Full text of DOI:10.1016/S0925-4773(02)00343-X

Mechanisms of Development 119 (2002) 91–108
www.elsevier.com/locate/modo
Fgf8 and Fgf3 are required for zebrafish ear placode induction,
maintenance and inner ear patterning
*
´
Sophie Leger, Michael Brand
Max-Planck-Institute of Molecular Cell Biology and Genetics (Dresden), Pfotenhauerstrasse 108, 01307 Dresden, Germany
Received 13 March 2002; received in revised form 9 July 2002; accepted 3 September 2002
Abstract
The vertebrate inner ear develops from initially ‘simple’ ectodermal placode and vesicle stages into the complex three-dimensional
structure which is necessary for the senses of hearing and equilibrium. Although the main morphological events in vertebrate inner ear
development are known, the genetic mechanisms controlling them are scarcely understood. Previous studies have suggested that the otic
placode is induced by signals from the chordamesoderm and the hindbrain, notably by fibroblast growth factors (Fgfs) and Wnt proteins. Here
we study the role of Fgf8 as a bona-fide hindbrain-derived signal that acts in conjunction with Fgf3 during placode induction, maintenance
and otic vesicle patterning. Acerebellar (ace) is a mutant in the fgf8 gene that results in a non-functional Fgf8 product. Homozygous mutants
for acerebellar (ace) have smaller ears that typically have only one otolith, abnormal semi-circular canals, and behavioral defects. Using
gene expression markers for the otic placode, we find that ace/fgf8 and Fgf-signaling are required for normal otic placode formation and
maintenance. Conversely, misexpression of fgf8 or Fgf8-coated beads implanted into the vicinity of the otic placode can increase ear size and
marker gene expression, although competence to respond to the induction appears restricted. Cell transplantation experiments and expression
analysis suggest that Fgf8 is required in the hindbrain in the rhombomere 4–6 area to restore normal placode development in ace mutants, in
close neighbourhood to the forming placode, but not in mesodermal tissues. Fgf3 and Fgf8 are expressed in hindbrain rhombomere 4 during
the stages that are critical for placode induction. Joint inactivation of Fgf3 and Fgf8 by mutation or antisense-morpholino injection causes
failure of placode formation and results in ear-less embryos, mimicking the phenotype we observe after pharmacological inhibition of Fgf-
signaling. Fgf8 and Fgf3 together therefore act during induction and differentiation of the ear placode. In addition to the early requirement for
Fgf signaling, the abnormal differentiation of inner ear structures and mechanosensory hair cells in ace mutants, pharmacological inhibition
of Fgf signaling, and the expression of fgf8 and fgf3 in the otic vesicle demonstrate independent Fgf function(s) during later development of
the otic vesicle and lateral line organ. We furthermore addressed a potential role of endomesomerm by studying mzoep mutant embryos that
are depleted of head endomesodermal tissue, including chordamesoderm, due to a lack of Nodal-pathway signaling. In these embryos, early
placode induction proceeds largely normally, but the ear placode extends abnormally to midline levels at later stages, suggesting a role for the
midline in restricting placode development to dorsolateral levels. We suggest a model of zebrafish inner ear development with several
discrete steps that utilize sequential Fgf signals during otic placode induction and vesicle patterning. q 2002 Elsevier Science Ireland Ltd.
All rights reserved.
Keywords: Sensory organ; Otic placode; Otic vesicle; Inner ear; Hair cells; Lateral line; fgf8; fgf3; Acerebellar; Fibroblast growth factors; Induction; Zebrafish;
Danio rerio
1. Introduction
during early somitogenesis stages (Fritzsch et al., 1998;
Baker and Bronner-Fraser, 2001; Whitfield et al., 2002).
Transplantation studies suggested that the placode is
induced by a signal from neighbouring hindbrain (Harrison,
1935; Stone, 1931; Waddington, 1937; Woo and Fraser,
1998; Yntema, 1955), but the nature of the inducer(s) was
controversial (Chisaka et al., 1992; Deol, 1964; McKay et
al., 1996). Embryological and molecular evidence
suggested that successive waves of inducing signals overlap
in time and/or space during ear induction (Jacobson, 1966;
Groves and Bronner-Fraser, 2000; Baker and Bronner-
Fraser, 2001). Previous studies suggest that both the hind-
The vertebrate inner ear contains the main sensory appa-
ratus for detection of sound and gravitational stimuli. It
develops from the otic vesicle or otocyst, and much of its
structural complexity originates at early developmental
stages. In the embryo, the otocyst forms from the otic
placode, an ectodermal thickening adjacent to the hindbrain
* Corresponding author. Tel.: 149-351-210-2514; fax: 149-351-210-
1359.
E-mail address: brand@mpi-cbg.de (M. Brand).
0925-4773/02/$ - see front matter q 2002 Elsevier Science Ireland Ltd. All rights reserved.
PII: S0925-4773(02)00343-X

´
S. Leger, M. Brand / Mechanisms of Development 119 (2002) 91–108
92
brain and the mesoderm can induce an otocyst, but that the
otocyst will not differentiate unless hindbrain is adjacent to
it for a critical time period (Harrison, 1935, 1945; Jacobson,
1963a, 1966; Yntema, 1950, 1955). Furthermore, no single
gene is known that gives rise, when mutated, to an ear
placode-less phenotype either in mouse (Steel, 1995) or in
zebrafish (Malicki et al., 1996; Whitfield et al., 1996, 2002),
consistent with the involvement of multiple events during
placode formation.
Although the evidence is strong that a signal from hind-
brain is involved in induction (reviewed in Van de Water
and Represa, 1991; Torres and Giraldez, 1998; Baker and
Bronner-Fraser, 2001), this signal has been difficult to find.
Genetic evidence for hindbrain factors regulating otic devel-
opment comes from mouse (Deol 1966; Frohman et al.,
1993; Cordes and Barsh, 1994;) and zebrafish mutants
(Malicki et al., 1996; Whitfield et al., 1996). Mouse mutants
for Hoxa-1, pax3 or kreisler, among others, and zebrafish
mutants such as valentino, mindbomb, snakehead, otter,
fullbrain and spiel-ohne-grenzen show primary defects
affecting the hindbrain and associated inner ear defects
(Chisaka et al., 1992; Cordes and Barsh, 1994; Dolle et
al., 1993; Epstein et al., 1991; Lufkin et al., 1991; Mark et
al., 1993; Moens et al., 1996, 1998; Burgess et al., 2002).
The hindbrain is therefore believed to influence the devel-
opment of the inner ear either directly or indirectly.
vertebrates (Haddon and Lewis, 1996; Fig. 1). Shortly after
it becomes visible around 16 h of development, the ear
placode forms the ovoid-shaped otocyst by cavitation.
Within it, sensory patches with overlying otoliths form at
each pole which then develops into sensory maculae
containing numerous mechanosensory hair cells. Neuronal
precursors delaminating from the ventral aspect of the
otocyst form the statoacoustic (VIIIth) ganglion. In certain
positions, the otocyst wall grows inwards forming epithelial
protrusions that fuse at their tips, thus subdividing the vesi-
cle and initiating the formation of the semicircular canals.
The small islands of hair cells, or cristae, which form at the
base of the semicircular canals are thought to detect angular
acceleration. After about 1 week, all major components of
the inner ear are present.
The optical clarity and experimental accessibility of
In several vertebrates, fibroblast growth factor 3 (Fgf3) is
expressed in rhombomeres (rh) adjacent to the site of initial
ear placode formation (Wilkinson et al., 1988; Mahmood et
al., 1995; McKay et al., 1996; Lombardo et al., 1998; Phil-
lips et al., 2001; Raible and Brand, 2001; Maroon et al.,
2002). Fgfs in general function in several important cell
interaction and inductive events. Fgf3 is expressed in the
hindbrain and is able to induce ectopic formation of vesicles
expressing some otic markers (Vendrell et al., 2000). More-
over, antisense oligonucleotide inhibition of Fgf3 caused
defects in otic vesicle formation but not placode induction
(Represa et al., 1991), and Fgf2 or Fgf3-coated beads can
elicit formation of otic vesicles at early neural plate stages in
Xenopus (Lombardo and Slack, 1998; Lombardo et al.,
1998). Target genes for Fgf-signaling, like sprouty2, erm
and pea3, are expressed in the otic placode during induction
stages and require Fgf signaling for their expression (Cham-
bers and Mason, 2000; Raible and Brand, 2001). However, a
‘knock-out’ of Fgf3 in the mouse causes only a mild ear
phenotype, variable disruption of the endolymphatic duct
(Mansour et al., 1993; McKay et al., 1994). These studies
suggested that Fgf3 might be dispensable during early inner
ear induction and positioning. More recent evidence impli-
cated chick Fgf19 signaling in otic placode induction, which
may signal from the paraxial mesoderm adjacent to the
hindbrain or from the hindbrain primordium itself (Ladher
et al., 2000), raising the possibility that Fgf3 might mimic
the action of Fgf19 or other Fgfs in misexpression assays.
Morphologically, early development of the zebrafish
inner ear is very similar to inner ear development of other
Fig. 1. Morphological development of the inner ear requires ace/fgf8.
Lateral views (anterior to the left, dorsal to the top) of the developing
inner ear in wild-type and ace live embryos at different developmental
stages (as indicated). (o) otolith, (asc) protrusion of the anterior, (lsc)
lateral, and (psc) posterior semicircular canal; (ac) anterior crista, (lc)
lateral crista, (pc) posterior crista, (am) anterior macula.

´
S. Leger, M. Brand / Mechanisms of Development 119 (2002) 91–108
93
zebrafish embryos has allowed the isolation of numerous
mutants affecting different aspects of development or func-
tion of the inner ear (Malicki et al., 1996; Riley and Grun-
wald, 1996; Whitfield et al., 1996). Only a few of these
mutants affect the initial formation of the otic vesicle. Acer-
ebellar (ace) is a loss-of-function mutant in the gene encod-
ing the Fgf8 signaling protein, and affects morphology of
the inner ear (Brand et al., 1996; Reifers et al., 1998). Here
we study the activity and function of Ace/Fgf8 and Fgf3 in
detail, and of Fgf signaling more generally, in development
of the inner ear. Our work partially confirms and extends the
results of two other independent studies (Phillips et al.,
2001; Maroon et al., 2002). Similar to these studies, we
find that pharmacological inhibition and morpholino-anti-
sense-induced knock-down (Nasevicius and Ekker, 2000) of
fgf3 and fgf8 in wild-type and ace mutant backgrounds
reveals that Fgf3 and Fgf8 are redundantly required for
early ear placode formation. We furthermore find that this
requirement includes onset of pax8 expression as the earliest
known marker of otic development, as well as other
markers. In addition, we provide evidence from cell trans-
plantation studies showing that Fgf8 is required in the hind-
brain including rhombomere 4, a site of fgf8 and fgf3
expression, in close neighbourhood to the forming placode.
Consistent with the results of our transplantation experi-
ments, and differing from the results obtained by Phillips
et al. (2001), we find that placode induction can proceed
largely normally in the absence of cephalic endomesoderm
in embryos lacking Nodal-signaling. We also find that Fgf8
is sufficient for normal induction of the ear placode in a
subset of the ectoderm, and that Fgf’s function indepen-
dently of early induction again during later stages in pattern-
ing the otic vesicle. We suggest that Fgf8 and Fgf3 are bona-
fide hindbrain-derived inducers of inner ear development
and differentiation that induce otic placode formation and
control patterning of the otic vesicle at later stages.
ing that they might result from the fusion of the otoliths (see
below). The epithelial protrusions subdividing the vesicle
into semicircular canals (scc) are abnormal in 40% of ace2
ears (n ¼ 120); they usually arise in abnormal positions and
some protrusions are missing. The protrusions often fail to
elongate and fuse (Figs. 1E–H). However, ears with only
one otolith can develop a relatively normal scc system and
vice versa, suggesting independent functions for Ace/Fgf8
in different parts of the inner ear. Towards later stages of
development, ace2 ears become increasingly normal
compared to the wild-type, although the morphological
defects always remain apparent (see Figs. 1G,H, and
below).
2.2. Abnormal behavior of ace/fgf8 mutants
Abnormalities of the vestibular system are often linked to
abnormal motor behavior, and indeed the morphological
defects of ace2 larvae are associated with abnormal beha-
vior. On day 5 of development, wild-type larvae have
straight tails, inflated swim bladders, and swim dorsal side
up. Sibling mutant larvae typically have a slightly undulated
tail and swim or lie on their sides. In addition, ace2 larvae
react abnormally to tactile stimuli: wild-type embryos, upon
touching the head or tail with a blunt glass capillary, move
straight for at least one full body length, or dash off alto-
gether. In contrast, 80% of the ace2 larvae (n ¼ 54) do not
escape but rather circle on the spot. Such circling behavior
often reflects a dysfunction of the vestibular system (Nicol-
son et al., 1998). In addition, ace2 larvae appear less sensi-
tive to stimulation on the head than on the tail, whereas
wild-type larvae react to both stimuli with a similar escape
response. In response to a vibrational stimulus (gentle
tapping on the rim of the petri dish) ace2 larvae hardly
react: 80% (n ¼ 48) fail to swim away from the position
of the vibrational stimulus as wild-type larvae normally do.
A total of 38% of ace2 embryos (n ¼ 48) respond only by
half a turn, and 43% (n ¼ 48) do not move at all. Given the
absence of the cerebellum and abnormal brain development
in ace2, these behavioral defects probably have multiple
origins, but we note that they are also consistent with the
defects in the auditory-vestibular and lateral line system
described below.
2. Results
2.1. Morphological development in the inner ear requires
ace/fgf8
In wild-type embryos, the otic placode becomes first visi-
ble around 16 h of development as a thickening of the ecto-
derm which then cavitates to generate the otic vesicle
containing two otoliths in stereotype locations at the oppo-
site end of the vesicle (Figs. 1A,C). Ear development in
homozygous ace mutants (ace2) is morphologically abnor-
mal from the beginning, since the size of the otic placode
and vesicle is variably reduced to about half the wild-type
size (Fig. 1B; Brand et al., 1996). At 28 h, 50% of ace2 ears
(n ¼ 157) have only one otolith, and even if both are
present, they are typically misplaced and very close to or
even touching each other (Figs. 1D,F). In 24% of the
mutants, the single otoliths are of abnormal shape, suggest-
2.3. fgf8 expression in the otic vesicle
We examined fgf8 expression by in situ hybridization
(ISH) during ear development, and compared it to expres-
sion in ace2 ears. Importantly, fgf8 is initially not expressed
in the forming placode itself (see Section 2.4). In the otic
vesicle, expression is first observed from 18 h of develop-
ment onwards. fgf8 is expressed in an anterior patch from
which the anterior macula will develop and more weakly
and transiently, at the posterior-medial pole (Figs. 2A,B).
Anterior expression is initially normal in ace2 compared to
wild-type otic vesicles. Posterior expression is extinguished
in wild type vesicles at 24 h, but remains detectable in some

