Human Molecular Genetics, 2006, Vol. 15, No.

17 2569–2587
doi:10.1093/hmg/ddl184
Advance Access published on July 18, 2006

Pathogenetic role of the deafness-related M34T

mutation of Cx26
Massimiliano Bicego1,{, Martina Beltramello2,{, Salvatore Melchionda3, Massimo Carella3,
Valeria Piazza2, Leopoldo Zelante3, Feliksas F. Bukauskas4, Edoardo Arslan5, Elona Cama5,
Sergio Pantano2,7, Roberto Bruzzone6,*, Paola D’Andrea1,{ and Fabio Mammano2,7,8,{
1
Dipartimento di Biochimica, Biofisica e Chimica delle Macromolecole, University of Trieste, 34127 Trieste, Italy,
2
Istituto Veneto di Medicina Molecolare (VIMM), Fondazione per la Ricerca Biomedica Avanzata, 35129 Padova, Italy,
3
Servizio di Genetica Medica, IRCCS-Ospedale Casa Sollievo della Sofferenza, San Giovanni Rotondo, Italy,
4
Department of Neuroscience, Albert Einstein College of Medicine, Bronx, NY 10461, USA, 5Servizio di Audiologia e
Foniatria, University of Padova, 35128 Padova, Italy, 6Département de Neuroscience, Institut Pasteur, 75015 Paris,
France and 7Consorzio Nazionale Interuniversitario per le Scienze Fisiche della Materia (CNISM) and 8Dipartimento di
Fisica ‘G.Galilei’, Università di Padova, 35131 Padova, Italy

Received April 21, 2006; Revised July 5, 2006; Accepted July 12, 2006

Mutations in the GJB2 gene, which encodes the gap junction protein connexin26 (Cx26), are the major cause
of genetic non-syndromic hearing loss. The role of the allelic variant M34T in causing hereditary deafness
remains controversial. By combining genetic, clinical, biochemical, electrophysiological and structural mod-
eling studies, we have re-assessed the pathogenetic role of the M34T mutation. Genetic and audiological data
indicate that the majority of heterozygous carriers and all five compound heterozygotes exhibited an
impaired auditory function. Functional expression in transiently transfected HeLa cells showed that,
although M34T was correctly synthesized and targeted to the plasma membrane, it inefficiently formed inter-
cellular channels that displayed an abnormal electrical behavior and retained only 11% of the unitary conduc-
tance of the wild-type protein (HCx26wt). Moreover, M34T channels failed to support the intercellular
diffusion of Lucifer Yellow and the spreading of mechanically induced intercellular Ca21 waves. When
co-expressed together with HCx26wt, M34T exerted dominant-negative effects on cell – cell coupling.
Our findings are consistent with a structural model, predicting that the mutation leads to a constriction of
the channel pore. These data support the view that M34T is a pathological variant of Cx26 associated with
hearing impairment.

INTRODUCTION association of mutations in several human connexins with a

Connexins comprise a multi-gene family of proteins that
form the intercellular channels clustered at gap junctions,
through which ions, small metabolites and second messen-
gers are exchanged between connected cells (1). Intercellular
channels are formed by two half-channels, or connexons,
consisting of six connexin subunits that span the plasma
membranes of communicating cells and dock in the intercel-
lular space (2). The specific role of connexin channels in the
homeostasis of different organs has been illustrated by the
variety of genetic diseases and by the specific phenotypes
revealed by targeted connexin gene deletion in mice (3 – 5).
In the inner ear, three connexin isoforms have been found
to be expressed in overlapping patterns (6) and their
crucial role in organ physiology has been revealed by their
implication in different forms of hereditary deafness (7).
Thus, mutations in human Cx26 (HCx26), HCx30 and
HCx31 (GJB2, GJB6 and GJB3 genes, respectively) have
been linked to both syndromic and non-syndromic forms of
hearing loss (reviewed in 8).

*To whom correspondence should be addressed at: Département de Neuroscience, Institut Pasteur, 25, rue du Dr Roux, 75015 Paris, France.
Tel: þ33 140613436; Fax: þ33 140613421; Email: bruzzone@pasteur.fr
{
These authors have contributed equally.

# The Author 2006. Published by Oxford University Press. All rights reserved.
For Permissions, please email: journals.permissions@oxfordjournals.org
2570 Human Molecular Genetics, 2006, Vol. 15, No. 17
The arrangement of gap junctions between supporting cells
of the organ of Corti appears to be the structural basis for the
re-circulation of cochlear Kþ ions that flow from the endo-
lymph, where their concentrations are 150 mM , into hair
cells upon sound stimulation (9). In this scheme, Kþ ions
flow into the hair cells through mechanically gated channels
by the combined effect of the positive endolymphatic potential
(þ80 mV) and the negative intracellular potential and
depolarize them. Kþ ions exit hair cells reaching the intersti-
tial space of the organ of Corti, where they are partly taken
up by cochlear supporting cells (10). The excess Kþ concen-
tration that results from auditory signals or sustained deleter-
ious stimuli, such as ischemia or prolonged exposure to high
sound pressure levels (reviewed in 11), is thought to be
removed via an intercellular pathway delineated by intercellu-
lar channels composed of Cx26 and Cx30 that provide a
spatial buffering network similar to that of astrocytic gap junc-
tions in the brain (12). Genetic studies have demonstrated that
a specific single-base deletion in Cx26, named 35delG, is the
most common mutation worldwide, although its prevalence
varies with ethnical groups (8). In addition, several missense
mutations responsible for either recessive or dominant forms
of the disease have been identified (for an updated list, see
Rabionet et al. (2002) http://www.iro.es/cx26deaf.html). The
first deafness-related Cx26 mutation, which was detected in a
dominant pedigree of non-syndromic hearing loss, was a
methionine-to-threonine substitution at position 34 (M34T) (13).
The pathogenetic role of M34T was initially supported by
the results of its functional expression in Xenopus oocytes,
which showed a dominant-negative effect of the M34T
mutant (14). Further studies on the family first described by
Kelsell et al. (13), however, uncovered the association of der-
matological signs in deaf patients and identified another domi-
nant mutation in GJB2 segregating with the disease, casting
doubts on the significance of the M34T variant. Subsequently,
the same M34T substitution was also reported to be respon-
sible for a recessive form of hearing loss (15), whereas
several authors described normal hearing in heterozygous
carriers of M34T associated with either 35delG or other
Cx26 recessive mutations (16 – 19) and, therefore, considered
it a benign polymorphism.
More recent in vitro data have shown that in transfected
HeLa cells, the M34T variant disrupts either channel for-
mation or its oligomerization (20,21). In contrast, Thonnissen
et al. (22) observed low levels of dye transfer between HeLa
cells expressing M34T, providing the first evidence that this
mutant could traffic to the cell membrane and form intercellu-
lar channels, although with reduced efficiency. More recently,
Oshima et al. (23) reported that assembly of M34T in HeLa
and Sf9 cells resembles that of wild-type human Cx26
(HCx26wt) and that dye transfer in these cells is close to
normal. Finally, electrophysiological studies performed in
paired Xenopus oocytes concluded that M34T was capable
of forming with Cx32 functional heterotypic channels with
abnormal gating properties, which allowed to suggest that
the channels are not fully open at rest but are activated
when positive transjunctional voltages are applied to the
M34T side (24). In addition, co-injection of M34T and
HCx26wt in oocytes supported a recessive role for the
M34T allele, as a 1:1 ratio of mutant to wild-type RNA
(as found in heterozygous individuals) did not significantly
reduce wild-type coupling.
Given the apparent contradictions of these reports, we
sought to re-assess the pathogenetic role of the M34T
variant by combining genetic, auditory, biochemical, electro-
physiological and structural modeling studies. We provide
here genetic and audiological evidence that although heterozy-
gous carriers are found among normal hearing subjects, the
majority of them (62.5%) exhibited a mild to moderate
hearing impairment. Furthermore, all compound heterozygotes
were affected by hearing loss, supporting the view that M34T
is a pathological variant. We also show that M34T is correctly
synthesized and targeted to HeLa cells membrane, although it
forms intercellular channels that retain an abnormal electrical
activity. Interestingly, these channels fail to sustain the inter-
cellular diffusion of Lucifer Yellow (LY) and largely inhibit
the spreading of intercellular Ca2þ waves. Finally, when
co-expressed with HCx26wt at a 1:1 protein ratio, M34T
invariably shows dominant-negative effects on cell – cell com-
munication. These results fit well with our structural model,
predicting that the mutation induces a constriction of the
channel pore.
RESULTS
Genetic and audiological features
We have studied seven families in which one or more
hearing-impaired members had the M34T mutation. Besides
carrying out the routine genetic tests, we examined the audio-
logical features of all members of the families to disclose
even mild auditory defects potentially associated with the
mutation. Cx26 mutations in heterozygous or compound het-
erozygous condition were found in 21 patients, without any
instance of homozygosity (Fig. 1). The M34T mutation was
identified in 16 individuals, either as heterozygotes or com-
pound heterozygotes. Moreover, in 14 of them, we found
the presence of a 10 bp deletion in the 50 -untranslated
region of the gene (2493del10), in cis with M34T. This del-
etion was absent in all other individuals enrolled in this study.
Patients were divided into three groups: group I, consisting of
patients negative for GJB2/GJB6 mutations; group II, which
included patients carrying mutations different from the
M34T and group III, with patients carrying at least the
M34T mutation.
In group I, hearing loss was present only in one patient
(Fig. 1; family 5, subject I.2), whose hearing threshold was
higher than that of controls. The other six exhibited normal
audiological features and their hearing thresholds were well
in line with those of age- and sex-matched controls. A flat
audiogram was recorded in four subjects, whereas high-
frequency sloping and low-frequency ascending curves were
observed in the other two (Fig. 2).
In group II, three subjects (all heterozygotes) had mild
hearing loss (delGJB6-D13S1830/wt, L90P/wt, 35delG/wt).
In two of them, males having 64 and 67 years (family 2,
subject I.2 and family 3, subject I.2; Fig. 1), hearing impair-
ment was similar to that of the appropriate control group,
whereas in the third one (delGJB6-D13S1830/wt), hearing
threshold was higher than expected across all frequencies.
 Human Molecular Genetics, 2006, Vol. 15, No. 17 2571

Figure 1. Pedigrees and genotypes of the seven families studied. Individuals with hearing loss are shown in black, whereas those with normal audiological
parameters are in white. Gray symbols indicate individuals in which at least one frequency threshold was impaired; asterisks denote individuals with a
degree of hearing loss that matches the one expected by age and sex.

