Discontinuous organization and specification of the

01 (2007) 117–129www.elsevier.com/locate/ydbio

CORE Metadata, citation and similar papers at core.ac.uk

Provided by Elsevier - Publisher Connector

Developmental Biology 3

Discontinuous organization and specification of the lateralfloor plate in zebrafish

Matthias Schäfer 1, Doris Kinzel 2, Christoph Winkler ⁎

Department of Physiological Chemistry I, Biocenter, University of Würzburg, Am Hubland, 97074 Würzburg, Germany

Received for publication 30 March 2006; revised 9 September 2006; accepted 11 September 2006Available online 16 September 2006

Abstract

The floor plate is a signaling center in the ventral neural tube of vertebrates with important functions during neural patterning and axonguidance. It is composed of a centrally located medial floor plate (MFP) and a bilaterally positioned lateral floor plate (LFP). While the role of theMFP as source of signaling molecules like, e.g., Sonic Hedgehog (Shh) is well understood, the exact organization and function of the LFP arecurrently unclear. Based on expression analyses, the one cell wide LFP in zebrafish has been postulated to be a homogenous structure. We insteadshow that the zebrafish trunk LFP is discontinuously arranged. Single LFP cells alternate with p3 neuronal precursor cells, which develop V3interneurons along the anteroposterior (AP) axis. Our mutant analyses indicate that both, formation of LFP and p3 cells require Delta-Notchsignaling. Importantly, however, the two cell types are differentially regulated by Hedgehog (HH) and Nkx2.2 activities. This implicates a novelmechanism of neural tube patterning, in which distinct cell populations within one domain of the ventral neural tube are differently specified alongthe AP axis. We conclude that different levels of HH and Nkx2.2 activities are responsible for the alternating appearance of LFP and p3 neuronalprogenitor cells in the zebrafish ventral neural tube.© 2006 Elsevier Inc. All rights reserved.

Keywords: Neural patterning; Floor plate formation; Hedgehog signaling; Nkx2.2; Homeobox genes

Introduction

The floor plate is an epithelial group of cells positioned in theventral neural tube of vertebrates. It functions as a signalingcenter, which specifies the identity of neuronal subtypes(Echelard et al., 1993; Marti et al., 1995a; Chiang et al.,1996; Ericson et al., 1997a; Briscoe et al., 2000) andoligodendrocytes (Poncet et al., 1996; Pringle et al., 1996;Agius et al., 2004) during early embryonic development. Later,the floor plate guides the trajectory of outgrowing axons(Charron et al., 2003; Lyuksyutova et al., 2003; Bourikas et al.,2005). Floor plate cells are characterized by a cuboidal shapeand expression of specific molecular marker genes. In recent

⁎ Corresponding author. Fax: +49 931 888 4150.E-mail address: [email protected] (C. Winkler).

1 Present address: Institute of Cell Biology, ETH Zurich, Schafmattstr. 18,8093 Zurich, Switzerland.2 Present address: GSF-Institute of Stem Cell Research, Ingolstädter Land-

straβe 1, 85764 Neuherberg, Germany.

0012-1606/$ - see front matter © 2006 Elsevier Inc. All rights reserved.doi:10.1016/j.ydbio.2006.09.018

years, differences in the spatial expression of these markers hasbeen observed in several vertebrate species (Tanaka and Obata,1984; Yamada et al., 1991; Placzek et al., 1991, 1993; Marti etal., 1995b; Odenthal and Nusslein-Volhard, 1998). In mice andzebrafish, for example, the floor plate marker genes shh andnetrin-1 are restricted to the innermost cells of the floor plate(Marti et al., 1995b; Odenthal et al., 2000), while the forkheadgene foxa2 is more broadly expressed (Marti et al., 1995b;Odenthal and Nusslein-Volhard, 1998; Odenthal et al., 2000). Inchicken, Foxa2 is transiently restricted to the innermost floorplate cells, while Shh and Netrin-1 extend more laterally (Martiet al., 1995b; Charrier et al., 2002). This led to the definition ofthe inner cells as medial floor plate (MFP), while the outer cellswere named lateral floor plate (LFP).

The structure and mechanisms of floor plate formation areapparently different in vertebrate species (reviewed in Strähle etal., 2004). While in higher vertebrates, the floor plate widthencompasses several cells, in zebrafish both trunk MFP andLFP are only one cell broad. In chicken and zebrafish, cell fateand mutant analyses have indicated that MFP cells of the neural

https://core.ac.uk/display/82062363?utm_source=pdf&utm_medium=banner&utm_campaign=pdf-decoration-v1

mailto:[email protected]

http://dx.doi.org/10.1016/j.ydbio.2006.09.018

118 M. Schäfer et al. / Developmental Biology 301 (2007) 117–129

tube derive from the embryonic organizer, the Hensens node orshield, respectively, and integrate into the midline of the neuralplate (reviewed in Le Douarin and Halpern, 2000; Strähle et al.,2004; Placzek and Briscoe, 2005). Therefore, a node derivedMFP and a neuroectodermally derived LFP can be distin-guished in chicken and zebrafish. In mouse, in contrast, all floorplate cells are apparently derived from the neuroectoderm(Jeong and Epstein, 2003). Besides the underlying mechanisms,also the signals controlling MFP and LFP formation differ invertebrates. In mice and chicken, the floor plate together withother ventral cells of the neural tube are induced by Sonichedgehog (Shh; reviewed in Placzek and Briscoe, 2005). Inzebrafish, in contrast, Shh only induces LFP and ventralneurons. Induction of the MFP, on the other hand, is controlledby shield-derived factors, like, e.g., Cyclops (Nodal-related 2;Halpern et al., 1997; Sampath et al., 1998; Schauerte et al.,1998; Appel et al., 1999; Tian et al., 2003; Schäfer et al., 2005b;Latimer et al., 2005).

Recently, it has been shown that in the avian neural tube thefloor plate markers Shh and Netrin-1 overlap at least transientlywith Nkx2.2 and Sim1 (Charrier et al., 2002). These markers areexpressed in the p3 neuronal progenitor domain in highervertebrates and differentiated V3 interneurons, respectively. Thep3 neuronal progenitor domain is also characterized byexpression of the homeobox gene Nkx2.9, but lack of Pax6expression (Briscoe et al., 1999; Ericson et al., 1997b; Pabst etal., 1998; Qiu et al., 1998; Shimamura et al., 1995). It has beenshown that cells of the p3 neuronal domain later differentiate intoV3 interneurons (Briscoe et al., 1999; Fan et al., 1996, Yamada etal., 1991; Briscoe et al., 2000). Also in other higher vertebratespecies, floor plate defining markers apparently overlap with thep3 neuronal domain (Placzek et al., 1993; Marti et al., 1995b).This raises the question, how cells of the LFP are organized andformed during early embryonic development. In zebrafish, a p3neuronal domain has not been described so far. However,GABAergic Kolmer–Agdhur neurons that ascend ipsilaterallyfrom the LFP into the ventral longitudinal fasciculus have beenreported (Bernhardt et al., 1992). Therefore, it is currently notclear whether cells between the most ventral MFP and themotoneuronal domain in the vertebrate neural tube are non-neuronal floor plate cells or neuronal cells.

We have addressed this question in the developing zebrafishembryo. In zebrafish, the one cell wide LFP has been describedas a homogenous structure (Odenthal et al., 2000). We, however,show that the zebrafish LFP is a discontinuous row of cells thatalternate with p3 neuronal progenitor cells along the AP axis inthe ventral neural tube. Furthermore, we show that both cellpopulations are dependent on Delta-Notch signaling, but aredifferently regulated by Hedgehog (HH) and Nkx2.2 activities.

Material and methods

Maintenance of wild-type and mutant zebrafish

Zebrafish were maintained as described in Westerfield (2000). Embryoswere obtained from mating of wild-type and heterozygous mutant carriers andraised at 26–30°C until they reached the desired developmental stage. For stagesolder then 42 h post fertilization (hpf), embryos were raised in 0.003%

phenylthiourea (PTU) to inhibit pigment formation. Staging was done accordingto Kimmel et al. (1995). The following mutant alleles were used: you-too (yot,gli-2; ty17a and ty119, Karlstrom et al., 1999), detour (dtr; te370a, Karlstrom etal., 2003), sonic you (syu, shh; t4, Schauerte et al., 1998) and slow muscleomitted (smo, smoothened; b641, Varga et al., 2001), as well as a transgenicgata-2:GFP line (Meng et al., 1997).

