Oncogene (1997) 14, 879 ± 889
 1997 Stockton Press All rights reserved 0950 ± 9232/97 $12.00
The zebra®sh homologue of the ret receptor and its pattern of expression
during embryogenesis
Camelia V Marcos-GutieÂrrez1, Stephen W Wilson2, Nigel Holder2 and Vassilis Pachnis1
1
Division of Developmental Neurobiology, National Institute for Medical Research, The Ridgeway, Mill Hill, London NW7 1AA;
Developmental Biology, Research Centre, Division of Biomedical Sciences, Randall Institute, King's College, 26-29 Drury Lane,
London WC2B 5RL, UK
2
The c-ret proto-oncogene, a member of the receptor
tyrosine kinase gene superfamily, plays a critical role in
the development of the excretory system and the enteric
and autonomic nervous systems of mammalian embryos.
To study the potential function of the c-ret locus in lower
vertebrates, we have isolated its zebra®sh homologue, ret1
and established its expression pattern during embryogenesis. Ret1 mRNA ®rst appears during early somitogenesis
in the presumptive brain, spinal cord and excretory
system. Within the CNS, expression of ret1 is detected in
primary motor and sensory (Rohon ± Beard) neurons.
Ret1 transcripts are also expressed in subsets of neural
crest cells and cranial ganglia as well as in the enteric
nervous system. In the excretory system, expression is
detected in the developing nephric duct and the
pronephros. Our ®ndings reveal a remarkable similarity
in the expression pattern of c-ret between higher and
lower vertebrates, suggesting that the function of this
locus has been conserved throughout vertebrate evolution.
Furthermore, the conservation of ret1 expression in cell
types which remain una€ected by the mammalian c-ret
mutations, such as motor and sensory neurons, suggests a
function of this receptor in these cell lineages.
Keywords: tyrosine kinase receptor; zebra®sh; c-ret
proto-oncogene; nervous system; excretory system
Introduction
Receptor tyrosine kinases (RTKs) are integral membrane proteins which play a critical role in intercellular
communication. All RTKs have a similar structure
characterized by the presence of an extracellular
ligand-binding domain, a transmembrane segment
and an intracellular region which contains the tyrosine
kinase catalytic domain (Ullrich and Schlessinger,
1990; van der Geer et al., 1994). Genetic studies in
invertebrates and vertebrates have established that
RTKs play a crucial role in regulating a wide range
of developmental processes (Pawson and Bernstein,
1990). For example, in Drosophila melanogaster, the
speci®cation of the R7 photoreceptor cell fate requires
the activation of the sevenless RTK (Rubin, 1991),
whilst in Caenorhabditis elegans, activation of let-23, an
RTK of the EGF receptor subfamily, is necessary for
normal larval growth and vulval development (Sternberg and Horvitz, 1991). In mice, mutations in the W
Correspondence: V Pachnis
Received 28 October 1996; revised 16 January 1997; accepted 17
January 1997
and Sl loci, which encode the kit RTK and its ligand
(steel) respectively, result in abnormal di€erentiation of
melanocytes, blood cells and germ cells (Geissler et al.,
1988; Green, 1989).
c-ret, ®rst isolated as an oncogene using the NIH3T3
cell transformation assay, encodes a member of the
RTK superfamily (Takahashi et al., 1985). To date,
closely related homologues of c-ret have been cloned
from three vertebrate species, namely human, mouse
and chicken (Takahashi et al., 1988; Iwamoto et al.,
1993; Pachnis et al., 1993; Robertson and Mason, 1995;
Schuchardt et al., 1995). In mammalian and avian
embryos, high levels of c-ret mRNA have been detected
predominantly in the developing excretory and nervous
system (Pachnis et al., 1993). In the excretory system, cret is expressed in the mesonephric (Wolan) duct and
the epithelium of the ureteric bud and the tips of its
branches. In the central nervous system (CNS), c-ret is
expressed in the somatic and visceral motor neuron
groups of the hindbrain and the spinal cord, as well as
in speci®c subsets of cells of the neuroretina. In the
peripheral nervous system (PNS), c-ret mRNA is
present at high levels in subsets of cells of the sensory,
autonomic and enteric neuronal lineages (Pachnis et al.,
1993; Tsuzuki et al., 1995; CVM-G and VP, unpublished data). Recently, a number of genetic studies have
indicated that the ret receptor has a critical function
during mammalian embryogenesis. In humans, germline mutations in the c-RET proto-oncogene have been
identi®ed in a high percentage of patients with
congenital megacolon, a condition characterized by
the loss of enteric neurons from the terminal part of the
colon (Hirschsprung's disease; Edery et al., 1994;
Romeo et al., 1994). Also, speci®c germ-line mutations
have been identi®ed in two dominantly inherited cancer
syndromes, Multiple Endocrine Neoplasia type 2A and
2B (MEN2A and MEN2B), which are characterized by
tumours of various neural crest derivatives (Mulligan et
al., 1993; Hofstra et al., 1994). In mice targeted
inactivation of c-ret results in death of the animals at
the newborn stage due to severe renal hypodysplasia or
agenesis and absence of subsets of sympathetic and
enteric neurons (Schuchardt et al., 1994; Durbec et al.,
1996a). However, several cell types of the CNS and
PNS which express high levels of c-ret mRNA, such as
motor neurons and sensory neurons, remain una€ected
in mutant embryos, suggesting that the ret receptor has
no function in these cell lineages. This contrasts with
the conserved pattern of c-ret expression between avian
and mammalian species in the majority of cell types
analysed, including sensory and motor neurons. It was
therefore of interest to determine the pattern of
expression of c-ret in a lower vertebrate species.
ret-1 expression in zebrafish
CV Marcos-Gutie rrez et al
880
The zebra®sh (Danio rerio), a member of the teleost
family, is an excellent model system for gene function
studies during vertebrate embryogenesis and for
evolutionary comparison of gene expression between
higher and lower vertebrates (Weinberg, 1992). Here,
we report the cloning and preliminary characterization
of the spatial and temporal pattern of expression
during embryogenesis of the zebra®sh homologue of
the c-ret proto-oncogene. Our data indicate a high
degree of conservation of both the primary sequence
and the pattern of expression of the ret receptor
between lower and higher vertebrates. Interestingly,
expression is also conserved in lineages which remain
apparently una€ected by the mammalian c-ret mutations.