Finally, the two remaining subjects (35delG/wt) exhibited
normal hearing. Flat audiograms were found in two patients,
whereas the other three were of the high-frequency steeply
sloping type. In one of the patients with a mild phenotype
(delGJB6-D13S1830/wt; family 1, subject I.1), hearing loss
was asymmetrical.
In group III, five M34T/wt and five compound heterozy-
gotes were hearing impaired. Mild hearing loss was diagnosed
in four patients (one M34T/wt, one M34T/35delG, one M34T/
delGJB6-D13S1830 and one M34T/L90P). Moderate hearing
loss was present in five patients (four M34T/wt genotypes
and one M34T/35delG), whereas severe hearing loss, in the
only hearing ear, was observed in one patient with an
M34T/delGJB6-D13S1830 genotype (Fig. 1; family 1,
subject II.1). In all M34T/wt hearing impaired patients,
thresholds across all frequencies were greater than that
expected for an age- and sex-matched control population
(Fig. 2). Even among normal hearing [PTA ,20 dB hearing
level (HL)] M34T heterozygotes (six cases), thresholds were
greater than that in control subjects for at least one frequency
in three patients (family 3, subject I.2; family 6, subject I.2;
family 7, subject I.1). Five patients had flat audiograms and
10 showed high-frequency sloping curves (six cases gently
sloping and four steeply sloping). A low-frequency ascending
audiogram was observed in only one patient (family 5, subject
II.1). Among the 16 patients carrying the M34T mutation, six
(five of which were compound heterozygotes) had asymmetri-
cal hearing loss.
2572 Human Molecular Genetics, 2006, Vol. 15, No. 17

Figure 2. Audiometric analysis of the seven families studied. Air conduction thresholds were measured across the indicated frequencies, and the degree of
hearing impairment was classified by the pure-tone average applied to the better ear. Among M34T-carrying patients, five had flat audiograms, all of which
showed the M34T/wt genotype, (families 2, 4, 6) and 10 showed high-frequency sloping curves (six gently sloping: family 1, M34T/delGJB6-D13S1830 geno-
type with the poorer threshold; family 2, M34T/L90P genotype; family 3, M34T/wt; families 6 and 7, with one and two members showing M34T/wt genotype,
respectively; four steeply sloping: family 1, the other subject with M34T/delGJB6-D13S1830 genotype and the subject M34T/wt; families 3 and 4, males with
the M34T/35delG genotype). For each individual, audiograms of the better ear are shown.

Protein expression and localization
To analyze the functional impact of the M34T mutation, which
is positioned in the first transmembrane domain (TM1) of Cx26,
we have compared its channel-forming ability with that of the
recessive R127H (25), which maps to the cytoplasmic loop
(Fig. 3A). We chose these two variants because previous
studies indicated that both mutated proteins form plaques in
the junctional membrane (20,22,26). To examine to which
extent these mutations affected the level of protein expression
and their cellular localization, HCx26wt and these two variants
were transiently transfected in HeLa cells, which were

Figure 3. Topology, expression and localization of HCx26wt and of the two
variants, M34T and R127H. (A) Schematic topology of Cx26 relative to the
plasma membrane, with the approximate distribution of the mutations
studied. (B) Western blot analysis of the expression of wild-type and
mutated Cx26 proteins in HeLa cells. Cells were transfected with the specified
constructs, lysed in SDS sample buffer 24 h post-transfection and further pro-
cessed for immunoblotting, as described in the Materials and Methods section.
A representative blot shows that the levels of expression of the mutated pro-
teins were comparable with those of HCx26wt. (C) Immunolocalization of
HCx26wt and deafness-associated mutations. HeLa cells were transfected
with the specified constructs, fixed and labeled with an anti-Cx26 antibody
24 h post-transfection, as described in the Materials and Methods section. A
secondary antibody conjugated with tetramethylrhodamine isothiocyanate
was used for visualization. GFP: Expression of GFP in transfected HeLa
cells shows a diffuse cytoplasmic staining. Cx26: Immunolocalization of
Cx26 in GFP-expressing cells shows a characteristic punctate staining, but
also the presence of cytoplasmic fluorescence in the case of M34T.
Bar ¼ 10 mm.
processed for western blot and immunofluorescence analysis
24 h post-transfection. Western blots performed on total cellu-
lar lysates demonstrated that proteins of the predicted molecular
mass were detected in each experimental condition (Fig. 3B).
Levels of R127H appeared comparable with those of wild-type
protein, whereas M34T showed slightly increased expression,
thus confirming our previous quantitative analysis (20).
Transfection with HCx26wt produced, in green fluorescent
Human Molecular Genetics, 2006, Vol. 15, No. 17 2573
protein (GFP) positive cells, a dashed, membrane-associated
staining typical of gap-junctional plaques, indicating that it
was correctly targeted to the plasma membrane (Fig. 3C),
whereas in cells transfected with vector alone, no staining
was detected (20). As previously shown (20,22,23), the M34T
mutant exhibited not only a strong staining at the plasma mem-
brane, but also the presence of cytoplasmic fluorescence, which
was suggestive of partial intracellular retention (Fig. 3C). The
immunolocalization of the R127H mutation gave results
similar to HCx26wt, with junctional plaques present at zones
of cell-to-cell apposition (Fig. 3C).
Biophysical properties
To compare the electrophysiological behavior of HCx26wt
and the M34T mutant, experiments were initially carried out
on cultures in which the two cDNAs were expressed through
the pIRES construct (see Materials and Methods). Cell pairs
expressing HCx26wt were held at a transjunctional voltage
(Vj) of 295 mV, under uncoupling conditions in a medium
saturated with 100% CO2 (27). Once junctional current (Ij)
was reduced to below-detectable levels, the superfusion
medium was switched back to normal extracellular solution
(ECS) to allow Ij to recover, and a series of discrete events
of similar amplitude, reflecting the opening of junctional chan-
nels, was observed (Fig. 4A). Off-line analysis yielded a
unitary conductance (g) of 114 pS for the fully open state.
In contrast, when the M34T mutant was challenged under
the same uncoupling conditions with a sustained Vj of
2100 mV, gap junction channels tended to open more
rapidly, which hampered the detection of discrete events.
However, in some cases, small current transitions were suffi-
ciently well resolved to define them as unitary gating events
of 13 pS in magnitude (Fig. 4B). Therefore, single-channel
conductance of M34T is substantially (9-fold) smaller than
that of HCx26wt (Fig. 4A and B). Although such small
signals were at the limit of resolution for our dual whole-
cell recording conditions, this analysis indicates that the
M34T variant retained partial activity and that functional
homotypic/homomeric channels were assembled.
Interactions between connexins may lead to the formation
of heterotypic channels assembled by docking of two homo-
meric hemichannels formed from different connexins (1). To
study the behavior of heterotypic channels, we used cells
expressing chimeric proteins in which color variants of GFP
were fused at the C-terminus of the specified constructs.
When HeLa cell pairs transfected with HCx26wt-Venus/L1
recovered from CO2-induced uncoupling with a sustained Vj
of 295 mV, unitary gating events indicative of the opening
of single-junctional channels yielded a g of 115 pS for the full-
open state (Fig. 5A). This initial characterization indicates that
the electrophysiological behavior of the tagged protein was
virtually indistinguishable from that of HCx26wt.
Because Cx26 and Cx30 are co-localized in gap junctions of
the inner ear (28,29), we next determined the biophysical
properties of HCx30wt homotypic and HCx26wt/HCx30wt
heterotypic channels. Pairs of HeLa cells transiently trans-
fected with HCx30wt-CFP/L1 displayed single-channel open-
ings when subjected to a Vj of 90 mV (Fig. 5B). Transitions
occurred among the closed state, the main conductance state
2574 Human Molecular Genetics, 2006, Vol. 15, No. 17

Figure 4. Electrical properties of HCx26wt and M34T mutant. (A) Conduc-

tance changes (gj, top trace) and corresponding discrete current transitions
(Ij, bottom trace) due to gating of homotypic channels formed by HCx26wt.
Several opening and closing of gap junction channels were recorded with
a transjunctional voltage (Vj) of 295 mV (V1 ¼ 25 mV; V2 ¼ 2100 mV;
n ¼ 4). The calculated unitary conductance (g) was 114 pS. (B) Conductance
changes (gj, top trace) and corresponding rapid current transitions (Ij, bottom
trace) due to opening of homotypic channels formed by M34T mutant
recorded under a sustained Vj of 2100 mV (V1 ¼ 0 mV; V2 ¼ 2100 mV;
n ¼ 2). The estimated g was 13 pS.

with gmain ¼ 160 pS and a sub-conductance state with

gresidual ¼ 27 pS. Similar results were obtained when unitary
currents (middle trace) were elicited by applying voltage
ramps from þ95 to 2100 mV (bottom trace). Ij (top trace)
was essentially linear with Vj, as expected for a normal homo-
typic channel, with gmain ¼ 160 pS, as well as gresidual ¼ 27 pS
(Fig. 5C). These observations are consistent with the findings
reported for mouse Cx30 by Valiunas et al. (30).
Mouse Cx26 and Cx30 have been shown to assemble
heterotypic channels in functional expression systems (31).
To study the behavior of HCx26wt/HCx30wt heterotypic
channels, we used co-culture of cells expressing HCx26wt-
Venus/L1 and HCx30wt-CFP/L1. Pairs of cells that displayed
at least one bi-color fluorescent plaque at the appositional
membrane were examined for functional coupling and
voltage gating. Junctional currents (Ij) were recorded from
the cell expressing HCx26wt-Venus/L1 while alternate nega-
tive and positive voltage steps (Vj) were applied to the cell
expressing HCx30wt-CFP/L1. We assume that at both the
Figure 5. Electrical properties of tagged HCx26wt and HCx30wt. (A) Con-
ductance changes (gj, top trace) and corresponding discrete current transitions
(Ij, bottom trace) due to gating of homotypic channels formed by HCx26wt–
Venus/L1 (Cx26, n ¼ 8). Ij was recorded with a transjunctional voltage (Vj)
of 295 mV (V1 ¼ 25 mV; V2 ¼ 2100 mV). (B) Conductance changes
(gj, top trace) and corresponding discrete transitions of junctional current
(Ij, bottom trace) due to gating of a single homotypic channel formed by
HCx30wt-CFP/L1 (Cx30, n ¼ 4). Vj (bottom trace) was initially set at
290 mV and stepped to zero by the end of the recording period, confirming
the complete closure of the channel. Conductance of the residual state, gresidual
was 27 pS (tilted arrow); conductance of the main state, gmain was 160 pS.
(C) Shown are two Vj ramps from þ95 to 2100 mV, 2 s in duration (Vj,
bottom trace), applied to a cell pair expressing homotypic HCx30wt-CFP/L1
channels (Cx30, n ¼ 3). The Ij /Vj relationship is linear. Estimated unitary con-
ductance (gj, upper trace) shows both main and residual (tilted arrow) states, as
in (B).

