Dev Genes Evol (1997) 206:389396

Springer-Verlag 1997

O R I G I NA L A RT I C L E

&roles:Toshiro Mizuno Etsuro Yamaha Fumio Yamazaki

Localized axis determinant in the early cleavage embryo

of the goldfish, Carassius auratus

&misc:Received: 25 May 1996 / Accepted: 19 September 1996

&p.1:Abstract The teleost dorsoventral axis cannot be distinguished morphologically before gastrulation. In order to
examine whether the yolk cell affects axis determination,
we bisect early cleavage embryos of the goldfish, Carassius auratus. When the vegetal yolk hemisphere is removed by bisection along the equatorial plane at the 2cell stage, the embryos develop abnormally and exhibit a
symmetrical morphology. No dorsal structures, such as
notochord, somites and neural tube, differentiate and no
embryonic shield is formed during gastrulation. In addition, no goosecoid mRNA is expressed before gastrulation. The frequency of abnormality decreases as the age
at which the vegetal yolk hemisphere is removed increases. Most embryos removed at the 32-cell stage develop
normally. Their morphological phenotype is similar to
that of a Xenopus ventralized embryo generated by ultraviolet irradiation on the vegetal hemisphere soon after
fertilization. We also observed that, when the embryos
were bisected along the first cleavage plane at the 2-cell
stage, the proportion of pairs of embryos of which one
embryo developed normally was 44.8%. These results
indicate that the vegetal yolk hemisphere of the early
cleavage embryo of the goldfish contains axis determination factor(s), which are necessary for generation of dorsal structures. Furthermore, it is suggested that these determinant(s) are distributed asymmetrically within the
vegetal yolk hemisphere.
Edited by D. Tautz
T. Mizuno ()
Department of Molecular Biology, School of Science,
Nagoya University, Furo-cho, Chikusa-ku, Nagoya,
464-01 Japan
E. Yamaha
Nanae Fish Culture Experimental Station, Faculty of Fisheries,
Hokkaido University, 498 Sakura, Nanae, Kameda-gun, Hokkaido,
041-11 Japan
F. Yamazaki
Laboratory of Embryology and Genetics, Faculty of Fisheries,
Hokkaido University, Minato-cho, Hakodate, Hokkaido,
041 Japan&/fn-block:

&kwd:Key words Dorsoventral axis Goldfish goosecoid

no tail Zebrafish&bdy:

Introduction
Is the mechanism underlying axis determination conserved among vertebrate species, the eggs of which are
morphologically diverse? During embryonic development, the formation of the dorsal lip of the blastopore
(Hensens node) during gastrulation is the first sign of
formation of the dorsoventral (D-V) axis. It is thought
that the dorsal lip region, the organizer, is the centre of
embryonic induction involved in morphogenesis. The
functions and molecular properties of the organizer are
well conserved among vertebrates (reviewed by De
Robertis et al. 1992). However, it is still uncertain what
mechanisms regulate the development of the organizer.
In amphibians, the cytoplasmic factor(s) which induce
formation of the dorsal structures are present in the eggs
just after fertilization, and this cytoplasm around the vegetal pole is necessary for determination of the D-V axis.
Ultraviolet (UV) irradiation of eggs just after fertilization results in development of radially symmetric, axisdefected embryos without dorsal structures (Malacinski
et al. 1974; Scharf and Gerhart 1980, 1983), and these
embryos have no dorsal lip on the blastopore. It has been
shown that UV irradiation blocks cortical rotation after
fertilization in the African clawed frog Xenopus laevis
(reviewed by Gerhart et al. 1989). This rotation is a cue
for the translocation of cytoplasmic factor(s) [determinant(s)] from the vegetal pole region to the prospective
dorsal side (Fujisue et al. 1993).
In teleosts, early development appears more regulative than that of amphibians (reviewed by Driever 1995).
Several studies on zebrafish have revealed intermixing of
deep layer (DEL) cells at the blastula stage (Kimmel and
Law 1985; Warga and Kimmel 1990). Furthermore, cell
transplantation studies have shown that the blastomeres
remain pluripotent and uncommitted throughout the late
blastula and early gastrula stages, since the fate of a sin-

