Hair cells use active zones with different voltage dependence of Ca 2+ influx to decompose sounds into complementary neural codes

Abstract

Significance
We hear sounds varying in intensity over six orders of magnitude using spiral ganglion neurons (SGNs), each of which changes its firing rates over only a fraction of this range. Somehow, the SGNs with different dynamic ranges collectively encode the full range of sound levels represented in the receptor potential of the inner hair cell (IHC) in the mammalian cochlea. Our study, combining subcellular imaging, mouse genetics, and auditory systems physiology, offers a unifying synaptic hypothesis for wide dynamic range sound encoding in the spiral ganglion. We propose that IHCs, from one receptor potential but via presynaptic active zones that vary in the voltage dependence of Ca ²⁺ influx, generate complementary codes on sound pressure level in functionally distinct SGNs.

Full text

Hair cells use active zones with different voltage
dependence of Ca
2+
influx to decompose sounds
into complementary neural codes
Tzu-Lun Ohn
a,b,c,d
, Mark A. Rutherford
a,e,1
, Zhizi Jing
b,d,f,1
, Sangyong Jung
a,g,h,1
, Carlos J. Duque-Afonso
a,b
,
Gerhard Hoch
a
, Maria Magdalena Picher
a,b
, Anja Scharinger
i,j
, Nicola Strenzke
b,d,f
, and Tobias Moser
a,b,c,d,g,h,k,2
a
Institute for Auditory Neuroscience & InnerEarLab, University Medical Center Göttingen, 37099 Goettingen, Germany;
b
Göttingen Graduate School for
Neurosciences and Molecular Biosciences, University of Göttingen, 37073 Goettingen, Germany;
c
Bernstein Focus for Neurotechnology, University of
Göttingen, 37073 Goettingen, Germany;
d
Collaborative Research Center 889, University of Göttingen, 37099 Goettingen, Germany;
e
Department of
Otolaryngology, Washington University School of Medicine, St. Louis, MO 63110;
f
Auditory Systems Physiology Group, InnerEarLab, Department
of Otolaryngology, University Medical Center Göttingen, 37099 Goettingen, Germany;
g
Synaptic Nanophysiology Group, Max Planck Institute for
Biophysical Chemistry, 37077 Goettingen, Germany;
h
Center for Nanoscale Microscopy and Molecular Physiology of the Brain, University of Göttingen,
37073 Goettingen, Germany;
i
Institute of Pharmacy, Department of Pharmacology and Toxicology, University of Innsbruck, A-6020 Innsbruck, Austria;
j
Center
for Chemistry and Biomedicine, University of Innsbruck, A-6020 Innsbruck, Austria; and
k
Bernstein Center for Computational Neuroscience, University of
Göttingen, 37073 Goettingen, Germany
Edited by A. J. Hudspeth, The Rockefeller University, New York, NY, and approved June 21, 2016 (received for review April 8, 2016)
For sounds of a given frequency, spiral ganglion neurons (SGNs) with
different thresholds and dynamic ranges collectively encode the
wide range of audible sound pressures. Heterogeneity of synapses
between inner hair cells (IHCs) and SGNs is an attractive candidate
mechanism for generating complementary neural codes covering the
entire dynamic range. Here, we quantified active zone (AZ) proper-
ties as a function of AZ position within mouse IHCs by combining
patch clamp and imaging of presynaptic Ca
2+
influx and by immu-
nohistochemistry. We report substantial AZ heterogeneity whereby
the voltage of half-maximal activation of Ca
2+
influx ranged over
∼20 mV. Ca
2+
influx at AZs facing away from the ganglion activated
at weaker depolarizations. Estimates of AZ size and Ca
2+
channel
number were correlated and larger when AZs faced the ganglion.
Disruption of the deafness gene GIPC3 in mice shifted the activation
of presynaptic Ca
2+
influx to more hyperpolarized potentials and
increased the spontaneous SGN discharge. Moreover, Gipc3 disrup-
tion enhanced Ca
2+
influx and exocytosis in IHCs, reversed the spatial
gradient of maximal Ca
2+
influx in IHCs, and increased the maximal
firing rate of SGNs at sound onset. We propose that IHCs diversify
Ca
2+
channel properties among AZs and thereby contribute to
decomposing auditory information into complementary representa-
tions in SGNs.
auditory system
|
spiral ganglion neuron
|
dynamic range
|
synaptic
strength
|
presynaptic heterogeneity
The auditory system enables us to perceive sound pressures
that vary over six orders of magnitude. This is achieved by
active amplification of cochlear vibrations at low sound pressures
and compression at high sound pressures. The receptor potential
of inner hair cells (IHCs) represents the full range (1), whereas
each postsynaptic type I spiral ganglion neuron (hereafter termed
SGN) encodes only a fraction (2–6). SGNs with comparable fre-
quency tuning but different spontaneous spike rates and sound
responses are thought to emanate from neighboring, if not the
same, IHC at a given tonotopic position of the organ of Corti (2, 5,
7, 8). Even in silence, IHC active zones (AZs) release glutamate at
varying rates, evoking “spontaneous”spiking in SGNs. SGNs with
greater spontaneous spike rates respond to softer sounds (high-
spontaneous rate, low-threshold SGNs), than those with lower
spontaneous spike rates (low-spontaneous rate, high-threshold
SGNs) (2, 9). This diversity likely underlies the representation of
sounds across all audible sound pressure levels in the auditory
nerve, to which neural adaptation also contributes (10).
How SGN diversity arises is poorly understood. Candidate
mechanisms include the heterogeneity of ribbon synapses that
differ in pre- and/or postsynaptic properties even within indi-
vidual IHCs (7, 11–14). IHC AZs vary in the number (11, 15) and
voltage dependence of gating (11) of their Ca
2+
channels regardless
of tonotopic position (16). Lateral olivocochlear efferent projec-
tions to the SGNs regulate postsynaptic excitability (17) and con-
tribute to the establishment of a gradient of presynaptic ribbon size
along the “modiolar–pillar”axis (18), where the modiolar side faces
the ganglion and the pillar side is away from the ganglion, toward
the pillar cells.
In postnatal development of the mouse, high-spontaneous-rate
SGNs coemerge with AZs that exhibit stronger maximal Ca
2+
influx.
Moreover, IHCs deficient for the AZ protein bassoon lack the
population of AZs with strongest Ca
2+
influx and, concurrently lack
SGNs with high spontaneous rates (19). Based on these correlations
and given the eminent role of presynaptic Ca
2+
influx in controlling
synaptic strength (20–22), we proposed that the varying Ca
2+
influx
at a given IHC AZ largely determines the difference in spontaneous
and evoked spiking among SGNs (19). Larger and more complex
AZs (7, 13, 14) with stronger Ca
2+
influx (16) tend to reside on
the modiolar IHC side. However, according to classical findings
from the cat cochlea, modiolar synapses seem weaker as they drive
Significance
We hear sounds varying in intensity over six orders of magni-
tude using spiral ganglion neurons (SGNs), each of which
changes its firing rates over only a fraction of this range.
Somehow, the SGNs with different dynamic ranges collectively
encode the full range of sound levels represented in the re-
ceptor potential of the inner hair cell (IHC) in the mammalian
cochlea. Our study, combining subcellular imaging, mouse ge-
netics, and auditory systems physiology, offers a unifying
synaptic hypothesis for wide dynamic range sound encoding in
the spiral ganglion. We propose that IHCs, from one receptor
potential but via presynaptic active zones that vary in the
voltage dependence of Ca
2+
influx, generate complementary
codes on sound pressure level in functionally distinct SGNs.
