Evolutionary Conservation of the Presumptive Neural Plate Markers Amphisox1/2/3 and Amphineurogenin in the Invertebrate Chordate Amphioxus

Developmental Biology 226, 18–33 (2000)

View metadata,doi:10.1006/dbio.2000.9810, citation and similar papers available at onlinecore.ac.uk at http://www.idealibrary.com on brought to you by CORE

provided by Elsevier - Publisher Connector Evolutionary Conservation of the Presumptive Neural Plate Markers AmphiSox1/2/3 and AmphiNeurogenin in the Invertebrate Chordate Amphioxus

Linda Z. Holland,* M. Schubert,*,1 N. D. Holland,* and T. Neuman† *Marine Biology Research Division, Scripps Institution of Oceanography, University of California at San Diego, La Jolla, California 92093-0202; and †Department of Surgery, Burns and Allen Research Institute, Cedars–Sinai Medical Center, University of California at Los Angeles School of Medicine, Los Angeles, California 90048

Amphioxus, as the closest living invertebrate relative of the vertebrates, can give insights into the evolutionary origin of the vertebrate body plan. Therefore, to investigate the evolution of genetic mechanisms for establishing and patterning the neuroectoderm, we cloned and determined the embryonic expression of two amphioxus transcription factors, AmphiSox1/ 2/3 and AmphiNeurogenin. These genes are the earliest known markers for presumptive neuroectoderm in amphioxus. By the early neurula stage, AmphiNeurogenin expression becomes restricted to two bilateral columns of segmentally arranged neural plate cells, which probably include precursors of motor neurons. This is the earliest indication of segmentation in the amphioxus nerve cord. Later, expression extends to dorsal cells in the nerve cord, which may include precursors of sensory neurons. By the midneurula, AmphiSox1/2/3 expression becomes limited to the dorsal part of the forming neural tube. These patterns resemble those of their vertebrate and Drosophila homologs. Taken together with the evolutionarily conserved expression of the dorsoventral patterning genes, BMP2/4 and chordin, in nonneural and neural ectoderm, respectively, of chordates and Drosophila, our results are consistent with the evolution of the chordate dorsal nerve cord and the insect ventral nerve cord from a longitudinal nerve cord in a common bilaterian ancestor. However, AmphiSox1/2/3 differs from its vertebrate homologs in not being expressed outside the CNS, suggesting that additional roles for this gene have evolved in connection with gene duplication in the vertebrate lineage. In contrast, expression in the midgut of AmphiNeurogenin together with the gene encoding the insulin-like peptide suggests that amphioxus may have homologs of vertebrate pancreatic islet cells, which express neurogenin3. In addition, AmphiNeurogenin, like its vertebrate and Drosophila homologs, is expressed in apparent precursors of epidermal chemosensory and possibly mechanosensory cells, suggesting a common origin for protostome and deuterostome epidermal sensory cells in the ancestral bilaterian. © 2000 Academic Press Key Words: body plan evolution; neural patterning; pancreas; olfactory placodes; motor neurons; Sox; neurogenin; tap; dichaete; embryogenesis; neural plate.

INTRODUCTION gin, and chordin in Xenopus), bind to and inhibit the TGF-␤ family members BMP2 and BMP4 to ensure a neural fate in Comparisons of developmental gene expression patterns the dorsal ectoderm (Sasai and De Robertis, 1997; Wein- between vertebrates and their closest living invertebrate stein and Hemmati-Brivanlou, 1997). In amphioxus, as in relative, amphioxus, have suggested that the genetic vertebrates, BMP2/4 is expressed in the nonneural ecto- mechanism for distinguishing neuro- and nonneuroecto- derm of the gastrula, becoming down-regulated in the derm in chordates evolved before the amphioxus/vertebrate neural plate (Panopoulou et al., 1998). In Drosophila as split (Holland and Holland, 1999). In the vertebrate gastrula, well, the expression and function of decapentaplegic and secreted proteins from the organizer (e.g., follistatin, nog- short gastrulation are similar to those of their respective vertebrate homologs (BMP2/4 and chordin), suggesting an- 1 Contributed the work on AmphiSox1/2/3. cient roles for these genes in dorsoventral patterning of

0012-1606/00 $35.00 Copyright © 2000 by Academic Press 18 All rights of reproduction in any form reserved. Amphioxus Presumptive Neural Plate Markers 19 bilaterian organisms (Holley et al., 1995; Marque´s et al., cerebrellar region appears to be lacking (reviewed in Hol- 1997). In contrast, the early molecular events downstream land and Holland, 1999). of chordin/sog that establish neural fates in the neuroecto- Amphioxus embryology differs in some respects from derm are less well understood (Streit and Stern, 1999). that of vertebrates. The blastula has but a single layer of In Xenopus, chordin, follistatin, and noggin secreted by cells, gastrulation consists of a flattening and invagination the organizer appear to act together with other factors in the of one side of the blastula, and there is little involution of induction and maintenance of several early neuroectoder- cells around the lips of the blastopore (Zhang et al., 1997). mal markers, including Sox2 and 3 (Mizuseki et al., 1998; At the end of gastrulation as the neural plate forms, the Streit et al., 1998). Expression of these markers together amphioxus embryo becomes more vertebrate-like. How- with that of genes such as FGF leads to the initiation of ever, neural folds do not form; instead, the nonneural nerve cell differentiation. Genes in the Sox1/2/3 group are ectoderm detaches from the edges of the open neural plate, Sry-related HMG box transcription factors most closely moves laterally over the neural plate, and fuses in the dorsal related to Drosophila Dichaete (ϭfish-hook, ϭSox70D) midline. Only then do the lateral edges of the neural plate curl up dorsally into the neural tube (Hirakow and Kajita, (Collignon et al., 1996; Nambu and Nambu, 1996; Russell 1994; N. D. Holland et al., 1996). Nevertheless, despite et al., 1996; Vriz et al., 1996; Penzel et al., 1997; Rex et al., these differences, our results show that AmphiSox1/2/3 and 1997; Sakai et al., 1997; Y. Ma et al., 1998). Vertebrate Sox2 AmphiNeurogenin, like their vertebrate homologs, are is first expressed throughout the presumptive neuroecto- broadly expressed very early in presumptive neuroectoderm of the very late blastula, turning on after chordin and derm. Furthermore, expression of AmphiNeurogenin, like before the proneural gene neurogenin-1 (Xngnr-1), a ho- that of vertebrate neurogenin-1, persists in a segmental molog of Drosophila tap (biparous) (Bush et al., 1996; subset of neural plate cells in the presumptive hindbrain, Gautier et al., 1997; Rex et al., 1997; Mizuseki et al., 1998; which probably includes presumptive motor neurons. Ex- Wood and Episkopou, 1999). neurogenins encode basic pression of Sox1/2/3 and neurogenin homologs in the de- helix-loop-helix (bHLH) transcription factors that bind to veloping CNS also appears to be somewhat conserved E-box sequences of target genes (Gradwohl et al., 1996). between amphioxus and Drosophila. These results suggest neurogenin-1 is first expressed in the vertebrate gastrula in evolutionary conservation of genetic mechanisms for estab- broad domains of presumptive neuroectoderm that give rise lishing the neural plate, for maintenance of neural fate, and to both neurons and nonneuronal cells (Ma et al., 1996, for neural induction in amphioxus and the vertebrates and, 1997; Blader, 1997; Korzh et al., 1998; Eisen, 1999). In to some extent, in Drosophila as well. Drosophila, the Sox1/2/3 homolog Dichaete and the neurogenin homolog tap (biparous) are also initially expressed quite broadly in the neuroectoderm (Bush et al., 1996; MATERIALS AND METHODS Nambu and Nambu, 1996; Russell et al., 1996). The rela- Amphioxus Material tion between these genes and short gastrulation (sog) has not been tested, although in sog mutants, expression of Adults of the Florida amphioxus (Branchiostoma floridae) were several neuroectodermal markers is nearly normal (Jaz´win´- collected from Old Tampa Bay, Florida. Genomic DNA was ex- ska et al., 1999). tracted in guanidinium isothiocyanate and purified as previously To date, no markers of presumptive neuroectoderm have described (L. Z. Holland et al., 1996). Ripe adults were electrically stimulated to spawn in the laboratory, and embryos and larvae been cloned from amphioxus, and it is not clear to what were raised at 23°C (Holland and Holland, 1993). extent gene networks acting downstream of chordin in the neuroectoderm are conserved across major taxonomic Cloning groups. To address this question, we have cloned and determined the embryonic expression of amphioxus ho- For amphioxus Sox genes, 250,000 clones of a cDNA library in mologs of Sox1/2/3 and neurogenin. Amphioxus is Lambda ZAP II (Stratagene, Inc., La Jolla, CA) made from 5- to 18-h vertebrate-like, but genomically and structurally simpler. embryos of B. floridae were screened under moderate stringency (hybridization in 7.25ϫ SSC, 0.24% SDS, 12.2ϫ Denhardt’s at Homologs of most vertebrate gene families are probably 60°C; washes in 1ϫ SSC, 0.1% SDS at 60°C) with mixed probes for represented in the amphioxus genome, but amphioxus has two amphioxus Wnt genes—a 354-bp fragment of Wnt6 (Wnt-B; not undergone the genome duplications that occurred early Holland et al., 1994) and a 405-bp piece of Wnt 7A, which was in vertebrate evolution (Holland et al., 1994). Structurally, obtained in screening a gridded cDNA library from 26-h embryos in amphioxus and vertebrates share features such as a dorsal pSport1 (Gibco BRL, Inc., Rockville, MD) under moderately low hollow nerve cord, notochord, muscular somites, and pha- stringency with mixed probes for the amphioxus Wnt6 fragment ryngeal gill slits. However, amphioxus lacks paired append- plus a 360-bp piece of amphioxus Wnt4 (Wnt-A; Holland et al., ages, paired eyes, auditory organs, and a skeleton. Micro- 1994). Two very weakly positive clones obtained corresponded to full-length cDNAs of the same Sox gene. anatomical and genetic studies have suggested that the To isolate a 174-bp fragment of amphioxus neurogenin, PCR was amphioxus nerve cord contains regions comparable to the used with degenerate primers corresponding to the bHLH region of vertebrate diencephalon and hindbrain and possibly the mammalian helix-loop-helix transcription factors and genomic DNA anteriormost portion of the midbrain, although an isthmo- of B. floridae. The forward primer corresponded to the amino acid

