Induction of anterior neural fates in the ascidian Ciona intestinalis

Clare Hudson*, Patrick Lemaire*

Laboratoire de GeÂneÂtique et Physiologie du DeÂveloppement, Institut de Biologie du DeÂveloppement de Marseille, CNRS-INSERM-Universite de la

Mediterranee-AP de Marseille, Parc Scienti®que de Luminy, Case 907, F-13288, Marseille Cedex 9, France

Received 26 September 2000; received in revised form 24 October 2000; accepted 25 October 2000

Abstract

The sensory vesicle of ascidians is thought to be homologous to the vertebrate forebrain and midbrain (Development 125 (1998) 1113).

Here we report the isolation of two sensory vesicle markers in the ascidian Ciona intestinalis, which are homologs of vertebrate otx and gsx

homeobox genes. By using these markers to analyze the induction of anterior neural tissue in Ciona, we ®nd that the restriction of anterior

neural fate to the progeny of the anterior animal blastomeres is due to a combination of two factors. The vegetal blastomeres show a

differential inducing activity along the anterior-posterior axis, while the competence to respond to this inducing signal is markedly higher in

the anterior animal blastomeres than in the posterior animal blastomeres. This differential competence to respond is also observed in response

to bFGF, a candidate neural inducer in ascidians (J. Physiol. 511.2 (1998) 347) and can be detected by the gastrula stage. Our results,

however, indicate that bFGF can only induce a subset of the responses of the endogenous inducer, suggesting that additional signals in the

embryo are necessary to induce a fully patterned nervous system. q 2001 Elsevier Science Ireland Ltd. All rights reserved.

Keywords: Gsx; Gsh; otx; Ciona intestinalis; Ascidians; Urochordate; Sensory vesicle; Neural induction; Neural competence; bFGF; Anterior-posterior pattern

1. Introduction

Ascidians are primitive invertebrate chordates that have

been classically used for developmental biology studies

(Conklin, 1905; Nishida, 1997). They are classed into two

orders, determined by the position of the gonad: the Enter-

ogona, with the gonads in the gut loop, including Ciona,

Ascidia and Phallusia, and the Pleurogona, with paired

gonads, including Styela, Halocynthia and Mogula

(reviewed in Satoh, 1994). During larval stages ascidians

exhibit a remarkably simple chordate body plan, possessing

a notochord and dorsal hollow nerve cord (for a general

review see Satoh, 1994; Nishida, 1997). Both the cleavage

pattern and the cell lineage are invariant in early ascidians.

Gastrulation starts at the 110-cell stage, when the majority of

cells are already restricted to form one tissue fate (Nishida,

1997). By larval stages, Ciona contains approximately 2500

cells of which 266 constitute the central nervous system

(Nicol and Meinertzhagen, 1991). Despite the high degree

of conservation of cleavage pattern and cell lineage between

the different ascidian species and an apparent conservation of

gene expression pro®les, regulation of gene expression may

have diverged signi®cantly in some cases. This was recently

highlighted in a study showing that the upstream regulatory

sequences of brachyury genes in Halocynthia and Ciona are

not conserved (Takahashi et al., 1999).

The nervous system of ascidian tadpoles consists of the

sensory vesicle (possibly equivalent to the forebrain and

midbrain), the neck region (midbrain or anterior hindbrain),

and the visceral ganglion and nerve cord (hindbrain and

spinal cord) (Wada et al., 1998). The sensory vesicle

contains two sense organs, each containing a melanocyte

or pigment cell: the posterior pigment cell is part of the

ocellus within a light-sensitive eye-like structure, and the

anterior pigment cell forms the gravity sensitive otolith.

Fate mapping experiments have determined that the

central nervous system mainly derives from the anterior

half of the eight-cell embryo (the A4.1 and a4.2 pairs;

Fig. 1) (Nishida, 1987; Nicol and Meinertzhagen,

1988a,b) (from henceforth blastomeres will be referred to

simply by their name, rather than as a pair). The sensory

vesicle forms from a4.2 derivatives (Fig. 1). The neck, visc-

eral ganglion and nerve cord predominantly develops from

A4.1 derivatives, with the dorsal-most cells being derived

from the b4.2 blastomere (Fig. 1).

Early experiments in ascidians revealed that the majority

of blastomeres, when isolated from eight-cell embryos,

continue to develop according to their normal lineage, high-

Mechanisms of Development 100 (2001) 189±203

0925-4773/01/$ - see front matter q 2001 Elsevier Science Ireland Ltd. All rights reserved.

PII: S0925-4773(00)00528-1

www.elsevier.com/locate/modo

* Corresponding authors. Tel.: 133-4-91-829248; fax: 133-4-91-

820682.

E-mail addresses: hudson@ibdm.univ-mrs.fr (C. Hudson),

lemaire@ibdm.univ-mrs.fr (P. Lemaire).

lighting the importance of cell autonomous maternal deter-

minants in ascidian development. One notable exception to

this mosaic development was a lack of neural differentiation

in animal half (a4.2 1 b4.2) embryos or animal blastomeres

cultured in isolation. It was further shown that induction of

neural fate in animal blastomeres, monitored by formation

of pigment cells, required an interaction with A4.1 blasto-

mere derivatives (Rose, 1939; Reverberi and Minganti,

1946, 1947). Analyzing membrane excitability in clea-

vage-arrested embryos of Halocynthia aurantium and Halo-

cynthia roretzi con®rmed that a signal was required from the

inducing cell A4.1 for the responding cell a4.2 to adopt a

neural fate and that this induction occurs within a speci®c

time window, being completed around the early gastrula

stage (Okado and Takahashi, 1990a). Although b4.2

normally does not differentiate into pigment cells or

neurons, it was shown that it could respond to the A4.1-

derived signals and develop both neural-type membrane

excitability in Halocynthia and form pigment cells in Styela

and Aspetto (Rose, 1939; Reverberi and Minganti, 1947;

Okado and Takahashi, 1990a). In all cases, however, the

competence of b4.2 to adopt a neural fate was signi®cantly

lower than that of a4.2 (Rose, 1939; Reverberi and

Minganti, 1947; Okado and Takahashi, 1990a). Finally,

more recent experiments have revealed that serine protease

or basic FGF treatment of the a4.2 blastomere (Ortolani et

al., 1979; Okado and Takahashi, 1990a, 1993; Inazawa et

al., 1998) can mimic the inductive properties of A4.1. The

action of bFGF was restricted to a similar time window as

induction via cell contact with the endogenous inducer,

while proteases showed slightly different kinetics, invoking

a neural response in a much shorter time window (Inazawa

et al., 1998).

These studies, which initiated the analysis of neural

induction in ascidians, had two major limitations. First, in

these experiments neural membrane excitability and

pigment cell formation identify only a subset of the a4.2-

derived neural cells. It is therefore important to use addi-

tional neural markers to test whether the results described

above can be extended to all a4.2-derived neural lineages.

Second, these studies focused on terminal differentiation

and therefore little is known about the intermediate steps

of neural tissue formation. For example, it is unclear when

the response of a4.2 and b4.2 to the signal starts to differ. Is

the inducing signal initially interpreted differently by each

cell, or is an identical early response translated later into the

adoption of distinct fates? The response to this important

question requires the isolation of early neural markers.

Here, we have cloned Ciona homologs of vertebrate otx

and gsx which are expressed in the a4.2-derived neural

lineages and which we used to address these issues.

2. Results

2.1. Identi®cation of Ciona homeobox-containing genes

belonging to the gsx/ind and otx families

We have cloned a Ciona intestinalis homeobox gene

encoding for a protein whose homeodomain is most similar

to that of mouse and medaka gsh (78% amino acid identity),

Amphioxus AmphiGsx (75% identity) and Drosophila ind

(73% amino acid identity) (Fig. 2b). Gsh (Genomic Screen-

ing Homeobox) genes were ®rst identi®ed in mouse (gsh-1

and 2; Singh et al., 1991; Valerius et al., 1995; Hsieh-Li et al.,

1995) and subsequently in medaka (Ol-Gsh 1; Deschet et al.,

1998), Amphioxus (AmphiGsx; Brooke et al., 1998) and

Drosophila (intermediate neuroblasts defective (ind);

Weiss et al., 1998) and expression of these molecules is

restricted to the nervous system. Outside of the gsh/ind

family, the most similar homeodomain is that of the

HoxD4 gene (65% amino acid identity). A shorter fragment

of the same cDNA has previously been cloned and named Ci-

Hbox2 (Di Gregorio et al., 1995). On the basis of sequence

comparisons and expression data (see below) we propose to

rename Ci-Hbox2, Ci-gsx. Outside of the homeodomain, the

predicted Ci-gsx amino acid sequence shares little homology

with its homologs in other species and lacks the polyalanine

tracts found in vertebrate gsh proteins (Valerius et al., 1995;

Hsieh-Li et al., 1995; Deschet et al., 1998) (Fig. 2a).