Author contributions: T.M. designed research; T.-L.O., M.A.R., Z.J., S.J., C.J.D.-A., and M.M.P.
performed research; G.H. and A.S. contributed new reagents/analytic tools; T.-L.O., M.A.R.,
Z.J., S.J., C.J.D.-A., and N.S. analyzed data; and T.-L.O., M.A.R., Z.J.,S.J., C.J.D.-A., M.M.P.,N.S.,
and T.M. wrote the paper.
The authors declare no conflict of interest.
This article is a PNAS Direct Submission.
1
M.A.R., Z.J., and S.J. contributed equally to this work.
2
To whom correspondence should be addressed. Email: tmoser@gwdg.de.
This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.
1073/pnas.1605737113/-/DCSupplemental.
www.pnas.org/cgi/doi/10.1073/pnas.1605737113 PNAS Early Edition
|
1of10
NEUROSCIENCE PNAS PLUS

low-spontaneous-rate, high-threshold SGNs (2, 7). This discrep-
ancy suggests that factors other than AZ size and amplitude of Ca
2+
influx contribute to differences in SGN properties.
Here, we live-imaged most, if not all, AZs of individual IHCs
for analyzing the amplitude and voltage dependence of Ca
2+
influx
as well as AZ size and position within the IHC. Combined with
immunohistochemical estimation of ribbon size and Ca
2+
-channel
abundance of AZs, our data indicate opposing gradients of max-
imal amplitude and voltage dependence of Ca
2+
influx along the
modiolar–pillar axis: modiolar AZs, thought to drive low-sponta-
neous-rate, high-threshold SGNs, are, on average, larger and have
more Ca
2+
channels but operate in a more depolarized range. We
propose that the more hyperpolarized activation range of Ca
2+
influx at pillar AZs poises them to enable high spontaneous rates
and low sound thresholds of SGN firing. We studied candidate
regulators of Ca
2+
influx at IHC AZs and found that disruption of
the deafness gene GIPC3 in mice causes a hyperpolarizing shift of
Ca
2+
-channel activation and increases spontaneous SGN firing.
Results
Heterogeneity of Presynaptic Ca
2+
Influx Among the AZs of IHCs. The
strength of presynaptic Ca
2+
influx and the large (∼2μm) distance
between hair cell AZs (16, 23) enable spatiotemporally resolved
optical analysis of presynaptic Ca
2+
signaling (11, 24, 25). How-
ever, given the limited duration of stable Ca
2+
influx in whole-cell
patch-clamp recordings, the low speed of laser-scanning confocal
microscopy had prohibited a comprehensive comparison of the
Ca
2+
-influx properties among the AZs within a given IHC. To
overcome this technical limitation we combined spinning disk
microscopy with fast piezoelectric focusing to sequentially and
rapidly (frame rate 100 Hz) image confocal IHC sections. This
enabled the analysis of voltage dependence and maximal ampli-
tude of the Ca
2+
influx, visualized as local fluorescence increase of
the low-affinity Ca
2+
indicator Fluo-8FF (K
d
=10 μM) at most if
not all synapses of a given IHC (Fig. 1A)whiletheIHCCa
2+
current remained stable (rundown <25%; a few IHCs were ex-
cluded from the analysis because of rundown >25%).
We first imaged the fluorescence of the TAMRA (tetrameth-
ylrhodamine)-labeled RIBEYE (major protein constituent of
the ribbon)-binding peptide (26) to localize AZs and to measure
the fluorescence intensity of labeled ribbons before unavoidable
TAMRA bleaching by the strong blue laser light used for Ca
2+
imaging. Another set of images was collected after completing
the time-critical Ca
2+
imaging to capture IHC morphology based
on the RIBEYE-peptide background fluorescence (Fig. 1A).
Supported by our previous work we assume that, in conditions of
strong cytosolic Ca
2+
buffering (10 mM EGTA in the pipette), the
fluorescence change (ΔF/F
0
,Fig.1B) of the low-affinity Ca
2+
indicator at the AZ faithfully reports synaptic Ca
2+
influx (11), but
note that this assumption could be violated should Ca
2+
-indicator
saturation or Ca
2+
-induced Ca
2+
release, nonetheless, occur. For
simplicity, we hence refer to the Fluo-8FF ΔF/F
0
at AZs as
“synaptic Ca
2+
influx.”We depolarized the IHC to −7mVto
maximize Ca
2+
-channel-open probability. As shown in Fig. 1B,the
“maximal synaptic Ca
2+
influx”varied among the AZs of this
representative IHC, whereas the whole-cell Ca
2+
current remained
stable. We postulate that this heterogeneity largely arises from
differences in the number of Ca
2+
channels among the AZs within
a given IHC. The coefficient of variation (c.v.) of maximal AZ
Ca
2+
influx per IHC was on average similar to the c.v. of maximal
AZ Ca
2+
influx across AZ pooled from all IHCs (c.v. =0.47 ±
0.18, N=28 IHCs vs. c.v.=0.59, n=331 AZs of the same 28 IHCs;
Fig. 1 Dand F), whereas the c.v. of maximal AZ Ca
2+
influx across
all IHCs was smaller (0.38). Thus, most of the AZ population
variance was explained by heterogeneity among the AZs within the
individual IHCs. However, IHCs varied in the extent of this het-
erogeneity: The c.v. of maximal Ca
2+
influx among AZs per IHC
ranged from 0.15 to 0.81 (Fig. 1F;seeFig. S1Afor statistics of
individual IHCs).
We directly tested for heterogeneous abundance of Ca
2+
channels at AZs by analyzing the relative fluorescence intensities
of IHC AZs immunolabeled for Ca
v
1.3 L-type Ca
2+
channels
(Fig. 1 Cand D, blue), which mediate >90% of IHC Ca
2+
influx
(27, 28). To measure the relative abundance of Ca
v
1.3 Ca
2+
channels among AZs, we identified synapses in fixed tissue by
juxtaposition of immunolabeled presynaptic ribbons and post-
synaptic glutamate receptors (Fig. 1C) and integrated the syn-
aptic Ca
v
1.3 immunofluorescence. Interestingly, the variability of
the synaptic Ca
V
1.3 immunofluorescence was smaller (c.v. =0.28)
than that of maximal Ca
2+
influx in live-cell imaging experiments
(c.v. =0.59, Fig. 1D; see statistics of individual cells in Fig. S1 A
and B). Variance in maximal Ca
2+
influx exceeding that of Ca
V
1.3
immunofluorescence could be explained, for example, by AZs
having Ca
V
1.3 channels with different open probabilities.
