Deafness is one of the most common hereditary diseases
(reviewed by Petit et al., 2001; Morton, 2002). One out of 1000
children suffers from congenital deafness and inherited
progressive hearing loss affects another 2.5% of the population.
Late onset age-related hearing loss is thought to be due to a
combination of genetic and environmental causes. To date,
over 500 syndromic deafness loci have been identified which
confer hearing loss in association with defects in other organ
systems (Gorlin et al., 1995). The most common forms of
syndromic deafness cause abnormalities in craniofacial and
skeletal structures (Branchio-oto-renal syndrome, Stickler
syndrome), retinitis pigmentosa (Usher syndrome) or heart
defects (Jervell and Lange-Nielsen syndrome) (Abdelhak et al.,
1997; Ahmad et al., 1991; Bitner-Glindzicz and Tranebjaerg,
2000; Petit, 2001). In addition, over 90 loci for non-syndromic
hearing loss have been identified thus far. For more than 30 of
these loci, the affected gene is known (Morton, 2002). In many
cases, these genes are responsible for both syndromic and non-
syndromic deafness, depending on the nature of the mutation
(Astuto et al., 2002).
DFNA5 is a non-syndromic autosomal dominant form of
progressive hearing loss that was identified in an extended
Dutch family (van Camp et al., 1995; DeLeenheer et al., 2002).
High frequency hearing loss starts in the first or second decade
of life. The patients carry a complex intronic insertion/deletion
mutation that leads to truncation of the protein (Van Laer et al.,
1998). DFNA5 encodes a protein of 496 amino acids that is
found in human cochlea, brain, placenta and kidney cDNA
preparations. DFNA5 is a novel protein, sharing no obvious
homology to other proteins, and as there is no animal model,
nothing is known about its molecular function or what role it
plays during vertebrate development. To gain insight into a
possible function for Dfna5, we examined its expression
pattern in developing zebrafish embryos and its function using
a morpholino knock-down strategy. Our phenotypic analysis
suggests that Dfna5 function is essential for development of
the semicircular canals and the pharyngeal cartilage, and acts
at the level of extracellular matrix production.
The extracellular matrix plays an important role in cell
migration, differentiation and morphogenesis of organs such as
the ear and jaw. The ear and the pharyngeal cartilage arise from
the same region of the embryo. In zebrafish, the otic placode
develops from the cells overlying the second cranial neural
crest (CNC) stream, adjacent to rhombomeres 5 and 6, at
about 14 hpf (hours post fertilization). At around 45 hpf, the
Over 30 genes responsible for human hereditary hearing
loss have been identified during the last 10 years. The
proteins encoded by these genes play roles in a diverse
set of cellular functions ranging from transcriptional
regulation to K+recycling. In a few cases, the genes are
novel and do not give much insight into the cellular or
molecular cause for the hearing loss. Among these poorly
understood deafness genes is DFNA5. How the truncation
of the encoded protein DFNA5 leads to an autosomal
dominant form of hearing loss is not clear. In order to
understand the biological role of Dfna5, we took a reverse-
genetic approach in zebrafish. Here we show that
morpholino antisense nucleotide knock-down of dfna5
function in zebrafish leads to disorganization of the
developing semicircular canals
pharyngeal cartilage. This phenotype closely resembles
previously isolated zebrafish
and reduction of
including the mutant jekyll. jekyll encodes Ugdh [uridine 5′-
diphosphate (UDP)-glucose dehydrogenase], an enzyme
that is crucial for production of the extracellular matrix
component hyaluronic acid (HA). In dfna5 morphants,
expression of ugdh is absent in the developing ear and
pharyngeal arches, and HA levels are strongly reduced in
the outgrowing protrusions of the developing semicircular
canals. Previous studies suggest that HA is essential for
differentiating cartilage and directed outgrowth of the
epithelial protrusions in the developing ear. We hypothesize
that the reduction of HA production leads to uncoordinated
outgrowth of the canal columns and impaired facial
Key words: dfna5, Deafness, Zebrafish, Semicircular canals,
Pharyngeal cartilage, UDP-glucose dehydrogenase, Hyaluronic acid
The deafness gene dfna5 is crucial for ugdh expression and HA
production in the developing ear in zebrafish
Elisabeth Busch-Nentwich1, Christian Söllner1, Henry Roehl2and Teresa Nicolson3,*
1Max-Planck-Institut für Entwicklungsbiologie, Spemannstr. 35, 72076 Tübingen, Germany
2Centre for Developmental Genetics, Department of Biomedical Science, University of Sheffield, Firth Court, Sheffield S10 2TN,
3Oregon Hearing Research Center and Vollum Institute, Oregon Health & Science University, Portland, OR 97201, USA
*Author for correspondence (e-mail: firstname.lastname@example.org)
Accepted 30 October 2003
Development 131, 943-951
Published by The Company of Biologists 2004
Development and disease
epithelial projections of the developing semicircular canal
begin to grow out. At 52 hpf, the opposing projections of the
anterior and posterior canals touch and fuse. This process is
completed with the fusion of the ventral column at 64 hours.
The directed outgrowth of the protrusions is dependent on
secretion of extracellular matrix (ECM), including hyaluronic
acid (HA), into the extending lumen between the epithelium
and the underlying mesenchyme (Haddon and Lewis, 1991).
The ECM is later replaced by invading fibroblasts. (Haddon
and Lewis, 1991; Haddon and Lewis, 1996; Waterman and
The cartilage that gives rise to the jaw and gills is called
pharyngeal cartilage and is of neural-crest origin (Schilling and
Kimmel, 1994). Pharyngeal cartilage is derived from three
streams of cranial neural crest (CNC) that begin to migrate at
12 hpf. The most anterior two CNC streams give rise to the
cartilage of the upper and lower jaw, and the posterior stream
is subdivided by endodermal pouches to give rise to the five
sets of branchial (gill) cartilage. The pharyngeal cartilage
condensations begin to differentiate starting at 53 hpf as seen
by Alcian Blue staining (reviewed by Kimmel et al., 2001;
Schilling and Kimmel, 1994). The early steps of jaw
development are regulated by Sox9, under the control of bone
morphogenetic proteins (BMPs) and sonic hedgehog (Shh)
(de Crombrugghe et al., 2001). After condensation, the
differentiating chondrocytes begin to express col2a1, the main
collagen of differentiated cartilage, col11a2, aggrecan and
other ECM proteins under the control of Sox5 and Sox6. In
differentiated cartilage, only 5% of the volume is occupied by
chondrocytes, with the remainder consisting of ECM.
