Expression of Zebrafish Nk2.2 Is Influenced by Sonic Hedgehog

Development 121, 1755-1768 (1995) 1755 Printed in Great Britain © The Company of Biologists Limited 1995

Expression of zebraﬁsh nk2.2 is inﬂuenced by sonic hedgehog/vertebrate hedgehog-1 and demarcates a zone of neuronal differentiation in the embryonic forebrain

Katrin Anukampa Barth and Stephen W. Wilson Developmental Biology Research Centre, Randall Institute, King’s College London, 26-29 Drury Lane, London WC2B 5RL, UK

SUMMARY

We have isolated zebraﬁsh nk2.2, a member of the Nk-2 and eyes. Moreover, cyclops mutant embryos, which family of homeobox genes. nk2.2 is expressed in a continu- initially lack neurectodermal expression of shh/vhh-1, show ous narrow band of cells along a boundary zone demar- a concomitant lack of nk2.2 expression. Together, these cating the location at which two of the earliest nuclei in the results suggest a requirement of shh/vhh-1 protein for the brain differentiate. This band of cells is located within a spatial regulation of nk2.2 expression. few cell diameters of cells expressing the signalling molecule sonic hedgehog/vertebrate hedgehog-1 (shh/vhh- 1). Injection of shh/vhh-1 RNA results in ectopic expression Key words: boundary, Nk2, hedgehog, axial, zebraﬁsh, forebrain, of nk2.2 and concomitant abnormalities in the forebrain neuronal differentiation

INTRODUCTION families, the Nk-2 family is characterised by an additional conserved motif, the Nk-2 domain (Price et al., 1992). This The last few years have seen significant advances in our under- motif, the prototype of which is found in the Drosophila NK- standing of the mechanisms underlying anteroposterior and 2 gene (Kim and Nirenberg, 1989), consists of at least 17 dorsoventral patterning of the hindbrain and spinal cord amino acids located carboxyterminal to the homeobox. Since (Krumlauf et al., 1993; Smith, 1994). However, the morpho- the isolation of the first vertebrate family member Nkx- genesis of more rostral brain regions is less well understood, 2.1/TTF-1 (Guazzi et al., 1990; Lazzaro et al., 1991; Price et and there is still debate over such basic issues as defining the al., 1992), five other family members have been isolated in neural axes in the forebrain. It has been demonstrated that, mice. Nkx-2.1 to Nkx-2.4 are all closely related (Price et al., similar to the hindbrain, distinct neuromeres are present in the 1992; Price, 1993), while Nkx-2.5 and Nkx-2.6 represent more developing forebrain (Puelles et al., 1987). However, contro- divergent members of the family (Lints et al., 1993). In versy remains as to the number and exact positions of the addition to being expressed in the thyroid and lung, Nkx- forebrain neuromeres and whether they correspond to true 2.1/TTF-1 is also transcribed in restricted regions of the segmental subdivisions (Figdor and Stern, 1993; Puelles and forebrain, as is Nkx-2.2 (Lazzaro et al., 1991; Price et al., Rubenstein, 1993; Macdonald et al., 1994). 1992). Nkx-2.5 is thought to be involved in heart development Because of its relative simplicity, the embryonic zebrafish (Lints et al., 1993), while no detailed expression data have been CNS is well suited for studies of early forebrain development. reported for Nkx-2.3, Nkx-2.4 and Nkx-2.6. Zebrafish nk2.2 is By 24 hours of development (h), a simple scaffold of axon most closely related to Nkx-2.2 and the Xenopus gene XeNk2 tracts has been established by a small number of neurons at (Saha et al., 1993). invariant locations within the brain (Chitnis and Kuwada, In this study, we suggest that the signalling molecule 1990; Wilson et al., 1990). We have recently shown that shh/vhh-1 and the transcription factor axial are involved in the neurons that pioneer this scaffold differentiate at boundaries of spatial regulation of expression of nk2.2. shh/vhh-1 is the gene expression domains (Macdonald et al., 1994). For zebrafish homologue of the Drosophila hedgehog gene (Krauss example, the nuclei of the tract of the postoptic commissure et al., 1993; Roelink et al., 1994), and axial (Strähle et al., (nTPOC) and of the medial longitudinal fasciculus (nMLF) 1993) is the homologue of mouse HNF-3β, a member of the both develop and extend axons along the ventral boundary of winged-helix family of transcription factors (Pani et al., 1992; pax6 and rtk1 expression. Lai et al., 1993). shh/vhh-1 and axial/HNF-3β are both In this study, we report the isolation of a member of the Nk- involved in regulating the patterning of midline structures in 2 family of homeobox genes, termed nk2.2, which is expressed mesoderm and in the ventral CNS (Ang and Rossant, 1994; along the boundary zone at which the nTPOC and nMLF dif- Smith, 1994; Strähle and Blader, 1994; Weinstein et al., 1994). ferentiate. Like several other homeobox-containing gene The notion that axial/HNF-3β is a key regulator of floorplate 1756 K. A. Barth and S. W. Wilson development is supported by the finding that ectopic resulted in a spatially identical signal, but gave higher background expression of HNF-3β results in the ectopic appearance of staining. For axial, shh/vhh-1 and hlx-1, full-length cDNA probes floorplate markers (Sasaki and Hogan, 1994; see also Ruiz i were synthesized. To decrease background staining, probes were frac- Altaba and Jessel, 1992). It seems likely that shh/vhh-1 is tionated over a G-50 (Sigma) drip column to remove unincorporated responsible for the induction of axial/HNF-3β in the pre- DIG-UTP. Whole-mount in situ hybridisations were carried out as sumptive floorplate since ectopic expression of shh/vhh-1 described (Xu et al., 1994). After staining with NBT/X-phosphate β (Boehringer Mannheim), embryos were refixed overnight in 4% results in ectopic axial/HNF-3 expression (Echelard et al., paraformaldehyde/PBS, washed in PBS and cleared in 70% glycerol. 1993; Krauss et al., 1993; Roelink et al., 1994), while COS Embryos were dissected from the underlying yolk and mounted in cells secreting shh/vhh-1 induce floorplate differentiation in 70% glycerol for photography. Immunohistochemistry was carried adjacent neuroectoderm (Roelink et al., 1994). out according to standard procedures (Wilson et al., 1990). In zebrafish, analysis of embryos carrying the cyclops mutation has also provided results consistent with the possi- RNA injections bility that shh/vhh-1 and axial/HNF3β regulate floorplate shh/vhh-1 RNA for injections was derived from the pSP64T-shh development. The cyclops mutation affects specification of the plasmid kindly provided by J.-P. Concordet and P. Ingham; see Krauss ventral midline of the CNS such that homozygous mutant et al., 1993. RNA for injections was transcribed in vitro and several embryos lack a floorplate and exhibit fusion of the eyes (Hatta picoliters were injected at a concentration of 0.1 mg/ml into blastomeres of 1- to 4-cell stage embryos using a pressure-pulsed Picospritzer II et al., 1991). The cyclops gene may be involved in the sig- (General Valve Corp.). To assess the extent of chimerism, lacZ RNA nalling pathway between mesoderm and neuroectoderm that was co-injected at a lower concentration of 20 µg/ml. For control injec- specifies ventral CNS cell types (Hatta et al., 1991, 1994). In tions, RNA encoding β-galactosidase was injected at the same concen- agreement with this interpretation, neither shh/vhh-1 nor axial tration (0.1 mg/ml). Analysis of β-galactosidase activity was performed are initially expressed in the neurectoderm, while mesodermal on embryos that had been fixed for 10 minutes in 4% paraformalde- expression of these genes is present (Krauss et al., 1993; hyde, 0.5% glutaraldehyde at stages between 12 and 24 hours. After Strähle et al., 1993). several washes in PBS, 0.1% Tween 20 (Sigma), embryos were rinsed The timing and spatially restricted expression of nk2.2 in buffer A (1 mM MgCl2, 15 mM K3Fe(CN)6, 12 mM K4Fe(CN)6) and incubated at 37¡C in buffer A containing X-gal (Stratagene) to a suggests that this gene may play a role in the regulation of a µ zone of neuronal differentiation within the embryonic zebrafish final concentration of 800 g/ml. After staining, embryos were washed several times, refixed and processed for in situ hybridisation. forebrain. Furthermore, we present evidence suggesting that shh/vhh-1 may be involved in the spatial regulation of nk2.2 expression. We show that nk2.2 expression is initially absent from the neuroectoderm of cyclops mutant embryos which con- RESULTS comitantly lack shh/vhh-1 expression and that overexpression of shh/vhh-1 results in ectopic expression of nk2.2. nk2.2 is homologous to mouse Nkx-2.2 and Xenopus XeNk-2 Through screening a zebrafish neurula stage cDNA library with MATERIALS AND METHODS a fragment of XeNk-2, a single clone containing 1.5 kb of cDNA was isolated and termed nk2.2 based on its homology Fish stocks to murine Nkx-2.2 (Price et al., 1992) and Xenopus XeNk-2 Breeding fish were maintained at 28.5°C and embryos were collected (Saha et al., 1993). The sequence of the 1470 bp nk2.2 clone by natural spawning and staged up to 24h (30 somites) according to shows a single open reading frame with a coding potential of Westerfield (1993); beyond this time, embryonic stage is given as 269 amino acids (Fig. 1A). hours post fertilisation. Cyclops (cycb16) mutant carrier fish were To define the extent of homology between nk2.2 and other obtained from C. Kimmel and C. Nüsslein-Volhard. Nk-2 gene family members, we compared the nk2.2 translation product to other sequences. Within the homeobox and Nk-2 Isolation of the nk2.2 cDNA clone domain, nk2.2 is 100% identical to mouse Nkx-2.2 and To isolate zebrafish NK-2 homologues, primers to the region flanking Xenopus XeNk-2 proteins, and 93% identical to the same the homeobox and Nk-2 domain of the Xenopus XeNk-2 gene (Saha domains of the Drosophila NK-2 protein (Kim and Nirenberg, et al., 1993) were used to amplify a 370 bp cDNA fragment from 1989; Fig. 1B,C). Comparison of nk2.2 to the published 472 Xenopus cDNA (stage 17). The PCR fragment obtained was cloned and used to screen 1.2×106 recombinant clones of a zebrafish neurula bp genomic fragment of Nkx-2.2 shows that, allowing for a stage cDNA library at low stringency (50% formamide, 6× SSC at possible frameshift in the published Nkx-2.2 sequence, nk2.2 37¡C). A single clone containing 1.5 kb of cDNA was obtained, and Nkx-2.2 are 93% identical over the entire published Nkx- subcloned and sequenced using internal primers and the Sequenase 2.2 sequence. Version 2.1 sequencing kit (USB). Sequence data were analysed using The region of high identity (94%) between nk2.2 and the Genetics Computer Group Sequence Analysis Software Package, Xenopus XeNK-2 comprises amino acid sequences both Version 7.0 (Devereux et al., 1984). aminoterminal (20 aa) and carboxyterminal (30 aa) to the domain containing the homeobox and Nk-2 domain. To either In situ hybridisation and immunohistochemistry side of this core region, the sequence diverges and the putative Antisense digoxigenin-labelled RNA probes were synthesized using proteins differ in size, with nk2.2 being 64 amino acids longer the digoxigenin (DIG) RNA labelling kit (Boehringer Mannheim). For nk.2.2, probes either comprising the entire 1.5 kb cDNA clone, than the published XeNk-2 protein. However, 58 amino acids or comprising a 630 bp 3′ region starting immediately downstream of upstream from the reported translational start site of XeNk-2, the homeobox and including the Nk-2 domain gave best results. A there is another in frame ATG, which is identical to the 710 bp probe derived from the 5′ region upstream of the homeobox proposed translational start site for nk2.2. Zebrafish nk2.2 gene 1757

