A fundamental character of the vertebrate brain is itspartition into discrete territories, often referred to as seg-ments, along the anterior–posterior (AP) axis. Segmentationis overt in the hindbrain, where distinct boundaries sepa-rate segments known as rhombomeres. In the diencepha-lon, segmentation is cryptic, and may be better described ascompartmentalisation as it is inferred by landmarks, geneexpression patterns, and cell lineage studies rather thanmarked by reiterated structures (Rubenstein et al., 1998;Larsen et al., 2001). These diencephalic segments areknown as prosomeres, while neural segments are knowncollectively as neuromeres. The spinal cord of vertebrates isalso segmented, in the sense that it includes serially re-peated structures, such as dorsal and ventral root exitpoints. This, however, is conceptually different from rhom-bomeric segmentation in that boundaries do not seem to bepresent. There are thus at least three different types of
1 To whom correspondence should be addressed. Fax: (�44) 118
segmentation in the vertebrate central nervous system,prosomeric, rhombomeric, and spinal, and each has its ownmorphological and developmental character.
Neural segmentation is also prominent in several proto-stome taxa, such as arthropods and annelids, suggestingthat some degree of subdivision of the central nervoussystem into segments is primitive for the Bilateria. Whetherthese primitive segments are homologous to the neuro-meres of the vertebrate brain is, however, highly debatablesince, beyond the basic conservation of AP organisation ofbilaterians reflected by Otx and Hox, there is little evidenceof molecular similarity between the two. This suggests thatneuromeric brain organisation evolved specifically on thevertebrate lineage, a view supported by the observation thattwo outgroups to the chordates, the echinoderms and hemi-chordates, show little evidence of neural segmentation(Benito and Pardos, 1997). A key question in vertebrateevolution, therefore, is how (and from what) have thedifferent types of neural segmentation evolved?
The vertebrates share Phylum Chordata with two othersubphyla, the Cephalochordata (amphioxus) and the Uro-
chordata (ascidians and their relatives). Amphioxus is the
closest living invertebrate relative of the vertebrates (Wadaand Satoh, 1994). It shares primitive vertebrate characterslike the notochord, endostyle, and segmented mesoderm.The central nervous system itself appears similar to that ofvertebrates, though considerably simpler. Both the AP anddorsoventral (DV) axes of the amphioxus neural tube havecomparable gene expression patterns with those of thevertebrate neural tube (Holland et al., 1992, 1996; Williamsand Holland, 1996; Shimeld, 1997a, 1999, 2000; Sharman etal., 1999; Wada et al., 1999). The amphioxus nervoussystem posterior to the cerebral vesicle is also segmented,as shown by the regular repetition of dorsal roots and theexpression of several genes (Bone, 1959, 1960; Jackman etal., 2000; Ferrier et al., 2001b). This has led several authors
to hypothesise homologous relationships between regionsof the amphioxus and vertebrate nervous systems. Forexample, the amphioxus anterior cerebral vesicle is pro-posed to be homologous to the vertebrate diencephalon andmidbrain, while the nerve cord, approximately down to thesomite 8 level, might correspond to the vertebrate hind-brain (Williams and Holland, 1998; Holland and Holland,1999).
These studies show that some basic molecular patterningof the chordate nervous system predates the evolution ofvertebrates. However, there are also major differences innervous system patterning between amphioxus and verte-brates. First, there is little evidence of a telencephalon inamphioxus, since interpretation of the expression of the
FIG. 1. Alignment of AmphiFoxB with other FoxB family members from Mouse (Mm), Xenopus (Xl), zebrafish (zf), and Drosophila (Dm).The forkhead domain is shaded, and two other conserved domains are boxed. GenBank Accession Nos. of protein sequences used are Musmusculus FoxB1: NP_071773; M. musculus FoxB2: NP_032049; Xenopus laevis FoxB1: AAC62623; Danio rerio foxb1.2: AF052246; D. reriofoxb1.1: AF052248; D. melanogaster fd4: P32028; D. melanogaster fd5: P32029. The AmphiFoxB sequence is available from the EMBLdatabase, Accession No. AJ506162.
only telencephalic marker characterised to date (the BF-1ortholog AmphiBF-1) is equivocal (Toresson et al., 1998).Second, homologs of individual prosomeres have not beenidentified in amphioxus. Third, no molecular markers forthe midbrain have been identified in amphioxus, and thereappears to be no equivalent of the vertebrate midbrain–hindbrain boundary (MHB), a key organising centre of theposterior vertebrate brain (Holland et al., 1997, 2000a;Kozmik et al., 1999; Ferrier et al., 2001a). Fourth, the overtsegmentation of vertebrate rhombomeres is absent from theregion of the amphioxus brain proposed to be homologousto the hindbrain (Holland et al., 1992; Knight et al., 2000).
These apparent differences in nervous system organisa-tion between amphioxus and vertebrates may reflect genu-ine vertebrate innovations. They might also be artefacts,resulting from the limited set of amphioxus marker genescurrently characterised, or even lineage-specific losses inamphioxus, as suggested for the MHB (Holland et al., 1997,2000; Kozmik et al., 1999; Ferrier et al., 2001a). To under-stand the evolutionary origin of the complex vertebratebrain, we must discriminate between these possibilities, asonly then can genuine novelty be identified and appropriatehypotheses for the evolution of brain segmentation con-structed. We also need to discriminate between the differ-ent types of vertebrate neural segmentation, since theseprobably arose in different ways and at different times. Toresolve some of these issues, we have characterised am-phioxus orthologs of two further markers of neural devel-
opment. The first, amphioxus FoxB (AmphiFoxB), is ex-pressed throughout the brain and spinal cord early indevelopment but is excluded from the presumptive fore-brain. Later in development, AmphiFoxB expression be-comes restricted to specific regions of the cerebral vesicle.The second, amphioxus single-minded (AmphiSim), is amarker of the posterior diencephalon and midbrain. Theexpression of AmphiFoxB suggests that segmentation ofmuch of the neural plate is regulated by adjacent somites.These observations lead us to propose a model for theevolution of the vertebrate hindbrain in which the keymechanistic change was the transfer of mechanisms con-trolling neural segmentation from extrinsic regulation tomechanisms intrinsic to the neurectoderm. Patterns ofFoxB and Sim gene expression in later development suggestthat amphioxus also has a brain that is subdivided intoregions homologous to the diencephalon and midbrain, andthat the diencephalon is further divided into subregions atthe molecular level.