Results
Structural conservation of the ret receptor between lower
and higher vertebrates
To isolate the zebra®sh homologue of the mammalian
c-ret proto-oncogene, two neurula stage cDNA
libraries were screened with the full length murine cret cDNA (Pachnis et al., 1993; Takahashi et al., 1993).
Two overlapping clones were isolated which, upon
sequencing, appeared to represent the zebra®sh
homologue of the mammalian c-ret locus, called
henceforth ret1. Despite the overall homology of the
ret1 sequence, as determined from our cDNA clones,
to that of the human, mouse and chicken c-ret
mRNAs, a 112-base pair segment of clone pzfret7
(region x, Figure 1a) appeared to disrupt the overall
colinearity between these mRNAs and introduce an in
frame termination codon which would result in a
truncated protein containing only part of the extracellar domain of the ret receptor. Closer inspection of
the sequence of this segment indicated that it is ¯anked
by sequences which match the splice donor and the
splice acceptor consensus sequences (not shown),
raising the possibility that pzfret7 represents a
partially processed form of the ret1 primary transcript
containing an unspliced intron. Since further cDNA
library screens failed to identify additional cDNA
clones which would correspond to the region containing the apparent insertion, we combined reverse
transcription and polymerase chain reaction (RT ±
PCR) to amplify the relevant sequences of ret1 mRNA.
Using primers ¯anking the apparent insertion (Figure
1a, oli 1 and oli 2), we generated an ampli®cation
product which was shorter than the size expected from
the cDNA clone pzfret7 by approximately 112 base
pairs (not shown). Sequencing of several clones
representing this ampli®cation product indicated that
they all lacked the 112-base pair insertion which led us
to conclude that the majority of the zebra®sh ret1
mRNA molecules lack this insertion and therefore are
capable of encoding a mature receptor. The sequence
presented in Figure 1b is based on both the cDNA
cloning and the PCR ampli®cation experiments. At this
point, however, we cannot exclude the possibility that a
small percentage of ret1 mRNA molecules contain this
insertion region and therefore are encoding a truncated
and presumably soluble form of the extracellular
domain of the receptor. The functional signi®cance of
such a potential form of the ret receptor is at present
unclear.
Figure 1b compares the amino acid sequence of the
zebra®sh, chicken, mouse and human ret receptors.
The zebra®sh ret1 mRNA is capable of encoding a
1106 amino acid-long protein of Mr 125 000. The
predicted protein shows 60% and 55% overall identity
to the amino acid sequence of the avian and
mammalian ret receptors, respectively. As is the case
for the avian and mammalian receptors, a region with
homology to the cell adhesion proteins, cadherins, is
also found in the extracellular domain of the zebra®sh
ret1 receptor. This region includes the (V/
L)DREXXXXYXL motif, involved in the coordination of Ca2+ and the DEDD and DXD motifs which
function as `linker' sequences (Figure 1b; Schneider,
1992; Takeichi, 1991). Another feature of the ret
receptor, is the presence of a cysteine-rich domain in
the C-terminal region of the extracellular domain.
Overall, out of 25 cysteine residues conserved between
the chick, mouse and human extracellular domains, 23
are also conserved in the zebra®sh (Figure 1b). In the
intracellular part of the zebra®sh protein, a tyrosine
kinase domain (which is approx 85 ± 88% identical to
that of the mammalian and avian ret receptors), is split
by a highly conserved insert region (Hunter et al.,
1990). Di€erential splicing at the 3'-end of the primary
c-ret transcript in mammals and birds, leads to the
generation of two protein isoforms, the ret9 and ret51
isoforms. The ret51 isoform is generated by replacement of the 9 C-terminal amino acids of the ret9
isoform by 51 unrelated amino acids (Ishizaka et al,
1989; Tahira et al., 1990). The predicted protein
encoded by the isolated zebra®sh cDNA clones
corresponds to the ret51 isoform present in mammals
and birds. Taken together, our data strongly suggest
that we have isolated the zebra®sh homologue of the
mammalian and avain c-ret genes.
Expression of ret1 during zebra®sh embryogenesis
Northern blot analysis established that ret1 is
expressed during embryogenesis and generates a
complex set of abundant transcripts which are of
similar size to the mammalian transcripts reported
previously (data not shown; Ishizaka et al., 1989;
Pachnis et al., 1993). To determine the spatial and
temporal pattern of expression of the ret1 gene during
zebra®sh development, embryos of various stages were
analysed by in situ hybridization using cRNA probes
that detect all transcripts. ret1 mRNA was ®rst
detected in speci®c groups of cells at the 3-somite
stage and was maintained throughout the embryonic
period and in the adult. As is the case for the
mammalian and avian c-ret genes, the main sites of
expression of ret1 are the excretory and the nervous
systems. The conservation of the overall pattern of
expression of the ret locus between lower and higher
vertebrates is indicated by comparing Figure 2c and f.
mRNA distribution in the developing excretory system
In the excretory system, ret1 transcripts were ®rst
detected in the pronephric ducts of 4 ± 6-somite stage
(10 ± 12 hpf) embryos (not shown). During the following 6 h of development and up to the 18-somite stage,
ret-1 expression in zebrafish
CV Marcos-GutieÂrrez et al
881
a
b
Figure 1 Structure of the zebra®sh ret tyrosine kinase receptor and comparison of its amino acid sequences to the chicken, mouse
and human counterparts. (a) Partial restriction enzyme map of the cDNAs identi®ed. Region x indicates the 112-base pair insertion
found in the pzfret7 clone. Oli 1 (5'-CCT CGG CAC GTG CCC TGA TGG TTA CTG-3') and oli 2 (5'-GAG CCG GAA AGC
TGT GGC CTT CAC CAC C-3') show the primers used in the RT ± PCR experiments. Shown at the bottom is a diagrammatic
representation of the various domains of the ret receptor as deduced from the amino acid sequence. Abbreviations: SP, signal peptide;
UTR, untranslated region; c, cysteine residue; TM, transmembrane domain; TK, tyrosine kinase domain; ret51, ret51-speci®c
sequence. (b) Amino acid sequence of the zebra®sh ret1 receptor and its comparison to the chicken, mouse and human proteins. The
conserved cysteine residues present in the extracellular domain and the cadherin-like motifs (L)DREXXXXYXL, DX(N)D and
DXD, are highlighted. The shaded region following the cystein-rich domain is the transmembrane segment of the receptor
ret-1 expression in zebrafish
CV Marcos-Gutie rrez et al
882
the distribution of ret1 mRNA was similar throughout
the length of the nephric (Wolan) duct of the
opisthonephros, the main excretory organ of the
embryonic and adult zebra®sh (Figure 2a and b).