We also cloned a Ciona homolog of vertebrate otx and

Drosophila otd, which are involved in anterior neural

C. Hudson, P. Lemaire / Mechanisms of Development 100 (2001) 189±203190

Fig. 1. Fate map of the ascidian nervous system (taken and adapted from Nishida, 1987). On the right of the ®gure, the central nervous system is shown, and on

the left a lateral view of an eight-cell stage embryo is shown. The anterior nervous system, the sensory vesicle, is derived from a4.2. The nerve cord is derived

from A4.1, with the dorsal-most cells being descendants of b4.2.

formation from Drosophila to mouse (Finkelstein and

Boncinelli, 1994). The protein sequence of Ci-otx is very

similar to Hroth, the Halocynthia counterpart (Wada et al.,

1996), sharing 52% identity over the entire protein

sequence. The homeodomain of Ci-otx shares a high degree

of conservation with other Otx homeodomains (Fig. 2c).

2.2. Ciona gsx is expressed in the posterior sensory vesicle

Expression of Ci-gsx in early embryos was ®rst detected

by in situ hybridization at the mid-gastrula stage in two

blastomeres of the neural plate (Fig. 3a). The neural plate

of ascidians is a simple structure consisting of a monolayer

of cells, of which the cleavage pattern is well documented

allowing one to precisely identify each cell and know its

lineage (Nicol and Meinertzhagen, 1988a,b). Expression of

Ci-gsx is restricted to the neural plate (Fig. 3i,j). In order to

determine precisely which cells express Ci-gsx, we

collected embryos at sequential time points during

gastrula-neurula stages and revealed Ci-gsx expression by

in situ hybridization. We then stained the embryos with a

nuclear dye, which allowed us to identify each blastomere

by comparing the cell number and pattern of cell division

with the neural plate fate map established by Nicol and

Meinertzhagen (1988b). Ci-gsx expression was ®rst

detected in two cells in the third row of neural plate cells

(Fig. 3a,e). The precise identity of these cells was con®rmed

by following the pattern of cell cleavages. First the medial

four cells in row II divide to give the 44-cell neural plate

(Nicol and Meinertzhagen, 1988b) (Fig. 3b,f). This is

followed by the medial four cells in row I (Fig. 4c,g).

Next the medial four cells in rows III and V divide, along

with the lateral cells in rows I and II (Fig. 3d,h). From this

analysis, we conclude that Ci-gsx expression is ®rst detected

in the a9.33 pair of blastomeres, which are fated to form

posterior sensory vesicle (Nishida, 1987; Nicol and

Meinertzhagen, 1988b). Expression is absent from the ante-

rior sister of a9.33, the a9.34 blastomere which is also of

sensory vesicle lineage, suggesting that the onset of Ci-gsx

expression may be coincident with cell fate restriction of

posterior sensory vesicle. When the a9.33 precursor divides,

expression of Ci-gsx is maintained in both sister blasto-

meres, the a10.65 and a10.66 (Fig. 3d,h). This expression

pattern persists and by early tailbud stages, expression is

restricted to two domains bordering the midline (Fig. 3k).

By mid-tailbud stages in embryos co-stained with tyrosi-

nase, a marker for pigment cell precursors, the strongest

expression of Ci-gsx can be seen in the posterior sensory

vesicle bordering the posterior pigment cell (Fig. 3l). We

conclude that at neural plate stages Ci-gsx is a marker of

posterior a4.2-derived sensory vesicle precursors and,

following this lineage, subsequently marks the posterior

sensory vesicle during tailbud stages.

2.3. Expression of Ciona otx

Expression of Ci-otx is essentially the same as described

in Halocynthia with a few differences (Fig. 4 and for a

detailed description in Halocynthia, Wada et al., 1996).

Expression is detected at the early 32-cell stage in two

posterior endoderm blastomeres (B6.1) and the muscle/

mesenchyme/notochord precursors (B6.4, B6.2; Fig. 4a,b).

C. Hudson, P. Lemaire / Mechanisms of Development 100 (2001) 189±203 191

Fig. 2. (a) Predicted amino acid sequence of Ci-gsx. An in-frame stop codon was present upstream of the methionine, indicating that the protein sequence is full

length. The homeobox is shown in bold. (b) Sequence comparison between homeoboxes of Ci-gsx, mouse mgsh-1 and -2, medaka olgsh-1, amphioxus

AmphiGsx, and Drosophila ind. (c) Sequence comparisons of the homeobox sequences between Ci-otx, Halocynthia Hroth, Drosophila otd, rat otx2 and

Zebra®sh otx3.

By the late 32-cell stage (Fig. 4c,d), expression begins in

b6.5, the derivatives of which gives rise to epidermis, tail tip

muscle and endoderm and to the most dorsal ependymal cell

of the nerve cord. Expression also appears in a6.5 which

gives rise to the sensory vesicle, palps and pharynx. Forty-

four-cell stage embryos show expression in a6.5, b6.5 and

B6.4 (Fig. 4e,f). In some embryos expression could still be

detected in B6.1 (for example, see Fig. 7o). This is slightly

different to Halocynthia embryos where expression of Hroth

is no longer detectable in B6.4 mesenchyme precursors at

the 44-cell stage (Wada et al., 1996). By the 110-cell stage,

expression of Ci-otx is detected throughout the endoderm,

mesenchyme and trunk lateral cell precursors, but becomes

barely detectable in the neural precursors (Fig. 4g,h). This is

also different from what is observed in Halocynthia and

Herdmania where expression is strongly maintained in the

a4.2-derived neural precursors at the 110-cell stage (Wada

et al., 1996; Hinman and Degnan, 2000). Shortly after

gastrulation commences, Ci-otx is strongly upregulated in

the neural plate precursors (see Fig. 7j) and subsequently

downregulated in the endoderm and mesenchyme. By

neural plate stages, expression is observed as an intense

band across the neural plate (Fig. 4i). As for Ci-gsx, using

double staining with a nuclear dye, we determined that this

site of expression corresponds to the six cells in rows III and

IV of the neural plate, i.e. the sensory vesicle and pharynx

precursors (Nicol and Meinertzhagen, 1988b; Nishida,

1987) (Fig. 4l,m). Subsequently, expression becomes stron-

ger throughout the a4.2-derived neural plate (rows III±VI)

(Fig. 4j,n,o) persisting until tailbud stages when expression

is detected in the forming sensory vesicle and the anterior

epidermal cap (Fig. 4k). Comparison of the expression

patterns of Ci-otx and Ci-gsx indicates that from its onset

of expression, transcripts of Ci-gsx are found within the

most posterior expression domain of Ci-otx.

Therefore, from mid-gastrula stages these two molecules

speci®cally mark the anterior nervous system, with Ci-otx

marking the entire sensory vesicle and part of the anterior

C. Hudson, P. Lemaire / Mechanisms of Development 100 (2001) 189±203192

Fig. 3. Expression pattern of Ci-gsx. (a±d) In situ hybridization of gastrula-neurula stage embryos co-stained with a nuclear dye to identity cell positions. In

each case dorsal views are shown; anterior is up. Cell rows that have just undergone cleavage are marked with an arrow. (e±h) Schematic representation of the

neural plate; anterior is up. Newly cleaved cells are marked with an arrow. For simplicity, the b4.2-derived neural plate and the lateral neural plate cells derived

from A4.1 in rows I and II are omitted. Rows I and II are A4.1 derivatives and rows III±VI are a4.2 derivatives. Cell cleavages were compared to Nicol and

Meinertzhagen (1988b). Cells that express Ci-gsx are coloured blue and cell names are indicated. (i) Lateral view of an early neural plate stage embryo

hybridized with Ci-gsx probe; dorsal is up, anterior is to the left. (j) Schematic representation of the embryo shown in (i); neu.pl., neural plate; noto., notochord;

ecto., ectoderm; endo., endoderm. (k) Early tailbud stage, dorsal view, anterior is to the left. (l) Late tailbud co-stained for tyrosinase (in brown) and Ci-gsx

expression (in blue), dorsal view, anterior is to the left. The scale bar throughout is 50 mm.

ectoderm and Ci-gsx marking the posterior sensory vesicle.