Not surprisingly, many genes expressed in the posterior
region of the head regulate the development of both the ear and
pharyngeal cartilage. For example, expression of fgf3 and fgf8
in rhombomere 4 is required for induction of the ear placode
and formation of branchial cartilage (Maroon et al., 2002;
Leger and Brand, 2002; Liu et al., 2003). A mutation in another
gene, jekyll, leads to interrupted ear columns and a lack of
cranial cartilage (Neuhauss et al., 1996). jekyll encodes Udgh,
an enzyme required for synthesis of proteoglycans including
HA (Walsh and Stainier, 2001). We show that a knock-down
of dfna5 function in zebrafish causes the loss of ugdh
expression and reduction of HA levels, leading to
malformation of the semicircular canals of the ear and
Materials and methods
Cloning and physical mapping
We blasted HsDFNA5 protein sequence against EST data from
GenBank and obtained an EST clone containing full-length zebrafish
dfna5 cDNA sequence (Accession Number: BE015795). The forward
primer 5′-TTGCAAAAGCCACTAAGAACC-3′ and reverse primer
5′-ATCGGTTACGCTTGATGACC-3′ were used on the T51 radiation
hybrid panel to physically map zebrafish dna5 to linkage group 16.
We analyzed the genomic organization of zebrafish dfna5 partially
by blasting the cDNA sequence against genomic zebrafish sequence
and partially by PCR on genomic sequence with subsequent cloning
and sequencing of the obtained PCR-products.
Whole-mount in situ hybridization
We amplified full-length dfna5 using the forward primer A5-bf 5′-
AGCACGAGCTGATCCTCAAA-3′ and reverse primer A5-br 5′-
AAGCCTCGTCTGTTTCGTGC-3′, cloned the 1990 bp fragment and
used it as a template to synthesize a digoxigenin labeled in situ RNA
probe. For col2a1 we used the forward primer 5′-CTGGATCT-
GATGGTCCACCT-3′ and reverse primer 5′-ACATGGCTGGATT-
TCAGAGG-3′ to amplify a 1593 bp fragment and for ugdh the
forward primer 5′-GACGTACGGTATGGGCAAAG-3′ and reverse
primer 5′-TTGATTTCCAGCAATGGTCA-3′ to amplify a 1304 bp
fragment. Both PCR-product were used as described for the dfna5 in
situ probe. Whole-mount in situ hybridization was performed as
described previously (Thisse et al., 1993).
Antisense morpholino oligonucleotide and DNA injections
We designed one antisense morpholino oligonucleotide (MO, Gene
Tools) directed against the 5′ sequence around the putative start codon
to block dfna5 translation (“ATG-MO”) and one MO against the splice
donor site of exon 8 to interfere with splicing (“GT-MO”). The
sequences for the ATG-MO and the GT-MO were 5′-TGCAAAC-
ATCTTCAATGCTGACAAG-3′ and 5′-TGATTTAACTGAACTCA-
CCGTCTAG-3′, respectively. An additional morpholino with four
base pair exchanges was designed for control injections. The sequence
for the mismatch MO was 5′-TGCGAACACCTTTAATGCTAAC-
AAG-3′. We injected concentrations between 10 and 40 ng in single
cell embryos in a volume of 5 nl.
For rescue experiments we amplified full-length dfna5 with the
primers used for generation of the in situ probe template, cloned it in
pCS2+ vector and identified a mutation free clone by sequencing. The
plasmid was co-injected in single cell embryos with different
concentrations of ATG-MO in a concentration of 50 pg/nl.
Histology and electron microscopy
Whole larvae (day 5) were anesthetized with 0.02% MESAB (3-
aminobenzoic acid ethyl ester) and then fixed in 2.0% glutaraldehyde
and 1.0% paraformaldehyde in PBS overnight to several days at 4°C.
Specimens were fixed with 1.0% OsO4in H2O for 10 minutes on ice,
followed by fixation and contrast with 1.0% uranyl acetate for 1 hour
on ice, and then dehydrated with several steps in ethanol and
embedded in Epon. Sections (5 µm) were stained with Toluidine Blue.
Ultrathin sections were stained with lead citrate and uranyl acetate.
Alcian-Blue cartilage staining
Cartilage was stained with Alcian Blue as described previously
(Schilling et al., 1996).
Whole mount labeling for hyaluronic acid
Freshly fixed larvae (4% PFA in PBS overnight at 4°C) were
permeabilized for 6 hours at room temperature with 2% Triton, 4%
PFA in PBS. After washing several times in TBS, larvae were
incubated with 10 µg/ml biotinylated hyaluronic acid binding protein
(HABP) from Seikagaku corporation in TBS overnight at 4°C. The
HABP was removed and the larvae washed in TBS three times for 20
minutes. After 4 hours blocking with 2% BSA in TBS at room
temperature, the larvae were incubated with anti-biotin rabbit
antibody (Heinz Schwarz, Tübingen), diluted 1:50 in TBS for 3 hours
at room temperature. The larvae were washed in TBS three times for
20 minutes and incubated overnight at 4°C with alexa-Fluor-coupled
anti-rabbit antibody, diluted 1:500. After rinsing several times, larvae
were mounted on a glass slide in 1.5% low melting point agarose and
Sequence analysis and expression of dfna5
To explore the function of dfna5 in vertebrates, we cloned the
zebrafish orthologue of DFNA5 by blasting the human protein
sequence against EST databases. Zebrafish Dfna5 has 472
amino acids with 33% identity and 51% similarity to human
Development 131 (3)Research article
945 The role of dfna5 in ear and jaw developmentDevelopment and disease
DFNA5 with 496 amino acids (Mouse Dfna5 values are 30%
and 49%, respectively) (Fig. 1). With respect to genomic
organization, the human and zebrafish genes are highly
conserved (Fig. 2A). Both genes have ten exons with identical
exon/intron boundaries. The first exon in each gene encodes a
short 5′UTR of about 40 bp, and the last exon contains the 3′
end of the open reading frame and a large 3′UTR of about 680
The temporal and spatial expression pattern of DFNA5 has
not been determined. We carried out in situ hybridization to
establish where dfna5 is expressed in the zebrafish. At stages
earlier than 20 hpf, dfna5 is weakly expressed in a ubiquitous
fashion (data not shown). At 20 hpf, it is expressed in the
intermediate cell mass, the olfactory placode, the ventral
diencephalon and the CNC (Fig. 3A,B). At 48 hpf, dfna5
mRNA is present in the brain and in the ear, predominantly in
the projections of the developing semicircular canals (Fig. 3C).