MATERIALS AND METHODS
Isolation of Amphioxus FoxB and Sim Orthologs
An amphioxus genomic phage library (a gift from J. Garcia-Fernandez and C. Minguillon) was screened with an amphioxusFoxF PCR fragment (F.M. and S.M.S., unpublished data) at 55°C(Church and Gilbert, 1984), then washed twice for 15 min with 2�SSC, 0.1% SDS at 55°C. Hybridising clones were plaque purifiedand analysed by Southern hybridisation. One phage clone includeda 1.29-kb SalI/HindIII fragment encoding a FoxB-related gene. Thisfragment was fully sequenced on both strands.
Degenerate oligonucleotide primers of sequence 5�-GCAC-CGGAYGGNAARATHATG-3� (forward primer) and 5�-TTAGC-ACTGNACCANACCCA-3� (reverse primer) were designed to theSim subset of the bHLH-PAS family and used to amplify by PCR a644-bp fragment from amphioxus cDNA. This fragment was clonedand multiple copies sequenced. All were found to derive from thesame gene.
Isolation of AmphiBF-1
We designed oligonucleotide primers (forward primer: 5�-ATGG-TGAGAACGGAGGACCG-3�, and reverse primer: 5�-CTATCC-CGTTAGGCGAGGTA-3�) to the published cDNA sequence ofAmphiBF-1 (Toresson et al., 1998) and amplified a 1210-bp frag-ment running from the predicted start codon to the terminationcodon of the open reading frame. This was cloned, the sequenceverified to ensure it derived from the same gene as the publishedcDNA, and used to generate an antisense riboprobe for in situhybridisation.
Phylogenetic Analysis
Amino acid sequence alignments were constructed withCLUSTALW (Thompson et al., 1994) and edited by eye to resolvesites of questionable homology. Maximum likelihood phylogenetictrees were constructed by using PUZZLETREE (Strimmer and vonHaessler, 1996). We used one invariable and eight gamma-
FIG. 2. Molecular phylogeny of the FoxB family, showing thatAmphiFoxB lies basal to the vertebrate FoxB genes. The tree hasbeen rooted with the Drosophila FoxB orthologs fd4 and fd5. Thescale indicates number of inferred substitutions per site, and thevalues are percentage quartet puzzling support values.
distributed rates, although varying these parameters did not affecttree topology significantly.
Embryo Collection and Whole-Mount in SituHybridisation
Adult amphioxus (Branchiostoma floridae) were collected fromold Tampa Bay, FL in August of 1998. In vitro fertilisation, embryoculture, and fixation were performed as described (Holland andHolland, 1993). Whole-mount in situ hybridisation was carried outas previously described (Shimeld, 1999).
RESULTS
Cloning and Characterisation of an AmphioxusFoxB Ortholog
We isolated and fully sequenced a 1.29-kb genomic frag-ment encoding an amphioxus Fox superfamily gene. The
amino acid sequence of the longest open reading frame(ORF) was deduced, and initial sequence comparisons byalignment and molecular phylogenetics showed it to bemost closely related to the FoxB group of vertebrate andinvertebrate Fox genes (not shown). We therefore alignedthe sequence to other FoxB family members (Fig. 1). TheORF runs to the 3� end of the genomic clone, suggestingthat a few amino acids at the carboxy terminus of theprotein may be encoded by an adjacent clone. Molecularphylogenetic analysis of the FoxB family shows the positionof this gene basal for the vertebrate FoxB genes (Fig. 2), andwe therefore name it AmphiFoxB. Like characterised verte-brate FoxB genes, AmphiFoxB appears to lack introns in thecoding sequence. Comparison of AmphiFoxB with otherFoxB amino acid sequences showed the forkhead domain tobe highly conserved and identified additional conserveddomains (Fig. 1). These included a short amino acid se-quence (consensus FAIENIIA) that is similar to the eh-1
FIG. 3. Alignment of AmphiSim with Sim family members from Human (Hs), Mouse (Mm), Xenopus (Xl), zebrafish (zf), and Drosophila(Dm). Accession Nos. of sequences used were as follows: HsSim1, XP_004324; HsSim2, XP_009755; MmSim1, NP_035506; MmSim2,NP_035507; zfSim, AAK27261; Xlsim, AAG42690; DmSim, P05709. Alignment of sequences beyond the carboxy-terminal extent ofAmphiSim is not shown. The AmphiSim sequence is available from the EMBL database, Accession No. AJ506161.
domain, a domain found in the N-terminal part of severalhomeodomain proteins and in the C-terminal part of theFoxA proteins (Smith and Jaynes, 1996; Shimeld, 1997b;Williams and Holland, 2000).
Cloning and Characterisation of an AmphioxusSim Ortholog
We isolated a 644-bp fragment by degenerate PCR usingprimers designed to recognise the Sim subgroup of thebHLH-PAS gene superfamily. Following pruning of primer-derived sequence, conceptual translation showed it to en-code a 202-amino-acid sequence with a high level of iden-tity to vertebrate and Drosophila Sim proteins (Fig. 3).Initial phylogenetic analysis of the bHLH-PAS genesshowed that this sequence grouped robustly with the Simgenes (data not shown). To increase resolution within theSim clade, we redid the analysis with only Sim genes, androoted the resulting tree with Drosophila sim (Fig. 4). Theresults show that the amphioxus sequence is basal to thetwo clades of vertebrate Sim genes, and hence, we name itAmphiSim.
Expression of AmphiFoxB and AmphiSim duringEmbryonic and Larval Development
The expression of AmphiFoxB was examined by whole-mount in situ hybridisation. No expression was detectedprior to the neurula stage (data not shown). At the neurulastage, in embryos with five somites, expression was de-tected in two lateral stripes along the neural plate, with ananterior boundary at the level of the first somite (Fig. 5A). In
later neurulae, these twin stripes resolved into a series ofblocks of expression, separated by narrow regions of low orabsent expression (Fig. 5B). These blocks were aligned withthe somites. Interestingly, at the level of somite 5, no blockof expression was observed, with transcripts instead beingconfined to a faint medial strip adjacent to the floor plate(Fig. 5B). Expression was also activated in presomitic me-soderm at this stage. Expression in the nervous systemcontinued until the eight-somite stage, when seven blockswere visible, one opposite each somite with the exceptionof somite 5. As embryos reached the end of neurulation,expression faded from the neural tube, with the exceptionof a small number of cells at the very posterior end. In allthe stages examined, expression was excluded from themidline.