Whole-mount antibody staining and in situ hybridization

Antibody staining was essentially performed as described in Westerfield,2000. A polyclonal GABA antibody (1:500; Sigma) was used to detect Kolmer–Agdhur neurons and a mouse monoclonal anti-huC/D antibody (1:1000;Molecular Probes) to detect postmitotic neurons. Proliferating cells in M-phasewere detected with an anti-phospho-histone H3 (PhH3) antibody (1:1000). Forvisualization of antibody reaction, the Vectastain ABC kit (Vector Laboratories)and Fast DAB (Sigma) or an anti-mouse-Cy3 antibody was used. In situhybridization was done as described before in Winkler and Moon (2001). Inembryos where double in situ hybridization was performed, the first colorstaining was photographed before and after staining of the second color tomonitor possible quenching effects. Sections of whole-mount embryos weremanually prepared with a razor blade and mounted in glycerol for photography.For co-localization of p3 and V3 neuronal markers with the neuron-specifichuC/D antigen, 10 μM cryosections were obtained from embryos stained by insitu hybridization and subsequently subjected to immunostaining. The followingprobes were used for in situ hybridization, which were all described before:eng2 (Ekker et al., 1992), f-spondin (Higashijima et al., 1997), foxa2 (Strähleet al., 1993), isl1 (Inoue et al., 1994), isl2 (Appel et al., 1995), nkx2.2a (Barthand Wilson, 1995), nkx2.2b (Schäfer et al., 2005a), olig2 (Park et al., 2002),sim1 (Serluca and Fishman, 2001) and tal2 (Pinheiro et al., 2004).

Cyclopamine treatment

To inhibit endogenous HH signaling, embryos were incubated from shieldstage to 18 s stage in 0.5×10−6, 1×10−6, 2×10−6 and 3×10−6 M cyclopamine(Toronto Research Chemicals) diluted from a 10−2 M stock in DMSO. Thechorion was manually removed prior to incubation.

Morpholino injection

Two morpholino antisense oligonucleotides (Gene Tools) were designedagainst the upstream regions of nkx2.2a (CGTCTTTGTGTTGGTCAACGA-CATG) and nkx2.2b (TCTTTAGGGACATTTTCCAAACCAG). These mor-pholinos were injected at 12 ng/nl (nkx2.2a) and 10 ng/nl (nkx2.2b),respectively, into single blastomeres of 1–2 cell stage embryos. The specificityand functionality of both morpholinos were tested by coinjection of eachmorpholino with the corresponding target sequence fused to GFP (nkx2.2-GFP)and the target sequence containing five mismatches fused to GFP (MM nkx2.2-GFP; see Supplementary material).

Results

Different expression profiles of LFP cells in the zebrafishventral neural tube

In most vertebrate species, the floor plate comprises a medialand a lateral part. While the MFP has been intensivelycharacterized, the organization and function of the LFP iscurrently not clear. In zebrafish, the LFP is described as ahomogenous structure (Odenthal et al., 2000). In chicken, incontrast, LFP markers apparently overlap with markers of thep3 neuronal progenitor domain and differentiated V3 inter-neurons (Charrier et al., 2002). To get insight into the structureof the LFP in zebrafish, we examined the expression of genesspecific for non-neuronal floor plate cells and neuronal

119M. Schäfer et al. / Developmental Biology 301 (2007) 117–129

precursor cells during neurogenesis of the ventral neural tube. Inthe trunk of 18–22 somite (s) stage (18–20 h post fertilization,hpf) zebrafish embryos, the homeobox gene nkx2.2b ishomogenously expressed in the two parallel rows of LFP cells(Fig. 1B; Schäfer et al., 2005a). The winged-helix transcriptionfactor foxa2 (Strähle et al., 1993) is homogenously expressedin the MFP (Fig. 1C), like other known genes, for instancef-spondin (Fig. 1A; Higashijima et al., 1997). However, inthe LFP, different levels of foxa2 expression were detected.

Fig. 1. LFP cells show different expression profiles and identities. (A, B) Dorsal viewand nkx2.2b in the LFP (B). (C) foxa2 is discontinuously expressed in the LFP, with(D) Overlay of a double in situ hybridization showing nkx2.2b (in red) that is strongregions, nkx2.2b detection is quenched. (E) tal2 is expressed in single cells of the Lhybridization with tal2 (blue) and foxa2 (red). foxa2 analysis shows no significant diof tal2 (G). Expression of tal2 only (H) and overlay of tal2 with foxa2 (I) showing coimmunostaining (in brown) is only detected in LFP cells that do not express tal2 (in b(K, L) tal2 expression in the LFP (indicated by arrow) partly co-localizes with the posof nkx2.2b-expressing LFP cells (M) also expresses huC/D (arrow in N). The same trarespectively. LFP, lateral floor plate; MFP, medial floor plate. Scale bar: J, 10 μM; N

Along the AP axis, we observed randomly distributed 1–2 cellbroad areas, in which foxa2was only weakly or not expressed innkx2.2b-positive LFP cells (Figs. 1C, D). Transcription of thebasic–loop–helix gene tal2 (Pinheiro et al., 2004), in contrast,was found only in single cells of the LFP. These cells werediscontinuously arranged along the AP axis (Fig. 1E). A similarexpression was found for gata-2 (Detrich et al., 1995; seeSupplementary material). Double in situ hybridization of foxa2and tal2 showed that tal2-positive cells are located in areas of

s of trunk regions showing homogenous expression of f-spondin in the MFP (A)regions of weak expression levels along the AP axis (indicated by arrowheads).ly expressed within regions of weak foxa2 expression (in blue). In overlappingFP that are discontinuously distributed along the AP axis. (F–I) Double in situfference in fluorescence detection when compared before (F) and after detectionmplementary localization of tal2 in regions of weak foxa2 expression. (J) PhH3lue). Panels A–J are dorsal views of trunk at 18 to 22 s stage, anterior to the left.tmitotic neuronal marker huC/D at 24 hpf (arrows in L). (M, N) A subpopulationnsverse sections are shown for bright-field (K, M) and Cy3 fluorescence (L, N),, 5 μM.


weak or no foxa2 expression (Figs. 1F–I). PhH3 antibodystaining for cells in M-phase showed that tal2-expressing cellsdo not divide at the 20 s (18–20 hpf) stage. The alternatingfoxa2-expressing cells, however, proliferate extensively (Fig.1J). This indicates that tal2-positive cells are postmitotic andhave started to differentiate at this stage of development.Moreover, the postmitotic neuron marker huC/D (Marusich etal., 1994) could be detected in a fraction of tal2-expressing LFPcells at 24 hpf (Figs. 1K, L). This suggests that tal2-expressingcells start to differentiate into neurons at this stage ofdevelopment. As tal2 is expressed in a subpopulation of thenkx2.2b-expressing LFP, huC/D expression was also detectedin a few nkx2.2b-expressing cells (Figs. 1M, N).

Taken together, this shows that the two parallel rows of cellsin the zebrafish ventral neural tube, which have originally beendescribed as the LFP, have different expression profiles at 18–20 hpf. It furthermore indicates that two different cellpopulations alternate along the AP axis within this domain.One cell population expresses the floor plate marker gene foxa2and nkx2.2b (foxa2+, nkx2.2b+), while the other cell populationexpresses tal2 and nkx2.2b (tal2+, nkx2.2b+). These cells arepostmitotic and start to differentiate into neurons.