A nk2.2 is expressed from late gastrula in the presumptive forebrain adjacent to cells expressing shh/vhh-1 and axial nk2.2 expression is ﬁrst detected around 95% epiboly (9.5h) as a small patch of cells at the animal pole in the presumptive brain (Fig. 2A). Between bud/1 somite (10h) and 3 somites (11h), the domain of expression is a narrow column of cells at the midline of the condensing neural keel (Fig. 2B,C). At later stages, cavitation of the neural keel bisects this column of cells to generate bilaterally symmetrical stripes of expression. Pre- liminary data indicated that expression of axial (Strähle et al., 1993), shh/vhh-1 (Krauss et al., 1993) and the homeobox-containing gene, hlx-1 (Fjose et al., 1994) may be localized to regions neighbouring nk2.2 expression and so the evolving pattern of nk2.2 expression was examined with respect to the expression domains of these genes. At 5 somites (11.7h), the nk2.2 expression domain can be divided into two components. The rostral domain of expression extends from the anterior end of the neural keel to the mid-diencephalon (Fig. 2E) and lies adjacent and dorsal to cells that express both shh/vhh-1 and hlx-1 (Fig. 2D,F), but anterior to the

Fig. 1. Sequence, gene structure and deduced amino acid sequence of the zebraﬁsh nk2.2 gene. (A) The nucleotide sequence of nk2.2 is shown along with the conceptual translation of the open reading frame. The putative translational initiation site (ATG) is indicated, the homeodomain is boxed, and the Nk-2 motif underlined (thick line). The 3′ polyadenylation consensus sequence is marked by a thin line. (B) Gene structure and partial restriction map of nk2.2. Thick bars at the 5′ and 3′ ends, representing 280 bp and 368 bp respectively, indicate untranslated regions, while the open box represents the protein coding region. The homeobox (light striped box) and Nk-2 domain (dark striped box) are highlighted. (C) Amino acid sequence comparison of the nk2.2 homeobox and Nk-2 domain to other members of the Nk-2 family in mouse, Xenopus and Drosophila as well as to a more distantly related mouse gene, Dlx-1. Percentage of amino acid identity is given for the core 17 amino acids of the Nk-2 domain and to the slightly larger region of 21 amino acid that is identical between nk2.2, Nkx-2.2 and XeNK-2. Adapted from Price et al. (1992) and references therein and Saha et al. (1993). 1758 K. A. Barth and S. W. Wilson