At 24 h of development, following the completion ofneurulation, expression was maintained in the cerebralvesicle (the swelling at the anterior end of the nerve cord)as a single patch (Figs. 5C and 5D). Expression in poste-rior mesoderm and nerve cord at this stage was stillpresent though fading. At about 36 h of development, thelarval mouth opens and the animal begins to feed. At thisstage, two areas of the cerebral vesicle express Amphi-FoxB: one in the anterior ventral cerebral vesicle justposterior to the neuropore, the second in the posteriorhalf of the cerebral vesicle (Fig. 5E). Both domains haveclear anterior and posterior boundaries. The expression inthe posterior nerve cord was no longer detected at thisstage, and the only site of expression other than thecerebral vesicle was a small population of subectodermalcells at the posterior tip of the embryo, which areprobably homologous to the vertebrate tail bud. Thispattern of expression continued until the latest stageexamined in which larvae had developed two gill slits andan obvious anterior eye spot (Fig. 5F). Expression did notreach the most anterior part of the cerebral vesicle at anystage. To verify the location of AmphiFoxB expression inthe cerebral vesicle, we also examined the expression ofAmphiBF-1, a marker of the anterior cerebral vesicle inlarvae. We identified AmphiBF-1 expression in an iden-tical domain to that already described (Toresson et al.,1998), ventral to the eyes and anterior to the expressionof AmphiFoxB (data not shown).
To refine our understanding of the relationship of theblocks of AmphiFoxB expression to other regional markersin the amphioxus nervous system, we also examined theexpression of AmphiSim in a staged developmental seriesranging from gastrulae to early larvae. No expression ofAmphiSim was detected in gastrulae. In very early neuru-lae, AmphiSim expression was detected as a broad band ofcells at the dorsal midline of the inner (mesendodermal) celllayer (Fig. 6A). By the midneurula stage, this expression hadbecome confined to the notochord, with low levels ofexpression also detected in future dorsal endoderm (Figs. 6Band 6C; and data not shown). In late neurulae, with five toseven somites, expression was only detected in the noto-chord and endoderm by prolonged staining (data not
FIG. 4. Molecular phylogeny of the Sim gene family producedusing maximum likelihood implemented by quartet puzzling. Thesequences used for phylogenetic reconstruction extend from posi-tion 103 to 330 inclusive as shown in Fig. 3. AmphiSim lies basalto the two well-supported clades of vertebrate Sim genes, and thisrelationship is supported by a value of 100. The position of the root,as deduced by preliminary analyses with a wider range of bHLH-PAS sequences, is shown with a dotted line. The scale indicatesnumber of inferred substitutions per site, and the values arepercentage quartet puzzling support values.
shown). Three areas of more intense AmphiSim expressionwere also observed at this stage, in the pharynx roof anteriorto the first somite and lateral to the anterior notochord, inposterior mesendoderm, and in an isolated block of neuralcells adjacent to the centre of the first somite. This patternof AmphiSim expression essentially continues into earlylarvae, with the exception that the anterior pharynx roofexpression is lost by about 30 h of development (Figs. 6Dand 6E).
DISCUSSION
The Evolution of the Vertebrate Hindbrain andRhombomeric Segmentation
Rhombomeric segmentation is found in all living verte-brates and is of fundamental importance to the develop-ment of the vertebrate head. Comparisons between theamphioxus neural tube posterior to the cerebral vesicle and
FIG. 5. Expression of AmphiFoxB during amphioxus embryonic and early larval development. Anterior is to the left. (A) AmphiFoxBis first detected in neurulae as continuous bands of expression in the neural plate on either side of the midline. (B) Embryo with 6/7somites (the first 7 are numbered) in which neural plate expression has resolved into a series of blocks, each aligned with a somite.Somite boundaries on both sides are marked by black lines; boundaries between the blocks of AmphiFoxB expression are marked bywhite lines on the left side of the embryo only. The gap in expression adjacent to somite 5 is arrowed. Expression is also activated inpresomitic mesoderm (PSM) at this stage. (C, D) Lateral and dorsal views, respectively, of a 10-somite embryo. Expression in the neuraltube (nt) is now confined to posterior cells and is not segmented. A patch of cells in the cerebral vesicle continues to expressAmphiFoxB at this stage (arrows); expression is also detected in the presomitic mesoderm. (E, F) Heads of larvae at 36 and 48 h,respectively. Expression at these stages has resolved into two patches in the cerebral vesicle. The anterior patch (arrow) is justposterior to the neuropore (NP), and later, to the eyespot (ES), while the posterior patch (double arrowhead), at 36 h, extends to theposterior limit of Hatschek’s pit (HP).
the vertebrate hindbrain have been made previously, pri-marily on the basis of Hox gene expression (Holland et al.,1992; Holland and Holland, 1996, 1999; Wada et al., 1999).These studies have suggested that an extensive region of theamphioxus neural tube, stretching from at least the secondsomite to the seventh, is homologous to the vertebratehindbrain. The expression in amphioxus embryos of motorneuron markers and other genes also suggests segmentalorganisation of cell types in the region, an interpretationsupported by the segmental organisation of dorsal nerveroots in larvae and adults (Bone, 1959, 1960; Jackman et al.,2000; Ferrier et al., 2001b). Evidence, however, for segmen-tation in the form of specification of repeated blocks of cellsseparated by boundaries (like rhombomeres) rather thanindividual cell types and nerve exit points has been lacking,and as such, the organisation of this region of the am-phioxus neural tube more closely resembles the vertebratespinal cord than the hindbrain.
