertebral column of neo- Ϫ/Ϫ ϩve natal mice that included dual ossification centers and formation at the cellular level. In E14.5 paraxis /lacZ random fusions of the vertebral bodies along the A/P axis embryos, boundaries between adjacent DRG were not ap- (Burgess et al., 1996). A more in-depth analysis of the parent (Fig. 3B). In wild-type embryos, axons from neurons cartilage associated with the vertebral bodies revealed a in the DRG project back toward the dorsal horn of the more dramatic loss of segmentation than previously de- neural tube as a single bundle, forming the dorsal sensory scribed (Figs. 2A and B). Hyaline cartilage of the vertebrae root. However, in the absence of paraxis, multiple smaller can be distinguished from the fibrocartilage of the interver- axon bundles were seen along the length of the axis. This tebral discs based on the selective staining of Alcian blue to suggests that paraxis is required for the proper segmental glycosaminoglycans present only in the former. In the pattern of the peripheral nerves. Additionally, several of the Ϫ/Ϫ sensory projections into the periphery were absent in the paraxis neonates, there was no clear evidence of inter- Ϫ/Ϫ vertebral discs along the length of the vertebral column. paraxis embryos. The loss of the sensory projections was To determine whether the vertebral fusions were due to coincidental with fusions in the vertebral column and thus degeneration of the vertebrae or alternatively resulted from likely due to a mechanical obstruction. a failure of the sclerotome to properly resegment, we examined the formation of these structures in wild-type Paraxis Is Not Required for Transcription of Ϫ Ϫ and paraxis / embryos at E14.5. At this stage, sclerotomal Members of the Notch Signaling Pathway in the derivatives have undergone chondrogenesis and resegmen- Presomitic Mesoderm tation such that the metameric pattern of the cartilaginous precursors of the vertebral bodies is clearly discernible (Figs. The loss of segmentation of the peripheral nerves and the 2C and E). In paraxisϪ/Ϫ embryos, the ventral cartilage failed axial skeleton suggests that paraxis plays a role upstream of to segment into distinct vertebral bodies (Figs. 2D and F). chondrogenesis during the establishment of A/P polarity in Over the same axial region, the neural arches segmented individual somites. Gene targeting experiments have re- reliably, though the ventral aspects of some of these seg- vealed that specification of polarity is dependent on activa- ments were absent (Fig. 2D). A dorsal view of this tissue tion of the Notch signaling pathway (de la Pompa et al., also revealed a failure of the cartilage to form at the ventral 1997; Evrand et al., 1998; Kusumi et al., 1998; Barrantes et midline, suggesting a deficiency in sclerotomal cell migra- al., 1999). We were interested in determining whether tion in the paraxis mutants (Fig. 2F). This was consistent paraxis was necessary for signaling through Notch. It has with the observation that there are two ossification sites on been reported that paraxis transcription was not altered by either side of the ventral midline instead of a single site. the absence of Dll1 (Hrabe de Angelis et al., 1997). Since the Overall, these data indicate that paraxis is required for inactivation of any one of the genes blocks signaling in the morphogenic patterning of the vertebrae along the A/P axis. Notch pathway (Pourquie´, 1999), it is unlikely that the transcription of paraxis is directly regulated by Notch signaling. The ability of paraxis to regulate the transcrip- The Peripheral Nerves Fail to Segment in the tion of members of the Notch signaling pathway was Absence of Paraxis determined by whole-mount in situ hybridization of Neural crest cell migration during peripheral nerve devel- paraxisϪ/Ϫ embryos. At E9.5, both Notch1 and Dll1 are opment is responsive to the A/P polarity of the sclerotome normally expressed throughout the presomitic mesoderm of individual somites. To determine whether paraxis is with the highest concentration near the anterior end (Figs. necessary for the A/P patterning of the peripheral nerves, 4A and C). In paraxisϪ/Ϫ embryos, a similar expression