Fig. 2. Comparison of the developmental time course of nk2.2 expression in the rostral brain with that of shh/vhh-1 and axial. Whole-mount embryos hybridised with antisense RNA to nk2.2, shh/vhh-1, axial or hlx-1. Lateral views (except A,B) are shown with rostral to the left. In D- Y, the skin, yolk and eyes have been removed. (A,B) Frontal views (with dorsal up) showing nk2.2 expression (arrowheads) at 95% epiboly (9.5h) (A) and bud/1s (10h) stage (B). Dots outline the yolk plug in A. (C) Lateral view of nk2.2 expression at 3 somites (11h) and 5 somites (11.7h). (D) hlx-1 expression in the forebrain of a 5 somites (11.7h) embryo. (E-Y) Comparison of rostral brain expression domains of nk2.2 (E,H,K,N,Q,T,W), shh/vhh-1 (F,I,L,O,R,U,X) and axial (G,J,M,P,S,V,Y) from 5 somites (11.5h) to 44-48h. The arrowheads in E-G indicate a small groove in the mid-diencephalon at which the cephalic flexure will later form. Arrowhead in Q indicates the gap between rostral and caudal nk2.2 expression domains. In V, the embryo was also labelled with an antisense RNA probe to wnt1 which is expressed in cells beneath the epiphysis (Macdonald et al., 1994). Abbreviations: cb, cerebellum; cf, cephalic flexure; e, epiphysis; fp, floorplate; hy, hypothalamus; mb, midbrain; mdb, mid-diencephalic boundary; or, optic recess; p; anlage of the anterior pituitary; rd and cd, rostral and caudal domains of nk2.2 expression; t, telencephalon; te, tegmentum; III, third ventricle. Scale bar=100 µm Zebrafish nk2.2 gene 1759 domain of axial expression (Fig. 2G). The weaker caudal domains of shh/vhh-1 and axial expression extend further domain of forebrain expression of nk2.2 (Fig. 2E,H) is directly dorsally at the MDB (axial expression expands dorsally several dorsal to cells expressing both shh/vhh-1 and axial in the caudal hours before shh/vhh-1), with the dorsal tip of expression diencephalon and midbrain. The junction between the rostral coming to underlie the anterior epiphysis (Fig. 2L,M,O,P). and caudal domains overlies a small transverse groove in the During these stages, cavitation of the neural keel begins to ventral neuroepithelium (Fig. 2E-G) at which the cephalic generate the ventricular system of the CNS and it becomes flexure will later form (see Fig. 2P), and corresponds to the apparent that the anterior domain of nk2.2 and shh/vhh-1 anterior boundary of both axial expression and the presumptive expression is located directly ventral to the optic recess/third floorplate. We have previously described this position along the ventricle (Fig. 2K,L). The rostral domain of nk2.2 expression rostrocaudal axis as the mid-diencephalic boundary (MDB, thus overlaps the ventralmost cells within the diencephalic Macdonald et al., 1994), which may, at later stages, correspond expression domains of both pax6 and rtk1 (see Macdonald et to the zona limitans interthalamica described in other species al., 1994). (Puelles and Rubenstein, 1993; Rubenstein et al., 1994). By 26-27h, a small gap between the rostral and caudal By 15 somites (16.5h), a dorsally directed deflection in the domains of nk2.2 expression is visible (Fig. 2Q and see Fig. band of nk2.2-expressing cells at the MDB becomes apparent 5G). This gap overlies the narrow dorsally directed finger-like (Fig. 2H). By this stage, shh/vhh-1 is no longer detectable in projection of shh/vhh-1- and axial-expressing cells at the MDB the ventralmost cells of the rostral forebrain (Fig. 2I), and the (Fig. 2R,S, and see Fig. 5H). Throughout later developmental rostralmost domains of shh/vhh-1 and nk2.2 expression stages, the expression domains of the three genes maintain partially overlap. However, within the caudal forebrain, nk2.2 similar spatial relationships as the forebrain undergoes further continues to be restricted to cells dorsal to the domains of both morphogenesis (Fig. 2T-Y). Finally, nk2.2 transcripts are shh/vhh-1 and axial (compare Fig. 2H to Fig. 2I and J). From detected in the anlage of the anterior pituitary (Fig. 2Q) though this stage onwards, the pattern of hlx-1 expression becomes expression is transient and decreases during further develop- complex and highly dynamic (Fjose et al., 1994) and shows no ment (Fig. 2T). obvious correlation with nk2.2 expression (not shown). Between 22 (20h) and 28 somites (23h), the dorsal deflec- Low levels of nk2.2 transcripts are present in the tion of nk2.2 expression at the MDB becomes more pro- hindbrain and in cells ventral to the notochord nounced (Fig. 2K-N). Concurrent with this change, the Although the most prominent site of nk2.2 expression lies within 1760 K. A. Barth and S. W. Wilson

Fig. 3. nk2.2 expression in the hindbrain and in a group of cells ventral to the notochord. Lateral views (A,B,E) with rostral to the left and transverse sections (C,D) of 20-22 somites (19-20h) embryos from which the eyes and yolk have been removed. The alkaline phosphatase colour reaction was developed 5-6 times longer than usual to reveal weak expression. The approximate levels of the sections shown in C and D are indicated in A. (A) Low magnification view of the entire embryo. The arrow indicates very faint staining in the caudal spinal cord, and the arrowhead points to the group of cells shown in D and E. (B) nk2.2 expression in the brain. The arrow indicates the discontinuity in the band of nk2.2-expressing cells. (C) Transverse section through the caudal hindbrain revealing expression in cells adjacent to the floorplate. (D) Transverse section near the hindbrain/spinal cord junction showing nk2.2 expression in cells beneath the notochord and hypochord (arrow). (E) Lateral view of the same group of cells as (D). Abbreviations: fb, forebrain; fp, floorplate; h, hypochord; hb, hindbrain; hy, hypothalamus; mb, midbrain; n, notochord; s, somite; sc, spinal cord; t, telencephalon. Scale bar: A,B=100 µm, C-E=20 µm. the forebrain, lower levels of mRNA were also detected in more of expression at the boundary between midbrain and hindbrain caudal parts of the CNS, as well as in a cluster of cells ventral (Fig. 3B). Within the hindbrain, nk2.2 is expressed in several to the notochord (Fig. 3). Within the CNS, the column of nk2.2- cells to either side of the floorplate (Fig. 3C). expressing cells extends from the ventral optic stalk to the caudal The only site of nk2.2 expression outside the neuroepithe- spinal cord with highest expression rostrally, lower transcript lium is a patch of cells ventral to the hypochord at the levels in the hindbrain and only barely detectable expression in hindbrain/spinal cord boundary (Fig. 3A). Expression is first the spinal cord (Fig. 3A). There is one small gap in this column detected in this location around 15 somites (16.5h), peaks at