However, subsequent to the 18-somite stage, expression
became gradually downregulated in a rostral to caudal
sequence and by the 26-somite stage (22 hpf), ret1
expression was restricted to the most posterior part of
the nephric duct (Figure 2c and d). In addition to the
expression in the pronephric and nephric ducts, ret1
mRNA was also detected in a group of mesodermal
cells located ventral to the notochord at the level of the
third somite (Figure 2c and e). This is the area where
the pronephric kidneys form (Wester®eld, 1994),
suggesting that ret1 mRNA is expressed in these
structures during zebra®sh embryogenesis.
Figure 2 Expression of ret1 in the excretory system of the zebra®sh embryo. (a) and (b) represent lateral and dorsal views
respectively of a 12-somite embryo hybridized with a ret1 cRNA probe. In addition to expression in ventrally located cells of the
forebrain (fb), hindbrain (hb) and the spinal cord (sc), high levels of ret1 mRNA are also detected in the posterior 2/3rds of the
nephric duct (nd). (c) In the prim-5 stage (24 hpf) embryo expression in the nephric duct is restricted to its most caudal part. Lower
levels of ret1 mRNA are also detected in the pronephros (pn). (d) Close up view of the caudal part of the nephric duct of a similar
stage embryo (22 hpf). Note the sharp boundary between the ret1-positive and ret1-negative domains of the nephric duct. (e)
Transverse section of a prim-5 (24 hpf) embryo at the level of the third somite. Expression of ret1 mRNA is detected in
ventrolaterally located cells in the spinal cord (sc) and in the pronephroi (pn). (f) Whole-mount in situ hybridization histochemistry
of a E10.5 mouse embryo with a c-ret riboprobe. The expression of c-ret mRNA is also restricted at this stage to the caudal-most
portion of the mesonephric duct (md). Compare also the expression of c-ret mRNA in the cranial ganglia of the mouse embryo with
expression of ret1 mRNA in the granial ganglia of the zebra®sh embryo shown in Figure 5c and d. ov, otic vesicle. VII, IX and X
represent the facial, glossopharyngeal and nodose ganglia. Scale bar is 210 mm in a, b, c, 43.5 mm in d and 30 mm in e
ret-1 expression in zebrafish
CV Marcos-GutieÂrrez et al
ret1 expression in the developing CNS
ret1 speci®c signal was observed in the presumptive
brain from the 3-somite stage. In the forebrain region
of 12- and 18-somite embryos, bilaterally symmetrical
clusters of ret1-positive cells were localized ventral to
the optic stalks, in a region that included the
developing hypothalamus of the diencephalon (Figure
3a and b). In addition to the positive diencephalic
clusters, ret1 expression was also observed in groups of
cells located dorsal to the optic stalks in cells of the
nasal epithelium and the dorsorostral telencephalon
(Figure 3a and b, op and t). Also, at this stage, ret1
expressing cells were observed in the epiphysis (Figure
3b, e).
In mice, high levels of c-ret mRNA are present in
subsets of cells of the developing retina (Pachnis et al.,
1993). However, ret1 mRNA was not detected in the
neuroretina of the zebra®sh embryo. To investigate the
possibility of ret1 expression in the fully developed
retina, we performed in situ hybridization on cryosections of adult zebra®sh. As shown in Figure 3c, ret1 is
Figure 3 Expression of ret1 in the brain of the zebra®sh embryo. (a) Flat-mount preparation of the brain region of a 12-somite
(15 hpf) embryo. Signal is detected in the telencephalon (t) and the diencephalon (d). (b) In the 18-somite embryo, ret1 mRNA is
also present in the olfactory placode (op) of the telencephalon and the hypothalamus (hy) of the diencephalon. At this stage,
expression is also detected in the dorsally located epiphysis (e) and in ventrally located groups of cells of the midbrain (mb). (c) in
situ hybridization on a section of adult zebra®sh retina. ret1-speci®c signal is present predominantly over the layer of amacrine (am)
cells. The few positive cells in the ganglion cell layer (g; arrowheads) are likely to represent displaced amacrine cells. The
photoreceptor (ph) cell layer is negative. (d) High-power view of the nucleus of the MLF (nMLF) from a prim-5 (24 hpf) embryo
labelled for ret1 expression by in situ hybridization (blue precipitate: arrow) and HNK-1 antibody (brown signal; arrowhead).
Coexpression of the two markers is restricted to the anterior group of neurons of the nMLF. The posterior groups of nMLF
neurons are positive for the HNK-1 marker, but do not express ret1 mRNA. (e) Flat-mount preparation of the hindbrain (hb)
region of a prim-5 embryo. ret1 expressing cells are present in the most rostral (arrowhead) and caudal (arrows) rhombomeres. ret1
mRNA is also present in cranial ganglia of the PNS, such as the facial (fg) and posterior lateral line (pllg) ganglia. ov, otic vesicle.
(f) High-power view of the ®rst rhombomere of the anterior hindbrain of a prim-5 embryo. The axons of the ret1-positive neurons
join the MLF. Scale bar is 87 mm in a, b; 43.5 mm in c, e and 17.5 mm in d, f
883
ret-1 expression in zebrafish
CV Marcos-Gutie rrez et al
884
predominantly expressed in a single cell layer, the
location of which suggests that it corresponds to the
amacrine cells (am). No expression was detected in the
other layers of the retina, including the photoreceptors
(ph; Figure 3c). A small number of ret1-positive cells
present in the ganglion cell layer (g) are likely to
represent displaced amacrine cells (Masland et al.,
1993). ret1 transcripts were also found in speci®c cells
in the larval and adult forebrain, the identity of which
remains unknown (not shown).