the organization of the dorsal root ganglia (DRG) was pattern was observed in the presomitic mesoderm for both examined in paraxisϪ/Ϫ embryos. Paraxisϩ/Ϫ mice were bred genes (Figs. 4B and D). with a transgenic mouse line that expresses the Lunatic Fringe (L-fng), a modulator of Notch signaling, is ␤-galactosidase gene (lacZ) in cells of the DRG. This expressed in a dynamic pattern in the presomitic meso- DRG-LacZ line was generated by the random insertion of derm, the proper spatiotemporal regulation of which is lacZ under the control of a minimal thymidine kinase intimately associated with segmentation (Forsberg et al., promoter into an uncharacterized gene locus. In these 1998; McGrew et al., 1998; Aulehla and Johnson, 1999). A embryos, lacZ is expressed in the DRG and trigeminal predominant feature is a band of expression that demarcates ganglia between E11.5 and E14.5. Embryos from a paraxisϩ/ the second presumptive somite and a broader band that

FIG. 6. Paraxis is required for proper transcription of ephrin-B2 in the posterior half of the somite. In situ hybridizations were performed on E9.5 wild-type (A and C), and paraxisϪ/Ϫ (B and D) embryos, using antisense RNA probes specific for EphA4 (A and B) and ephrin-B2 (C and D). Posterior-restricted transcription of ephrin-B2 (red arrowheads) was noted in the somites of wild-type embryos. However, it was diffuse in the paraxisϪ/Ϫ embryos. Transcription of EphA4 in the anterior half of the first somite (yellow arrowhead) and in the next presumptive somite (blue arrowhead) at the anterior end of the presomitic mesoderm was normal in the paraxisϪ/Ϫ embryos. FIG. 7. A model of paraxis function during anterior/posterior patterning of the somite. (a) Normal A/P patterning is dependent on specification of alternating domains of cells to an anterior or posterior fate. Specification occurs in the region of the anterior presomitic mesoderm that will give rise to at least the next two somites. Tightly associated with specification in the presomitic mesoderm and somites is the maintenance of a cellular boundary. A failure for the presomitic cells to become specified (b) or the adoption of a single fate (c or d) will result in the loss of somite polarity and disruption of the axial skeleton and peripheral nerves. Alternatively, specification could occur normally and the boundary between anterior and posterior compartments is disrupted (e). This would produce a similar fused phenotype in the axial skeleton and peripheral nerves. Paraxis Regulation of Somite A/P Polarity 183 progressively moves in an anterior direction starting from ences were observed (Figs. 4G and H). Two conclusions can the tail bud. In embryos at E9.0, two bands of L-fng be drawn from this observation, paraxis is not required for transcription were detectable. These bands in the paraxisϪ/Ϫ the normal transcription of Mesp2 and Notch signaling is embryos were comparable in intensity and position to those not disrupted in the mutant embryos. Since Mesp2 forms a observed in wild-type littermates (Figs. 4E and F). The positive feedback loop with Notch1, a block in the activity unaltered pattern of L-fng, Notch1, and Dll1 suggests that of one of these genes will lead to a decrease in the transcrip- paraxis is not required to regulate the transcription of genes tion of the other. in the Notch signaling pathway in the presomitic mesoderm. Restriction of ephrin-B2 Transcription to the Paraxis Is Required for Spatial Regulation of Posterior Half of the Somite Requires Paraxis Notch and Dll1 Transcription in the Somites EphA4 is a member of the Eph family of receptor tyrosine In wild-type embryos, transcription of Dll1 is maintained kinases, which has been implicated in the regulation of in the posterior half of the somite (Fig. 4C; Bettenhausen et both somite formation and A/P patterning. Ephrin-B2, a al., 1995). Transcription of Dll1 in paraxisϪ/Ϫ embryos ligand for EphA4, is expressed in two diffuse bands at the appeared diffuse in the somites, without any clear distinc- anterior end of the presomitic mesoderm and in the poste- tion between the anterior and the posterior halves (Fig. 4D). rior half of somites (Bergemann et al., 