sequence 5Ј-ANA(D,N)RERR(N,T)R, nucleotide sequence 5Ј- plasmid DNA, via the calcium phosphate precipitation method. GCIAAC(T)G(A)C(A)IC(A)GIGAA(G)C(A)G(A,C)IC(A)G-3Ј, and the The medium was changed to normal growth medium 16 h after reverse primer to amino acid sequence NYIWALA(T,S)E, nucleotide transfection. Cells were harvested 48 h later, and CAT assays were sequence 5Ј-TCIC(G)C(T,A)IAG(A)IGCCCAT(G,A)ATG(A)TAG(A)- performed as previously described (Chiaramello et al., 1995b). TT-3Ј (I is inosine). PCR conditions were denaturation—93°C, 1 min; Quantification of acetylation ratios was obtained by PhosphorIm- annealing—45°C, 3 min; synthesis—72°C, 2 min. PCR products were ager (Molecular Dynamics, Sunnyvale, CA) analysis. To normalize cloned, and one, corresponding to neurogenin, was used to screen a transfection efficiencies, cells were cotransfected with 0.5 ␮gofthe cDNA library of 2- to 4-day larvae in Lambda ZAP II (Stratagene Inc.) plasmid pRcRSVlacZ. All reported CAT activities were normalized under conditions of medium stringency (hybridization in 2ϫ SSC at to total protein and lacZ activity. Values represent the means of at 42°C). least three independent transfections. Embryos and larvae were fixed and processed for whole-mount in situ hybridization by the method of L. Z. Holland et al. (1996). Southern Blot Analysis Antisense riboprobes were synthesized from full-length cDNA Sixteen 10-␮g aliquots of genomic DNA were each digested with clones. Hybridized embryos and larvae were photographed as a different restriction enzyme, subjected to electrophoresis, and whole mounts, counterstained with Ponceau S, embedded in plas- transferred to Hybond Nϩ (Amersham Life Sciences, Cleveland, tic, and sectioned at 3 ␮m. OH) according to L. Z. Holland et al. (1997). The Southern blot was hybridized at low stringency (55°C) with a 447-bp EcoO109I/EcoRV Phylogenetic Analysis and Amino Acid Alignments fragment of the AmphiSox1/2/3 cDNA located in the coding region 164 bp 3Ј of the HMG box. Washes were 2ϫ 20 min at 50°C in 2ϫ Phylogenetic trees were constructed heuristically with 100 ran- SSC, 0.1% SDS. These conditions are comparable to those previ- dom stepwise additions (PAUP 3.1.1). Outgroups were mouse SRY ously used for demonstrating the presence of two muscle actin for Sox proteins and MATH-1 for neurogenins. Bootstrap values genes in amphioxus (Kusakabe et al., 1997). After being probed, the were calculated in 1000 cycles with 10 random stepwise additions blot was stripped in boiling 0.5% SDS and reprobed under identical per cycle. For the Sox tree, 13 proteins were used. Because of the conditions with a 347-bp SmaI/SphI fragment of AmphiNeuroge- large number of amino acid identities in the HMG box among nin, located 5Ј of the bHLH domain, overlapping the 5Ј end of the different Sox proteins, it was necessary to base the tree on the HMG domain by 40 bp. HMG box (80 characters) plus conserved regions C-terminal of the HMG box (45 characters). Only one most parsimonious tree, which had a length of 561, was retained. The tree of neurogenins is based Electrophoretic Mobility Shift Assay (EMSA) only on the 67 amino acids of the bHLH domain. The analysis For expression of AmphiNeurogenin, a full-length cDNA was yielded a single tree with a length of 121. All sequences used and cloned into the eukaryotic expression vector pcDNA3 (Invitrogen, their GenBank accession numbers are listed in the figure legends. Inc., Carlsbad, CA). Amphioxus neurogenin protein was synthesized Amino acid alignments were performed with ClustalW (Baylor in vitro using the TNT Quick Coupled Transcription/Translation College of Medicine) and manually adjusted. System (Promega, Inc., Madison, WI). The resulting translation product was analyzed by SDS–polyacrylamide gel electrophoresis. EMSA were carried out as described by Chiaramello et al. (1995a). Briefly, 3 RESULTS ␮l of the translation product was incubated for 15 min at room temperature in binding buffer (10 mM Tris–HCl, pH 7.5, 50 mM KCl, Characterization of AmphiSox1/2/3

0.1 mM EDTA, 1 mM dithiothreitol, 1 mM MgCl2, 5% glycerol) with There are seven groups (A–G) in the vertebrate Sox gene 100 ng of poly(dI–dC) (Roche Molecular Biochemicals, Indianapolis, family. Screening a cDNA with mixed probes for two Wnt 32 IN) and 40 fmol of a P-labeled double-stranded oligonucleotide probe genes (405 and 354 bp long) yielded two identical clones corresponding to the E-box sequence of the rat p75 gene promoter. coding for an amphioxus Sox B group gene. Fortuitously, the Complementary single-stranded probes [5Ј-GCATTGCCTTCACC- nucleotide sequence between bases 156 and 590 of this Sox CAGCTGCTCCCGCCCGC (E-box sequence is underlined)] were synthesized, allowed to anneal, and labeled at the 3Ј recessed ends gene is 56% identical to the sequence of heterologous with [␣-32P]dCTP and the large fragment of DNA polymerase I probes used for screening (data not shown). Both Sox clones (Klenow). Unlabeled competitor DNA was prepared by annealing are 1878 bp long, including an open reading frame of 735 bp complementary oligonucleotides. After incubation, the binding reac- coding for a protein of 245 amino acids. The amino- tion was subjected to electrophoresis on a 5% native polyacrylamide terminal half of the protein includes an HMG box of 79 gel. After electrophoresis the gel was dried and exposed to X-ray ﬁlm. amino acids characteristic of Sox proteins. The HMG box is a DNA-binding domain, which resembles that in high- Cell Culture, Transfection, CAT Assay, and in Situ mobility group nonhistone chromosomal proteins (Wegner, Hybridization 1999). As Fig. 1 shows, the amphioxus Sox protein shares 7 of the 14 amino-terminal amino acids that are conserved in Transcriptional activity of AmphiNeurogenin was measured in the vertebrate Sox B group proteins Sox1, Sox2, and Sox3. In an in vitro cell culture system. The rat p75 promoter CAT addition, the HMG box shares 75% identities with chick construct (1.4 CAT) (Metsis et al., 1992) and GAP-43 promoter CAT construct (GXC; Chiaramello et al., 1996) were used as Sox1, 2, and 3 proteins. Several domains C-terminal to the reporter plasmids. Mouse neuroblastoma Neuro2A cells (ATCC) HMG domain, including the 12-amino-acid group B homol- were grown in Dulbecco’s modiﬁed Eagle’s medium containing ogy domain adjacent to the HMG box, are also conserved 10% fetal bovine serum (Gibco). Cells in 35-mm diameter dishes at between the amphioxus Sox protein and the vertebrate a density of 5 ϫ 104 cells per dish were transfected with 5 ␮g of total homologs. However, the C-terminal 54 amino acids, rich in

FIG. 1. Amino acid alignment of Sox1/2/3 proteins. Double underlining indicates the HMG domain. Single underlining indicates regions used in addition to the HMG domain for constructing the phylogenetic tree in Fig. 3. Identities are indicated by dark shading, conserved substitutions by light shading. Accession numbers are as in Fig. 3 except for chickSox1 (AB01802), chickSox2 (U12532), and chickSox3 (U12467). proline (15.5%), serine (15.5%), and threonine (11%), sug- Phylogenetic analysis of the Sox B clade (Fig. 3) shows gestive of a transactivation domain, are not conserved that the amphioxus Sox B protein is located at the base of between the amphioxus Sox B protein and the homologs of the vertebrate Sox1, Sox2, and Sox3 proteins. We, therefore, other species. Furthermore, this region is over 40 amino acids shorter in our amphioxus Sox protein than the corresponding region in other species (Fig. 1). Southern blot analysis demonstrates a single major band with some enzymes, e.g., BglI, EcoRI, NcoI, XhoI (Fig. 2). However, one or more fainter bands are present in all lanes with major bands less than 10 kb. Some of these bands may be due to a restriction site in an intron or to polymorphism, which is very high in amphioxus. Since AmphiSox1/2/3 has many identities with vertebrate Sox1/2/3 genes over the region of the probe (Fig. 1), we would expect the additional bands to be quite intense if there were another AmphiSox1/ 2/3 homolog. It is most likely, therefore, that the fainter bands are due to the recognition of a more distantly related Sox B class gene, e.g., homologs of sea urchin SoxB2 (Fig. 3) or chick and human Sox14 and Sox21 (Malas et al., 1999) FIG. 2. Genomic Southern blot analysis of pooled amphioxus DNA. Numbers at top indicate digestion in 1, BamHI; 2, BglI; 3, or to unrelated genes like the Wnts; the AmphiSox1/2/3 BstEII; 4, EcoO109I; 5, EcoRI; 6, BstXI; 7, HindIII; 8, KpnI; 9, PstI; cDNAs were obtained by screening with Wnt4/6 10, NotI; 11, NcoI; 12, PvuI; 13, SalI; 14, StuI; 15, XbaI; 16, XhoI. probes, and sequence alignment of AmphiWnt4 and the Blot probed under low stringency with a 447-bp EcoO109I, EcoRV AmphiSox1/2/3 probe used for the Southern blot demon- fragment of AmphiSox1/2/3 located in the coding region just 3Ј of strates 48% nucleotide identity. the HMG box. Molecular size standards in kilobases at left.