At 55 hpf, expression in the brain decreases, whereas it is
elevated in the ear projections (Fig. 3D,F). We detected dfna5
mRNA at very low levels in the fused ear columns and the brain
at 72 hpf (Fig. 3E). At later stages, dna5 mRNA was no longer
detectable (data not shown). A sense probe did not produce a
signal at any stage (data not shown).
A splice site morpholino directed against exon 8
Individuals with a mutation in DFNA5 carry a complex
insertion/deletion mutation on chromosome 7p15 in intron 7 of
DFNA5, which leads to skipping of exon 8 (Fig. 2A, part a,
red), causing a premature stop codon after an aberrant stretch
of 40 additional amino acids (Van Laer et al., 1998). In order
to target the same exon affected in the human disease, we
designed a morpholino directed against the donor site (‘GT-
MO’) of exon 8 in zebrafish dfna5 (Fig. 2A, part b, red).
Blocking the donor site of an exon can cause skipping of the
targeted exon and/or partial retention of the intron. We traced
efficiency of the dfna5 GT-MO by RT-PCR with primers
flanking the targeted exon on wild type and GT-MO injected
embryos at different time-points (Fig. 2B). RT-PCR on RNA
from 28 hpf GT-MO-injected embryos produced only PCR
products that lacked exon 8 (Fig. 2B, indicated with ∆ exon 8)
Fig. 1. Protein alignment of zebrafish, human and mouse Dfna5. Black shading indicates identity and gray shading indicates similarity. The
overall identity of zebrafish compared with human and mouse Dfna5 is 33% and 30%, respectively. Similarity of zebrafish compared with
human and mouse Dfna5 is 51% and 49%, respectively.
Fig. 2. Target sites of the dfna5 morpholinos and aberrant splicing of dfna5 transcript. (A) Genomic organization of human DFNA5 and
zebrafish Dfna5. The exon/intron boundaries of DFNA5 are entirely conserved between human and zebrafish. (a) Individuals with DFNA5
mutations carry an insertion/deletion mutation in intron 7, which leads to skipping of exon 8 (indicated in red). Absence of exon 8 causes a
frameshift after amino acid 330, resulting in an aberrant stretch of 41 amino acids followed by a premature stop. (b) The two morpholino
antisense oligos designed against Dfna5 mRNA are indicated in red. The splice site antisense oligo (‘GT-MO’) directed against the donor site of
exon 8 leads to skipping of the targeted exon (indicated in red), resembling the human mutation. (B) RT-PCR time course of aberrant splicing
of dfna5 mRNA caused by the dfna5 GT-MO. At 28 hpf, wild-type dfna5 transcript is not detectable by RT-PCR, only PCR products lacking
exon 8 (indicated with ∆ exon 8) or with partial retention of the following intron (arrowhead) are present. Wild-type transcript recovers starting
at day 2, but the morpholino-modulated transcripts are still present at day 7.
or partially retained the downstream intron (Fig. 2B,
arrowhead). Recovery of wild-type dfna5 message started at 48
hpf, but as late as day 7 the GT-MO modulated mRNAs were
still detectable. Sequencing of the PCR product lacking exon
8 showed that the loss of exon 8 did not cause a frameshift in
the dfna5 mRNA. A second morpholino was designed against
the start codon to block translation of the mRNA (‘ATG-MO’)
and thereby knock down the full-length gene.
Knock-down of dfna5 leads to ear and jaw
To determine the function of dfna5 in zebrafish, we analyzed
the phenotype in animals injected with either ATG or GT dfna5
morpholinos. Injection of the ATG-MO (Fig. 4) or the GT-MO
caused malformation of the semicircular canals of the ear.
Either anterior, posterior or both columns of the developing
semicircular canals were not fused (Fig. 4E,G,I). At higher
doses the lateral canal also did not form (Fig. 4J). The
projections, which grow out and fuse to form the columns,
were present, but short and thickened. Other structures of the
ear, like the cristae or maculae, were unaffected.
With increasing ATG-MO dose, starting at 7.5 ng, we
saw an additional phenotype, namely a deformation of
the ventral jaw. Injections of 15 ng of ATG-MO (Fig.
4B) resulted in a mildly affected, smaller ventral jaw; 25
ng of ATG-MO led to a severely malformed jaw (Fig.
4C). Meckel’s cartilage was ventrally displaced, leading
to a widely opened mouth. This phenotype, unfused ear
columns and malformed jaw, is similar to the phenotype
seen in a particular class (Group II) of zebrafish cartilage
differentiation and morphogenesis mutants (Neuhauss et
al., 1996). At doses higher than 30 ng, we obtained
unspecific side effects such as heart edema with the
Development 131 (3)Research article
Fig. 3. Expression of dfna5 in zebrafish embryos using whole-mount
in situ hybridization. (A,B) 22 hpf embryo. Lateral (A) and dorsal
view of the head region (B), showing expression in the intermediate
cell mass (arrowhead in A), the olfactory placodes (arrows), ventral
diencephalon and migrating neural crest (arrows in B). (C,D) 48 hpf
embryo (C) and 55 hpf embryo (D) with expression in the developing
semicircular canals of the ear (arrowheads). (E) 72 hpf embryo
indicating low level of expression in the mature projections of the
semicircular canals in the ear (arrowheads). (F) Dorsal view of the
ear at 48 hpf with expression at the tip of the outgrowing projections
of the semicircular canals (arrowheads). Scale bar: 300 µm in A,B;
100 µm in C-E; 40 µm in F.