beginning and the end of the negative Vj step, two channels
were open and that, during the pulse, one channel exhibited
opening and closing transitions with a calculated unitary con-
ductance (gmain) of 82 pS (Fig. 6A). Ij recordings during a
positive Vj step showed two closing events from open to
residual states of 130 pS in magnitude and one closing event
of 30 pS in magnitude (gresidual), presumably from the sub-
conductance to the closed state (Fig. 6A). The different
values of gmain at positive and negative Vjs, indicative of rec-
tification, were also measured by delivering voltage ramps
from þ95 to 295 mV (Fig. 6B). In both experimental proto-
cols, single-channel conductance was smaller when the
HCx26wt side was made relatively positive and vice versa.
The fitting curve (solid line of the gj – Vj plot in Fig. 6B)
showed that the conductance of the main state at a
Vj ¼ 0 mV was 110 pS. This value is reasonably close to the
expected gmain value of 130 pS [geq ¼ (1/g26 þ 1/g30)]21 cal-
culated from the arrangement in series of one homotypic
HCx26wt connexon (g26 ¼ 230 pS) with one homotypic
HCx30wt connexon (g30 ¼ 320 pS), thus supporting the
notion that these heterotypic channels retain the properties of
composing hemichannels.
To obtain heterotypic channels containing one mutated con-
nexon, HeLa cells were transiently transfected with M34T,
expressed through the pIRES vector, HCx26wt-Venus/L1 or
HCx30wt-Venus/L1 (Fig. 6C and D). Separate cultures were
eventually mixed and pairs for dual whole-cell patch clamp
analysis were identified by the simultaneous presence of dif-
fused GFP fluorescence in the cytosol of one cell, indicating
positive M34T transfectants, and punctate Venus fluorescence
on the other side of the junction, reflecting expression of
tagged forms of either HCx26wt or HCx30wt. Our data
show that M34T hemichannels were able to dock with either
HCx26wt or HCx30wt, leading to the formation of functional
heterotypic gap junction channels. In both cases Ij recordings
obtained by applying voltage ramps showed a rectification that
was clearly more pronounced for the M34T/HCx26wt combi-
nation (Fig. 6C) than for M34T/HCx30wt (Fig. 6D).
Steady-state Gj – Vj dependence was analyzed by applying pro-
longed voltage steps (see Materials and Methods), and pooled
data were obtained by normalizing the steady-state conduc-
tance, measured just prior to command offset, to the conduc-
tance values at command onset (Fig. 6E). These data show
that M34T homotypic junctions were relatively insensitive to
Vj, similarly to HCx26wt channels (32), a feature maintained
in both heterotypic configurations for relative positivity of
the cytoplasmic side of the M34T connexon (Fig. 6E). Thus,
although M34T channels were endowed with a residual func-
tionality, their properties were significantly altered in both
homotypic and heterotypic configurations.
As a control, we examined heterotypic junctions formed by
pairing HCx26wt or HCx30wt with V84L, a deafness-related
recessive mutant of Cx26 that retains the ability to sustain
electrical communication (32,33). Homotypic V84L channels
exhibited a unitary conductance of 110 pS, very close to the
value measured for HCx26wt channels (33). Heterotypic
combinations (HCx26wt/V84L and HCx30wt/V84L) were
obtained by mixing transiently transfected HeLa cells, as
described previously. In contrast to what observed with
M34T, HCx26wt/V84L channels showed a linear Ij – Vj
Human Molecular Genetics, 2006, Vol. 15, No. 17 2575
dependence of the open state with a main conductance of
120 pS and no residual state (Fig. 7A). Results obtained
by delivering voltage commands in the range from 2100 to
þ100 mV for 10 s (not shown) were consistent with those
generated with ramp protocols. As expected on the basis of
the properties of HCx26wt/HCx30wt heterotypic junctions,
HCx30wt/V84L channels exhibited rectification and single-
channel conductance increased from 70 to 150 pS when
voltage ramps from 2100 to þ100 mV were delivered to
the cell expressing V84L while recording Ij from the cell
expressing HCx30wt (Fig. 7B). Thus, in contrast to what
observed with M34T, the electrical characteristics of
HCx30wt/V84L channels did not differ appreciably from
those of HCx26wt/HCx30wt channels.
Intercellular dye transfer
To assess the permeability of mutated gap junction channels,
we examined the cell-to-cell transfer of LY (MWionic
form ¼ 443; charge 22) by loading one cell of a pair with a
patch pipette. We did not detect LY diffusion across M34T
channels despite the presence of sizeable junctional conduc-
tance (gj ¼ 4.5 + 1.3 nS, n ¼ 5), measured by dual whole-cell
patch-clamp (Fig. 8). It should be noted that this level of gj
corresponds to approximately 450 open channels (4.5/
0.013 nS) at any given time and that robust LY transfer is con-
sistently detected with a similar number of functional
HCx26wt homotypic channels (34). LY diffusion was then
assayed in cultures of parental and transiently transfected
HeLa cells expressing either the wild-type or the mutated con-
nexins (Table 1). Dye coupling was scored as positive if the
probe diffused to at least two neighboring cells. As expected,
non-transfected HeLa cells did not display intercellular dye
diffusion (not shown). Microinjection of LY into an individual
cell of an HCx26wt-positive cluster resulted in intercellular
coupling in all instances with 100% of GFP-positive cells
filled with the probe. In contrast, no dye transfer was observed
between cells expressing the M34T mutation, whereas R127H
showed some incidence of dye transfer, although drastically
reduced in comparison with HCx26wt, with only one or two
cells of the R127H cluster receiving LY (Table 1).
The original report of the heterozygous M34T mutation in a
family with dominant deafness (13) suggested that the mutant
allele would act as a dominant-negative subunit of connexin
channels, and functional expression studies performed on
pairs of Xenopus oocytes (14) initially supported this view.
To determine whether this mutation could interfere with the
function of wild-type channels, we examined dye coupling
in HeLa cells co-transfected with the same amount of
HCx26wt and M34T plasmids and found that LY transfer
was abolished under these experimental conditions
(Table 1). In contrast, the R127H mutation, segregating with
autosomal recessive deafness (25), failed to affect the ability
of HCx26wt to form functional channels when the two
constructs were co-transfected at an equal ratio (Table 1).
These results demonstrate that the M34T mutant, but not the
recessive R127H mutant, significantly inhibited the functional
activity of HCx26wt when expressed in a manner mimicking
the heterozygous genotype.
2576 Human Molecular Genetics, 2006, Vol. 15, No. 17

Figure 6. Gating behavior of heterotypic channels. (A) A transjunctional voltage (Vj ¼ VCx30 2 VCx26, bottom trace) was imposed by stepping the HCx30wt
(Cx30)-expressing cell, and junctional current (Ij, middle trace) was recorded from the HCx26wt cell (Cx26, n ¼ 4). Top trace: junctional conductance trace
(gj). (B) Junctional current (Ij, top left) was recorded from the Cx30 cell during five consecutive Vj (VCx26 2 VCx30) ramps from þ95 to 295 mV, 3 s in duration
(bottom left, n ¼ 4). The right-hand panel shows rectification in the junctional conductance gj ¼ Ij /Vj, derived from the ramp responses and plotted as a function
of Vj. (C) Junctional current (Ij, middle trace) was recorded from the cell expressing Cx26 in response to four consecutive Vj ramps (VM34T 2 VCx26, bottom trace)
from þ100 to 2100 mV, 3 s in duration (n ¼ 4). The calculated junctional conductance (gj, upper trace) of heterotypic M34T/HCx26wt channels shows pro-
nounced rectification (n ¼ 4). (D) Same protocol as in (C) shows milder rectification for heterotypic M34T/HCx30wt channels (Vj ¼ VM34T 2 VCx30). Ij (middle
trace) was recorded from the cell expressing Cx30 (n ¼ 4). (E) Plot of the relationship of Vj to steady-state junctional conductance (Gjss, calculated as the ratio of
steady-state to instantaneous junctional conductance measured at the offset and onset of voltage steps, respectively) obtained in cell pairs expressing homotypic
M34T channels (filled circles), or heterotypic M34T/HCx26wt (open circles) and M34T/HCx30wt (filled squares) combinations (pooled data from five indepen-
dent experiments). In heterotypic channels, Vj is defined as positive for depolarization and negative for hyperpolarization of the side expressing the wild-type
connexins cell relative to the M34T cell. Lines through data are least square fits with quadratic functions to aid data trend visualization. Heterotypic channels
display asymmetric Gj 2 Vj relationships because M34T is poorly voltage dependent, whereas HCx26wt and HCx30wt close at positive voltages.

Figure 7. Gating behavior of heterotypic channels containing the V84L
mutation. (A) Conductance changes (gj, top trace) and corresponding discrete
current transitions (Ij, middle trace) due to channel opening during transjunc-
tional voltage ramps (Vj ¼ VV84L 2 VCx26, bottom trace) from þ100 to 2100 mV
(n ¼ 3) (B) Two junctional voltage ramps (Vj ¼ VV84L 2 VCx30, bottom trace)
from þ100 to 2100 mV, 3 s in duration, are shown (n ¼ 3). In both panels,
junctional current (Ij, middle trace) shows an ohmic behavior (viz. a linear
dependence) on Vj.
Intercellular calcium waves
In many cell types, connexins are involved in the radial propa-
gation of intercellular calcium waves induced by mechanical
stimulation of a single cell in a monolayer (reviewed in 35).
We performed calcium-imaging experiments to determine
whether the M34T variant could interfere with the intercellular
diffusion of second messengers such as calcium ions and ino-
sitol 1,4,5-trisphosphate (InsP3). To minimize the contribution
of ATP release to the propagation of calcium signals (36), in
all experiments, the extracellular medium was supplemented
with 40 U/ml apyrase grade VII.
In control cells transfected with the empty vector, mechan-
ical stimulation induced a [Ca2þ]i increase that remained
Human Molecular Genetics, 2006, Vol. 15, No. 17 2577
Figure 8. Permeability to LY of M34T mutant channels. Top left: bright field
image of one representative pair of HeLa cells with superimposed regions of
interest (ROIs) used to compute average pixel fluorescence. Top right: image
of the same field illuminated at lex ¼ 480 nm to visualize EGFP positive cells
(lem ¼ 535 nm). Bottom left: fluorescence image obtained at the end of the
experiment by exciting the field at lex ¼ 425 nm to quantify LY spread
(lem ¼ 535 nm). Cell 1 was contacted by a patch pipette containing 4% LY.
Intracellular delivery of LY started 20 s after the onset of the recording.
Bottom right: normalized fluorescence from the corresponding ROIs (traces
numbered 1 and 2) measured concurrently with junctional conductance (gj,
lowermost trace). The arrowhead marks the time at which the LY fluorescence
image was acquired. Results are representative of five independent exper-
iments. Scale bar: 10 mm.
Table 1. Qualitative comparison of dye diffusion (LY; MW ¼ 457 Da;
charge 22) between HeLa cells transfected with either HCx26wt or the
specified mutations
Transfection Percentage of positive injections n
HCx26wt 100 135
M34T 0 67
R127H 14 95
HCx26wt þ M34T 0 102
HCx26wt þ R127H 100 47
The incidence of cell coupling (second column) was scored by recording
the number of instances in which the dye diffused from the injected cell to
at least two neighboring ones. In the third column, the total numbers of
injections (n) is indicated.
largely confined to the stimulated cell, and in the few instances
in which the signal propagated to neighboring cells, it reached
only one cell of the first order (not shown). As expected, in
monolayers expressing HCx26wt, the [Ca2þ]i response of a
mechanically stimulated cell was consistently followed by
the intercellular spreading of a calcium wave (Fig. 9A;
n ¼ 33) that involved all first-order cells (91% of the obser-
vations), and in 21% of the cases, cells of the second order.
The propagation of calcium waves was completely inhibited
by the gap junction blocker carbenoxolone (CBX, 100 mM )
in all experiments (Fig. 9B; n ¼ 13). In M34T-expressing
cells, calcium spreading was largely blocked, being observed
only in 27% of the cases (Fig. 9C; n ¼ 37), with only one
cell of the first order being recruited. Similar results (blockade
2578 Human Molecular Genetics, 2006, Vol. 15, No. 17

Figure 9. Mechanically-induced intercellular calcium waves in transfected HeLa cells. HeLa cells expressing the specified constructs were analyzed by calcium
imaging 24 h post-transfection. Monolayers loaded with fura-2/AM were subjected to mechanical stimulation of single cells (cell 1 in all panels) by briefly
deforming the cell surface with a micropipette, as described in the Materials and Methods section. Apyrase (40 U/ml) was added in all experiments. For all
conditions, intercellular calcium waves shown are representative of at least 12 experiments from four independent transfections. (A) HCx26wt transfected
cells (n ¼ 32). (B) HCx26wt transfectants incubated in the presence of the gap junction blocker CBX (100 mM , n ¼ 18). (C) M34T transfected cells
(n ¼ 45). (D) HCx26 wt þ M34T co-transfected cells (n ¼ 12). In all panels, insets show the clusters of EGFP-positive cells included in the experiment with
the cell numbers corresponding to the individual fluorescence traces (see color-coding below the abscissa). Scale bar: 10 mm.