Using these markers, we carried out a series of blastomere

isolation and recombination experiments to study anterior

neural induction.

2.4. Induction of Ci-otx and Ci-gsx requires interaction with

the vegetal pole

To con®rm that a signal is required from the vegetal cells

to the animal cells for induction of anterior neural gene

expression, we bisected embryos into animal and vegetal

halves at the eight-cell stage (Fig. 5a), cultured the partial

embryos until early tailbud stages and assayed for gene

expression by in situ hybridization. Similar to the results

found for Hroth in Halocynthia (Wada et al., 1999) we

found that neither Ci-otx (98% negative, n � 45) nor Ci-

gsx (100% negative, n � 53) are expressed in animal half

explants cultured in isolation, suggesting that interaction

with the vegetal half is required for expression of these

genes (Fig. 5c and data not shown). In this experiment we

also observed very little expression in vegetal half explants

(Fig. 5d and data not shown). Embryo quarter explants

(a4.2, b4.2 and B4.1) cultured in isolation from the eight-

cell stage failed to express Ci-otx at tailbud stages (data not

shown). However, we found strong expression of Ci-otx in

A4.1 derivatives in 17% (n � 76) of cases with a further

20% showing a much smaller domain of expression which

we designated as a `patch' of expression (see Fig. 6c for a

C. Hudson, P. Lemaire / Mechanisms of Development 100 (2001) 189±203 193

Fig. 4. Expression pattern of Ci-otx. (a±h) Expression of Ci-otx during cleavage stages. (b,d,f,h) Embryo schematics representing the embryos in (a,c,e,g),

respectively. The thicker line indicates the division of animal and vegetal cells. Blastomeres expressing Ci-otx are marked with a red dot and their blastomere

names are shown. (a) Early 32-cell stage, expression can be seen in B6.4, B6.2 and B6.1, animal view, anterior is up. Expression in B6.1 can be seen through the

embryo due to its transparency. (b) Animal and vegetal views of the posterior of the embryo to show expression in all cells which express Ci-otx. (c,d) Late 32-

cell stage, expression is now detected in a6.5 and b6.5, animal views, anterior is up. (e,f) 44-cell stage, animal views, anterior is up. In some cases expression in

B6.2 was still detectable. (g,h) 110-cell stage, expression is found strongly throughout the endoderm precursors, and more weakly in the mesenchyme and trunk

lateral cell precursors, vegetal views, anterior is up. (i,j) Neural plate stages, dorsal views, anterior is to the left. (k) Early tailbud, dorsal view, anterior is to the

left. (l±o) Detailed analysis of the expression pattern of Ci-otx during the neural plate stages in embryos co-stained with a nuclear dye, dorsal views, anterior is

up. (m,o) schematic representations of the neural plates found in (l,n), respectively. Blastomeres expressing Ci-otx are shown in blue with weaker expression

shown in pale blue. The scale bar in (a) is 50 mm and applies to all panels.

description of the scheme used for scoring expression

domains). In these studies we observed many vegetal half

embryos (n � 355); these showed an average positive score

of 25% and a patch score of 10% (Fig. 6d). Expression in

A4.1 derivatives may represent residual expression in the

anterior endoderm, which is obscured by the neural expres-

sion in whole embryos.

As mentioned above, the earliest expression of Ci-otx in

a4.2 derivatives was detected at the late 32-cell stage. To

determine whether the absence of Ci-otx expression

detected in animal explants at tailbud stages was due to a

lack of initiation of Ci-otx expression or to a lack of main-

tenance of this expression, we bisected embryos into animal

and vegetal halves at the eight-cell stage and assayed at the

44-cell stage for gene expression. No expression of Ci-otx

was observed in the animal explants (94% negative,

n � 101), whereas the expression in the vegetal explants

appeared unaffected (81% positive, n � 121); Fig. 5e±g).

This implies a failure of initiation of Ci-otx expression in

the neural lineage and demonstrates that vegetal to animal

cell signalling is required prior to the 44-cell stage for the

initiation of Ci-otx expression. Similar results were found in

a4.2 or b4.2 explants isolated at the eight-cell stage and

cultured until the 44-cell stage, gastrula stage, and tailbud

stage (see Fig. 7).

These experiments show that cell contacts between

animal and vegetal blastomeres are required for the induc-

tion of anterior neural markers. Analysis at the 44-cell stage

demonstrates that this cell interaction is required for the

early induction of Ci-otx in the anterior neural precursors.

2.5. Anterior neural induction involves a differential

inductive capacity of the vegetal blastomeres and a

difference in the competence of the responding animal

blastomeres

The next question asked was whether restriction of Ci-otx

and Ci-gsx to the anterior neural plate relies on a differential

inducing ability of the vegetal blastomeres or a differential

competence of the animal blastomeres to respond. This was

addressed using a series of blastomere co-isolations and

recombinations (Fig. 6a), in which partial embryos were

cultured until the tailbud stage and then assayed for gene

C. Hudson, P. Lemaire / Mechanisms of Development 100 (2001) 189±203194

Fig. 5. (a) Schematic representation of the manipulation carried out: bisection into animal (a4.2 1 b4.2) and vegetal (A4.1 1 B4.1) halves. For simplicity only

one of each blastomere type is shown both for the embryo and explants. In reality, the half embryos contained four cells, two of each type. (b±j) Animal±vegetal

halves, as in (a), assayed for expression of Ci-otx at tailbud stages (b±d), or at the 44-cell stage (e±g). The percentage values indicate the proportion of explants

that the panel represents for each sample. The scale bar in (b) is 50 mm and applies to all panels.

expression by in situ hybridization. A summary of the data

is represented as graphs in Fig. 6d.

In anterior half explants (A4.1 and a4.2), we found that a

large area was induced to express Ci-otx (78%, n � 45; Fig.

6cii) while a smaller area expressed Ci-gsx (86%, n � 74;

Fig. 6bii) in a manner strikingly similar to that observed in

whole embryos. This suggests that the posterior half of the

embryo was not required for anterior neural induction.

Recombinations of isolated A4.1 and a4.2 blastomeres

were carried out as a control for the recombination techni-

C. Hudson, P. Lemaire / Mechanisms of Development 100 (2001) 189±203 195

Fig. 6. Recombination assays. (a) Schematic representation of manipulations carried out. For simplicity only one of each blastomere type is shown in both the

embryo view and the explants. In reality, each quarter embryo contained two cells, so that recombinations ®nally contained four cells, two of each type. (b) Ci-

gsx expression: (i) control embryo, anterior is to the left; (ii) A4.1 1 a4.2 anterior half embryo; (iii) A4.1 1 a4.2 recombinations; (iv) A4.1 1 b4.2 recombina-

tions; (v) B4.1 1 a4.2 recombinations; and (vi) B4.1 1 b4.2 posterior half embryo. The percentage values indicate the proportion of explants that the panel

represents for each sample. The scale bar in (i) is 50 mm and applies to all panels. (c) Scoring chart used for Ci-otx. (i) Control embryo, anterior is to the left.

Explants were scored as follows: positive (ii); the lowest domain of expression scored as positive is shown in (iii); less than that seen in (iii) was scored as a

patch of expression, with the lowest expression domain scored as a patch shown in (iv); less than (iv) to (v) was scored as negative. (d) Graphs showing the

average value from the pooled data from three experiments for each marker in the ®rst ®ve bars; the following bars are the average score for data pooled from a

number of experiments. Blocked colour indicates explants scored as positive; hatched boxes refer to explants scored as a patch; W, whole embryos; A, A4.1; B,

B4.1; a, a4.2; b, b4.2; rec., recombined blastomeres.

que and gave essentially the same result as co-isolation, with

high induction of Ci-otx (80%, n � 65) and Ci-gsx (68%,

n � 69; Fig. 6biii), showing that the recombination techni-

que was reliable.