Ca
2+
-Channel Abundance and Maximal Ca
2+
Influx Scale with Ribbon
Size. Do larger AZs contain more Ca
2+
channels and display
stronger maximal Ca
2+
influx? To address this, we related our
estimates of Ca
2+
-channel abundance from live imaging and im-
munohistochemistry to the corresponding fluorescence intensity of
RIBEYE-peptide or immunolabeled ribbons (Fig. 1 Aand C;see
statistics of individual cells in Fig. S1 Aand B). We assume that
ribbon fluorescence scales with the number of RIBEYE molecules
per ribbon and, hence, with ribbon size (26, 29). Moreover, we
assume that AZ size scales with ribbon size (29, 30), which was
further supported by a positive correlation of synaptic immuno-
fluorescence of RIBEYE and the AZ marker bassoon, a pre-
synaptic scaffold, in separate experiments (r=0.46, n=77 AZs,
P<0.001). The distributions of ribbon fluorescence intensity were
broad and slightly skewed in live imaging (Fig. 1 Eand F; c.v. of
0.46 for the entire population and 0.31 as the mean c.v. of individual
cells, Fig. S1D) and immunofluorescence (c.v. of 0.46 for the entire
population and 0.42 as the mean c.v. of individual cells) analyses,
respectively.
Maximal AZ Ca
2+
influx showed a positive correlation with
the RIBEYE-peptide fluorescence in live-imaging experiments,
suggesting that, indeed, larger AZs exhibit stronger maximal Ca
2+
influx (Fig. 1G, green). We found a stronger correlation between
Ca
V
1.3 and RIBEYE immunofluorescence (Fig. 1G, blue), in-
dicating that larger AZs contain more Ca
2+
channels. This weaker
correlation in live imaging might relate to greater variability of
maximal synaptic Ca
2+
influx, which next to the number of Ca
2+
channels depends on their open probability that can also be en-
hanced by calmodulin-mediated cooperative gating (31). Positive
correlations were also found within most individual IHCs in both
live-cell imaging and immunohistochemistry (Fig. S1 Dand E),
indicating that larger AZs tended to have more Ca
2+
channels
and, consequently, stronger maximal Ca
2+
influx. This was further
supported by finding a positive correlation of synaptic Ca
V
1.3 and
bassoon immunofluorescence in separate experiments (Fig. S1C,
r=0.625, n=77 AZs, P<0.001). Most cells contained an AZ
whose Ca
2+
influx was substantially stronger than that of the
others: On average the ΔF/F
0
of the strongest AZ (“winner”)was
2.5 times greater than the average of the others (Fig. S1F,P<
0.001, Wilcoxon rank sum test).
Maximal Ca
2+
Influx Varies with AZ Position in a Modiolar–Pillar
Gradient. To study whether structural and functional AZ proper-
ties vary with position in mouse IHCs, we maintained the native
morphology and position of the IHC within the organ of Corti as
much as possible, by patching the modiolar IHC face with minimal
disruption of the surrounding cells. To combine live-imaging data
from several cells, we reconstructed the morphology of individual
IHCs and the positions of their synapses based on the fluorescence
of the TAMRA-conjugated RIBEYE-binding peptide and then
2of10
|
www.pnas.org/cgi/doi/10.1073/pnas.1605737113 Ohn et al.

transformed the Cartesian coordinates onto cell-aligned cylindri-
cal coordinates (Fig. S2). In brief, for each cell we identified the
plane of maximized mirror symmetry orthogonal to the tonotopic
axis (Fig. S2A). Then, we optically resectioned the IHC orthogonal
to a straight line fit to the pillar edge of the plane of symmetry
(Fig. S2B). We estimated the center of mass for each section and
connected those of the bottom-most and of the largest section to
define the central axis for our cylindrical coordinate system. We
projected the AZ coordinates of several cells along their central
axis for the polar charts, with the four sides annotated as modiolar
or pillar (facing toward or away from the ganglion), or tonotopic-
apical or tonotopic-basal (toward the cochlear apex or base, Fig.
S2C). The analogous fixed-tissue volumes were also transformed
into cylindrical coordinates (Fig. S3).
PolarchartsofmaximalCa
2+
influx and ribbon size for 202 AZs
from 14 IHCs as a function of AZ cylindrical coordinates are
displayed in Fig. 2 Aand B. Our analysis revealed a modiolar–pillar
gradient: large AZs with stronger Ca
2+
influx tended toward the
modiolar face (Fig. 2 A–D,F,andG). There was a nonsignificant
trend for maximal Ca
2+
influx to be greater in the modiolar half of
the AZ population. This difference became significant when ex-
cluding the AZs of the basal IHC pole (r≤3μm, Fig. S4A,P=
0.024, Wilcoxon rank sum test), for which modiolar/pillar sorting is
less certain. In 13 out of 14 IHCs the AZ with the strongest Ca
2+
influx was located on the modiolar side. Ca
V
1.3 immunofluores-
cence was significantly more intense for modiolar AZs (Fig. 2 C
and E,P<0.001, Wilcoxon rank sum test). RIBEYE-peptide
ribbon fluorescence (Fig. 2 B,F,andG,P<0.05, Wilcoxon rank
sum test) and RIBEYE immunofluorescence (Fig. 2 Fand H,P<
0.01, Wilcoxon rank sum test) consistently revealed significant
modiolar–pillar differences, indicating that modiolar AZs are
larger. In contrast, we did not detect gradients along the tonotopic
or the central axes of the IHCs for any of the analyzed parameters
(Fig. S4 Band C). In conclusion, IHC AZs follow a modiolar–pillar
gradient for size and Ca
2+
-channel number, whereby modiolar AZs
are larger and have more Ca
2+
channels.
Voltage Dependence of Ca
2+
Influx Varies with AZ Position in a Pillar–
Modiolar Gradient. Next, we examined the voltage dependence of
synaptic Ca
2+
influx for most if not all AZs of a given IHC. We
applied voltage ramps (from −87 mV to 63 mV) to the IHC
while sequentially imaging the AZs (Fig. 3A), thereby estab-
lishing the voltage dependence of AZ Ca
2+
influx in single de-
polarizations. Using ramps instead of step depolarizations, we
limited stimulation to at most 80 depolarizations per IHC. We
estimated the voltage of half-maximal activation of the whole-
cell Ca
2+
current of 25 IHCs by fitting a Boltzmann function to
their activation curves [V
0.5
=−28.1 ±1.7 mV (SD), Fig. 3B].
Given the noise of the fluorescence–voltage relationships of the
individual AZs we approximated the data by a fit function (SI
Methods) and then analyzed the derived fractional activation
curves by fitting a Boltzmann function. The mean V
0.5
of
synaptic Ca
2+
influx at 210 AZs of these IHCs (−27.4 mV) was
very similar to that of the mean whole-cell Ca
2+
influx. However,
fractional activation varied substantially among the AZs (Fig.
3C). Accordingly, the V
0.5
of Ca
2+
influx showed a wide distri-
bution for the entire AZ population, ranging from −38.4 mV to
−18.1 mV (SD: 4.7 mV, Fig. 3 Cand D) and within the individual
IHCs [mean: −27.3 ±3.2 mV (SD), Fig. S1G]. The voltage
sensitivity of Ca
2+
influx reported by the slope factor k was
similar for whole-cell and synaptic Ca
2+
influx (7.2 ±0.4 mV vs.
8.2 ±1.5 mV, Fig. 3 Band E). The scatter plot of slope factor
50 ms
AB
C
DE
c.v.
FG
0.4 a.u.
100 pA
80 mV
Cav1.3 CtBP2 GluA2 Cav1.3 CtBP2
GluA2
0
20
30
10
∆F-7mv / F0
3210
35
25
15
5
0
3210
c.v. = 0.59, 0.28
n = 331 (28)
187 (12)
# of synapses
# of synapses
Cav1.3 IF (a.u.)