1995). Later, Notch1, which is also expressed at its highest level in the ephrin-B2 is expressed in the posterior half of the scle- posterior half of the somite (Fig. 4A), was diffuse in the rotome. Disruption of EphA4 or ephrin-B2 in zebrafish paraxisϪ/Ϫ embryos (Fig. 4B). This demonstrates that paraxis results in aberrant somite formation and a block in myo- is required to restrict transcription of Notch1 and Dll1 to genesis (Durbin et al., 1998). Later, during the establish- the posterior half of the somite. ment of the peripheral nerves, neural crest cell migration is Next, we tested whether the loss of posterior restriction inhibited in the posterior half of the sclerotome. In vitro of Notch1 and Dll1 transcription was associated with a studies have demonstrated that ephrin-B2 protein is suffi- change in the level of transcription of these genes in cient to inhibit neural crest cell migration and axon guid- paraxisϪ/Ϫ embryos. Semiquantitative RT/PCR was per- ance in stripe assays (Wang and Anderson, 1997). We were formed using total RNA isolated from the newest three interested in determining whether paraxis regulated the epithelial somites. To ensure that the PCR amplification level or pattern of transcription of these two genes. EphA4 was performed within the linear range, parallel reactions is expressed in the anterior half of the first somite and the were performed with half the amount of input cDNA. In next presumptive somite in the presomitic mesoderm addition, the amount of input RNA was normalized by (Becker et al., 1994). Recently, it was demonstrated that performing PCR amplifications on each sample with L7, a EphA4 transcription in the presomitic mesoderm and ribosomal RNA. Under these conditions, a 2- to 3-fold somites required signaling through the Notch pathway reduction in the level of transcription Dll1 and Notch1 was (Barrantes et al., 1999). In E9.5 paraxis-deficient embryos, detectable in the paraxis mutants when compared to wild EphA4 transcription was not altered in either the first type (Fig. 5). somite or the presomitic mesoderm (Figs. 6A and B). In paraxisϪ/Ϫ embryos, transcription of ephrin-B2 in the presomitic mesoderm was similar to that of wild-type litter- Transcription of Mesp2 in the Presomitic mates at E8.5 (not shown) and E10.5 (Fig. 6D). However, the Mesoderm Is Not Dependent on Paraxis level of ephrin-B2 transcription in the mutant somites Mesp2 is a bHLH transcription factor that is expressed in appears to be dramatically reduced. a single band that demarcates the anterior half of the second Semiquantitative RT/PCR was used to measure the level presumptive somite in the presomitic mesoderm. Mesp2 is of EphA4 and ephrin-B2 transcripts in the first three believed to play an important role in specifying the anterior somites (Fig. 5). EphA4 transcription was not affected by the half of the somite (Saga et al., 1997). In Mesp2Ϫ/Ϫ neonates, absence of paraxis, consistent with the in situ hybridization the axial skeleton is fused along the A/P axis and the dorsal data. Interestingly, the level of ephrin-B2 transcription was root ganglia fail to segment. In addition, a positive feedback also unchanged in the paraxisϪ/Ϫ somites. Though this loop exists between Mesp2 and Notch signaling, such that appears inconsistent, the difference can be reconciled with Mesp2 expression is lost in Dll1Ϫ/Ϫ and NotchϪ/Ϫ and the expression study if ephrin-B2 transcription was distrib- conversely, Dll1 and Notch1 transcription is downregulated uted throughout the somite, reducing the level of mRNA in in Mesp2Ϫ/Ϫ embryos (Saga et al., 1997; Barrantes et al., any one area below the sensitivity threshold for whole- 1999). It is possible that the phenotype of the paraxis-null mount in situ hybridizations. These data suggest that could be due to the downregulation of Mesp2 transcription paraxis is necessary to restrict the domain of ephrin-B2- in the absence of paraxis. However, when the level and expressing cells to the posterior half of the somite, similar pattern of Mesp2 transcription in the paraxisϪ/Ϫ embryos to the expression patterns observed for both Notch1 and were compared to those of wild-type littermates, no differ- Dll1.