FIG. 3. Phylogenetic tree of Sox1/2/3 proteins. Drosophila Dichaete is used as an outgroup. A single most parsimonious tree was obtained. Accession numbers are Drosophila Dichaete (X96419), sea urchin SoxB1 (AF157389), mouse Sox1 (X94126), mouse Sox2 (X94127), mouse Sox3 (X94125), mouse SRY (AF070933), mouse Sox15 (AB014474), mouse Sox4 (S37303), mouse Sox17 (Q61473), mouse Sox10 (Q048880), and AmphiSox1/2/3 (AF271787).

term this amphioxus gene AmphiSox1/2/3. This tree sug- midline and in the anterior part of the CNS (Figs. 4J–4L). At gests that the Sox1, Sox2, and Sox3 genes arose from gene the midneurula stage (Fig. 4M), expression is limited to the duplication in the vertebrate lineage. The analysis further extreme posterior-dorsal portion of CNS and then ceases indicates that sea urchin SoxB1 is more closely related to entirely. However, in much later larvae at the one- and AmphiSox1/2/3 than is sea urchin SoxB2 and suggests that two-gill-slit stages (2–5 days), prolonged staining revealed these sea urchin genes arose from an independent gene extremely weak expression in a few cells scattered in the duplication within the echinoderm lineage. CNS rostral to the pigment spot at the level of somite 5 (data not shown). Expression was not detected in tissues other than neuroectoderm. Embryonic Expression of AmphiSox1/2/3 AmphiSox1/2/3 is a highly specific marker for presump- Characterization of AmphiNeurogenin tive neuroectoderm. Its expression is first detectable by in situ hybridization in the early gastrula (cap-shaped stage) in Of 52 clones obtained by amplification of genomic DNA the dorsal epiblast, including the presumptive neuroecto- of B. floridae with degenerate primers to the bHLH region derm (Figs. 4A and 4B). Expression remains uniformly and sequenced, 14 corresponded to a 174-bp fragment of a strong in the presumptive neuroectoderm throughout gas- single amphioxus neurogenin gene. No other clones corre- trulation (Figs. 4C–4E). Whether the expression domain sponding to neurogenins were obtained. cDNA library includes any cells in adjacent nonneural ectoderm at these screening with the neurogenin fragment yielded four am- early stages is not clear because of the lack of anatomical phioxus neurogenin clones. The longest was 1240 bp, in- markers. However, at the onset of neurulation when the cluding the entire coding region but lacking the C-terminal neural plate flattens and becomes distinct from nonneural portion of the 3Ј UTR. The gene codes for a protein of 255 ectoderm (Figs. 4F and 4G), it is evident that expression is aa, including a bHLH region of 60 aa. C-terminal of the limited to neural ectoderm and is excluded from nonneural bHLH domain there are several proline, serine, and glycine ectoderm. As the edges of the neural plate curl dorsally, residues conserved among the amphioxus, vertebrate, and expression initially remains panneural (Figs. 4H and 4I). Drosophila homologs (Fig. 5). Later, before the edges of the neural plate have fused Southern blot analysis with a probe 5Ј of the bHLH region dorsally, expression becomes down-regulated along the reveals a single major band with most enzymes (Fig. 6). Two

FIG. 4. AmphiSox1/2/3 expression during development. For side and dorsal views of whole mounts, anterior is to the left. (A) Animal pole view of an early gastrula with the blastopore opening away from viewer; expression is in the epiblast at the presumptive dorsal side of the embryo. (B) Cross section of very early gastrula with blastopore opening toward the right (approximate posterior end of embryo). Expression is in the dorsal epiblast (arrow). (C) Whole mount of midgastrula in dorsal view; expression is in the dorsal epiblast (presumptive neuroectoderm) including the dorsal lip of the blastopore, opening toward the right. (D) Cross section of midgastrula in side view. Expression is in the presumptive neuroectoderm. (E) Side view of late gastrula/early neurula with expression throughout early neural plate. (F) Same gastrula as in E in dorsal view showing expression in early neural plate. (G) Cross section at the level of the arrowhead in F showing expression throughout early neural plate. (H) Early neurula with expression limited to neural plate. (I) Cross section through level of arrowhead in H showing expression in the neural plate but not in the dorsal epidermis (arrows) overgrowing the neural plate. (J) Side view of an early to midneurula with expression in neural plate. (K) Dorsal view of neurula in J. Expression is becoming restricted to the edges of the neural plate. (L) Optical cross section though level of arrowhead in K. Expression is down-regulated in the midline of the neural plate (ﬂoor plate). (M) Side view of midneurula. Expression is down-regulated except in the most posterior region of the neural tube (arrow). Scale for whole mounts, 50 ␮m, and for cross sections (counterstained pink), 25 ␮m.

bands of equal intensity obtained with some enzymes, e.g., with the neurogenins and, thus, NeuroD and Neurogenin SalI and StuI, are probably due to polymorphism or to the evidently separated before the deuterostome/protostome presence of introns. Thus, the amphioxus neurogenin gene split. we cloned probably has no close relatives, a conclusion To examine the ability of AmphiNeurogenin to activate supported by the results from PCR ampliﬁcation of transcription from two promoters of neuronal genes that genomic DNA with degenerate primers. Phylogenetic anal- contain E-box sequences, rat p75 and GAP-43, transient ysis shows that this gene is located at the base of the CAT assays were performed with mouse neuroblastoma vertebrate neurogenins (Fig. 7). We, therefore, term this Neuro2A cells. Cotransfection of an AmphiNeurogenin amphioxus gene AmphiNeurogenin. The tree indicates that expression vector with a reporter plasmid resulted in sig- the vertebrate neurogenins arose from gene duplication in niﬁcant increases in CAT activity in transfected cells (Fig. the vertebrate lineage. In addition, Drosophila tap groups 8A), indicating that AmphiNeurogenin can act as a tran-

FIG. 5. Amino acid alignment of neurogenin proteins. Single underlining indicates the bHLH, used for constructing the phylogenetic tree in Fig. 7. Identities are indicated by dark shading, conserved substitutions by light shading. Accession numbers as in Fig. 7.

scriptional activator in mammalian cells. To determine if Embryonic Expression of AmphiNeurogenin AmphiNeurogenin can bind to E-box sequences and acti- Expression of AmphiNeurogenin is first detectable in the vate transcription we performed EMSA and transient CAT very early gastrula about the stage at which AmphiSox1/2/3 assays. Binding of AmphiNeurogenin to an oligonucleotide is first expressed (Figs. 9A and 9B). Expression is restricted carrying the p75 gene promoter E-box (CAGCTG) produced to the presumptive neural plate in a domain that largely a shifted AmphiNeurogenin–DNA complex (Fig. 8B), sug- overlaps that of AmphiSox1/2/3; however, AmphiNeuroge- gesting that AmphiNeurogenin functions as a DNA binding nin, unlike AmphiSox1/2/3, is not expressed in the dorsal protein. lip of the blastopore (Fig. 9B). As gastrulation proceeds, expression remains strong in the presumptive neural plate (Figs. 9C–9F). In the early neurula, while AmphiSox1/2/3 is still uniformly expressed in the neural plate, AmphiNeuro- genin becomes down-regulated in the future floor plate and expression begins to break up into transverse stripes (Figs. 9G–9I). In addition, expression becomes down-regulated in the anteriormost portion of the neural plate. By the midneurula stage, expression in the CNS is in dorsolateral rows of transverse stripes except in the cerebral vesicle and in the posteriormost portion of the nerve cord (Figs. 9J–9L). There is no expression in the floor plate. A dorsal view (Fig. 9K) shows that cells strongly expressing AmphiNeurogenin are in approximate register with somite boundaries; however, there are two areas of strong expression on each side at the level of somite 2. In the late neurula (Fig. 9M), these stripes FIG. 6. The same Southern blot as in Fig. 2, stripped and reprobed of expression persist in dorsolateral cells of the nerve cord with a 347-bp SmaI, SphI fragment of AmphiNeurogenin located in (Figs. 9N and 9O); in addition, expression has become the coding region just 5Ј of the bHLH region. Restriction enzymes detectable in the ventral endoderm cells of the midgut (Figs. numbered at top as in Fig. 2. Molecular size standards in kilobases 9M, arrowhead, and 9N). In embryos with an incipient at left. mouth opening, expression in the nerve cord has been