LFP cells alternate with p3 neuronal progenitor cells in thezebrafish ventral neural tube

To characterize the two different cell populations within theventral neural tube in more detail, we analyzed the expressionprofiles of both cell types during late neurogenesis. At 48 hpf,the expression patterns of foxa2 (Fig. 2A) and tal2 (Fig. 2B)were not significantly different when compared to earlierstages (Figs. 1C, E). Both markers, however, had significantlydecreased expression levels. We found that the bHLH genesim1 (single-minded1; Serluca and Fishman, 2001), which is amarker for differentiated V3 interneurons in higher vertebrates(Briscoe et al., 1999; Fan et al., 1996; Yamada et al., 1991), isexpressed in single cells of the LFP (Fig. 2C), similar to tal2.Double in situ hybridization at 44 hpf showed that a subset oftal2-positive cells expresses sim1 (Figs. 2D, E, K). sim1-expressing cells are apparently postmitotic neurons as theyexpress the neuronal marker huC/D (Figs. 2F, G). Thisindicates that sim1 is a marker for differentiated V3interneurons, similar to the situation in higher vertebrates.Thus, at least a subset of the tal2+, nkx2.2b+ cell populationdifferentiates into V3 interneurons. We also observed a similardistribution of GABAergic Kolmer–Agdhur (KA) neurons(Figs. 2H, L; Bernhardt et al., 1992) and gata-2-positiveventral interneurons (Fig. 2I; transgenic gata-2:GFP embryos;Meng et al., 1997) in the ventral neural tube. At 48 hpf, thetwo parallel rows of cells, defined as LFP, are still one cell inwidth and directly neighbored by cells expressing themotoneuronal domain marker olig2 and the differentiatedmotoneuron marker islet2 (isl2; Figs. 2M–P). Thus, noadditional cell types have formed between these cells andmotoneuronal cells.

Our expressions analysis, therefore, indicates that the twoparallel rows of cells flanking the MFP in zebrafish consist of

two distinct cell populations that differ in their expressionprofiles. Cells, which are tal2+, nkx2.2b+ appear to be neuronalprogenitor cells. A subset of these cells differentiates intosim1+/huC/D+ V3 interneurons and we thus postulate thatthese cells are p3 neuronal pregenitor cells equivalent to cells ofthe p3 neuronal domain in higher vertebrates (Briscoe et al.,1999; Ericson et al., 1997b; Pabst et al., 1998; Qiu et al., 1998;Shimamura et al., 1995). Possibly these cells also differentiateinto KA and/or other ventral interneurons. foxa2+, nkx2.2b+

cells, on the other hand, are not expressing any analyzedneuronal marker genes at 48 hpf and therefore can be consideredas non-neuronal floor plate cells (Fig. 2Q). Hence, we concludethat during early neurogenesis the zebrafish one cell wide LFP isnot a homogenous structure as described earlier but that LFPcells rather alternate along the AP axis with p3 neuronalprogenitor cells.

Delta-Notch signaling regulates LFP and p3 neuronalprogenitor cell formation

Proliferation and specification of neurons and glia cellswithin the developing vertebrate neural tube are controlled byDelta-Notch-dependent lateral inhibition. This process regulatesdifferential neural development by progressive specification ofonly a subset of cells within the neural ectoderm as neuralprogenitor cells. The neighboring cells, on the other hand, stayundifferentiated and acquire an alternative fate (Appel et al.,2001). We have investigated whether in zebrafish also LFP andp3 neuronal progenitor cells are specified by Delta-Notchsignaling. For this, we analyzed development of both cell typesin the mutant mindbomb (mib), which is deficient for a ring typeubiquitin ligase required for Delta activity (Itoh et al., 2003). Inhomozygous mib mutants, we found a strong reduction ofnkx2.2b (Fig. 3B) as marker for LFP and p3 neuronal cells. Inaddition, also foxa2 in LFP cells (Fig. 3D) and tal2 in p3neuronal progenitor cells (Figs. 3H, J) were absent. foxa2expression in the MFP, in contrast, was normal (Fig. 3D). Thisindicates that most LFP and ventral neuronal progenitor cellsfail to form in homozygous mib mutants. The number ofprimary neurons expressing islet1 (isl1, Inoue et al., 1994), onthe other hand, was extremely increased in the neural tube. isl1-positive cells were detected directly adjacent to the MFP (Fig.3F). This indicates that mib function is required for LFP and p3marker gene expression and suggests that these cells are notspecified in mib mutants. Delta-Notch signaling thereforeseems to be required for LFP as well as p3 neuronal progenitorcell formation and the inhibition of primary neurogenesis insideboth cell types. Heterozygous embryos showed no LFP or p3neuronal phenotype (data not shown), which indicates thatreduced Delta-Notch activity is still sufficient to specify bothLFP and p3 neuronal progenitor cells.

Formation of LFP and p3 neuronal progenitor cells isdifferently regulated by Hedgehog signaling

To investigate the mechanisms of differential specificationof LFP and p3 neuronal progenitor cell development, we

Fig. 2. Two different cell populations alternate in the zebrafish ventral neural tube. (A, B) At 48 hpf, foxa2 is discontinuously expressed in cells adjacent to the MFP(A). tal2 expression is weakly detectable in single cells at 44 hpf (B). Arrowheads in panel (A) indicate regions with no detectable expression of foxa2. (C) sim1 isexpressed in groups of cells located adjacent to the MFP, which are discontinuously distributed along the AP axis, similar to tal2 in panel B. (D, E) Some sim1-positivecells (in red in D and E) co-express tal2 (in blue in D, indicated by arrowheads). (F–G) sim1-expressing cells (F) are positive for the neuronal marker huC/D (G). Thesame transverse section is shown for bright-field (F) and Cy3 fluorescence (H). (H, I) GABA protein in wild-type (H) and gata-2 gene expression in gata-2:GFPtransgenic embryos (I). Focal plane shows the most ventral cells of the neural tube and depicts Kolmer–Agdhur (KA) neurons and ventral interneurons (VI),respectively, in the ventral-most domain (also shown in transverse section in panel L). Note similar expression of tal2 (B), sim1 (C), GABA (H) and gata-2 (I). PanelsA–E and H, I are dorsal views of trunk, anterior to the left. (J–N) Transverse sections showing co-expression of sim1 (J), tal2 (in blue) and sim1 (in red; K), GABA (L),foxa2 (M) and nkx2.2b (N) in the ventral neural tube. (O, P) Expression of olig2 in the motoneuronal domain (M) and isl2 in differentiated motoneurons (N) is locatedadjacent to markers detected in panels J–N. (Q) Model of the zebrafish ventral neural tube during early neurogenesis in which non-neuronal LFP (foxa2+, nkx2.2b+)and p3 neuronal progenitor cells (tal2+, nkx2.2b+) alternate along the AP axis in cells adjacent to the MFP. Panels F, G and J–P are transverse sections of middle trunkregions. VI, ventral interneurons; KA, Kolmer–Agdhur neurons; MN, motoneurons; VeLD, ventral lateral distal neurons. Scale bars: E,G 5 μM; I,P 10 μM.


analyzed the role of Hedgehog (HH) signaling during thisprocess. Cell explant experiments as well as knockoutanalyses have shown that distinct neuronal progenitor cellsare specified along the DV axis of the vertebrate ventralneural tube, according to an activity gradient of HH signaling(Briscoe and Ericson, 2001; Ericson et al., 1997a; Jessell,2000, Lewis and Eisen, 2001). In zebrafish, LFP and p3neuronal progenitor cells apparently alternate along the AP

axis but are positioned at the same level along the DV axis.To investigate how these two cell types are specified, weanalyzed expression of specific marker genes in mutants ofthe ligand Shh (sonic-you, syu; Schauerte et al., 1998), thesignal transducer Smoothened (slow-muscle-omitted, smo;Varga et al., 2001), as well as of the downstream transcriptionfactors Gli1 (detour, dtr; Karlstrom et al., 2003) and Gli2(you-too, yot; Karlstrom et al., 1999). In all analyzed

Fig. 3. Delta-Notch signaling is required for LFP and p3 neuronal progenitor cell development. (A–D) nkx2.2b (A, B) and foxa2 (C, D) expression in wild-typecontrol (A, C) and mib−/− embryos (B, D). In mib−/− mutants, nkx2.2b and foxa2 are both strongly reduced in the LFP and p3 neuronal progenitor cells incomparison to control (for nkx2.2b in n=12 (22%) embryos from a clutch of 55 embryos obtained from identified, heterozygous carriers; n=5/19, 26% forfoxa2, n=8/34, 24% for tal2). Only single cells are detectable. (E, F) Expression of isl1 (in red) and shh (in blue) in wild-type (E) and mib−/− mutants (F). Incontrol embryos, isl1 is expressed in primary motor and sensory neurons (E). In mib−/−, isl1 is strongly expanded and is located in cells directly adjacent to theMFP. (G–J) tal2 expression in wild-type (G, I) and mib−/− (H, J) mutants. In mib−/− mutants, tal2 expression is strongly reduced in the most ventral cells(arrowhead marks region with absent p3 cells). Panels A–D are dorsal views, panels E, F, I, J are transverse sections and panels G, J are lateral views of embryosat the 16- to 20 s stage. Scale bars: H, 20 μM; J, 10 μM.