In the midbrain, expression of ret1 mRNA was ®rst
detected in a ventrally located group of cells at the 18somite stage (Figure 3b, mb). To identify the ret1positive cells, the expression of the HNK-1 epitope was
examined in prim-5 (24 hpf) embryos that had been
previously hybridised with ret1 cRNA. As shown in
Figure 3d, the ret1 positive region in the midbrain
coincides with the most anterior neurons in the nucleus
of the medial longitudinal fasciculus (nMLF).
In the hindbrain, ret1 expression was ®rst detected at
the 18-somite stage (18 hpf). By the prim-5 stage
(24 hpf), a few strongly expressing cells were present
in the anterior and posterior rhombomeres (Figure 3e).
Double labelling with the HNK-1 antibody and the ret1
cRNA probe, showed that the ret1-positive neurons
present in rhombomere 1 extend their axons along the
MLF pathway (Figure 3f). The identity of ret1-positive
cells in the posterior hindbrain was not established,
although their location suggests that they belong to the
group of reticulospinal neurons (Mendelson, 1986a,b).
Expression in the spinal cord was ®rst detected at
the 3-somite stage (not shown). Between the 3 to 12somite stages, ret1 mRNA was restricted to ventrally
located cells along the anterior-posterior axis (Figure
4a), while by the prim-5 stage (24 hpf), ret1 expression
was detected in both ventrally as well as dorsally
located groups of cells (Figure 4b). By the 14-somite
stage (16 hpf), the ventrally located ret1-expressing
cells were arranged in a segmental pattern relative to
the borders of hemisegments 8 to 20 (2 to 3 ret1
positive cells per hemisegment; Figure 4c). These cells
were likely to be primary motor neurons as indicated
by their early appearance, ventrolateral location and
the relatively large size of their cell bodies (Figure 4b
and c; Eisen et al., 1986; Myers et al., 1986; Eisen,
1991). Vital dye labelling and immunohistochemistry
using several neuron-speci®c markers, has indicated
that three primary motor neurons are present in each
hemisegment: the CaP, (Caudal Primary, caudally
located); RoP, (Rostral Primary, rostrally located)
and MiP, (Middle Primary, between CaP and RoP).
A fourth primary motor neuron, which is variably
present, is found in about half of the spinal
hemisegments (VaP, Variable Primary, adjacent to
CaP; Eisen et al., 1990). To determine whether the
ret1-expressing cells were indeed primary motor
neurons, in situ hybridization histochemistry (with a
ret1 cRNA probe) and immunohistochemistry with an
isl-1/2 antiserum were combined on prim-5 to prim-16
(24 to 30 hpf) embryos (Figure 4d). Isl-1/2 speci®c
antibodies have been shown to identify all primary
motor neurons of the zebra®sh embryo (Korzh et al.,
1993). As shown in Figure 4d, the RoP and MiP
primary motor neurons, as de®ned by their position
just rostral to the hemisegment border and by
expression of isl-1/2, were also expressing ret1
mRNA, while the CaP and VaP neurons were ret1negative (Figure 4d). In the hemisegments caudal to
somite 20, most ventrolaterally located cells expressed
ret1, but did not show a segmental arrangement at any
of the stages examined. This was also the case of
ventrolateral ret1-expressing cells in the spinal cord
between somites to 4 to 7. ret1 transcripts were not
observed in the spinal cord at the level of somites 1 to
3 at any of the stages analysed (not shown).
In addition to the primary motor neurons, ret1
positive cells were also located in the dorsal aspect of
the spinal cord (Figure 4b). These large and
nonsegmentally arranged cells were thought to
represent the primary sensory Rohon ± Beard (RB)
neurons. This was further supported by double staining
experiments on a 18-somite stage (18 hpf) embryo in
which in situ hybridization histochemistry was combined with immunohistochemistry using the HNK-1
monoclonal antibody (Figure 4e). The HNK-1 epitope
has been shown previously to be present in many
primary neurons during early zebra®sh development,
including RB neurons (Metcalfe et al., 1990). As shown
in Figure 4e, the ret1 expressing cells of the dorsal
spinal cord were also positive for the HNK-1 epitope.
The number of ret1-positive RB neurons was found to
be variable and in some embryos we were unable to
detect ret1 mRNA in the RB neurons.
ret1 expression in the developing PNS
Expression of ret1 mRNA was also detected in groups
of cranial cells outside the CNS. Double label in situ
hybridization histochemistry using a combination of
ret1 and krox-20 cRNA probes (Oxtoby and Jowett,
1993), indicated that at the 6-somite stage, a small
number of ret1-positive cells was located adjacent to
the dorsal hindbrain (Figure 5a). The position of these
cells adjacent to rhombomeres 3 and 4, suggested that
they were derived from the neural crest. At later stages
(12-somite stage), the population of ret1-positive cells
adjacent to the hindbrain expands and occupies an
area extending from rhombomeres 3 to 5. A cross
section at the level of rhombomere 5 of a 12-somite
stage embryo is shown in Figure 5b.
During the next stages analysed (18 ± 24 somites),
ret1 expression was detected in speci®c subsets of
cranial ganglia of the PNS. To identify the ret1-positive
cranial ganglia, in situ hybridization for ret1 mRNA
was combined in some embyros with HNK-1
immunostaining. As shown in Figure 5c, d and e at
these stages, ret1 mRNA was expressed in the
trigeminal, facial and posterior lateral line ganglia,
while the acoustic ganglia were negative. High levels of
ret1 expression in the facial and posterior lateral line
ganglia were also seen at the long-pec stage (48 hpf;
Figure 5f). The expression of ret1 mRNA in the cranial
ganglia of the zebra®sh embryo is highly reminiscent of
that of c-ret in the mammalian embryo. In the 9.5 ±
10.5 day mouse embryo, c-ret mRNA is expressed in
the facial (VII), glossopharyngeal (IX) and vagal (X)
ganglia, but is absent from the acoustic (VIII) ganglia
(Figure 2f). Although at this stage of mouse
embryogenesis, no expression was detected in the
trigeminal ganglion, c-ret transcripts were detected in
a subset of neurons one to two days later (Pachnis et
al., 1993).