In amphioxus neurulae, we observed seven blocks ofAmphiFoxB expression, one next to each of the first eightsomites with the exception of somite 5. These blocks wererelatively broad, separated by narrow bands of non- orlow-expressing cells, and were aligned with the adjacentsomites. The absence of expression at the level of somite 5is important, as it shows that segmental expression ofAmphiFoxB does not simply reflect a physical imprint ofthe somites on the neural tube (for example, a thicker andthinner neural plate next to somites and boundaries, respec-tively), but instead reflects molecular regionalisation. Thislevel is the site of development of the first major pigmentspot in amphioxus, and segmental expression of othergenes, for example, Mnx and Islet, is altered here. Amphi-FoxB expression is different from these segmentally ex-pressed genes, however, in that it forms broad blocks ofcells rather than isolated, reiterated patches of cells. We donot believe that these blocks of AmphiFoxB-expressingcells can be regarded as homologous to rhombomeres,despite their similar location in the nervous system withrespect to Hox gene expression (Fig. 7A). The first block ofexpression, adjacent to somite 1, includes either only theamphioxus posterior diencephalon homolog, or the dien-cephalon and midbrain homologs, as defined by AmphiSimand AmphiOtx, and correspondingly cannot be consideredhomologous to rhombomere 1 (Williams and Holland, 1996,1998; Holland and Holland, 1999). Therefore, a one-to-onerelationship between these blocks of expression and rhom-bomeres is not supported. The alignment of AmphiFoxBexpression with adjacent somites, however, gives us animportant insight into differences in neural tube patterningmechanisms between vertebrates and amphioxus, and cor-respondingly into hindbrain evolution. AmphiFoxB expres-sion, in common with other markers of early neural platesegmentation, only resolves into a segmented pattern ofexpression following segmentation of the underlying meso-derm (Jackman et al., 2000; Ferrier et al., 2001b). Indeed,there is no evidence of neural segmentation prior to meso-dermal segmentation, since the segmental expression of
earliest markers of neural segmentation, amphioxus or-thologs of islet and neurogenin, follows molecular andmorphological mesoderm segmentation (Zhang et al., 1997;Holland et al., 2000b; Jackman et al., 2000). These data, andspecifically the alignment of the blocks of AmphiFoxBexpression to somites, suggest that the amphioxus neuraltube posterior to the region homologous to the diencepha-lon develops segmentation under direct, one-to-one controlby adjacent segmented mesoderm (Fig. 7B). This is similarto the mechanisms proposed to pattern vertebrate spinalcord segmentation, in which mesodermal signals regulateadjacent segmental compartmentalisation such that neuro-meric boundaries (as judged by lineage restriction) alignwith the centre of adjacent somites (Detwiler, 1934; Keynesand Stern, 1984; Stern et al., 1991). It is, however, funda-mentally different to the regulation of rhombomeric seg-mentation. Current studies suggest that rhombomeric seg-mentation is regulated by signalling from two areas (Fig.7C; reviewed by Gavalas and Krumlauf, 2000). Posteriorsegmentation, up to the r3/r4 boundary, is regulated byretinoic acid (RA). Considerable evidence shows that thesource of RA is the somitic mesoderm adjacent to theposterior hindbrain and spinal cord, and suggests that RAacts in a graded manner with progressively lower concen-trations more anteriorly (Dupe and Lumsden, 2001). Inter-pretation of this primary signal appears to be via mecha-nisms intrinsic to the neurectoderm, acting through Krox-20, MafB/kr, and Hoxa1, and results in both segmentationand specification of segment identity (Dupe and Lumsden,2001). Cells anterior to the hindbrain also have a regulatoryrole. Midbrain cells express Cyp26, which encodes anenzyme that inactivates RA and presumably acts as an RAsink, maintaining an RA gradient and protecting anteriorrhombomeres from RA overexposure (Swindell et al., 1999).The MHB is also a source of signals that regulate theidentity of anterior rhombomeres and the position ofboundaries between them (Irving and Mason, 2000).
Based on these comparisons, we propose that a majorinnovation underlying the evolution of the vertebrate brainwas the transfer of mechanisms regulating segmentation ofthe brain posterior to the diencephalon from extrinsic,vertical signalling (that is, direct induction of segments byadjacent segmented tissue, as inferred in amphioxus) tohorizontal signalling within the neurectoderm. The keycandidate molecule in this transfer is RA. It is important tonote that regulation of neural Hox expression by RA hasbeen previously demonstrated in both amphioxus and as-cidians, showing that the involvement of RA in regulatingAP neural identity predates vertebrate evolution (Ka-tsuyama et al., 1995; Holland and Holland, 1996). What isnovel, therefore, is not the involvement of RA in AP neuralpatterning, but the elaboration of this system to includeregulation of segmentation as well as identity. This newmechanism presumably supplanted the ancestral, seg-mented mesoderm-derived mechanism and involved therecruitment of new factors such as Krox-20, which does not
play a role in amphioxus neural segmentation (Knight et al.,2000).
Our results also highlight a previously unrecognisedconnection between two key innovations underlying verte-brate head evolution: unsegmented cranial paraxial meso-derm and hindbrain segmentation. Cryptic segmentation ofvertebrate preotic mesoderm, in the form of somitomeres oras suggested by the head cavities of cartilaginous fish, isdebatable (reviewed by Holland, 2000). Most authors, how-ever, assume that segmentation of this tissue in some formis primitive for vertebrates, although the precise relation-ship between amphioxus mesodermal segments and verte-brate somitomeres and somites is controversial. Irrespec-tive of this controversy, our results suggest thatmesodermal segmentation was primitively required toregulate neural segmentation. Acquisition by vertebrates ofa new method of hindbrain segmentation may have freedhead mesoderm from this constraint, clearing the way forloss of head mesoderm segmentation and further adaptationof the vertebrate head.
Compartmentalisation of the Amphioxus Brain
All vertebrates develop a distinctive midbrain and fore-brain. The forebrain is historically divided into two areas,the anterior telencephalon and the posterior diencephalon.Recent studies, however, have redefined the longitudinalaxis of the brain and shown that these two areas should beprobably not be considered as lying in AP series (Rubensteinet al., 1998; Cobos et al., 2001). Specifically, these authorspropose models detailing the topographic relationship be-tween the early anterior neural plate and the telencephalicand diencephalic structures into which it develops. Thesemodels show the telencephalon as derived exclusively fromthe alar plate, and it could therefore be considered a dorsaloutgrowth rather than an anterior compartment.
Gene expression patterns, anatomical landmarks, andaxon tracts described in the brains of mouse, chick, andzebrafish embryos suggest that the diencephalon can befurther divided into prosomeres (Rubenstein et al., 1998;but see Larsen et al., 2001 for an alternative view). Morerecent studies show the same subdivisions in the embry-onic brains of lampreys, suggesting that this organisation isprimitive for extant vertebrates (Pombal and Puelles, 1999;Murakami et al., 2001). Many authors have addressed thequestion of homology between the amphioxus cerebralvesicle and these various vertebrate brain regions (Hollandet al., 1992, 1996, 1997; Lacalli, 1996; Williams and Hol-land, 1996; Toresson et al., 1998; Kozmik et al., 1999;Jackman et al., 2000; Ferrier et al., 2001a). The emergingconsensus of these studies is that the amphioxus anteriorcerebral vesicle is homologous to the vertebrate diencepha-lon, although the possibility of homology between subre-gions of the amphioxus cerebral vesicle and vertebrateprosomeres has not been explored. The presence of a telen-cephalic homolog in amphioxus has been considered un-likely by most authors (Holland and Holland, 1998, 1999;
Toresson et al., 1998). Posterior to the proposed diencepha-lon homolog is a region that may correspond to the mid-brain, and ultrastructural studies have been interpreted insupport of this assumption (Lacalli, 1996). However, theexpression of genes that specifically mark the posteriormidbrain and MHB of vertebrates, including amphioxusorthologs of Evx, Engrailed, Pax-2/5/8, and Wnt1, has failedto provide evidence for a midbrain or MHB homolog inamphioxus (Holland et al., 1997, 2000a; Kozmik et al.,1999; Ferrier et al., 2001a). This has been particularlysurprising, since ascidians clearly possess a neural domainof Pax-2/5/8 expression separating anterior Otx and poste-rior Hox expression, suggesting that this tripartite structureis ancestral for chordates (Wada et al., 1998).