390

Fig. 1AI Bisection of embryos using a nylon fibre. AD Removal of vegetal yolk hemisphere. This process was completed within
25 min. E Yolk fragments after removal of the vegetal yolk hemisphere. A blastodisc-like structure is observed. Photographed
69 h after removal of vegetal yolk hemisphere. FI Bisection
along the first cleavage plane. Completed within 35 min. (Bar
1 mm)&ig.c:/f

gle cell is changed by the transplantations of cells from

other fate-mapped regions (Ho and Kimmel 1993).
On the other hand, results of other studies suggest that
the yolk cell contains the mosaically distributed cytoplasmic factor(s) at the early cleavage stage, that are important in later morphogenesis. Tung and co-workers
have shown that fragmentation of the goldfish yolk
hemisphere at the early cleavage stage results in various
abnormalities in the embryo (Tung and Tung 1944; Tung
et al. 1945; reviewed by Devillers 1961). Recently, Yamaha and Yamazaki (1993) demonstrated that the fusion of
two goldfish eggs just after fertilization gives rise to
complete double embryos with high frequency. This suggests that cytoplasmic determinant(s) are firmly localized in the egg, which are not re-organized after the fusion.
These observations have led us to examine the localization of cytoplasmic factor(s) in teleosts, in order to investigate the role of cytoplasmic factor(s) in axis formation at the early cleavage stage. We fragmented early
cleavage embryos of the goldfish and analysed the development of these embryos. Here, we show that the vegetal
yolk hemisphere of the early cleavage embryo of the
goldfish plays an important role in the D-V axis formation.

Materials and methods

Eggs and embryo cultivation
Goldfish, Carassius auratus, were reared in the Faculty of Fisheries, Hokkaido University. Ovulation was induced, and artificial insemination and fertilization were performed as previously de-

scribed (Yamaha and Yamazaki 1993). Dechorionation was carried

out before blastodisc formation using a method described by Yamaha et al. (1986) with slight modification. Activated eggs were
dechorionated with 0.1% trypsin (DIFCO) and 0.4% urea in Ringers solution (128 mM NaCl, 2.8 mM KCl, 1.8 mM CaCl2) for
about 10 min, and then washed with culture Ringers solution containing 1.6% albumen, 0.01% penicillin, 0.01% kanamycin and
0.01% streptomycin. Fragmented (see below) and control embryos
were cultured in separated wells filled with culture Ringers solution for 1 day, and then transferred to wells filled with 1.8 mM
CaCl2, 1.8 mM MgCl2 and the same antibiotics as in the culture
Ringers solution, at the same concentrations, and incubated at
20C for about 3 days until the intact sibling embryos hatched.
Fragmentation of embryos
All procedures were performed in a glass dish coated with 1%
agar and filled with culture Ringers solution. Embryos were fragmented in the following two ways:
1. The vegetal yolk hemisphere was removed by horizontal bisection along the equatorial plane at the 2-cell to 32-cell stages
(Fig. 1AD). Nylon fibres (0.08 mm diameter) were used to bisect
the embryos. Bisection was completed within about 30 min, aided
by pressure with tips of cover-glass fragments on both ends of the
fibre. Formation of a blastodisc-like structure (Fig. 1E) in the removed yolk fragments and rhythmic contractile movements of the
yolk fragments were observed. Cell division was not observed.
2. Two-cell embryos were bisected laterally along the first cleavage plane in a manner similar to that for the horizontal bisection
(Fig. 1F,G). Two or three fibres were used in order to prevent adhesion of bisected blastomeres (Fig. 1H,I). The resulting two halfembryos were incubated separately.
We did not analyse the embryos which showed yolk leakage in all
operations and pairs of embryos of which one embryo died in the
case of bisection along the first cleavage plane.
Whole mount in situ hybridization
Digoxigenin (DIG)-11-UTP-labelled single-stranded RNA probes
were generated by in vitro transcription from a linearized plasmid
clone, containing the cDNA fragment of no tail (ntl) or goosecoid
(gsc), according to the manufacturers instructions for the DIG
RNA labelling kit (Boehringer Mannheim, Mannheim).
The HincII fragment of zebrafish ntl cDNA (Schulte-Merker et
al. 1992), which excludes the 3 end of the noncoding region, was
subcloned into pBluescript II SK+ and used as a template for the