In order to investigate if the animal blastomeres (a4.2 and

b4.2) showed differential ability to respond to the anterior

neural inducers emitted from A4.1 descendants, we recom-

bined A4.1 with b4.2 blastomeres. In this experiment, strong

Ci-otx expression was observed in 28% (n � 86) of cases

with a further 19% showing a small patch of expression

(see Fig. 6c for description of scoring procedure). Taking

into account that around 20% of A4.1 explants cultured

alone expressed Ci-otx, we conclude that b4.2 can only acti-

vate Ci-otx in response to signals from A4.1 descendants in a

small number of cases. Ci-gsx was expressed in 6% (n � 70)

of these recombinations and in these cases the area expres-

sing Ci-gsx was small (Fig. 6biv). The marked reduction in

gene expression in these recombinations compared to

A4.1 1 a4.2 recombinations shows that b4.2 has a much

lower competence than a4.2 to adopt an anterior neural fate

in response to the A4.1-derived inducer, thus supporting and

extending the work of Rose (1939), Reverberi and Minganti

(1946, 1947) and Okado and Takahashi (1990a).

The next question addressed was whether there was a

difference in ability to induce anterior markers between

the two vegetal blastomere descendants (A4.1 and B4.1).

In this set of experiments, we recombined the responding

blastomere a4.2 with B4.1. Ci-otx was strongly induced in

11% (n � 95) of cases with a further 20% showing a patch

of expression. Ci-gsx was never observed in these recombi-

nations (n � 79; Fig. 6bv). We conclude that B4.1 deriva-

tives have some ability to induce anterior markers but much

C. Hudson, P. Lemaire / Mechanisms of Development 100 (2001) 189±203196

Fig. 7. Ci-otx expression in bFGF-treated animal explants. (a) Early tailbud control embryo, lateral view, anterior is to the left. (b) Animal half (a4.2 1 b4.2)

explant cultured alone. (c) Animal explant treated with bFGF showing expression of Ci-otx and altered morphology. (d) Dose response of Ci-otx expression to

bFGF concentrations (in mg/ml); Os, sea water (no bFGF); OB, BSA in sea water. Explant type and culture conditions for (e±s) are shown along the top of each

column of panels. (e) As for (a). (f±i) Animal explants cultured until tailbud stages. (j) Gastrula stage, vegetal view, anterior is up; expression is strong

throughout the endoderm, mesenchyme and brain/pharynx precursors. (k±n) Animal explants cultured until gastrula stage; the explant in (l) appears much

smaller than that seen in (k) due to the explant having formed a ball of cells in (l) rather than being ¯attened during mounting as in (k). (o) 44-cell stage,

orientation as for (j). (p±s) Animal explants cultured until the 44-cell stage. The percentage values indicate the proportion of explants that the panel represents

for each sample. The scale bars in (a) are for (a±c), those in (e) are for (e,j,o) and those in (f) are for (f±i,k±n,p±s). All scale bars are 50 mm.

less than is found in A4.1. The lack of expression of Ci-gsx

in B4.1 1 a4.2 recombinations may indicate that the induc-

tion of Ci-otx in these explants represents anterior ectoderm,

rather than sensory vesicle. Posterior half (B4.1 and b4.2)

explants showed no expression of Ci-otx (98% negative,

n � 111) or Ci-gsx (100% negative, n � 94; Fig. 6bvi).

These results show that there is a difference in the compe-

tence of the animal blastomeres to respond to anterior neural

inducers as well as a difference in the ability of the vegetal

blastomeres to induce anterior neural markers, and we

propose that both of these factors play an important role

in restricting anterior neural induction to the a4.2 descen-

dants. Our results indicate that the conclusions drawn from

the work of Rose (1939), Reverberi and Minganti (1946,

1947) and Okado and Takahashi (1990a,b) may be generally

applicable to anterior neural induction in ascidians.

2.6. bFGF induces Ci-otx in a4.2, but not b4.2 derivatives

In the recombination experiments described in the

previous section, we analyzed expression of Ci-otx and

Ci-gsx at the tailbud stage. It was therefore not possible to

distinguish between a loss of maintenance of anterior neural

induction in b4.2 derivatives or a lack of induction. This

would require analysis at earlier time points, but the endo-

dermal expression of Ci-otx in A4.1 and B4.1 derivatives

during the cleavage and gastrula stages would render this

experiment dif®cult to interpret. To overcome this dif®culty,

we tested, at different time points after the onset of treat-

ment, the response of a4.2 and b4.2 blastomeres to bFGF, a

growth factor which has been shown to mimic the neural

inducing activity of A4.1 derivatives (Inazawa et al., 1998)

and to be critical for neural development in chick embryos

(Wilson et al., 2000; Streit et al., 2000). To con®rm that

bFGF mimicked the activity of the endogenous inducer,

we ®rst tested whether bFGF could induce expression of

Ci-otx in animal half explants by the tailbud stage. Animal

half explants were isolated at the eight-cell stage and treated

with sea water or various concentrations of bFGF (1, 10 or

100 ng/ml) until the early tailbud stages. Animal explants

cultured to the tailbud stage in sea water alone have a

rounded, disc-shape morphology (Fig. 7b). bFGF-treated

explants appeared more compact and had striking bumps

and protrusions (Fig. 7c). These morphological changes

were accompanied by induction of Ci-otx expression and

were dose-dependent, occurring more frequently with

increasing doses of bFGF (Fig. 7d). Half of the explants

were induced to express Ci-otx at around 10 ng/ml bFGF

and almost 100% expressed Ci-otx after treatment with 100

ng/ml bFGF, a result very similar to that seen in Halo-

cynthia (Inazawa et al., 1998). For all subsequent experi-

ments, unless otherwise stated, 100 ng/ml bFGF was used

because this dose gave the most robust induction of Ci-otx.

As expected, bFGF, like the endogenous inducer, elicited

different responses by the tailbud stage in a4.2 and b4.2.

Treatment with bFGF induced Ci-otx in a4.2 derivatives

(93%, n � 54), whereas b4.2 derivatives showed a much

reduced response (10%, n � 31; Fig. 7e±i). These results

indicate that bFGF can be used as a replacement to the

A4.1-derived inducer to study the regulation of Ci-otx. This

enabled us to look at an earlier stage at the response of a4.2

and b4.2 to bFGF. bFGF-treated explants were ®rst analyzed

at the early gastrula stages. Eighty-®ve percent (n � 46) of

a4.2 explants and 24% (n � 38) of b4.2 explants expressed

Ci-otx (Fig. 8m,n). Expression was not seen in a4.2 explants

(95% negative, n � 40) or b4.2 explants (100% negative,

n � 40) cultured in sea water alone (Fig. 7k,l). To analyze

this further, we looked at the response to bFGF at the 44-cell

stage. We found that incubation in the presence of bFGF

induced expression of Ci-otx in a4.2 (96%, n � 27) and

b4.2 (100%, n � 27) derivatives when assayed at the 44-

cell stage (Fig. 7r,s), whereas isolated a4.2 and b4.2 blasto-

meres cultured in sea water alone showed no expression

(a4.2, 100% negative, n � 26; b4.2, 100% negative,

n � 19; Fig. 7p,q). It appears that both a4.2 and b4.2 animal

blastomere derivatives can respond to bFGF signalling by

expressing Ci-otx at the 44-cell stage, but that by the early

gastrula stage, this expression is only detected in a4.2 deri-

vatives. There are two possibilities to account for the expres-

sion of Ci-otx at the 44-cell stage in b4.2 explants treated with

bFGF; either the blastomeres are partially converted to an

anterior neural fate which they cannot maintain until the early

gastrula stage, or the expression is indicative of the b4.2

derivative which normally expresses Ci-otx at the 44-cell

stage, i.e. the b6.5 blastomere (Figs. 3e,f and 7o). The

absence of a b6.5 marker does not allow us to con®rm

which of these events is occurring. This result suggests,

however, that FGF can act very early (prior to the 44-cell

stage) to induce Ci-otx in the a4.2 and b4.2 lineages.

Although bFGF seems to mimic the endogenous signal in

its ability to induce the expression of Ci-otx, a signi®cant

difference was seen when analyzing Ci-gsx expression. We

found no induction of Ci-gsx in animal half embryos

isolated at the eight-cell stage and incubated in bFGF until

the tailbud stage (10 ng/ml bFGF 0% positive, n � 38; 100

ng/ml 0% positive, n � 19). Ci-gsx was also not observed in

bFGF-treated or untreated a4.2 or b4.2 explants (data not

shown). This indicates that bFGF is able to elicit in animal

cap cells only part of the genetic programme activated by

the signals emitted by the progeny of A4.1, which suggests

the existence of additional inducing molecules.