ret-1 expression in zebrafish
CV Marcos-GutieÂrrez et al
In the mammalian and avian embryos, high levels
of c-ret mRNA have been detected in the developing
autonomic nervous system (sympathetic and parasympathetic) and the ENS (Pachnis et al., 1993;
Robertson and Mason, 1995; Schuchardt et al.,
1995; Tsuzuki et al., 1995). To determine whether
ret1 is expressed in the ENS of teleosts, the gut of
larvae and adult zebra®sh was analysed by in situ
hybridization with ret1 cRNA probes. Whilst no ret1
mRNA was detected in the gut of adult zebra®sh (not
shown), ret1-positive cells were present in the gut wall
of 3 ± 8 day old larvae (Figure 5g and h). The
punctate nature of the signal and their localization
in the mesenchyme surrounding the gut endoderm,
suggests that the ret1-positive cells represent the
precursors of the zebra®sh ENS. The distribution of
the ret1-positive cells along the anterior posterior axis
of the gut was not homogeneous during the stages
analysed, with more cells present in the anterior
segments of the ®sh gastrointestinal tract compared to
the posterior (Figure 5g and h).
Discussion
Evolutionary conservation of c-ret
We report here the cloning and initial characterization
of the zebra®sh ret1 gene, the teleost homologue of the
mammalian and avian c-ret proto-oncogenes. Our data
indicate that both the primary structure and the overall
pattern of expression of the ret receptor have been
conserved between lower and higher vertebrates. The
predicted amino acid sequence of the zebra®sh ret1
Figure 4 ret1 expression in the spinal cord of the zebra®sh embryo. (a) Wax section of the anterior spinal cord of a 9-somite (13 hpf)
embryo. Signal is present in a small group of ventrolaterally located cells (arrowhead) in the ventral spinal cord (sc). (b) A manually
generated transverse section through the spinal cord of a prim-5 (24 hpf) embryo. ret1-speci®c signal is present in both ventrally (mn)
and dorsally (rb) located groups of cells in the spinal cord. High levels of ret1 expression are also detected in the bilaterally located
groups of cells of the nephric duct (nd). (c) Flat-mount preparation of a prim-5 (24 hpf) embryo. ret1 is expressed in segmentally
arranged groups of cells in the ventral spinal cord (sc). Note also the high levels of expression in the caudal end of the nephric duct
(nd). (d) Combined immunohistochemistry and in situ hybridization histochemistry using an isl-1/2 antiserum and the ret1 cRNA
probe respectively. The RoP and MiP primary motor neurons coexpress both markers as indicated by the nuclear signal (isl-1/2;
highlighted by dotted circles) and the cytoplasmic signal (ret1). The VaP and CaP primary motor neurons are negative for ret1 but
positive for isl-1/2. (e) Combined immunohistochemistry and in situ hybridization histochemistry using a HNK-1 monoclonal
antibody (arrowheads) and the ret1 cRNA probe (arrows). In this preparation coexpression of both markers is observed in two of the
primary sensory Rohon-Beard (rb) neurons. mn, motor neurons. Scale bar is 43.5 mm in a, b, c and 17.4 mm in d, e
885
ret-1 expression in zebrafish
CV Marcos-Gutie rrez et al
886
Figure 5 Expression of ret1 mRNA in the cranial ganglia of the zebra®sh embryo. (a) Flat-mount preparation of the hindbrain of
a 3-somite (11 hpf) zebra®sh embryo double stained with krox-20 and ret1 cRNA probes. krox-20 is expressed in rhombomeres 3
and 5 (r3 and r5), while ret1 is expressed in a small group of cells (arrowhead) emerging from r3. (b) Manually generated transverse
section at the level of the anterior hindbrain (hb) of a 12-somite embryo. A larger group of putative neural crest cells are positive for
ret1 mRNA (arrow). (c) Flat-mount preparation of the head of a prim-5 (24 hpf) embryo double stained with HNK-1 antibody and
ret1 cRNA probe. (d) A higher magni®cation of the boxed region in c, of a same stage double stained embryo as in c. ret1 is
expressed in the facial ganglia (fg) and the posterior lateral line ganglia (pllg). No expression is detected in the acoustic ganglion
(ag). (e) Flat-mount preparation of the hindbrain (hb) region of an 18-somite (18 hpf) embryo. (f) Flat-mount preparation of the
posterior hindbrain region of a long-pec (48 hpf) zebra®sh larva. ret1 expression still persists in the facial (fg) and posterior lateral
line (pllg) ganglia. (g and h) Transverse sections of day 4 zebra®sh larva at the anterior (f) or posterior (g) gastrointestinal tract.
ret1-positive cells (arrowheads) are present outside the gut endoderm (e). A higher number of ret1-expressing cells is present in the
anterior segments of the gut. Scale bar is 43.5 mm in a, d, f, g, h; 21.7 mm in b; 87 mm in c and 52.6 mm in e
ret-1 expression in zebrafish
CV Marcos-GutieÂrrez et al
protein shows an overall identity of 55 ± 60% when
compared to the mammalian and avian receptors,
while comparison of the intracellular tyrosine kinase
domains revealed over 85% identity. These ®gures are
comparable to those reported for the homologues of
other members of the RTK superfamily (Xu et al.,
1994).
In both the mammalian and the avian ret proteins,
a subdomain has been identi®ed in the extracellular
part of the receptor, which shows clear homology to
the extracellular portion of cadherins, a family of cell
surface proteins mediating homophilic Ca2+-dependent
cell-cell adhesion (Takeichi, 1991; Schneider, 1992). A
similar cadherin-like subdomain was also present in
the zebra®sh ret1 protein. The role of this domain in
mediating cell-cell adhesion is at present unclear, since
aggregation assays failed to show ret-dependent
homophilic interaction between cells expressing the
murine ret receptors (Takahashi et al., 1993). Another
conserved feature of the extracellular region of the ret
receptor, is the presence of a cysteine-rich domain.