FIG. 7. Phylogenetic tree of neurogenin and neuroD proteins. The distantly related bHLH protein MATH1 (mouse atonal-related protein 1) is used as an outgroup. GenBank accession numbers are MATH-1 (D43694), mouse NeuroD-1 (D83507), mouse NeuroD-2 (U58471), zebraﬁsh NeuroD-1a (AF115772), mouse neurogenin-1 (U76207), mouse neurogenin-2 (U76207), mouse neurogenin-3 (U76208), zebraﬁsh Neurogenin-1 (AF036149), AmphiNeurogenin (AF271788), Drosophila Tap (O16867).

largely down-regulated anteriorly and now occurs mostly posterior to the primary pigment spot (Fig. 9P, arrow). Expression in the midgut endoderm has expanded (Fig. 9P, arrowhead), and new domains have appeared in a few dorsal cells of the cerebral vesicle (Fig. 9Q, arrowhead) and in a few epidermal cells that may be precursors of the ciliary tuft cells, which differentiate just ventral to the mouth (Fig. 9Q, arrow). At 2 days of development, expression continues in the cerebral vesicle, in the neural tube posterior to the primary pigment spot, and in the midgut endoderm, but is down-regulated in the presumed ciliary tuft precursors. However, expression is up-regulated in the rostral ectoderm (Figs. 9R and 9S, tandem arrowheads). At 6 days of age, the larvae have elongated considerably, and expression continues in the cerebral vesicle, in the nerve cord (in a few widely spaced cells anteriorly and more closely spaced cells posteriorly), and in the midgut endoderm (Fig. 9T). In larvae more than a week old, expression persists only in the endoderm of the midgut, where it is both dorsal and ventral (Fig. 9T). Expression in a subset of midgut cells continues at least FIG. 8. (A) Transactivation potential of AmphiNeurogenin. Ampi- until metamorphosis (data not shown). Neurogenin stimulates activity of rat p75 and GAP-43 promoters in Neuro2A cell line. Plasmids p75 CAT and GAP43 CAT were transfected together with pRcCMV (control, C) or pRcCMV- AmphiNeurogenin (A) into neuroblastoma Neuro2A cells. CAT DISCUSSION activities are corrected for transfection efﬁciency and expressed relative to the value obtained by transfection of p75-CAT and GAP-43-CAT promoter reporter plasmids and expression plasmid Structure, Function, and Evolution of Sox1/2/3 and pRcCMV. CAT activities are presented as means Ϯ standard Neurogenin Genes deviations and represent results of at least three independent experiments. (B) Electrophoretic mobility shift assay showing bind- Vertebrate Sox1/2/3 proteins are transcriptional activa- ing of AmphiNeurogenin to the E-box sequence. 1, control; 2, tors. The transactivation domain is serine, threonine, and AmphiNeurogenin.

Copyright © 2000 by Academic Press. All rights of reproduction in any form reserved. Amphioxus Presumptive Neural Plate Markers 27 proline rich and comprises the C-terminal 90 amino acids neurogenins, it is likely that the binding and activation (Kamachi et al., 1998; Uchikawa et al., 1999). The corre- characteristics of vertebrate neurogenins are like those of sponding domain of AmphiSox1/2/3 is also serine, proline, AmphiNeurogenin. and threonine rich (44%), including several regions with a Southern blot and phylogenetic analyses suggest that high percentage of identity to vertebrate Sox1, Sox2, and AmphiSox1/2/3 and AmphiNeurogenin are the amphioxus Sox 3 sequences. Thus, although AmphiSox1/2/3 lacks 40 homologs, respectively, of three vertebrate Sox1/2/3 genes amino acids at its C-terminus compared to its vertebrate and three vertebrate neurogenin genes. Although the boot- and sea urchin homologs, its C-terminal region is also strap values for the neurogenin tree are low due to the high probably a transactivation domain. Since this region in percentage of identities in the bHLH region, the presence of vertebrate Sox1, but not Sox2, Sox3, or AmphiSox1/2/3, a single amphioxus neurogenin gene is additionally sup- also includes several stretches of alanine repeats, these ported by the failure of PCR with degenerate primers to repeats probably evolved only in vertebrate Sox1. amplify more than one neurogenin gene. Thus, the three Vertebrate Sox1/2/3 proteins typically bind to target vertebrate homologs of each gene probably evolved by genes, such as Fgf-4, in concert with POU domain proteins duplication of a single gene in the common ancestor of like Oct (Ambrosetti et al., 1997; Fraidenraich et al., 1998; amphioxus and the vertebrates. This presumably reﬂects reviewed in Bianchi and Beltrame, 1998; Uchikawa et al., the two rounds of genome duplication that occurred early in 1999) and initiate pathways leading to expression of neuro- vertebrate evolution (Holland et al., 1994). Unexpectedly, genins (Mizuseki et al., 1998), which in turn results in however, the analyses show that Drosophila tap (biparous) activation of a cascade of downstream bHLH factors, in- is more closely related to amphioxus and vertebrate neuro- cluding NeuroD, leading to neuronal differentiation (Q. Ma genins than it is to NeuroD. Until the discovery of tap et al., 1998). DNA-binding studies have shown that mouse (Gautier et al., 1997), it was thought that Drosophila had no neurogenin-2 (MATH4A) can bind as a heterodimer with close relative of neurogenin and NeuroD, the nearest rela- the ubiquitous bHLH protein E12 to the insulin E-box tive being atonal, most closely related to vertebrate sequence (Gradwohl et al., 1996). Activation studies of MATH-1 (Akazawa et al., 1995; reviewed in Lee, 1997). neurogenins have not been done; however, the related However, our tree shows that Drosophila tap groups with NeuroD genes are transcriptional activators (McCormick et neurogenins, not with NeuroD, suggesting that neurogenin al., 1996). Similarly, our results show that AmphiNeuroge- and NeuroD arose by gene duplication before the nin can bind to E boxes and act as a transcriptional activator deuterostome–protostome split. Thus, both amphioxus and in a mammalian cell culture system. Given the sequence Drosophila may have homologs of NeuroD which have yet conservation between AmphiNeurogenin and vertebrate to be discovered.

FIG. 9. AmphiNeurogenin expression during development. For side and dorsal views of whole mounts, anterior is to the left. (A) Animal pole view of a very early gastrula with blastopore opening away from viewer; expression in the dorsal epiblast (presumptive neuroectoderm). (B) Cross section of very early gastrula as in A with blastopore opening toward right (approximate posterior end of embryo). Expression is in the presumptive neuroectoderm (arrow) except just anterior to the blastopore. (C) Midgastrula in dorsal view with expression in the dorsal epiblast. Blastopore opens to the right. (D) Midgastrula in side view. Expression is in the epiblast except just anterior to the blastopore (arrowhead). (E) Dorsal view of midgastrula with expression in dorsal epiblast (presumptive neuroectoderm) except just anterior to the blastopore. (F) Gastrula in E viewed from the blastopore (approximate posterior side). Expression is in dorsal epiblast. (G) Side view of very early neurula with nearly closed blastopore (arrow). Expression is in the anterior 7/8 of the neural plate. (H) Dorsal view of neurula in G. Expression is restricted to two rows of cells with a segmental arrangement on either side of the midline of the neural plate. (I) Cross section through x in H showing the dorsal nonneural ectoderm (arrowheads) overgrowing the neural plate; cells in the midline of the neural plate (presumptive floor plate) do not express the gene. (J) Side view of an early/midneurula showing segmental expression in two rows of cells on either side of the floor plate. (K) Dorsal view of embryo in J. Segmental expression in cells (probable motor neuron precursors) on either side of the midline in approximate register with the somites (numbered). (L) Cross section through x in K. Expression is at either side of the neural plate, which has been overgrown by the epidermis dorsally. The notochord (n) is dorsal to the gut (g) and flanked by somites. (M) Side view of late neurula with expression in the neural tube and in the midgut endoderm. (N) Cross section through x in M. Expression is in the nerve cord, but not the floor plate and in ventral endoderm cells of the gut. (O) Cross section through y in M. Expression is in dorsolateral cells of the neural tube. (P) Side view of embryo with incipient mouth opening. Expression is in scattered groups of cells in the neural tube and in ventral endodermal cells of the midgut (arrowhead). The arrow indicates the primary pigment spot in the nerve cord. (Q) Enlargement of the anterior end of an embryo with incipient mouth opening; expression is in a few cells of the cerebral vesicle (single arrowhead) and in likely precursor cells of the ciliary tuft (arrow). (R) Side view of 2-day larva with primary pigment spot (arrow). Expression is in the neural tube, including the cerebral vesicle (single arrowhead), in midgut cells, and in the anterior epidermis (tandem arrowhead). (S) Enlargement of the anterior end of a 2-day larva with expression in the anterior epidermis (tandem arrowhead) and in a few cells of the cerebral vesicle (single arrowhead). (T) Side view of a 6-day larva with primary pigment spot (arrow). Expression is in endoderm cells of the midgut (inset) and in cells of the neural tube. Scale for whole mounts, 50 ␮m, and for cross sections (counterstained pink), 25 ␮m.