20
10
07531
25
15
5
0
3210
c.v. = 0.46, 0.46
n = 217 (19)
187 (12)
# of synapses
# of synapses
Fribbon / Fnearby
CtBP2 IF (a.u.)
0.6
0.4
0.2
0
1.00.80.60.40.20
proportion of cells
c.v. (mean ± S.D.)
0.47 ± 0.18 (28)
0.25 ± 0.06 (12)
0.31 ± 0.09 (19)
0.42 ± 0.11 (12)
2.5
1.7
0.9
0.1
7.55.53.51.5
2.2
1.5
0.8
0.1
4.13.12.11.10.1
Fribbon / Fnearby
∆F-7mv/F0
CtBP2 cluster intensity (a.u.)
Cav1.3 IF (a.u.)
r = 0.55
(n=208)
r = 0.65
(n=187)
Fig. 1. AZ heterogeneity: Those with larger ribbons have more Ca
2+
chan-
nels. (A,Left) Stack of confocal sections covering an exemplary IHC filled
with TAMRA-RIBEYE-binding peptide via the patch pipette. (A,Upper Right)
Fluo-8FF fluorescence reports Ca
2+
influx (arrowheads) as spot-like maxima
near the membrane during depolarization. (A,Lower Right) Corresponding
image of TAMRA-RIBEYE-binding peptide fluorescence labeling synaptic
ribbons (arrowheads), where Ca
2+
influx occurs. White square indicates
where the background fluorescence (F
nearby
) was measured. (Scale bar:
1μm.) (B) Measurement of Ca
2+
influx at individual AZs in an exemplary IHC.
Depolarizations to −7mV(Upper) evoked similar whole-cell Ca
2+
currents
with each stimulus (superimposed, Middle). Focusing on single AZs during
each repetition revealed major heterogeneity of maximal Ca
2+
-indicator
fluorescence amplitudes (Lower,ΔF
−7mV
/F
0
, 100-Hz frame rate; panel su-
perimposes one trace per AZ, each marked by a different color). (C) Confocal
projections of immunolabeled presynaptic Ca
2+
channels (Ca
v
1.3), pre-
synaptic ribbons (CtBP2/RIBEYE), and postsynaptic AMPA receptors (GluA2)
with overlay on far right. (Scale bar: 2 μm.) (D) Green: distribution of the
maximal ΔF
−7mV
/F
0
for each AZ from live imaging. Blue: distribution of the
integrated Ca
v
1.3 immunofluorescence pixel intensities for each AZ in fixed
tissue. (E) Red: distribution of the maximal fluorescence intensity of TAMRA-
peptide labeled ribbons, normalized to the cytosolic fluorescence (F
ribbon
/
F
nearby
). Brown: distribution of the integrated CtBP2/RIBEYE immunofluo-
rescence pixel intensites for each AZ. (F) Distributions of the c.v. estimated
within individual IHCs for AZ parameters: maximal Ca
2+
influx (ΔF
−7mV
/F
0
,
dashed green), TAMRA-RIBEYE-binding peptide fluorescence (solid red),
Ca
v
1.3 immunofluorescence (solid blue), and CtBP2/RIBEYE immunofluores-
cence (solid brown). (G) Scatter plot of maximal ΔF
−7mV
/F
0
vs. TAMRA-
RIBEYE-binding fluorescence intensity (green) and of Ca
v
1.3 vs. CtBP2/RIBEYE
immunofluorescence (blue). Dashed lines are linear regressions and r is
Pearson’s correlation coefficient.
Ohn et al. PNAS Early Edition
|
3of10
NEUROSCIENCE PNAS PLUS

and V
0.5
(Fig. 3F) also presents the maximal Ca
2+
influx (color
scale) for a total of 210 AZs: The synapses with the strongest
maximal Ca
2+
influx showed intermediate V
0.5
. We found a tendency
for synapses with more hyperpolarized V
0.5
to have greater voltage
sensitivity (i.e., smaller k; r=0.47).
We found that, on average, Ca
2+
influx at pillar AZs had
slightly more hyperpolarized V
0.5
than that of the modiolar
ones (Fig. 4 Aand B,∼1.4 mV, P<0.001, Wilcoxon rank sum
test), whereas there was no significant difference in voltage
sensitivity between pillar and modiolar synapses (Fig. 4 Cand
D,P>0.05, Wilcoxon rank sum test). We did not detect gra-
dients along the tonotopic and central axes of the IHCs for
either V
0.5
or k (Fig. S4 Band C). In summary, our data in-
dicate major differences in presynaptic Ca
2+
signaling for a
given receptor potential. This may translate into differences in
transmitter release and postsynaptic spiking. The more nega-
tive operating point of pillar AZs might account for the high
spontaneous rates and low sound thresholds of SGNs emanat-
ing from pillar synapses reported for the cat. How the pillar–
modiolar gradient of the voltage dependence of Ca
2+
influx
012
pillar
(abneural)
modiolar
(neural)
apical
basal
∆F-7mV / F0
pillar
(n=82)
modiolar
(n=140)
∆F-7mV / F0
0
1.6
0.4
0.8
1.2
Fribbon / Fnearby
135
apical
basal
pillar
(abneural)
modiolar
(neural)
6
μm
pillar
(abneural)
modiolar (μm)
(neural)
∆F-7mV / F0
4-4-8 80
0
0.5
1.5
2.5
0.3
1
1.7
2.4
FCav1.3
Ca2+ influx
(r=0.21,n=202)
Cav1.3 antibody
(r=0.35, n=187)
pillar
(n=78)
modiolar
(n=109)
0.4
0.8
1.2
1.6
2.0
FCav1.3
***
FCtBP2
Fribbon / Fnearby
5
1
3
9
7
0
4
3
1
5
2
4-4-8 8
0
CtBP2 peptide
(r=0.24, n=147)
CtBP2 antibody
(r=0.47, n=187)
pillar
(abneural)
modiolar (μm)
(neural)
pillar
(n=57)
modiolar
(n=90)
1
5
2
3
4
Fribbon / Fnearby
pillar
(n=78)
modiolar
(n=109)
***
0.4
0.8
1.2
1.6
2.0
FCtBP2
*
6
μm
AB
CD
E
FGH
Fig. 2. Maximal synaptic Ca
2+
influx varies with AZ position in IHCs:
modiolar–pillar gradient. (A) The polar chart displays intensity of maximal AZ
Ca
2+
influx (ΔF
−7mV
/F
0
) as a function of AZ position in live-imaging experi-
ments. Modiolar and pillar refer to facing toward or away from the ganglion
in the modiolus; apical and basal refer to the tonotopic axis of the organ of
Corti. The radius of the inner circle is 3 μm. Data were pooled from 21 IHCs.