Fig. 4. Gene expression boundaries of nk2.2, shh/vhh-1 and axial and the positions of the sections shown in H and I are indicated in C. demarcate sites of neuronal differentiation and axogenesis in the (E-N) Transverse sections. (E-G) nk2.2 (E,F) and shh/vhh-1 (G) forebrain and midbrain. Embryos are hybridised with nk2.2 expression at the level of the nTPOC. The arrowheads in F indicate (A,B,E,F,L,M), shh/vhh-1 (C,G-I), axial (D,J,K,N) antisense RNA immunoreactive processes within the nk2.2 expression domain. and HNK-1 antibody (brown labelling of neurons and axons). (H-J) shh/vhh-1 (H,I), axial (J,K,N) and nk2.2 (L,M) expression at (A-D) Lateral views of sagittal hemisections with rostral to the left the level of the nMLF (H and J are through the rostral part of the and eyes removed. (A) nk2.2 expression with respect to the nTPOC nucleus and I,K and L are through the caudal part of the nucleus). and nMLF. The dark blue alkaline phosphatase reaction product The arrowheads in M and N indicate immunoreactive processes masks the nTPOC in A and C. (B) High magnification of nk2.2 connecting to the ventricle. The section shown in M is from a expression with respect to the nTPOC. The arrowheads indicate the slightly older embryo than in L. Abbreviations: cf, cephalic flexure; course of the axons in the TPOC, and the white arrow indicates hy, hypothalamus; mb, midbrain; mdb, mid-diencephalic boundary; HNK1 labelling within the nk2.2 expression domain. (C,D) nMLF and MLF, the nucleus of the medial longitudinal fasciculus Correlation of shh/vhh-1 (C) and axial (D) expression domains with and its associated tract; nTPOC and TPOC, the nucleus of the tract of the locations of the nTPOC and nMLF. The arrows in D indicate a the postoptic commissure and its associated tract; or, optic recess; t, few axial-expressing cells dorsal and ventral to axons in the TPOC telencephalon. Scale bars for A,C,D,E, G-L and B,F,M,N=100 µm. Zebrafish nk2.2 gene 1761 1762 K. A. Barth and S. W. Wilson

Table 1. Alterations in gene expression following injection perhaps all of the mature nMLF neurons lie just lateral to the of RNA encoding shh/vhh-1 domain and do not express nk2.2 (Fig. 4M). Conversely, many % Embryos affected Severely of the mature neurons lie just within the axial expression abnormal domain and do express this gene (Fig. 4J,K,N) whereas many n* Total MDB MDB+eyes Mb % WT embryos** of the radial processes lie just dorsal to the axial expression nk2.2 114 63 37 26 37 7 domain (compare Fig. 4M to N). The shh/vhh-1 expression shh/vhh-1 59 34 34 66 11 domain in the midbrain does not extend quite as far dorsal as Axial 71 64 64 20 36 2 the axial expression domain with the result that there are hlx-1 12 67 67 33 usually a few non-expressing cells between the shh/vhh-1 Total n= 256 145 101 30 14 107 20 expression domain and the neurons of the nMLF (Fig. 4C,H,I). In total, 256 shh/vhh-1-injected embryos were analysed for changes in the Overexpression of shh/vhh-1 RNA results in pattern of gene expression. ‘total n=’ represents the number of embryos in each category. elevated and ectopic nk2.2 expression n* are the numbers of embryos examined for each gene. The observation that all sites of nk2.2 expression are within ** are the numbers of severely abnormal embryos that were not included in several cell diameters of cells expressing shh/vhh-1 raises the the analysis (see text). Scored embryos were either wild type (WT), or fell into one of three classes of altered expression patterns: ‘MDB’ corresponds to possibility that shh/vhh-1 may be involved in the regulation of changes in gene expression domains at the mid-diencephalic boundary; nk2.2 expression. In order to determine if shh/vhh-1 can induce ‘MDB+eyes’ indicates ectopic gene expression in the eyes in addition to nk2.2, we analysed embryos that ectopically expressed altered expression at the MDB, and ‘Mb’ denotes widespread ectopic gene shh/vhh-1 after injection of synthetic shh/vhh-1 RNA. expression in the midbrain (and sometimes the hindbrain) as well as at the In total, 256 shh-injected embryos were examined for alter- MDB. In a few cases, embryos that showed ectopic expression of nk2.2 in the eyes and at the MDB also exhibited a small patch of ectopic expression in the ations in expression of nk2.2, shh/vhh-1, axial and hlx-1 (see midbrain (see text). Figures are given as percentages. Table 1). More than half of the injected embryos had specific alterations in CNS expression domains (see below), a few had minor deficiencies in the body axis (such as kinked noto- about 20-22 somites (19-20h) (Fig 3D,E) and diminishes by chords), while the remainder did not show any obvious defects. 28-30 somites (23-24h). Because expression is transient, we A further 20 injected embryos showed severely perturbed have not determined the fate of this group of cells. development and were not included in our detailed analysis. Similarly disturbed development was occasionally seen in nk2.2 expression demarcates a zone of neuronal control injections or in the wild-type background. differentiation in the rostral brain nk2.2 and axial expression was noticeably altered in about Boundaries between gene expression domains demarcate the two thirds, and endogenous shh/vhh-1 expression changed in sites at which the first neurons in the forebrain differentiate and one third of injected embryos (Table 1). Because widespread extend axons (Macdonald et al., 1994). The dorsoventral ectopic expression of shh/vhh-1 was not detected, we assume position of nk2.2 expression suggested that it may overlie the that the injected shh/vhh-1 RNA was already degraded by the boundary at which neurons in the nTPOC and the nMLF dif- stage at which embryos were fixed. Almost invariably, alter- ferentiate. To test this possibility, we examined the formation ations in the expression of all genes examined were apparent of the nMLF and nTPOC with respect to sites of nk2.2, at the MDB. For nk2.2, the expression domain at the MDB was shh/vhh-1 and axial expression. broader, extended further dorsal and mRNA levels were Mature neurons of the TPOC differentiate within the nk2.2 elevated as compared to controls (compare Fig. 5A to 2Q). In expression domain (Fig. 4A,B,E,F). Indeed, many HNK1- some cases, the gap between the rostral and caudal domains of immunoreactive radial processes connected to the ventricle lie nk2.2 expression was enlarged (compare Fig. 5D to 2Q and within the nk2.2 expression domain; these processes probably 5G). Ectopic nk2.2 transcripts were also observed in the eyes belong to young neurons that still retain ventricular connec- (see below), and occasionally in the midbrain (Fig. 5D). No tions (Fig. 4F). There is considerable overlap between the ectopic nk2.2 expression was observed posterior to the shh/vhh-1 and nk2.2 expression domains in the rostral midbrain or outside the CNS. forebrain though shh/vhh-1 expression extends further Paralleling the changes observed for nk2.2 expression, axial ventrally and nk2.2 expression extends more laterally into the and shh/vhh-1 expression domains also extended further optic stalk (compare Fig. 4E to G). While at least some of the dorsal, and mRNA levels were higher and detected in a wider neurons of the nTPOC appear to differentiate just within the stripe of cells at the MDB of injected embryos (Fig. 5B,C,E,F). shh/vhh-1 expression domain (compare Fig. 4F to G), mature In severe cases, the width of the band of tissue expressing neurons and axons are positioned at the edge of this domain shh/vhh-1 and axial at the MDB expanded from the 1-2 cells (Fig. 4G). The axons of the TPOC initially trace a course along normally observed (Fig. 5H) to 10 or more cell diameters (Fig. the ventral edge of cells expressing nk2.2 (Fig. 4B) and as they 5E,F,I). Embryos examined for changes in axial expression at approach the mid-diencephalon, they extend into a domain of earlier stages indicated that cells in the mid-diencephalon cells expressing axial (Fig. 4D). A small region not express- expressed the gene earlier in injected embryos than in controls ing axial (Fig. 4D) was usually observed at the point of entry (Fig. 5J). About one third of injected embryos examined for of the leading TPOC axons into the domain of axial expression. axial expression also exhibited ectopic expression in the The nMLF differentiates along the ventral edge of the caudal midbrain and/or hindbrain as has been previously observed domain of nk2.2 expression (Fig. 4A,L,M). While many (Krauss et al., 1993). hlx-1 was also overexpressed at the MDB immunoreactive radial processes, probably belonging to young in 8 out of 12 embryos examined (not shown). Embryos that neurons, lie within the nk2.2 expression domain, most and exhibited changes in gene expression domains also showed Zebrafish nk2.2 gene 1763