Conservation of Early Expression of Sox1/2/3 Conservation of Expression of AmphiSox1/2/3 and Genes and neurogenins in Amphioxus and AmphiNeurogenin during Neurulation in Vertebrates Amphioxus and Vertebrates

AmphiSox1/2/3 and AmphiNeurogenin are the earliest As neurulation begins, AmphiSox1/2/3 and AmphiNeu- known markers for the presumptive neuroectoderm in rogenin are both down-regulated in specific areas of the amphioxus, being up-regulated in the very early gastrula neural plate. AmphiSox1/2/3 turns off in the midline, as soon as the mesendoderm has invaginated. The becoming restricted to the edges of the neural plate, while panneural expression of AmphiSox1/2/3 resembles that AmphiNeurogenin expression becomes restricted to cells in of vertebrate Sox2. In vertebrates, initial expression of two rows on either side of the midline. Within each row, Sox2 may be more extensive than presumptive neuroec- the pattern of expression is striped, indicating a segmental toderm (Rex et al., 1997) and may be so in amphioxus. organization in the very early neural plate. The striped pattern extends posteriorly after the nerve cord has formed. The panneural expression of Sox1/2/3 genes was at first In later larvae, cells in the cerebral vesicle and several thought to confer neural competence on ectoderm (Streit domains outside the CNS also express the gene. In contrast, et al., 1997). However, more recent evidence suggests expression of AmphiSox1/2/3 largely ceases at the four- to that it may modulate the responsiveness to additional five-somite stage, well before the neural tube closes dor- signals in cells which have gained neural competence but sally, and it is not expressed in any other tissues during et al., are not yet committed to a neural fate (Mizuseki development. 1998; Pevny et al., 1998). These expression patterns are similar to those of all three Although early expression of amphioxus and vertebrate of their respective vertebrate homologs considered together. Sox1/2/3 genes are similar, initial expression of am- Vertebrate Sox1, Sox2, and Sox3 typically have overlapping phioxus and vertebrate neurogenins differ. The vertebrate patterns of expression in the early vertebrate CNS and may genes are never expressed throughout the entire anterior have considerable functional redundancy (Collignon et al., 7/8 of the dorsal epiblast. This difference, however, may 1996; Vriz et al., 1996; Rex et al., 1997; Sakai et al., 1997; simply be quantitative, not qualitative. Although neuro- Streit et al., 1997; Mizuseki et al., 1998; Pevny et al., 1998; genins were initially thought to be strictly neuronal Zygar et al., 1998; Uchikawa et al., 1999; Wood and determination genes, acting downstream of Sox1/2/3 Episkopou, 1999). Like AmphiSox1/2/3, the vertebrate (Sommer et al., 1996; Fode et al., 1998; Q. Ma et al., 1998, genes turn off ventrally in the floor plate (Penzel et al., 1999), at least in vertebrates, neurogenin-expressing cells 1997; Pevny et al., 1998; Zygar et al., 1998; Uchikawa et al., can be mitotic and biased toward, but not irreversibly 1999; Wood and Episkopou, 1999). However, expression of committed to, a neural fate (Ma et al., 1996; Blader et al., AmphiSox1/2/3 in the dorsal part of the CNS ceases early 1997; Kim et al., 1997; Korzh et al., 1998). Similarly, in compared to that of its vertebrate counterparts. The differ- amphioxus, since nonneuronal cells as well as neurons ence may be due to early differentiation of some neuronal differentiate anteriorly in the nerve cord (Lacalli et al., cell types in amphioxus. Amphioxus embryos display a 1994), the neuroectodermal cells initially expressing Am- positive phototropism as soon as they hatch at 10 h, and phiNeurogenin presumably differentiate into both cell nerve cell bodies of the peripheral nervous system in types. amphioxus have already begun to differentiate by the 8- to In vertebrates, the number of cells expressing neuroge- 10-somite stage (15–17 h), as shown by labeling with ␣ nin is regulated by lateral inhibition mediated by Notch/ anti-acetylated -tubulin (Yasui et al., 1998). In addition, the embryos begin muscular movements by 18–20 h (10–12 Delta signaling. Consequently, overexpression of verte- somites) (Stokes, 1997), indicating that motor neurons brate neurogenin outside the neural plate, where Notch/ differentiate relatively early. Delta signaling does not operate, induces ectopic While vertebrate Sox1/2/3 genes apparently become neurons, while overexpression within the neural plate down-regulated as cells cease dividing and embark on can reduce the number of neurons (Ma et al., 1996; Blader differentiation pathways, neurogenins apparently turn off et al., 1997; Takke et al., 1999) evidently due to activa- in cells that have lost the competence to become neurons tion of Delta/Notch signaling (Takke et al., 1999). Thus, and adopted nonneuronal fates. Although the function of the differences between amphioxus and vertebrates in the AmphiNeurogenin in vitro has not been determined, its in percentage of cells initially expressing neurogeninsinthe vivo behavior and expression in the neural plate are similar presumptive neuroectoderm may be due to differences in to those of its vertebrate counterparts. Thus, by compari- levels of Delta/Notch signaling. Delta has not been son, the two rows of cells with a segmental organization cloned from amphioxus, but the expression of Amphi- expressing AmphiNeurogenin in the neural plate are prob- Notch in presumptive neuroectoderm is complementary ably neuronal precursors. The pattern in 18-h embryos to that of AmphiNeurogenin (our unpublished data), strongly suggests that the most intensely expressing cells suggesting that Notch/Delta signaling might be involved are the dorsal compartment (DC) somatic motor neurons, in modulating the expression of AmphiNeurogenin in the which have been mapped in late larvae by Lacalli and Kelly neural plate. (1999) and which express the motor neuron marker Islet

(Jackman et al., 2000). Like the most intensely labeled cells pressed in tissues for which amphioxus apparently lacks a in Fig. 5K, the DC cells are located in pairs that are counterpart, such as the lens and retina, the otic placode, staggered left to right on either side of the floor plate. There and several neural crest derivatives (vertebrate Sox1/2/3 are two pairs in somite 2 and one pair per somite in 3, 4, and genes are not expressed in migrating neural crest) (Vriz et 5. The DC cells innervate the superficial muscle cells, al., 1996; Zygar et al., 1998; Uchikawa et al., 1999; Wood which are responsible for slow movement. The ventral and Episkopou, 1999). compartment somatic motor neurons are less segmentally The Drosophila Sox1/2/3 homolog Dichaete is also ex- arranged and extend farther forward than the DC cells. pressed outside the CNS, where it is involved in segmenta- These may be the more weakly AmphiNeurogenin- tion, regulating pair-rule genes (Nambu and Nambu, 1996; expressing cells visible in Fig. 5K. Russell et al., 1996; Y. Ma et al., 1998). In contrast, neither In vertebrates, neurogenin is also one of the first tran- amphioxus nor vertebrate Sox1/2/3 genes are segmentally scription factors expressed in prospective motor neurons expressed in any tissue. Thus, the extraneural expression (Eisen, 1998, 1999). In addition, other genes, like islet, have domains of Dichaete and vertebrate Sox1/2/3 genes do not conserved expression in presumptive motor neurons in appear comparable. Sea urchins also express SoxB1 in the amphioxus, vertebrates, and also Drosophila (Thor and larval ectoderm (Kenny et al., 1999); however, this expres- Thomas, 1997; reviewed in Eisen, 1998, 1999; Jackman et sion seems more comparable to expression of amphioxus al., 2000). Moreover, Sonic hedgehog (shh), which can and vertebrate homologs in the dorsal presumptive neuro- induce neurogenin expression and is important for motor ectoderm than to expression outside the CNS (see below). neuron formation (Blader et al., 1997; reviewed in Eisen, Thus, expression of vertebrate Sox1/2/3 homologs outside 1998), is expressed in the floor plate and underlying noto- the CNS may signify new functions for these genes. chord in both amphioxus (Shimeld, 1999) and vertebrates. Unlike Sox1/2/3 genes, vertebrate neurogenins are ex- Nevertheless, the expression of some downstream genes pressed in migrating neural crest cells, as well as in epi- differs between amphioxus and vertebrates, suggesting that branchial placodes and cranial sensory ganglia (Sommer et downstream genetic programs operating in motor neurons al., 1996; Anderson et al., 1997; Q. Ma et al., 1998, 1999; may differ between amphioxus and vertebrates. For ex- Perez et al., 1999). In contrast, AmphiNeurogenin is not ample, Nkx2.2 and Pax6 are expressed in subsets of motor expressed outside the CNS in a pattern suggestive of mi- neurons in vertebrates (Eisen, 1998), but not in amphioxus grating neural crest or neural crest derivatives. Although et al., et al., (Glardon 1998; Holland 1998). amphioxus appears to have an evolutionary precursor of Vertebrate neurogenins are also expressed in precursors of neural crest in the cells at the edges of the neural plate and sensory neurons and interneurons (Ma et al., 1996). Simi- adjacent nonneural ectoderm (reviewed in Baker and larly, AmphiNeurogenin is expressed in many dorsal cells Bronner-Fraser, 1997), cells in that region do not appear to which are unlikely to be motor neurons (Figs. 5M–5O and migrate as individuals (N. D. Holland et al., 1996; reviewed 5R). Dorsal cells in the cerebral vesicle develop into the in Holland and Holland, 1999). In contrast, neurogenins are lamellar body, a likely homolog of the pineal eye in verte- expressed in epidermal chemosensory tissues not only in brates, and neurons just posterior to it receive input from vertebrates but also in amphioxus and Drosophila. For the frontal photoreceptor (Lacalli, 1996). More posteriorly example, both the timing and the expression of AmphiNeu- are the photoreceptive Joseph cells (Welsch, 1968); Rhode rogenin and AmphiPax6 (but not, as noted above, cells, apparently involved in swimming; and several types AmphiSox1/2/3) in the anterior epidermis, which may be of apparent sensory neurons, some similar to Rohon–Beard chemosensory (Baatrup, 1981; Lacalli et al., 1999a), are cells in lamprey larvae (Bone, 1960). Since only the cerebral similar to those of their vertebrate homologs in the olfac- vesicle and the motor neurons just posterior to it have been et al., et al., mapped in detail (Lacalli et al., 1994; Lacalli, 1996; Lacalli tory epithelium (Cau 1997; Glardon 1998). and Kelly, 1999), many of the AmphiNeurogenin-expressing AmphiNeurogenin is also probably expressed in precursor cells cannot be correlated with particular cell types. cells of the ciliary tuft, near the mouth, for which a sensory function has been proposed (Stokes and Holland, 1995; Lacalli et al., 1999b), while tap, the Drosophila neurogenin Expression of Sox1/2/3 and neurogenins homolog, is expressed in chemosensory organs (Gautier et outside the CNS al., 1997; Ledent et al., 1998). These comparisons suggest an Unlike vertebrate Sox1/2/3 genes, AmphiSox1/2/3 is not inheritance of an ancient and conserved program for differ- expressed outside the amphioxus CNS during embryogen- entiation of chemosensory cells in chordates and arthro- esis. Vertebrate Sox1/2/3 genes are expressed in some pods. tissues with likely amphioxus counterparts, including the In addition, expression of AmphiNeurogenin and the endoderm of the gut, branchial arches, nasal epithelium, amphioxus insulin-like peptide (P. W. H. Holland et al., and gonad (Collignon et al., 1996; Uwanogo et al., 1995; 1997; L. Z. Holland and S. Chan, unpublished) suggests that Ishii et al., 1998; Uchikawa et al., 1999; Wood et al., 1999). cells in the middle region of the simple tubular gut of We did not determine AmphiSox1/2/3 expression in am- amphioxus may be homologous to the pancreatic islet cells phioxus gonads since they do not develop until after meta- of vertebrates, which express insulin and neurogenin3 morphosis. In addition, vertebrate Sox1/2/3 genes are ex- (Sommer et al., 1996; Apelqvist, 1999). However, the Dro-