DISCUSSION Paraxis Regulation of the Peripheral Nerve Segmentation through ephrin-B2 Expression Ϫ/Ϫ The segmental pattern of the axial skeleton and periph- In paraxis embryos, the DRG are fused along the eral nerves in vertebrates is dependent on the division of the length of the embryo, predicting a loss in the inhibition of sclerotome along the anterior/posterior axis. This process is neural cell migration into the posterior half of the scle- initiated before the formation of somites when nonoverlap- rotome. Inhibition of neural crest cell migration into the ping domains of the presomitic mesoderm become specified posterior sclerotomal compartment is believed to be medi- to either an anterior or a posterior fate. Associated with ated, in part, through the interaction of EphB3 receptor on specification is the onset of domain-specific expression of the surface of the neural crest cells with ephrin-B1 and genes involved in the Notch signaling pathway. A/P polar- ephrin-B2 expressed in the posterior sclerotome (reviewed ity is maintained in the somites and then the sclerotome by Bronner-Fraser, 2000). In vitro, neural crest cell migra- through compartment-specific expression of signaling fac- tion and axon guidance were biased toward stripes of the tors (e.g., Notch1, Dll1, FGFR1, and members of the Eph culture dish devoid of either ephrin-B1 or ephrin-B2 (Krull family of tyrosine kinases and their ligands), transcription et al., 1997; Wang and Anderson, 1997). Interestingly, a factors (e.g., Uncx4.1), and extracellular matrix proteins continuous field of either ephrin did not inhibit cell migra- (e.g., Peanut lectin binding molecules, collagen IX, tion. This suggests that a discontinuous field of the Eph T-cadherins, and chondroitin sulfate proteoglycans; re- ligands are required for the proper patterning of cell migra- Ϫ/Ϫ viewed in Rawls et al., 2000). However, it remains to be tion. In paraxis embryos, the expression of ephrin-B2 was determined how these genes may direct A/P polarity or diffuse across the anterior and posterior sides of the somite. direct resegmentation of the sclerotome. Here, we report It is tempting to predict that the loss of DRG segmentation Ϫ/Ϫ that paraxis is also necessary for normal specification or in the paraxis neonates is due to the loss of polar maintenance of somite A/P polarity. Paraxis-deficient em- expression of ephrin-B2. However, mice deficient in bryos fail to develop distinct segments in the vertebral ephrin-B2 do not display segmental defects in the peripheral column and dorsal root ganglia. Examination of gene ex- nerves (Wang et al., 1998; Adams et al., 1999). It is possible pression in the presomitic mesoderm and somites revealed that a combination of Eph ligands is required to direct a failure of mutant somites to maintain a distinct A/P normal neural crest cell migration and the absence of any boundary. one can be compensated for. Further analysis will be required to evaluate the potential causative link between misexpression of Eph ligands and DRG fusion in paraxisϪ/Ϫ mice. Paraxis Regulates the Patterning of the Ventral Structures of the Vertebrae Paraxis Functions Independently of the Notch The defects observed in the vertebrae of paraxisϪ/Ϫ em- Signaling Pathway bryos are consistent with a defect in resegmentation. The The overlapping pattern of expression of paraxis, Notch1, most dramatic fusions occurred in the vertebral bodies, and Dll1 in the anterior half of the presomitic mesoderm while the neural arches are much less affected. This is and the phenotypic similarities of the null mutations pre- concordant with the relative dependence of the two tissues dict a potential genetic interaction between these genes. It upon resegmentation for proper development. The nature of has been reported that paraxis transcription is not regulated the defect in resegmentation is less clear. The presence of through the Notch signaling pathway (Hrabe de Angelis et the neural arches and the absence of the intervertebral discs al., 1997; Shen et al., 1997). Here we report that paraxis does in paraxis mutants predict that the entire sclerotome has not regulate the pattern or level of transcription of Notch1, adopted a posterior fate. However, in mutants where the Dll1, or L-fng in the presomitic mesoderm. In addition, the sclerotome has been shown to be posteriorized, such as in normal expression of the Notch target genes, Mesp2 and Ϫ Ϫ Mesp2 / neonates, the vertebral structures derived from EphA4, in paraxisϪ/Ϫ embryos indicates that paraxis does the neural arches are fused (Saga et al., 1997). It would be not affect Notch signaling in the presomitic mesoderm. In predicted that if cells of the anterior compartment adopted contrast, paraxis appears to play a much more important a posterior fate, there would be an increase in the level of role in regulating the spatial pattern of gene expression in transcription of genes that are normally restricted to the newly formed somites. Notch1 and Dll1 continued to be posterior half of the somite. In contrast, we observed no expressed in the posterior half of the somite, though their change in the level of Dll1, Notch1, or ephrin-B2 in the first function in these cells remains to be determined. In the three somites. Further, we did not observe a reduction in absence of paraxis, both genes were transcribed evenly the level of an anterior marker, EphA4. Therefore, the across the A/P axis of the somite. This revealed a specific phenotype of the paraxisϪ/Ϫ embryos is more consistent role for paraxis in the spatial regulation of genes that specify with a role for paraxis in regulating the maintenance of polarity in the somite and is in contrast to Notch signaling, separate anterior and posterior compartments after they which is required to regulate compartment-specific gene have been specified. expression in both the presomitic mesoderm and somites.