These studies leave several key questions of brain homol-ogy unresolved. First, is the telencephalon a vertebrateinnovation, deriving from a dorsal outgrowth of the anteriorbrain, or did it evolve from a preexisting anterior neuralcompartment? Second, does amphioxus have a definitivemidbrain? Third, is the amphioxus homolog of the dien-cephalon divisible into further subregions? Our analysis ofAmphiFoxB and AmphiSim expression in the cerebralvesicle suggests answers to some of these questions.
Early in development, AmphiFoxB is widely expressed inthe nervous system, before becoming confined to twodomains in the cerebral vesicle of early larvae. The anteriordomain is just posterior and ventral to the eyes, posterior tothe domain of cells expressing AmphiBF-1 (Toresson et al.,1998). This expression pattern resembles that in vertebrateembryos, where FoxB gene expression has been described inmouse, Xenopus, and zebrafish (Ang et al., 1993; Kaestneret al., 1996; Labosky et al., 1997; Wehr et al., 1997; Grinblatet al., 1998; Odenthal and Nusslein-Volhard, 1998; Gamseand Sive, 2001). In both zebrafish and Xenopus, the FoxBgene fkh5 (also known as fkd5 and foxb1.1 in zebrafish;Odenthal and Nusslein-Volhard, 1998) is widely expressedin the nervous system during early development but isexcluded from the presumptive telencephalon, as shown bycomparison with the telencephalic marker opl (Grinblat etal., 1998; Gamse and Sive, 2001). A second zebrafish FoxBgene, fkd3 (also known as foxb1.2), is also excluded fromthe future telencephalon (Odenthal and Nusslein-Volhard,1998). In mice, the expression of two FoxB genes has beendescribed, Foxb1 (also known as fkh5/MF3/HFH-E5.1) andFoxb2 (also known as fkh4) (Ang et al., 1993; Kaestner et al.,1996; Labosky et al., 1997; Wehr et al., 1997). Foxb1 iswidely expressed in the early neurectoderm, but is excludedfrom the telencephalon before neural tube closure (Ang etal., 1993; Kaestner et al., 1996; Labosky et al., 1997; Wehr etal., 1997). Foxb2 expression is more restricted than that ofFoxb1, but is also excluded from telencephalic cells early indevelopment (Kaestner et al., 1996).
Therefore, a consistent feature of vertebrate FoxB genes iswidespread expression in the diencephalon but exclusionfrom the telencephalon, a territory that is marked by theexpression of another Fox family gene, BF-1 (FoxG1) (Tor-esson et al., 1998). Thus, both amphioxus and vertebrates
have domains of BF-1 and FoxB expression in their anteriorbrains, organised from anterior to posterior. Do these rep-resent homologous territories, and if so, are the amphioxusterritories homologous to the telencephalon and diencepha-lon, respectively? Comparative brain anatomy shows thatsuch an interpretation may be too simplistic, since topo-graphical considerations, as discussed above, suggest thatthe vertebrate telencephalon is a dorsal structure. Con-versely, the AmphiBF-1 territory is situated under the eyespot and should probably be considered ventral; the only
exception to this would be if much of the anterior am-phioxus cerebral vesicle, including ventrally situated cells,in fact had dorsal character, a possibility since amphioxusanterior midline cells do not express the midline signallingmolecules characteristic of their vertebrate counterparts(Shimeld, 1999, 2000). However, in the absence of corrobo-rating anatomical or gene expression data, such a conclu-sion is premature. Therefore, while the domains of BF-1 andFoxB expression are intriguingly similar between am-phioxus and vertebrates and do indicate that the anterior
FIG. 6. Expression of AmphiSim during amphioxus embryonic and early larval development. Anterior is to the left. (A) Early amphioxusneurula in ventral view with the blastopore (bp) indicated. Expressing cells are localised along the dorsal midline of the internal(mesendodermal) cell layer. (B) Neurula with five somites seen in lateral view. The neural plate (np) is indicated. Expressing cells arelocalised to the future notochord, as shown by the optical cross-section shown in (C). This is shown with dorsal to the top, and is througha posterior region of the embryo where the future somites (s) have yet to separate from the future endoderm and notochord (n). (D) Neurulawith six somites. Expressing cells are localised to the posterior mesendoderm (pme), the anterior pharynx roof, and a patch of cells (arrowed)in the future cerebral vesicle adjacent to somite 1. Somites 1 and 2 are indicated. (E) Embryo which has finished neurulation. Expressionis maintained in the cerebral vesicle (arrow) and in posterior mesendoderm. (F) Larva at 36 h of development, in which the mouth has juststarted to open. Expression of AmphiSim is seen in a patch of cells (arrow) in the cerebral vesicle (cv), dorsal to the posterior part ofHatschek’s pit (hp).
FIG. 7. (A) Schematic comparison of gene expression patterns in the neural plate of an amphioxus neurula. Somites are numbered. Notethe overlap of AmphiSim and AmphiOtx expression with the first block of AmphiFoxB expression. (B, C) Diagram of the proposedmechanisms of neural segmentation in amphioxus (B) and vertebrates (C). In amphioxus, neural segments align with the underlyingsomites, suggesting vertical induction of neural segmentation as seen in the vertebrate spinal cord. In contrast, in the vertebrate hindbrain,segmentation is regulated by horizontal signalling from both anterior and posterior. Mesoderm adjacent to the majority of the hindbrainis not segmented. Note this is a schematic and does not necessarily reflect the precise positioning of somites and rhombomeres relative toone another. (D) Comparison of gene expression patterns in the amphioxus cerebral vesicle. The eye spot is in black. Below the expressiondomains are our interpretations of the extent of territories homologous to the diencephalon (D), midbrain (M), and hindbrain (H). Anteriorto the diencephalon is an additional domain (?) of uncertain homology (see text). AmphiBF-1 expression data from Toresson et al. (1998).AmphiOtx expression data from Williams and Holland (1996, 1998). AmphiHox expression data from Wada et al. (1999). Amphioxus Isletexpression data from Jackman et al. (2000). The approximate positions of the floor plate (fp) and where the infundibular cells (i) and lamellarbody (lb) will develop are indicated.
amphioxus cerebral vesicle is divided into discrete territo-ries at the molecular level, extrapolating this to infer thatamphioxus has a definitive telencephalon homolog is alsopremature.