391
synthesis of a DIG-labelled RNA probe. The probe for gsc (Stachel et al. 1993) was obtained by reverse transcriptase-polymerase
chain reaction (RT-PCR), followed by screening of a zebrafish
ovary cDNA library. The fragment from nucleotide 132 to nucleotide 570, which excludes the homeodomain, was subcloned into
pBluescript KS- and used as a template for the probe (courtesy of
Dr. H. Takeda; ntl was originally cloned by Dr. S. Schulte-Merker
and gsc by Drs. M. Tada and N. Ueno).
The embryos were fixed with 4% paraformaldehyde in phosphate-buffered saline (PBS) overnight at 4C, washed three times
with PBS, transferred to 100% methanol, and stored at 20C until
further treatment. All further steps were performed at room temperature unless otherwise stated. The embryos were washed two
times with 0.1% Tween-20 in PBS, soaked in hybridization buffer
[50% formamide, 5 sodium sodium citrate (SSC), 100 g/ml
sonicated calf thymus DNA, 50 g/ml heparin, 0.1% Tween-20,
10 g/ml yeast tRNA], and then prehybridized for at least 1 h, and
then the hybridization buffer was removed and an equal volume of
hybridization buffer containing an equal amount of DIG-labelled
RNA probe (25 ng/ml) was added.
After an overnight incubation at 52C the probe was removed
and the embryos were rinsed in 2 SSC, and then washed with
50% formamide in 2 SSC at 65C for 30 min, and treated with
RNase A (20 g/ml) at 37C for 1 h. Thereafter, the embryos were
washed sequentially in 2 SSC for 10 min, 50% formamide in
2 SSC at 65C for 30 min, and 0.2 SSC at 55C for 15 min.
The embryos were then washed with maleic acid buffer
(100 mM maleic acid, 150 mM NaCl, pH 7.5) for 10 min and were
blocked for at least 1 h in blocking buffer (2% fetal calf serum,
0.2% Tween-20, 0.2% Triton X-100). The blocking buffer was
then removed and 0.0125% alkaline-phosphatase conjugated antidigoxigenin Fab-fragments (Boehringer Mannheim) in blocking
buffer were added. After an overnight incubation at 4C, the embryos were washed twice for 15 min and three times for 30 min
with maleic acid buffer, and then three times for 10 min with
staining buffer (100 mM TRIS, 100 mM NaCl, 50 mM MgCl2,
0.1% Tween-20, 1 mM levamisole, pH 8.0). Detection was performed in BM purple AP-substrate (Boehringer Mannheim).
When the signal intensity reached a level sufficient for observation, the reaction was stopped and the embryos were washed three
times for 5 min with staining buffer and then fixed with 4% paraformaldehyde in PBS.
Histology
Embryos were fixed with Bouins fixative for 2 h, dehydrated in a
butyl alcohol series and embedded in a paraffin block. Serial sections were cut at 8 m and haematoxilin-eosin staining was performed using a standard procedure.

Results
Morphological series of abnormal embryos
Both the removal of vegetal yolk hemispheres and the
lateral bisection along the first cleavage plane resulted in
a characteristic series of abnormal morphologies, although some fragments developed normally. The abnormalities were categorized into four main types: rotational
symmetry, acephaly, cyclopia and joined eyes. From the
results of careful examination we concluded that these
four types are not independent of each other but represent degrees of antero-dorsal abnormality. The features
are described in detail below.