Copyright © 2000 by Academic Press. All rights of reproduction in any form reserved. 30 Holland et al. sophila neurogenin homolog, tap, is apparently not ex- organism-specific pathways are involved in the first steps of pressed in the gut and its derivatives (Gautier et al., 1997; neurogenesis, a more thorough comparison of the expres- Sommer et al., 1996), suggesting that new roles for neuro- sion and function of genes in presumptive neuroectoderm genins in gut development may have evolved within the in amphioxus and other nonmodel systems is in order. chordates. In summary, expression of Sox1/2/3 and neurogenin in the CNS appears to be highly conserved in evolution. Sox1/2/3 and neurogenin Genes: Insights into Together with the evolutionary conservation of genetic Evolution of Bilaterian Nervous Systems pathways involving BMP2/4 and chordin for distinguishing neural from nonneural ectoderm, these results lend addi- It is now generally accepted that the genetic mechanism tional support to the idea that the dorsal nerve cord of involving antagonism of BMP2/4/dpp by chordin/sog for chordates and the ventral nerve cord of insects are homolo- distinguishing neuroectoderm and nonneural ectoderm is gous structures that evolved from the nerve cord of an evolutionarily conserved in Drosophila and vertebrates. ancestral bilaterian (Holley et al., 1995; Schmidt et al., The present study shows that expression of two genes 1995; De Robertis and Sasai, 1996; Arendt and Nu¨bler-Jung, (Sox1/2/3 and neurogenin) with very early roles in the 1997). presumptive neuroectoderm is conserved between amphioxus and vertebrates and to some extent between amphioxus and Drosophila as well (Bush et al., 1996; Nambu ACKNOWLEDGMENTS and Nambu, 1996; Russell et al., 1996; Q. Ma et al., 1998). Factors controlling expression of Sox1/2/3 homologs in We thank J. M. Lawrence of the University of South Florida, amphioxus and Drosophila have not been determined. Tampa, for providing facilities for summer field research on am- However, in vertebrates, Sox2 transcription is up-regulated phioxus. This research was supported by NSF Grants IBN 96- 309938 and INT 97-07861 to N.D.H. and L.Z.H and NASA–Ames by chordin and suppressed by BMP4 (Sasai et al., 1995; Grant NAG 2-376 to L.Z.H. Mizuseki et al., 1998; reviewed in Chitnis, 1999). Even in sea urchins, which are nonchordate deuterostomes without a neural plate, Sox1/2/3 and BMP2/4 homologs appear to REFERENCES play roles in axial patterning (Angerer and Angerer, 1999, 2000; Kenny et al., 1999). Akazawa, C., Ishibashi, M., Shimizu, C., Nakanishi, S., and In vertebrates, neurogenin appears to be not far down- Kageyama, R. (1995). A mammalian helix-loop-helix factor struc- stream of Sox1/2/3. In amphioxus the two genes are coex- turally related to the product of Drosophila proneural gene pressed in most cells of the presumptive neuroectoderm. In atonal is a positive transcriptional regulator expressed in the vertebrates, Sox1/2/3 homologs cooperate with partners developing nervous system. J. Biol. Chem. 270, 8730–8738. Ambrosetti, D.-C., Basilico, C., and Dailey, L. (1997). Synergistic like Oct3 in activating genes such as FGF which in turn can activation of the fibroblast growth factor 4 enhancer by Sox2 and induce expression of neurogenin-1 and nerve cell differen- Oct-2 depends on protein–protein interactions facilitated by a tiation (Vriz et al., 1996; Ambrosetti et al., 1997; Mizuseki specific spatial arrangement of factor binding sites. Mol. Cell. et al., 1998; Streit et al., 1998). Correspondingly, the Dro- Biol. 17, 6321–6329. sophila Sox1/2/3 homolog Dichaete also appears to func- Anderson, D. J., Groves, A., Lo, L., Ma, Q., Rao., M., Shah, N. M., tion in nerve cell induction (Nambu and Nambu, 1996). and Sommer, L. (1997). Cell lineage determination and the Thus, it seems likely that in the presumptive neuroecto- control of neuronal identity in the neural crest. Cold Spring derm, genetic pathways leading from chordin to Sox1/2/3 Harbor Symp. Quant. Biol. 62, 493–504. and thence to neurogenin predate the deuterostome– Angerer, L. M., and Angerer, R. C. (1999). Regulative development protostome split. In contrast, expression of some genes in of the sea urchin embryo: Signalling cascades and morphogen gradients. Semin. Cell Dev. Biol. 10, 327–334. presumptive neuroectoderm is not conserved between Dro- Angerer, L. M., and Angerer, R. C. (2000). Animal–vegetal axis sophila and vertebrates. For example, neural expression patterning mechanisms in the early sea urchin embryo. Dev. patterns of Snail and Distalless are similar in vertebrates Biol. 218, 1–12, and amphioxus (N. D. Holland et al., 1996; Langeland et al., Apelqvist, Å., Li, H., Sommer, L., Beatus, P., Anderson, D. J., Honjo, 1998), but not in Drosophila (Mann, 1994). Moreover, while T., Hrabe de Angelis, M., Lendahl, U, and Edland, H. (1999). ZicR-1 is induced by chordin and coexpressed with Sox2 in Notch signalling controls pancreatic cell differentiation. Nature the vertebrate neuroectoderm, its Drosophila homolog, 400, 877–881. odd-paired, is a segmentation gene, which influences ex- Arendt, D., and Nu¨bler-Jung, K. (1997). Dorsal or ventral: Similari- pression of genes such as en and wg involved in specifying ties in fate maps and gastrulation patterns in annelids, arthro- neuroblast identity in each segment (Benedyk et al., 1994; pods and chordates. Mech. Dev. 61, 7–21. Baatrup, E. (1981). Primary sensory cells in the skin of amphioxus Natkata et al., 1997; Bhat, 1999). Zic/odd-paired homologs Branchiostoma lanceolatum (P). Acta Zool. 62, 147–157. Wnt have not been cloned from amphioxus, but neither nor Baker, C. V. H., and Bronner-Fraser, M. (1997). The origins of the engrailed genes are segmentally expressed in the nerve cord neural crest. Part II. An evolutionary perspective. Mech. Dev. 69, (L. Z. Holland et al., 1997; Schubert et al., 2000). Thus, 13–29. while the present research helps to address the question Benedyk, M. J., Mullen, J. R., and DiNardo, S. (1994). Odd-paired: raised by Bier (1997) as to what degree conserved versus A zinc finger pair-rule protein required for the timely activation