Fig. 4. Reduction of dfna5 activity leads to abnormal ear and
cartilage development. (A-C) Lateral views of wild-type (A),
15 ng dfna5 ATG-MO-injected (B) and 25 ng dfna5 ATG-
MO-injected (C) embryos, indicating a dose-dependent
malformation of the lower jaw (arrows). (D,E) Dorsal view of
wild-type (D) and 20 ng dfna5 ATG-MO-injected (E)
embryos at day 5, showing malformation of the anterior and
posterior column in the morphant ear (arrows). (F,G) Close up
of the right ear in D and E, respectively. (G) The anterior
column (ac, arrow) of the morphant is malformed, the
posterior column (pc, large arrows) is interrupted.
(H-J) Lateral view of the ear at day 5 (anterior towards the
left, ventral towards the bottom) of wild type (H), 15 ng dfna5
ATG-MO injected (I) and 25 ng dfna5 ATG-MO injected (J)
embryos. In morphants, either one or more columns do not
fuse (arrows), depending on the injected dose of dfna5 ATG-
MO. ao, anterior otolith; po, posterior otolith. Scale bar:
300 µm in A-C; 250 µm in D,E; 80 µm in F,G; 125 µm in H-
J. ac, anterior column; pc, posterior column.
947The role of dfna5 in ear and jaw developmentDevelopment and disease
ATG-MO (data not shown). In fish injected with the GT-MO
(20 ng-30 ng), we observed only the ear phenotype. A mild
jaw phenotype in addition to the ear malformation (but no
unspecific effects) was obtained with very high doses (40 ng,
data not shown), suggesting that the GT-MO phenotype
represents a hypomorphic situation.
As dfna5 mRNA is highly expressed in the intermediate cell
mass which gives rise to blood cells, we analyzed this tissue
in more detail. We did not see any change in expression of scl,
an early blood marker (data not shown). In addition, staining
of blood cells with o-dianisidine did not reveal any differences
(data not shown).
To exclude the possibility that toxicity of injecting high
amounts of morpholino led to the jaw and ear phenotype, we
examined cell death in MO-injected animals. Embryos injected
with two different doses of ATG-MO (20-40 ng) were stained
with Acridine Orange at 35 hpf and 48 hpf. We did not observe
elevated levels of cell death in any tissue compared with
uninjected embryos (data not shown).
We observed a strong correlation between the amount of MO
injected and the resulting phenotype, indicating the specificity
of the morpholino knockdown of dfna5 (Fig. 5A). In addition,
injection of a 4 bp mismatch control oligonucleotide did not
produce any effects on ear or jaw development (Fig. 5B). To
gain further evidence of specific interference with dfna5
function, we co-injected dfna5 ATG-MO with an expression
construct containing the full-length coding sequence of dfna5
under the control of the cytomegalo virus (CMV) promoter into
one cell-stage wild-type embryos. We found a partial rescue of
the phenotype (Fig. 5C,D) at day 5 in 20% of the embryos with
strongly affected ears and jaw (n=20). In partially rescued
larvae, one ear and the lower jaw showed severe malformations
typical for injection of high doses of ATG-MO, whereas
the other ear was completely normal. A partial rescue was
expected because of the highly mosaic expression of injected
DNA. Injection of ATG-MO alone never led to the above
‘rescued’ phenotype of abnormal jaw morphology combined
with normal ear morphology. Only low doses of ATG-MO or
GT-MO yielded defects in one ear accompanied by normal
morphology in the other ear. However, this unilateral defect
was never present in combination with a jaw defect, as seen in
larvae co-injected with the expression construct.
dfna5 is essential for development of the inner ear
epithelial columns and pharyngeal cartilage
We examined both thick and ultrathin sections of day 5 larval
ears to analyze the morphant phenotype in the columns of the
ear at the cellular level. We stained 5 µm cross-sections of the
lateral canal of the ear at day 6 with Toluidine Blue to
visualize nuclei and membranes (Fig. 6A,B). We detected two
major defects in the GT-MO injected embryos compared with
wild-type embryos. First, the epithelial monolayer of the
developing semicircular canal is disrupted (Fig. 6B, arrows).
The cells appear to be stretched in an apicobasal direction as
opposed to the rather round cells of the wild-type epithelium.
In addition, the cells leave the layer and form stacks, which is
never observed in the mature wild-type column. Second, the
basal lamina of the epithelium is missing or disorganized. In
the Toluidine Blue-stained sections, the basal lamina appears
in the wild type as a thick blue stripe running down the lateral
column (Fig. 6A, arrowhead). In morphant sections, we saw
only a few faint light blue lines (Fig. 6B, arrowhead). TEM
sections revealed that the basal lamina is composed of parallel,
densely packed fibers (Fig. 6C). The morphant basal lamina
consists of much fewer and disorganized fibers, which do not
run parallel to the epithelial sheet, but point in all directions
The zebrafish larval head skeleton can be divided into four
parts, the neurocranium, the mandibular, the hyoid and the
branchial cartilage (Schilling and Kimmel, 1997). We used
Alcian Blue staining to determine which of these elements are
affected in dfna5 ATG-MO-injected larvae. At day 5, the
neurocranium (ethmoid plate, trabeculae, and the anterior and
posterior basicranial commissures) is largely unaffected in
morphant embryos (Fig. 7A,B). The most obvious defect in the
lateral view is the ventrally displaced Meckel’s cartilage, which
leads to the ‘screamer’ phenotype (Fig. 7B). The ventral view
reveals that although the branchial cartilage is present, the
Fig. 5. Specificity of the ATG morpholino-
induced phenotype. (A) Dose dependency of the
observed phenotype. (B) Injection of up to 25 ng
4 bp mismatch morpholino does not cause any
mutant phenotype. (C,D) Partial rescue of the
morphant phenotype by co-injection of dfna5
DNA. Dorsal (C, anterior towards the top) and
lateral (D, anterior towards the right) view of one
d5 larva injected with 25 ng dfna5 ATG-MO and
70 pg dfna5 pCS2+. Left ear (C) and jaw (D)
show malformations typical for injections of
relatively high doses of ATG-MO, whereas the
right ear is rescued and shows no morphological
abnormalities. The anterior and posterior
columns are properly fused, which is never
observed in larvae injected with 25 ng ATG-MO
only. Scale bar: 130 µm in B,C; 200 µm in D.
cartilage from branchial arches 1-4 is strongly reduced
(Fig. 7D,E, indicated with 1-4). The cells of the
ceratobranchials are irregularly arranged as opposed to the
penny stack-like ordered cells in wild-type larvae (Fig. 7C).