Fig. 5. Injection of shh/vhh-1 RNA results in elevated and ectopic expression of nk2.2, axial and shh/vhh-1. Whole-mount embryos with rostral to the left. (A-F) Lateral views of 24h shh/vhh-1-injected embryos showing nk2.2 (A,D), axial (B,E) and shh/vhh-1 (C,F) expression. Eyes have been removed. (D-F) Examples of embryos affected more severely than those in A-C. (G,H) Dorsal views showing the gap in nk2.2 expression at the MDB of uninjected wild-type embryos (G) and the complementary expression of axial at the same location (H). The eyes are removed in H. (I) Dorsal view of expanded axial expression domain at the MDB in an shh/vhh-1-injected embryo. (J) Lateral view of axial expression in an control (left) and shh/vhh-1-injected 12 somite embryos. (K) Detection of β-galactosidase activity (blue) in embryos injected with β- galactosidase-encoding RNA. Some embryos were also examined for nk2.2 expression (eg. dark blue label in the forebrain of embryo at bottom left). (L) Higher magnification of the tail region of embryo seen bottom right in K. Blue cells are positive for β-galactosidase. Abbreviations: bl, blood; cf, cephalic flexure; h, hypochord; hy, hypothalamus; mb, midbrain; mdb, mid-diencephalic boundary; n, notochord; or, optic recess; sc, spinal cord; sk, skin; t, telencephalon; y, yolk; III, third ventricle. Scale bar=100 µm. 1764 K. A. Barth and S. W. Wilson morphological defects in the anterior brain. In particular, the column of neurons to being a much more widely scattered cavity of the third ventricle appeared reduced (compare Fig. 5I group of cells (not shown). and H) and the development of the eyes was abnormal (see To ascertain that changes in gene expression were not due below and Krauss et al., 1993). to the effects of injecting RNA per se, we examined embryos Changes in gene expression were usually not apparent at the injected with RNA encoding β-galactosidase for changes in sites at which the nTPOC and nMLF differentiate. For nk2.2 expression. Of 99 injected control embryos, 94 were instance, nk2.2 expression never expanded ventrally into the morphologically normal with unchanged expression patterns, hypothalamus or floorplate. However, in a few cases, nk2.2 while 5 embryos showed non-specific defects. expression was disrupted in the midbrain and in these embryos We examined the distribution of injected RNA in 42 we also observed disruption of the nMLF from being a tight embryos that had been injected with both RNA encoding

Fig. 6. Ectopic expression of nk2.2 in the optic primordia of shh/vhh-1-injected embryos correlates with impaired eye development. (A-D) Whole-mount 22-24h shh/vhh-1-injected embryos hybridised with antisense RNA to nk2.2. (A,B) Lateral views showing ectopic nk2.2 expression throughout the optic primordia. (A) Focussed at the level of the eye and (B) focussed through the eye and onto the brain. The white arrowhead indicates the normal position of the optic stalk and the arrow indicates the dorsocaudal limit of fusion of the optic primordia to the brain. (C) Ventral view of an shh/vhh-1-injected embryo with ectopic nk2.2 expression in the anterior part of the optic primordia. (D) Frontal view of an shh/vhh-1-injected embryo with nk2.2 expression throughout the optic primordia. (E,F) Dorsal (E) and frontal (F) views of nk2.2 expression in normal 22-26 somites (20-22h) embryos. (G-H) Eye morphology in living normal (G) and shh/vhh-1-injected (H,I) 30h embryos. The lens is reduced in H and absent in I. Ventrorostral eye development and pigment formation is affected in both embryos. Abbreviations: cf, cephalic flexure; ch, choroid fissure; hy, hypothalamus; l, lens; mb, midbrain; mdb, mid-diencephalic boundary; nr, neural retina; op, optic primodia; or, optic recess; os, optic stalk; pe, pigment epithelium; pnr, presumptive neural retina; ppe, presumptive pigment epithelial layer; se, surface ectoderm; t, telencephalon. Scale bar: A-F=100 µm, G-I=50 µm. Zebrafish nk2.2 gene 1765 shh/vhh-1 and β-galactosidase by assaying the distribution of and pigment layers of the retina (compare Fig. 6D with F). enzyme activity. In all cases, β-galactosidase-positive cells Indeed, the abnormal optic primodia of injected embryos more were widely distributed throughout the embryo (Fig. 5K) and closely resembled the undifferentiated optic vesicles of much detected in all tissue layers (Fig. 5L). younger normal embryos. Possibly as a consequence of abnormal optic cup formation, the lens was frequently reduced Ectopic expression of nk2.2 in the eyes of shh/vhh- in size or sometimes even absent from the eyes of injected 1-injected embryos correlates with abnormal eye embryos (Fig. 6G-I). development In normal embryos, nk2.2 is expressed in the proximal, ventral cyclops mutant embryos that lack shh/vhh-1 and part of the optic stalk, but not within the eyes (Fig. 6E,F). In axial expression in the neuroectoderm exhibit a contrast, 41% of shh/vhh-1-injected embryos in which nk2.2 concomitant loss of nk2.2 expression expression was altered, exhibited ectopic nk2.2 expression in That overexpression of shh/vhh-1 leads to ectopic induction of the eyes. The extent of this expression was variable, sometimes nk2.2 suggests that shh/vhh-1 may be required for the normal being detected throughout the optic primodia (Fig. 6A,D), in induction of nk2.2 expression. To investigate this possibility, other cases restricted to more medial and ventral parts of the we examined nk2.2 expression in embryos homozygous for the developing eyes (Fig. 6C). Although nk2.2 expression spread cyclops mutation. The cyclops mutation prevents specification laterally into the eyes of injected embryos, it was never of the ventral midline in the CNS (Hatta et al., 1991) and, at detected in the hypothalamus or within dorsal regions of the early stages, mutant embryos do not express shh/vhh-1 within telencephalon (Fig. 6A,C,D). the CNS (Krauss et al., 1993). In embryos that exhibited ectopic nk2.2 expression in the nk2.2 expression was absent from the forebrain of all cyclops optic primordia, normal eye development was impaired, and mutant embryos examined between 10 somites (14h) and 24 eyes remained fused to the brain (Fig. 6A,B,D). The area of somites (21h) (n=23; Fig. 7A,B,C). shh/vhh-1 and axial fusion extended dorsocaudally from the normal position of the expression were also absent at comparable stages confirming optic stalk to near the MDB (Fig. 6A,B). In addition, the optic previous results (Krauss et al., 1993; Strähle et al., 1993). primordia frequently failed to invaginate to form an optic cup However, in 30 somites (24h) and older mutant embryos, a and showed abnormal development of the presumptive neural small patch of cells expressed nk2.2 (12/16 embryos examined),