1. Rotational symmetry (no axis)

No axial structures formed externally (Fig. 2AD). The
large body-like structure at the vegetal side had a red
blood-island near the yolk cell, a somite-like segmented
structure in the middle and several fin-like structures distally. A long projection with a cavity elongated from the
animal pole. Little pigmentation was visible.
Histologically the irregular gut-like structure was situated between the yolk and blood-island. The spherical
erythroblast-like cells each exhibited incomplete differentiation and had a large nucleus. The segmented structure had a large number of cells with condensed nuclei
and showed no features of muscle and no fibrous tissue
with eosin-positive staining. The periderm had enveloped
the whole body and the mesenchyme lay between the
periderm and inner structures. The yolk syncytial layer
(YSL) had completely enveloped the yolk mass as in
normal embryos.
This type of embryo exhibited only an animal-vegetal
axis since latitudinal sections mostly show rotationally
symmetrical images (Fig. 2BD).
2. Truncated head
No retinal pigments were externally visible (Fig. 2EG).
Otic vesicles were visible. Dorsal body flexures were observed. The pericardial cavity was expanded and contained a frail-looking heart. The fin fold was split at the
end of the tail.
Histologically no notochord was present. The somites
had fused across the midline and were less developed in
the anterior part of the trunk than in posterior one. The
spinal cord frequently had no lumen, and its white matter
was restricted to the ventral side. There was a body cavity with blood cells and mesenchyme with condensed nuclei at the ventral part of the trunk behind the anus.
3. Cyclops
The cyclopic eye was externally visible on the ventral
side of the head (Fig. 2HJ). Dorsal body flexure was
observed.
Histologically the forebrain was incompletely formed
and the telencephalon was lost. The optic cup covered
one lens or rarely two lenses. No notochord was observed. The somites and the spinal cord showed the same
features as those in the acephalic embryo (see above, 2.
Truncated head).
4. Joined eyes
The eyes were joined on the ventral side and two lenses
were externally visible (Fig. 2KM). Internally the optic
cups had fused across the midline and included a confused optic chiasma. The nervous tissues showed almost

392

Fig. 2AP Series of axis-deficient embryos after removal of vegetal yolk hemisphere and bisection along first cleavage plane. A
Median section of non-axial (dorso-anterior index 1) embryo. Left
column (B, E, H, K, N): external views of embryos. Middle column (C, F, I, L, O): transverse sections of head regions (except C:
vegetal proximal region). Right column (D, G, J, M, P): transverse
sections of trunk regions. BD Rotational symmetry (no axis, DAI
1). EG Truncated head (DAI 2). HJ Cyclops (DAI 3). KM

Joined eyes (DAI 4). NP Normal (DAI 5). Embryos were obtained from removal of vegetal yolk hemisphere at 2-cell (C, D),
4-cell (A), 8-cell (B, O, P) and 16-cell stage (EJ), or from bisection along first cleavage plane (KN). Observed and fixed at 3
days (AJ, OP) or 4 days (KN); g gut-like structure, b bloodlike cells, s segmented structure, n notochord; bar 1 mm B, E,
H, K, N, 0.1 mm others&ig.c:/f

393

normal shapes except that the forebrain was incomplete.

The notochord was normally formed. The somites contained some deformed muscle fibres.
5. Normal
All of the externally or internally visible structures of the
embryos were normal, except that dorsal flexures were
only rarely present (Fig. 2NP). These embryos from the
removal of the vegetal yolk hemisphere were able to
swim and feed, grew when incubated for a long period,
and matured the following year. By contrast, most of the
embryos from lateral bisections along the first cleavage
plane, even if they looked normal, did not float and feed,
and soon died.
The dorso-anterior index (DAI) is used to estimate the
extents of dorsoanterior development and body axial
structures in X. laevis embryos (Scharf and Gerhart
1983; modified by Kao and Elinson 1988). Using a modified DAI grading system in Xenopus, we categorized the
degrees of abnormality of the goldfish which developed
from fragmented embryos as follows:
DAI 1 Rotational symmetry (no axis)
DAI 2 Truncated head
DAI 3 Cyclops
DAI 4 Joined eyes
DAI 5 Normal.
This goldfish DAI grading system does not exactly correspond to the one for Xenopus. In the case of Xenopus,
the DAI grading system is comprised of six grades (0: no
axis5: normal; Kao and Elinson 1988), with acephalic
embryos sorted into two classes (DAI 1 and 2). In the
present study, the acephalic embryos were assigned to
one class because the external critical features of head
structure were unclear and the frequency of appearance
was low (Fig. 3).
The abnormalities of the DAI 1 embryos and some
abnormalities of the DAI 25 embryos were consistent
morphologically and histologically with those described
by Tung et al. (1945, 1955). Other types of abnormalities, which could not be classified using the DAI grading
system, were sometimes observed (others in Fig. 3).
Many of the abnormal embryos consisted of yolk and
blastoderm-like structures with irregular spinal cord-like
tissues, notochords and an undifferentiated cell mass.
Frequencies of axis-defective embryos by fragmentation
1. Removal of vegetal yolk hemisphere
Figure 3 shows the frequencies of the DAI 15 embryos
induced by the removal of the vegetal yolk hemisphere at
the 2- to 32-cell stage. Approximately 70% of the embryos developed into DAI 1 (no-axis) embryos due to
vegetal yolk hemisphere removal at the 2-cell stage. The
frequency of DAI 5 (normal) embryos increased as the