in 70% of the experiments) were also obtained when cells
were co-transfected with a 1:1 ratio of HCx26wt and M34T
plasmids (Fig. 9D; n ¼ 12).
The evidence that in a minority of cases a calcium wave was
observed in non-transfected and in M34T-expressing cells was
not entirely surprising, given the complexity of the molecular
mechanisms governing this phenomenon. On the one hand, the
contribution of paracrine signals other than ATP could not be
entirely excluded, and on the other hand, the recruitment of
stretch-activated channels and downstream pathways could
also account for the occasional calcium response of
the cell(s) tightly connected to the stimulated one. To
exclude an involvement of connexin-unrelated mechanisms
in the propagation of calcium waves, we refined our investi-
gation by restricting the analysis to neighboring cells involved
in the wave. In M34T transfectants, the [Ca2þ]i elevation in
 Human Molecular Genetics, 2006, Vol. 15, No. 17 2579

the cell(s) of the first order relative to the stimulated one

(D[Ca2þ]Io/D[Ca2þ]s) decreased slightly with respect to that
of HCx26wt (Fig. 10A). In addition, the rate of the rising
phase in the first-order cells (D[Ca2þ]Io/Dt) was reduced to
less than 50% compared with that of HCx26wt pairs and
similar results were obtained in control cells transfected with
the empty vector (Fig. 10A). The delay in the calcium
response of the first-order cell(s), measured as the time
required to reach the half maximal change (EC50) in
fluorescence ratio with respect to that of the stimulated cell
(tIo 2 ts) (37), was also altered, being significantly reduced
in M34T transfectants with respect to that observed in
HCx26wt-expressing cells (Fig. 10B). Finally, to complete
this quantitative analysis, we derived a coupling index
among different transfectants by multiplying the percentage
of successful stimulations (those inducing an intercellular
Ca2þ wave) by the number of cells displaying an increase of
cytosolic Ca2þ. The value calculated for M34T-expressing
cells was similar to that of controls and five times lower
than that of HCx26wt transfected cells (Fig. 10C).

Structural model of the M34T mutant

These genetic and functional data indicate that the M34
residue in Cx26 is critical for channel function. Fleishman
et al. (38) have recently published a structural model of the
transmembrane (TM) region of Cx32 (PDB code 1TXH) on
the basis of a combination of theoretical methods and low-
resolution electron cryo-microscopy map of Cx43 (39).
Given the high conservation of the TM helices among the con-
nexin family, these data are expected to serve as a general
template for the TM regions of other connexins. In this
model, M34 mediates putative contacts between two consecu-
tive connexins (38). However, because only the positions of
the Ca atoms were determined, the evaluation of the possible
consequences of point mutations is not immediate. To gain
further structural insights into the consequence of the M34T
mutation on channel structure, we undertook the theoretical
construction of all-atoms models of HCx26wt and M34T
using the Ca structure of Cx32 as a reference template.
In this atomic model, the side chains of Tyr152 within TM3,
which is considered to be the main pore-lining helix (2), deter-
mine the maximum constriction region of the pore of HCx26wt
channels near the extracellular side (Fig. 11A). M34 does not
face the pore lumen, but is partially exposed to the lipid face,
where it also establishes hydrophobic contacts with residues
of the adjacent connexin chain (Fig. 11B, inset).
Both HCx26wt and M34T models showed alike root mean
square deviation (RMSD) behaviors, reaching a stabilization
plateau after nearly 250 ps, thereafter oscillating around a
value of 0.3 nm (Fig. 11C). The average RMSD of each con-
nexin was 0.15 nm, suggesting that, although the helical
conformation was fairly well maintained, the relative orien-
tation of the chains suffered significant changes upon relax-
ation. Comparison of the average distances between the
backbone extremities of two opposite TM3 helices at the intra-
cellular sides of both variants indicated that HCx26wt suffered
a decrease of 0.2 nm, whereas M34T showed an opposite
behavior increasing the internal diameter by 0.1 nm
(Fig. 11D). A similarly opposite behavior was also retrieved
Figure 10. Quantitative analysis of critical parameters of calcium wave propa-
gation. HeLa cells expressing the specified constructs were analyzed by
calcium imaging in response to mechanical stimulation, as described in the
Materials and Methods section. Apyrase (40 U/ml) was added in all exper-
iments. Changes in intracellular Ca2þ concentration were estimated by calcu-
lating the excitation ratio of fura-2 fluorescence at 340 and 380 nm (see
Materials and Methods). (A) D[Ca2þ]Io/D[Ca2þ]s describes the calcium con-
centration change in cells of the first order relative to the stimulated cell.
D[Ca2þ]Io/Dt calculates the rise rate of the calcium signal in cells of the
first order relative to the stimulated cell. (B) tIo 2 ts indicates the delay in
the calcium response of first order cells, measured as the time required to
reach the half maximal (EC50) fluorescence ratio change with respect to that
of the stimulated cell. (C) The coupling index of different transfectants is
derived by multiplying the percentage of positive experiments (i.e. the exper-
iments in which intercellular calcium spreading was detected) by the number
of coupled cells. Results are shown as means + SD of the number of
experiments indicated in the legend of Figure 9. Statistical significance was
assessed by the unpaired Student’s t-test.
P , 0.05;

P ¼ 0.05; ns, not
significant.
at the extracellular side, where the backbone extremities of
the TM3 helices remained 0.2 nm closer in M34T
(Fig. 11D). Thus, the inter-helical angle between opposite
TM3 helices of HCx26wt decreased to almost 50 degrees,
whereas in the M34T mutant, it increased up to 65
degrees (Fig. 11E). This implies a sort of angular bending
between the principal angles of the helices (Fig. 11F)
that leads, as a consequence of this movement, to a higher
constriction of the pore in the M34T mutant. Hence,
the average distance between couples of OH atoms in
2580 Human Molecular Genetics, 2006, Vol. 15, No. 17

Figure 11. Structural models of HCx26 wt and M34T channels. (A) Atomic model of HCx26wt as seen from the intracellular side after full EM. The different
colors of the ribbons depict the six subunits forming the connexon. The Y152 residue corresponding to the minimum diameter of the pore lumen is represented in
space-filling spheres. Hydrogen atoms are not shown for visual clarity. (B) Lateral view of the transmembrane part of the HCx26wt connexon obtained rotating
the structure in (A) by 90 degrees. The position of M34 is indicated with its corresponding space-filling representation. Inset: a representative close-up of the
M34 neighborhood. Residues within 0.5 nm are shown in sticks. M34 may establish hydrophobic interactions with the lipid phase and also with residues belong-
ing to the second transmembrane helix (TM2) from the nearby connexin. (C) Root mean square deviation (RMSD) calculated over the Ca atoms plotted versus
simulated time for Cx26 wild-type (blue) and M34T (red) connexons. (D) Top panel: averaged distance between the center of the mass of the last four Ca atoms
of couples of opposite TM3 helices (see Materials and Methods). Middle panel: averaged distance between the center of the mass of the first four Ca atoms
of couples of opposite TM3 helices. Bottom panel: average distance between OH atoms belonging to opposite Y152 for HCx26wt (dark blue) and M34T
(red). (E) Trajectory of the inter-helical angle between opposite TM3 helices within the connexon. (F) The central scheme illustrates the conformational
change produced by the M34T substitution (red cylinders) with respect to wild-type (blue cylinders).

opposite Tyr152 residues remained nearly within its initial
value of 2 nm for HCx26wt, whereas a more pronounced
reduction to 1.7 nm was observed for the M34T mutant
(Fig. 11D).
deafness. This conclusion is supported by comprehensive bio-
physical and functional studies showing that, although M34T
retains partial functionality, the mutant channels are defective
with respect to wild-type in both electrical and metabolic
couplings.

DISCUSSION Genotype – phenotype correlation

Our data demonstrate that the presence of the M34T variant Although M34T has been described both as an autosomal
of HCx26 is indeed associated with mild-to-severe forms of dominant (13) and recessive mutation (15,40 – 42), this