Fig. 7. Expression of nk2.2, shh/vhh-1 and axial in homozygous mutant cyclops embryos. Lateral views (A,D,E,F) with rostral to the left, and transverse sections (B,C) of embryos hybridised with nk2.2, axial or shh/vhh-1 anti-sense RNA. (A-C) nk2.2 expression in 18 somite (18h) wild-type and cyclops mutant embryos. The transverse sections shown in B and C are at the level of the diencephalon. The small dark patch on the dorsal surface of the embryo in B is an artefact. (D) 30 somites (24h) cyclops mutant embryos hybridised with antisense RNA to nk2.2, axial and shh/vhh-1. The arrowheads indicate a small cluster of axial and shh/vhh-1-expressing cells in the forebrain. (E) shh/vhh-1 expression at the dorsal tip of the mid-diencephalic furrow of a 30 somite (24h) cyclops mutant embryo. (F) nk2.2 expression at the tip of the mid- diencephalic furrow of a 40-44h cyclops mutant embryo. Abbreviations: cb, cerebellum; e, epiphysis; fe, fused eye; l, lens; mdb, mid- diencephalic boundary; mdf, mid-diencephalic furrow; ov, optic vesicle; t, telencephalon; te, tectum. Scale bar=100 µm. 1766 K. A. Barth and S. W. Wilson shh/vhh-1 (14/19) and axial (14/15) at the tip of the furrow that neurons (Jimenez and Modolell, 1993). Several members of the forms in place of the MDB (see Macdonald et al., 1994; Patel basic helix-loop-helix family of transcription factors act as et al., 1994) in cyclops mutant embryos (Fig. 7D-F). proneural genes (Jan and Jan, 1993), while signalling Low levels of nk2.2 expression were also detected in more molecules including Notch and Delta function in the neuro- caudal regions of the CNS of cyclops embryos from about 16 genic pathway (Ghysen et al., 1993). Although nk2.2 somites (14h) (data not shown). Similar observations have expression defines several regions where neurons differentiate, been made for shh/vhh-1 (Krauss et al., 1993) and axial it is unlikely that it functions as a neurogenic gene. For (Macdonald et al., 1994). instance, nk2.2 is expressed in a continuous longitudinal band within the rostral brain at which early neuronal differentiation is only observed in two discrete sites. Therefore, many of the DISCUSSION cells that express nk2.2 do not appear to be in the developmental pathway leading to early neuronal differentiation. We have described the isolation and characterisation of the However, it is possible that nk2.2 functions in combination zebrafish nk2.2 gene. In common with all members of the Nk- with other genes to regulate the distribution of the earliest 2 family of homeobox genes, nk2.2 contains a conserved neurons in the brain (Barth and Wilson, 1994). sequence characteristic for this family, the Nk-2 domain. The Our results indicate that cells at a boundary region are high conservation of the Nk-2 domain suggests that it is distinct from adjacent cells in terms of gene expression, though important for the function of Nk-2 proteins. The nature of this it remains unknown if this distinction extends to differences in function is unknown although it has been suggested that the morphology or cell surface properties, as has been documented Nk-2 domain could be involved in mediating protein-protein for boundary cells between rhombomeres in the hindbrain interactions (Price et al., 1992). (Heyman et al., 1994). Although there are no published nk2.2 is most closely related to mouse Nkx-2.2 (Price et al., descriptions of cell surface proteins restricted to boundary cells 1992) and Xenopus XeNk-2 (Saha et al., 1993). The observed in the forebrain, several such proteins are expressed in spatially homology at the amino acid level appears to be paralleled by restricted domains that respect these boundaries (Allendoerfer the conservation of the expression patterns among nk2.2, Nkx- et al., 1994; Redies and Gänzler, 1994. 2.2 and XeNk-2, although there are some differences between our interpretation of expression patterns and others. However, shh/vhh-1 influences the expression of nk2.2 these differences probably reflect the fact that previous studies All sites of nk2.2 expression in the CNS lie within several cell have not analysed expression patterns in such great detail. diameters of cells that express shh/vhh-1. The observations that Indeed, recent reanalysis of Nkx-2.2 expression in mouse all changes in the pattern of shh/vhh-1 expression are accom- suggests a very close similarity in expression between this gene panied by complementary changes in nk2.2 expression, and and nk2.2 within the developing forebrain (Rubenstein et al., that overexpression of shh/vhh-1 ectopically induces nk2.2, 1994; Rubenstein, personal communication). suggest that secreted shh/vhh-1 may be required for nk2.2 While expression domains of nk2.2, Nkx-2.2 and XeNk-2 expression. Hence, the spatially restricted domain of nk2.2 appear to be similar in the CNS, nk2.2 exhibits one additional expression may arise due to the limited diffusion of shh/vhh- site of expression not reported for the mouse and frog homo- 1 protein within the neuroectoderm. The Drosophila hedgehog logues. This patch of nk2.2-expressing cells is located in a protein has recently been shown to be cleaved into two active region ventral to the hypochord near the hindbrain/spinal cord forms and it is likely that one has short-range, and one has boundary. The transient and weak nature of expression at this longer range activities (Lee et al., 1994). Similar cleavage of site may explain why it has not been described in other species. zebrafish shh/vhh-1 occurs (Lee et al., 1994) though it remains unknown over what range the two protein species may signal nk2.2 expression delineates a zone of neuronal and so it is premature to speculate which of the two proteins differentiation in the rostral brain may be involved in regulating nk2.2 expression. Many of the early neurons in the rostral zebrafish CNS differ- Further support for a possible requirement of shh/vhh-1 for entiate at boundaries between regulatory gene expression the induction of nk2.2 expression is derived from the observed domains (Macdonald et al., 1994). For instance, the nTPOC lack of nk2.2 transcripts in young cyclops mutant embryos, and nMLF are both positioned at the ventral boundary of which lack shh/vhh-1 expression in the CNS. Although recent expression of the receptor tyrosine kinase, rtk1, and the paired results have shown that the mesoderm of cyclopic embryos is box transcription factor, pax6. These observations raised the affected (Thisse et al., 1994), the primary consequence of the possibility that cells at the interface between adjacent mutation appears to be the incorrect specification of ventral expression domains may have an identity distinct from that of midline cells in the CNS (Hatta et al., 1991, 1994). We suggest either of the neighbouring domains (Wilson et al., 1993). nk2.2 that a secondary consequence of the failure to specify ventral is expressed in a band of cells at the interface where both the midline tissue is a failure to induce nk2.2 in more lateral cells. nTPOC and nMLF differentiate suggesting that this gene may In older homozygous cyclops mutant embryos, the partial be involved in the establishment or maintenance of the identity recovery of shh/vhh-1 and axial expression is accompanied by of cells at a zone of neuronal differentiation. late expression of nk2.2. The recovery of ventral midline gene From studies performed mainly in Drosophila, at least two expression in the neuroectoderm of cyclops mutant embryos is classes of genes have been shown to be important in regulat- not understood but suggests that signalling between mesoderm ing neurogenesis; proneural genes influence whether ectoder- and ectoderm may not be completely blocked by the mutation. mal cells become epidermis or neural tissue while neurogenic It also remains unknown if the recovery of gene expression is genes influence which of the neural cells differentiate as accompanied by changes in the phenotype of midline cells. Zebrafish nk2.2 gene 1767