Importantly, in these experiments we could not distin-

guish between induction of sensory vesicle or anterior ecto-

derm since both tissues express Ci-otx in whole embryos.

This prompted us to characterize in more detail the fate of

bFGF-treated blastomeres.

2.7. Fate of bFGF-treated a4.2 explants

We ®rst looked for neural tissue in bFGF-treated a4.2

blastomere derivatives by looking for the presence of differ-

entiated neurons at swimming larvae stages. To do this, we

C. Hudson, P. Lemaire / Mechanisms of Development 100 (2001) 189±203 197

utilized anti-acetylated tubulin antibodies which, in whole

embryos, label cilia (Crowther and Whittaker, 1992) and

neurons along the entire neuraxis (Nakatani et al., 1999).

We observed strong staining in both neurons (Fig. 8a) and

the cilia pairs along the tail (Fig. 8b). These cilia pairs

originate from cell bodies of peripheral neurons, which

run under the dorsal and ventral epidermis (Crowther and

Whittaker, 1994).

In untreated a4.2 explants (95% negative, n � 61; Fig. 8f)

or A4.1 explants cultured alone (n � 46; data not shown) we

observed very few neurons or cilia. In a4.2 blastomeres co-

isolated with the natural inducer A4.1, neurons were

observed in 40% (n � 118) of cases (Fig. 8c). In these ante-

rior half explants we observed very few cilia (7%). bFGF

treatment of a4.2 explants induced the formation of neurons

in 28% (n � 92) of cases, the majority of which appeared

inside the explant (Fig. 8g), although a few neurons with

large cell bodies appeared closer to or at the surface of the

C. Hudson, P. Lemaire / Mechanisms of Development 100 (2001) 189±203198

Fig. 8. Acetylated tubulin staining. (a) Control embryo sensory vesicle and anterior part of tail, showing labelling of neurons (short thin arrow) in the brain and

nerve cord, anterior is to the left, dorsal up. An asterisk is placed just posterior to the otolith to mark its position. The ocellus is not in the plane of focus.

Neurons are also present in the posterior sensory vesicle, but are out of the plane of focus in this picture. (b) Whole embryo tail showing labelling of neurons in

the nerve cord (short thin arrow) and cilia pairs in the tail (arrows). (c) Anterior A4.1 1 a4.2 halve showing neurons (short thin arrows). (d) Posterior

B4.1 1 b4.2 showing cilia pairs (arrows). (e) A4.1 1 b4.2 recombination showing neuron formation (short thin arrows). (f) Untreated a4.2 explant. (g,h) a4.2

explant treated with bFGF showing neuron (short thin arrows) formation. (i) Untreated b4.2 explant. (j) b4.2 explant treated with bFGF showing neuron (short

thin arrows) formation. (k) b4.2 explant treated with bFGF showing cilia pair (arrows) formation. Scale bar in (a) applies to all panels, 50 mm.

explant and may represent a different type of neuron (Fig.

8h). We also observed induction of cilia in a small number

of cases (18%). Neurons formed only rarely in B4.1 1 a4.2

recombinations (8%, n � 153), although cilia formed in

30% of cases (data not shown). Neither neurons nor cilia

formed B4.1 explants cultured alone (n � 66; data not

shown).

These results show that bFGF can induce neurons in a4.2

albeit to a lesser extent than the endogenous A4.1-derived

inducer. a4.2 can also respond in a number of cases to B4.1

and bFGF by producing cilia.

The use of acetylated beta tubulin antibodies did not

allow us to distinguish between the formation of neurons

from the central or peripheral nervous systems. To test for

the presence of sensory vesicle-derived tissue in the bFGF-

treated explants, we looked in treated blastomeres for the

formation of pigment cells, a late differentiation marker for

the sensory vesicle lineage, which originates from the same

anterior-posterior position as Ci-gsx during neural plate

stages. Pigment cells did not form in untreated isolated

a4.2 blastomeres cultured to the swimming larval stage

(0%, n � 71; Fig. 9d). However, bFGF treatment induced

pigment cells in 36% of a4.2 explants (one to four pigment

cells per explant, n � 137; Fig. 9f), similar to the frequency

observed in A4.1-a4.2 explants (35%, one to three pigment

cells, n � 60; Fig. 9b). The majority of pigment cells which

formed in these experiments contained large pigment gran-

ules, suggesting that they are of the otolith type. Recombi-

nations of a4.2 with B4.1 led to induction of pigment cells in

only 4% (n � 153) of cases.

Taken together, these results show that bFGF-treated a4.2

blastomeres do indeed contain sensory vesicle tissue,

suggesting that while bFGF cannot fully replace A4.1 as

an anterior neural inducer, the induction of Ci-otx by

bFGF does not solely re¯ect induction of anterior ectoderm.

Additionally, these results further show that similar to the

results found for Ci-otx and Ci-gsx, B4.1 is a much less

competent inducer of neurons and pigment cells in a4.2

derivatives than A4.1.

2.8. Fate of bFGF-treated b4.2 explants

In the majority of cases involving b4.2 as the responding

blastomere, we observed no induction of Ci-otx or Ci-gsx.

Therefore, we looked to see if b4.2 explants could respond

to bFGF or A4.1 and produce some neural tissue. In both

uninduced a4.2 and b4.2 explants epidermis differentiation

was seen as judged by the presence of the tunic (Fig. 9d,e,

arrows). However, in either a4.2 or b4.2 explants treated

with bFGF, the surface of the whole or part of the explant

was devoid of tunic (Fig. 9f,g), suggesting that in b4.2

explants as in a4.2 explants, bFGF may suppress epidermal

differentiation.

Analysis of acetylated tubulin in b4.2 gave unexpected

results. While neurons did not form in untreated b4.2

explants (97% negative, n � 99; Fig. 8i), recombinations

with A4.1 induced neuronal formation in 40% (n � 129;

Fig. 8e) of recombinations. Posterior (B4.1 1 b4.2) explants

possessed neurons in fewer cases (11%, n � 123) and these

were generally signi®cantly less numerous than those seen

in A4.1 1 b4.2 blastomere recombinations. Explants treated

with bFGF showed neuronal differentiation in 62% (n � 6;

C. Hudson, P. Lemaire / Mechanisms of Development 100 (2001) 189±203 199

Fig. 9. Pigment cell formation at hatching/swimming tadpole stage. (a) Control embryo with two pigment cells, ocellus and otolith, dorsal is up, anterior is to

the left. (b) A4.1 1 a4.2 anterior explant with two otolith-type pigment cells. (c) B4.1 1 b4.2 posterior explant. (d) Untreated a4.2 explant. (e) Untreated b4.2

explant. (f) a4.2 treated with bFGF containing three otolith-type pigment cells. (g) b4.2 treated with bFGF. Arrows indicate the presence of tunic. Arrowheads

mark the border between the tunic positive and tunic negative parts of the explant. The scale bar in (a) is 50 mm and applies to all panels.

Fig. 8j) of cases. Both the proportion of explants induced to

form neurons and the numbers of neurons per positive

explants were higher than that found in bFGF-treated a4.2

explants.

In posterior explants, cilia pairs formed (63%, n � 123;

Fig. 8d). Recombinations of A4.1 1 b4.2 could also

produce cilia, although less frequently (24%, n � 129).

We also observed induction of cilia in b4.2 by bFGF

(68%, n � 66; Fig. 8k). We observed very few cases of

pigment cell formation in A4.1 1 b4.2 recombinations or

bFGF-treated b4.2 (less than 4% in both cases, data not

shown) and never in B4.1 1 b4.2 posterior explants.

Taken together, these data show that b4.2 is competent to

respond to bFGF and A4.1 and adopt a neural fate. As judged

by the lack of Ci-otx, Ci-gsx and pigment cell formation in

b4.2, the neural tissue induced is not anterior in character.

While B4.1 does not induce high levels of Ci-gsx, Ci-otx,

pigment cells or neurons in either a4.2 or b4.2, B4.1 appears

to induce peripheral neurons, as characterized by cilia, in

both animal blastomeres at higher levels than A4.1. These

results also show that b4.2 is more competent to produce cilia

in response to B4.1 or bFGF than is a4.2.