interpretation has remained controversial due to the frequent
findings of normal hearing in individuals heterozygous for
the M34T variant (16 – 19) and even in compound heterozy-
gotes with 35delG (15). Thus, it has been suggested that the
pathological effect of the M34T allele might depend on the
mutations segregating in the opposing allele, such as
167delT (43,44) or 35delG (45). However, in a recent multi-
center study, all patients with a 35delG/M34T genotype
were found to have mild-to-moderate hearing impairment
(46). These discrepancies may reflect a reduced penetrance
of M34T or be the consequence of a diagnostic bias toward
more severe forms of hearing impairment, as persons with
mild deficits are less likely to undergo audiological or
genetic testing (46).
The audiometric features of the seven families reported here
revealed that hearing loss was present in 14 subjects. Consid-
ering heterozygous subjects carrying the M34T mutation, 10/
16 (62.5%) were mildly to severely impaired with respect to
the audiometric parameters of age- and sex-matched controls.
Of note, none of the five M34T compound heterozygotes dis-
played normal hearing, being classified as mildly to severely
impaired. The most compromised frequencies were in the
range of 2000 – 8000 Hz, which resulted in an audiogram con-
figuration of the sloping type. This observation may account
for the discrepancy with many previous studies in which the
degree of hearing impairment was assessed by pure tone
average of only three frequencies (500, 1000, 2000 Hz). A
clinical re-assessment of those subjects with a wider range
of frequencies may help uncover mild cases of hearing loss
associated with M34T mutation. In fact, the only other
report that examined air conduction thresholds across the
same range of frequencies, as in this study, found that the
two subjects with M34T/wt genotype displayed
mild-to-moderate hearing loss (47). A moderate hearing
loss affected the heterozygous M34T child (48); in this
study, the authors included only deaf children with hearing
thresholds .56 dB HL. Together, these observations
indicate a pathogenetic role of the M34T mutation and are
consistent with most recent surveys associating M34T com-
pound heterozygosis to mild-to-moderate hearing loss
(15,46,47,49,50).
The finding that 60% of patients carrying only the M34T
mutation were also hearing impaired, thus suggesting the
involvement of this variant even in dominant forms of deaf-
ness, is not easily explained. It has been noted, however,
that M34T segregates with a 10 bp deletion in the
50 -untranslated regions of the Cx26 gene (15). It is not
known whether this deletion is of any significance, but its
association with M34T may account for the variability in the
phenotype of M34T heterozygous, as it could alter the
expression level of the M34T allele in vivo. Of note, this del-
etion was present, in cis, in 14 out of the 16 M34T subjects,
whereas it was absent in all other individuals enrolled in the
study. Hence, this study and functional studies (14) suggest
that M34T could be regarded as a dominant allele, but its
dominant-negative activity may be affected by modifying
factors such as the level of RNA expression, as recently
suggested (51), thus explaining the observed phenotypic
variability. Although a population-based study may also
give information on the penetrance of the M34T allele, the
Human Molecular Genetics, 2006, Vol. 15, No. 17 2581
pedigrees of our families suggest that environmental factors
alone cannot explain the hearing deficits of M34T heterozy-
gotes (see, for example, family 7 in Fig. 1).
Impact of M34T on channel function
By combining electrophysiology, dye transfer and dynamic
calcium imaging, our data demonstrate that M34T mutant
channels, although correctly synthesized and targeted to the
plasma membrane, are functionally defective. This conclusion
is based on several lines of evidence. First, unitary
conductance of M34T was 10-fold lower than that of
HCx26wt gap junction channels. Secondly, heterotypic
HCx26wt/M34T channels exhibited a strong Ij –Vj rectification
that can have functional implications for communicating cells
residing at different transmembrane potentials. Thirdly,
intercellular transfer of LY was prevented, and finally,
M34T channels failed to sustain the propagation of inter-
cellular Ca2þ waves. These results demonstrate that the
mutation alters significantly the molecular cut-off size of the
pore and therefore should be considered as a pathological
variant of Cx26.
In addition, we found that M34T exerted a dominant-
negative inhibition of metabolic coupling when expressed
in a manner mimicking a heterozygous genotype, an effect
not displayed by the recessive R127H mutation. Our data
support the conclusion reached by previous functional
studies performed in paired Xenopus oocytes expressing
M34T (14). Although this dominant-negative effect was
later attributed to an artifact because of the lack of
control of the expression levels of mutant and wild-type
RNA in the exogenous system (40), a dominant
inhibition of channel function has been confirmed in a
recent study conducted on Cx26 hemichannels in single
oocytes (52).
These observations are difficult to reconcile with a recent
study demonstrating that M34T, although not capable of
forming functional gap junction channels on its own in
Xenopus oocytes, did not exhibit a dominant-negative effect
on cell – cell coupling when co-expressed at a 1:1 ratio with
HCx26wt (24). Such a discrepancy cannot be explained
easily. It should be noted, however, that White et al. (14)
and we have tested the dominant-negative action of M34T
in cells that were all programmed to express this variant
together with HCx26wt, whereas Skerrett et al. (24) have
paired oocytes co-injected with HCx26wt and M34T in hetero-
typic fashion, with oocytes expressing only the wild-type
protein. Thus, these studies are not directly comparable. An
additional difference in the observed functional properties of
M34T may stem from the relative expression levels of the
wild-type and mutant proteins, a parameter that cannot be
fully controlled in cells co-expressing the two HCx26
sequences, as there are no antibodies that can distinguish
between them. As found from this study and from the work
of D’Andrea et al. (20), although the amount of M34T
expression was found to be comparable with that of HCx26wt
in either single transfectants or in oocytes injected with the
cognate RNA (14), it is possible that changes in the protein
ratios may also account for the differences between
2582 Human Molecular Genetics, 2006, Vol. 15, No. 17
these three studies. Indeed, increasing the ratio of mutated to
wild-type RNA injection (M34T:HCx26wt ¼ 2:1) led to a
considerable decrease of intercellular coupling (24).
Differences in protein levels are an inherent drawback of in
vitro functional expression systems. Thus, the residual
permeability of the M34T variant to fluorescent tracers, as
previously reported (22,23), is likely to depend on the over-
expression of the low conductance channel, as suggested by
the same authors.
A structural model of M34T channels
The stringency of the requirements of distinct connexin chan-
nels in different tissues strongly suggests that altering such
composition in vivo would result in the development of func-
tional abnormalities (53). Indeed, targeted ablation of Cx26 in
the epithelial network of the organ of Corti has elegantly
illustrated its essential role for cochlear function and cell sur-
vival (54). Our functional analysis demonstrates that a meth-
ionine residue at position 34 in Cx26 substantially influences
homotypic, as well as heterotypic and heteromeric channels
formed by association with both HCx26wt and HCx30wt and,
together with the genetic and clinical data, further suggests
that the consequences of this single-amino acid substitution
cannot be compensated by other connexin co-expressed in the
cochlea.
M34 is located in the TM1 domain of connexins, which is
likely to be one of the pore-lining helices (38,55,56) and
interacts with other transmembrane helices, stabilizing the
TM1 structure and ensuring the open-channel state (23).
Our molecular dynamic simulations demonstrate that the
M34T substitution would cause a reduction of the side
chain size at this position, leading to a partially closed
channel, consistent with decreased channel permeability
and unitary conductance. Caution should be exerted,
however, in the interpretation of these models, which are
affected by several approximations whose consequences are
difficult to estimate because of the lack of input data at
atomic resolution (38). Nonetheless, the results of these
simulations are in good agreement with our own, as well
as previously published experimental data, suggesting that
this model captures at least the gross determinants of the
M34T mutation.
Calcium wave propagation and M34T pathogenicity
Intercellular calcium waves propagating in the network of sup-
porting cells have been observed in the organ of Corti in
response to mechanical stimulation of single-hair cells (57).
This phenomenon depends on two underlying mechanisms:
(i) the release of an extracellular messenger, believed to be
ATP because purinergic receptor antagonists and apyrase,
which hydrolyzes ATP, block the propagation of Ca2þ
waves in cochlear organotypic cultures (57); and (ii) the
diffusion of inositol trisphosphate (InsP3) through gap junc-
tions (35,58), because agents that interfere with InsP3 signal-
ing or gap junction communication effectively inhibit wave
propagation between many cell types (59 –61), including
supporting cells of the organ of Corti (33). We speculate,
therefore, that the disruption of this intercellular signaling
pathway by the M34T mutation may interfere with re-cycling
of Kþ ions back into the endolymph, leading to accumulation
of Kþ ions and excitotoxic death of hair and supporting
cells in vivo, which would eventually result in hearing
impairment (62). Our data are consistent with a
mechanistic explanation of the pathogenesis of deafness
which postulates that InsP3 permeability between the epi-
thelial cell network of the cochlea is crucial for normal
hearing (33).
MATERIALS AND METHODS
Reagents
LY, apyrase grade VII, aprotinin, chymostatin and leupeptin
were obtained from Sigma (St Louis, MO). The polyclonal
anti-rat Cx26 antibody employed was from Zymed Laboratories
(San Francisco, CA). For immunoblot detection, the secondary
antibody was a goat anti-rabbit IgG conjugated with alkaline
phosphatase from Calbiochem (San Diego, CA). For immuno-
fluorescence, the secondary antibody was a tetramethylrhoda-
mine isothiocyanate (TRITC)-conjugated affinity purified
goat anti-rabbit IgG from Jackson ImmunoResearch
Laboratories (West Grove, PA). The chemiluminescence
substrate (CSPD) and enhancer (Nitro-Block) for alkaline
phosphatase were from Tropix (Bedford, MA). Restriction
enzymes were from New England Biolabs (Beverly, MA).
Nitrocellulose was from Schleicher & Schuell (Dassel,
Germany). All other reagents were from Sigma, unless
otherwise specified.
DNA analysis
DNA was extracted from peripheral blood according to stan-
dard protocols after obtaining informed consent. The whole
Cx26 coding region was amplified by polymerase chain reac-
tion (PCR) using the following forward (CX26ORF-F:
TGCTTACCCAGACTCAGAGAA) and reverse
(CX26ORF-R: GACTGAGCCTTGACAGCTGAG) primers.
PCR reaction was performed in a total volume of 50 ml con-
taining 15 pmol of each primer, 100 mM dNTP, 5 ml 10 reac-
tion buffer (100 mM Tris pH 8.3, 500 mM KCl, 15 mM MgCl2,
0.01% gelatin), 2.5 U Ampli Taq Gold Polymerase (Perkin
Elmer, Foster City, CA) and 100 ng of DNA template, using
an automated Thermal Cycler 9700 (Perkin Elmer). The
cycling profile consisted of an initial denaturation at 948C
for 12 min, which was followed by 35 cycles of 948C 
45 s, 608C  45 s, 728C  80 s and a final extension step at
728C for 10 min. Exon I and flanking splice sites were ampli-
fied using oligonucleotides EX1CX26F (tcaaaggaactagga-
gatcgg) and EX1CX26R (aaggacgtgtgttggtccag) as forward
and reverse primers, respectively.
When genetic analysis was either negative for GJB2 mutations
or a heterozygous condition was found, samples were further
tested for GJB6 mutations. The D(GJB6-D13S1830) deletion
of the GJB6 gene was detected by PCR amplification as
suggested by del Castillo et al. (63). Amplicons were directly
sequenced with the same PCR primers, using an ABI-PRISM
big-dye terminator cycle reaction kit (Perkin Elmer), and
samples were then analyzed on an automated sequencer (ABI