homozygous mutants, the markers nkx2.2b, foxa2, tal2 andsim1 were not detectable in the LFP and p3 neuronal cellsof the trunk indicating that these cells fail to form (Figs. 4C,F, K and data not shown; Schäfer et al., 2005a). Heterozygoussyu and smo embryos showed no altered expression whencompared to wild-type embryos (data not shown). However,in heterozygous dtr and yot mutants, we observed modulatedexpression of LFP and p3 neuronal marker genes. Nkx2.2b asa marker for both cell populations was reduced to single cells,which were irregularly distributed along the AP axis (Fig. 4Band data not shown). Expression of foxa2 as a marker forLFP cells, in contrast, was only detectable in the MFP (Fig.4E), while expression of tal2 in p3 neuronal progenitor cellswas normal (Fig. 4J). Also the V3 neuron marker sim1showed no altered expression in heterozygous yot embryoswhen compared to wild type (data not shown; n=24 out of aclutch of 34 embryos obtained from identified yot carriers).This suggests that in heterozygous dtr and yot embryos, LFPcells (foxa2+, nkx2.2b+) fail to form, while p3 neuronal cells(tal2+, nkx2.2b+) are still present. Therefore, we concludethat LFP and p3 neuronal cells require different HH activities.Non-neuronal LFP cells appear to require high levels of HHactivity and consequently are not formed in heterozygous gli

mutants. P3 neuronal progenitor cells, in contrast, requirelower levels of HH activity and are therefore induced.

The loss of LFP marker genes in heterozygous yot and dtrmutants raises the question, whether these cells are specifiedto develop into any other neuronal subtype. To investigatethis, we analyzed expression of the motoneuronal domainmarker olig2 in dtr mutants. In homozygous dtr mutantembryos, we found a ventral shift of olig2 expression intocells neighboring the MFP in the entire trunk (Figs. 5C, F anddata not shown). This has also been shown before inhomozygous syu mutants (Park et al., 2004). In heterozygousdtr embryos, we observed a similar ventral shift of the olig2domain in regions between remaining nkx2.2b-expressingcells (Figs. 5B, E). However, no expression of themotoneuronal marker isl2 (Appel et al., 1995) could beobserved in the most ventral cells of homozygous andheterozygous yot mutants (Figs. 5G, H). This indicates thatLFP cells fail to develop their characteristic features whenHH activity is reduced. They instead acquire characteristicsindicative for the motoneuronal domain but do not differen-tiate into isl2-positive motoneurons. Whether these cellsdevelop into any other subtype of motoneuron or into gliacells remains unclear.

Fig. 4. Differential regulation of LFP and p3 neuronal progenitor cells in hedgehogmutants. (A–C) In wild-type control embryos, nkx2.2b is homogenously expressedin LFP and p3 neuronal progenitor cells along the AP axis (A). nkx2.2b expression is reduced to single cells in dtr+/− (detour, gli1) mutants (in n=15 (56%) out of aclutch of 27 embryos obtained from identified carriers; B) and completely absent in dtr−/− (n=7/27, 26%; C). Similar results were obtained in yotmutants (yot+/−, n=22/40, 55%; yot−/−, n=11/40, 28%). (D) foxa2 is expressed in the MFP and LFP cells in wt embryos (D). In yot+/− (you-too, gli2; E) and yot−/− embryos (F), foxa2 is onlydetectable in theMFP (n=23/34, 68%). The samewas observed in dtr+/− and dtr−/−mutants (n=11/18, 61%). (G–I) Identification of homozygous and heterozygous yotmutants. In wild-type embryos, eng2a is expressed in muscle pioneer cells (MPC; G). In yot+/− embryos, where foxa2 expression is lost in the LFP, expression of eng2ais normal (H). yot−/− embryos, in contrast, lack both, foxa2 in the LFP, as well as eng2a in the trunk (I). (J, K) Transverse sections of middle trunk regions. In wt and yot+/−

embryos, tal2 is expressed in the p3 neuronal progenitor cells and in dorsally located interneurons (IN) within the ventral neural tube (J). yot−/− embryos lack tal2expression in p3 neuronal cells, expression in more dorsally located interneurons is unaffected (n=6/26, 23%; K). Scale bars: F, K, 10 μM.


Cyclopamine treatment reveals a differential regulation of LFPand p3 neuronal progenitor cells by Hedgehog activities

To directly test whether LFP and p3 neuronal progenitor cellsare regulated by different activities of HH signaling, wegradually inhibited the HH signal transducer Smoothened usingthe alkaloid cyclopamine (Cooper et al., 1998). While treatmentof embryos with DMSO did not affect nkx2.2b expression(Figs. 6A, E), embryos incubated in 1–1.5 μM cyclopaminefrom shield stage to the 22 s stage resulted in a significantreduction of nkx2.2b expression to single cells (Figs. 6B, F).This is similar to the situation in heterozygous dtr and yotembryos (Fig. 4B). As observed in these heterozygous mutants,foxa2 in LFP cells was not detectable (Fig. 6J), but expressionof tal2 in p3 neuronal progenitor cells was normal (Fig. 6N). Inembryos incubated with 2 μM cyclopamine or higherconcentrations, nkx2.2b, foxa2 and tal2 completely disappearedin LFP and p3 neuronal cells (Figs. 6C, D, G, H, K, L, O, P), asobserved in homozygous hh mutants.

The expression pattern of nkx2.2b in the head of cyclopa-mine-treated embryos also resembles that of hh mutants.Treatment of embryos with 1–1.5 μM cyclopamine resultedin normal expression of nkx2.2b in the head, as observed inheterozygous gli mutants, in which also only nkx2.2bexpression in the trunk was reduced (Fig. 6B). Embryosincubated with 2 μM cyclopamine showed no nkx2.2bexpression in the midbrain–hindbrain boundary (MHB) andin parts of the hindbrain (Fig. 6C). This is similar to nkx2.2bexpression in homozygous yot embryos (Schäfer et al., 2005a).

Treatment of embryos with 3 μM cyclopamine resulted in acomplete repression of nkx2.2b in the entire embryo (Fig. 6D),which resembles the situation in homozygous smo embryos(Schäfer et al., 2005a). Altogether, this shows that gradualinhibition of Smoothened activity with the alkaloid cyclopa-mine can mimic the phenotypes observed in homozygous andheterozygous gli mutants. These results therefore stronglyindicate that LFP and p3 neuronal progenitor cells aredifferentially regulated by HH activity in a dosage-dependentmanner. LFP cells (foxa2+, nkx2.2b+) require high levels of HHactivity, while p3 neuronal progenitor cells (tal2+, nkx2.2b+)require lower levels.

Interestingly, in embryos treated with 1.5 μM cyclopamine,nkx2.2b was gradually repressed along the AP axis with threeregions of different expression. In hindbrain and anterior trunk,nkx2.2b expression was normal. In mid-trunk regions, nkx2.2bwas reduced to single cells, while in the posterior trunk nkx2.2bexpression was completely blocked (Fig. 6R). This indicatesthat at a given developmental stage, HH signaling has ananterior to posterior gradient of activity during induction ofnkx2.2b expression. This is in line with earlier observations ofthe HH-dependent induction of motoneurons (Lewis and Eisen,2001; Varga et al., 2001).

nkx2.2a and nkx2.2b genes are required for LFP formation

Nkx genes represent important targets for HH signaling in theventral neural tube. As member of the class II homeobox genes,Nkx2.2 is activated by high doses of HH activity. Consequently,

Fig. 5. The motoneuronal domain is ventrally expanded in hedgehog mutants.(A–F) Expression of nkx2.2b (blue) and olig2 (red) in lateral view of trunkregions (A–C) and transverse sections (D–F). In wt embryos, nkx2.2b isexpressed in LFP and p3 neuronal progenitor cells, olig2 is expressed in themotoneuronal domain (A, D). Olig2 expression is ventrally expanded in regionswith no nkx2.2b in dtr+/− (E; n=35 out of a clutch of 71 embryos obtained frommutant carriers, 49%) or along the entire axis in dtr−/− embryos (n=17/71,24%). The section shown in panel (E) was made in a region between remainingnkx2.2b expression. (G, H) Differentiated motoneurons expressing isl2 (in blue)are formed within the lateral motoneuronal domain in wt and yot+/− embryos(n=16/22, 73%; G). The same position of isl2 is observed in yot−/− (n=6/22,27%). Most notably, no expression in cells neighboring the MFP expressingcol2A1 (in red) are found (H). Homozygous yot embryos were identified byloss of eng2 (in blue) in muscle pioneer cells (MPC). Panels A–C are lateralviews of trunk, anterior to the left. Panels D–H are transverse sections of middletrunk. MPC, muscle pioneer cells. Scale bars: C, F, 10 μM.