Fig. 3 Frequencies of the DAI 15 embryos developed from embryos from which the vegetal yolk hemisphere had been removed
at the 2- to 32-cell stages. Results are totalized from 5 independent
experiments. Dechorionated embryos were used as controls&ig.c:/f

age at which the vegetal yolk hemisphere was removed

increased, and approximately 90% of the embryos developed normally when the vegetal yolk hemisphere was removed at the 32-cell stage. The frequency of each of the
DAI 24 embryos never exceeded 30%.
2. Lateral bisection along the first cleavage plane
The frequencies of DAI values for the pairs of embryos
by the lateral bisection along the first cleavage plane are
shown in Table 1. There was a tendency that only one of
the pair developed into a DAI 1 embryo, whilst the other
developed into a DAI 2, 3, 4 or 5 embryo. Ninety-five
percent of the dechorionated control embryos developed
normally (DAI 5), but only 44.8% of the pairs developed
normally in one or both sibling blastomeres. Therefore,
it seems that the potential for normal development was
greatly reduced by the lateral bisection.
Altered gastrulation of non-axial embryos
More than 70% of the embryos developed into non-axial,
DAI 1 embryos, when the vegetal yolk hemisphere was
removed before the 2-cell stage. We examined histologically the gastrulation of DAI 1 embryos.
At the early gastrula stage, as in normal embryos, a
germ ring formed in the DAI 1 embryos. The margin of
DEL cells of the germ ring formed a thick epiblast and
loose hypoblast as a result of the involution. However, no
embryonic shield formed and no extended invagination
was observed (Fig. 4A, B). Histologically, the hypoblast
cells did not reach the animal pole region. Thus, it was
impossible to tell which side was dorsal. Epiboly was

394
Table 1 Frequencies of dorso-antorior index (DAI) values for pairs that developed from half-embryos obtained by bisection along the
first cleavage plane at the 2-cell stage. The results were obtained from two independent experiments&/tbl.c:&
One part of pairs
Other part of pairs
1
2
DAI

3
4
5
Control

DAI

Total
number

No.
(%)
No.
(%)
No.
(%)
No.
(%)
No.
(%)

30
(28.6%)
7
(6.7%)
4
(3.8%)
5
(4.8%)
1
(1.0%)

15
(14.3%)
3
(2.9%)
2
(1.9%)
0
(0.0%)

18
(17.1%)
2
(1.9%)
2
(1.9%)

9
(8.6%)
1
(1.0%)

6
(5.7%)

No.
(%)

120
(95.2%)

0
(0.0%)

105
2
(1.6%)

3
(2.4%)

1
(0.8%)

126

&/tbl.:

Fig. 4AF Ventralization of prospective DAI 1 embryo. Ventralized embryos were obtained by removal of the vegetal yolk hemisphere. A, B Vertical sections of a ventralized (A) and control (B)
embryo at early gastrula stage (18 h after fertilization). In the control embryo, the dorsal hypoblast cells reached the animal pole region (arrowhead). C, D goosecoid (gsc) expression in ventralized

(C) and control (D) embryos at late blastula stage (11 h). No gsc
mRNA-positive cells were observed in the ventralized embryo
(C), whilst the no tail (ntl) expression pattern in the ventralized
embryo (E) was similar to that in the control embryo (F; Bar
0.1 mm)