3100, Perkin Elmer). Patients were eventually divided into three
groups: group I, negative for GJB2/GJB6 mutations; group II,
carrying mutations different from M34T and group III, carrying
at least the M34T mutation.
Patients
We have studied eight probands with mild-to-severe sensori-
neural hearing loss, in which genetic analysis had identified
the presence of the M34T mutation as a heterozygous or a
compound heterozygous condition and their respective
families (seven in total). A total of 28 subjects, aged
between 5 and 69 years (median ¼ 37 years and 5 months)
were studied. The follow-up period ranged from a minimum
of 18 months to a maximum of 12 years, and in all patients,
serial audiograms were available. None reported a relevant
history of otological disease, noise exposure or assumption
of ototoxic drugs.
A detailed history was collected from all patients, followed
by otoscopic and audiological examination comprehensive of
pure tone audiometry (or conditioned infantile audiometry in
one patient), tympanometry, stapedial reflex threshold and
auditory brainstem evoked potentials (ABR). Otoacoustic
emissions were not available in all patients. In all cases, the
audiological study revealed cochlear site of lesion, and there
were no clinical features suggestive of a syndromic deafness.
None of them accused vertigo or dizziness, so vestibular func-
tion was not tested. Audiological features of hearing loss were
then analyzed according to the GenDeaf study group rec-
ommendations (64). Air conduction thresholds were measured
across the following frequencies: 125, 250, 500, 1000, 2000,
4000 and 8000 Hz, whereas for bone conduction, the range
was between 250 and 4000 Hz. The dB values relate to
dB HL. The degree of hearing impairment was classified
by the pure-tone average (PTA) applied to the better ear
at 500, 1000, 2000 and 4000 Hz. In the case of
normal hearing, PTA is 20 dB; in mild hearing loss, PTA
is .20  40 dB; in moderate hearing loss, PTA is
.40  70 dB; severe hearing loss is characterized by a PTA
.70  95 dB, and hearing loss is defined as profound when
PTA .95 dB.
Hearing impairment is progressive when a deterioration
of 15 dB in the PTA occurs within 10 years. Depending on
its configuration, the audiometric curve is classified as flat, if
the difference between 125 and 8000 Hz is ,15 dB;
mid-frequency U-shaped, when difference is 15 dB between
the poorest thresholds in the mid-frequencies
(.500  2000 Hz) and those at higher (.2000  8000 Hz)
and lower (500 Hz) frequencies; low frequency-ascending,
when there are 15 dB between the poorer low frequency
thresholds and the higher frequencies; high frequency-gently
sloping, if the difference between the mean of 500 and
1000 Hz and the mean of 4000 and 8000 Hz is within
15 –29 dB; high frequency-steeply sloping, with 30 dB
difference between the latter frequency pairs. Hearing impair-
ment is classified as (i) asymmetrical, if .10 dB difference
occurs between the ears in at least two frequencies, with the
PTA in the better ear exceeding 20 dB HL and (ii) unilateral,
when one ear has either a PTA .20 dB or one frequency
threshold exceeding 50 dB while the other ear has a PTA
Human Molecular Genetics, 2006, Vol. 15, No. 17 2583
20 dB. Hearing thresholds were compared with those of
age- and sex-matched controls (ISO 7029:2000). Differences
in audiological parameters between groups were determined
by Fisher’s exact test and P-values ,0.05 were considered to
be significant.
Molecular cloning
The coding region of HCx30 (PubMed-NCBI accession
number HSA005585), HCx26wt (PubMed-NCBI accession
number AF281280) or HCx26 mutations was amplified by
PCR from genomic DNA of normal hearing subjects and
patients with different forms of sensorineural deafness. Ampli-
cons were subcloned into the bicistronic vector pIRES-EGFP
(Clontech, Palo Alto, CA), which permits an efficient selection
of transfected cells for functional studies by monitoring the
expression of the enhanced form of Aequorea victoria GFP,
as previously described (32). For electrophysiological record-
ings, we employed HCx26 proteins fused to either EYFP
(HCx26wt –EYFP) or the circularly permuted yellow mutant
of the GFP, called Venus (65) (HCx26wt –Venus), by two
linkers of different lengths. For the generation of fusion pro-
teins, the open reading frame (ORF) of the specified constructs
was amplified by PCR using oligonucleotide primers that
introduced the desired restriction enzyme sites at both ends
of the coding sequence and deleted the stop codon. The fol-
lowing primers, which introduced EcoRI –BamHI sites, were
used for Linker L1 (six amino acids: DPPVAT):
26forward: 50 -CGGAATTCAGATGGATTGGGGCACG-30 ;
26reverse: 50 -CGGGATCCACTGGCTTTTTTGACTT-30 ; 30-
forward: 50 -CGGAATTCCCAGCGCAATGGATTGG-30 ; 30-
reverse: 50 -CGGGATCCCTTGGGAAACCTGTGAT-30 . For
linker L2, which introduced EcoRI–XhoI sites and comprised
17 residues (RILQSTVPRARDPPVAT), the primers used
were 26forward: 50 -GGTCCTCGAGATGGATTGGGGCACG
CTGC-30 ; 26reverse: 50 -CCCGAATTCGAACTGGCTTTTTT
GACTTCCC-30 .
The resulting PCR products were digested with the two
enzymes and ligated into the respective sites of pEGFP-N1
(Clontech). All constructs were sequenced using the Dye ter-
minator (Perkin-Elmer), as recommended by the manufac-
turer, to verify that PCR amplification did not introduce
unwanted mutations. The length of the linker did not interfere
with the electrophysiological profile of the recombinant gap
junction channels, as assessed by measuring their unitary con-
ductance with a dual whole-cell patch clamp technique.
Cell culture and transfection
A clone of HeLa cells essentially devoid of connexins (66)
was kindly provided by Professor Klaus Willecke (University
of Bonn, Germany) and cultured according to standard
procedures. Twenty-four hours after plating, cells were trans-
fected with the expression vectors described previously, using
lipofectamine (Gibco/Invitrogen, Leek, The Netherlands).
Experiments were performed the following day.
Protein immunoblot analysis
Proteins were dissolved in lysis buffer (50 mM Tris pH 6.8, 2%
SDS, 10 mg/ml each of aprotinin, chymostatin and leupeptin)
2584 Human Molecular Genetics, 2006, Vol. 15, No. 17
from cells grown to confluence. Prior to electrophoresis,
protein concentration was measured by the bicinchoninic
acid method (Pierce, Rockford, IL) with bovine serum
albumin as the standard. For each sample, 20 mg of total pro-
teins were separated on 13% SDS gel and then transferred to
nitrocellulose. Coomassie blue staining of identical gels run
in parallel confirmed the equal loading of the samples. Nitro-
cellulose filters were blocked with 1% bovine serum albumin
in Tween-added PBS (TPBS: 137 mM NaCl, 2.7 mM KCl,
8.1 mM Na2PO4, 1.5 mM KH2PO4, pH 7.4, supplemented
with 0.1% Tween 20) and incubated overnight with the
primary anti-Cx26 antibody under continuous shaking. Blots
were further incubated for 1 – 2 h with the alkaline phospha-
tase (AP)-conjugated secondary antibody, processed for
protein visualization with ECL and exposed to CL-Xposure
Films (Pierce) for 10 or 20 min.
Immunofluorescence
Cells grown onto glass coverslips were fixed with 2% parafor-
maldehyde and incubated with the primary anti-Cx26 antibody
for 1 h at 48C. Following a 30 min incubation at 48C in the
presence of a secondary antibody, coverslips were mounted
onto glass slides and visualized under a Leica DMLS fluor-
escence microscope (Leica Microsystems, Wetzlar, Germany).
Dual-patch clamp recordings
For electrophysiological recordings, cells were grown on thin
(no. 0) coverslips and transferred to an experimental chamber
mounted on the stage of an upright microscope equipped with
an infinity-corrected water-immersion objectives 60,
0.90 numeric aperture (NA); LUMPlanFl, Olympus, Tokyo,
Japan. Cells were constantly superfused with ECS containing
(in mM ) 150 NaCl, 4 KCl, 1 MgCl2, 5 Hepes, 2 CaCl2,
2 Pyruvate, 5 Glucose, 2 CsCl and 1 BaCl2 (pH 7.4,
330 mOsm). Glass capillaries for patch clamp recordings
were formed on a vertical puller (PP-83, Narishige, Japan)
from 1.5 mm outer diameter soda glass (Harvard Apparatus,
Edenbridge, UK) and filled with an intracellular solution
(ICS-1) containing (in mM ) 130 KCl, 10 Na – Aspartate, 0.26
CaCl2, 5 TEA –Cl, 1 MgCl2, 5 Hepes, 2 EGTA and 3 ATP –
Mg (pH 7.2). Current and voltage were sampled at rates
between 1 and 20 kHz. In the whole-cell configuration, the
patch electrodes acquired an access resistance of 10 –
12 MOhm, on average, whereas the input resistance of
untransfected HeLa cells was .0.5 GOhm. To measure junc-
tional conductance, each cell of an isolated pair was voltage-
clamped independently with one of two List EPC-7 amplifiers
and kept at the same holding potential (Vh). By stepping the
voltage in one cell (cell 1) while keeping the potential of
cell 2 at Vh, thus establishing a transient transjunctional
voltage Vj ; V1 2 V2 ¼ DV1, junctional current (Ij) was
measured directly as the current change in the unstepped
cell (i.e. Ij ¼ 2DI2). Values of junctional conductance (gj)
were calculated by dividing Ij/Vj, corrected for the error due
to the access resistance of both pipettes. In the study of the
voltage dependence of normalized conductance (Gjss ¼ gjss/
gj(0), where gjss is the steady-state conductance measured
at the end of a prolonged voltage step and gj(0) is the
instantaneous conductance measured at the earliest time resol-
ution of Ij after the imposition of a voltage step), pairs coupled
by conductances larger than 10 nS were discarded, thereby
limiting the effect of series resistance on these measurements.
To block gap junction channels, cells were transiently super-
fused in ECS saturated with 100% CO2 to produce carbonic
acid (H2CO3), which, in its non-dissociated form, is membrane
permeable and causes a rapid closure of the gap junction
channels. The current flowing through single-gap junction
channels was detected as discrete step-like events while the
cells recovered from cytoplasm acidification during washout
of the CO2.
Intracellular delivery of LY and fluorescence imaging
For dye transfer assays, cell 1 (donor) was patch-clamped
using patch pipettes filled with a 4% LY solution in ICS-2,
containing (in mM ) 60 LiCl and 10 HEPES (adjusted to
pH 7.2 with KOH; 320 mOsm). Illumination wavelength
was set at 425 nm by a fast switching monochromator (Poly-
chrome IV, TILL Photonics, Martinsried, Germany). At
image recording onset, cell 1 was maintained in the
cell-attached configuration for a few seconds to establish a
baseline. Subsequently, the patch of membrane under the
pipette was ruptured, allowing LY to fill the cell while
leaving the seal intact (whole-cell recording conditions), and
fluorescence emission (F) was selected around 535 nm using
a D535/30 m filter (Chroma, Rockingham, VT). Images were
formed on a scientific grade CCD camera (SensiCam, PCO
Computer Optics, Kelheim, Germany) using either 40 or
60 water-immersion objectives (LumPlan FL, 0.8
and 0.9 NA, respectively; Olympus) on a motorized upright
microscope (BX61, Olympus) and displayed as (F 2 Fbck)/
(Fmax 2 Fbck), where Fmax is the asymptotic maximal value
reached in the injected cell and Fbck is background fluor-
escence. At the end of image recording (3 – 6 min), cell 2
(acceptor) was contacted by a second pipette containing an
identical intracellular solution. The whole-cell configuration
was achieved in cell 2 also, and the responses to a sequence
of 13 mV voltage pulses, fed to the patch clamp amplifier con-
nected to cell 1, were used to measure junctional conductance.
Data were analyzed offline using the Matlab 7.0 software
package (The MathWorks, Natick, MA).
Intercellular dye transfer
Glass capillaries were prepared with a dual-step puller (Nar-
ishige, Tokyo, Japan) and filled with a 5% solution of LY (dis-
solved in 0.33 M lithium chloride). Individual cells in
GFP-expressing clusters were injected with a pneumatic
PLI-100 pico-injector (Medical Systems Corp., Greenvale,
NY) and mounted onto a Zeiss-inverted Axiovert 35 TV
microscope (Carl Zeiss, Jena, Germany). For both GFP and
LY, the filter set was Zeiss 09 (BP450-490/LP520). Fluor-
escence images were collected through an oil immersion
objective (40, 1.8 NA), captured by a low light level CCD
camera (Hamamatsu Photonics, Tokyo, Japan) and fed into a
digital image processor developed in the laboratory. Intercel-
lular LY diffusion was revealed by capturing three different
images for each experiment: image 1 (GFP), showing