´
S. Leger, M. Brand / Mechanisms of Development 119 (2002) 91–108
94
and although fgf8 is not expressed in the forming placode
itself, it is expressed during gastrulation and early somito-
genesis in close proximity to it. From 70% epiboly onwards,
fgf8 is initially expressed throughout the anterior hindbrain
and becomes then restricted during early somitogenesis to
rhombomere 4 (r4), ventral r2, r1 and the midbrain-hind-
brain-boundary (MHB) (Fu¨rthauer et al., 1997; Reifers et
al., 1998). In addition, fgf8 is expressed in the mesodermal
heart field underlying the ectoderm just anterior to the site of
ear placode formation (Reifers et al., 2000a). Because fgf8 is
not expressed in the otic placode prior to the 18 somite
stage, and because placode size is reduced in ace2 embryos,
this suggested that fgf8 could induce placode formation
from surrounding tissue. We therefore studied marker
gene expression by ISH that reveal the early events of
placode formation prior to morphological differentiation
Fig. 2. fgf8 expression in the otic vesicle. Lateral views of wild-type and
ace2 otic vesicles at different developmental stages (as indicated). (am)
anterior macula, (ac) anterior crista, (lc) lateral crista, (pc) posterior crista,
(*) protrusions of the semicircular canals.
ace2 vesicles until 28 h (Fig. 2D), giving the vesicle a more
symmetric appearance in the mutants. Persistence of poster-
ior pole expression may be due to a general failure in feed-
back regulation in ace2 mutants (Reifers et al., 1998;
Shanmugalingam et al., 2000; Fu¨rthauer et al., 2001). At
30 and 48 h of development, fgf8 upregulation is also visible
in an apparently increased number of cells in the anterior
patch (Figs. 2E–H). From 48 h onwards, fgf8 is expressed
somewhat more strongly in the lumenal cell layer of the
anterior macula containing the hair cells, and in the cristae
and the epithelial protrusions of the scc system (Figs. 2G–J).
The dynamic and spatially ordered expression of fgf8 and
the defects observed in ace2 suggests that fgf8 might func-
tion during several distinct steps of otic vesicle differentia-
tion.
Fig. 3. Induction of the otic placode marker gene expression requires ace/
fgf8. (A–J) Expression of early otic markers in the otic primordium
(bracket) of wild-type and ace embryos (markers as indicated). Dorsal
views, anterior to the left, except for K, L and O, P (lateral views). (K,
L) Acridine orange staining does not detect increased numbers of dying
cells in the otic primordium (arrow) of ace mutants. (M, N) Staining with an
anti-phosphohistone antibody recognizing mitotic cells (in brown) and ISH
to pax8 to visualize the otic primordium; a similar number of mitotic cells is
apparent. (O–R) Expression of pax8 in the wild type and in the ace2 otic
primordium (bracket) at the indicated stages. (a) animal pole, (v) vegetal
pole, (d) dorsal side. (Q–X) Position of the otic placode relative to rhom-
bomeres forming in the hindbrain; double ISH with a placode marker (pax8
or pax2.1, as indicated) and a rhombomere marker (Kroxr20 or fgf8, as
indicated). (r3), (r4) and (r5), rhombomeres 3, 4 and 5, respectively.
2.4. Induction of the otic placode requires ace/Fgf8
On a gastrula-stage fate map, the otic primordium arises
adjacent to the anterior hindbrain (Kozlowski et al., 1997),

´
S. Leger, M. Brand / Mechanisms of Development 119 (2002) 91–108
95
(Fig. 3) and performed RNA injections, Fgf8-bead implan-
tations (Fig. 6), cell transplantations (Fig. 7), pharmacolo-
gical inhibition of Fgf-signaling (Fig. 9) and morpholino-
inactivation of fgf8 (Fig. 10); together, these studies provide
evidence that Fgf8 acts as a placode inducer acting from the
hindbrain primordium.
phospho-histone antibody recognizing mitotic cells at tail-
bud, 5, 10 and 20 somite-stage, and at 24 h, and find no
difference to wild-type embryos. Although difficult to quan-
titate, the amount of dying and of mitotic cells in the otic
primordium of ace mutants appears similar to that in wild-
type siblings (Figs. 3K–N). Once the placode is formed, its
cells appear to be of normal size in the mutants (not shown).
Together with the gene expression data, this suggests that
the defect in ace mutants is due to a failure in otic placode
induction.
pax2.1 (Krauss et al., 1991), dlx3 (Ekker et al., 1992),
pax8 (Pfeffer et al., 1998), eya1 (Sahly et al., 1999) and
six4.1 (Kobayashi, 2000) are among the first genes to be
specifically activated or upregulated in the otic primordium
from late gastrulation stages onwards. The otic placode
forms from the primordium by thickening within the dlx3,
eya1 and six4.1-expressing stripe bordering on the neural
plate (Akimenko et al., 1994; Sahly et al., 1999; Kobayashi,
2000). In ace mutants, the stripe is unaffected, but expres-
sion of all these markers in the placode arising from it
occurs in a smaller territory that is at most half the size of
the wild-type placode (Figs. 3A–J). pax8 may be the earliest
marker for ear placode formation (Pfeffer et al., 1998), and
is expressed in the otic primordium from 85 to 90% epiboly
onwards. pax8 expression is initially located adjacent to
much or all of the hindbrain primordium, roughly corre-
sponding to fate map position of the inner ear placode.
Subsequently, starting at 95% epiboly, pax8 expression
becomes progressively restricted to the placodal area
(Figs. 3O–R). Double-ISHs with pax2.1, fgf8, pax8 and
krox20 probes (Figs. 3Q–X) show that otic placode devel-
opment, as monitored by pax8 expression, is initiated imme-
diately adjacent to the fgf8-expressing cells in the hindbrain
primordium of both wild-type and ace2 embryos, covering
an area approximately adjacent to r2–6 (Figs. 3Q–T).
Within the domain of pax8 and pax2.1 positive cells, higher
levels of expression then develop next to r4 and r5 (Figs.
3S–V), and by the 8-somite stage, expression of pax2.1 is
concentrated next to r4 and r5 (Figs. 3U,V). Thus, during
gastrulation stages, placodal development starts next to
most of the hindbrain posterior to r2, and then becomes
progressively restricted during early somitogenesis stages
to an area next to r4–6, and thus closely follows fgf8 expres-
sion during these stages. During later somitogenesis stages,
the placode is located predominately next to r5 and r6 (Figs.
3W,X), perhaps due to an anterior shift of the hindbrain that
has been described previously (Moens et al., 1996). In ace2
mutants, placodal marker expression is initially reduced
throughout, then becomes reduced to about halve the normal
size transiently next to r4, and expression then persists next
to r5 (Figs. 3W,X). pax8 expression is already slightly
reduced in the otic primordium of ace mutants at its onset
during late epiboly stages, and more strongly so from tail-
bud stage onwards (Figs. 3O–T).
2.5. Requirement for ace/fgf8 during otic vesicle
differentiation and neurogenesis
The morphological defects of ace2 otic vesicles and the
fgf8 expression pattern suggested that fgf8, in addition to its
function during placode induction, might also function
during otic vesicle differentiation. Different parts of the
otic vesicle at 20–24 h of development are altered in ace2
embryos. pax5 expression marks an anterior-medial domain
of the otic vesicle, which is reduced in ace2, as is mshD
expression in the dorsal vesicle (Figs. 4A,B,E,F; Ekker et
al., 1992). otx1 (Li et al., 1994), gsc (Thisse et al., 1994) and
zdk1 (not shown) mark a ventromedial or posterior domains
that are almost eliminated in ace mutants (Figs. 4I,J,M,N).
Fig. 4. Requirement of Fgf8 and Fgf3 during otic vesicle differentiation and
neurogenesis. (A–P) Expression of marker genes for different parts of the
otic vesicle at 24 hpf in wild-type and ace embryos and wild type embryos
injected with Mo-fgf3 or treated with the SU5402 inhibitor from the 18-
somite stage to 24 hpf, as indicated. Dorsal views except for (E–H), lateral
views (anterior to the left). (U, V) Toluidin-blue stained parasagital section
of a wild-type (Q) and an ace (R) embryo at 36hpf, taken at the level of the
eighth ganglion (VIII). (Q–T, W–Z) Expression of neurogenic markers (as
indicated) in the vesicle and in the eighth ganglion (*) at 24 hpf in wild-type
and ace embryos and wild type embryos injected with Mo-fgf3 or treated
with the SU5402 inhibitor from the 18-somite stage to 24 hpf, as indicated.
Lateral views (anterior to the left).
Because Fgf8 might act as a survival factor or mitogen on
placodal cells, marker gene expression might be indirectly
reduced in ace mutants through cell death or altered prolif-
eration of placodal cells. We therefore examined ace2
embryos using acridine orange to detect dying cells between
70% of epiboly and the 6-somite stage, and with an anti-