(B) The polar chart displays locations and intensity of AZ TAMRA-peptide
fluorescence intensity in live-imaging experiments. Data were pooled
from 14 IHCs. (C) Maximal AZ Ca
2+
influx (red, 202 AZs of 21 cells) and in-
tegrated AZ Ca
v
1.3 immunofluorescence (black, 187 AZs of 12 cells) as a
function of position along the modiolar–pillar axis. Solid lines represent linear
regressions and r indicates the Pearson’s correlation coefficient. (D)Thebox
plot (10, 25, 50, 75, and 90% percentiles) shows the statistics of maximal AZ
Ca
2+
influx for the modiolar (140 AZs) and pillar (82 Azs) halves of the same
IHCs as in A. No significant difference was reported by Wilcoxon rank sum test
(P=0.33). (E) The box plot shows the statistic s (as in D)ofCa
v
1.3 immuno-
fluorescence intensity for AZs of 12 cells, which was stronger on the modiolar
(109 AZs) than on the pillar side (78 AZs, P<0.001). (F) Spatial gradient of AZ
TAMRA-peptide fluorescence in live-imaging (red, 14 IHCs) and CtBP2/RIBEYE
immunofluorescence (black circle, same AZs as in E) along the mod iolar–pillar
axis. (Gand H) The box plot shows the statistics (as in D) of the TAMRA-peptide
fluorescence intensity (G) and CtBP2/RIBEYE immunofluorescence intensity (H).
Both measures indicate greater CtBP2/RIBEYE abundance on the modiolar side
than on the pillar side: G,P<0.05 and H,P<0.001.
A
150 ms
150 mV
0.5 a.u.
100 pA
B
D
-90 -60 -30 0
0
1
0.5
N = 25 cells
Pactivation,ICa
Vm (mV)
# of synapses
-40 -30 -20 -10
0
40
80
416128
V0.5, mean = -27.4 ± 4.7 mV
(n = 226, N = 25)
kmean = 8.2 ± 1.5 mV
(n = 226, N = 25)
V0.5 (mV)
k (mV)
Pactivation,FCa
-90 -60 -30 0 30
0
1
0.5
Vm (mV)
C
n = 226
V0.5, mean =
-27.4 ± 4.7 mV
V0.5, mean =
-28.1 ± 1.7 m
V
kmean =
7.2 ± 0.4 mV
E
# of synapses
0
40
80
F
kmean =
8.2 ± 1.5 mV
-40 -30 -20 -15-35 -25
4
9
14
k (mV)
V0.5 (mV)
r = 0.47
0
2.5
2.0
1.5
1.0
0.5
ΔF-7mV / F0
0.8
2.0
1.4
ΔF-7mV / F0
Fig. 3. Heterogeneity of voltage-dependent activation of AZ Ca
2+
influx in
IHCs. (A) Representative experiment analyzing voltage-dependent activation
of AZ Ca
2+
influx. Ramp depolarizations were used (Upper: from −87 mV to 63 mV
in 150 ms, that is, 1 mV/ms), which elicited Ca
2+
influx at the AZs (Middle)
and whole-cell Ca
2+
currents (Lower) within an individual cell. (B)Fractional
activation of the whole-cell Ca
2+
current: data from 25 IHCs (gray) and average
(black). The red circle and error bars indicate the mean and SD of V
0.5
of these
25 IHCs. (C) Fractional activation of 210 AZs, which are coded by colors
according to their ΔF
−7mV
/F
0
. The white circle and bars represent the mean and
SD of the V
0.5
of Ca
2+
influx at these AZs. (Dand E) Histograms of V
0.5
(D)or
slope factor k (E) of all AZs along with mean and SD. (F) Relation of V
0.5
and
k of the 210 AZs with color indicating their maximal Ca
2+
influx. The gray
dashed line represents a line fit to the k vs. V
0.5
relationship with r=0.47.
4of10
|
www.pnas.org/cgi/doi/10.1073/pnas.1605737113 Ohn et al.

may be reconciled with the opposing modiolar–pillar gradient
of maximal Ca
2+
influx is discussed below.
Probing Molecular Candidate Mechanisms for Regulation of Synaptic
Ca
2+
Influx. How does the IHC, a compact presynaptic cell, estab-
lish heterogeneity of Ca
2+
influx among its AZs? Here, we con-
sidered two classes of candidate mechanisms: (i) differences in the
subunit composition of the Ca
2+
-channel complex and (ii)different
interaction partners. The pore-forming Ca
V
1.3αand the auxiliary
Ca
V
β
2
are the dominating subunits of IHC Ca
2+
-channel com-
plexes, but other subunits contribute (27, 28, 32, 33). Differences in
the biophysical properties of Ca
V
1.3 Ca
2+
channels among IHC
AZs might arise from the preferential abundance of specific aux-
iliary subunits (32, 33) or splice variants of the pore-forming α
subunit (34, 35).
Indeed, analysis of splice variants with short C terminus or long
C terminus, both expressed by IHCs (36), in HEK293 cells
revealed more hyperpolarized activation and higher open proba-
bility for short splice variants of the channel (34, 35). Here, we
studied synaptic Ca
2+
influx in IHCs of knock-in mice expressing
the pore-forming Ca
V
1.3αwith an HA-tag in exon 49 (36). The
HA-tag disrupts the C-terminal automodulatory domain in the
long Ca
V
1.3αsplice variant. This preserves potential C-terminal
protein interaction domains but when expressed in HEK293 cells
functionally converts the gating of “long”splice variants to that of
“short”ones. Therefore, we expected to find more hyperpolarized
activation of synaptic Ca
2+
influx. However, the mean V
0.5
(ttest)
and its variance (Levene’s test) were unchanged (Fig. 5 A–C),
whereas the maximal AZ Ca
2+
influx (Fig. 5 Dand E,P<0.005,
Wilcoxon rank sum test) and whole-cell Ca
2+
current (Fig. 5F)
V0.5 (mV)
apical
basal
k (mV)
# of synapses
511
10
20
30
40
0
pillar
(abneural)
modiolar
(neural)
-40 -30 -20
modiolar
(neural)
6810
pillar
(abneural)
k (mV)
AB
CD
apical
basal
6
μm
6
μm
-40 -30 -20
10
20
30
0
V0.5 (mV)
***
pillar
(n=102)
modiolar
(n=98)
# of synapses
V0.5, mean =
-27.2 ± 3.7 m
V
-28.6 ± 4.0 m
V
pillar
(n=102)
modiolar
(n=98)
Fig. 4. Spatial gra dients of voltage-dependent act ivation of AZ Ca
2+
influx.
(Aand C) Polar charts display V
0.5
(A)andk(C)of210AZsof25IHCsasa
function of AZs location within IHCs (as in Fig. 2A). (Band D)(Upper) Histograms
of V
0.5
(B)ork(D) of AZs on the modiolar side (blue) and pillar side (red). (Lower)
The box plots of V
0.5
or k for pillar and modiolar AZs. V
0.5
was more hyper-
polarized for pillar AZs than for modiolar AZs (P<0.001, Wilcoxon rank sum
test). Modiolar and pillar AZs did not show statistically significant differences in k.