In later development, vertebrate FoxB genes are expressedin specific areas of the diencephalon, midbrain, and hind-brain. Expression is dynamic, varying with developmentalstage and species. For valid comparison between theseexpression domains and gene expression in amphioxus, aspecific criterion must be met: specifically, for an expres-sion domain to be inferred to be primitive for the verte-brates and not a derived character of one vertebrate lineage,it should be present in multiple vertebrate taxa. Zebrafishfoxb1.2/fkd3 is expressed in the hypothalamus, thalamus,and midbrain, as well as by rhombomere boundary cells,while foxb1.1/fkd5 expression is also found in ventraldiencephalic cells (Grinblat et al., 1998; Odenthal andNusslein-Volhard, 1998). Similarly, Xenopus fkd5/Foxb1,following widespread early neural expression, becomes re-stricted to the diencephalon, midbrain, and rhombomere 5,although sublocalisation of expression within the dien-cephalon has not been described (Gamse and Sive, 2001).The detailed sublocalisation of FoxB genes has been mostextensively described in mice. Foxb1 expression has beendetected in the mammillary area, hypothalamus, thalamus,and midbrain, while Foxb2 is expressed in the mammillaryarea, zona limitans intrathalamica, hypothalamus, andmidbrain, as well as in the medullary area of the hindbrain(Ang et al., 1993; Kaestner et al., 1996; Labosky et al., 1997;Wehr et al., 1997). Therefore, two expression domains areconsistent between the three vertebrate lineages in whichFoxB genes have been characterised: the anterior diencepha-lon (specifically the thalamus and hypothalamus) and themidbrain.
AmphiFoxB expression in the amphioxus cerebral vesicleresolves into two domains. The more anterior, as describedabove, is posterior to the domain of AmphiBF-1 expressionand in the area proposed to be homologous to the vertebratediencephalon. Comparison with vertebrate FoxB expressionsuggests that this region of the amphioxus cerebral vesiclemay therefore contain homologous regions to the vertebratethalamus and/or hypothalamus, as these are the consistentanterior sites of expression in divergent vertebrate taxa. Thesecond, more posterior domain of AmphiFoxB expression isanterior to the Hox-expressing region and within theAmphiOtx-expressing region (Williams and Holland, 1998;Wada et al., 1999). Comparison with vertebrate FoxB ex-pression suggests that this area is homologous to thevertebrate midbrain, as this is the more posterior consistentsite of expression in divergent vertebrate taxa. In addition,we also analysed the expression of an amphioxus Simortholog. Two Sim genes have been identified in verte-brates, Sim1 and Sim2, and our analysis shows that theduplication that formed these occurred after the divergenceof the amphioxus and vertebrate lineages. Expression ofvertebrate Sim1 and Sim2 marks the posterior diencephalonand the anterior midbrain in zebrafish, Xenopus, chicks,
and mice (Ema et al., 1996; Fan et al., 1996; Moffett et al.,1996; Yamaki et al., 1996; Fernandez-Teran et al., 1997;Coumailleau et al., 2000; Serluca and Fishman, 2001; Wenet al., 2002). A detailed analysis of transcript localisation inthe diencephalon has only been undertaken in mice, whereSim-2 expression is reported in ventral prosomeres 2–4,Sim-1 expression in ventral prosomeres 1–4, and dorsally inthe mantle of prosomeres 5 and 6 (Fan et al., 1996).
We observed AmphiSim expression in a similar (thoughless extended posteriorly) domain to the posterior domainof AmphiFoxB. This could be hypothesised to be homolo-gous to Sim gene expression in the vertebrate midbrain,diencephalon, or both; to discriminate between these pos-sibilities, we compared these results with the expression ofAmphiOtx, AmphiFoxB, and AmphiHox, and positionedexpression domains relative to the somites in embryos andto the neuropore and Hatschek’s pit in larvae (Figs. 7A and7D). At both embryonic and larval stages, the posteriorlimit of AmphiSim never reaches the posterior limits of theAmphiFoxB and AmphiOtx domains. This order of geneexpression matches the vertebrate brain, suggesting thatthe very posterior part of the amphioxus cerebral vesiclecorresponds to the posterior midbrain of vertebrates. Evi-dence in support of this conclusion comes from the motorneuron marker Islet, which is expressed in the posteriorpart of the cerebral vesicle, within the AmphiOtx domainof expression (Jackman et al., 2000). As suggested by Jack-man et al., this is similar to the expression of vertebrateIslet-1 in the posterior midbrain. The comparison of APlocalisation of FoxB, Sim, and Otx orthologs in vertebratesand amphioxus thus shows that, at least at the molecularlevel, the subdivided diencephalic and midbrain areas of thevertebrate predate the separation of these two lineages andare primitive for vertebrates and amphioxus.
ACKNOWLEDGMENTS
We thank Graham Luke for technical assistance, Marty Cohnand Anthony Graham for discussions, Linda Holland and Jr-Kai Yufor sharing unpublished data on amphioxus Fox genes, and Dr. JohnLawrence and Professor Skip Pierce for generously loaning uslaboratory space in Tampa. We also thank an anonymous reviewerfor constructive comments on the manuscript. This work wassupported by the BBSRC.
REFERENCES
Ang, S. L., Wierda, A., Wong, D., Stevens, K. A., Cascio, S., Rossant,J., and Zaret, K. S. (1993). The formation and maintenance of thedefinitive endoderm lineage in the mouse: Involvement ofHNF3/forkhead proteins. Development 119, 1301–1315.
Benito, J., and Pardos, F. (1997). Hemichordata, chaetognatha, andthe invertebrate chordates. In “Microscopic Anatomy of Inverte-brates.” (E. E. Ruppert, Ed.), Vol. 15, pp. 15–101. Wiley-Liss.
Bone, Q. (1959). The central nervous system in larval acraniates.Q. J. Microsc. Sci. 100, 509–527.
Bone, Q. (1960). The central nervous system in amphioxus.J. Comp. Neurol. 115, 27–64.
Church, G. M., and Gilbert, W. (1984). Genomic sequencing. Proc.Natl. Acad. Sci. USA 81, 1991–1995.
Cobos, I., Shimamura, K., Rubenstein, J. L., Martinez, S., andPuelles, L. (2001). Fate map of the avian anterior forebrain at thefour-somite stage, based on the analysis of quail-chick chimeras.Dev. Biol. 239, 46–67.
Coumailleau, P., Penrad-Mobayed, M., Lecomte, C., Bollerot, K.,Simon, F., Poellinger, L., and Angelier, N. (2000). Characteriza-tion and developmental expression of xSim, a Xenopus bHLH/PAS gene related to the Drosophila neurogenic master genesingle-minded. Mech. Dev. 99, 163–166.