Nkx2.2 is expressed in the cells directly adjacent to the MFP.Zebrafish contain two duplicated nkx2.2 genes, which are theonly genes known so far that are expressed in both LFP andp3 neuronal cells (Barth and Wilson, 1995; Schäfer et al.,2005a). To analyze, if nkx2.2 genes are capable of mediatingdifferential HH activities in specifying p3 neuronal versusLFP cells, we used a morpholino based gene knock-downapproach.

Knock-down of nkx2.2a led to a severe reduction of the headsize with retardation of eyes, enlargement of the 3rd ventricleand aggregation of necrotic cells in the hindbrain (data notshown). nkx2.2b morpholino injected embryos were slightlyshortened and had enlarged pericardial sacs, but otherwise

looked normal (data not shown). Both morpholinos were testedfor functionality by coinjection with GFP sensor constructs (seeSupplementary material).

We investigated the function of nkx2.2a and nkx2.2b duringformation of the trunk LFP and p3 neuronal progenitor cells.First, we analyzed expression of nkx2.2b as marker of both cellpopulations. Knock-down of nkx2.2a resulted in a reduction ofnkx2.2b expression to single cells that were irregularlydistributed along the AP axis (n=19/44, 43%; Fig. 7B), similarto the situation in HH deficient embryos. As observed incyclopamine-treated embryos, nkx2.2b showed a gradualinhibition from anterior to posterior in nkx2.2a morphants(data not shown). The same reduction of nkx2.2b was observedin a small number of nkx2.2b morpholino injected embryos(n=16/142, 11%; data not shown). Coinjection of both nkx2.2aand nkx2.2b morpholinos did not reduce the number ofremaining nkx2.2b-expressing cells any further (n=30/57,52%, data not shown). nkx2.2a morpholino injected embryosfurthermore resulted in a strong reduction of foxa2 expressionin the LFP (n=27/57, 47%; Fig. 7D). A similar reduction wasobserved for nkx2.2b morpholino injection as well as coinjec-tion of nkx2.2a and nkx2.2b morpholinos (n=8/52, 15% fornkx2.2b; n=21/38, 55% for nkx2.2a+b morpholino; data notshown). On the other hand, no change in the expression of thep3 neuronal progenitor cell marker tal2 was observed in theventral neural tube (n=46/46 for nkx2.2a, Fig. 7F; n=55/55 fornkx2.2b and n=63/63 for nkx2.2a+b morpholino, data notshown). This indicates that nkx2.2a and nkx2.2b are requiredfor formation of LFP cells but do not affect formation of p3neuronal progenitor cells.

Discussion

LFP and p3 neuronal progenitor cells alternate along the APaxis in the zebrafish ventral neural tube

The floor plate is an important signaling center, whichspecifies neurons and glia cells in the ventral neural tube andguides the trajectory of outgrowing axons. It consists of an innerlocated MFP and an outer positioned LFP. Contrary results existfor the structure of the LFP in vertebrates. While in zebrafish theLFP is described as a homogenous structure, in chicken LFPand p3 neuronal markers overlap. Thus, the exact structure ofthe LFP in vertebrates is currently unclear.

We and others have previously shown that the homeoboxgenes nkx2.2a and nkx2.2b are continuously expressed in thezebrafish LFP (Fig. 1B; Barth and Wilson, 1995; Schäfer et al.,2005a). In the present study, we found that only a subset ofnkx2.2-positive cells expresses high levels of the floor platemarker gene foxa2 at 18–20 hpf (Figs. 1C, D). Furthermore, weshow that foxa2-positive cells proliferate during early neuro-genesis (Fig. 1J) and most likely stay undifferentiated until atleast 48 hpf (Fig. 2A). The other subset of cells, in contrast,expresses the genes tal2 and probably gata-2 (Fig. 1E andSupplementary material). During early neurogenesis, these cellsmost likely become postmitotic and start to differentiate intoneurons (Figs. 1J–N). At 48 hpf, a subset of these cells

Fig. 6. Dose-dependent effects of cyclopamine treatment on specification of LFP and p3 neuronal progenitor cells. (A–H) nkx2.2b expression in embryos treated withDMSO (A, E), 1–1.5 μM (B, F), 2 μM (C, G) and 3 μM (D, H) cyclopamine. (A, E) Embryos treated with DMSO show no effects on nkx2.2b expression. (B, F) 1–1.5 μM cyclopamine has no effect on head expression (B), but leads to discontinuous expression in the trunk (F; n=68/123, 55%). (C, G) At 2 μM cyclopamine,nkx2.2b expression is reduced in the head (C) and completely absent in the trunk (G; n=18/22, 82%). (D, H) Treatment with 3 μM cyclopamine blocks nkx2.2bexpression in the entire embryo (n=17/19, 89%). (I–L) Expression of foxa2 in embryos treated with DMSO (I), 1–1.5 μM (J), 2 μM (K) 3 μM cyclopamine (L). foxa2is absent in the LFP and only detectable in the MFP in embryos treated with 1 μM cyclopamine or higher (J–L; n=27/40, 67% for 1–1.5 μM), when compared to theDMSO control (I). (M–P) tal2 expression in embryos incubated with DMSO (M), 1–1.5 μM (N), 2 μM (O) and 3 μM cyclopamine (P). Expression of tal2 is notaffected in embryos treated with 1–1.5 μM cyclopamine (M; n=51/73, 70%), when compared to the DMSO control (N). Only higher concentrations of 2 μM (O) and3 μM cyclopamine (P) lead to a loss of tal2 in p3 neuronal cells (indicated by arrowhead). (Q, R) nkx2.2b expression in DMSO control (Q) and 1.5 μM cyclopamine-treated embryos, lateral view (R). Embryos treated with 1–1.5 μM cyclopamine have patchy nkx2.2b expression in the middle trunk regions. In the posterior trunk,nkx2.2b expression is completely blocked. Regions with different degrees of nkx2.2b repression are indicated by boxes. Panels A–D and Q, R are lateral views ofembryos in overview and trunk and tail regions respectively, anterior is to the left, panels E–L are dorsal view on trunk regions, anterior to the left, panels M–P aretransverse section of middle trunk region. IN, interneurons. Scale bars: L, P, 10 μM; R, 50 μM.


expresses sim1, which is a marker for differentiated V3interneurons in higher vertebrates (Briscoe et al., 1999). Inzebrafish, sim1 is expressed in postmitotic neurons in thezebrafish ventral neural tube (Figs. 2F, G). This suggests thatsimilar to higher vertebrates, sim1 is a marker for differentiated

V3 interneurons in zebrafish. Thus, at least a subset of tal2+,nkx2.2b+ cells differentiates into V3 interneurons during lateneurogenesis. A similar distribution as for tal2 and sim1 in theventral neural tube was shown for the neurotransmitter GABA,specific for Kolmer–Agdhur neurons (Figs. 2H, L; Bernhardt et

Fig. 7. Knock-down of nkx2.2a interferes with LFP formation. (A, B) nkx2.2a morpholino injected embryos show reduction of nkx2.2b expression to single cells(B; for details, see text). (C–F) The LFP marker foxa2 is almost completely repressed in the LFP after nkx2.2a morpholino knock-down (D), while expression oftal2 in p3 neuronal progenitor cells is not affected (F). All pictures are dorsal views with anterior to the left. Scale bar in panel F is 20 μM.


al., 1992), and gata-2 as marker for ventral interneurons (Fig.2I; Meng et al., 1997).