The biological signi®cance of this domain is highlighted by the fact that point mutations in the human
receptor, which replace speci®c cysteine residues
proximal to the transmembrane segment (conserved
in equivalent positions in the ret1 receptor), lead to
ligand-independent and constitutive activation of its
kinase domain (Mulligan et al., 1993; Asai et al.,
1995). The conservation of the cadherin-like and
cysteine-rich domains between teleosts and mammals,
separated by at least 300 ± 400 million years of
evolution, indicates that they are likely to play an
important role in ligand binding or signalling by the
ret receptor.
Analysis of the human (Ishizaka et al., 1989; Tahira
et al., 1990) and murine (CVM-G and VP unpublished
data) c-ret genes indicate that they are capable of
encoding two isoforms, termed ret9 and ret51, which
di€er exclusively at their COOH-termini. Amino acid
sequence comparison revealed that the zebra®sh
cDNAs we have isolated, encode the teleost homologue of the mammalian ret51 isoform. Although
additional library screens have failed to isolate ret9
isoform-speci®c cDNAs, we have been able to identify
within the zebra®sh ret1 locus a region which is highly
homologous to the mammalian ret9-speci®c cDNA
sequences (CVM-G and VP; unpublished data). This
suggests that, as in higher vertebrates, teleosts also
generate two isoforms of the ret receptor. Although
the di€erential regulation of expression and the
functional signi®cance of the ret receptor isoforms is
not clear at present, the evolutionary conservation of
their sequences among vertebrates suggests the
activation of conserved and isoform speci®c intracellular signalling pathways (Borrello et al., 1994).
Consistent with this hypothesis, we have recently
observed di€erential e€ects of the two mammalian
ret isoforms when ectopically expressed in animal cap
cells of blastula stage Xenopus embryos (P Durbec, C
Kilkenny, CVM-G and VP; unpublished observations).
Expression during embryogenesis
As found in mammalian embryos, one of the main sites
of expression of ret1 in zebra®sh embryos is the
developing excretory system. ret1 mRNA was detected
in both the pronephric kidney and the Wolan duct of
the opisthonephros, initially distributed uniformly
throughout its length, but subsequently con®ned to
its posterior end. The expression of ret1 in the
excretory system of the zebra®sh embryo, combined
with the established role of the receptor in the
development of the mammalian kidney (Schuchardt et
al., 1994, 1996), suggest that ret1 is required for the
normal development of the excretory system of lower
vertebrates. In mammlian embryos, the caudal part of
the Wolan duct (which also expresses the highest
levels of c-ret mRNA), gives rise to the ureteric bud
which, upon invasion of the adjacent metanephric
mesenchyme and branching, induces formation of the
excretory units of the permenant kidney, the nephrons.
At the same time, reciprocal signals from the
undi€erentiated mesenchyme induce proliferation of
the epithelial cells and further branching of the ureteric
bud (Saxen, 1987). Based on the pattern of expression
of the mammalian ret receptor and the analysis of the
phenotypic e€ects of mutations of the murine c-ret
locus on kidney development, we have previously
suggested that ret is the receptor for a mesenchyme
derived signal, which stimulates the initial formation
and the subsequent branching of the ureteric bud
(Schuchardt et al., 1994, 1996). In teleosts such as the
zebra®sh, there is no structure analogous to the
ureteric bud and the metanephros, and the excretory
functions of the adult animal are served by the
opisthonephros, a structure that could be viewed as
analogous to the avian and mammalian mesonephros.
It is therefore likely that in lower vertebrates the
function of the ret receptor is restricted to the growth
and di€erentiation of the cells of the Wolan duct and
the opisthonephric tubules. Interestingly, although the
most dramatic e€ects of c-ret ablation in mammals is
severe dysplasia or aplasia of the metanephros, we have
recently established an additional e€ect on the
development of the mesonephros (Schuchardt et al.,
1996). Although the complete understanding of the role
of the ret1 in the development of the zebra®sh
excretory system will require further experiments, our
analysis of ret1 expression in the kidney of the
zebra®sh, suggests that the basic molecular mechanisms underlying the development of the excretory
system have been conserved between lower and higher
vertebrates.
In the spinal cord, ret1 is expressed in both primary
motor and sensory (RB) neurons. The identity of the
ventromedially located ret1-expressing cells as primary
motor neurons was established by a combination of in
situ hybridization with ret1 cRNA probes and
immunohistochemistry with an antiserum identifying
two members of the LIM family of homeodomain
containing transcription factors, isl-1 and isl-2.
Interestingly, in the prim-5 stage embryo, ret1 mRNA
is expressed in a subset of motor neurons. Distinct
subclasses of motor neurons have been identi®ed in the
spinal cord of higher and lower vertebrate (chick and
zebra®sh) embryos on the basis of their columnar
organisation and selection of axonal pathway. Expression and genetic studies have suggested that members
of the LIM family of transcription factors (Tsuchida et
al, 1994; Appel et al., 1995), play a critical role in the
speci®cation and di€erentiation of subclasses of motor
887
ret-1 expression in zebrafish
CV Marcos-Gutie rrez et al
888
neurons. The expression of ret1 in the spinal cord of
the zebra®sh embryo suggests that this receptor is also
part of a signalling and molecular cascade necessary
for the di€erentiation or acquisition of identity of
speci®c subsets of primary motor neurons.
Another major group of neurons expressing ret1
mRNA are the primary sensory neurons mediating
touch sensitivity, the Rohon ± Beard neurons of the
spinal cord and the sensory neurons of the trigeminal
ganglion. The RB neurons are characterised by their
large size, expression of the HNK-1 epitope and their
non-segmental arrangement at the dorsal 1/3 of the
spinal cord. The trigeminal sensory neurons are bipolar
neurons forming compact ganglia, between the developing eye and the otic vesicle (Figure 5e). Coexpression
of ret1 mRNA by RB and trigeminal sensory neurons
further supports the suggestion (originally based on the
analysis of various phenotypic markers) that these two
groups of neurons share a common developmental
programme (Metcalfe et al., 1990). Expression of c-ret
mRNA has also been detected in a speci®c subpopulation of sensory neurons of mammalian embryos
although the sensory modality of these neurons has
not been established (Pachnis et al., 1993).