´
S. Leger, M. Brand / Mechanisms of Development 119 (2002) 91–108
96
Thus, in spite of the reduced vesicle size, many aspects of
differentiation proceed normally in ace2 vesicles on a
reduced scale; ventromedial (otx1-positive) and posterior
vesicle development may be somewhat more strongly
affected than other vesicle parts.
Neurons of the stato-acoustic (VIIIth) ganglion innervat-
ing the hair cells, detected by time-lapsing or using a
pax2.1-GFP transgenic line, delaminate from the ventral
wall of the otic vesicle mainly between 22 and 42 h of
development, where they collect underneath the otic epithe-
lium (Haddon and Lewis, 1996; Picker et al., 2002). Devel-
opment of the eighth ganglion is affected in ace2 mutants.
In parasagittal sections of 36 h wild-type and ace embryos,
ace2 otic vesicles are reduced in size and the eighth gang-
lion is much smaller than in wild-type vesicles (Fig. 4Q,R).
neurogenin1 is a marker for early neuronal development
(Korzh et al., 1998). In wild-type embryos neurogenin1 is
expressed in the ventral vesicle wall, the forming VIIIth
ganglion, and in unidentified cells posterior to it; expression
in the VIIIth ganglion is strongly reduced in ace2 embryos,
apparently labelling fewer cells (Figs. 4Q,R). Other markers
of early neurogenesis of the eighth ganglion are similarly
reduced, like snail2 (Thisse et al., 1995), six4.1 and nkx5.1
(Figs. 4W–Z; Adamska et al., 2000). We suggest that the
reduced ventral portion of the ace2 otic vesicle secondarily
leads to the smaller size of the eighth ganglion, due to a
smaller domain in which neurogenesis can occur.
Fig. 5. Hair cell organization is affected in ace/fgf8 mutants. (A–D) Expres-
sion of msxC in the developing cristae at 48 hpf in wild-type (A) and ace (B,
C) embryos and in wild type embryos injected with Mo-fgf3 (C) or treated
with the SU5402 inhibitor from 24 to 48 hpf (D). Lateral views. (E–G)
Fluorescent phalloidin staining of the actin-rich stereocillia of the hair cells
at day 5 of development in wild-type (E) and ace2 (F, G) inner ears. Lateral
views obtained by projection of several optical sections to show all the hair
bundles of the hair cells. Each hair bundle corresponds to one hair cell. (F)
weak ace2 phenotypes. (G) strongest ace2 phenotype found; see text. (ac)
anterior crista, (lc) lateral crista, (pc) posterior crista, (am) anterior macula,
(mm) medial macula, (m) macula, (*) epithelial protrusions of the semi-
circular canals, (o) otolith. (H–K) DASPEI staining of the hair cells of the
lateral line neuromasts at day 5 of development in wild type (H, I) and ace2
(J, K) embryos. (I, K) closer view of the neuromast indicated by an arrow in
(H) and (J), respectively.
2.6. Hair cell organization is affected in ace2/fgf8 mutants
The behavioral defects, abnormal otolith formation and
the fgf8 expression suggested a possible role for Ace/Fgf8 in
the development of mechanosensory hair cells. msxC is a
homeobox gene marking the developing cristae epithelium
(Ekker et al., 1992), and msxC expression in particular of the
lateral crista is strongly reduced in ace2 ears (Figs. 5A,B).
We also observed reduction of pax5 expression in the ante-
rior macula, and of zdk1 expression in the posterior macula
of ace2 ears at 24 and 48 h (Figs. 4A,B and data not shown).
To analyze hair cell development in ace2 mutants, we
stained the actin-rich stereocilia of the hair cells with fluor-
escent phalloidin (FITC-phalloidin) on day 5, and analyzed
them by confocal microscopy. In ace2 mutants the sensory
patches of the ear are misplaced, especially the three cristae,
probably reflecting the distorted morphology of the vesicle
(Figs. 5E–G). Also, the number of hair cells in ace2 ears is
variable, but always reduced: on day 5, the wild-type ante-
rior macula contains on average 75 ^ 3 (n ¼ 2) hair cells,
whereas ace2 ears contain 55 ^ 12 (n ¼ 8). Typically, ace2
ears with fewer hair cells also have only one otolith that is
always either anterior or medial, but never posterior in posi-
tion. Fig. 5G shows the strongest ace phenotype we
observed: this ear has only one otolith, and the number of
hair cells of the lateral and posterior cristae is strongly
reduced. Moreover, there is only one macula with about
40 hair cells which is located in the medial part of the
otocyst, but spreads more anteriorly than the normal medial
macula, probably reflecting a fusion of the anterior and the
medial macula. The single otolith appears composed of two
parts, which may reflect a partial fusion, since otoliths form
owing to and in association with the maculae (Riley et al.,
1997). In conclusion, positioning of the sensory patches,
hair cell number and otolith development are variably
abnormal in ace2 mutants, probably reflecting a mixture
of direct and indirect effects of the lack of fgf8 on otic
vesicle and sensory hair cell differentiation.
Placodally-derived mechanosensory hair cells similar to
those of the inner ear also occur in the neuromasts of the
lateral line, suggesting that they may share common devel-
opmental programs indicative of a common evolutionary
ancestry (Northcutt, 1986; Jorgensen, 1989; Platt et al.,
1989). We therefore asked if ace2 larvae have normal
neuromasts by staining the neuromast hair cells with the
fluorescent dye DASPEI (Whitfield et al., 1996) and count-
ing neuromasts on both sides of the body. Compared to
wild-type larvae (25 ^ 0 neuromasts, n ¼ 10 sides), ace2
larvae have strongly reduced numbers of neuromasts
(9.7 ^ 5 neuromasts per side, n ¼ 30 sides); head neuro-
masts appear somewhat less strongly reduced (Figs. 5H,J).
In addition, the number of hair cells per neuromast is vari-

´
S. Leger, M. Brand / Mechanisms of Development 119 (2002) 91–108
97
able, but reduced overall; a representative case is shown in
Figs. 5I,K. To determine if lateral line placode formation is
occurring normally, we examined eya1, six4.1 and nkx5.1 as
markers of the lateral line placode, but expression was unaf-
fected at 24 h in ace2 embryos, and the placode was in a
similar position along the anterior-posterior axis as in the
wild type siblings (not shown; eya1 is also normal at 48 h).
Thus, lateral line placode formation and migration appar-
ently proceeds normally in ace2 mutants, but a later, as yet
unknown step of neuromast development requires fgf8 func-
tion. Consistent with this possibility, the lateral line placode
and migrating primordium expresses the Fgf target gene
pea3 and Fgf-R1 (Mu¨nchberg et al., 1999; Raible and
Brand, 2001; C. Thisse and B. Thisse, personal communica-
tion).
2.7. Fgf8 can expand ear placode territory
Abnormal ear development of ace2mutants clearly has an
early origin during placodal induction stages, and we there-
fore focussed on understanding the role of fgf8 during the
initial, inductive step. Our above analysis indicated that fgf8
is required for induction of a normal sized placode. In order
to test whether fgf8 is also sufficient to specify placodal fate,
we injected fgf8 mRNA into one side of the embryo, or we
implanted Fgf8 protein coated beads prior to placode forma-
tion (Figs. 6A,D). Control injections with lacZ mRNA had
no effect. In 7 of 18 injected embryos, fgf8 mRNA injection
caused expanded pax2.1 expression in the placodal region at
the 12-somite stage. Fgf8 is thus sufficient to stimulate
pax2.1 expression ectopically. However, the ability to turn
on pax2.1 ectopically in response to Fgf8 misexpression is
limited to the ectoderm adjacent to the posterior hindbrain
rhombomeres (Figs. 6B,C). Likewise, Fgf8 beads implanted
at shield stage (n ¼ 3=14 that ended up in the hindbrain/otic
region) were able to induce expanded ear vesicles in wild-
type embryos (Fig. 6E), that expressed sprouty4, a target
gene for Fgf8 signalling, in its normal position at the ante-
rior vesicle pole (Figs. 6F,G). We have however not
observed additional ear vesicles in such embryos, suggest-
ing that Fgf8 may act in conjunction with other signals.
Fig. 6. Fgf8 misexpression expands pax2.1 positive territory in competent
cells next the hindbrain. (A) Co-injection of fgf8 and lacZ mRNAs in wild
type embryos at the 2-cell stage creates chimaeric embryos misexpressing
fgf8 on one side, which are then examined for pax2.1 expression. (B, C)
Expression of pax2.1 (ISH) in the otic primordium (bracket) of a wild-type
and a unilaterally injected embryo; the injected cells are located in the right
halve of the embryo, as revealed by the antibody to b-gal (brown) at the 12-
somite stage, dorsal views. Note the expansion of the pax2.1 positive otic
expression domain on the injected side in (C), indicated by a long bracket,
compared to the uninjected side, or the wild-type embryo in (B). The
midline is indicated by a dashed line. (D) A heparin bead covered with
Fgf8 is implanted into the hindbrain primordium of wild type embryos at
the shield stage, prior to placode formation. The effects are scored on otic
vesicle formation, and on activation of the Fgf target gene sprouty4. (E)
Live embryo at 30hpf that had received an Fgf8 bead; note the enlarged ear
vesicle on the side of the bead (arrow). (F, G) sprouty4 expression (ISH) at
the anterior pole of an enlarged, partially split ear vesicle at the 20-somite
stage. The split was observed in two cases out of three; the enlarged ear
apparently split secondarily, probably due to the mechanical hindrance of
the bead. Two focal planes of the same embryo showing the bead (F, arrow)
and the two lumena of the partially split otic vesicle (G, *). Dorsal views.
MHB, midbrain-hindbrain-boundary.
chimaeric embryos. When wild type cells were located in
the mesoderm, notochord, ectoderm or in the hindbrain
anterior to or posterior to r4–6, no rescue was observed
(Figs. 7C–I). A chimaera was scored as ‘rescued’ when
the size of placodal pax2.1 expression in ace2 mutants
was restored to wild-type size. Using this stringent criterion,
rescue was observed in two chimaeras out of 31 with clones
located in the hindbrain of ace mutants. In both cases, many
(more than about 40) transplanted wild-type cells were
located adjacent to the site of placode formation in hind-
brain rhombomeres 4–6 (Fig. 7J). Two additional clones
with few (,10) cells in r4–6 did not show visible rescue
(Fig. 7I), nor did four clones with cells in the otic placode
itself (Figs. 7E,F). In a crossection of the hindbrain-clone in
Fig. 7J we observed transplanted cells only in the hindbrain
2.8. Fgf8 is required in the adjacent hindbrain for placode
induction
While our mRNA injection and bead implantation experi-
ments showed an expansion of otic territory in response to
Fgf8, they did not allow us to address the mechanisms or the
normal source of Fgf8 signaling. To test more directly
whether Fgf8 emanating from r4 to 6 is responsible for
placode induction, we transplanted wild-type cells into the
hindbrain primordium of ace2 mutants at pregastrula stages.
The resulting chimaeras were stained for pax2.1 as a
placode marker, and with bgal-antibody to detect the loca-
tion of the transplanted cells (Figs. 7A,B). pax2.1 staining is
also reduced at the MHB in the mutants (Reifers et al.,
1998), allowing us to distinguish wild-type and ace2

´
S. Leger, M. Brand / Mechanisms of Development 119 (2002) 91–108
98
neuroepithelium, but not in the underlying endo- or meso-
derm, nor in the otic placode itself (Fig. 7K), although we
are normally able to detect overlapping labeling in the
placode, and observed a clear separation between the
labeled neural tube cells and the placode prior to flattening
of the embryo for the photograph in Fig. 7J. These results
strongly suggest that Fgf8 emanating from r4 to 6 during
early somitogenesis stages acts to induce otic placode devel-
opment.
To address the importance of endomesoderm as a source
for otic inducers further, we examined embryos lacking
cephalic endomesoderm because they are defective in
nodal signaling, due to lack of maternal and zygotic one-
eyed pinhead product, a crucial cofactor in Nodal signaling
(mzoep embryos; Gritsman et al., 1999; Schier and Shen,
2000). We find that in mzoep embryos otic vesicles do form
and express otx1 regionally, although they are typically of
abnormal shape (Figs. 8A,B,M,N). The placode markers
pax8, pax2.1 and dlx3 are activated in normal spatial rela-
tion to the forming rhombomeres stained with krox20 (Figs.
8C–H); however, marker expression is seen at more medial
levels from tailbud stages onwards. Because midline tissue
is absent develops abnormally in these embryos, this may
reflect a repressive influence of the midline on the medio-
lateral extent of the otic primordium. In the early hindbrain
primordium of mzoep embryos, fgf8 and fgf3 are initially
activated properly, and fgf3, but not fgf8 expression is main-
tained in rhombomere 4 (Figs. 8I–L). Consistent with the
results of our transplantation experiments, these findings
show that otic induction can proceed in the absence of
developing endomesoderm.
2.9. Successive requirements for Fgf signaling
To confirm the importance of Fgf signaling in otic induc-
tion, and to resolve temporal aspects, we treated wild-type
embryos for different periods with the pharmacological
inhibitor SU5402, which is thought to block all Fgf receptor
signal transduction (Mohammadi et al., 1997). The pheno-
type of living inhibitor-treated embryos closely resembles
that of ace2 mutants with respect to the developing inner
ear, midbrain-hindbrain boundary and heart (Figs. 9A–C;
Reifers et al., 2000a; Araki and Brand, 2001). We treated
embryos for different periods of development and find that
Fgf signaling is absolutely required for inner ear induction,
and that there is a separate requirement during later differ-
entiation. We used sprouty4 to confirm that the inhibition
was complete (Figs. 9D,E,L,M). Inhibition starting at 70%
epiboly, the tailbud or the 2-somite stage, results in
complete loss or strong reduction of the expression of
pax8, pax2.1 and dlx3 (Figs. 9F–K; Table 1), resulting in
Fig. 7. Fgf8 is required in the adjacent hindbrain for placode induction. (A)
Fluorescently labelled wild-type cells were transplanted into the primordia
of different tissues of ace2 embryos at shield stage, and the ability to rescue
the normal extend of placode formation was assayed by ISH. (B–K) Expres-
sion of pax2.1 (ISH) at the 12-somite stage. (B) wild-type embryo. (C–I)
ace2 chimeras carrying transplanted cells (brown) in the mesoderm. (C), in
the notochord and the trunk (D), in the otic primordium itself (E, F), in the
midbrain and the mesoderm (G) in the posterior hindbrain (H) and in the
midbrain and the hindbrain (few cells) (I). (J) Rescue of the pax2.1 expres-
sion in the otic primordium (bracket) in an ace2 chimera carrying many
transplanted cells (more than 40) in the hindbrain adjacent to the otic
domain. (K) transverse section at the level of the otic primordium in the
chimera shown in (J). Dorsal views, except (C, K), lateral views. (r4)
rhombomere 4, (opl) otic placode, (ntu) neural tube.