3. Discussion

3.1. Expression patterns of Ci-otx and Ci-gsx

Here, we report the cloning and expression patterns of

two sensory vesicle markers in the ascidian, Ciona intesti-

nalis. Ci-otx is expressed dynamically during early cleavage

and is restricted to a4.2-derived neural precursors from mid-

gastrula stages. The expression pattern of Ci-otx in endo-

derm and anterior neural precursors appears somewhat simi-

lar to that reported in vertebrates. In mouse, otx2 is

expressed in the endoderm which is critical for anterior

neural induction (Rhinn et al., 1998) and subsequently in

the fore- and midbrain (Simeone, 1998). A detailed compar-

ison between ascidian and vertebrate otx expression from

gastrula stages can be found in Wada et al. (1996).

Ci-gsx is expressed in a4.2-derived posterior sensory

vesicle precursors from early neurula to tailbud stages

within the posterior expression domain of Ci-otx. In

mouse and medaka, gsh genes are expressed in two bands

of cells running along the anterior-posterior axis of the

central nervous system located in the intermediate neuro-

genic region, with additional expression in speci®c regions

of the fore-, mid- and hindbrain (Hsieh-Li et al., 1995;

Szucsik et al., 1997; Valerius et al., 1995; Deschet et al.,

1998). In Drosophila, ind is expressed in lateral columns of

the ventral nerve cord, in a pattern remarkably similar to

that described for the mouse and medaka Gsh genes (Weiss

et al., 1998; Arendt and Nubler-Jung, 1999). In contrast,

amphioxus was found to have a more limited expression

domain, being restricted to the cerebral vesicle, the equiva-

lent of the vertebrate fore- and midbrain (Brooke et al.,

1998). We ®nd that expression of Ci-gsx is more reminis-

cent of the expression seen in amphioxus, being restricted to

the sensory vesicle (Brooke et al., 1998). Expression of Ci-

gsx appears much earlier in development than its counter-

parts in other chordate species.

3.2. Anterior neural induction in ascidians

Early studies showed that both a4.2 and b4.2 blastomeres

could respond to neural inducers producing pigment cells and

neuron-type membrane excitability, and that b4.2 had a lower

competence to respond than a4.2 (Rose, 1939; Reverberi and

Minganti, 1946, 1947; Okado and Takahashi, 1990a).

However, since these studies looked at differentiation of

speci®c cell types, it was not clear whether the reduced

competence of b4.2 was due to reduced general neural induc-

tion, or if the induced blastomeres lacked the ability to

produce these speci®c neural cell types. Additionally, in clea-

vage-arrested embryos, only one cell type can be adopted and

therefore a partial neural conversion would not be detected.

In this study we looked again at neural induction using

Ci-otx and Ci-gsx which enabled us to look at earlier steps of

neural tissue formation and to focus on anterior neural fates.

In addition, our analysis did not require cleavage arrest and

therefore we could analyze neural induction at a more

physiological level.

Using a series of blastomere co-isolations and recombi-

nations we found that high levels of anterior neural induc-

tion occurred only when both the anterior vegetal and

anterior animal blastomeres were recombined. Anterior

and posterior vegetal cells have a different inductive capa-

city while anterior and posterior animal cells show a differ-

ent competence to respond. Furthermore, we found that a4.2

and b4.2 blastomeres also have a strikingly different compe-

tence to respond to bFGF and that this difference in compe-

tence can be detected as early as the gastrula stage.

Although it seems that only a4.2 can adopt an anterior

neural fate, our experiments with A4.1 1 b4.2 and bFGF-

treated b4.2 suggest that the difference in competence

between a4.2 and b4.2 is not the result of an inability of

b4.2 to respond to neural inducing signals: the morphology

of bFGF-treated b4.2 explants is altered, tunic formation is

decreased and these explants produce many neurons and

cilia. The precise identity of the neurons produced in

response to A4.1 and bFGF demands a much more detailed

neuronal fate map and a collection of neuronal markers

speci®c to each lineage. Very little is known about the

formation of peripheral neurons in ascidians. We were

able to show that cilia pairs, which mark a speci®c type of

peripheral neuron (Crowther and Whittaker, 1994), could be

induced by B4.1 much more readily than by A4.1. Similarly,

b4.2 was more competent to produce cilia in response to

B4.1 and bFGF than a4.2.

3.3. Basis for a difference in competence

It is likely that the difference in competence to form

C. Hudson, P. Lemaire / Mechanisms of Development 100 (2001) 189±203200

anterior neural tissue between a4.2 and b4.2 descendants is

established maternally, since it is already apparent when

blastomeres are isolated at the eight-cell stage. After ferti-

lization, ascidian eggs undergo extensive cytoplasmic rear-

rangement, termed ooplasmic segregation, resulting in the

localization of a region of cytoplasm termed the myoplasm

or posterior cytoplasm to a sub-equatorial position in the

egg, the future posterior pole. This ooplasmic segregation

is responsible for the correct positioning of the embryonic

anterior-posterior axis, for positioning of endoderm, muscle

and ectoderm lineages, and also for the establishment of the

correct early cleavage asymmetries (Nishida, 1992, 1993,

1994a,b). It has been proposed that posterior fate is domi-

nant over anterior and is determined by maternal determi-

nants localized in the posterior cytoplasm. The default

anterior fate would only occur in the absence of these deter-

minants (Nishida, 1994b). This early segregation event may

also be responsible for the difference in competence

between the animal blastomeres in terms of anterior neural

induction. One could imagine that the posterior cytoplasm

inherited predominantly by B4.1 is also present in b4.2 and

confers upon this cell a different responsiveness to neural

inducers than a4.2. In embryos in which the posterior cyto-

plasm is removed at the one-cell stage, the embryo develops

in a radially symmetrical, anteriorized fashion (Nishida,

1994b). If the posterior vegetal cytoplasm is then fused

into the anterior vegetal side, the polarity of the subsequent

embryos is reversed. It would be very interesting to deter-

mine in normal embryos whether b4.2 cells inherit some

posterior vegetal cytoplasm, and whether this is responsible

for the difference in competence that we see across the

animal hemisphere.

A differential competence of animal blastomeres to

respond to neural inducing signals is also observed in verte-

brates. In Xenopus, dorsal (future neural plate) adopts a

neural fate much more readily than ventral (future epider-

mis) ectoderm (Sharpe et al., 1987; Sokol and Melton, 1991;

Otte and Moon, 1992). Recent evidence suggests that a

maternal event, implicating the stabilization of b-catenin,

enhances the competence of the dorsal ectoderm in Xenopus

embryos to become neural, the subsequent induction of

neural tissue relying on zygotic molecules secreted from

the organizer (Baker et al., 1999; Harland, 2000). However,

in Xenopus, this difference in competence is between neural

and epidermal fate, whereas in ascidians our results suggest

that this difference appears to be between anterior and

posterior neural fates. Additionally, in Xenopus, the differ-

ence in competence can be overridden and ventral ectoderm

can be converted to a neural fate. In ascidians, as we show

here, b4.2 explants very rarely adopt an anterior neural fate.

In ascidians, cell fates are determined in embryos with very

few cells, whereas in Xenopus cell fates are determined in

embryos with many more cells. One could imagine that if a

diffusible inducer was responsible for neural induction, in

embryos with a small cell number the competence to

respond would have to be much more tightly regulated

than in much larger embryos with many cells and a highly

regulative mode of development.

3.4. Role of bFGF in neural induction

It has been proposed that bFGF may be the natural neural

inducer in ascidians (Inazawa et al., 1998). The time

window of action and the response of animal blastomeres

to bFGF are similar to those observed with the natural indu-

cer (Inazawa et al., 1998). We con®rmed that in Ciona,

bFGF also has inducing activity and used this as a tool to

further investigate the induction of neural markers. The

early induction of Ci-otx in a4.2 and b4.2 explants by the

44-cell stage suggests that bFGF is acting directly to induce

Ci-otx. The lack of mesoderm (notochord and muscle) and

endoderm in these explants (data not shown) lends support

to this, although we cannot exclude the possibility that a

tissue for which we have no marker as yet is induced

which secondarily induces or maintains neural markers.

a4.2 blastomeres respond to bFGF by expressing Ci-otx

and developing pigment cells at levels comparable to induc-

tion by A4.1. However, high concentrations of bFGF were

required to obtain the same level of Ci-otx induction as that

seen in response to the natural inducer. It is very dif®cult to

imagine that this concentration of bFGF could be achieved

in an embryo, although it is possible that high levels are only

required because recombinant human bFGF used in this

study may be less potent than the endogenous FGF. Testing

this hypothesis will require the cloning and analysis of FGF

molecules in ascidians. Additionally, although a4.2 blasto-

meres responded similarly to bFGF and the endogenous

inducer, in terms of expression of Ci-otx and pigment

cells formation, the much reduced number of neurons

produced and the lack of induction of Ci-gsx in response

to bFGF indicates that even high concentrations of bFGF

cannot fully replace A4.1 as an anterior neural inducer.