Based on these data, we conclude that the LFP in zebrafish isnot a homogenous structure as described earlier (Odenthal et al.,2000), but that non-neuronal LFP cells (foxa2+, nkx2.2b+)alternate with p3 neuronal progenitor cells (tal2+, nkx2.2b+),which later at least partly develop into V3 interneurons (sim1+).In the zebrafish neural tube, genes expressed in LFP and p3neuronal cells do not show overlapping patterns, as it has beenshown for chicken (Charrier et al., 2002), but have clearlyseparated domains (Fig. 8). It remains open, whether the samecells express both, sim1 as V3 neuronal differentiation markerand GABA, indicative for KA neurons. Furthermore, it is stillnot clear, whether nkx2.2+/foxa2+ LFP cells serve as a pool ofundifferentiated progenitors, which develop into neurons or gliacells during later stages. As both markers are down-regulated at48 hpf, future cell labeling experiments are required todetermine the fate of this cell population.

Delta-Notch signaling is required for both LFP and p3neuronal progenitor formation

Delta-Notch-dependent lateral inhibition is involved inneural specification at different steps of development. By thismechanism, only a subset of cells within the neural ectodermdifferentiates, which transiently expresses high levels of theligand Delta. Neighboring cells that express Notch are inhibitedfrom differentiation and acquire a different fate at later stages ofdevelopment. In zebrafish, this mechanism for exampleregulates primary neurogenesis in a fraction of cells in thegastrula neural plate leading to a specific pattern of motor andsensory neurons (Appel et al., 2001). It has been shown beforethat notch genes are expressed in most ventral cells of thezebrafish neural tube during neurulation (Appel et al., 2001).We have, therefore, investigated whether development of theLFP and p3 neuronal progenitor cells is regulated by Delta-Notch signaling. We observed a strong reduction of LFP and p3neuronal markers in the entire trunk of mib mutants. On the

other hand, isl1-expressing primary motoneurons were expand-ed to the MFP (Fig. 3). This indicates that Delta-Notch signalingis required for the specification of LFP and p3 neuronal cells,which are located directly adjacent to the motoneuronal domain.It remains, however, unclear whether Delta-Notch-dependentlateral inhibition also regulates formation of the LFP versus p3neuronal progenitor cells at a later stage of development, afterprimary neurogenesis. This has to be analyzed in the future, e.g.,by inhibition of Delta-Notch signaling in the zebrafish embryoduring early neurulation.

Different Hedgehog activities specify LFP and p3 neuronalprogenitor cells

The development of distinct neurons along the DV axis ofthe vertebrate ventral neural tube is initiated by Shh secretedfrom the floor plate and underlying notochord (Echelard etal., 1993; Marti et al., 1995a; Chiang et al., 1996). Duringearly neurulation, distinct threshold concentrations of Shhinduce and repress homeobox transcription factors andthereby specify five neuronal progenitor domains at appro-priate positions along the DV axis. These are from ventral todorsal p3, motoneuron, p2, p1 and p0 neuronal progenitordomains. Later, a specific subtype of neuron develops in eachdomain (reviewed in Briscoe and Ericson, 2001). We haveshown that in zebrafish non-neuronal LFP cells and p3neuronal progenitor cells alternate along the AP axis, and arepositioned at the same level of the DV axis. Expressionanalyses of different floor plate markers in HH signalingmutants showed that the two subpopulations are differentlyregulated by HH signaling. No effects compared to the wild-type situation were observed in heterozygous syu and smomutants. In heterozygous dtr (gli1) and yot (gli2) mutants, onthe other hand, only p3 neuronal cells were formed, while LFPcells failed to form (Figs. 4B, E, J). This suggests a dosage-dependent regulation of these two subpopulations, which isapparently not as sensitive to reduced receptor and ligand levels,but very sensitive to the reduction of the HH downstream

Fig. 8. Schematic model of ventral neural tube organization during earlyneurogenesis in chicken and zebrafish. In chicken, non-neuronal LFP and p3neuronal progenitor cell markers overlap in the domain between MFP andmotoneuron progenitors (pMN). In zebrafish, LFP cells (in red) and p3 neuronalprogenitor cells (in grey) alternate along the AP axis within this domain. Duringearly embryonic development, distinct domains of cells that require differentlevels of Shh activity are generated along the DV axis of the vertebrate neuraltube. In zebrafish, LFP and p3 neuronal progenitor cells, which are positioned atthe same level along the DV axis, require different levels of HH activity. Notethat MFP, LFP and neural progenitor domains are several cells wide in chicken,while in zebrafish MFP and LFP are only one cell in width and the pMN domainis two to three cells wide.


components Gli1 and Gli2. This result was confirmed incyclopamine inhibitor experiments, in which Smoothenedsignaling was gradually inhibited. A distinct concentration ofcyclopamine was identified that resulted in a loss of non-neuronal floor plate cells, but had no effect on p3 neuronalprogenitor cells, similar to the situation in heterozygous yot anddtrmutants (Figs. 6B, F, J, N). This suggests that the loss of LFPcells in yot and dtr mutants is unlikely to be caused by adominant negative effect of truncated Glis. It rather opens thepossibility that LFP cells and p3 neuronal progenitor cellsexhibit different levels of competence to respond to a given HHconcentration. We speculate that LFP cells (foxa2+, nkx2.2b+)require high HH activities, while p3 neuronal progenitor cells(tal2+, nkx2.2b+) respond to lower HH activities. In heterozy-gous dtr and yotmutants, where the HH activity is reduced, LFPcells are therefore missing, while p3 neuronal progenitor cellsare still specified. In heterozygous dtr and yot mutants, LFPcells surprisingly establish an olig2-expressing motoneuronaldomain, rather than for example differentiating into p3 neuronalprogenitor cells, which are tal2+, nkx2.2b+ (Fig. 5). Obviously,the fate of non-neuronal LFP cells is strictly separated from thatof neuronal precursor cells of the p3 neuronal domain. Thissuggests that these cells are pre-specified by a so far unknownfactor to differently respond to a given HH concentration.Alternatively, discontinuous levels of HH activity and/or thedownstream factors Gli along the AP axis could be responsiblefor the specification of the two subpopulations of cells. Anactivity gradient for different Glis along the DVaxis of the avianneural tube has recently been reported (Stamataki et al., 2005).Future experiments investigating Gli activity along the AP axisin zebrafish will give insight into the mechanisms of LFP and p3neuronal progenitor cell specification. However, we cannotexclude the possibility that the pattern of the two distinct celltypes is determined in a process downstream of HH signaling.According to this possibility, equivalent and directly neighbor-ing cells of the unspecified ventral neural tube could responddifferently to the same concentration of HH ligand. This could be

mediated by cell-to-cell interactions downstream of the responseto HH signaling, like e.g., lateral inhibition by Delta-Notchsignaling.

The homeobox genes nkx2.2a and nkx2.2b are required fornon-neuronal LFP cells

Nkx2.2 is a member of the class II homeobox proteins,which is present in both, the LFP (Charrier et al., 2002) and p3neuronal progenitor domain (Briscoe et al., 1999; Qiu et al.,1998; Shimamura et al., 1995; Charrier et al., 2002) of theventral neural tube of vertebrates. It has been shown thatNkx2.2 represses the class I homeobox gene pax6 and thusindirectly specifies the p3 neuronal progenitor domain (Briscoeet al., 2000). However, since the p3 neuronal domain formsnormally in nkx2.2 knockout mice (Briscoe et al., 1999), thefunction of Nkx2.2 during early neuronal specification is notclear. Nkx2.2 might function redundantly with other ventrallyexpressed homoebox transcription factors, like e.g., Nkx2.9(Briscoe et al., 1999, 2000). During later development, Nkx2.2specifies V3 interneurons and inhibits formation of motoneur-ons. Nkx2.2 knockout mice therefore lack differentiated V3neurons and motoneurons are ventrally expanded (Briscoe et al.,1999).