This also predicts that the function of paraxis is temporally Gossler, and Dr. U. Lendahl for gene-specific probes, and Barb and spatially independent of Notch signaling. Backes for imaging support. In addition, we acknowledge the W. M. Keck Foundation for their generous support of the W. M. Keck Bioimaging Laboratory in which part of these experiments were Specification versus Boundary Maintenance carried out. This research was supported by National Science In theory, establishing A/P polarity within a somite Foundation Grant IBN-9905982. requires both the specification of anterior and posterior cell fates in the presomitic mesoderm and a mechanism by which the intrasomitic boundary between the two popula- tions is maintained (Fig. 7). The experimental definition of REFERENCES A/P specification is based on the observation that when presomitic mesoderm has been transplanted in the opposite Adams, R. H., Wilkinson, G. A., Weiss, C., Diella, F., Gale, N. W., orientation, the migration path of neural crest cells is Deutsch, U., Risau, W., and Klein, R. (1999). Roles of ephrinB reversed (Bronner-Fraser and Stern, 1991). However, this ligands and EphB receptors in cardiovascular development: De- functional definition does not distinguish between specifi- marcation of arterial/venous domains, vascular morphogenesis, cation and maintenance of the A/P boundary during further and sprouting angiogenesis. Genes Dev. 13, 295–306. embryonic development. Specification of A/P fate has been Aulehla, A., and Johnson, R. L. (1999). Dynamic expression of inferred from the restricted expression of genes to domains lunatic fringe suggests a link between notch signaling and an in the presomitic mesoderm fated to become either the autonomous cellular oscillator driving somite segmentation. anterior or the posterior halves of the next two somites. The Dev. Biol. 207, 49–61. Bagnall, K. M., Higgins, S. J., and Sanders, E. J. (1988). The observation that the expression of L-fng in the posterior contribution made by a single somite to the vertebral column: domains and Mesp2 and EphA4 in the anterior domains of Ϫ/Ϫ Experimental evidence in support of resegmentation using the the presomitic mesoderm in the paraxis embryos pre- chick-quail chimera model. Development 103, 69–85. dicts that paraxis does not regulate specification. However, Bagnall, K. M., Sanders, E. J., and Berdan, R. C. (1992). Communi- it remains to be determined whether gene expression in the cation compartments in axial mesoderm of the chick embryo. anterior and posterior domains of the presomitic mesoderm Anat. Embryol. 186, 195–204. predicts or specifies positional identity. It is possible that Barnes, G. L., Alexander, P. G., Hsu, C. W., Mariani, B. D., and Notch signaling and Mesp2 are necessary but not sufficient Tuan, R. S. (1997). Cloning and characterization of chicken for A/P specification. In such a model, paraxis could act as Paraxis: A regulator of paraxial mesoderm development and a necessary cofactor. This would also explain how the somite formation. Dev. Biol. 189, 95–111. broad, unsegmented expression of paraxis in the presomitic Barrantes, I. B., Elia, A. J., Wunsch, K., De Angelis, M. H., Mak, mesoderm could regulate segmentation. T. W., Rossant, J., Conlon, R. A., Gossler, A., and de la Pompa, J. L. (1999). Interaction between Notch signaling and Lunatic Alternatively, paraxis could be required to maintain the fringe during somite boundary formation in the mouse. Curr. integrity of the A/P boundary after specification has been Biol. 9, 470–480. established (Fig. 7). To date there is no direct evidence that