The conserved expression of ret in motor and
sensory neurons of lower and higher vertebrates,
supports the hypothesis that the receptor plays a role
in the di€erentiation or survival of these groups of
neurons during embryogenesis. This hypothesis is
further supported by the recent identi®cation of glial
cell line-derived neurotrophic factor (GDNF) as a
functional ligand for the ret RTK, since this molecule
has a dramatic trophic e€ect on several groups of CNS
neurons, including motor neurons (Durbec et al.,
1996b; Jing et al., 1996; Trupp et al., 1996).
However, phenotypic analysis of mouse embryos and
newborn animals mutated at the c-ret locus, failed to
reveal any major abnormalities of motor neuron pools
and sensory ganglia, although at this point we cannot
exclude the possibility that subtle defects are present
(CVM-G and VP; unpublished data). The ability to
express dominant negative forms of the ret1 receptor in
the zebra®sh embryo, will allow us to address directly
the potential role of this signalling pathway in the
motor and sensory neurons in an experimentally
accessible vertebrate.
In mammals, loss-of-function mutations of the c-ret
proto-oncogene lead to dramatic defects in the
development of the ENS and subsets of sympathetic
and parasympathetic ganglia. Despite the paucity of
data on the expression in the autonomic nervous
system of zebra®sh, the presence of ret1 mRNA in the
ENS of the larva suggests that, as in mammalian
embryos, the ret receptor plays a critical role in the
development of the ENS in lower vertebrates.
Furthermore, ret1 mRNA can serve as a useful
marker for studies on the development of the ®sh ENS.
cDNA cloning and sequencing
Materials and methods
Acknowledgements
We thank Qiling Xu and Stefan Schulte-Mercker for
technical advice and stimulating discussions. We also
thank Brent Bisgrove and David Grunwald for sharing
unpublished sequence data. This work was supported by
the Medical Research Council (MRC). EMBL data base
accession number is X94363.
Animals
Embryos were maintained as described previously
(Wester®eld, 1994) and staged according to Kimmel et al.
(1995) and by hours post-fertilization at 28.58C (hpf).
An oligo-dT primed neurula-stage cDNA library (gift from
Dr David Grunwald), was screened using the mouse full
length c-ret cDNA under low stringency hybridization and
washing conditions (hybridization: 56SSC, 106dextran
sulphate, 106Denhardt's solution, 0.1% SDS and 100 mg/
ml salmon sperm DNA at 608C; washes: 0.56SSC at 608C
for 30 min). A random primed neurula-stage cDNA library
(gift from Prof Jose Campos-Ortega) was screened under
high stringency conditions (hybridization: 36SSC,
106dextran sulphate, 106Denhardt's solution, 0.1%
SDS and 100 mg/ml salmon sperm DNA at 658C; washes:
0.16SSC at 658C for 1 h). Nested deletions of the cDNA
clones were generated using the Erase-a-Base kit (Promega)
and sequenced using the Sequenase Version 2.0 kit (USB).
In situ hybridization
Non-radioactive in situ hybridization on sections from
larval and adult zebra®sh was performed as described
(Schaeren-Wiemers and Ger®n-Moser, 1993). Whole mount
in situ hybridization was performed according to previously
described protocols by Xu et al. (1994). Double label whole
mount in situ hybridization using ret1 and krox-20 cRNAs
was performed as previously described (Jowett and Lettice,
1994). Following whole mount in situ hybridization,
embryos were re®xed with 4% paraformaldehyde in PBS,
dehydrated in a graded series of ethanol solutions,
embedded in paran wax at 608C and 14 mm sections
were collected onto Tespa-coated slides, rehydrated and
mounted in 80% glycerol. Manually section embryos were
cut with a tungsten needle. For all the in situ hybridization
experiments, pzfret7 was linearized with EcoRI and cRNA
probes were generated using digoxigenin labelled UTP and
T7 RNA polymerase as described previously (Wilkinson,
1992).
Stained embryos and sections were observed and
photographed with a microscope equipped with di€erential
interference contrast optics (DIC; Zeiss Axiophot) or with a
dissecting microscope (Leica Wild 10).
Immunohistochemistry
Dechorionated embryos were ®xed with 4% paraformaldehyde at 48C for 24 h and stored in methanol at 7208C
until used. After rehydration in a graded methanol/PBS
series (5 min each), the specimens were washed in four
changes of PBT (PBS, 0.1% Triton X-100) for 5 min each.
The tissues were incubated overnight at 48C with either
anti isl-1/2 (1 : 200; gift from T Jessell) or anti-HNK-1
(1 : 100; gift from G Rougon) antibodies. The primary
antibodies were detected with the ABC immunoperoxidase
system according to the manufacturer's speci®cations
(Vector Laboratories Inc.).
For ret1 and HNK-1 double-staining, embryos were ®rst
treated for in situ hybridization with ret1 cRNA, ®xed with
4% paraformaldehyde for 12 h at 48C and then incubated
with the HNK-1 antibody as described above. For ret1 and
isl-1/2 antibody double-labelling, embryos were ®rst stained
with the isl-1/2 antibody and then in situ hybrized with ret1
cRNA.
ret-1 expression in zebrafish
CV Marcos-GutieÂrrez et al
889
References
Appel B, Korzh V, Glasgow E, Thor S, Edlund T, Dawid IB
and Eisen JS (1995). Development, 121, 4117 ± 4125.
Asai N, Iwashita T, Matsuyama M and Takahashi M.
(1995). Mol. Cell Biol., 15, 1613 ± 1619.
Borrello MG, Pelicci G, Arighi E, De-Filippis L, Greco A,
Bongarzone I, Rizzetti M, Pelicci PG and Pierotti MA.
(1994). Oncogene, 9, 1661 ± 1668.
Durbec PL, Larsson Blomberg LB, Schuchardt A, Costantini F and Pachnis V. (1996a). Development, 122, 349 ± 358.
Durbec P, Marcos-Gutierrez CV, Kilkenny C, Grigoriu M,
Wartiowaara K, Suvanto P, Smith D, Ponder B,
Costantini F, Saarma M, Sariola H and Pachnis V.
(1996b). Nature, 381, 789 ± 793.
Edery P, Lyonnet S, Mulligan LM, Pelet A, Dow E, Abel L,
Holder S, Nihoul Fekete C, Ponder BA and Munnich A.
(1994). Nature, 367, 378 ± 380.