We have investigated the function of two nkx2.2orthologues, nkx2.2a and nkx2.2b, for formation of theLFP and p3 neuronal progenitor domain in zebrafish.Knock-down of nkx2.2a and nkx2.2b led to a loss of LFPcells but had no effect on formation of p3 neuronal cells(Fig. 7). Obviously, these genes are exclusively required forLFP formation, although they are strongly expressed also inp3 neuronal progenitor cells. It is possible that nkx2.2 genesact redundantly during formation of p3 neuronal progenitorcells, like it is predicted in higher vertebrates for nkx2.2 andnkx2.9. However, apparently also in the non-neuronal LFPsome level of redundancy exists. In our experiments, knock-down of nkx2.2a led to a LFP reduction in 43% of injectedembryos, while nkx2.2b even affected only 11%. Simulta-neous injection of both morpholino oligonucleotides led to asimilar reduction of the LFP and only slightly increased theefficiency (53%). Therefore, it remains unclear whethernkx2.2 genes act redundantly with each other and/or withother genes during patterning of the LFP and p3 neuronaldomain in zebrafish. Isolation and functional analysis of thenkx2.9 ortholog in zebrafish will shed further light on themechanisms of neural patterning in the zebrafish ventralneural tube.

Acknowledgments

We thank Bruce Appel, Matthias Carl, Jochen Holzschuh,Carl Neumann, Fabrizio Serluca, Josette Ungos, Zoltan Vargaand Steve Wilson for providing mutant zebrafish and probes.We are grateful to Andrea Wizenmann for discussion andcritical comments on the manuscript. This work was supportedby the Boehringer Ingelheim Fonds to M.S. and the DFG/Graduiertenkolleg GRK1048 to C.W. and M.S.


Appendix A. Supplementary data

Supplementary data associated with this article can be found,in the online version, at doi:10.1016/j.ydbio.2006.09.018.

References

Agius, E., Soukkarieh, C., Danesin, C., Kan, P., Takebayashi, H., Soula, C.,Cochard, P., 2004. Converse control of oligodendrocyte and astrocytelineage development by Sonic hedgehog in the chick spinal cord. Dev. Biol.270, 308–321.

Appel, B., Korzh, V., Glasgow, E., Thor, S., Edlund, T., Dawid, I.B., Eisen, J.S.,1995. Motoneuron fate specification revealed by patterned LIM homeoboxgene expression in embryonic zebrafish. Development 121, 4117–4125.

Appel, B., Fritz, A., Westerfield, M., Grunwald, D.J., Eisen, J.S., Riley, B.B.,1999. Delta-mediated specification of midline cell fates in zebrafishembryos. Curr. Biol. 9, 247–256.

Appel, B., Givan, L.A., Eisen, J.S., 2001. Delta-Notch signaling and lateralinhibition in zebrafish spinal cord development. BMC Dev. Biol. 1, 13.

Barth, K.A., Wilson, S.W., 1995. Expression of zebrafish nk2.2 is influencedby sonic hedgehog/vertebrate hedgehog-1 and demarcates a zone ofneuronal differentiation in the embryonic forebrain. Development 121,1755–1768.

Bernhardt, R.R., Patel, C.K., Wilson, S.W., Kuwada, J.Y., 1992. Axonaltrajectories and distribution of GABAergic spinal neurons in wildtype andmutant zebrafish lacking floor plate cells. J. Comp. Neurol. 326, 263–272.

Bourikas, D., Pekarik, V., Baeriswyl, T., Grunditz, A., Sadhu, R., Nardo, M.,Stoeckli, E.T., 2005. Sonic hedgehog guides commissural axons along thelongitudinal axis of the spinal cord. Nat. Neurosci. 8, 297–304.

Briscoe, J., Ericson, J., 2001. Specification of neuronal fates in the ventral neuraltube. Curr. Opin. Neurobiol. 11, 43–49.

Briscoe, J., Sussel, L., Serup, P., Hartigan-O'Connor, D., Jessell, T.M.,Rubenstein, J.L., Ericson, J., 1999. Homeobox gene Nkx2.2 andspecification of neuronal identity by graded Sonic hedgehog signalling.Nature 398, 622–627.

Briscoe, J., Pierani, A., Jessell, T.M., Ericson, J., 2000. A homeodomain proteincode specifies progenitor cell identity and neuronal fate in the ventral neuraltube. Cell 101, 435–445.

Charrier, J.B., Lapointe, F., Le Douarin, N.M., Teillet, M.A., 2002. Dual originof the floor plate in the avian embryo. Development 129, 4785–4796.

Charron, F., Stein, E., Jeong, J., McMahon, A.P., Tessier-Lavigne, M., 2003.The morphogen sonic hedgehog is an axonal chemoattractant thatcollaborates with netrin-1 in midline axon guidance. Cell 113, 11–23.

Chiang, C., Litingtung, Y., Lee, E., Young, K.E., Corden, J.L., Westphal, H.,Beachy, P.A., 1996. Cyclopia and defective axial patterning in mice lackingSonic hedgehog gene function. Nature 383, 407–413.

Cooper, M.K., Porter, J.A., Young, K.E., Beachy, P.A., 1998. Teratogen-mediated inhibition of target tissue response to Shh signaling. Science 280,1603–1607.

Detrich III, H.W., Kieran, M.W., Chan, F.Y., Barone, L.M., Yee, K.,Rundstadler, J.A., Pratt, S., Ransom, D., Zon, L.I., 1995. Intraembryonichematopoietic cell migration during vertebrate development. Proc. Natl.Acad. Sci. U. S. A. 92, 10713–10717.

Echelard, Y., Epstein, D.J., St-Jacques, B., Shen, L., Mohler, J., McMahon, J.A.,McMahon, A.P., 1993. Sonic hedgehog, a member of a family of putativesignaling molecules, is implicated in the regulation of CNS polarity. Cell 75,1417–1430.

Ekker, M., Wegner, J., Akimenko, M.A., Westerfield, M., 1992. Coordinateembryonic expression of three zebrafish engrailed genes. Development 116,1001–1010.

Ericson, J., Briscoe, J., Rashbass, P., van Heyningen, V., Jessell, T.M., 1997a.Graded sonic hedgehog signaling and the specification of cell fate in theventral neural tube. Cold Spring Harbor Symp. Quant. Biol. 62, 451–466.

Ericson, J., Rashbass, P., Schedl, A., Brenner-Morton, S., Kawakami, A., vanHeyningen, V., Jessell, T.M., Briscoe, J., 1997b. Pax6 controls progenitorcell identity and neuronal fate in response to graded Shh signaling. Cell 90,169–180.

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., Tessier-Lavigne,M., 1996. Expression patterns of twomurine homologs ofDrosophila single-minded suggest possible roles in embryonic patterning and in thepathogenesis of Down syndrome. Mol. Cell. Neurosci. 7, 1–16.

Halpern, M.E., Hatta, K., Amacher, S.L., Talbot, W.S., Yan, Y.L., Thisse, B.,Thisse, C., Postlethwait, J.H., Kimmel, C.B., 1997. Genetic interactions inzebrafish midline development. Dev. Biol. 187, 154–170.

Higashijima, S., Nose, A., Eguchi, G., Hotta, Y., Okamoto, H., 1997. Mindin/F-spondin family: novel ECM proteins expressed in the zebrafish embryonicaxis. Dev. Biol. 192, 211–227.

Inoue, A., Takahashi, M., Hatta, K., Hotta, Y., Okamoto, H., 1994.Developmental regulation of islet-1 mRNA expression during neuronaldifferentiation in embryonic zebrafish. Dev. Dyn. 199, 1–11.

Itoh, M., Kim, C.H., Palardy, G., Oda, T., Jiang, Y.J., Maust, D., Yeo, S.Y.,Lorick, K., Wright, G.J., Ariza-McNaughton, L., et al., 2003. Mind bomb isa ubiquitin ligase that is essential for efficient activation of Notch signalingby Delta. Dev. Cell 4, 67–82.

Jeong, Y., Epstein, D.J., 2003. Distinct regulators of Shh transcription in thefloor plate and notochord indicate separate origins for these tissues in themouse node. Development 130, 3891–3902.

Jessell, T.M., 2000. Neuronal specification in the spinal cord: inductive signalsand transcriptional codes. Nat. Rev., Genet. 1, 20–29.

Karlstrom, R.O., Talbot, W.S., Schier, A.F., 1999. Comparative synteny cloningof zebrafish you-too: mutations in the Hedgehog target gli2 affect ventralforebrain patterning. Genes Dev. 13, 388–393.

Karlstrom, R.O., Tyurina, O.V., Kawakami, A., Nishioka, N., Talbot, W.S.,Sasaki, H., Schier, A.F., 2003. Genetic analysis of zebrafish gli1 and gli2reveals divergent requirements for gli genes in vertebrate development.Development 130, 1549–1564.