Becker, N., Seitanidou, T., Murphy, P., Mattei, M. G., Topilko, P., a maintenance function is required for A/P polarity in the Nieto, M. A., Wilkinson, D. G., Charnay, P., and Gilardi- somite. However, there are examples of a cell-segregation Hebenstreit, P. (1994). Several receptor tyrosine kinase genes of mechanism being linked to specification during compart- the Eph family are segmentally expressed in the developing mentalization, including rhombomere formation and the hindbrain. Mech. Dev. 47, 3–17. subdivision of Drosophila imaginal discs along the anterior/ Bergemann, A. D., Cheng, H.-J., Brambilla, R., Klein, R., and posterior and dorsal/ventral axes (reviewed by Dahmann Flanagan, J. G. (1995). Elf-2, a new member of the Eph ligand and Basler, 1999). Since paraxis is required for cells in the family, is segmentally expressed in mouse embryos in the region somite to become epithelial (Burgess et al., 1996), it is of the hindbrain and newly forming somites. Mol. Cell. Biol. 15, possible that an increase in cell adhesion prevents migra- 4921–4929. Bettenhausen, B., Hrabe de Angelis, M., Simon, D., Guenet, J. L., tion across the A/P boundary. The diffuse pattern of gene and Gossler, A. (1995). Transient and restricted expression during expression in the paraxis mutants could be due to increased mouse embryogenesis of Dll1, a murine gene closely related to cell movement in the absence of that epithelium. The broad Drosophila Delta. Development 121, 2407–2418. expression pattern of paraxis in the newly formed somites Blanar, M. A., Crossley, P. H., Peters, K. G., Steingrimsson, E., could be reconciled with this model, since it does not Copeland, N. G., Jenkins, N. A., Martin, G. R., and Rutter, W. J. require specific anterior or posterior activities. Direct anal- (1995). Meso1, a basic-helix loop-helix protein involved in mam- ysis of cell migration in the somites will be required to malian presomitic mesoderm development. Proc. Natl. Acad. assess the role of paraxis in maintaining the A/P boundary. Sci. USA 92, 5870–5874. Brand-Saberi, B., and Christ, B. (2000). Evolution and development of distinct cell lineages derived from somites. Curr. Top Dev. ACKNOWLEDGMENTS Biol. 48, 1–42. Bronner-Fraser, M., and Stern, C. (1991). Effects of mesodermal We thank J. Wilson-Rawls for valuable discussion and critical tissues on avian neural crest cell migration. Dev. Biol. 143, reading of the manuscript, Dr. R. Johnson, Dr. Y. Saga, Dr. A. 213–217.

Bronner-Fraser, M. (2000). Rostrocaudal differences within the Interactions of Eph-related receptors and ligands confer rostro- somites confer segmental pattern to trunk neural crest migra- caudal pattern to trunk neural crest migration. Curr. Biol. 7, tion. Curr. Top. Dev. Biol. 47, 279–296. 571–580. Burgess, R., Cserjesi, P., Ligon, K. L., and Olson, E. N. (1995). Kusumi, K., Sun, E. S., Kerrebrock, A. W., Bronson, R. T., Chi, Paraxis: A basic helix loop-helix protein expressed in paraxial D. C., Bulotsky, M. S., Spencer, J. B., Birren, B. W., Frankel, mesoderm and developing somites. Dev. Biol. 168, 296–306. W. N., and Lander, E. S. (1998). The mouse pudgy mutation Burgess, R., Rawls, A., Brown, D., Bradley, A., and Olson, E. N. disrupts Delta homologue Dll3 and initiation of early somite (1996). Requirement of the paraxis gene for somite formation and boundaries. Nature Genet. 