Detwiler, S. R. (1934). An experimental study of spinal nervesegmentation in Amblystoma with reference to the plurisegmen-tal contribution to the brachial plexus. J. Exp. Zool. 67, 395–441.
Dupe, V., and Lumsden, A. (2001). Hindbrain patterning involvesgraded responses to retinoic acid signalling. Development 128,2199–2208.
Ema, M., Morita, M., Ikawa, S., Tanaka, M., Matsuda, Y., Gotoh,O., Saijoh, Y., Fujii, H., Hamada, H., Kikuchi, Y., and Fujii-Kuriyama, Y. (1996). Two new members of the murine Sim genefamily are transcriptional repressors and show different expres-sion patterns during mouse embryogenesis. Mol. Cell. Biol. 16,5865–5875.
Fan, C. M., Kuwana, E., Bulfone, A., Fletcher, C. F., Copeland,N. G., Jenkins, N. A., Crews, S., Martinez, S., Puelles, L.,Rubenstein, J. L., and Tessier-Lavigne, M. (1996). Expressionpatterns of two murine homologs of Drosophila single-mindedsuggest possible roles in embryonic patterning and in the patho-genesis of Down syndrome. Mol. Cell. Neurosci. 7, 1–16.
Fernandez-Teran, M., Piedra, M. E., Simandl, B. K., Fallon, J. F., andRos, M. A. (1997). Limb initiation and development is normal inthe absence of the mesonephros. Dev. Biol. 189, 246–255.
Ferrier, D. E., Minguillon, C., Cebrian, C., and Garcia-Fernandez, J.(2001a). Amphioxus Evx genes: Implications for the evolution ofthe Midbrain-Hindbrain Boundary and the chordate tailbud. Dev.Biol. 237, 270–281.
Ferrier, D. E. K., Brooke, N. M., Panopoulou, G., and Holland,P. W. H. (2001b). The Mnx homeobox gene class defined by HB9,MNR2 and amphioxus AmphiMnx. Dev. Genes Evol. 211, 103–107.
Gamse, J. T., and Sive, H. (2001). Early anteroposterior division ofthe presumptive neurectoderm in Xenopus. Mech. Dev. 104,21–36.
Gavalas, A., and Krumlauf, R. (2000). Retinoid signalling andhindbrain patterning. Curr. Opin. Genet. Dev. 10, 380–386.
Grinblat, Y., Gamse, J., Patel, M., and Sive, H. (1998). Determina-tion of the zebrafish forebrain: Induction and patterning. Devel-opment 125, 4403–4416.
Holland, L. Z., and Holland, N. D. (1996). Expression ofAmphiHox-1 and AmphiPax-1 in amphioxus embryos treatedwith retinoic acid: Insights into evolution and patterning of thevertebrate nerve cord and pharynx. Development 122, 1829–1838.
Holland, L. Z., and Holland, N. D. (1998). Developmental geneexpression in amphioxus: New insights into the evolutionaryorigin of vertebrate brain regions, neural crest and rostrocaudalsegmentation. Am. Zool. 38, 647–658.
Holland, L. Z., and Holland, N. D. (1999). Chordate origins of thevertebrate central nervous system. Curr. Opin. Neurobiol. 9,596–602.
Holland, L. Z., Holland, N. N., and Schubert, M. (2000a). Develop-mental expression of AmphiWnt1, an amphioxus gene in theWnt1/wingless subfamily. Dev. Genes Evol. 210, 522–524.
Holland, L. Z., Kene, M., Williams, N. A., and Holland, N. D.(1997). Sequence and embryonic expression of the amphioxusengrailed gene (AmphiEn): The metameric pattern of transcrip-tion resembles that of its segment-polarity homolog in Drosoph-ila. Development 124, 1723–1732.
Holland, L. Z., Schubert, M., Holland, N. D., and Neuman, T.(2000b). Evolutionary conservation of the presumptive neuralplate markers AmphiSox1/2/3 and AmphiNeurogenin in theinvertebrate chordate amphioxus. Dev. Biol. 226, 18–33.
Holland, N. D., and Holland, L. Z. (1993). Embryos and larvae ofinvertebrate deuterostomes. In “Essential Developmental Biol-ogy: A Practical Approach” (C. D. Stern and P. W. H. Holland,Eds.), pp. 21–32. IRL Press at Oxford University Press, Oxford.
Holland, N. D., Panganiban, G., Henyey, E. L., and Holland, L. Z.(1996). Sequence and developmental expression of AmphiDll, anamphioxus Distalless gene transcribed in the ectoderm, epider-mis and nervous system: Insights into the evolution of craniateforebrain and neural crest. Development 122, 2911–2920.
Holland, P. W. H. (2000). Embryonic development of heads, skel-etons and amphioxus: Edwin S. Goodrich revisited. Int. J. Dev.Biol. 44, 29–34.
Holland, P. W. H., Holland, L. Z., Williams, N. A., and Holland,N. D. (1992). An amphioxus homeobox gene: Sequence conser-vation, spatial expression during development and insights intovertebrate evolution. Development 116, 653–661.
Irving, C., and Mason, I. (2000). Signalling by FGF8 from theisthmus patterns anterior hindbrain and establishes the anteriorlimit of Hox gene expression. Development 127, 177–186.
Jackman, W. R., Langeland, J. A., and Kimmel, C. B. (2000). isletreveals segmentation in the Amphioxus hindbrain homolog.Dev. Biol. 220, 16–26.
Kaestner, K. H., Schutz, G., and Monaghan, A. P. (1996). Expressionof the winged helix genes fkh-4 and fkh-5 defines domains in thecentral nervous system. Mech. Dev. 55, 221–230.
Katsuyama, Y., Wada, S., Yasugi, S., and Saiga, H. (1995). Expres-sion of the labial group Hox gene HrHox-1 and its alterationinduced by retinoic acid in development of the ascidian Halo-cynthia roretzi. Development 121, 3197–3205.
Keynes, R. J., and Stern, C. D. (1984). Segmentation in the verte-brate nervous system. Nature 310, 786–789.
Knight, R. D., Panopoulou, G. D., Holland, P. W. H., and Shimeld,S. M. (2000). An amphioxus Krox gene: Insights into vertebratehindbrain evolution. Dev. Genes Evol. 210, 518–521.
Kozmik, Z., Holland, N. D., Kalousova, A., Paces, J., Schubert, M.,and Holland, L. Z. (1999). Characterization of an amphioxuspaired box gene, AmphiPax2/5/8: Developmental expressionpatterns in optic support cells, nephridium, thyroid-like struc-tures and pharyngeal gill slits, but not in the midbrain-hindbrainboundary region. Development 126, 1295–1304.