´
S. Leger, M. Brand / Mechanisms of Development 119 (2002) 91–108
99
reduction of vesicle size (Figs. 4D,H,L,P,T, 5D and 9L,M;
Table 1), whereas expression of neurogenin1 and snail2 is
not affected (Fig. 4T, and not shown). These observations,
together with the analysis of the expression pattern and the
ace2 phenotype, show that Fgfs in general, and Fgf8 in
particular, function during otic induction and again indepen-
dently during vesicle stages.
2.10. Fgf3 acts together with Fgf8 in ear induction
Fgf inhibition also results in phenotypes that are often
stronger and less variable than the ace2 phenotype, suggest-
ing that other Fgfs are involved in inner ear development in
addition to Fgf8. Consistent with this possibility we find that
fgf3 (Kiefer et al., 1996) is co-expressed with fgf8 from 85 to
90% epiboly onwards in the hindbrain primordium. Expres-
sion is seen initially in a single broad domain in the posterior
hindbrain primordium, that at the 2-somite stage becomes
confined to r4 where it remains expressed until the 18-
somite stage; expression is downregulated by 21 h (Figs.
10A–D; Phillips et al., 2001; Maroon et al., 2002). More-
over, fgf3, fgf8 and fgf17 (Reifers et al., 2000b) are co-
expressed in the anterior macula at vesicle stages (Figs.
10E–H). Consistent with the notion of redundancy, fgf3 is
expressed independently of ace/fgf8 during gastrulation and
somitogenesis stages (Fig. 10D), and fgf3 and fgf17 expres-
Fig. 8. Depletion of cephalic endomesoderm in mzoep embryos devoid of
Nodal signaling still allows normal placode induction. (A, B) Live pheno-
type of mzoep embryos around 24 hpf, lateral views; otic vesicles form in
spite of the distorted anatomy of these embryos. (B) Closer view of a
partially split vesicle shown in (A). (C–H) Early expression of pax8,
pax2.1 and dlx3 in the otic primordium (arrow) of mzoep embryos (stages
and markers as indicated). (D, F, H) Dorsal views of the lateral views shown
in (C, E, G), respectively. (I–L) Expression of fgf8 (I, J) and fgf3 (K, L) in
the hindbrain of mzoep embryos at the tail bud-stage and the 4-somite stage,
as indicated. Dorsal views. (M–O) Otx1 expression in the otic vesicle of
mzoep embryos at 24 h. (M) Lateral view. (N, O) Dorsal views of the same
embryo shown in (M), taken at two different focal planes. Notice that otx1
expression stretches across the midline, showing one large otic vesicle
stretching across the midline, or several interconnected smaller vesicles,
that form in mzoep embryos. (*) lumen of the otic vesicle. (r3), (r4) and (r5)
rhombomeres 3, 4 and 5, respectively, (mhb) midbrain-hindbrain-boundary,
(hb) hindbrain.
very tiny or no ears after 24 h (Figs. 9B,C). pax8 and pax2.1
staining is absent in Fgf-inhibited embryos, and dlx3 is not
upregulated in the otic region. Treatment for different times
between the tailbud- and 18-somite stage reveals that Fgf
signaling is not only required for induction, but also
required to maintain expression of pax8, pax2.1 and dlx3
in the otic placode (Table 1). In contrast, inhibition after the
18-somite stage up to 24 h has no effect on pax2.1 and dlx3
(Figs. 9N–Q); pax8 is downregulated in the otic vesicle after
the 10 somite-stage. Thus, between the 1 and 18 somite-
stage Fgf signaling is critically required for induction and
maintenance of otic placode markers.
To test whether Fgf signaling acts again during the vesi-
cle stage, as the fgf8 expression pattern and the phenotype of
ace2 mutants suggested, we inhibited wild-type embryos
during otic vesicle development, after the 18-somite stage.
sprouty4, pax5, otx1, gsc, msxC and mshD expression is
specifically lost from the otic vesicle, with no apparent
Fig. 9. Pharmalogical inhibition of Fgf signaling with the SU5402 inhibitor
during otic induction and otic vesicle patterning. See also Table 1. (A, K)
Inhibition during placodal induction. (A, C). Live phenotype at 30 hpf of a
wild-type embryo (A) and wild-type embryos inhibited from the tailbud
stage onwards (B, C). Lateral views. (D–K) ISH, expression at the 6-somite
stage of sprouty4 (D, E) and early otic markers (F–K, as indicated) in wild-
type embryos and wild-type embryos inhibited from the tail bud-stage to the
6-somite stage, as indicated. Dorsal views. (L–Q) Inhibition at the vesicle
stage. Expression at 24 hpf of sprouty4 (L, M) and early otic markers (N–Q,
as indicated) in wild-type embryos and wild-type embryos inhibited from
the 18-somite stage to 24 hpf, as indicated. Dorsal views.

´
S. Leger, M. Brand / Mechanisms of Development 119 (2002) 91–108
100
Table 1
Temporal requirement of Fgf signaling
Inhibition at placodal stages
Inhibition at vesicle stages
70–95% tb–4 s 4–6 s
6–9 s
10–15 s 14–18 s
18 s–24 h
24–48 h
Otic marker expression pax8
Absent Absent Absent Absent
Absent Absent Reduced Reduced Weakly reduced Wild type
pax2.1
dlx3
Absent Absent Absent
Reduced Weakly reduced Wild type
pax5; mshD otx1; goosecoid
Absent
neurogenin1; snail2; six4.1
mshC
Wild type
Absent
sion is reduced, but present in ace2 ears at later stages (Figs.
10F,H).
induction of the ear placode, but the nature of the signal
itself, its time of action, whether it serves an inductive or
permissive role, and through which downstream target
genes it acts have been enigmatic. Our results partly confirm
and extend, but also differ in some aspects (see below) from
those of two independent reports by Phillips et al. (2001)
and Maroon et al. (2002), describing redundant functions of
To test the notion of redundant Fgf functions further, we
determined the function of fgf3 in ear development, by
injecting morpholino antisense oligonucleotides (Nasevi-
cius and Ekker, 2000; Mo-fgf3 and Mo-fgf8) into wild-
type and into ace mutant embryos. We find evidence that
fgf3 and fgf8 are both required for placode induction (Figs.
10I–Z, Table 2). Injection of 10 or 15 ng of Mo-fgf3 into
wild-type embryos causes a reduction of both ear size and
gene expression of the otic markers pax8, pax2.1, dlx3,
pax5, mshD, msxC, otx1, goosecoid and neurogenin1, simi-
lar to, and at later stages stronger than that observed for ace
mutants (Figs. 4C,G,K,O,S, 5C,D and 10M–P, and not
shown). At late gastrula stages and 24 h, fgf8 expression
is unaffected in these embryos (Figs. 10Y,Z). In addition,
Mo-fgf3 injection causes absence of all epithelial protru-
sions of the semicircular canals (not shown); further defects
are seen in early forebrain development (Raible and Brand,
2001). Control injections with 15 ng of a four-basepair-
mismatched morpholino, or a random sequence morpholino,
or of morpholinos directed against a number of unrelated
genes, have no such effects (Table 2 and not shown). To test
the idea of a possible redundancy between Fgf8 and Fgf3,
we co-injected embryos with Mo-fgf3 and Mo-fgf8, which
causes either very strong reduction or complete absence of
the ear vesicle (n ¼ 73 absent of 96 injected; of the remain-
ing 23 ears, 21 were reduced to about half the size, and two
were wild type; Table 2, Figs. 10Q,R). The same phenotype
is observed for ace mutants injected with Mo-fgf3 (n ¼ 20
of 20; Table 1; Figs. 9U,V). Expression of pax8 and pax2.1
during placode induction stages is absent or very strongly
reduced under those conditions (n ¼ 33 of 38; Figs.
9S,T,W,X). As expected, and as we observed in inhibitor
treated embryos, the expression of dlx3 and eya1 is not
affected initially, but fails to be maintained after the placode
induction stage in these embryos (not shown). Thus, Fgf3
and Fgf8 redundantly specify placode induction.
Fig. 10. Fgf3 acts together with Fgf8 in otic induction and differentiation.
(A–D) Expression of fgf3 in the hindbrain (*) during placode induction
(stages as indicated) in wild-type (A–D) and ace (D) embryos. (B–D)
Double ISH with krox20 (red). (r3) and (r5) rhombomeres 3 and 5 respec-
tively. (E–H) Expression of fgf3 (E, F) and fgf17 (G, H) in the otic vesicle at
24 hpf in wild-type and ace embryos, as indicated. (I–Z) Morpholinos
injection. Live phenotype at 30 hpf (I–P) and early otic marker expression
(Q–X, stages and markers as indicated) in wild-type and ace embryos, wild-
type embryos injected with Mo-fgf8 and Mo-fgf3 and ace embryos injected
with Mo-fgf3, as indicated. (Y–Z0) fgf8 expression in the hindbrain (Y, Y0)
and in the otic vesicle (Z, Z) is normal in wild type embryos injected with
Mo-fgf3.
3. Discussion
Classical transplantation studies have provided evidence
for the existence for a hindbrain-derived signal involved in