19, 274–278. musculoskeletal patterning. Nature 384, 570–573. Loring, J. F., and Erickson, C. A. (1987). Neural crest cell migratory Cheng, T. C., Tseng, B. S., Merlie, J. P., Klein, W. H., and Olson, pathways in the trunk of the chick embryo. Dev. Biol. 121, E. N. (1995). Activation of the myogenin promoter during mouse 220–236. embryogenesis in the absence of positive autoregulation. Proc. Mansouri, A., Yokota, Y., Wehr, R., Copeland, N. G., Jenkins, Natl. Acad. Sci. USA 92, 561–565. N. A., and Gruss, P. (1997). Paired-related murine homeobox Christ, B., and Ordahl, C. P. (1995). Early stages of chick somite gene expressed in the developing sclerotome, kidney, and ner- development. Anat. Embryol. 191, 381–396. vous system. Dev. Dyn. 210, 53–65. Conlon, R. A., Reaume, A. G., and Rossant, J. (1995). Notch1 is Martin, J. F., Bradley, A., and Olson, E. N. (1995). The paired-like required for the coordinate segmentation of somites. Develop- homeobox gene MHox is required for early events of skeletogen- ment 121, 1533–1545. esis in multiple lineages. Genes Dev. 9, 1237–1249. Dahmann, C., and Basler, K. (1999). Compartment Boundaries at McGrew, M. J., Dale, J. K., Fraboulet, S., and Pourquie, O. (1998). the edge of development. Trends Genet. 15, 320–325. The lunatic fringe gene is a target of the molecular clock linked Dunwoodie, S. L., Henrique, D., Harrison, S. M., and Beddington, to somite segmentation in avian embryos. Curr. Biol. 8, 979–982. R. S. (1997). Mouse Dll3: A novel divergent Delta gene which Munsterberg, A. E., Kitajewski, J., Bumcrot, D. A., McMahon, A. P., may complement the function of other Delta homologues during and Lassar, A. B. (1995). Combinatorial signaling by Sonic early pattern formation in the mouse embryo. Development 124, hedgehog and Wnt family members induces myogenic bHLH 3065–3076. gene expression in the somite. Genes Dev. 9, 2911–2922. Durbin, L., Brennan, C., Shiomi, K., Cooke, J., Barrios, A., Shan- Nieto, M. A., Gilardi-Hebenstreit, P., Charnay, P., and Wilkinson, mugalingam, S., Guthrie, B., Lindberg, R., and Holder, N. (1998). D. G. (1992). A receptor protein tyrosine kinase implicated in the Eph signaling is required for segmentation and differentiation of segmental patterning of the hindbrain and mesoderm. Develop- the somites. Genes Dev. 12, 3096–3109. ment 116, 1137–1150. de la Pompa, J. L., Wakeham, A., Correia, K. M., Samper, E., Brown, Orioli, D., and Klein, R. (1997). The Eph receptor family: Axonal S., Aguilera, R. J., Nakano, T., Honjo, T., Mak, T. W., Rossant, J., guidance by contact repulsion. Trends Genet. 13, 354–359. and Conlon, R. A. (1997). Conservation of the Notch signalling Pourquie´, O. (1999). Notch around the clock. Curr. Opin. Genet. pathway in mammalian neurogenesis. Development 124, 1139– Dev. 9, 559–565. 1148. Quertermous, E. E., Hidai, H., Blanar, M. A., and Quertermous, T. Evrard, Y. A., Lun, Y., Aulehla, A., Gan, L., and Johnson, R. L. (1994). Cloning and characterization of a basic helix-loop-helix (1998). Lunatic fringe is an essential mediator of somite segmen- protein expressed in early mesoderm and the developing somites. tation and patterning. Nature 394, 377–381. Proc. Natl. Acad. Sci. USA 91, 7066–7070. Flenniken, A., Gale, N., Yancopoulos, and Wilkinson, D. (1996). Rawls, A., Morris, J. H., Rudnicki, M., Braun, T., Arnold, H. H., Distinct and overlapping expression patterns of ligands for Eph Klein, W. H., and Olson, E. N. (1995). Myogenin’s functions do related receptor tyrosine kinases during mouse development. not overlap with those of MyoD or Myf-5 during mouse embryo- Dev. Biol. 179, 382–401. genesis. Dev. Biol. 172, 37–50. Forsberg, H., Crozet, F., and Brown, N. A. (1998). Waves of mouse Rawls, A., Valdez, M. R., Zhang, W., Richardson, J., Klein, W. H., Lunatic fringe expression, in four-hour cycles at two-hour inter- and Olson, E. N. (1998). Overlapping functions of the myogenic vals, precede somite boundary formation. Curr. Biol. 8, 1027– bHLH genes MRF4 and MyoD revealed in double mutant mice. 1030. Development 125, 2349–2358. Goldstein, R. S., and Kalcheim, C. (1992). Determination of epithe- Rawls, A., Wilson-Rawls, J., and Olson, E. N. (2000). Genetic lial half-somites in skeletal morphogenesis. Development 116, regulation of somite formation. Curr. Top. Dev. Biol. 47, 131– 441–445. 