395

completed more rapidly in the DAI 1 embryos than in

normal embryos. This reflects the fact that the yolk
sphere was smaller than normal, and the speed of epiboly
was probably unchanged.
During this study, we also found that the zebrafish gsc
and ntl probes hybridized to goldfish mRNA. In zebrafish, gsc is expressed in the presumptive dorsal region as
a molecular marker, whilst ntl is expressed in the presumptive meso-endodermal region, the germ ring. In order to detect gsc and ntl mRNA in goldfish embryos,
whole mount in situ hybridization was performed with
the antisense zebrafish gsc and ntl probes. Sense probe
hybridization was used as a control and no signal was
detected in normal embryos of goldfish (data not shown).
During normal development of goldfish, no ntl and gsc
expression was detected before 7 h postfertilization. After 8 h, localized expression of gsc and ntl was detected
at the blastoderm margin as a patch of staining. After
10 h, ntl expression spread to all margins of the blastoderm (data not shown). These normal patterns of expression were very similar to these in the zebrafish (Stachel
et al. 1993; Schulte-Merker et al. 1992). We then examined gsc and ntl expression patterns in prospective DAI 1
embryos of goldfish. Seven out of ten of the prospective
DAI 1 embryos showed no gsc expression (Fig. 4C, D) at
the late blastula stage (1112 h after fertilization). Three
gsc mRNA positive embryos included two specimens
representing double positive regions irregularly. Of the
prospective DAI 1 embryos, seven out of nine showed
normal ntl expression patterns (Fig. 4E, F) but the other
two embryos showed elevated ntl expression.

Discussion
Role of axis determinant(s) in teleost development
Our results demonstrate that the vegetal yolk hemisphere
plays an important role in dorsal structure formation at
the early cleavage stage. In the embryos from which the
vegetal yolk hemisphere was removed, embryonic shield
and continuous dorsal structure formation were arrested.
In embryos with the most severe phenotype, the DAI 1
embryos, the head, spinal cord, notochord and somites
did not form. Less severe phenotypes (DAI 24) also can
be considered to result from antero-dorsal defects. Thus,
we concluded that the removal of the vegetal yolk hemisphere inhibited dorsoanterior formation. Furthermore,
the lack of gsc expression in DAI 1 embryos suggested
that the embryonic shield, equivalent to the amphibian
organizer (Oppenheimer 1936, 1955), was missing in
these embryos. These results suggest that, as in the case
of amphibian embryos, cytoplasmic factor(s) exist in the
vegetal yolk hemisphere of early cleavage embryos.
It seems that these factor(s) are only responsible for
dorso-ventral (D-V) patterning, because gastrulation and
germ layer differentiation (formation of epiblast and hypoblast) proceeded normally in the DAI 1 embryos
(Fig. 4). The phenotypes of germ layer differentiation are

similar to those of Xenopus UV ventralized embryos

(Scharf and Gerhart 1983).
First cleavage plane and axial determinant(s)
In the embryos obtained by lateral bisection along the
first cleavage plane, the frequency at which one part of
the pair shows no axis (DAI 1) is 74.3% (Table 1, top
row). These results suggest that the axis determinant(s)
are distributed asymmetrically in the vegetal yolk hemisphere. Is the first cleavage furrow formation involved in
the distribution of axis determinant(s) in teleosts?
Several groups have investigated possible relationships between the first cleavage plane and the embryonic
axis. However, the data they have obtained are conflicting. Cell lineage analysis results indicated that the embryonic axis is not correlated with the first cleavage furrow in the zebrafish (Kimmel and Law 1985). Similarly,
in the case of the janus mutation in zebrafish, in which
two blastomeres at the 2-cell stage detach along the first
cleavage furrow and the resultant blastula have two sideby-side blastoderms, the distribution of axis determinant(s) is also not expected to be related to the first
cleavage furrow because of the gsc expression of the mutant (Abdelilah et al. 1994). On the other hand, Iwamatsu
(1993) in the case of medaka and Strehlow and Gilbert
(1993) in the case of zebrafish demonstrated the existence of a close relationship between the first cleavage
plane and the embryonic axis: the D-V axis is at right angles to the first cleavage furrow.
Our present findings supports the former hypothesis.
If the distribution of axis determinant(s) is related to the
first cleavage furrow as suggested in the latter hypothesis, at least one part of the pairs should have developed
normally in the present study. However, the frequency at
which one part of a pair developed normally was only
44.8% (Table 1). This is consistent with the data reported
by Tung and Tung (1944). They reported that, when
blastomeres were separated along the first or second
cleavage plane at the 2- , 4- or 8-cell stage, the development of these isolated fragments varied from the formation of two normal embryos of half size to the formation
of two vesicles showing no histological differentiation.
Taken together, the above findings suggest that first
cleavage furrow formation is not to be related to the distribution of axis determinant(s) in goldfish.
Cellular processes of axis determination
Our results indicate that the animal hemisphere of goldfish embryos receives information for D-V axis formation after the 16-cell stage: the embryos developed normally even after removal of the vegetal yolk hemisphere
at this stage (Fig. 3). The cytoplasmic factor(s) have
been demonstrated to be upstream of the cascade of D-V
axis formation (leading to the expression of gsc) in goldfish. It has been shown that, in zebrafish, the cytoplasmic