of engrailed and wingless in Drosophila embryos. Genes Dev. 8, Hirakow, R., and Kajita, N. (1994). Electron microscopic study of 105–107. the development of amphioxus Branchiostoma belcheri tsing- Bhat, K. M. (1999). Segment polarity genes in neuroblast formation tauensei: The neurula and larva. Acta Anat. Nippon 69, 1–13. and identity specification during Drosophila neurogenesis. Holland, L. Z., Holland, P. W. H., and Holland, N. D. (1996). Trends Genet. 21, 472–485. Revealing homologies between body parts of distantly related Bianchi, M. E., and Beltrame, M. (1998). Protein complexes ‘98. animals by in situ hybridization to developmental genes: Am- Flexing DNA: HMG-box proteins and their partners. Am. J. phioxus versus vertebrates. In “Molecular Zoology: Advances, Hum. Genet. 63, 1573–1577. Strategies, and Protocols” (J. D. Ferraris and S. R. Palumbi, Eds.), Bier, E. (1997). Anti-neural-inhibition: A conserved mechanism for pp. 267–282. Wiley–Liss, New York. neural induction. Cell 89, 681–684. Holland, L. Z., Kene, M., Williams, N. A., and Holland, N. D. Blader, P., Fischer, N., Gradwohl, Guillemot, F., and Stra¨hle, U. (1997). Sequence and embryonic expression of the amphioxus (1997). The activity of neurogenin1 is controlled by local cues in engrailed gene (AmphiEn): The metameric pattern of transcrip- the zebrafish embryo. Development 124, 4557–4569. tion resembles that of its segment-polarity homolog in Drosoph- Bone, Q. (1960). The central nervous system in amphioxus. ila. Development 124, 1723–1732. J. Comp. Neurol. 115, 27–51. Holland, L. Z., Venkatesh, T. V., Gorlin, A., Bodmer, R., and Bush, A., Hiromi, Y., and Cole, M. (1996). Biparous: A novel bHLH Holland, N. D. (1998). Characterization and developmental ex- gene expressed in neuronal and glial precursors in Drosophila. pression of AmphiNk2-2, an NK2 class homeobox gene from Dev. Biol. 180, 759–772. amphioxus (Phylum Chordata; Subphylum Cephalochordata). Cau, E., Gradwohl, G., Fode, C., and Guillemot, F. (1997). Mash1 Dev. Genes Evol. 208, 100–105. activates a cascade of bHLH regulators in olfactory neuron Holland, L. Z., and Holland, N. D. (1999). Chordate origins of the progenitors. Development 124, 1611–1621. vertebrate central nervous system. Curr. Opin. Neurobiol. 9, Chiaramello, A., Neuman, K., Palm, K., Metsis, M., and Neuman, 596–602. T. (1995a). Helix-loop-helix transcription factors mediate activa- Holland, N. D., and Holland, L. Z. (1993). Embryos and larvae of Mol. Cell. Biol. tion and repression of the p75LNGFR gene. 15, invertebrate deuterostomes. In “Essential Developmental Biol- 6036–6045. ogy: A Practical Approach” (C. Stern and P. W. H. Holland, Eds.), Chiaramello, A., Soosaar, A., Neuman, T., and Zuber, M. X. pp. 21–32. IRL Press, Oxford. (1995b). Differential expression and distinct DNA-binding speci- Holland, N. D., Panganiban, G., Henyey, E. L., and Holland, L. Z. ficity of ME1a and ME2 suggest a unique role during differentia- (1996). Sequence and developmental expression of AmphiDll, an tion and neuronal plasticity. Mol. Brain Res. 29, 107–118. amphioxus Distal-less gene transcribed in the ectoderm, epider- Chiaramello, A., Neuman, T., Peavy, D. R., and Zuber, M. X. mis and nervous system: Insights into evolution of craniate (1996). The GAP-43 gene is a direct downstream target of the forebrain and neural crest. Development 122, 2911–2920. basic helix-loop-helix transcription factors. J. Biol. Chem. 271, Holland, P. W. H., Garcia-Ferna`ndez, J., Williams, N. A., and 22035–22043. Sidow, A. (1994). Gene duplications and the origins of vertebrate Chitnis, A. B. (1999). Control of neurogenesis—Lessons from frogs, development. Development Suppl. 120, 125–133. fish and flies. Curr. Opin. Neurobiol. 9, 18–25. Holland, P. W. H., Patton, S. J., Brooke, N. M., and Garcia- Collignon, J., Sockanathan, S., Hacker, A., Cohen-Tannoudji, M., Ferna`ndez, J. (1997). Genetic patterning of the ectoderm and Norris, D., Rastan, S., Stevanovic, M., Goodfellow, P. N., and endoderm in amphioxus: From homeobox genes to hormones. In Lovell-Badge, R. (1996). A comparison of the properties of Sox-3 with Sry and two related genes, Sox-1 and Sox-2. Development “Advances in Comparative Endocrinology: Proceedings of the 122, 509–520. 13th International Congress on Comparative Endocrinology” (S. De Robertis, E. M., and Sasai, Y. (1996). A common plan for Kawashima and S. Kikuyama, Eds.), pp. 247–252. Bologna. dorsoventral patterning in Bilateria. Nature 380, 37–40. Holley, S. A., Jackson, P. D., Sasai, Y., Lu, B., De Robertis, E. M., Eisen, J. S. (1998). Genetic and molecular analyses of motoneuron Hoffmann, F. M., and Ferguson, E. L. (1995). A conserved system development. Curr. Opin. Neurobiol. 8, 697–704. for dorsal–ventral patterning in insects and vertebrates involving Eisen, J. S. (1999). Patterning motor neurons in the vertebrate sog and Chordin. Nature 376, 249–377. nervous system. Trends Neurosci. 22, 321–326. Ishii, Y., Rex, M., Scotting, P. J., and Yasugi, S. (1998). Region- Fraidenraich, D., Lang, R., and Basilico, C. (1998). Distinct regula- specific expression of chicken Sox2 in the developing gut and tory elements govern Fgf4 gene expression in the mouse blasto- lung epithelium: Regulation by epithelial–mesenchymal interac- cyst, myotomes, and developing limb. Dev. Biol. 204, 197–209. tions. Dev. Dyn. 213, 464–475. Fode, C., Gradwohl, G., Morin, X., Dierich, A., Lemeur, M., Jackman, W. R., Langeland, J. A., and Kimmel, C. B. (2000). islet Goridis, C., and Guillemot, F. (1998). The bHLH protein neuro- reveals segmentation in the amphioxus hindbrain homolog. Dev. genin 2 is a determination factor for epibranchial placode-derived Biol. 220,16–26. sensory neurons. Neuron 20, 483–494. Jaz´win´ska, A., Rushlow, C., and Roth, S. (1999). The role of brinker Gautier, P., Ledent, V., Massaer, M., Dambley-Chaudie`re, and in mediating the graded response to Dpp in early Drosophila Ghysen, A. (1997). Tap, a Drosophila bHLH gene expressed in embryos. Development 126, 3323–3334. chemosensory organs. Gene 191, 15–21. Kamachi, Y., Uchikawa, M., Collignon, J., Lovell-Badge, R., and Glardon, S., Holland, L. Z., Gehring, W. J., and Holland, N. D. Kondoh, H. (1998). Involvement of Sox1, 2 and 3 in the early and (1998). Isolation and developmental expression of the amphioxus subsequent molecular events of lens induction. Development Pax-6 gene AmphiPax-6: Insights into eye and photoreceptor 125, 2521–2532. evolution. Development 125, 2701–2710. Kenny, A. P., Kozlowski, D., Oleksy, D. W., Angerer, L. M., and Gradwohl, G., Fode, C., and Guillemot, F. (1996). Restricted Angerer, R. C. (1999). SpSoxB1, a maternally encoded transcrip- expression of a novel murine atonal-related bHLH protein in tion factor asymmetrically distributed among early sea urchin undifferentiated neural precursors. Dev. Biol. 180, 227–241. blastomeres. Development 126, 5473–5483.