Eisen JS, Myers PZ and Wester®eld M. (1986) Nature, 320,
269 ± 271.
Eisen JS, Pike SH and Romancier B. (1990). J. Neurosci, 10,
34 ± 43.
Eisen JS. (1991). Science, 252, 569 ± 572.
Geissler EN, Ryan MA and Housman DE. (1988). Cell, 55,
185 ± 192.
Green MC. (1989). Genetic Variants and Strains of the
Laboratory Mouse: Catalog of mutant genes and polymorphic loci. Lyon MF and Searle AG (eds). Oxford
University Press: Oxford, pp. 12 ± 403.
Hofstra RM, Landsvater RM, Ceccherini I, Stulp RP,
Stelwagen T, Luo Y, Pasini B, Hoppener JW, van Amstel
HK, Romeo G et al. (1994). Nature, 367, 375 ± 376.
Hunter T, Lindberg RA and Middlemas DS. (1990). Adv.
Second Messenger Phosphoprotein Res., 24, 260 ± 265.
Ishizaka Y, Ochiai M, Tahira T, Sugimura T and Nagao M.
(1989). Oncogene, 4, 789 ± 794.
Iwamoto T, Taniguchi M, Asai N, Ohkusu K, Nakashima I
and Takahashi M. (1993). Oncogene, 8, 1087 ± 1091.
Jing S, Wen D, Yu Y, Holst PL, Luo Y, Fang M, Tamir R,
Antonio S, Hu Z, Cupples R, Louis JC, Hu S, Altrock BW
and Fox GM. (1996). Cell, 85, 1113 ± 1124.
Jowett T and Lettice L. (1994). Trends Genet., 10, 73 ± 74.
Kimmel CB, Ballard WW, Kimmel SR, Ullmann B and
Schilling TF. (1995). Dev. Dyn., 203, 253 ± 310.
Korzh V, Edlund T and Thor S. (1993). Development, 118,
417 ± 425.
Masland RH, Rizzo JF and Sandell JH. (1993). J. Neurosci,
13, 5194 ± 5202.
Mendelson B. (1986a). J. Comp. Neurol., 251, 160 ± 171.
Mendelson B. (1986b). J. Comp. Neurol., 251, 172 ± 184.
Metcalfe WK, Myers PZ, Trevarrow B, Bass MB and
Kimmel CB. (1990). Development, 110, 491 ± 504.
Mulligan LM, Kwok JB, Healey CS, Elsdon MJ, Eng C,
Gardner E, Love DR, Mole SE, Moore JK, Papi L et al.
(1993). Nature, 363, 458 ± 460.
Myers PZ, Eisen JS and Wester®eld M. (1986). J. Neurosci, 6,
2278 ± 2289.
Oxtoby E and Jowett T. (1993). Nucleic. Acids Res., 21,
1087 ± 1095.
Pachnis V, Mankoo BS and Costantini F. (1993). Development, 119, 1005 ± 1017.
Pawson T and Bernstein A. (1990). Trends Genet., 6, 350 ±
356.
Robertson K and Mason I. (1995). Mech. Develop., 53, 329 ±
344.
Romeo G, Ronchetto P, Luo Y, Barone V, Seri M,
Ceccherini I, Pasini B, Bocciardi R, Lerone M,
Kaariainen H et al. (1994). Nature, 367, 377 ± 378.
Rubin GM. (1991). Trends Genet., 7, 372 ± 377.
Saxen L. (1987). Organogenesis of the kidney. Cambridge
University Press: Cambridge.
Schaeren-Wiemers N and Ger®n-Moser A. (1993). Histochemistry, 100, 431 ± 440.
Schneider R. (1992). Trends Biochem. Sci., 17, 468 ± 469.
Schuchardt A, D'Agati V, Larsson Blomberg L, Costantini F
and Pachnis V. (1994). Nature, 367, 380 ± 383.
Schuchardt A, Srinivas S, Pachnis V and Costantini F.
(1995). Oncogene, 10, 641 ± 649.
Schuchardt A, D'Agati V, Pachnis V and Costantini F.
(1996). Development, 122, 1919 ± 1929.
Sternberg PW and Horvitz HR. (1991). Trends Genet., 7,
366 ± 371.
Tahira T, Ishizaka Y, Itoh F, Sugimura T and Nagao M.
(1990). Oncogene, 5, 97 ± 102.
Takahashi M, Ritz J and Cooper GM. (1985). Cell, 42, 581 ±
588.
Takahashi M, Buma Y, Iwamoto T, Inaguma Y, Ikeda H
and Hiai H. (1988). Oncogene, 3, 571 ± 578.
Takahashi M, Asai N, Iwashita T, Isomura T, Miyazaki K
and Matsuyama M. (1993). Oncogene, 8, 2925 ± 2929.
Takeichi M. (1991). Science, 251, 1451 ± 1455.
Trupp M, Arenas E, Fainzilber M, Nilsson AS, Sieber BA,
Grigoriou M, Kilkenny C, Salazar-Grueso E, Pachnis V,
Arumae U, Sariola H, Saarma M and Ibanez CF. (1996).
Nature, 381, 785 ± 789.
Tsuchida T, Ensini M, Morton SB, Baldassare M, Edlund T,
Jessell TM and Pfa€ SL. (1994). Cell, 79, 957 ± 970.
Tsuzuki T, Takahashi M, Asai N, Iwashita T, Matsuyama M
and Asai J. (1995). Oncogene, 10, 191 ± 198.
Ullrich A and Schlessinger J. (1990). Cell, 61, 203 ± 212.
Van-der-Geeer P, Hunter T and Lindberg RA. (1994). Annu.
Rev. Cell Biol., 10, 251 ± 337.
Weinberg ES. (1992). Results Probl. Cell Di€er., 18, 91 ± 150.
Wester®eld M. (1994). The zebra®sh book. Univ. Oregon
Press: Eugene.
Wilkinson D. (1992). In situ hybridization, a practical
approach: Whole mount in situ hybridization of vertebrate
embryos. Wilkinson D (ed). Oxford University Press:
Oxford, pp. 75 ± 83.
Xu Q, Holder N, Patient R and Wilson SW. (1994).
Development, 120, 287 ± 299.