Moreover, a recent analysis of the ability of A4.1 and

B4.1 line blastomeres to induce notochord or mesenchyme,

two FGF inducible cell fates, has led to the proposition that

an FGF-like signal is emitted by both A4.1 and B4.1 deri-

vatives (Kim et al., 2000). Our ®ndings that anterior and

posterior vegetal blastomeres have very dissimilar neural

inducing ability lends support to the proposition that FGF

signalling is not the sole requirement for induction of ante-

rior neural fates.

4. Experimental procedures

4.1. Embryos

Adults were obtained from the Station de Biologie

Marine in Roscoff and from the Saumaty industrial harbour

in Marseille. Eggs and sperm were collected by dissection

and cross fertilizations were made in petri dishes. Dechor-

ionated embryos (according to Mita-Miyazawa et al., 1985)

were raised in sea water at 13±188C on 1% agarose-coated

C. Hudson, P. Lemaire / Mechanisms of Development 100 (2001) 189±203 201

dishes in arti®cial sea water (420 mM NaCl, 90 mM KCl, 10

mM CaCl2, 24.5 mM MgCl2, 25.5 mM MgSO4, 2.15 mM

NaHCO3, 10 mM Hepes (pH 8.0)) supplemented with 50

mg/ml gentamycin. Explants were dissected with a ®ne glass

needle. Recombinations were carried out by placing blasto-

mere pairs in contact, in 30% PEG(8000) and then washing

in sea water.

bFGF (Sigma, F0291) was diluted in 0.1% BSA (bovine

serum albumin, fraction V, Sigma, A8022) in sea water.

Explants were placed in bFGF solution until the required

stage and ®xed.

4.2. Cloning of Ci-gsx and Ci-otx

Ci-gsx and Ci-otx were ampli®ed from mid-gastrula

embryo cDNA using the degenerate antennapedia class

primers previously described (Moretti et al., 1994) with a

508C annealing temperature and 2 mM MgCl2. Full-length

cDNAs were obtained by screening a tailbud cDNA library

using standard procedures. Sequencing was done by

Genome Express, Genoble. The Genbank database acces-

sion numbers are AF305499 (Ci-otx) and AF305500 (Ci-

gsx).

4.3. In situ hybridization

In situs were carried out as described (Wada et al., 1995).

For Fig. 3 embryos were collected every 15 min from early

gastrula (4.5 h at 188C) until the tailbud began to elongate (8

h). Nuclei were observed using Syto11 or 13 Live Cell

Nucleic Acid Stain (Molecular probes). For photography,

all embryos were mounted in 50±80% glycerol in PBS-

Tween. Often this led the embryos to appear rather bigger

than in unmounted specimens. Embryos co-stained with a

nuclear dye and those in Figs. 3i,k, 4e,g and 9b±e,g were

photographed prior to mounting.

4.4. Staining with anti-acetylated tubulin

Anti-acetylated tubulin (Sigma T6793) antibody staining

was carried out as follows. Brie¯y, embryos or explants

were ®xed in 2208C methanol for 10 min and washed

three times for 10 min in PBS-Triton X (0.05%). Primary

antibody was used at 1:1000 in PBS-Tween (0.1%) and

incubated overnight at 48C. Samples were washed for at

least 10 min three times in PBS-Tween at room temperature,

and then secondary antibody (AlexaFluore 488 goat anti-

mouse IgG, Molecular probes, used at 1:100 in PBS-Tween)

was incubated overnight at 48C. Finally, samples were

washed at least three times for 10 min in PBS-Tween and

mounted in 80% glycerol.

This antibody also stained another structure in explants

on the surface of both a4.2 and b4.2 explants treated with

bFGF or combined with A4.1 (data not shown), which was

not seen in similarly high numbers in whole embryos. These

structures were shorter than the neurons and cilia and may

represent a different kind or partially formed cilia or sensory

neurons. Since we could not de®ne these structures as either

neuron-like or cilia-like, neurons were scored as positive

only if they were present in a group of neurons. Cilia

were scored only if they were present as a pair or if they

could be seen projecting into the tunic.

Note added in proof

Ciona-otx has been independently cloned and reported:

Nanami Utsumi and Hidetoshi Saiga (2001). Comparison of

the Structure and Expression of otx Genes Between Ciona

intestinalis and Halocynthia roretzi. In: Biology of Asci-

dians, subtitle: Proceedings of the First International

Symposium on the Biology of Ascidians, Sappora, Japan

June 26±30, 2000, Springer±Verlag, Tokyo.

Acknowledgements

A big thank you to H. Yasuo for all his help when we

were starting to establish ascidian work in the lab. Many

thanks to A. Spagnuoli for the generous gift of the cDNA

library. We would also like to thank O. Cabaud, J.-P.

Bonamy, S. Marcellini and A. Ribas for their input into

the Ciona culture system. Many thanks to N. Utsumi, H.

Saiga, S. Darras and H. Nishida for discussions. We also

thank K. Dale, L. Kodjabachian, M. Maroto, M. McGrew

and H. Yasuo for their helpful comments on the manuscript.

This work was supported by grants to P.L. from the Centre

National de la Recherche Scienti®que (CNRS), the Associa-

tion pour la Recherche contre le Cancer, the Association

FrancËaise contre les Myopathies, the Human Frontier

Science Program Organisation, and the European Union

(TMR programme). C.H. was a EU-TMR fellow and then

the recipient of a CNRS Poste Rouge.

References

Arendt, D., Nubler-Jung, K., 1999. Comparison of early nerve cord devel-

opment in insects and vertebrates. Development 126, 2309±2325.

Baker, J.C., Beddington, R.S.P., Harland, R.M., 1999. Wnt signalling in

Xenopus embryos inhibits Bmp4 expression and activates neural devel-

opment. Genes Dev. 13, 3149±3159.

Brooke, N.M., Garcia-Fernandez, J., Holland, P.W., 1998. The ParaHox

gene cluster is an evolutionary sister of the Hox gene cluster. Nature

392, 920±922.

Conklin, E.G., 1905. The organisation and cell lineage in the ascidian egg.

J. Acad. Natl. Sci. (Philadelphia) 13, 1±119.

Crowther, R.J., Whittaker, J.R., 1992. Structure of the caudal neural tube in

an ascidian larva: vestiges of its possible evolutionary origin from a

ciliated band. J. Neurobiol. 23, 280±292.

Crowther, R.J., Whittaker, J.R., 1994. Serial repetition of cilia pairs along

the tail surface of an ascidian larva. J. Exp. Zool. 268, 9±16.

Deschet, K., Bourrat, F., Chourrout, D., Joly, J.S., 1998. Expression

domains of the medaka (Oryzias latipes) Ol-Gsh 1 gene are reminiscent

of those of clustered and orphan homeobox genes. Dev. Genes Evol.

208, 235±244.

Di Gregorio, A., Spagnuoli, A., Ristoratore, F., Pischetola, M., Aniello, F.,

Branno, M., Cariello, L., Di Lauro, R., 1995. Cloning of ascidian

C. Hudson, P. Lemaire / Mechanisms of Development 100 (2001) 189±203202

homeobox genes provides evidence for a primordial chordate cluster.

Gene 156, 253±257.

Finkelstein, R., Boncinelli, E., 1994. From ¯y head to mammalian fore-

brain: the story of otd and Otx. Trends Genet. 10, 310±315.

Harland, R., 2000. Neural induction. Curr. Opin. Genes Dev. 10, 357±362.

Hinman, V.F., Degnan, B.M., 2000. Retinoic acid perturbs Otx gene expres-

sion in the ascidian pharynx. Dev. Genes Evol. 210, 129±139.

Hsieh-Li, H.M., Witte, D.P., Szucsik, J.C., Weinstein, M., Li, H., Potter,

S.S., 1995. Gsh-2, a murine homeobox gene expressed in the develop-

ing brain. Mech. Dev. 50, 177±186.

Inazawa, T., Okamura, Y., Takahashi, K., 1998. Basic ®broblast growth

factor induction of neuronal ion channel expression in ascidian ecto-

dermal blastomeres. J. Physiol. 511.2, 347±359.