Kim, C.-H., Bae, Y.-K., Yamanaka, Y., Yamashita, S., Shimizu, T., Mann, R. S. (1994). Engrailed-mediated repression of Ultrabithorax Fujii, R., Park, H.-C., Yeo, S.-Y., Huh, T.-L., Hibi, M., and Hirano, is necessary for the parasegment 6 identity in Drosophila. R. (1997). Overexpression of neurogenin induces ectopic expres- Development 120, 3205–3212. sion of HuC in zebrafish. Neurosci. Lett. 239, 113–116. Marque´s, G., Musacchio, M., Shimell, M. J., Wu¨nnenberg- Korzh, V., Sleptsova, I., Liao, J., He, J., and Gong, Z. (1998). Stapleton, K., Cho, K. W. Y., and O’Connor, M. B. (1997). Expression of zebrafish bHLH genes ngn1 and nrd defines dis- Production of a DPP activity gradient in the early Drosophila tinct stages of neural differentiation. Dev. Dyn. 213, 92–104. embryo through the opposing actions of the SOG and TLD Kusakabe, R., Kusakabe, T., Satoh, N., Holland, N. D., and Holland, proteins. Cell 91, 417–426. L. Z. (1997). Differential gene expression and intracellular McCormick, M. B., Tamimi, R. M., Snider, L., Asakura, A., mRNA localization of amphioxus actin isoforms throughout Bergstrom, D., and Tapscott, S. J. (1996). NeuroD2 and neuroD3: development: Implications for conserved mechanisms of chor- Distinct expression patterns and transcriptional activation po- date development. Dev. Genes Evol. 207, 203–215. tentials within the neuroD gene family. Mol. Cell. Biol. 16, Lacalli, T. C. (1996). Frontal eye circuitry, rostral sensory pathways 5792–5800. and brain organization in amphioxus larvae: Evidence from 3D Metsis, M., Timmusk, T., Allikmets, R., Saarma, M., and Persson, reconstructions. Philos. Trans. R. Soc. London B 351, 243–263. H. (1992). Regulatory elements and transcriptional regulation by Lacalli, T. C., Holland, N. D., and West, J. E. (1994). Landmarks in testosterone and retinoic acid of the rat nerve growth factor the anterior central nervous system of amphioxus larvae. Philos. receptor promoter. Gene 121, 247–254. Trans. R. Soc. London B 344, 165–185. Mizuseki, K., Kishi, M., Matsui, M., Nakanishi, S., and Sasai, Y. Lacalli, T. C., Gilmour, T. H. J., and Hou, S. (1999a). A reexamina- (1998). Xenopus Zic-related-1 and Sox-2, two factors induced by tion of the epithelial sensory cells of amphioxus (Branchios- chordin, have distinct activities in the initiation of neural toma). Acta Zool. 80, 125–134. induction. Development 125, 579–587. Lacalli, T. C., and Kelly, S. J. (1999). Somatic motoneurons in Nakata, K., Nagai, T., Aruga, J., and Mikoshiba, D. (1997). Xenopus Zic3, amphioxus larvae: Cell types, cell position and innervation a primary regulator both in neural and neural crest development. Proc. Natl. Acad. Sci. USA 94, 11980–11985. patterns. Acta Zool. 80, 113–124. Nambu, P. A., and Nambu, J. R. (1996). The Drosophila fish-hook Lacalli, T. C., Gilmour, T. H. J., and Kelly, S. J. (1999b).The oral gene encodes a HMG domain protein essential for segmentation nerve plexus in amphioxus larvae: Function, cell types and and CNS development. Development 122, 3467–3475. phylogenetic significance. Proc. R. Soc. Biol. Sci. Ser. B 266, Panopoulou, G. D., Clark, M. D., Holland, L. Z., Lehrach, H., and 1461–1470. Holland, N. D. (1998). AmphiMBP2/4, an amphioxus bone mor- Langeland, J. A., Tomsa, J. M., Jackman, W. R., and Kimmel, C. B. phogenetic protein closely related to Drosophila decapentaplegic (1998). An amphioxus snail gene: Expression in paraxial meso- and vertebrate BMP2 and BMP4: Insights into evolution of derm and neural plate suggests a conserved role in patterning the dorsoventral axis specification. Dev. Dyn. 213, 130–139. chordate embryo. Dev. Genes Evol. 208, 569–577. Penzel, R., Oschwald, R., Chen, Y., Tacke, L., and Grunz, H. (1997). Ledent, V., Gaillard, F., Gautier, P., Ghysen, A., and Dambly- Characterization and early embryonic expression of a neural Chaudiere, C. (1998). Expression and function of tap in the specific transcription factor xSox3 in Xenopus laevis. Int. J. Dev. gustatory and olfactory organs of Drosophila. Int. J. Dev. Biol. 42, Biol. 41, 667–677. 163–170. Perez, S. E., Rebelo, S., and Anderson, D. J. (1999). Early specifica- Lee, J. E. (1997). Basic helix-loop-helix genes in neural develop- tion of sensory neuron fate revealed by expression and function ment. Curr. Opin. Neurobiol. 7, 13–20. of neurogenins in the chick embryo. Development 126, 1715– Ma, Q., Kintner, C., and Anderson, D. J. (1996). Identification of 1728. neurogenin, a vertebrate neuronal determination gene. Cell 87, Pevny, L. A., Sockanathan, S., Placzek, M., and Lovell-Badge, R. 43–52. (1998). A role for SOX1 in neural determination. Development Ma, Q., Sommer, L., Cserjesi, P., and Anderson, D. J. (1997). Mash1 125, 1967–1978. and neurogenin1 expression patterns define complementary do- Rex, M., Orme, A., Uwanogho, D., Tointon, K., Wigmore, P. M., mains of neuroepithelium in the developing CNS and are corre- Sharpe, P. T., and Scotting, P. J. (1997). Dynamic expression of lated with regions expressing notch ligands. J. Neurosci. 17, chicken Sox2 and Sox3 genes in ectoderm induced to form neural 3644–3652. tissue. Dev. Dyn. 209, 323–333. Ma, Q., Chen, Z., Del Barco Barrantes, I., De La Pompa, J.-L., and Russell, S. R. H., Sanchez-Soriano, N., Wright, C. R., and Ash- Anderson, D. J. (1998). neurogenin1 is essential for the determi- burner, M. (1996). The Dichaete gene of Drosophila melano- nation of neuronal precursors for proximal cranial sensory gan- gaster encodes a SOX-domain protein required for embryonic glia. Neuron 20, 469–482. segmentation. Development 122, 3669–3676. Ma, Q., Fode, C., Guillemot, F., and Anderson, D. J. (1999). Sakai, Y., Hiraoka, Y., Konishi, M., Ogawa, M., and Aiso, S. (1997). Neurogenin1 and Neurogenin2 control two distinct waves of Isolation and characterization of Xenopus laevis xSox-B1 cDNA. neurogenesis in developing dorsal root ganglia. Genes Dev. 13, Arch. Biochem. Biophys. 346, 1–6. 1717–1728. Sasai, Y., Lu, B., Steinbeisser, H., and De Robertis, E. M. (1995). Ma, Y., Niemitz, E. L., Nambu, P. A., Shan, X., Sackerson, C., Regulation of neural induction by the Chd and Bmp-4 antagonis- Fujioka, M., Goto, T., and Nambu, J. R. (1998). Gene regulatory tic patterning signals in Xenopus. Nature 376, 333–336. functions of Drosophila Fish-hook, a high mobility group domain Sasai, Y., and De Robertis, E. M. (1997). Ectodermal patterning in Sox protein. Mech. Dev. 73, 169–182. vertebrate embryos. Dev. Biol. 182, 5–20. Malas, S., Duthie, S., Deloukas, P., and Episkopou, V. (1999). The Schmidt, J., Francois, V., Bier, E., and Kimelman, D. (1995). isolation and high-resolution chromosomal mapping of human Drosophila short gastrulation induces an ectopic axis in Xeno- SOX14 and SOX21, two members of the SOX gene family related pus: Evidence for conserved mechanisms of dorsal–ventral pat- to SOX1, SOX2, and SOX3. Mamm. Genome 10, 934–937. terning. Development 121, 4319–4328.

Schubert, M., Holland, L. Z., and Holland, N. D. (2000). Character- Uchikawa, M., Kamachi, Y., and Kondoh, H. (1999). Two distinct ization of two amphioxus Wnt genes AmphiWnt4 and subgroups of Group B Sox genes for transcriptional activators and AmphiWnt7b with early expression in the developing central repressors: Their expression during embryonic organogenesis of nervous system. Dev. Dyn. 217, 205–215. the chicken. Mech. Dev. 84, 103–120. Shimeld, S. M. (1999). The evolution of the hedgehog gene family in Uwanogho, D., Rex, M., Cartwright, E. J., Pearl, G., Healy, C., chordates: Insights from amphioxus hedgehog. Dev. Genes Evol. Scotting, P. J., and Sharpe, P. T. (1995). Embryonic expression of 209, 40–47. the chicken Sox2, Sox3 and Sox11 genes suggests an interactive Sommer, L., Ma, Q., and Anderson, D. J. (1996). Neurogenins, a role in neuronal development. Mech. Dev. 49, 23–36. novel family of atonal-related bHLH transcription factors, are Vriz, S., Joly, C., Boulekbache, H., and Condamine, H. (1996). putative mammalian neuronal determination genes that reveal Zygotic expression of the zebrafish Sox-19, an HMG box- progenitor cell heterogeneity in the developing CNS and PNS. containing gene, suggests an involvement in central nervous Mol. Cell. Neurosci. 8, 221–241. system development. Mol. Brain Res. 40, 221–228. Stokes, M. D. (1997). Larval locomotion of the lancelet Branchio- Weinstein, D. C., and Hemmeti-Brivanlou, H. (1997). Neural in- stoma floridae. J. Exp. Biol. 200, 1661–1680. duction in Xenopus laevis: Evidence for the default model. Curr. Stokes, M. D., and Holland, N. D. (1995). Embryos and larvae of a Opin. Neurobiol. 7, 7–12. lancelet, Branchiostoma floridae, from hatching through meta- Wegner, M. (1999). From head to toes: The multiple facets of Sox morphosis: Growth in the laboratory and external morphology. proteins. Nucleic Acids Res. 27, 1409–1420. Acta Zool. 76, 105–120. Welsch, U. (1968). Die Feinstruktur der Josephschen Zellen im Streit, A., Sockanathan, S., Pe´rez, L., Rex, M., Scotting, P. J., Sharpe, Gehirn von Amphioxus. Z. Zellforsch. 86, 252–261. P. T., Lovell-Badge, R., and Stern, C. D. (1997). Preventing the Wood, H. B., and Episkopou, V. (1999). Comparative expression of loss of competence for neural induction: HGF/SF, L5 and Sox2. the mouse Sox1, Sox2 and Sox3 genes from pre-gastrulation to Development 124, 1191–1202. early somite stages. Mech. Dev. 86, 197–201. Streit, A., Lee, K. J., Woo, I., Roberts, C., Jessell, T. M., and Stern, Yasui, K., Tabata, S., Ueki, T., Uemura, M., and Zhang, S.-C. (1998). C. D. (1998). Chordin regulates primitive streak development Early development of the peripheral nervous system in a lancelet and the stability of induced neural cells, but is not sufficient for species. J. Comp. Neurol. 393, 415–425. neural induction in the chick embryo. Development 125, 507– Zhang, S.-C., Holland, N. D., and Holland, L. Z. (1997).Topographic 519. changes in nascent and early mesoderm in amphioxus embryo Streit, A., and Stern, C. D. (1999). Neural induction. A bird’s eye studies by Dil labeling and by in situ hybridization for a view. Trends Genet. 15, 20–24. Brachyury gene. Dev. Genes Evol. 206, 532–535. Takke, C., Dornseifer, P., Weiza¨cker, V., and Campos-Ortega, J. A. Zygar, C. A., Cook, T. L., and Grainger, R. M. (1998). Gene (1999). Her4, a zebrafish homologue of the Drosophila neuro- activation during early stages of lens induction in Xenopus. genic gene Espl, is a target of Notch signalling. Development 126, Development 125, 3509–3515. 1811–1821. Thor, S., and Thomas, J. A. B. (1997). The Drosophila islet gene Received for publication November 30, 1999 governs axon pathfinding and neurotransmitter identity. Neuron Revised May 3, 2000 18, 397–409. Accepted May 23, 2000