154. Hrabe de Angelis, M., McIntyre, J., and Gossler, A. (1997). Mainte- Rickmann, M., Fawcett, J. W., and Keynes, R. J. (1985). The nance of somite borders in mice requires the Delta homologue migration of neural crest cells and the growth of motor axons Dll1. Nature 386, 717–721. through the rostral half of the chick somite. Embryol. Exp. Huang, R., Zhi, Q., Neubuser, A., Muller, T. S., Brand-Saberi, B., Morphol. 90, 437–455. Christ, B., and Wilting, J. (1996). Function of somite and somi- Saga, Y., Hata, N., Koseki, H., and Taketo, M. M. (1997). Mesp2: A tocoele cells in the formation of the vertebral motion segment in novel mouse gene expressed in the presegmented mesoderm and avian embryos. Acta Anat. 155, 231–241. essential for segmentation initiation. Genes Dev. 11, 1827–1839. Krull, C. E., Collazo, A., Fraser, S. E., and Bronner-Fraser, M. (1995). Shanmumgalingam, S., and Wilson, S. W. (1998). Isolation, expres- Segmental migration of trunk neural crest: Time-lapse analysis sion, and regulation of a zebraﬁsh paraxis homologue. Mech. reveals a role for PNA-binding molecules. Development 121, Dev. 78, 85–89. 3733–3743. Shen, J., Bronson, R. T., Chen, D. F., Xia, W., Selkoe, D. J., and Krull, C. E., Lansford, R., Gale, N. W., Collazo, A., Marcelle, C., Tonegawa, S. (1997). Skeletal and CNS defects in Presenilin-1- Yancopoulos, G. D., Fraser, S. E., and Bronner-Fraser, M. (1997). deﬁcient mice. Cell 89, 629–639.

Sosic, D., Brand-Saberi, B., Schmidt, C., Christ, B., and Olson, E. N. Williams, R., Lendahl, U., and Lardelli, M. (1995). Complementary (1997). Regulation of paraxis expression and somite formation by and combinatorial patterns of Notch gene family expression ectoderm- and neural tube derived signals. Dev. Biol. 185, during early mouse development. Mech. Dev. 53, 357–368. 229–243. Wilson-Rawls, J., Hurt, C. R., Parsons, S. M., and Rawls, A. (1999). Swiatek, P. J., Lindsell, C. E., del Amo, F. F., Weinmaster, G., and Differential regulation of epaxial and hypaxial muscle develop- Gridley, T. (1994). Notch1 is essential for postimplantation ment by paraxis. Development 126, 5217–5229. development in mice. Genes Dev. 8, 707–719. Wong, P. C., Zheng, H., Chen, H., Becher, M. W., Sirinathsinghji, Teillet, M. A., Kalcheim, C., and Le Douarin, N. M. (1987). D. J., Trumbauer, M. E., Chen, H. Y., Price, D. L., Van der Ploeg, Formation of the dorsal root ganglia in the avian embryo: L. H., and Sisodia, S. S. (1997). Presenilin 1 is required for Notch1 Segmental origin and migratory behavior of neural crest progeni- and DII1 expression in the paraxial mesoderm. Nature 387, tor cells. Dev. Biol. 120, 329–347. 288–292. Wang, H. U., and Anderson, D. J. (1997). Eph family transmem- Zhang, N., and Gridley, T. (1998). Defects in somite formation in brane ligands can mediate repulsive guidance of trunk neural lunatic fringe-deﬁcient mice. Nature 394, 374–377. crest migration and motor axon outgrowth. Neuron 18, 383–396. Wang, H. U., Chen, Z. F., and Anderson, D. J. (1998). Molecular Submitted for publication July 11, 2000 distinction and angiogenic interaction between embryonic arter- Revised September 8, 2000 ies and veins revealed by ephrin-B2 and its receptor Eph-B4. Cell Accepted October 3, 2000 93, 741–753. Published online December 5, 2000