The basibranchial cartilage is also strongly reduced in size,
which most likely leads to the posterior displacement of the
ceratohyal cartilage (Fig. 7E, labeled with 2).
The jaw phenotype in dfna5 morphants indicated a potential
defect within the ECM. To determine whether ECM proteins
such as collagen were affected in the dfna5 morphants, we
examined the expression of col2a1, a marker for differentiating
cartilage. At 55 hpf, col2a1 is expressed at equal levels in wild-
type (Fig. 8A) and morphant (Fig. 8D) embryos in the otic
vesicle. In contrast to the developing ear, col2a1 expression
was strongly reduced in the developing pharyngeal cartilage
(Fig. 8D), suggesting a defect in differentiation.
A previously identified zebrafish mutant that shows a
striking similarity to the dfna5 morphants is the jekyll (jek)
mutant. The ear in jek shows a phenotype very similar to dfna5
morphants. The overall structure and size of the ear is normal.
However, although the protrusion of the developing canals
form, they fail to elongate and do not fuse (Neuhauss et al.,
1996). In addition, jek larvae show a strong reduction of all
cartilage in the head. jek encodes Udgh, an enzyme required
for synthesis of proteoglycans including HA. It catalyzes the
formation of glucuronic acid, which (together with N-
acetylglucosamine) forms the repeating disaccharide unit of
hyaluronic acid. To characterize the relationship between
jek/udgh and dfna5, we examined the expression of ugdh in
dfna5 morphants. At 55 hpf, ugdh is highly expressed in the
epithelial columns in the otic vesicle and pharyngeal arches of
the jaw in wild-type embryos (Fig. 8B). In dfna5 morphants,
ugdh expression is absent in the otic vesicle and pharyngeal
arches (Fig. 8E,F). Expression in the developing trabeculae and
ethmoid plate is, however, normal (arrows in Fig. 8C,F). This
is consistent with the finding that the neurocranium is
unaffected in dfna5 morphants.
Hyaluronic acid levels are strongly reduced in dfna5
The biosynthesis of HA depends upon the activities of a
number of enzymes, including Ugdh. We tested whether the
loss of ugdh in the developing semicircular canals affected the
production of HA in these structures. We examined the levels
of HA in vivo using a biotinylated HA-binding protein (Fig. 9).
As shown in Fig. 9C, HA is highly abundant in the semicircular
projections of 54 hpf wild-type larvae. Morphant embryos,
however, have reduced levels of HA in the protrusions (Fig.
9D-F). This strongly suggests that reduced HA levels lead to
the failure of semicircular canal formation in the ear and
supports previous findings that HA is important for projection
Development 131 (3)Research article
Fig. 6. Histological analysis of the inner ear epithelial columns in
dfna5 morphants. Cross-sections of lateral semicircular canals at day
6 of wild-type (A,C) and dfna5 GT-MO injected (B,D) embryos.
(A,B) Toluidine Blue staining of 5 µm sections, indicating a loss of
the monolayer of the epithelium (A,B, arrow) and the basal lamina
(A,B, arrowheads) in morphants (B) compared with wild type (A).
(C,D) Structure of the basal lamina in TEM cross-sections. (C) The
wild-type lamina has tightly packed, parallel fibers. (D) dfna5
morphant basal lamina is disrupted with loose, disorganized fibers.
Scale bar: 20 µm in A,B; 1 µm in C,D.
Fig. 7. Abnormal jaw development
in dfna5 morphants. (A-E) Alcian
Blue staining of cartilage.
(A,C) Lateral (A) and ventral (C)
view of a day 5 wild-type larval
head. (A) Meckel’s cartilage is
indicated with a white arrowhead. In
C, the ceratobranchials are indicated
with 1-5. (B,D,E) Lateral (B) and
ventral (D,E) view of a day 5
morphant larval head. (B) Meckel’s
cartilage (arrowhead) is malformed
and the ceratohyal cartilage is
inverted, most probably owing to
reduction of the branchial cartilages.
(E) Cartilage derived from brachial arches 1-5 are present, but strongly reduced. The ceratobranchial cartilage in morphants is less intensely
stained than in wild-type. Scale bar: 125 µm in A-D; 200 µm in E.
949The role of dfna5 in ear and jaw developmentDevelopment and disease
DFNA5 plays a crucial role in hearing as its loss of function
leads to a progressive form of dominant inherited hearing loss
in humans. Despite a role in deafness, little is known about
the biological function of DFNA5. Using reverse genetics
in zebrafish, we found that dfna5 is required for normal
development of the ear.
dfna5 is essential for semicircular canal formation in
the inner ear and cartilage formation in the jaw
During development of the inner ear, three projections emerge
from the epithelium lining the otic vesicle (Haddon and Lewis,
1996; Waterman and Bell, 1994). These projections or
epithelial columns then grow towards the center of the otic
vesicle where they fuse and form the rudiments of the three
semicircular canals. In larvae injected with either ATG or GT-
MOs directed against dfna5, the columns fail to fuse. Instead,
they appear as short and thickened masses, with a loss of an
epithelial monolayer. As judged by thick sections, the number
of cells is not reduced, but rather the space in between cells
is increased. This phenotype suggests that these cells were
unable to coordinate their movements to form elongated
An additional phenotype occurs in animals injected with the
ATG-MO that is predicted to mimic a loss-of-function allele.
The pharyngeal cartilage is reduced in size and distorted. The
cells stain weakly with Alcian Blue, suggesting that cartilage
differentiation and production of the cartilage-specific
extracellular matrix is disrupted. During differentiation,
chondrocytes express high levels of collagens. Our results
indicate that a collagen highly expressed by chondrocytes,
col2a1, is reduced in dfna5 morphants, suggesting that dfna5
is required for differentiation of chondrocytes.