GFP-expressing cells before injection; image 2 (GFP þ LY),
capturing the same field of cells 3 –5 min after LY injection
and containing the fluorescence of both GFP and LY; image
3 (LY), resulting from the subtraction of image 1 from
image 2, thus displaying LY fluorescence alone. Subsequent
processing for false color display was performed with Corel
Photo-Paint software (Corel Corporation, Ottawa, Canada).
Ca21 imaging
Transfected HeLa cells grown onto coverslips were incubated
in modified Krebs solution (KRH), containing (in mM)
25 Hepes/NaOH buffer, pH 7.4, 125 NaCl, 5 KCl, 1 MgSO4,
1.2 KH2PO4, 2 CaCl2, 10 glucose, supplemented with 0.5%
BSA and loaded at room temperature with a mixture (1:2,
vol:vol in DMSO) of fura-2/AM (final concentration in
KRH ¼ 1 mM ) and 20% Pluronic gel (Molecular Probes,
Eugene, OR; final concentration in KRH ¼ 0.04%). After
45 min, the loading solution was removed and the cells
washed and kept in KRH for 45 min to allow complete
de-esterification of the dye. Videomicroscopy and Ca2þ
measurements were carried out at room temperature. The
fura-2 loaded cultures were observed on an inverted micro-
scope (Zeiss Axiovert 35 TV) with an oil immersion 40
objective (1.8 NA; Carl Zeiss). Cells were excited at wave-
lengths between 340 and 380 nm with a monochromator
device equipped with integrated light source (Polychrome
IV, Till Photonics). Excitation light was separated from the
light emitted from the sample using a 395 nm dichroic
mirror. Images of emitted fluorescence .510 nm were
recorded at a frequency of one image per second by a
cooled slow-scan interline transfer camera (IMAGO CCD
camera, Till Photonics) and simultaneously displayed on a
19 inch Vivitron color monitor. The imaging system was con-
trolled by an integrating imaging software package (TILLvi-
sION, Till Photonics) using a personal computer. Video
frames were then digitized, integrated and processed offline
to convert fluorescence data into Ca2þ maps by computing a
ratio of 340/380 nm excitation wavelength values. Mechanical
stimulation of single cells in confluent monolayer was per-
formed by briefly deforming the cell surface with a micropip-
ette. A fire-polished glass micropipette with a tip diameter
1 mm was positioned over a single cell by a micromanipula-
tor (Narishige). The pipette was briefly deflected downward
manually to transiently distort the membrane. Experiments
were either discontinued or discarded when there was evi-
dence of damage to the plasma membrane, as revealed by
leak of the fluorescent probe from the cell. Results are
shown as mean + SD. Comparisons between two populations
of data were made using the Student’s unpaired t-test and
P-values of 0.05 or less were considered to be significant.
Generation of the all-atoms model
Our analysis is based on the Ca model of mouse Cx32
reported by Fleishman et al. (38; PDB entry code 1TXH).
This model includes the transmembrane (TM) a-helices corre-
sponding to the segments Ala19 – Ala40 (TM1), Leu76 –Ala96
(TM2), Leu131 – Val152 (TM3) and Val189 – Val209 (TM4).
After sequence alignment using T-COFFEE (67), these
Human Molecular Genetics, 2006, Vol. 15, No. 17 2585
fragments corresponded in HCx26 to: Ser19 –Glu42 (TM1),
Arg75– Ala96 (TM2), Leu132 –Val153 (TM3) and Glu187 –
Leu209 (TM4). Charged residues found immediately next to
those segments were also added in order to consider eventual
electrostatic interactions. Mapping of the Ca coordinates from
the original mouse Cx32 model generated a rough Ca model
of HCx26. Subsequently, all the remaining atoms of the
amino acids were added using the canonical conformation
with the XLEAP module of AMBER8.0 software suite (68).
Additionally, all helices were capped with acetyl and
N-methyl groups in order to avoid the spurious effects of term-
inal zwitterionic charges.
This procedure generated an all-atoms model with a large
number of steric clashes and internal tensions that were
relaxed using energy minimization (EM) and molecular
dynamics (MD). Simulations were performed in the presence
of implicit solvent with the generalized Born approach (69)
as implemented in the SANDER module of AMBER8.0
with a 2 ps time-step. After EM, 50 ps of constrained
MD were performed, in which the initial harmonic constraints
on the Ca carbons were linearly reduced from 1 to
0 kcal/mol-Å2, followed by 450 ps of unconstrained MD.
Simulations were stopped when the systems reached stabiliz-
ation of the RMSD. A dielectric constant of 78, a cutoff
distance of 1.8 nm and a salt concentration of 0.15 M were
used. The simulations used the Amber94 force field (70)
keeping the temperature at 100 K using Langevin dynamics
with a collision frequency of 2 ps21. The shake algorithm
was used to constraint all chemical bonds. This procedure is
expected to allow the entire complex to explore the energy
landscape more efficiently than a simple EM, obtaining a
better relaxation of the protein complex. A similar strategy
has already been proven to produce reliable results in other
protein– RNA and protein – protein complexes (71,72).
RMSDs were calculated from the last minimization step
over the Ca- atoms. The distances between the extremes of
the pore-lining TM3 helices were calculated between the
centers of mass of the first (last) four Ca atoms of two oppo-
site TM3 helices. The inter-helix angles were calculated
between the lines defined by centers of mass of the first
and last four Ca carbons of each helix TM3. The distances
between OH atoms of Tyr152, corresponding to the
maximum constriction of the pore, were calculated between
couples of residues in opposite TM3 helices in the connexon.
All the distances and angles were averaged over the three
possible pairs of opposite helices in the hexamer.
ACKNOWLEDGEMENTS
This work has been funded by grants from Regione Friuli
Venezia-Giulia (to P.D’A.), Telethon Italy (GP0043Y02 to
F.M. and P.D’A. and GGP05131 to F.M.), the Sixth Research
Frame Program of the European Union (FP6 Integrated Project
EuroHear, LSHG-CT-20054-512063 to F.M.), Fondazione
CARIPARO (to S.P.) and National Institutes of Health
(RO1-NS036706 to F.F.B.).
Conflict of Interest statement. None declared.
2586 Human Molecular Genetics, 2006, Vol. 15, No. 17
REFERENCES
1. Bruzzone, R., White, T.W. and Paul, D.L. (1996) Connections with
connexins: the molecular basis of direct intercellular signaling.
Eur. J. Biochem., 238, 1 –27.
2. Sosinsky, G.E. and Nicholson, B.J. (2005) Structural organization of gap
junction channels. Biochim. Biophys. Acta, 1711, 99 –125.
3. Willecke, K., Eiberger, J., Degen, J., Eckardt, D., Romualdi, A.,
Guldenagel, M., Deutsch, U. and Sohl, G. (2002) Structural and functional
diversity of connexin genes in the mouse and human genome. Biol.
Chem., 383, 725–737.
4. Gerido, D.A. and White, T.W. (2004) Connexin disorders of the ear, skin,
and lens. Biochim. Biophys. Acta, 1662, 159 –170.
5. Wei, C.J., Xu, X. and Lo, C.W. (2004) Connexins and cell signaling in
development and disease. Annu. Rev. Cell Dev. Biol., 20, 811 –838.
6. Kikuchi, T., Kimura, R.S., Paul, D.L., Takasaka, T. and Adams, J.C.
(2000) Gap junction systems in the mammalian cochlea. Brain Res. Brain
Res. Rev., 32, 163–166.
7. Petit, C., Levilliers, J. and Hardelin, J.P. (2001) Molecular genetics of
hearing loss. Annu. Rev. Genet., 35, 589–646.
8. Sabag, A.D., Dagan, O. and Avraham, K.B. (2005) Connexins in hearing
loss: a comprehensive overview. J. Basic Clin. Physiol. Pharmacol., 16,
101 –116.
9. Jentsch, T.J. (2000) Neuronal KCNQ potassium channels: physiology and
role in disease. Nat. Rev. Neurosci., 1, 21–30.
10. Boettger, T., Hubner, C.A., Maier, H., Rust, M.B., Beck, F.X. and
Jentsch, T.J. (2002) Deafness and renal tubular acidosis in mice lacking
the K–Cl co-transporter Kcc4. Nature, 416, 874–878.
11. Nuttall, A.L. (1999) Sound-induced cochlear ischemia/hypoxia as a
mechanism of hearing loss. Noise Health, 2, 17 –32.
12. Rio, C., Dikkes, P., Liberman, M.C. and Corfas, G. (2002) Glial fibrillary
acidic protein expression and promoter activity in the inner ear of
developing and adult mice. J. Comp. Neurol., 442, 156– 162.
13. Kelsell, D.P., Dunlop, J., Stevens, H.P., Lench, N.J., Liang, J.N.,
Parry, G., Mueller, R.F. and Leigh, I.M. (1997) Connexin 26 mutations in
hereditary non-syndromic sensorineural deafness. Nature, 387, 80–83.
14. White, T.W., Deans, M.R., Kelsell, D.P. and Paul, D.L. (1998) Connexin
mutations in deafness. Nature, 394, 630–631.
15. Houseman, M.J., Ellis, L.A., Pagnamenta, A., Di, W.L., Rickard, S.,
Osborn, A.H., Dahl, H.H., Taylor, G.R., Bitner-Glindzicz, M., Reardon, W.
et al. (2001) Genetic analysis of the connexin-26 M34T variant:
identification of genotype M34T/M34T segregating with mild-moderate
non-syndromic sensorineural hearing loss. J. Med. Genet., 38, 20 –25.
16. Denoyelle, F., Weil, D., Maw, M.A., Wilcox, S.A., Lench, N.J.,
Allen-Powell, D.R., Osborn, A.H., Dahl, H.H., Middleton, A.,
Houseman, M.J. et al. (1997) Prelingual deafness: high prevalence of a
30delG mutation in the connexin 26 gene. Hum. Mol. Genet., 6,
2173–2177.
17. Kelley, P.M., Harris, D.J., Comer, B.C., Askew, J.W., Fowler, T.,
Smith, S.D. and Kimberling, W.J. (1998) Novel mutations in the connexin
26 gene (GJB2) that cause autosomal recessive (DFNB1) hearing loss.
Am. J. Hum. Genet., 62, 792 –799.
18. Scott, D.A., Kraft, M.L., Carmi, R., Ramesh, A., Elbedour, K., Yairi, Y.,
Srisailapathy, C.R., Rosengren, S.S., Markham, A.F., Mueller, R.F. et al.
(1998) Identification of mutations in the connexin 26 gene that cause
autosomal recessive nonsyndromic hearing loss. Hum. Mutat., 11,
387 –394.
19. Feldmann, D., Denoyelle, F., Loundon, N., Weil, D., Garabedian, E.N.,
Couderc, R., Joannard, A., Schmerber, S., Delobel, B., Leman, J. et al.
(2004) Clinical evidence of the nonpathogenic nature of the M34T variant
in the connexin 26 gene. Eur. J. Hum. Genet., 12, 279– 284.
20. D’Andrea, P., Veronesi, V., Bicego, M., Melchionda, S., Zelante, L.,
Di Iorio, E., Bruzzone, R. and Gasparini, P. (2002) Hearing loss:
frequency and functional studies of the most common connexin26 alleles.
Biochem. Biophys. Res. Commun., 296, 685– 691.
21. Martin, P.E., Coleman, S.L., Casalotti, S.O., Forge, A. and Evans, W.H.
(1999) Properties of connexin26 gap junctional proteins derived from
mutations associated with non-syndromal heriditary deafness. Hum. Mol.
Genet., 8, 2369–2376.
22. Thonnissen, E., Rabionet, R., Arbones, M.L., Estivill, X., Willecke, K.
and Ott, T. (2002) Human connexin26 (GJB2) deafness mutations affect
the function of gap junction channels at different levels of protein
expression. Hum. Genet., 111, 190–197.
23. Oshima, A., Doi, T., Mitsuoka, K., Maeda, S. and Fujiyoshi, Y. (2003)
Roles of Met-34, Cys-64 and Arg-75 in the assembly of human connexin
26. Implication for key amino acid residues for channel formation and
function. J. Biol. Chem., 278, 1807–1816.
24. Skerrett, I.M., Di, W.L., Kasperek, E.M., Kelsell, D.P. and Nicholson, B.J.
(2004) Aberrant gating, but a normal expression pattern, underlies the
recessive phenotype of the deafness mutant Connexin26M34T. FASEB J.,
18, 860–862.
25. Estivill, X., Fortina, P., Surrey, S., Rabionet, R., Melchionda, S.,
D’Agruma, L., Mansfield, E., Rappaport, E., Govea, N., Mila, M. et al.
(1998) Connexin-26 mutations in sporadic and inherited sensorineural
deafness. Lancet, 351, 394–398.
26. Wang, H.L., Chang, W.T., Li, A.H., Yeh, T.H., Wu, C.Y., Chen, M.S. and
Huang, P.C. (2003) Functional analysis of connexin-26 mutants associated
with hereditary recessive deafness. J. Neurochem., 84, 735– 742.
27. Peracchia, C. (2004) Chemical gating of gap junction channels; roles of
calcium, pH and calmodulin. Biochim. Biophys. Acta, 1662, 61 –80.
28. Lautermann, J., ten Cate, W.J., Altenhoff, P., Grummer, R., Traub, O.,
Frank, H., Jahnke, K. and Winterhager, E. (1998) Expression of the
gap-junction connexins 26 and 30 in the rat cochlea. Cell Tissue Res., 294,
415–420.
29. Forge, A., Marziano, N.K., Casalotti, S.O., Becker, D.L. and Jagger, D.
(2003) The inner ear contains heteromeric channels composed of cx26 and
cx30 and deafness-related mutations in cx26 have a dominant negative
effect on cx30. Cell Commun. Adhes., 10, 341–246.
30. Valiunas, V., Manthey, D., Vogel, R., Willecke, K. and Weingart, R.
(1999) Biophysical properties of mouse connexin30 gap junction channels
studied in transfected human HeLa cells. J. Physiol., 519, 631 –644.
31. Dahl, E., Manthey, D., Chen, Y., Schwarz, H.J., Chang, Y.S., Lalley, P.A.,
Nicholson, B.J. and Willecke, K. (1996) Molecular cloning and functional
expression of mouse connexin-30, a gap junction gene highly expressed in
adult brain and skin. J. Biol. Chem., 271, 17903–17910.
32. Bruzzone, R., Veronesi, V., Gomes, D., Bicego, M., Duval, N., Marlin, S.,
Petit, C., D’Andrea, P. and White, T.W. (2003) Loss-of-function and
residual channel activity of connexin26 mutations associated with
non-syndromic deafness. FEBS Lett., 533, 79– 88.
33. Beltramello, M., Piazza, V., Bukauskas, F.F., Pozzan, T. and Mammano, F.
(2005) Impaired permeability to Ins(1,4,5)P3 in a mutant connexin
underlies recessive hereditary deafness. Nat. Cell Biol., 7, 63–69.
34. Beltramello, M., Bicego, M., Piazza, V., Ciubotaru, C.D., Mammano, F.
and D’Andrea, P. (2003) Permeability and gating properties of human
connexins 26 and 30 expressed in HeLa cells. Biochem. Biophys. Res.
Commun., 305, 1024– 1033.
35. Allbritton, N.L. and Meyer, T. (1993) Localized calcium spikes and
propagating calcium waves. Cell Calcium, 14, 691 –697.
36. Osipchuk, Y. and Cahalan, M. (1992) Cell-to-cell spread of calcium
signals mediated by ATP receptors in mast cells. Nature, 359, 241–244.
37. Scemes, E., Dermietzel, R. and Spray, D.C. (1998) Calcium waves
between astrocytes from Cx43 knockout mice. Glia, 24, 65 –73.
38. Fleishman, S.J., Unger, V.M., Yeager, M. and Ben-Tal, N. (2004) A
Calpha model for the transmembrane alpha helices of gap junction
intercellular channels. Mol. Cell., 15, 879–888.
39. Unger, V.M., Kumar, N.M., Gilula, N.B. and Yeager, M. (1999)
Three-dimensional structure of a recombinant gap junction membrane
channel. Science, 283, 1176–1180.
40. Wilcox, S.A., Saunders, K., Osborn, A.H., Arnold, A., Wunderlich, J.,
Kelly, T., Collins, V., Wilcox, L.J., McKinlay Gardner, R.J., Kamarinos, M.
et al. (2000) High frequency hearing loss correlated with mutations in the
GJB2 gene. Hum. Genet., 106, 399–405.
41. Kenneson, A., Van Naarden Braun, K. and Boyle, C. (2002) GJB2
(connexin 26) variants and nonsyndromic sensorineural hearing loss: a
HuGE review. Genet. Med., 4, 258– 274.
42. Wu, B.L., Lindeman, N., Lip, V., Adams, A., Amato, R.S., Cox, G.,
Irons, M., Kenna, M., Korf, B., Raisen, J. et al. (2002) Effectiveness of
sequencing connexin 26 (GJB2) in cases of familial or sporadic childhood
deafness referred for molecular diagnostic testing. Genet. Med., 4,
279–288.
43. Cucci, R.A., Prasad, S., Kelley, P.M., Green, G.E., Storm, K., Willcox, S.,
Cohn, E.S., Van Camp, G. and Smith, R.J. (2000) The M34T allele variant
of connexin 26. Genet. Test, 4, 335–344.
44. Kenna, M.A., Wu, B.L., Cotanche, D.A., Korf, B.R. and Rehm, H.L.
(2001) Connexin 26 studies in patients with sensorineural hearing loss.
Arch. Otolaryngol. Head Neck Surg., 127, 1037–1042.