400
C
WT
600
200
0
806040200
Spontaneous rate (spikes/s)
Time ( ms)
WT
GH
I
Cumulative fraction
400
300
200
100
0
Spike rate (spikes/s)
100806040200-20
Dynamic range
x10%
x
90%
100
Steady state
rate (spikes/s)
Onset rate
(spikes/s)
500
400
300
200
300200
0 50 100 150
0
1
Dynamic range (dB SPL)
CaV1.3HA/HA
CaV1.3HA/HA
Spike rate (spikes/s)
25
15
5
0
-30 -10
WT
CaV1.3HA/HA
# of active zones
A
DE
2
1.5
1
0.5
0
**
CaV1.3HA/HA
∆F-7mV / F0
25
15
5
0
43210
WT
# of active zones
CaV1.3HA/HA
∆F-7mV / F0
-80 -40 0 40
-50
-150
Vm (mV)
ICa (pA)
WT
Cav1.3HA/HA
***
F
-80 -40 0 20-20-60
0
1.0
0.5
Vm (mV)
WT
Cav1.3HA/HA
Pactivation, ICa
V0.5, mean=
-26. 2 ± 1. 3 m
V
-26. 9 ± 2. 2 m
V
k =
7.4 ± 0. 3 mV
7.5 ± 0. 7 mV
-30
-25
-20
0
B
0
10
20
30
40
50
WT Ca
V
1.3
HA/HA
WT CaV1.3HA/HA
V0.5(mV)
V0.5(mV)
0.5
Sound pressure level (dB SPL)
WT
-20-40
Fig. 5. Disrupting the C-terminal regulatory domain of Ca
V
1.3 does enhance
maximal synaptic Ca
2+
influx but does not alter V
0.5
heterogeneity and sound
encoding. (A) Distributions of V
0.5
oftheAZCa
2+
influx in Ca
v
1.3
HA/HA
(red, 76 AZs
of 10 IHCs) and WT (black, 89 AZs of 10 IHCs) mice. Experiments were performed
as in Fig. 3. (B) Box-whisker charts of V
0.5
of the AZ Ca
2+
influx: nonsignificant
differences for Ca
v
1.3
HA/HA
.(C) Comparable fractional activation of Ca
2+
current as
a function of voltage for IHCs of both genotypes. Neither V
0.5
nor k has significant
differences observed between the two different mouse strains. Experiments and
analysis were conducted as described in SI Methods.(D) Distributions of maximal
AZ Ca
2+
influx in Ca
v
1.3
HA/HA
(red, 79 AZs of 10 IHCs) and control (black, 94 AZs of
10 IHCs) mice. Experiments were performed as in Fig. 1B.(E) Box-whisker charts of
maximal AZ Ca
2+
influx in Ca
v
1.3
HA/HA
IHCs, which was significantly stronger than
that of WT IHCs (P<0.01, Wilcoxon rank sum test). (F) Current–voltage re-
lationship of the whole-cell Ca
2+
current of IHC in WT (black) or Ca
v
1.3
HA/HA
(red).
The Ca
v
1.3
HA/HA
IHC has larger maximal Ca
2+
current than the WT cells (P<0.001,
I
Ca,−7mV,WT
=−194.6 ±6.6 pA, I
Ca,−7mV,HA
=−234.6 ±10.6 pA). The errors in this
panel are SEM. (G) Histogram showing the spontaneous rate distribution of
Ca
V
1.3
HA/HA
and WT SGNs. (H) Average poststimulus time histogram of SGN re-
sponses to 50-ms suprathreshold tone bursts at the characteristic frequency. Both
the onset (average firing rate in a 1-ms window after median first spike latency)
and steady-state rate (average firing rate over the last 5-ms window of 50-ms tone
bursts) did not differ significantly between Ca
V
1.3
HA/HA
and WT SGNs. (I) Rate-level
functions (Right) and dynamic range (Left)forCa
V
1.3
HA/HA
and WT SGNs. Dynamic
range is defined as the range of sound pressure levels in which the rate-level
function showed a rate increase between 10% and 90% of the difference be-
tween spontaneous and maximum rate. No statistically significant difference
between the dynamic ranges was observed between Ca
V
1.3
HA/HA
and WT SGNs
(19 for Ca
V
1.3
HA/HA
vs. 10 for WT SGNs, not significant by Wilcoxon rank sum test).
Ohn et al. PNAS Early Edition
|
5of10
NEUROSCIENCE PNAS PLUS

were larger. Next, we studied sound encoding at the single SGN
level in Ca
V
1.3
HA/HA
mice and found a normal distribution of
spontaneous firing rates (Fig. 5G), normal sound-evoked firing
rates (Fig. 5H), and normal sound-pressure dependence of firing
(Fig. 5I). The observed lack of effect of the C-terminal Ca
V
1.3α
manipulation on the voltage dependence of Ca
2+
current activa-
tion may be due to strong regulation of Ca
V
1.3 in IHCs by in-
teractions with Ca
2+
-binding proteins (37–39). The increased
maximal Ca
2+
influx, likely resulting from an increased open
probability (34), apparently did not change SGN firing behavior.
AZ proteins known to directly or indirectly interact with the
Ca
2+
-channel complex include bassoon (40) and rab-interacting
protein (RIM) (41, 42). Both positively regulate the number of
Ca
2+
channels at the IHC AZ (15, 43). Bassoon, moreover, con-
tributes to establishing heterogeneity of maximal Ca
2+
influx
whereas RIM2αdoes not seem to be required. Another candidate
interacting partner is harmonin, which regulates Ca
V
1.3 Ca
2+
channels, likely by binding of one of its PDZ domains to the ITTL
motif of the extreme Ca
V
1.3 C terminus (44, 45). Here, we
revisited the regulation of IHC AZs by harmonin in deaf-circler
mice (dfcr) (46), reasoning that site-specific expression of har-
monin might contribute to the heterogeneity of maximal Ca
2+
influx and voltage dependence of activation. Dfcr IHCs had
weaker RIBEYE-peptide fluorescence regardless of the AZ po-
sition (Fig. S5A,P<0.05, Wilcoxon rank sum test). AZ Ca
2+
influx
and whole-cell Ca
2+
current activated at more hyperpolarized
potentials (Fig. S5 Band C,P<0.05, ttest and Wilcoxon rank sum
test, respectively). However, this V
0.5
shift was observed for both
modiolar and pillar dfcr AZs (both P<0.05, one-way ANOVA
with Tukey’s post hoc test), arguing against a substantial contri-
bution of harmonin to presynaptic heterogeneity.
The whole-cell Ca
2+
influx was enhanced (P<0.05, Wilcoxon
rank sum test), but the mean synaptic Ca
2+
influx was not (Fig.
S5D), suggesting greater extrasynaptic Ca
2+
influx in dfcr IHCs. To
test for IHC immaturity and the greater extrasynaptic Ca
2+
influx
that accompanies it (29), we immunolabeled for Ca
V
1.3 and large-
conductance Ca
2+
-activated K
+
channels (BK channels, Fig. S5 E
and F). However, Ca
V
1.3 was confined to synapses, which seemed
present at normal numbers, and BK channels clustered at the IHC
neck, both indicating IHC maturity. We speculate that the V
0.5
shift
reflects the absence of Ca
V
1.3 regulation by harmonin or might
relate to impaired mechanoelectrical transduction in dfcr mice (47).
Gipc3 Disruption Causes a Hyperpolarized Shift in Ca
2+
Channel Activation
and Enhances Spontaneous SGN Firing. Another PDZ protein and
candidate regulator of Ca
2+
channels and AZ heterogeneity is Gipc3
(GAIP interacting protein, C terminus 3). Defects of the human
GIPC3 gene cause human deafness (48, 49) and Gipc3 disruption in
mice lead to audiogenic seizures and progressive hearing loss (48).