The role of HA in the developing ear
The phenotype seen in dfna5 morphants is similar to the
phenotype observed in zebrafish jekyll mutants (Neuhauss et
al., 1996). These loss-of-function mutants have short and
thickened epithelial columns that fail to elongate and fuse and
cartilage differentiation is also affected. The jekyll gene
encodes Ugdh, an enzyme necessary for the production of the
dissaccharide unit of HA (Walsh and Stainier, 2001). Based on
the similar phenotypes, we hypothesized that dfna5 may be
involved in the ugdh pathway. In MO-injected animals, we find
that loss of dfna5 function leads to loss of ugdh expression in
the ear and the developing pharyngeal arches. We also find that
HA is reduced in the developing semicircular canals of the
dfna5 morphants. Moreover, loss of HA is associated with
reduced levels of col2a1 expression, affecting differentiation
of the branchial cartilage. By contrast, col2a1 expression in
morphant ears is normal, supporting our hypothesis that loss
of ugdh in the developing ear is not affecting differentiation of
Fig. 8. In situ analysis of cartilage
differentiation in dfna5 morphants.
(A,D) Lateral view of col2a1 expression at 55
hpf. White arrows indicate the otic vesicle;
white arrowheads indicate pharyngeal arches.
Expression in the morphant otic vesicle (D) is
unaffected compared with wild type (A).
(B,C,E,F) ugdh expression in wild-type (B,C)
and morphant (E,F) 55 hpf embryos.
(B,E) Lateral view. (E) Expression in the
developing morphant ear columns (indicated on
wild type) and pharyngeal arches (white
arrowhead) is reduced compared with wild-type
(B). (C,F) Ventral view of expression in the
developing neurocranium reveals no difference
between wild-type and morphant embryos
(arrows in C and F). Scale bar: 200 µm.
c, column; e, eye; fb, fin bud; nc, neurocranium;
ov, otic vesicle; pa, pharyngeal arches.
Fig. 9. Reduction of hyaluronic acid (HA) in the
developing semicircular canals in dfna5 morphants
(54 hpf larvae). (A,D) Outline of the structures in a
dorsolateral view of wild-type (B,C) and dfna5
ATG-MO injected larval ears (E,F). The
differential interference contrast (DIC) and
corresponding fluorescent images are focussed at
the level of the anterior column. An outgrowing
protrusion of the anterior column (ac) in the wild-
type ear (B) is filled with HA as shown by staining
with biotinylated HA-binding protein (C). The
lateral (lc) and posterior columns (pc) are out of
focus in C. HA is reduced in both protrusions of
the anterior column in a dfna5 ATG-MO injected
larval ear (E,F). Scale bar: 35 mm.
the ear cavity, but only the directed outgrowth of the canal
In the ear, HA has been shown to play a role in the outgrowth
of projections that fuse to form columns (Haddon and Lewis,
1991). HA is made of variable numbers of disaccharide units
capable of forming filaments up to several micrometers in
length. In the growing columns, epithelial cells at the tip of
each protrusion are thought to secrete these long molecules that
can act as a ‘propellant’. Secretion of HA by a subset of cells
within the projection may provide a driving force for growth
in a particular direction. When HA is enzymatically removed
in the epithelial columns, a similar phenotype of defective
outgrowth is seen (Haddon and Lewis, 1991). In dfna5
morphants, the reduction of HA levels results in an
uncoordinated outgrowth of the epithelial projections. The
epithelial cells are still proliferating, but the projection appears
to lack the directional force of HA secretion by the cells at the
leading edge. This causes disorganization of the epithelium as
reflected by the missing basal lamina and stacking of cells.
Controlled breakdown of the basal lamina underlying the
outpocketing epithelium appears to be crucial for proper
outgrowth. In wild-type canal projections, only the cells at the
leading edge of the protrusion lack a basal lamina. In netrin 1
mouse mutants, the opposite situation is found: the basal
lamina does not break down at all. However, this defect also
results in unfused columns (Salminen et al., 2000), suggesting
that only spatially and temporally correct remodeling of the
basal lamina allows proper outgrowth and fusion of the
Members of the protein family that share motifs with
DFNA5 that are described in the literature are cancer-
associated (Saeki et al., 2000; Thompson and Weigel,
1998; Lage et al., 2001; Watabe et al., 2001) (see
DFNA5 family). This is particularly compelling because high
levels of HA are a prognostic indicator for malignancy in
clinical cases of human breast, ovarian and colon carcinomas.
HA may promote invasiveness of cancer cells by providing a
highly hydrated microenvironment that facilitates detachment
and migration of cells (Toole, 2002). One possibility is that
proteins of this family are able to regulate HA biosynthesis.
All members of this protein family notably contain one (in the
case of DFNA5) or more putative HA-binding sites. This motif
is called BX7B domain because it contains two basic amino
acids (arginine or lysine) separated by seven non-acidic amino
acids (Yang et al., 1994). This domain is found in all
hyaladherins, proteins that interact with HA, such as aggrecan
or CD44. Because this is a very common motif, it is not clear
whether this domain is sufficient for binding to HA.
If Dfna5 regulates the HA biosynthetic pathway, it may act
at the transcriptional level. Evidence for this notion is
supported by a recent heterologous expression study in which
human DFNA5 was expressed in S. pombe cells (Gregan et al.,
2003). Gregan and coworkers found a genetic interaction with
the S. pombe gene mcm10 (cdc23), which is involved in DNA
replication. Interestingly, both DFNA5 and Mcm10p contain
similar zinc-finger-like motifs that define the Mcm10 family of
DNA replication proteins.