´
S. Leger, M. Brand / Mechanisms of Development 119 (2002) 91–108
101
Table 2
Fgf3 and Fgf8 are required for otic induction
No ear
0
Tiny ears
0
Reduced ear sizea
92b
Wild-type ears
0
Total number
Mo-fgf3
15 ng Mo-fgf3
92
92
Mo-fgf3-MM
Mo-fgf3 1 Mo-fgf8
15 ng
0
0
0
92
10 ng Mo-fgf3 1 4 ng Mo-fgf8
15 ng Mo-fgf3 1 8 ng Mo-fgf8
0
n.a.c
41
32
21
0
2
0
64
32
ace2 1 Mo-fgf3
10 ng Mo-fgf3
10
10
0
0
20
a
Size as seen in ace mutants.
All ears had two otoliths.
b
c
In these injections more than 50% of the embryos do not develop to the ear vesicle stage. When assayed for pax2.1 expression at eight somites, 18 of 21
embryos had no pax2.1 expression in the ear placode. Control injections with a four pair-base mismatch morpholino against fgf3 or fgf8 (Araki and Brand,
2001), and a random sequence morpholino have no effect on ear development.
Fgf8 and Fgf3 in otic induction. Our work provides consis-
tent evidence from expression studies, cell transplantations,
gain-of-function experiments, loss-of-function analysis of
acerebellar/fgf8 mutants, pharmacological inhibition of
Fgf-signaling and morpholino-inactivation of fgf3 and
fgf8. This evidence suggests that Fgf8 and Fgf3 are the
molecules mediating activity of the endogenous hindbrain-
derived ear inducer, acting in a redundant fashion during the
earliest stages of placode induction. Furthermore, we find
that Fgf8 and Fgf3 serve additional function(s) in patterning
and differentiation of the otic vesicle at later stages.
Whereas our results strongly argue that Fgf3 and Fgf8 act
as hindbrain-derived signals, our analysis of mzoep embryos
lacking cephalic endomesoderm does not support a major
role of this tissue in otic induction.
consistent with an involvement of Fgfs as signals specifi-
cally during late-gastrulation to early-somitogenesis stages,
the time of initial otic induction. Importantly, inhibition
during this period leads, in our hands, to complete loss of
otic induction as observed by pax2.1, dlx3 and pax8 expres-
sion, whereas Maroon et al. (2002) report that expression of
pax8 is not affected until the 2–3 somite stage. The reasons
for this different finding is not clear, but we tentatively
suggest a lower effective concentration of SU5402 may
have been employed by these authors. Alternatively, it
may matter whether Fgf signaling is blocked starting at
30% epiboly, as reported by Maroon et al., or from 70%
of epiboly (our work), shortly before onset of the endogen-
ous pax8 expression. We were not able to obtain reliable
development of embryos when FGF inhibition was started at
30%, due to an independent requirement for Fgf signaling in
maintaining early mesoderm development. Phillips et al.
(2001) did not study the effects of SU5402 inhibition. We
also observe complete loss of pax8 expression when fgf3
and fgf8 are knocked-down via morpholino-injection, as is
also reported by Phillips et al. (2001), and, for one embryo,
also by Maroon et al. (2002). Together, these results argue
that already the first stage of placode induction requires
Fgf3 1 Fgf8 signaling. Examination of another target gene
for Fgf8-signaling during gastrulation, gbx2, after SU5402
inhibition further supports this notion: gbx2 is expressed
from tailbud-stage onwards in the otic primordium, only
slightly later than pax8, and is completely dependent on
Fgf signaling (Reim and Brand, 2002; M. Rhinn, K. Lun
and M.B., unpublished). (iv) Our transplantation chimaeras
show that the hindbrain primordium, and in particular r4–6,
is the site where wild-type activity of Fgf8 is needed for
normal otic induction to occur. Fgf8 is also expressed in the
mesendodermal primordium during gastrulation, called the
germring, and during early somitogenesis in the lateral plate
mesoderm and in the heart field, i.e. close to the forming otic
placode (Reifers et al., 1998, 2000a). In accordance with our
findings, grafts of rhombomere 4 indicate that this rhombo-
mere carries the capacity of the hindbrain to induce the inner
3.1. The initial stage of otic induction requires Fgf8 and
Fgf3 signaling from the hindbrain primordium
Transfilter experiments have suggested that the induction
of the inner ear by the neural tube is due to a diffusible
molecule from the neural tube, rather than direct cell-cell
contact (Van de Water and Conley, 1982). Fgf3 and Fgf8 are
thought to encode secreted factors, which fits with their
proposed role to mediate the hindbrain action on inner ear
development. Our analysis of acerebellar/fgf8 and fgf3
function provides clear evidence for an Fgf requirement as
a signal acting from the hindbrain: (i) during placodal induc-
tion, fgf8 and fgf3 are not expressed in the placode itself, but
the expression in the hindbrain during late gastrulation/early
somitogenesis- stages fits the expected timing and tissue
distribution of the signal. (ii) The expression of placode
marker gene expression is reduced to about half the size
when fgf3 or fgf8 are singly inactivated, and completely
lost when both are inactivated. At otic vesicle stages this
results in smaller otic vesicles of the ace mutants or fgf3-
morphants. This clearly demonstrates the requirement for
Fgf3 and Fgf8 at the appropriate stage for induction. (iii)
More generally, our SU5402 inhibition experiments are

´
S. Leger, M. Brand / Mechanisms of Development 119 (2002) 91–108
102
ear (Sechrist et al., 1994). Transplantation experiments in
chick and amphibians suggested that endomesoderm also
has otic inducing capacity in the absence of hindbrain tissue
(reviewed in Baker and Bronner-Fraser, 2001). For instance,
in amphibian embryos cardiac mesoderm can act as a source
for otic placode specifying signals (Harrison, 1935).
However, in our transplantation chimaeras, wild type cells
located in the mesoderm did not rescue the otic induction
defect in ace mutants, although it remains formally possible
that a larger number of wild-type cells would be needed in
the mesoderm to achieve rescue. However, although fgf8 is
expressed in the heart field, we do not detect expression
there before the 3-somite stage (Reifers et al., 2000a), i.e.
after the onset of ear placode induction. The Fgf8 signal
responsible for otic specification around the tail bud stage
is therefore unlikely to come from the lateral plate meso-
derm. Harrison’s observation could however be explained
by assuming that heart tissue, due to its Fgf8 expression, can
under certain conditions mimic the activity of the endogen-
ouse hindbrain-derived Fgf8, in analogy to what was
suggested to explain the ability of olfactory placode tissue
(an unlikely endogenous inducing tissue) to induce limb
formation (Slack, 1995).
supports the notion derived from our transplantation analy-
sis, that mesendoderm is unlikely to be a major source for
the otic inducing signal missing in acerebellar mutants or
fgf3 1 fgf8 singly- or doubly-inhibited embryos. The endo-
mesodermal primordium may however indirectly influence
the formation of the otic primordium, e.g. by controlling
patterning of the gastrula neuroectoderm.
3.3. Fgf3 as a redundant signal with Fgf8
In spite of the important and early function of Fgf8 in ear
induction, acerebellar mutants do have an, albeit smaller,
ear, and placodal marker gene expression is reduced to
about 50% of the size of the wild type placode. Why?
Importantly, blockade of all Fgf signaling by SU5402 inhi-
bition results in the formation of tiny ears, but more typi-
cally of no ear at all, suggesting that additional Fgfs are
important for otic induction. As reported by Phillips et al.
(2001) and Maroon et al. (2002), we find that Fgf3 acts
redundantly with Fgf8. fgf3 is also expressed in hindbrain
rhombomere 4 from 85 to 90% of epiboly onwards, and its
expression is normal in ace mutants. Fgf3 is therefore avail-
able to partially compensate the lack of Fgf8 in ace mutants,
and the results of the joint depletion of Fgf8 and Fgf3
suggests this is indeed the case.
3.2. Otic induction and fgf3 and fgf8 expression occur
normally in embryos depleted of endomesoderm
In other species Fgf3 (int-2) is involved in inner ear
development, but its precise role is unclear. Expression
studies in mouse, chicken and Xenopus demonstrate the
presence of fgf3 in the hindbrain, close to the otic domain
at early somite stages (Wilkinson et al., 1988; Tannahill et
al., 1992; Mahmood et al., 1995); our studies on fgf3 expres-
sion confirm this notion for the zebrafish gene. Functional
studies for fgf3 have however given conflicting results.
Consistent with a requirement for Fgf3, Hoxa-1 (Hox1.6)
mutant mice show a delayed formation and gross morpho-
logical alteration of the otic vesicle (Chisaka et al., 1992;
Dolle et al., 1993; Lufkin et al., 1991), and also have
reduced Fgf3 expression (Mark et al., 1993; McKay et al.,
1994). Moreover, Fgf3 antisense oligonucleotides and anti-
Fgf3 antibody inhibit the formation of chick otic vesicles
(Represa et al., 1991). More recently, overexpression in
chick showed that fgf3 is capable of mimicking the activity
mediating early induction (Vendrell et al., 2000). Concei-
vably, however, the antibody used in this study might also
inhibit other members of the Fgf family. Furthermore,
targeted disruption of fgf3 in mice does not lead to absence
or reduction of the otocyst, but rather affects the differentia-
tion of the otocyst with incomplete penetrance (Mansour et
al., 1993). One possibility is that in gain-of-function situa-
tions, Fgf3 mimics all or part of the activity of another Fgf.
Our studies on Fgf3 function are consistent with a require-
ment for Fgf3 in otic induction, but with an important differ-
ence. As previously suggested (Kinoshita et al., 1995;
Mansour et al., 1993) also for other tissues (Fu¨rthauer et
al., 2001; Raible and Brand, 2001; Reifers et al., 2000b),
functional redundancy between different Fgf’s is a likely
In zebrafish, several mutants affecting early mesoderm
development, including mutants lacking chordamesoderm,
develop normal otic vesicles, whereas a slight temporal
delay can be observed for others (Mendonsa and Riley,
1999). Specifically, morpholino-inactivation of the Nodal-
cofactor Oep was reported to strongly delay pax8 expression
in the otic primordium in a portion of the embryos to the six-
somite stage (Phillips et al., 2001). Because morpholinos
can cause unspecific side effects, we used in our experi-
ments mzoep embryos to inactivate both the maternal and
the zygotic function of oep genetically by homozygosity for
an oep null allele (Gritsman et al., 1999), which should
formally give the same result as the morpholino injections.
However, in mzoep embryos, which lack cephalic endome-
soderm, we did not observe the strong delay of pax8 expres-
sion reported by Phillips et al. (2001), nor of the other early
placodal markers pax2.1 and dlx3. Instead, we observe a
normal temporal and spatial correlation of their expression
sites, and hence the site of placode formation, in relation to
the hindbrain. Therefore, genetic depletion of cephalic
endomesoderm in mzoep embryos does not suppress otic
induction. In addition, we observe normal expression of
fgf8 and fgf3 in the early hindbrain of mzoep embryos,
which we suggest to be the explanation for normal otic
induction in these embryos. We do however observe a
medial expansion of otic marker expression, suggestive of
a lack of inhibitory signals from the midline, and it will be
interesting to determine if this a direct or an indirect effect
of Nodal signaling. Overall, our analysis of mzoep embryos

´
S. Leger, M. Brand / Mechanisms of Development 119 (2002) 91–108
103
explanation: our results show that the full phenotypic
requirement, visible as absence of otic vesicle formation
and lack of early placodal marker expression, only becomes
apparent when Fgf8 and Fgf3 are both inactivated, either
through mutation, or by co-injection of morpholinos. Given
that the double-inactivated embryos show the same pheno-
type as the SU5402 inhibited embryos, Fgf3 and Fgf8
together probably account for a large portion or all of the
hindbrain-derived placode inducing activity. Interestingly,
expression of Fgf8 in the early hindbrain primordium has
not been reported in mice and chick, raising the possibility
that in these animals another Fgf cooperates with Fgf3 in ear
induction, such as Fgf4, which is expressed in the early
neuroectoderm in chick (Shamim and Mason, 1999). Addi-
tional Fgfs or other signaling molecules may serve addi-
tional functions, for instance Fgf19 might act as a paraxial
mesoderm-derived signal in chick otic placode induction;
the timing of expression from the 6-somite stage onwards
suggests this may be a function following the initial induc-
tion (Ladher et al., 2000). The different phenotypes we
observe following fgf8 inactivation and fgf3 inactivation
in zebrafish suggests that the two components of the indu-
cing signal are not identical; we have not addressed whether
the difference is merely quantitative or also qualitative in
nature. Interestingly, Fgf8 bead implantations in chick indi-
cate that not all otic placode markers can be equally acti-
vated ectopically in response to Fgf8 alone (Adamska et al.,
2001). Our results suggest that simultaneous expression of
Fgf8 and Fgf3, and possibly other factors, is necessary for
ectopic activation of the full ear program.
The spatial limitation of the site of placode formation
may largely be due to the hindbrain derived Fgf signal,
including Fgf8 and Fgf3. A key question is whether Fgf3
and Fgf8 act directly, or via a relay mechanism on prepla-
codal ectoderm. Several studies demonstrate that Fgf8 is
also needed for normal patterning events in the hindbrain
around the time of placode induction (Fu¨rthauer et al.,
2001; Raible and Brand, 2001; Maves et al., 2002; Reim
and Brand, 2002). Our finding that two ETS-type transcrip-
tion factors acting downstream of Fgf signaling require
Fgf8 for their expression in the ear placode (Raible and
Brand, 2001) favours in addition the possibility of a direct
action. Furthermore, FgfR1, as the likely receptor for Fgf8
(Fu¨rthauer et al., 2001) is expressed in the ear (unpublished
observations), as are FgfR2 and FgfR4 (C. Thisse and B.
Thisse, personal communication). Although we did not
assay for Fgf3 and Fgf8 protein directly, these proteins
are presumably present in limiting quantities during ear
induction, since our gain of function experiments either
with RNA injection, or with bead implantation, could
clearly expand the size of the area where otic placode is
formed. The area responding to Fgf exposure in our experi-
ments is however limited, defining a zone of competence,
which appears to be restricted to a stripe of cells adjacent
to the hindbrain. A similar restricted competence was
observed for the response to Fgf3 bead implantation in
chick (Vendrell et al., 2000), suggesting that the ability
to generate an otic placode in response to Fgf signals is
generally limited; see also Baker and Bronner-Fraser
(2001), for further discussion on competence.
3.4. A model for Fgf function during early zebrafish ear
development
We propose the following sequence of events for Fgf-
dependent induction and patterning of the zebrafish inner
ear (Fig. 11).
3.4.1. Induction phase
Initially, the entire ventral cephalic ectoderm is compe-
tent to respond to placode inducing signals, but competence
becomes increasingly restricted (Gallagher et al., 1996;
Groves and Bronner-Fraser, 2000; Yntema, 1950). Our
analysis similarly suggest that the cells responding to Fgf
signals are to some extent pre-specified. dlx3, eya1 and
six4.1 are initially expressed in a stripe of cells at the
neural-non-neural ectoderm border that gives rise to several
placodes (Akimenko et al., 1994; Kobayashi, 2000; Sahly et
al., 1999), and this expression is not affected by absence of
Fgf8 and Fgf3. During otic induction, the placode appears to
form from within this stripe (Fig. 11), concomitant with
upregulation of the above genes, and the placode-specific
expression of markers like pax8 and pax2.1. At more ante-
rior levels, the olfactory placode is thought to arise from the
same stripe (Whitlock and Westerfield, 2000). These genes
are therefore good candidates to define a state of mulitpla-
codal competence (Jacobson, 1963a, and discussion in
Baker and Bronner-Fraser, 2001). Importantly, upregulation
of dlx3 and eya1 in the otic region fails to occur in the
absence of Fgf8 and Fgf3 signaling, suggesting that the
transition from a competent to an induced state is deficient
in the absence of Fgf8 and Fgf3. This transition does not
require the presence of cephalic endomesoderm and its deri-
vatives. We therefore suggest that ear placode is specified
after definition of the multiplacodal ground state, with Fgf3
and Fgf8 together defining the signal allowing transition to
an otic placode. In agreement with this model, the acquisi-
tion of placodal competence is thought to start in the early
gastrula or in the midgastrula stage (Gallagher et al., 1996;
Jacobson, 1963a; Servetnick and Grainger, 1991). Then, as
otic field specification occurs, the placodal competence
becomes restricted to the prospective ear region (Gallagher
et al., 1996; Zwilling, 1941).
3.4.2. Maintenance phase: Fgf’s in patterning and
differentiation of the otic vesicle
At induction, placode markers like pax2.1, pax8, eya1,
dlx3 and six4.1 show relatively uniform expression through-
out the placode primordium, arguing that initially there is no
distinction between subregions of the placode. After going
through this uniform otic ground state, the otic placode
becomes polarized, visible for instance by the polar expres-