GIPC proteins share a central PDZ domain flanked by GIPC ho-
mologous domains. Gipc1 has broad roles in intracellular protein
trafficking and in hair cells is required for development of stereo-
ciliary bundles and planar cell polarity (50–52). We studied Black
Swiss (BLSW, hereafter Gipc3 mutant) mice that carry a missense
mutation in the Gipc3 gene (c.343G >A), which replaces a highly
conserved glycine with an arginine at position 115 (Gly115Arg) lo-
cated within the PDZ domain. Early-onset hearing impairment in
Gipc3 mutant mice has been attributed to dysfunction of hair cell
stereocilia, although its localization within hair cells resembles a
cytoplasmic vesicular pattern similar to myosin VI (48). In the
brain, glutamate release was shown to depend on interaction be-
tween myosin VI and its binding partner Gipc1 (50). In hair cells,
a synaptic function for Gipc proteins is yet unknown.
In Gipc3 mutant IHCs we found an increased whole-cell Ca
2+
current (Fig. 6 Aand D,P<0.05, Wilcoxon rank sum test) and a
more hyperpolarized V
0.5
of activation [Fig. 6B,P=0.004, Wilcoxon
rank sum test; Gipc3 mutant IHCs: V
0.5
=−31.2 ±1.4 mV, n=
19 compared with C57BL/6J control IHCs (WT): V
0.5
=−24.1 ±
1.8 mV, n=20]. Moreover, we found Ca
2+
current inactivation
to be reduced in Gipc3 mutants when comparing Ca
2+
currents
elicited by 500-ms-long depolarizations (I
res
/I
peak
0.59 ±0.04 for
10 Gipc3 mutant IHCs vs. 0.41 ±0.03 for 8 WT IHCs, P=0.007,
Wilcoxon rank sum test). We also observed enhanced exocytosis
reported as membrane capacitance increments (Fig. 6 Cand D,
significant for 50- and 100-ms-long depolarizations), likely as a
consequence of the increased Ca
2+
current. At the AZ level, the
distributions of maximal Ca
2+
influx (Fig. 7A) and RIBEYE-
peptide fluorescence (Fig. 7B) and were comparable to WT, but
the V
0.5
distribution was broader in Gipc3 mutant IHCs and AZs
activated at more hyperpolarized potentials on average (Fig. 6C,
P<0.01 for variance, Brown–Forsythe test and P<0.01 for
mean, Wilcoxon rank sum test). Opposite that in WT, maximal
AZ Ca
2+
influx (Fig. 7D) exhibited a pillar–modiolar gradient in
A
CD
Vm (mV) 50
-50
-150
-200 ICa (pA)
-80 -40 40
-100
0
Gipc3
*
WT
1.32 x WT
1.0
0.8
0.6
0.4
0.2
0
-80 -60 -40 -20 0
Vm (mV)
Fractional activation
1.2
WT
Gipc3
Boltzman fit WT
Boltzman fit Gipc3
ΔC
m
(fF)
0
50
100
150 *
*
B
WT
Gipc3
Duration of depolarization (ms)
*
*
020406080100
0
5
10
15
20
Q
Ca
(pC)
*
*
Duration of depolarization (ms)
020406080100
WT
Gipc3
*
WT Gipc3 WT Gipc3
50 ms
100 pA
200 ms
50 fF
Fig. 6. Disruption of Gipc3 results in increased amplitude and hyperpolarized
shift of activation of IHC Ca
2+
influx as well as enhanced exocytosis. (A)Av-
erage steady-state I
Ca
-V of WT IHCs (black trace, 11 cells) and Gipc3 IHCs (red
trace, 7 cells) acquired with 10-ms-long depolarizations (steady-state Ca
2+
currents during depolarization): hyperpolarized voltage dependence and in-
creased amplitude (I
Ca,−24 mV, WT
=−110.6 ±13 pA, I
Ca,−24 mV, Gipc3
=−166.2 ±
17.3 pA, <0.05, Wilcoxon rank sum test). Data are shown as mean ±SEM.
(B) Fraction al activation of C a
2+
current as a function of voltage for IHCs of WT
IHCs (black trace, 19 cells) and Gipc3 IHCs (red trace, 20 cells). V
0.5
of Gipc3 IHCs
is significantly different from that of WT IHCs (<0.05, Wilcoxon rank sum test).
Data are shown as mean ±SEM. (C)(Upper) Average capacitance increments
(ΔCm) traces of WT IHCs (black trace, 11 cells) and Gipc3 IHCs (red trace, 8 cells)
in response to 100-ms-long step depolarization to −14 mV. (Lower)Average
ΔCm of p14-16 WT IHCs (black, 11 cells) and Gipc3 IHCs (red, 8 cells) for de-
polarizations of different durations. Capacitance increments are significantly
increased in Gipc3 IHCs at depolarizations of 50 ms (WT: 32.4 ±3.4 fF, Gipc3:
65.9 ±8.7 fF, <0.01, Wilcoxon rank sum test) and 100 ms (WT: 54.3 ±4.7 fF,
Gipc3:111.7±19.4 fF, <0.01, Wilcoxon rank sum test). Data are shown as
mean ±SEM. (D)(Upper)AverageCa
2+
currents (Q
Ca
) traces of WT IHCs (black
trace, 11 cells) and Gipc3 IHCs (red trace, 8 cells) in response to 100-ms-long
step depolarization to −14 mV. (Lower)AverageQ
Ca
corresponding to C). Ca
2+
currents are significantly increased in Gipc3 IHCs (red) at depolarizations of
50 ms (WT: 6.1 ±0.4 pC, Gipc3:8.3±0.6 pC, <0.01, Wilcoxon rank sum test)
and 100 ms (WT: 11.2 ±0.8 pC, Gipc3:15.8±1.3 pC, <0.05, Wilcoxon rank sum
test). Data are shown as mean ±SEM.
6of10
|
www.pnas.org/cgi/doi/10.1073/pnas.1605737113 Ohn et al.

Gipc3 mutants. As a result, the pillar AZs, operating at even
more hyperpolarized potentials than in WT, also had stronger
maximal Ca
2+
influx on average (Fig. 7 D–F).
To relate these presynaptic observations to SGN firing prop-
erties, we performed in vivo recordings of SGN firing in Gipc3
mutant mice after onset of hearing [postnatal days (p) 14–25].
Gipc3 mutant mice showed elevated auditory brainstem response
(ABR) thresholds at p14 and p21 (Fig. S6 Aand B). Single SGN
frequency tuning was broad and thresholds at the characteristic
frequency in Gipc3 mutant mice were elevated on average by
28 dB (2 wk) and 72 dB (3 wk) (Fig. S6 B–D). These effects are
attributable to impaired mechanoelectrical transduction and co-
chlear amplification (48). This stereociliary impairment precluded
assessment of the influence of the observed hyperpolarized acti-
vation of Ca
2+
channels on sound-response properties. However,
in the absence of sound the spontaneous spike rate inSGNs comes
from Ca
2+
-dependent transmitter release from the hair cell at its
resting potential (53), allowing us to assess the effects of shifted
voltage dependence of AZ Ca
2+
influx on baseline release. De-
spite impaired transduction of sound, spontaneous SGN firing
01 32
-40
-10
-20
-30
∆F-7mV / F0
V
0.5
(mV)
50
40
30
20
10
0
# of active zones
25
20
15
10
5
0
# of active zones
01 3254
∆F-7mV / F0
WT: n = 233 (25)
Gipc3: n = 152 (19)
30
20
10
0
-40 -30 -20 -10
60
40
20
0
35
28
21
14
7
0
7654321
20
15
10
5
0
# of synapses
# of synapses
Fribbon / Fnearby
V
0.5
(mV)
# of active zones
# of active zones
V
0.5
(mV)
**
V
0.5
(mV)
-35
-30
-25
-20
-15
-40
***
AB C
∆F-7mV / F0 < 1 1 < ∆F-7mV / F0 < 2
WT: n = 253 (25)
Gipc3: n = 166 (19)
F
2.5
2.01.51.0
0.5
0
Gipc3
modiolar
pillar
pillar - modiolar axis (μm)
∆F-7mV / F0
-40
-35
-30
-25
-20
-15
x10
-3
pillar - modiolar axis (μm)
-40 -35 -30 -25 -20 -15
Gipc3
modiolar
pillar
V0.5 (mV)
V
0.5
(mV)
***
***
DE
-6 -4 -2 0 2 4 6
4
3
2
1
0
-6 -4 -2 0 2 4 6
V
0.5
(mean ± S.D.)