Human DFNA5 hearing loss
How does our study add to the understanding of the human
DFNA5 disease? The anatomy of the vestibular inner ear,
including the semicircular canals, is highly conserved among
vertebrates (Riley and Phillips, 2003). Certain aspects of the
development and morphogenesis of semicircular circular
canals are similar. For example, in fish, frogs, birds and
mammals, apposing epithelial walls or protrusions contact and
fuse to form the rudiments of the semicircular canals (Haddon
and Lewis, 1991; Martin and Swanson, 1993; Haddon and
Lewis, 1996; Fekete et al., 1997). Development of the
pharyngeal arches is also conserved among vertebrates
(Graham, 2003). However, individuals with a mutation in
DFNA5 display neither obvious vestibular defects nor
craniofacial malformations. Only one family has been found
with a single mutation in DFNA5 thus far. Whether the
truncation of DFNA5 in humans results in haplo-insufficiency
remains to be determined. Typically, only severe mutations
such as null mutations cause syndromic hearing loss (Astuto
et al., 2002). Syndromic deafness associated with craniofacial
and/or skeletal deformations have been reported for three
genes, COL2A1, COL11A1 and COL11A2 (reviewed by
Morton, 2002). All three collagens are highly abundant in
cartilage extracellular matrix (de Crombrugghe, 2001).
Mutations in COL11A2 cause syndromic as well as
nonsyndromic hearing loss: congenital hearing loss is
accompanied by various skeletal abnormalities in the case of
Stickler syndrome (which can also be caused by mutations in
COL2A1 and COL11A1), whereas individuals with a mutation
in DFNA13 suffer from nonsyndromic progressive hearing loss
at high and middle frequencies (Sirko-Osadsa, 1998; McGuirt
et al., 1999; Kunst et al., 2000). The Marshall syndrome
is associated with splicing mutations in COL11A1 and
characterized by craniofacial and skeletal abnormalities,
cataracts and progressive sensorineural hearing loss. Computer
tomography did not detect any malformations of ear bones.
Therefore, hearing loss is thought to be due to direct effects on
COL11A1 loss in the labyrinth and CNS (Griffith et al., 2000).
In all three cases, genes affecting cartilage differentiation can
cause progressive sensorineural hearing loss without causing
cartilage malformations in the human ear.
This leads us to speculate that a complete loss of DFNA5
function in humans could cause cartilage phenotypes as seen
in individuals with Stickler syndrome. Determination of the
expression pattern of Dfna5 in mammals will shed some light
on this paradox. Nevertheless, HA is highly abundant in many
areas of the developing human inner ear. It is suggested to serve
there as a friction-reducing lubricant and molecular filter
(Anniko and Arnold, 1995). If HA biosynthesis is reduced or
lost in the ear, it is possible that this disruption of extracellular
matrix causes increased mechanical stress on hair cells. This
stress may lead to premature aging of hair cells and a
progressive hearing loss, reminiscent of age-related hearing
loss that also starts with high frequency sound.
Abdelhak, S., Kalatzis, V., Heilig, R., Compain, S., Samson, D., Vincent, C.,
Weil, D., Cruaud, C., Sahly, I., Leibovici, M. et al. (1997). A human
homologue of the Drosophila eyes absent gene underlies branchio-oto-renal
(BOR) syndrome and identifies a novel gene family. Nat. Genet. 15, 157-
Ahmad, N. N., Ala-Kokko, L., Knowlton, R. G., Jimenez, S. A., Weaver,
E. J., Maguire, J. I., Tasman, W. and Prockop, D. J. (1991). Stop codon
Development 131 (3)Research article
951 The role of dfna5 in ear and jaw developmentDevelopment and disease
in the procollagen II gene (COL2A1) in a family with the Stickler syndrome
(arthro-ophthalmopathy). Proc. Natl. Acad. Sci. USA 88, 6624-6627.
Anniko, M. and Arnold, W. (1995). Hyaluronic acid as a molecular filter and
friction-reducing lubricant in the human inner ear. ORL J. Otorhinolaryngol.
Relat. Spec. 57, 82-86.
Bitner-Glindzicz, M. and Tranebjaerg, L. (2000). The Jervell and Lange-
Nielsen syndrome. Adv. Otorhinolaryngol. 56, 45-52.
de Crombrugghe, B., Lefebvre, V. and Nakashima, K. (2001). Regulatory
mechanisms in the pathways of cartilage and bone formation. Curr. Opin.
Cell Biol. 13, 721-727.
Fekete, D. M., Homburger, S. A., Waring, M. T., Riedl, A. E. and Garcia,
L. F. (1997). Involvement of programmed cell death in morphogenesis of
the vertebrate inner ear. Development 124, 2451-2461.
Graham, A. (2003) Development of the pharyngeal arches. Am. J. Med. Genet.
Gregan, J., van Laer, L., Lieto, L. D., van Camp, G. and Kearsey, S. E.
(2003). A yeast model for the study of human DFNA5, a gene mutated in
nonsyndromic hearing impairment. Biochim Biophys Acta. 1638, 179-186.
Griffith, A. J., Gebarski, S. S., Shepard, N. T. and Kileny, P. R. (2000).
Audiovestibular phenotype associated with a COL11A1 mutation in
Marshall syndrome. Arch. Otolaryngol. Head Neck Surg. 126, 891-894.
Haddon, C. M. and Lewis, J. H. (1991). Hyaluronan as a propellant for
epithelial movement: the development of semicircular canals in the inner ear
of Xenopus. Development 112, 541-550.
Haddon, C. and Lewis, J. (1996). Early ear development in the embryo of
the zebrafish, Danio rerio. J. Comp. Neurol. 365, 113-128.
Kimmel, C. B., Miller, C. T. and Moens, C. B. (2001). Specification and
morphogenesis of the zebrafish larval head skeleton. Dev. Biol. 233, 239-257.
Knudson, C. B. and Knudson, W. (2001). Cartilage proteoglycans. Semin.
Cell Dev. Biol. 12, 69-78.
Kunst, H., Huybrechts, C., Marres, H., Huygen, P., van Camp, G. and
Cremers, C. (2000). The phenotype of DFNA13/COL11A2: nonsyndromic
autosomal dominant mid-frequency and high-frequency sensorineural
hearing impairment. Am. J. Otol. 21, 181-187.
Lage, H., Helmbach, H., Grottke, C., Dietel, M. and Schadendorf, D.
(2001). DFNA5 (ICERE-1) contributes to acquired etoposide resistance in
melanoma cells. FEBS Lett. 494, 54-59.