´
S. Leger, M. Brand / Mechanisms of Development 119 (2002) 91–108
104
Fig. 11. Fgf function and inductive events during successive steps of zebrafish inner ear induction and patterning. (A) During the induction phase (1), the otic
placode is induced within the domain of multiplacodal competence by Fgf8 and Fgf3 acting from the hindbrain primordium, resulting in expression of otic
placode markers in a specific temporal order (given in B). The domain of multiplacodal competence (expressing dlx3, eya1 and six4.1) is indicated in blue. (2)
Fgf3 and Fgf8 become successively restricted to rhombomere 4, and are required to maintain placodal development in its vicinity. (3) The otic vesicle is
subdivided from an initially symmetric state during the patterning phase. Fgfs signal a second time from the anterior vesicle pole (dark blue, arrow) during the
patterning phase of the otic vesicle, and (4) during feedback regulation in the anterior sensory macula. For further discussion see text.
sion of markers like pax5, pax2.1, gsc, fgf8 and others at the
vesicle stage. From comparison of gene expression domains
at this stage, it has been suggested that these domains might
provide the information needed to confer positional infor-
mation and a cell-type specification code for sense organs in
the otic vesicle (Fekete, 1996), but it is unclear how the
domains are initially set up and how they are further elabo-
rated. Our results suggest that Fgf’s in general, and Fgf8 and
Fgf3 in particular, act a second time in setting up this pattern
at the otic vesicle stage. Fgf8 and Fgf3 expression is now
detected for the first time in the otic vesicle, at the anterior
pole and, for Fgf8 very weakly and transiently, at the poster-
ior pole. This symmetric fgf8 expression is more
pronounced in ace mutant vesicles, and ace mutants some-
times even show a fusion of the maculae (macula commu-
nis). We cannot rule out that this is simply the consequence
of the reduced vesicle size. However, symmetrical arrange-
ments with only two semicircular canal protrusions and with
a macula communis is typical for the inner ears of the more
primitive myxinoids, whereas the typical split maculae of
higher vertebrates are first observed in Petromyzontes
(Gegenbaur, 1898; Fritzsch et al., 1998). The more symme-
trical ace mutant phenotype might therefore reflect an
atavistic condition of an otic vesicle with symmetric
morphological traits. In teleosts a bipartite macula is
observed only rarely (e.g. in Chimaera monstrosa), three
maculae being the more typical condition (Gegenbaur,
1898). The basis for this difference in ear organization is
not known, but we tentatively suggest, in analogy to the
MHB organizer (Sharman and Brand, 1998), that this
involves the development of an ear organizer at the anterior
pole of the ear vesicle that utilizes Fgf8, probably in
conjunction with other Fgf’s, as the signaling molecule. It
is interesting to note that in both the MHB and ear vesicle,
this also involves otx1 gene function (Morsli et al., 1999),
suggesting that part of the patterning machinery operating in
development of the MHB organizer may also act during ear
patterning. Ear specific inactivation of fgf8 and otx1 at this
stage can be used to address their detailed functions in otic
vesicle patterning and differentiation.
Expression of fgf8, spry4, as well as erm and pea3 (Raible
and Brand, 2001) correlates at the vesicle stage with defects
in otic vesicle patterning, neurogenesis, and abnormal semi-
circular canal protrusions that we observed in the mutants.
Due to the smaller size of the ace vesicle, aspects of these
phenotypic traits are likely to be a secondary consequence
of the size reduction. The selective inhibition with SU5402
at this stage (from 18 somite-stage onwards) has provided
evidence both for such indirect effects, and for likely direct
effects of Fgf signalling. Expression of spry4, pax5, gsc,

´
S. Leger, M. Brand / Mechanisms of Development 119 (2002) 91–108
105
mshD and otx1 is absent after inhibitor treatment, showing
that Fgf signaling specifically acts at the vesicle stage in
regional subdivision of the vesicle. Similarly, msxC expres-
sion in the crista epithelium is affected both in ace mutants
and after inhibition. In contrast, neurogenin1 and snail2
(and nkx5.1; Adamska et al., 2000) expression is only
affected in ace mutant vesicles, but not after late SU5402
inhibition, showing that the effect on reduction of the VIIIth
ganglion is probably a secondary consequence of the
reduced ear vesicle size in ace mutants. We therefore
suggest that Fgf-dependent patterning of the otic vesicle
starts during late somitogenesis/early pharyngula stages
(Fig. 11), and requires Fgf8, probably in conjunction with
other Fgf’s, to achieve proper patterning of the otic vesicle.
At otic vesicle stages additional Fgfs like Fgf17 and Fgf3
are expressed in the same ventro-anterior domain as Fgf8
(Reifers et al., 2000b; this paper). These fgfs are also not
expressed in the ear before the vesicle stage, but may, as we
have shown here during placode induction, cooperate with
Fgf8 during otic vesicle differentiation. For instance,
expression of Fgf8 is upregulated in ace mutant vesicles,
probably reflecting the lack of feedback regulation in the
mutants (Fu¨rthauer et al., 2001; Reifers et al., 2000b),
whereas spry4 and fgf3 are downregulated, and fgf17 is
unaffected (this paper). These data also provide further
evidence that Fgf8 is still signaling at the vesicle stage,
also in the developing macula, where its function remains
to be determined.
were cut at 1 mm and stained with toluidinblue/methylene-
blue as described in Kuwada et al. (1990).
4.2. Fluorescein-phalloidin staining of actin rich stereocilia
The following treatments were done at 48C: 5 day old
larvae were fixed in 4% paraformaldehyde (PFA) in phos-
phate buffered saline (PBS) for 2 days. They were then
rinsed and permeabilized for 4 £ 30 min in 2% Triton X-
100 (Sigma) in PBS. Embryos were stained with 2.5 mg/ml
fluorescein isothiocyanate (FITC)-labelled phalloidin
(Sigma) in PBS for 2 h in the dark. The embryos were rinsed
several times in PBS over 2 h and fixed in 4% PFA for half
an hour and then rinsed in PBS. The ears were dissected out,
mounted in PBS and viewed on a scanning confocal micro-
scope.
4.3. Daspei live staining of the lateral line hair cells
Five day old embryos were immersed in 1 mM DASPEI
(2-(4-dimethyl-aminostyryl)-N-ethyl pyridinium iodide,
Molecular Probes) in E2 for 1 h. They were then rinsed
several times in E2, anaesthetized with tricaine (0.5 mM
3-aminobenzoic acid ethyl ester, 2 mM Na2HPO4), and
mounted in methylcellulose for observation.
4.4. RNA injections
fgf8, subcloned into pCS21 (Rupp et al., 1994) was line-
arized and transcribed using the SP6 message machine kit
(Ambion). The amount of RNA injected was estimated from
the concentration and volume of a sphere of RNA injected
into oil at the same pressure settings. Typically, about 25 pg
of fgf8 RNA were injected; RNA was dissolved in 0.25 M
KCl with 0.2% of phenol red and backloaded into borosili-
cate capillaries prepared on a Sutter puller. During injection,
RNA was deposited into the cytoplasm of two cells stage
embryos; in embryos after the first cleavage, the RNA
usually stays in the progeny of the injected blastomere, as
judged from the often unilateral distribution of control lacZ
RNA, as detected with anti-b-gal antibody (Promega,
1:500) after ISH. Embryos were fixed prior to ISH and anti-
body staining.
4. Experimental procedures
Zebrafish were raised and kept under standard laboratory
conditions at about 278C (Brand et al., 2002; Westerfield,
1994). To obtain ace mutant embryos, two heterozygous
aceti282/1 carriers were crossed to one another. Typically,
the eggs were spawned synchronously at dawn of the next
morning, and embryos were collected, sorted, observed and
fixed at different times of development at 28.58C. In addi-
tion, morphological features were used to determine the
stage of the embryos, as described by (Kimmel et al.,
1995). In some cases, 0.2 mM phenylthiourea was added
to prevent melanization. Isolation and characterization of
acerebellar is described in Brand et al. (1996), Reifers et
al. (1998) and Picker et al. (1999). A fraction of ace2 larvae
die due to abnormal heart development (Reifers et al.,
2000a). For the behavioral studies, we therefore concen-
trated on larvae with relatively normal morphology and
circulation. Mzoep embryos were obtained as described
previously (Gritsman et al., 1999).
4.5. Bead implantations, inhibitor treatment, morpholino
injections
Bead implantations at shield stage and inhibitor treat-
ments were done as described (Mohammadi et al., 1997;
Reifers et al., 2000a). SU5402 inhibitor (Calbiochem) treat-
ments were done at a final concentration of 16 mM before 24
hpf, or at 8 mM after 24 hpf, diluted into embryo medium
from an 8 mM stock in dimethyl sulfoxide (DMSO); control
treatments with DMSO dilution without inhibitor had no
effect (not shown). Morpholino injections for Mo-fgf8
were done as described in Araki and Brand (2001), for
Mo-fgf3 in Raible and Brand (2001); injections with four-
base pair-mismatched (MM) control Mo’s had no effect.
4.1. In situ hybridization and histology
ISH were done as described in Reifers et al. (1998).
Digoxigenin (DIG) labelled probes were prepared according
to manufacturers instructions (Roche). Semithin sections