Labosky, P. A., Winnier, G. E., Jetton, T. L., Hargett, L., Ryan, A. K.,Rosenfeld, M. G., Parlow, A. F., and Hogan, B. L. (1997). Thewinged helix gene, Mf3, is required for normal development ofthe diencephalon and midbrain, postnatal growth and the milk-ejection reflex. Development 124, 1263–1274.
Lacalli, T. (1996). Frontal eye circuitry, rostral sensory pathwaysand brain organisation in amphioxus larvae: Evidence from 3Dreconstructions. Philos. Trans. R. Soc. Lond. B Biol. Sci. 351,243–263.
Larsen, C. W., Zeltser, L. M., and Lumsden, A. (2001). Boundaryformation and compartition in the avian diencephalon. J. Neu-rosci. 21, 4699–4711.
Moffett, P., Dayo, M., Reece, M., McCormick, M. K., and Pelletier,J. (1996). Characterization of msim, a murine homologue of theDrosophila sim transcription factor. Genomics 35, 144–155.
Murakami, Y., Ogasawara, M., Sugahara, F., Hirano, S., Satoh, N.,and Kuratani, S. (2001). Identification and expression of thelamprey Pax6 gene: Evolutionary origin of the segmented brain ofvertebrates. Development 128, 3521–3531.
Odenthal, J., and Nusslein-Volhard, C. (1998). fork head domaingenes in zebrafish. Dev. Genes Evol. 208, 245–258.
Pombal, M. A., and Puelles, L. (1999). Prosomeric map of thelamprey forebrain based on calretinin immunocytochemistry,Nissl stain, and ancillary markers. J. Comp. Neurol. 414, 391–422.
Rubenstein, J. L., Shimamura, K., Martinez, S., and Puelles, L.(1998). Regionalization of the prosencephalic neural plate. Annu.Rev. Neurosci. 21, 445–477.
Serluca, F. C., and Fishman, M. C. (2001). Pre-pattern in thepronephric kidney field of zebrafish. Development 128, 2233–2241.
Sharman, A. C., Shimeld, S. M., and Holland, P. W. H. (1999). Anamphioxus Msx gene expressed predominantly in the dorsalneural tube. Dev. Genes Evol. 209, 260–263.
Shimeld, S. M. (1997a). Characterisation of amphioxus HNF-3genes: Conserved expression in the notochord and floor plate.Dev. Biol. 183, 74–85.
Shimeld, S. M. (1997b). A transcriptional modification motif en-coded by homeobox and fork head genes. FEBS Lett. 410, 124–125.
Shimeld, S. M. (1999). The evolution of the hedgehog gene family inchordates: Insights from amphioxus hedgehog. Dev. Genes Evol.209, 40–47.
Shimeld, S. M. (2000). An amphioxus netrin gene is expressed inmidline structures during embryonic and larval development.Dev. Genes Evol. 210, 337–344.
Smith, S. T., and Jaynes, J. B. (1996). A conserved region ofengrailed, shared among all en-, gsc-, Nk1-, Nk2- and msh-classhomeoproteins, mediates active transcriptional repression invivo. Development 122, 3141–3150.
Stern, C. D., Jaques, K. F., Lim, T. M., Fraser, S. E., and Keynes, R. J.(1991). Segmental lineage restrictions in the chick embryo spinalcord depend on the adjacent somites. Development 113, 239–244.
Strimmer, K., and von Haessler, S. (1996). Quartet puzzling: Aquartet maximum likelihood method for reconstructing treetopologies. Mol. Biol. Evol. 13, 964–969.
Swindell, E. C., Thaller, C., Sockanathan, S., Petkovich, M., Jessell,T. M., and Eichele, G. (1999). Complementary domains of
retinoic acid production and degradation in the early chickembryo. Dev. Biol. 216, 282–296.
Thompson, J. D., Higgins, D. G., and Gibson, T. J. (1994).CLUSTAL W: Improving the sensitivity of progressive multiplesequence alignment through sequence weighting, positions-specific gap penalties and weight matrix choice. Nucleic AcidsRes. 22, 4673–4680.
Toresson, H., Martinez-Barbera, J. P., Bardsley, A., Caubit, X., andKrauss, S. (1998). Conservation of BF-1 expression in amphioxusand zebrafish suggests evolutionary ancestry of anterior celltypes that contribute to the vertebrate telencephalon. Dev.Genes Evol. 208, 431–439.
Wada, H., Garcia-Fernandez, J., and Holland, P. W. (1999). Colinearand segmental expression of amphioxus Hox genes. Dev. Biol.213, 131–141.
Wada, H., Saiga, H., Satoh, N., and Holland, P. W. H. (1998).Tripartite organization of the ancestral chordate brain and theantiquity of placodes: Insights from ascidian Pax-2/5/8, Hox andOtx genes. Development 125, 1113–1122.
Wada, H., and Satoh, N. (1994). Details of the evolutionary historyfrom invertebrates to vertebrates, as deduced from the sequencesof 18S rDNA. Proc. Natl. Acad. Sci. USA 91, 1801–1804.
Wehr, R., Mansouri, A., de Maeyer, T., and Gruss, P. (1997).Fkh5-deficient mice show dysgenesis in the caudal midbrain andhypothalamic mammillary body. Development 124, 4447–4456.
Wen, H.-J., Wang, Y., Chen, S.-H., and Hu, C.-H. (2002). Expressionof the single-minded gene in zebrafish embryos. Mech. Dev. 110,231–235.
Williams, N. A., and Holland, P. W. H. (1996). Old head on youngshoulders. Nature 383, 490.
Williams, N. A., and Holland, P. W. H. (1998). Molecular evolutionof the brain of chordates. Brain Behav. Evol. 52, 177–185.
Williams, N. A., and Holland, P. W. H. (2000). An amphioxus Emxhomeobox gene reveals duplication during vertebrate evolution.Mol. Biol. Evol. 17, 1520–1528.
Yamaki, A., Noda, S., Kudoh, J., Shindoh, N., Maeda, H.,Minoshima, S., Kawasaki, K., Shimizu, Y., and Shimizu, N.(1996). The mammalian single-minded (SIM) gene: Mouse cDNAstructure and diencephalic expression indicate a candidate genefor Down syndrome. Genomics 35, 136–143.
Zhang, S.-C., Holland, N. D., and Holland, L. Z. (1997). Topo-graphic changes in nascent and early mesoderm in amphioxusembryos studied by DiI labeling and by in situ hybridisation fora Brachyury gene. Dev. Genes Evol. 206, 532–535.
Received for publication April 10, 2002Revised August 28, 2002
Accepted August 28, 2002Published online October 15, 2002