Leger, S. and Brand, M. (2002). Fgf8 and Fgf3 are required for zebrafish ear
placode induction, maintenance and inner ear patterning. Mech. Dev. 119,
Liu, D., Chu, H., Maves, L., Yan, Y. L., Morcos, P. A., Postlethwait, J. H.
and Westerfield, M. (2003). Fgf3 and Fgf8 dependent and independent
transcription factors are required for otic placode specification. Development
Magee, C., Nurminskaya, M. and Linsenmayer, T. F. (2001). UDP-glucose
pyrophosphorylase: up-regulation in hypertrophic cartilage and role in
hyaluronan synthesis. Biochem. J. 360, 667-674.
Maroon, H., Walshe, J., Mahmood, R., Kiefer, P., Dickson, C. and Mason,
I. (2002). Fgf3 and Fgf8 are required together for formation of the otic
placode and vesicle. Development 129, 2099-2108.
Martin, P. and Swanson, G. J. (1993). Descriptive and experimental analysis
of the epithelial remodellings that control semicircular canal formation in
the developing mouse inner ear. Dev. Biol. 159, 549-558.
McGuirt, W. T., Prasad, S. D., Griffith, A. J., Kunst, H. P., Green, G. E.,
Shpargel, K. B., Runge, C., Huybrechts, C., Mueller, R. F., Lynch, E. et
al. (1999). Mutations in COL11A2 cause non-syndromic hearing loss
(DFNA13). Nat. Genet. 23, 413-419.
Morton, C. C. (2002). Genetics, genomics and gene discovery in the auditory
system. Hum. Mol. Genet. 11, 1229-1240.
Neuhauss, S. C., Solnica-Krezel, L., Schier, A. F., Zwartkruis, F., Stemple,
D. L., Malicki, J., Abdelilah, S., Stainier, D. Y. and Driever, W. (1996).
Mutations affecting craniofacial development in zebrafish. Development
Petit, C., Levilliers, J. and Hardelin, J. P. (2001). Molecular genetics of
hearing loss. Annu. Rev. Genet. 35, 589-646.
Piotrowski, T., Schilling, T. F., Brand, M., Jiang, Y. J., Heisenberg, C. P.,
Beuchle, D., Grandel, H., van Eeden, F. J., Furutani-Seiki, M., Granato,
M. et al. (1996). Jaw and branchial arch mutants in zebrafish II: anterior
arches and cartilage differentiation. Development 123, 345-356.
Riley, B. B. and Phillips, B. T. (2003). Ringing in the new ear: resolution of
cell interactions in otic development. Dev. Biol. 261, 289-312.
Saeki, N., Kuwahara, Y., Sasaki, H., Satoh, H. and Shiroishi, T. (2000).
Gasdermin (Gsdm) localizing to mouse Chromosome 11 is predominantly
expressed in upper gastrointestinal tract but significantly suppressed in
human gastric cancer cells. Mamm. Genome 11, 718-724.
Salminen, M., Meyer, B. I., Bober, E. and Gruss, P. (2000). Netrin 1 is
required for semicircular canal formation in the mouse inner ear.
Development 127, 13-22.
Schilling, T. F. and Kimmel, C. B. (1994). Segment and cell type lineage
restrictions during pharyngeal arch development in the zebrafish embryo.
Development 120, 483-494.
Schilling, T. F. and Kimmel, C. B. (1997). Musculoskeletal patterning in the
pharyngeal segments of the zebrafish embryo. Development 124, 2945-
Schilling, T. F., Piotrowski, T., Grandel, H., Brand, M., Heisenberg, C. P.,
Jiang, Y. J., Beuchle, D., Hammerschmidt, M., Kane, D. A., Mullins, M.
C. et al. (1996). Jaw and branchial arch mutants in zebrafish I: branchial
arches. Development 123, 329-344.
Sirko-Osadsa, D. A., Murray, M. A., Scott, J. A., Lavery, M. A., Warman,
M. L. and Robin, N. H. (1998). Stickler syndrome without eye involvement
is caused by mutations in COL11A2, the gene encoding the alpha2(XI)
chain of type XI collagen. J. Pediatr. 132, 368-371.
Spicer, A. P., Kaback, L. A., Smith, T. J. and Seldin, M. F. (1998). Molecular
cloning and characterization of the human and mouse UDP-glucose
dehydrogenase genes. J. Biol. Chem. 273, 25117-25124.
Thisse, C., Thisse, B., Schilling, T. F. and Postlethwait, J. H. (1993).
Structure of the zebrafish snail1 gene and its expression in wild-type,
spadetail and no tail mutant embryos. Development 119, 1203-1215.
Thompson, D. A. and Weigel, R. J. (1998). Characterization of a gene that
is inversely correlated with estrogen receptor expression (ICERE-1) in
breast carcinomas. Eur. J. Biochem. 252, 169-177.
Toole, B. P. (2001). Hyaluronan in morphogenesis. Semin. Cell Dev. Biol. 12,
Van Laer, L., Huizing, E. H., Verstreken, M., van Zuijlen, D., Wauters, J.
G., Bossuyt, P. J., van de Heyning, P., McGuirt, W. T., Smith, R. J.,
Willems, P. J. et al. (1998). Nonsyndromic hearing impairment is associated
with a mutation in DFNA5. Nat. Genet. 20, 194-197.
Walsh, E. C. and Stainier, D. Y. (2001). UDP-glucose dehydrogenase
required for cardiac valve formation in zebrafish. Science 293, 1670-1673.
Watabe, K., Ito, A., Asada, H., Endo, Y., Kobayashi, T., Nakamoto, K.,
Itami, S., Takao, S., Shinomura, Y., Aikou, T. et al. (2001). Structure,
expression and chromosome mapping of MLZE, a novel gene which is
preferentially expressed in metastatic melanoma cells. Jpn. J. Cancer Res.
Waterman, R. E. and Bell, D. H. (1984). Epithelial fusion during early
semicircular canal formation in the embryonic zebrafish, Brachydanio rerio.
Anat. Rec. 210, 101-114.
Yang, B., Yang, B. L., Savani, R. C. and Turley, E. A. (1994). Identification
of a common hyaluronan binding motif in the hyaluronan binding proteins
RHAMM, CD44 and link protein. EMBO J. 13, 286-296.