396

connection between the yolk cell and vegetal end of the

blastoderm remains at least until the tenth cleavage stage
(Ho 1992), and we obtained similar results in the case of
the goldfish (data not shown). After YSL formation, the
blastodermal cells are disconnected from the yolk cell. If
the vegetal blastomeres receive the cytoplasmic factor(s)
before the disconnection, then gsc can be expressed cellautonomously. If not, intercellular communication between the yolk cell and blastoderm is necessary for D-V
patterning. In this context, Stachel et al. (1993) have proposed a hypothetical axis determination centre responsible for gsc expression which is localized in the early
yolk cell just under the overlying blastoderm. This centre
could be equivalent to the Nieuwkoop centre in Xenopus
(Nieuwkoop 1973). By contrast, Lemaire and Gurdon
(1994) showed that gsc expression could be controlled
via an intracellular cascade after the 2-cell stage, even
without intercellular induction in Xenopus. The results of
our recent yolk cell transplantation experiments favour
the former possibility. The zebrafish yolk cell, when
transplanted to the animal-pole region, induces ectopic
expression of gsc (Mizuno et al. 1996).
In summary, the present findings suggest that the
mechanism underlying D-V axis formation are conserved between teleosts and amphibians. However, how
the mechanism is conserved in the morphological diversity of the eggs is still unknown. Analysis of axis-determining mechanisms in teleosts is expected to provide
useful information for this question.
&p.2:Acknowledgements We wish to thank Drs. T. S. Yamamoto, N.
Ueno, M. Tada and W. Kobayashi for valuable discussions at
manuscript preparation. We thank Dr. H. Takeda for critical reading of the manuscript. We are grateful to Dr. C. B. Kimmel for
technical suggestions. Thanks are also due to the members of Zoological Institute, Graduate School of Science and Laboratory of
Embryology and Genetics, Faculty of Fisheries, Hokkaido University for suggestive discussions. We are grateful to Nippon Suisan
Co. Ltd. for a grant for this study. This study was supported in part
by a grant-in-aid from the Ministry of Education, Science and Culture (05760142, 06760165).

References
Abdelilah S, Solnica-Krezel L, Stainer DYR, Driever W (1994)
Implications for dorsoventral axis determination from the zebrafish mutation janus. Nature 370: 468471
De Robertis EM, Blum M, Niehrs C, Steinbeisser H (1992)
goosecoid and the organizer. Development Suppl 1992: 167
171
Devillers C (1961) Structural and dynamic aspects of the development of the Teleostean egg. In: Abercrombie M, Brachet J
(eds) Advances in morphogenesis, vol 1. Academic Press,
New York London pp 379428
Driever W (1995) Axis formation in zebrafish. Curr Opin Genet
Dev 5: 610618
Fujisue M, Kobayakawa Y, Yamana K (1993) Occurence of dorsal
axis-inducing activity around the vegetal pole of an uncleaved
Xenopus egg and displacement to the equatorial region by cortical rotation. Development 118: 163170
Gerhart J, Danilchik M, Doniach T, Roberts S, Rowning B, Stewart R (1989) Cortical rotation of the Xenopus egg: consequenc-