Although we suggest that shh/vhh-1 influences nk2.2 genesis. Over time, there is a gradual dorsal extension in axial expression, other molecules may also be involved in the spatial expression, followed by a comparable change in shh/vhh-1 regulation of expression of this gene since overexpression of expression and complemented by a dorsal deflection of nk2.2 shh/vhh-1 results in spatially restricted ectopic expression of expression either side of the MDB. At least two possibilities nk2.2; thus only a subset of neural cells exposed to shh/vhh-1 could account for the changes in gene expression at the MDB. respond by inducing nk2.2. However, we cannot rule out the Either cells could migrate from ventral regions to more dorsal possibility that spatially restricted ectopic nk2.2 expression positions at the MDB or, alternatively nk2.2, axial and may be explained by position-specific differences in shh/vhh-1 expression may be gradually induced in progres- exogenous shh/vhh-1 RNA or protein stability or processing. sively more dorsal cells within the mid-diencephalon. Indeed, The spatial relationship between cells that express nk2.2 and the precocious expression of axial in dorsal cells at the MDB cells expressing shh/vhh-1 differs slightly between the rostral of shh/vhh-1-injected embryos suggests that dorsal cells may and caudal domains of nk2.2 expression in the forebrain. In the be responsive to inductive signals at stages before axial is caudal domain, cells expressing nk2.2 and shh/vhh-1 are normally expressed. This would suggest that in normal discrete populations whereas rostrally the expression domains embryos, the temporal availability of inductive signals may of these genes overlap. If shh/vhh-1 is involved in the induction contribute to the regulation of gene expression at the MDB. of nk2.2 transcription, then it is of interest to consider why nk2.2 expression is not induced within all shh/vhh-1-express- We thank Phil Ingham, Jean-Paul Concordet, Stefan Krauss, Uwe ing cells. One possibility is that other gene(s) may repress Strähle, Anders Fjose, Denis Duboule and Claudio Stern for probes nk2.2 expression within many of the cells that express shh/vhh- or antibodies, David Grunwald and R. Riggleman for the cDNA 1 and thus it could be the absence of such factors rostrally that library and Nigel Holder, Rachel Macdonald and Roger Patient for allows overlap of nk2.2 and shh/vhh-1 expression. comments on the manuscript. This study was initiated with funds from the Medical Research Council and supported by the Wellcome Trust. Do nk2.2 and shh/vhh-1 expression domains define S. W. was a Science and Engineering Research Council Advanced the rostro-caudal axis of the brain? Research Fellow and is a Wellcome Senior Research Fellow. The databank accession number for nk2.2 cDNA sequence is Temporal analysis of shh/vhh-1 and nk2.2 expression indicates X85977. that both genes are transcribed to the anterior tip of the condensing neural keel. As the forebrain differentiates, it becomes apparent that the anteriormost point of expression lies imme- REFERENCES diately ventral to the optic recess. These observations support the hypothesis that the anterior end of the brain is the optic Allendoerfer, K. L., Tole, S. and Patterson, P. H. (1994). Biochemical stalk region and that the telencephalon is a dorsal structure characterization of FORSE-1, a positionally-restricted proteoglycan in the (Ross et al., 1992; Puelles and Rubenstein, 1993; Hatta et al., developing CNS. Soc. Neurosci. Abstr. 20, 691. 1994; Rubenstein et al., 1994), not a discrete neuromere rostral Ang, S.-L. and Rossant, J. (1994). HNF-3β is Essential for node and notochord formation in mouse development. Cell 78, 561-574. to the diencephalon. Barth, K. A. and Wilson, S. W. (1994). Specification of neuronal identity in Although nk2.2 is expressed further dorsally in rostral the embryonic CNS. Semin. Dev. Biol. 5, 349-358. compared to caudal regions of the brain, we believe that the Chitnis, A. B. and Kuwada, J. Y. (1990). Axonogenesis in the brain of position at which this gene is expressed may be equivalent zebrafish embryos. J. Neurosci. 10, 1892-1905. throughout the CNS in terms of the molecular mechanisms that Devereux, J., Haeberli, P. and Smithies, O. (1984). A Comprehensive set of sequence analysis programs for the VAX. Nucl. Acids Res. 12, 387-395. regulate dorsoventral patterning. Indeed, in the early neural Eagleson, G. W. and Harris, W. A. (1990). Mapping of the presumptive brain keel, there is little difference in the dorsoventral position of regions in the neural plate of Xenopus laevis. J. Neurobiol. 21, 427-440. shh/vhh-1 or nk2.2 expression between the rostral brain and Echelard, Y., Epstein, D. J., St-Jacques, B., Shen, L., Mohler, J., more caudal regions. It is only at later stages that the distinc- McMahon, J. A. and McMahon, A. P. (1993). Sonic Hedgehog, a member tion between more dorsally positioned expression in the dien- of a family of putative signalling molecules is implicated in the regulation of CNS polarity. Cell 75, 1417-1430. cephalon and ventral expression in more caudal regions Figdor, M. and Stern, C. D. (1993). Segmental organization of embryonic becomes apparent. diencephalon. Nature 363, 630-633. Two prominent features of forebrain morphogenesis con- Fjose, A., Izpisua-Belmonte, J. C., Fromental-Ramain, C. and Duboule, D. tribute to the spatial changes in gene expression in the dien- (1994). Expression of the zebrafish gene hlx-1 in the prechordal plate and cephalon. The first is the expansion of the hypothalamus and during CNS development. Development 120, 71-81. Ghysen, A., Dambly-Chaudiere, C., Jan, L. Y. and Jan, Y. N. (1993). Cell the associated development of the cephalic flexure; the second interactions and gene interactions in peripheral neurogenesis. Genes Dev. 7, is the development of the MDB. Fate mapping studies of the 723-733. neural plate in Xenopus suggest that the hypothalamus derives Guazzi, S., Price, M., De Felice, M., Damante, G., Mattei, M.-G. and Di from midline cells (Eagleson and Harris, 1990), and so we Lauro, R. (1990). Thyroid nuclear factor 1 (TTF-1) contains a assume that the hypothalamic cells derive from the early pool homeodomain and displays a novel DNA binding specificity. EMBO J. 9, 3631-3639. of shh/vhh-1-expressing midline cells. The later expansion of Hatta, K., Kimmel, C. B., Ho, R. K. and Walker, C. (1991). The cyclops the hypothalamic region is not well understood in any species mutation blocks specification of the floorplate of the zebrafish central and it is unknown if it occurs by recruitment of more dorsal nervous system. Nature 350, 339-341. forebrain cells or by exaggerated proliferation of the most Hatta, K., Püschel, A. W. and Kimmel, C. B. (1994). Midline signalling in the ventral cells. primordium of the zebrafish anterior central nervous system. Proc. Natl. Acad. Sci. USA 91, 2061-2065. The shift in gene expression associated with the develop- Heyman, I., Kent, A. and Lumsden, A. (1994). Cellular morphology and ment of the MDB takes place after the neural keel has extracellular space at rhombomere boundaries in the chick embryo condensed and continues throughout early forebrain morpho- hindbrain. Devel. Dyn. 198, 241-253. 1768 K. A. Barth and S. W. Wilson