45. Marlin, S., Garabedian, E.N., Roger, G., Moatti, L., Matha, N., Lewin, P.,
Petit, C. and Denoyelle, F. (2001) Connexin 26 gene mutations in
congenitally deaf children: pitfalls for genetic counseling. Arch.
Otolaryngol. Head Neck Surg., 127, 927– 933.
46. Snoeckx, R.L., Huygen, P.L., Feldmann, D., Marlin, S., Denoyelle, F.,
Waligora, J., Mueller-Malesinska, M., Pollak, A., Ploski, R., Murgia, A.
et al. (2005) GJB2 mutations and degree of hearing loss: a multicenter
study. Am. J. Hum. Genet., 77, 945–957.
47. Roux, A.F., Pallares-Ruiz, N., Vielle, A., Faugere, V., Templin, C.,
Leprevost, D., Artieres, F., Lina, G., Molinari, N., Blanchet, P. et al.
(2004) Molecular epidemiology of DFNB1 deafness in France. BMC Med.
Genet., 5, 5.
48. Levi, H., Tell, L. and Cohen, T. (2004) Sensorineural hearing loss in
Jewish children born in Jerusalem. Int. J. Pediatr. Otorhinolaryngol., 68,
1245–1250.
49. Liu, X.Z., Pandya, A., Angeli, S., Telischi, F.F., Arnos, K.S., Nance, W.E.
and Balkany, T. (2005) Audiological features of GJB2 (connexin 26)
deafness. Ear Hear., 26, 361 –369.
50. Cryns, K., Orzan, E., Murgia, A., Huygen, P.L., Moreno, F.,
del Castillo, I., Chamberlin, G.P., Azaiez, H., Prasad, S., Cucci, R.A. et al.
(2004) A genotype–phenotype correlation for GJB2 (connexin 26)
deafness. J. Med. Genet., 41, 147–154.
51. Wilch, E., Zhu, M., Burkhart, K.B., Regier, M., Elfenbein, J.L.,
Fisher, R.A. and Friderici, K.H. (2006) Expression of GJB2 and GJB6 is
reduced in a novel DFNB1 allele. Am. J. Hum. Genet., 79, 174–179.
52. Palmada, M., Schmalisch, K., Bohmer, C., Schug, N., Pfister, M., Lang, F.
and Blin, N. (2006) Loss of function mutations of the GJB2 gene detected
in patients with DFNB1-associated hearing impairment. Neurobiol. Dis.,
22, 112 –118.
53. White, T.W. and Bruzzone, R. (2000) Gap junctions: fates worse than
death? Curr. Biol., 10, R685– R688.
54. Cohen-Salmon, M., Ott, T., Michel, V., Hardelin, J.P., Perfettini, I.,
Eybalin, M., Wu, T., Marcus, D.C., Wangemann, P., Willecke, K. et al.
(2002) Targeted ablation of connexin26 in the inner ear epithelial gap
junction network causes hearing impairment and cell death. Curr. Biol.,
12, 1106–1111.
55. Zhou, X.W., Pfahnl, A., Werner, R., Hudder, A., Llanes, A., Luebke, A.
and Dahl, G. (1997) Identification of a pore lining segment in gap junction
hemichannels. Biophys. J., 72, 1946–1953.
56. Kronengold, J., Trexler, E.B., Bukauskas, F.F., Bargiello, T.A. and
Verselis, V.K. (2003) Single-channel SCAM identifies pore-lining
residues in the first extracellular loop and first transmembrane domains of
Cx46 hemichannels. J. Gen. Physiol., 122, 389–405.
57. Gale, J.E., Piazza, V., Ciubotaru, C.D. and Mammano, F. (2004) A
mechanism for sensing noise damage in the inner ear. Curr. Biol., 14,
526 –529.
58. Sanderson, M.J., Charles, A.C., Boitano, S. and Dirksen, E.R. (1994)
Mechanisms and function of intercellular calcium signaling. Mol. Cell.
Endocrinol., 98, 173–187.
Human Molecular Genetics, 2006, Vol. 15, No. 17 2587
59. Boitano, S., Dirksen, E.R. and Sanderson, M.J. (1992) Intercellular
propagation of calcium waves mediated by inositol trisphosphate. Science,
258, 292–295.
60. Newman, E.A. and Zahs, K.R. (1997) Calcium waves in retinal glial cells.
Science, 275, 844 –847.
61. Giaume, C. and Venance, L. (1998) Intercellular calcium signaling and
gap junctional communication in astrocytes. Glia, 24, 50–64.
62. Bruzzone, R. and Cohen-Salmon, M. (2005) Hearing the messenger:
Ins(1,4,5)P3 and deafness. Nat. Cell Biol., 7, 14 –16.
63. del Castillo, I., Villamar, M., Moreno-Pelayo, M.A., del Castillo, F.J.,
Alvarez, A., Telleria, D., Menendez, I. and Moreno, F. (2002) A deletion
involving the connexin 30 gene in nonsyndromic hearing impairment.
N. Engl. J. Med., 346, 243– 249.
64. GenDeaf Study GroupMazzoli, M., Van Camp, G., Newton, V.,
Giarbini, M., Declau, F. and Parving, A. (2003) Recommendations for the
description of genetic and audiological data for families with
nonsyndromic hereditary hearing impairment. Audiological Medicine, 1,
148–150.
65. Miyawaki, A., Llopis, J., Heim, R., McCaffery, J.M., Adams, J.A.,
Ikura, M. and Tsien, R.Y. (1997) Fluorescent indicators for Ca2þ
based on green fluorescent proteins and calmodulin. Nature, 388,
882–887.
66. Elfgang, C., Eckert, R., Lichtenberg-Frate, H., Butterweck, A., Traub, O.,
Klein, R.A., Hulser, D.F. and Willecke, K. (1995) Specific permeability
and selective formation of gap junction channels in connexin-transfected
HeLa cells. J. Cell Biol., 129, 805– 817.
67. Notredame, C., Higgins, D.G. and Heringa, J. (2000) T-Coffee: a novel
method for fast and accurate multiple sequence alignment. J. Mol. Biol.,
302, 205–217.
68. Case, D.A., Darden, T.A., Cheatham, T.E., III, Simmerling, C.L.,
Wang, J., Duke, R.E., Lou, R., Merz, K.M., Wang, B., Pearlman, D.A.
et al. (2004) AMBER 8.0. Amber 8.0.
69. Tsui, V. and Case, D.A. (2000) Theory and applications of the generalized
Born solvation model in macromolecular simulations. Biopolymers, 56,
275–291.
70. Cornell, W.D., Cieplak, P., Bayly, C.I., Gould, I.R., Merz, K.M.,
Ferguson, D.M., Spellmayer, D.C., Fox, T., Caldwell, J.W. and
Kollman, P.A. (1995) A second generation force field for the simulation of
proteins, nucleic acids, and organic molecules. J. Am. Chem. Soc., 117,
5179–5197.
71. Ayala, Y.M., Pantano, S., D’Ambrogio, A., Buratti, E., Brindisi, A.,
Marchetti, C., Romano, M. and Baralle, F.E. (2005) Human, Drosophila,
and C. elegans TDP43: nucleic acid binding properties and splicing
regulatory function. J. Mol. Biol., 348, 575 –588.
72. Pantano, S., Marcello, A., Ferrari, A., Gaudiosi, D., Sabo, A., Pellegrini, V.,
Beltram, F., Giacca, M. and Carloni, P. (2005) Insights on HIV-1 Tat:
P/CAF bromodomain molecular recognition from in vivo experiments and
molecular dynamics simulations. Proteins, 62, 1062–1073.