es for the anteroposterior pattern of embryonic dorsal development. Development Suppl 1989: 3751
Ho RK (1992) Axis formation in the embryo of the zebrafish,
Brachydanio rerio. Semin Dev Biol 3: 5364
Ho RK, Kimmel CB (1993) Commitment of cell fate in the early
zebrafish embryo. Science 261: 109111
Iwamatsu T (1993) On the first cleavage plane and embryonic axis
in medaka eggs (in Japanese, with English abstract). Bull Aich
Univ Educ 42: 7986
Kao KR, Elinson RP (1988) The entire mesodermal mantle behaves as Spemanns organizer in dorsoanterior enhanced Xenopus laevis embryos. Dev Biol 127: 6477
Kimmel CB, Law RD (1985) Cell lineage of zebrafish blastomeres. III. Clonal analysis of the blastula and gastrula stages.
Dev Biol 108: 94101
Lemaire P, Gurdon JB (1994) A role for cytoplasmic determinants
in mesoderm patterning: cell-autonomous activation of the
goosecoid and Xwnt-8 genes along the dorsoventral axis of
early Xenopus embryos. Development 120: 11911199
Malacinski GM, Allis CD, Chung H-M (1974) Correction of developmental abnormalities resulting from localized ultra-violet
irradiation of an amphibian egg. J Exp Zool 189: 249254
Mizuno T, Yamaha E, Wakahara M, Kuroiwa A, Takeda H (1996)
Mesoderm induction in zebrafish embryos. Nature 383: 131
132
Nieuwkoop PD (1973) The organization center of the amphibian
embryo: its origin, spatial organization, and morphogenetic action. In: Abercrombie M, Brachet J, King TJ (eds) Advances
in morphogenesis, vol 10. Academic Press, New York London
pp 139
Oppenheimer JM (1936) Transplantation experiments on developing Teleosts (Fundulus and Perca). J Exp Zool 72: 409437
Oppenheimer JM (1955) The differentiation of derivatives of the
lower germ layers in Fundulus following the implantation of
shield grafts. J Exp Zool 128: 525559
Scharf SR, Gerhart JC (1980) Determination of the dorsal-ventral
axis in eggs of Xenopus laevis: complete rescue of uv-impaired eggs by oblique orientation before first cleavage. Dev
Biol 79: 181198
Scharf SR, Gerhart JC (1983) Axis determination in eggs of Xenopus laevis: a critical period before first cleavage, identified by
the common effects of cold, pressure and ultraviolet irradiation. Dev Biol 99: 7587
Schulte-Merker S, Ho RK, Herrmann BG, Nsslein-Volhard C
(1992) The protein product of the zebrafish homologue of the
mouse T genes is expressed in nuclei of the germ ring and the
notochord of the early embryo. Development 116: 1021
1032
Stachel SE, Grunward DJ, Myers PZ (1993) Lithium pertubation
and goosecoid expression identify a dorsal specification pathway in the pregastrula zebrafish. Development 117: 1261
1274
Strehlow D, Gilbert W (1993) A fate map for the first cleavages of
the zebrafish. Nature 361: 451453
Tung TC, Tung YFY (1944) The development of egg-fragments,
isolated blastomeres and fused eggs in the goldfish. Proc Zool
Sci 114: 4664
Tung TC, Chang CY, Tung YFY (1945) Experiments on the developmental potencies of blastoderms and fragments of Teleostean eggs separated latitudinally. Proc Zool Sci 115: 175188
Tung TC, Lee CY, Tung YFY (1955) Further studies on the developmenal potencies of Carassius eggs (in Chinese, with English abstract). Acta Biol Exp Sinica 4: 107128
Warga RM, Kimmel CB (1990) Cell movements during epiboly
and gastrulation in zebrafish. Development 108: 569580
Yamaha E, Usui K, Onozato H, Hamada K (1986) A method for
decholionation in goldfish, Carassius auratus. Nippon Suisan
Gakkaishi 52: 19291934
Yamaha E, Yamazaki F (1993) Electrically fused-egg induction
and its development in the goldfish, Carassius auratus. Int J
Dev Biol 37: 291298