Jan, Y. N. and Jan, L. Y. (1993). HLH proteins, fly neurogenesis, and Roelink, H., Augsburger, A., Heemskerk, J., Korzh, V., Norlin, S., Ruiz i vertebrate myogenesis. Cell 75, 827-830. Altaba, A., Tanaba, Y., Placzek, M., Edlund, T., Jessel, T. M. and Dodd, Jimenez, F. and Modolell, J. (1993). Neural fate specification in Drosophila. J. (1994). Floor plate and motor neuron induction by vhh-1, a vertebrate Curr. Op. Gen. Devel. 3, 626-632. homolog of hedgehog expressed by the notochord. Cell 76, 761-755. Kim, Y. and Nirenberg, M. (1989). Drosophila NK-homeobox genes. Proc. Ross, L. S., Parrett, T. and Easter, S. S. Jr. (1992). Axonogenesis and Natl. Acad. Sci. USA 86, 7716-7720. morphogenesis in the embryonic zebrafish brain. J.Neurosci. 12, 467-482. Krauss, S., Concordet, J.-P. and Ingham, P. (1993). A functionally Rubenstein, J. L. R., Martinez, S., Shimamura, K. and Puelles, L. (1994). conserved homolog of the Drosophila segment polarity gene hh is expressed The embryonic vertebrate forebrain: the prosomeric model. Science 266, in tissues with polarizing activity in zebrafish embryos. Cell 75, 1431-1444. 578-580. Krumlauf, R., Marshall, H., Studer, M., Nonchev, S., Sham, M. H. and Ruiz i Altaba, A. and Jessel, T. M. (1992). Pintallavis, a gene expressed in the Lumsden, A. (1993). Hox homeobox genes and regionalisation of the organizer and midline cells of frog embryos: involvement in the development nervous system. J. Neurobiol. 24, 1328-1340. of the neural axis. Development 116, 81-93. Lai, E., Clark, K. L., Burley, S. K. and Darnell, J. E. Jr. (1993). Hepatocyte Ruiz i Altaba, A., Prezioso, V. R., Darnell, J. E. and Jessell, T. M. (1993). nuclear factor 3/fork head or ‘winged helix’ proteins: a family of Sequential expression of HNF-3β and HNF-3α by embryonic organizing transcription factors of diverse biologic function. Proc. Natl. Acad. Sci. USA centers: the dorsal lip/node, notochord and floorplate. Mech. Devel. 44, 91- 90, 10421-10423. 108. Lazzaro, D., Price, M., De Felice, M. and Di Lauro, R. (1991). The Saha, M. S., Michel, R. B., Goulding, K. M. and Grainger, R. M. (1993). A transcription factor TTF-1 is expressed at the onset of thyroid and lung Xenopus homeobox gene defines dorsal-ventral domains in the developing morphogenesis and in restricted regions of the foetal brain. Development brain. Development 118, 193-202. 113, 1093-1104. Sasaki, H. and Hogan, B. L. M. (1993). Differential expression of multiple Lee, J. J., Ekker, S. C., von Kessler, D. P., Porter, J. A., Sun, B. I. and fork head related genes during gastrulation and axial pattern formation in the Beachy, P. I. (1994). Autoproteolysis in hedgehog protein biogenesis. mouse embryo. Development 118, 47-59. Science 266, 1528-1537. Sasaki, H. and Hogan, B. L. M. (1994). HNF-3β as a regulator of floor plate Lints, T. J., Parsons, L. M., Hartley, L., Lyons, I. and Harvey, R. P. (1993). development. Cell 76, 103-116. Nkx-2.5: a novel murine homeobox gene expressed in early heart progenitor Smith, J. C. (1994). Hedgehog, the Floorplate, and the Zone of Polarizing cells and their myogenic descendants. Development 119, 419-431. Activity. Cell 76, 193-196. Macdonald, R., Xu, Q., Barth, K. A., Mikkola, I., Holder, N., Fjose, A., Strähle, U. and Blader, P. (1994). Early neurogenesis in the zebrafish embryo. Krauss, S. and Wilson, S. W. (1994). Regulatory gene expression FASEB J. 8, 692-698. boundaries demarcate sites of neuronal differentiation and reveal neuromeric Strähle, U., Blader, P., Henrique, D. and Ingham, P. (1993). Axial, a organisation of the zebrafish forebrain. Neuron 13, 1039-1053. zebrafish gene expressed along the developing body axis, shows altered Pani, L., Overdier, A., Porcella, A., Qian, X., Lai, E. and Costa, R. H. expression in cyclops mutant embryos. Genes Dev. 7, 1436-1446. (1992). Hepatocyte nuclear factor 3b contains two transcriptional activation Thisse, C., Thisse, B., Halpern, M. and Postlethwait, J. H. (1994). goosecoid domains, one of which is novel and conserved with the Drosophila fork head expression in neurectoderm and mesoderm is disrupted in zebrafish cyclops protein. Mol. Cell. Biol. 12, 3723-3732. gastrulas. Dev. Biol. 164, 420-429. Patel, C., Rodriquez, L.C. and Kuwada, J.Y. (1994). Axonal outgrowth Weinstein, D. C., Ruiz i Altaba, A., Chen, W. S., Hoodless, P., Prezioso, V. within the abnormal scaffold of brain tracts in a zebrafish mutant. J. R., Jessel, T. M. and Darnell, J. E. Jr. (1994). The winged-helix Neurobiol. 25, 345-360. transcription factor HNF-3β is required for notochord development in the Price, M., Lazzaro, D., Pohl, T., Mattei, M.-G., Ruther, U., Olivo, J.-C., mouse embryo. Cell 78, 575-588. Duboule, D. and Di Lauro, R. (1992). Regional expression of the Westerfield, M. (1993). The Zebrafish Book. University of Oregon Press. homeobox gene Nkx-2.2 in the developing mammalian forebrain. Neuron 8, Wilson, S. W., Placzek, M. and Furley, A. (1993). Border disputes: do 241-255. boundaries play a role in growth cone guidance? Trends in NeuroScience 16, Puelles, L., Amat, J. A. and Martinez-de-la-Torre, M. (1987). Segment- 316-322. related, mosaic neurogenetic pattern in the forebrain and mesencephalon of Wilson, S. W., Ross, L., Parrett, T. and Easter, S. S., Jr. (1990). The early chick embryos: i. topography of AChE-positive neuroblasts up to stage development of a simple scaffold of axon tracts in the brain of the embryonic HH18. J. Comp. Neurol. 266, 247-268. zebrafish, Brachydanio rerio. Development 110, 121-145. Puelles, L. and Rubenstein, J. L. R. (1993). Expression patterns of homeobox Xu, Q., Holder, N., Patient, R. and Wilson, S. W. (1994). Spatially regulated and other putative regulatory genes in the embryonic mouse forebrain expression of three receptor tyrosine kinase genes during gastrulation in the suggest a neuromeric organization. Trends in NeuroScience 16, 472-479. zebrafish. Development 120, 287-289. Redies, C. and Gänzler, S. (1994). Neuromeric expression of R-cadherin and the formation of nuclei in the developing chicken forebrain. Soc. Neurosci. Abstr. 20, 252. (Accepted 17 February 1995)