´
S. Leger, M. Brand / Mechanisms of Development 119 (2002) 91–108
106
Mo-fgf8: GAGTCTCATGTTTATAGCCTCAGTA;
Mo-fgf8-MM:GAGTCTGATCTTTTTAGCCACAGTA;
Acknowledgements
Mo-fgf3: CAGTAACAACAAGAGCAGAATTATA;
Mo-fgf3-MM:CACTAACAAGAAGACCACAATTATA.
We thank the past and present members of the Brand
laboratory for stimulating discussions and advice, in parti-
cular Frank Reifers and Alexander Picker. We furthermore
thank Catherine Haddon for many a lobster, and her, Tanya
Whitfield and Julian Lewis for stimulating discussions about
ear development, and Tanya Whitfield for insightful
comments on the manuscript. We also thank many zebrafish
researchers, particularly Christine and Bernard Thisse and
M. Kobayashi for kindly providing probes employed in this
study, and Muriel Rhinn and Anukampa Barth for providing
mzoep embryos. This work was supported by grants from
the Deutsche Forschungsgemeinschaft, the EU (QLG3-CT-
2001-02310), and the Max-Planck-Society.
In selected cases, the results were confirmed using a second
Mo-fgf3 (Phillips et al., 2001): CATTGTGGCATGGCGG-
GATGTCGGC.
4.6. Cell transplantations
Fertilized wild-type eggs, to be used as donors, were
fluorescently labelled by injection of 7.5% tetramethylrho-
damine and biotin, lysine fixable dextran (mini-ruby),
10.000 MW (Molecular Probes D-3312) in 0.25 M KCL.
They were raised together with an unlabelled ace lay (to be
used as hosts) to the shield stage. Cells from the wild-type
donors embryos were then transplanted into the hosts. Host
embryos were raised until the 10–12 somites stage and fixed
overnight in 4% paraformaldehyde and a whole-mount
RNA in situ hybridization to pax2.1 was performed. The
biotin lineage tracer in the transplanted cells was then
detected in the host embryos using the ABC kit (Vector
Labs). The host embryos were mounted in 70% glycerol
for observation. For photography, embryos were dissected
using electrolytically sharpened tungsten needles.
References
´
Adamska, M., Leger, S., Brand, M., Hadrys, T., Bober, E., 2000. Expres-
sion of a new zebrafish inner ear and lateral line specific gene, Nkx5-1,
is significantly reduced in the ears of Fgf8 mutant, ace. Mech. Dev. 97,
161–165.
Adamska, M., Herbrand, H., Adamski, M., Kruger, M., Braun, T., Bober,
E., 2001. FGFs control the patterning of the inner ear but are not able to
induce the full ear program. Mech. Dev. 109, 303–313.
Akimenko, M.A., Ekker, M., Wegner, J., Lin, W., Westerfield, M., 1994.
Combinatorial expression of three zebrafish genes related to distal- less:
part of a homeobox gene code for the head. J. Neurosci. 14, 3475–3486.
Araki, I., Brand, M., 2001. Antisense-morpholino-induced knockdown of
fgf8 phenocopies the acerebellar (ace) phenotype. Genesis 30, 157–
159.
4.7. Anti-phospho histone H3 staining
After ISH, embryos were washed 3 £ 5 min in
PBS 1 0.1% Tween 20 (PBSTw) and blocked with 10%
normal goat serum (NGS) in PBS 1 0.8% Triton X-100
(PBSTx) for at least 1 h at room temperature. They were
then incubated overnight at 48C in the anti-phospho histone
H3 primary antibody (rabbit polyclonal IgG; Upstate
Biotechnology; 1:300 in 1% NGS-PBSTx) on a shaker.
Embryos were washed 2 £ 5 min and then 4 £ 30 min in
PBSTw. The endogenous peroxidases were blocked by
washing 5 min in 50% MeOH/PBS, 5 min in 100%
MeOH, 10 min in 0.30% H2O2/MeOH, 5 min in 50%
MeOH/PBS and 5 min in PBS. Then, embryos were incu-
bated overnight with the secondary antibodies (anti-rabbit
IgG peroxidase conjugated, Sigma #A-6154; 1:200 in 1%
NGS-PBSTx) on a shaker. Embryos were washed 2 £ 5 and
4 £ 30 min in PBSTw and 5 min in Tris 0.1 M pH 7.4.
Embryos were then incubated with DAB peroxidase
substrate (sigma, #D-4293). The color reaction was stopped
by washing with PBSTw several times. The embryos were
fixed 30 min in 4% PFA at room temperature, washed 2 £ 5
min in PBSTw and cleared in 70% glycerol/PBSTw.
Baker, C.V.H., Bronner-Fraser, M., 2001. Vertebrate cranial placodes. I.
Embryonic induction. Dev. Biol. 232, 1–61.
Brand, M., Heisenberg, C.-P., Jiang, Y.-J., Beuchle, D., Lun, K., van Eeden,
F.J.M., Furutani-Seiki, M., Granato, M., Haffter, P., Hammerschmidt,
M., Kane, D.A., Kelsh, R.N., Mullins, M.C., Odenthal, J., Nu¨sslein-
Volhard, C., 1996. Mutations in zebrafish genes affecting the formation
of the boundary between midbrain and hindbrain. Development 123,
179–190.
Brand, M., Granato, M., Nu¨sslein-Volhard, C., 2002. Keeping and raising
zebrafish. In: Nu¨sslein-Volhard, C., Dahm, R. (Eds.). Zebrafish: A Prac-
tical Approach, Oxford University Press, Oxford.
Burgess, S., Reim, G., Chen, W., Hopkins, N., Brand, M., 2002. The zebra-
fish spiel-ohne-grenzen (spg) gene encodes the POU homeodomain
protein Pou2 related to mammalian Oct4 and is essential for formation
of the midbrain-hindbrain territory and early morphogenesis. Develop-
ment 129, 905–916.
Chambers, D., Mason, I., 2000. Expression of sprouty2 during early devel-
opment of the chick embryo is coincident with known sites of FGF
signalling. Mech. Dev. 91, 361–364.
Chisaka, O., Musci, T.S., Capecchi, M.R., 1992. Developmental defects of
the ear, cranial nerves and hindbrain resulting from targeted disruption
of the mouse homeobox gene Hox-1.6 [see comments]. Nature 355,
516–520.
Cordes, S.P., Barsh, G.S., 1994. The mouse segmentation gene kr encodes a
novel basic domain-leucine zipper transcription factor. Cell 79, 1025–
1034.
4.8. Acridine orange (AO)
Deol, M.S., 1964. The abnormalities of the inner ear in kreisler mice. J.
Embryol. Expt. Morphol. 12, 475–490.
AO was applied to dechorionated embryos in E3 medium
at a final concentration of 2 mg/ml for between 1 and 3 h.
Excess AO was washed off by rinsing several times in E3
after this period. Fluorescence was monitored microscopi-
cally with FITC filter set (Hoechst).
Deol, M.S., 1966. Influence of the neural tube on the differentiation of the
inner ear in the mammalian embryo. Nature 209, 219–220.
Dolle, P., Lufkin, T., Krumlauf, R., Mark, M., Duboule, D., Chambon, P.,
1993. Local alterations of Krox-20 and Hox gene expression in the

´
S. Leger, M. Brand / Mechanisms of Development 119 (2002) 91–108
107
hindbrain suggest lack of rhombomeres 4 and 5 in homozygote null
Hoxa-1 (Hox-1.6) mutant embryos. Proc. Natl. Acad. Sci. USA 90,
7666–7670.
fied growth cones in the spinal cord of zebrafish embryos. J. Neurosci.
10, 1299–1308.
Kozlowski, D.J., Murakami, T., Ho, R.K., Weinberg, E.S., 1997. Regional
cell movement and tissue patterning in the zebrafish embryo revealed
by fate mapping with caged fluorescein. Biochem. Cell Biol. 75, 551–
562.
Ekker, M., Akimenko, M.A., Bremiller, R., Westerfield, M., 1992. Regional
expression of three homeobox transcripts in the inner ear of zebrafish
embryos. Neuron 9, 27–35.
Epstein, D.J., Vekemans, M., Gros, P., 1991. Splotch (Sp2H), a mutation
affecting development of the mouse neural tube, shows a deletion
within the paired homeodomain of Pax-3. Cell 67, 767–774.
Fekete, D.M., 1996. Cell fate specification in the inner ear. Curr. Opin.
Neurobiol. 6, 533–541.
Ladher, R.K., Anakwe, K.U., Gurney, A.L., Schoenwolf, G.C., Francis-
West, P.H., 2000. Identification of synergistic signals initiating inner
ear development. Science 290, 1965–1967.
Li, Y., Allende, M.L., Finkelstein, R., Weinberg, E.S., 1994. Expression of
two zebrafish orthodenticle-related genes in the embryonic brain. Mech.
Dev. 48, 229–244.
Fritzsch, B., Barald, K.F., Lomax, M.I., 1998. Early embryology of the
vertebrate ear. In: Rubel, R.W., Popper, A.N., Fay, R.R. (Eds.). Devel-
opment of the Auditory System, 9. Springer, New York, pp. 80–145.
Frohman, M.A., Martin, G.R., Cordes, S.P., Halamek, L.P., Barsh, G.S.,
1993. Altered rhombomere-specific gene expression and hyoid bone
differentiation in the mouse segmentation mutant, kreisler (kr). Devel-
opment 117, 925–936.
Lombardo, A., Slack, J.M., 1998. Postgastrulation effects of fibroblast
growth factor on Xenopus development. Dev. Dyn. 212, 75–85.
Lombardo, A., Isaacs, H.V., Slack, J.M., 1998. Expression and functions of
FGF-3 in Xenopus development. Int. J. Dev. Biol. 42, 1101–1107.
Lufkin, T., Dierich, A., LeMeur, M., Mark, M., Chambon, P., 1991. Disrup-
tion of the Hox-1.6 homeobox gene results in defects in a region corre-
sponding to its rostral domain of expression. Cell 66, 1105–1119.
Mahmood, R., Kiefer, P., Guthrie, S., Dickson, C., Mason, I., 1995. Multi-
ple roles for FGF-3 during cranial neural development in the chicken.
Development 121, 1399–1410.
Fu¨rthauer, M., Thisse, C., Thisse, B., 1997. A role for Fgf-8 in the dorso-
ventral patterning of the zebrafish gastrula. Development 124, 4253–
4264.
Fu¨rthauer, M., Reifers, F., Brand, M., Thisse, B., Thisse, C., 2001. sprouty4
acts in vivo as a feedback-induced antagonist of Fgf signaling in zebra-
fish. Development 128, 2175–2186.
Malicki, J., Schier, A.F., Solnica Krezel, L., Stemple, D.L., Neuhauss, S.C.,
Stainier, D.Y., Abdelilah, S., Rangini, Z., Zwartkruis, F., Driever, W.,
1996. Mutations affecting development of the zebrafish ear. Develop-
ment 123, 275–283.
Gallagher, B.C., Henry, J.J., Grainger, R.M., 1996. Inductive processes
leading to inner ear formation during Xenopus development. Dev.
Biol. 175, 95–107.
Mansour, S.L., Goddard, J.M., Capecchi, M.R., 1993. Mice homozygous
for a targeted disruption of the proto-oncogene int-2 have developmen-
tal defects in the tail and inner ear. Development 117, 13–28.
Mark, M., Lufkin, T., Vonesch, J.L., Ruberte, E., Olivo, J.C., Dolle, P.,
Gorry, P., Lumsden, A., Chambon, P., 1993. Two rhombomeres are
altered in Hoxa-1 mutant mice. Development 119, 319–338.
Maroon, H., Walshe, J., Mahmood, R., Kiefer, P., Dickson, C., Mason, I.,
2002. Fgf3 and Fgf8 are required together for formation of the otic
placode and vesicle. Development 129, 2099–2108.
Gegenbaur, C., 1898. Vergleichende Anatomie der Wirbeltiere mit Beru¨ck-
sichtigung der Wirbellosen. Wilhelm Engelmann, Leipzig, I, 874–910.
Gritsman, K., Zhang, J., Cheng, S., Heckscher, E., Talbot, W.S., Schier,
A.F., 1999. The EGF-CFC protein one-eyed pinhead is essential for
nodal signaling. Cell 97, 121–132.
Groves, A.K., Bronner-Fraser, M., 2000. Competence, specification and
commitment in otic placode induction. Development 127, 3489–3499.
Haddon, C., Lewis, J., 1996. Early ear development in the embryo of the
zebrafish, Danio rerio. J. Comp. Neurol. 365, 113–128.
Harrison, R.G., 1935. Factors concerned in the development of the ear in
Amblystoma punctatum. Anat. Rec. 64, 38–39.
Maves, L., Jackman, W., Kimmel, C.B., 2002. FGF3 and FGF8 mediate a
rhombomere 4 signaling activity in the zebrafish hindbrain. Develop-
ment 129, 3825–3837.
Harrison, R.G., 1945. Relations of symmetry in the developing embryo.
Trans. Conn. Acad. Arts Sci. 36, 277–330.
McKay, I.J., Muchamore, I., Krumlauf, R., Maden, M., Lumsden, A.,
Lewis, J., 1994. The kreisler mouse: a hindbrain segmentation mutant
that lacks two rhombomeres. Development 120, 2199–2211.
McKay, I.J., Lewis, J., Lumsden, A., 1996. The role of FGF-3 in early inner
ear development: an analysis in normal and kreisler mutant mice. Dev.
Biol. 174, 370–378.
Jacobson, A.G., 1963a. The determination and positioning of the nose, lens
and ear. I. Interactions within the ectoderm, and between the ectoderm
and underlying tissues. J. Exp. Zool. 154, 273–283.
Jacobson, A.G., 1966. Inductive processes in embryonic development.
Science, 152, 25–34.
Mendonsa, E.S., Riley, B.B., 1999. Genetic analysis of tissue interactions
required for otic placode induction in the zebrafish. Dev. Biol. 206,
100–112.
Jorgensen, J.M., 1989. Evolution of octavolateralis sensory cells. In:
¨
¨
Coombs, S., Gorner, P., Munz, H. (Eds.). The Mechanosensory Lateral
Line: Neurobiology and Evolution, Springer-Verlag, New York, pp.
187–216.
Moens, C.B., Yan, Y.L., Appel, B., Force, A.G., Kimmel, C.B., 1996.
valentino: a zebrafish gene required for normal hindbrain segmentation.
Development 122, 3981–3990.
Kiefer, P., Strahle, U., Dickson, C., 1996. The zebrafish Fgf-3 gene: cDNA
sequence, transcript structure and genomic organization. Gene 168,
211–215.
Moens, C.B., Cordes, S.P., Giorgianni, M.W., Barsh, G.S., Kimmel, C.B.,
1998. Equivalence in the genetic control of hindbrain segmentation in
fish and mouse. Development 125, 381–391.
Kimmel, C.B., Ballard, W.W., Kimmel, S.R., Ullmann, B., Schilling, T.F.,
1995. Stages of embryonic development of the zebrafish. Dev. Dyn.
203, 253–310.
Mohammadi, M., McMahon, G., Sun, L., Tang, C., Hirth, P., Yeh, B.K.,
Hubbard, S.R., Schlessinger, J., 1997. Structures of the tyrosine kinase
domain of fibroblast growth factor receptor in complex with inhibitors.
Science 276, 955–960.
Kinoshita, N., Minshull, J., Kirschner, M.W., 1995. The identification of
two novel ligands of the FGF receptor by a yeast screening method and
their activity in Xenopus development. Cell 83, 621–630.
Kobayashi, M., 2000. Six family genes as key factors in the sensory organ
formation. Tanpakushitsu Kakusan Koso 45, 2775–2781.
Korzh, V., Sleptsova, I., Liao, J., He, J., Gong, Z., 1998. Expression of the
zebrafish bHLH genes ngn1 and nrd defines distinct stages of neural
differentiation. Dev. Dyn. 213, 92–104.
Morsli, H., Tuorto, F., Choo, D., Postiglione, M.P., Simeone, A., Wu, D.K.,
1999. Otx1 and Otx2 activities are required for the normal development
of the mouse inner ear. Development 126, 2335–2343.
Mu¨nchberg, S.R., Ober, E.A., Steinbeisser, H., 1999. Expression of the Ets
transcription factors erm and pea3 in early zebrafish development.
Mech. Dev. 88, 233–236.
Krauss, S., Johansen, T., Korzh, V., Fjose, A., 1991. Expression of the
zebrafish paired box gene pax[zf-b] during early neurogenesis. Devel-
opment 113, 1193–1206.
Nasevicius, A., Ekker, S.C., 2000. Effective targeted gene ‘knockdown’ in
zebrafish. Nat. Genet. 26, 216–220.
Nicolson, T., Rusch, A., Friedrich, R.W., Granato, M., Ruppersberg, J.P.,
Nusslein-Volhard, C., 1998. Genetic analysis of vertebrate sensory hair
Kuwada, J.Y., Bernhardt, R.R., Chitnis, A.B., 1990. Pathfinding by identi-