Kim, G.J., Yamada, A., Nishida, H., 2000. An FGF signal from endoderm

and localized factors in the posterior-vegetal egg cytoplasm pattern the

mesodermal tissues in the ascidian embryo. Development 127, 2853±

2862.

Mita-Miyazawa, I., Ikegami, S., Satoh, N., 1985. Histospeci®c acetylcho-

linesterase development in the presumptive muscle cells isolated from

16-cell-stage ascidian embryos with respect to the number of DNA

replications. J. Embryol. Exp. Morphol. 87, 1±12.

Moretti, P., Simmons, P., Thomas, P., Haylock, D., Rathjen, P., Vadas, M.,

D'Andrea, R., 1994. Identi®cation of homeobox genes expressed in

human haemopoietic progenitor cells. Gene 144, 213±219.

Nakatani, Y., Moody, R., Smith, W.C., 1999. Mutations affecting tail and

notochord development in the ascidian Ciona savignyi. Development

126, 3293±3301.

Nicol, D., Meinertzhagen, I.A., 1988a. Development of the central nervous

system of the larva of the ascidian, Ciona intestinalis L. I. The early

lineages of the neural plate. Dev. Biol. 130, 721±736.

Nicol, D., Meinertzhagen, I.A., 1988b. Development of the central nervous

system of the larva of the ascidian, Ciona intestinalis L. II. Neural plate

morphogenesis and cell lineages during neurulation. Dev. Biol. 130,

737±766.

Nicol, D., Meinertzhagen, I.A., 1991. Cell counts and maps in the larval

central nervous system of the ascidian Ciona intestinalis (L.). Comp.

Neurol. 309, 415±429.

Nishida, H., 1987. Cell lineage analysis in ascidian embryos by intracellular

injection of a tracer enzyme. III. Up to the tissue restricted stage. Dev.

Biol. 121, 526±541.

Nishida, H., 1992. Regionality of egg cytoplasm that promotes muscle

differentiation in embryo of the ascidian, Halocynthia roretzi. Devel-

opment 116, 521±529.

Nishida, H., 1993. Localized regions of egg cytoplasm that promote expres-

sion of endoderm-speci®c alkaline phosphatase in embryos of the asci-

dian Halocynthia roretzi. Development 118, 1±7.

Nishida, H., 1994a. Localization of egg cytoplasm that promotes differen-

tiation to epidermis in embryos of the ascidian Halocynthia roretzi.

Development 120, 235±243.

Nishida, H., 1994b. Localization of determinants for formation of the ante-

rior-posterior axis in eggs of the ascidian Halocynthia roretzi. Devel-

opment 120, 3093±3104.

Nishida, H., 1997. Cell fate speci®cation by localized cytoplasmic determi-

nants and cell interactions in ascidian embryos. Int. Rev. Cytol. 176,

245±306.

Okado, H., Takahashi, K., 1990a. Induced neural-type differentiation in the

cleavage-arrested blastomere isolated from early ascidian embryos. J.

Physiol. 427, 603±623.

Okado, H., Takahashi, K., 1990b. Differentiation of membrane excitability

in isolated cleavage-arrested blastomeres from early ascidian embryos.

J. Physiol. 427, 583±602.

Okado, H., Takahashi, K., 1993. Neural differentiation in cleavage arrested

ascidian blastomeres induced by proteolytic enzyme. J. Physiol. 463,

269±290.

Ortolani, G., Patricolo, E., Mansueto, C., 1979. Trypsin-induced cell

surface changes in ascidian embryonic cells. Exp. Cell Res. 122,

137±147.

Otte, A.P., Moon, R.T., 1992. Protein kinase C isozymes have distinct roles

in neural induction and competence in Xenopus. Cell 68, 1021±1029.

Reverberi, G., Minganti, A., 1946. Fenomeni di evocazione nello sviluppo

dell'uovo di Ascidie. Riulti dell'indagine sperimentale sull'uovo di

Ascidiella aspersa e do Ascidia malaca allo stadio di otto blastomeri.

Pubbl. Stn. Zool. Napoli 20, 199±252.

Reverberi, G., Minganti, A., 1947. La distribuzione delle potenze nel germe

di Ascidie allo stadio di otto blastomeri, analizzata mediante le combi-

nazioni e i trapianti di blastomeri. Pubbl. Stn. Zool. Napoli 21, 1±35.

Rhinn, M., Dierich, A., Shawlot, W., Behringer, R.R., Le Meur, M., Ang,

S.-L., 1998. Sequential roles for Otx2 in visceral endoderm and

neuroectoderm for forebrain and midbrain induction and speci®cation.

Development 125, 845±856.

Rose, S.M., 1939. Embryonic induction in the ascidia. Biol. Bull. 76, 216±

232.

Satoh, N., 1994. Developmental Biology of Ascidians, Cambridge Univer-

sity Press, Cambridge.

Sharpe, C.R., Fritz, A., De Robertis, E.M., Gurdon, J.B., 1987. A homeo-

box-containing marker of posterior neural differentiation shows the

importance of predetermination in neural induction. Cell 50, 749±758.

Simeone, A., 1998. Otx1 and Otx2 in the development and evolution of the

mammalian brain. EMBO J. 17, 6790±6798.

Singh, G., Kaur, S., Stock, J.L., Jenkins, N.A., Gilbert, D.J., Copeland,

N.G., Potter, S.S., 1991. Identi®cation of 10 murine homeobox genes.

Proc. Natl. Acad. Sci. USA 88, 10706±10710.

Sokol, S., Melton, D.A., 1991. Pre-existent pattern in Xenopus animal pole

cells revealed by induction with activin. Nature 351, 409±411.

Streit, A., Berliner, A.J., Papanayotou, C., Sirulnik, A., Stern, C.D., 2000.

Initiation of neural induction by FGF signalling before gastrulation.

Nature 406, 74±78.

Szucsik, J.C., Witte, D.P., Li, H., Pixley, S.K., Small, K.M., Potter, S.S.,

1997. Altered forebrain and hindbrain development in mice mutant for

the Gsh-2 homeobox gene. Dev. Biol. 191, 230±242.

Takahashi, H., Mitani, Y., Satoh, G., Satoh, N., 1999. Evolutionary altera-

tions of the minimal promoter for notochord-speci®c Brachyury expres-

sion in ascidian embryos. Development 126, 3725±3734.

Valerius, M.T., Li, H., Stock, J.L., Weinstein, M., Kaur, S., Singh, G.,

Potter, S.S., 1995. Gsh-1: a novel murine homeobox gene expressed

in the central nervous system. Dev. Dyn. 203, 337±351.

Wada, H., Saiga, H., Satoh, N., Holland, P.W., 1998. Tripartite organisation

of the ancestral chordate brain and the antiquity of placodes: insights

from ascidian Pax-2/5/8, Hox and Otx genes. Development 125, 1113±

1122.

Wada, S., Katsuyama, Y., Yasugi, S., Saiga, H., 1995. Spatially and tempo-

rally regulated expression of the LIM class homeobox gene Hrlim

suggests multiple distinct functions in development of the ascidian,

Halocynthia roretzi. Mech. Dev. 51, 115±126.

Wada, S., Katsuyama, Y., Sato, Y., Itoh, C., Saiga, H., 1996. Hroth, an

orthodenticle-related homeobox gene of the ascidian, Halocynthia

roretzi: its expression and putative roles in the axis formation during

embryogenesis. Mech. Dev. 60, 59±71.

Wada, S., Katsuyama, Y., Saiga, H., 1999. Anteroposterior patterning of the

epidermis by inductive in¯uences from the vegetal hemisphere cells in

the ascidian embryo. Development 126, 4955±4963.

Weiss, J.B., Von Ohlen, T., Mellerick, D.M., Dressler, G., Doe, C.Q., Scott,

M.P., 1998. Dorsoventral patterning in the Drosophila central nervous

system: the intermediate neuroblasts defective homeobox gene speci-

®es intermediate column identity. Genes Dev. 12, 3591±3602.

Wilson, S.I., Graziano, E., Harland, R., Jessell, T.M., Edlund, T., 2000. An

early requirement for FGF signalling in the acquisition of neural cell

fate in the chick embryo. Curr. Biol. 10, 421±429.

C. Hudson, P. Lemaire / Mechanisms of Development 100 (2001) 189±203 203