Kimmel, C.B., Ballard, W.W., Kimmel, S.R., Ullmann, B., Schilling, T.F., 1995.Stages of embryonic development of the zebrafish. Dev. Dyn. 203, 253–310.

Latimer, A.J., Shin, J., Appel, B., 2005. her9 promotes floor plate developmentin zebrafish. Dev. Dyn. 232, 1098–1104.

Le Douarin, N.M., Halpern, M.E., 2000. Discussion point. Origin andspecification of the neural tube floor plate: insights from the chick andzebrafish. Curr. Opin. Neurobiol. 10, 23–30.

Lewis, K.E., Eisen, J.S., 2001. Hedgehog signaling is required for primarymotoneuron induction in zebrafish. Development 128, 3485–3495.

Lyuksyutova, A.I., Lu, C.C., Milanesio, N., King, L.A., Guo, N., Wang, Y.,Nathans, J., Tessier-Lavigne, M., Zou, Y., 2003. Anterior–posteriorguidance of commissural axons by Wnt-frizzled signaling. Science 302,1984–1988.

Marti, E., Bumcrot, D.A., Takada, R., McMahon, A.P., 1995a. Requirement of19K form of Sonic hedgehog for induction of distinct ventral cell types inCNS explants. Nature 375, 322–325.

Marti, E., Takada, R., Bumcrot, D.A., Sasaki, H., McMahon, A.P., 1995b.Distribution of Sonic hedgehog peptides in the developing chick and mouseembryo. Development 121, 2537–2547.

Marusich, M.F., Furneaux, H.M., Henion, P.D., Weston, J.A., 1994. Hu neuronalproteins are expressed in proliferating neurogenic cells. J. Neurobiol. 25,143–155.

Meng, A., Tang, H., Ong, B.A., Farrell, M.J., Lin, S., 1997. Promoteranalysis in living zebrafish embryos identifies a cis-acting motif requiredfor neuronal expression of GATA-2. Proc. Natl. Acad. Sci. U. S. A. 94,6267–6272.

Odenthal, J., Nusslein-Volhard, C., 1998. Fork head domain genes in zebrafish.Dev. Genes Evol. 208, 245–258.

Odenthal, J., van Eeden, F.J., Haffter, P., Ingham, P.W., Nusslein-Volhard, C.,2000. Two distinct cell populations in the floor plate of the zebrafish areinduced by different pathways. Dev. Biol. 219, 350–363.

Pabst, O., Herbrand, H., Arnold, H.H., 1998. Nkx2-9 is a novel homeoboxtranscription factor which demarcates ventral domains in the developingmouse CNS. Mech. Dev. 73, 85–93.

Park, H.C., Mehta, A., Richardson, J.S., Appel, B., 2002. olig2 is required forzebrafish primary motor neuron and oligodendrocyte development. Dev.Biol. 248, 356–368.

Park, H.C., Shin, J., Appel, B., 2004. Spatial and temporal regulation of ventral

http://dx.doi.org/doi:10.1016/j.ydbio.2006.09.018


spinal cord precursor specification by Hedgehog signaling. Development131, 5959–5969.

Pinheiro, P., Gering, M., Patient, R., 2004. The basic helix–loop–helixtranscription factor, Tal2, marks the lateral floor plate of the spinal cord inzebrafish. Gene Expr. Patterns 4, 85–92.

Placzek, M., Briscoe, J., 2005. The floor plate: multiple cells, multiple signals.Nat. Rev., Neurosci. 6, 230–240.

Placzek, M., Yamada, T., Tessier-Lavigne, M., Jessell, T., Dodd, J., 1991.Control of dorsoventral pattern in vertebrate neural development: inductionand polarizing properties of the floor plate. Development (Suppl. 2),105–122.

Placzek, M., Jessell, T.M., Dodd, J., 1993. Induction of floor plate differentiationby contact-dependent, homeogenetic signals. Development 117, 205–218.

Poncet, C., Soula, C., Trousse, F., Kan, P., Hirsinger, E., Pourquie, O., Duprat,A.M., Cochard, P., 1996. Induction of oligodendrocyte progenitors in thetrunk neural tube by ventralizing signals: effects of notochord and floor plategrafts, and of sonic hedgehog. Mech. Dev. 60, 13–32.

Pringle, N.P., Yu, W.P., Guthrie, S., Roelink, H., Lumsden, A., Peterson, A.C.,Richardson, W.D., 1996. Determination of neuroepithelial cell fate:induction of the oligodendrocyte lineage by ventral midline cells andsonic hedgehog. Dev. Biol. 177, 30–42.

Qiu, M., Shimamura, K., Sussel, L., Chen, S., Rubenstein, J.L., 1998. Control ofanteroposterior and dorsoventral domains of Nkx-6.1 gene expressionrelative to other Nkx genes during vertebrate CNS development. Mech. Dev.72, 77–88.

Sampath, K., Rubinstein, A.L., Cheng, A.M., Liang, J.O., Fekany, K., Solnica-Krezel, L., Korzh, V., Halpern, M.E., Wright, C.V., 1998. Induction of thezebrafish ventral brain and floorplate requires cyclops/nodal signalling.Nature 395, 185–189.

Schäfer, M., Kinzel, D., Neuner, C., Schartl, M., Volff, J.N., Winkler, C., 2005a.Hedgehog and retinoid signalling confines nkx2.2b expression to the lateralfloor plate of the zebrafish trunk. Mech. Dev. 122, 43–56.

Schäfer, M., Rembold, M., Wittbrodt, J., Schartl, M., Winkler, C., 2005b.Medial floor plate formation in zebrafish consists of two phases and requirestrunk-derived Midkine-a. Genes Dev. 19, 897–902.

Schauerte, H.E., van Eeden, F.J., Fricke, C., Odenthal, J., Strahle, U., Haffter, P.,1998. Sonic hedgehog is not required for the induction of medial floor platecells in the zebrafish. Development 125, 2983–2993.

Serluca, F.C., Fishman, M.C., 2001. Pre-pattern in the pronephric kidney field ofzebrafish. Development 128, 2233–2241.

Shimamura, K., Hartigan, D.J., Martinez, S., Puelles, L., Rubenstein, J.L., 1995.Longitudinal organization of the anterior neural plate and neural tube.Development 121, 3923–3933.

Stamataki, D., Ulloa, F., Tsoni, S.V., Mynett, A., Briscoe, J., 2005. A gradient ofGli activity mediates graded Sonic Hedgehog signaling in the neural tube.Genes Dev. 19, 626–641.

Strähle, U., Blader, P., Henrique, D., Ingham, P.W., 1993. Axial, a zebrafish geneexpressed along the developing body axis, shows altered expression incyclops mutant embryos. Genes Dev. 7, 1436–1446.

Strähle, U., Lam, C.S., Ertzer, R., Rastegar, S., 2004. Vertebrate floor-platespecification: variations on common themes. Trends Genet. 20,155–162.

Tanaka, H., Obata, K., 1984. Developmental changes in unique cell surfaceantigens of chick embryo spinal motoneurons and ganglion cells. Dev. Biol.106, 26–37.

Tian, J., Yam, C., Balasundaram, G., Wang, H., Gore, A., Sampath, K., 2003. Atemperature-sensitive mutation in the nodal-related gene cyclops reveals thatthe floor plate is induced during gastrulation in zebrafish. Development 130,3331–3342.

Varga, Z.M., Amores, A., Lewis, K.E., Yan, Y.L., Postlethwait, J.H., Eisen, J.S.,Westerfield, M., 2001. Zebrafish smoothened functions in ventral neuraltube specification and axon tract formation. Development 128, 3497–3509.

Westerfield, M., 2000. The Zebrafish Book. A Guide for the Laboratory Use ofthe Zebrafish Danio (Brachydanio) rerio, 4th ed. Univ. of Oregon Press,Eugene.

Winkler, C., Moon, R.T., 2001. Zebrafish mdk2, a novel secreted midkine,participates in posterior neurogenesis. Dev. Biol. 229, 102–118.

Yamada, T., Placzek, M., Tanaka, H., Dodd, J., Jessell, T.M., 1991. Control ofcell pattern in the developing nervous system: polarizing activity of the floorplate and notochord. Cell 64, 635–647.

Documents

Discontinuous organization and specification of the