´
S. Leger, M. Brand / Mechanisms of Development 119 (2002) 91–108
108
cell mechanosensation: the zebrafish circler mutants. Neuron 20, 271–
283.
Shamim, H., Mason, I., 1999. Expression of Fgf4 during early development
of the chick embryo. Mech. Dev. 85, 189–192.
Northcutt, R.G., 1986. Evolution of the octavolateralis system: evaluation
and heuristic value of phylogenetic hypotheses. In: Rubel, R.W., Van de
Water, T.R., Rubel, E.W. (Eds.). The Biology of Change in Otolaryn-
gology, Elsevier, New York, pp. 3–14.
Shanmugalingam, S., Houart, C., Picker, A., Reifers, F., MacDonald, R.,
Barth, A.K., Brand, M., Wilson, S.W., 2000. Ace/Fgf8 is required for
forebrain commissure formation and patterning of the telencephalon.
Development 127, 2549–2561.
Pfeffer, P.L., Gerster, T., Lun, K., Brand, M., Busslinger, M., 1998. Char-
acterization of three novel members of the zebrafish Pax2/5/8 family:
dependency of Pax5 and Pax8 expression on the Pax2.1(noi) function.
Development 125, 3063–3074.
Sharman, A.C., Brand, M., 1998. Evolution and homology of the nervous
system: cross-phylum rescues of otd/Otx genes. Trends Genet. 14, 211–
214.
Slack, J.M., 1995. Developmental biology. Growth factor lends a hand.
Nature 374, 217–218.
Phillips, B.T., Bolding, K., Riley, B.B., 2001. Zebrafish fgf3 and fgf8
encode redundant functions required for otic placode induction. Dev.
Biol. 235, 351–365.
Steel, K.P., 1995. Inherited hearing defects in mice. Annu. Rev. Genet. 29,
675–701.
Picker, A., Brennan, C., Reifers, F., Clarke, J.D., Holder, N., Brand, M.,
1999. Requirement for the zebrafish mid-hindbrain boundary in
midbrain polarisation, mapping and confinement of the retinotectal
projection. Development 126, 2967–2978.
Stone, L.S., 1931. Induction of the ear by the medulla and its relation to
experiments on the lateralis system in amphibia. Science 74, 577.
Tannahill, D., Isaacs, H.V., Close, M.J., Peters, G., Slack, J.M., 1992.
Development expression of the Xenopus int-2 (FGF-3) gene: activation
by mesodermal and neural induction. Development 115, 695–702.
Thisse, C., Thisse, B., Postlethwait, J.H., 1995. Expression of snail2, a
second member of the zebrafish snail family, in cephalic mesendoderm
and presumptive neural crest of wild-type and spadetail mutant
embryos. Dev. Biol. 172, 86–99.
¨
Picker, A., Scholpp, S., Bohli, H., Takeda, H., Brand, M., 2002. A novel
positive transcriptional feedback loop in midbrain-hindbrain boundary
development is revealed through analysis of the zebrafish pax2.1
promoter in transgenic lines. Development 129, 3227–3239.
Platt, C., Popper, A.N., Fay, R.R., 1989. The ear as a part of the octavola-
teralis system. In: Coombs, S. (Ed.). The Mechanosensory Lateral Line,
Springer, Berlin, Heidelberg, New York, pp. 633–651.
Thisse, C., Thisse, B., Halpern, M.E., Postlethwait, J.H., 1994. Goosecoid
expression in neurectoderm and mesendoderm is disrupted in zebrafish
cyclops gastrulas. Dev. Biol. 164, 420–429.
Raible, F., Brand, M., 2001. Tight transcriptional control of the ETS
domain factors Erm and Pea3 by FGF signaling during early zebrafish
nervous system development. Mech. Dev. 107, 105–117.
Torres, M., Giraldez, F., 1998. The development of the vertebrate inner ear.
Mech. Dev. 71, 5–21.
¨
Reifers, F., Bohli, H., Walsh, E.C., Crossley, P.H., Stainier, D.Y.R., Brand,
Van de Water, T.R., Conley, X., 1982. Neural inductive message to the
developing mammalian inner ear: contact mediated versus extracellular
matrix interaction. Anat. Rec., 195A.
M., 1998. Fgf8 is mutated in zebrafish acerebellar mutants and is
required for maintenance of midbrain-hindbrain boundary development
and somitogenesis. Development 125, 2381–2395.
Van de Water, T.R., Represa, J., 1991. Tissue interactions and growth
factors that control development of the inner ear. Neural tube-otic
anlage interaction. Ann. NY Acad. Sci., 630, 116–128.
Vendrell, V., Carnicero, E., Giraldez, F., Alonso, M.T., Schimmang, T.,
2000. Induction of inner ear fate by FGF3. Development 127, 2011–
2019.
Reifers, F., Adams, J., Mason, I., Schulte-Merker, S., Brand, M., 2000a.
Overlapping and distinct functions provided by fgf17, a new zebrafish
member of the Fgf8/17/18 subgroup of Fgfs. Mech. Dev. 99, 39–49.
´
Reifers, F., Walsh, E.C., Leger, S., Stainier, D.Y.R., Brand, M., 2000b.
Induction and differentiation of the zebrafish heart requires fibroblast
growth factor 8 (Fgf8/acerebellar). Development 127, 225–235.
Reim, G., Brand, M., 2002. spiel-ohne-grenzen/pou2 mediates regional
competence to respond to Fgf8 during zebrafish early neural develop-
ment. Development 129, 917–933.
Waddington, C.H., 1937. The determination of the auditory placode in the
chick. J. Exp. Biol. 14, 232–239.
Westerfield, M., 1994. The Zebrafish Book, 2.1 Edition. University of
Oregon Press, Oregon.
Represa, J., Leon, Y., Miner, C., Giraldez, F., 1991. The int-2 proto-onco-
gene is responsible for induction of the inner ear. Nature 353, 561–563.
Riley, B.B., Grunwald, D.J., 1996. A mutation in zebrafish affecting a
localized cellular function required for normal ear development. Dev.
Biol. 179, 427–435.
Whitfield, T.T., Riley, B.B., Chiang, M.Y., Phillips, B., 2002. Development
of the zebrafish inner ear. Dev. Dyn. 223, 427–458.
Whitfield, T., Granato, M., van Eeden, F.J.M., Schach, U., Brand, M.,
Furutani-Seiki, M., Haffter, P., Hammerschmidt, M., Heisenberg, C.-
P., Jiang, Y.-J., Kane, D.A., Kelsh, R.N., Mullins, M.C., Odenthal, J.,
Nu¨sslein-Volhard, C., 1996. Mutations affecting development of the
zebrafish inner ear and lateral line. Development 123, 241–254.
Whitlock, K.E., Westerfield, M., 2000. The olfactory placodes of the zebra-
fish form by convergence of cellular fields at the edge of the neural
plate. Development 127, 3645–3653.
Riley, B.B., Zhu, C., Janetopoulos, C., Aufderheide, K.J., 1997. A critical
period of ear development controlled by distinct populations of ciliated
cells in the zebrafish. Dev. Biol. 191, 191–201.
Rupp, R.A.W., Snider, L., Weintraub, H., 1994. Xenopus embryos regulatie
the nuclear localization of XMyoD. Genes Dev. 8, 1311–1323.
Sahly, I., Andermann, P., Petit, C., 1999. The zebrafish eya1 gene and its
expression pattern during embryogenesis. Dev. Genes Evol. 209, 399–
410.
Wilkinson, D.G., Peters, G., Dickson, C., McMahon, A.P., 1988. Expres-
sion of the FGF-related proto-oncogene int-2 during gastrulation and
neurulation in the mouse. EMBO J. 7, 691–695.
Schier, A.F., Shen, M.M., 2000. Nodal signalling in vertebrate develop-
ment. Nature 403, 385–389.
Woo, K., Fraser, S.E., 1998. Specification of the hindbrain fate in the
zebrafish. Dev. Biol. 197, 283–296.
Sechrist, J., Scherson, T., Bronner-Fraser, M., 1994. Rhombomere rotation
reveals that multiple mechanisms contribute to the segmental pattern of
hindbrain neural crest migration. Development 120, 1777–1790.
Servetnick, M., Grainger, R.M., 1991. Changes in neural and lens compe-
tence in Xenopus ectoderm: evidence for an autonomous developmen-
tal timer. Development 112, 177–188.
Yntema, C.L., 1950. An analysis of induction of the ear from foreign
ectoderm in the salamander embryo. J. Exp. Zool. 113, 211–240.
Yntema, C.L., 1955. Ear and nose. Analysis of Development, Saunders,
Philadelphia/London, pp. 415–428.
Zwilling, E., 1941. The determination of the otic vesicle in Rana pipiens. J.
Exp. Zool. 86, 333–342.