-29.0 ± 5.2
(n=142)(19)
-27.4 ± 4.7
(n=225)(25)
r = -0.43
r = 0.19
r = 0.37
r = 0.23
Gipc3
WT
Gipc3
WT
Gipc3WT
-35
-30
-25
-20
-15
-40
-35
-30
-25
-20
-15
-40
∆F-7mV / F0
Fig. 7. Disruption of Gipc3 in mice shifts the activation of Ca
2+
influx to more hyperpolarized potentials and reverses the modiolar–pillar gradient of maximal
Ca
2+
influx. (A) Distribution of maximal AZ Ca
2+
influx in IHCs of Gipc3 mutant (red trace, n=152, 19 cells) and WT mice (blue trace, 225 AZs of 25 cells). The
experimental protocol is the same as Fig. 1B.(B) Distribution of the AZ TAMRA-peptide fluorescence in Gipc3 mutant IHCs (red trace, 166 AZs, 19 IHCs) and WT
IHCs (blue trace, 253 AZs of 25IHCs). (C,Left)DistributionofV
0.5
of the voltage-dependent activation of AZ Ca
2+
influx in Gipc3 mutant IHCs (red trace, 142 AZs of
19 IHCs) and WT IHCs (blue trace, n=225, 25 cells). The experimental protocol is the same as Fig. 3A.(C,Right) Box plot of V
0.5
of voltage-dependen t activatio n of
AZ Ca
2+
influx in Gipc3 mutant IHCs: V
0.5
was significantly more hyperpolarized in Gipc3 mutant IHCs (P<0.01, Wilcoxon rank sum test). (D,Upper) Opposing
gradients of maximal AZ Ca
2+
influx in Gipc3 (red circle, 128 AZs of 19 IHCs) and WT (blue circle, 226 AZs of 25 IHCs) along the modiolar–pillar axis. Solid lines are
their fitting lines, r indicates their correlation coefficients. (D,Lower) The box plots summarize the distributions of maximal Ca
2+
influx of pillar and modiolar AZs
for Gipc3 mutant IHCs. The maximal AZ Ca
2+
influx of pillar AZs was significantly stronger than that of pillar AZs (P<0.001, Wilcoxon rank sum test). (E)(Upper)
Spatial distribution of the V
0.5
in Gipc3 mutant IHCs (red circle, 119 AZs of 19 IHCs) and WT IHCs (blue, 194 AZs, 21 IHCs) along the modiolar–pillar axis. Solid lines
arelinefitstothedata.(E,Lower) Box plots summarize the V
0.5
distributions of pillar and modiolar AZs in Gipc3 mutant IHCs. The V
0.5
of pillar AZs was sig-
nificantly more hyperpolarized than that of modiolar AZs (P<0.001, Wilcoxon rank sum test). (F,Left) The relationship of maximal AZ Ca
2+
influx and V
0.5
(red: Gipc3,
blue: WT). (F,Right) Box-whisker charts summarize the distributions of V
0.5
in subgroups of their ΔF
−7mV
/F
0
: smaller than 1 (Left) and between 1 and 2 (Right). V
0.5
of
AZs from Gipc3 mutant IHCs were more hyperpolarized (−30.2 ±4.9 mV) than those of WT IHCs (−28.4 ±3.5 mV, P<0.001, Wilcoxon rank sum test) for the subgroup
of AZs with ΔF
−7mV
/F
0
is between 1 and 2, whereas it was statistically indistinguishable for the AZ subgroup with smaller maximal Ca
2+
influx.
Ohn et al. PNAS Early Edition
|
7of10
NEUROSCIENCE PNAS PLUS

rates were elevated in Gipc3 mutant mice (P<0.05, one-sided
Kolmogorov–Smirnov test). In contrast to WT animals, where
about 50% of SGNs had a spontaneous rate below 20 Hz, only
16% of SGNs in Gipc3 mutant mice exhibited spontaneous rates
below 20 Hz (Fig. 8A). Some low-spontaneous-rate SGNs, which
generally have higher thresholds, may not have been activated by the
highest-intensity search stimuli (100-dB sound pressure level). Re-
gardless, we found a greater abundance of SGNs with high spon-
taneous firing rates (28% with rates >50 Hz vs. only 12% in WT).
Despite elevated thresholds in Gipc3 mutant mice that most
likely resulted from impaired mechanotransduction and cochlear
amplification, firing rates at sound onset were elevated when
probed at 30 dB above threshold with tone bursts at the char-
acteristic frequency in 2- to 3-wk-old mice. SGNs had onset rates
of 451.4 ±27.0 Hz for Gipc3 mutant (n=25) vs. 347.1 ±26.1 Hz
for WT (n=24, P<0.05, Wilcoxon rank sum test). In contrast,
the adapted firing rates were indistinguishable from WT (Fig.
8B). Consistent with the defect of cochlear amplification, rate vs.
level functions had steeper slopes and significantly narrower
dynamic ranges (Fig. 8C). Recovery from adaptation, most likely
reflecting presynaptic replenishment of the readily releasable
pool, proceeded at normal pace in Gipc3 mutant mice (Fig. S6F).
In principle, the increased spontaneous discharge rates of SGNs
in Gipc3 mutant mice could have resulted from a more depolar-
ized IHC resting membrane potential, potentially due to reduced
basolateral K
+
conductances (48). However, IHC resting mem-
brane potentials were unaltered (Fig. S7A), BK channels clustered
at the neck as in WT (Fig. S7B), and BK currents were normally
sized (Fig. S7C)inGipc3 mutant IHCs at p14–16, indicating
normal IHC maturation (54). This notion was further corrobo-
rated by the normal number of ribbon synapses (Fig. S7D). In
conclusion, disruption of Gipc3 produced a shift of Ca
2+
-current
activation toward hyperpolarized potentials, reversed the modiolar–
pillar gradient of maximal Ca
2+
influx at IHC AZs, and increased
spontaneous firing rates in SGNs.
Discussion
The ear encodes changes in sound pressure over six orders of
magnitude using SGNs that change their firing rates over different
fractions of the audible range, but the mechanisms underlying
such dynamic range fractionation are unknown. Therefore, we
used fluorescence imaging to characterize heterogeneity among
AZs in cochlear IHCs as a candidate mechanism. Our study in-
dicates that IHCs decompose information on sound intensity into
different outputs by varying the Ca
2+
influx among their AZs. Two
key determinants of presynaptic Ca
2+
influx, the voltage dependence
and the number of Ca
2+
channels, differed greatly among the AZs
within IHCs and exhibited shallow opposing spatial gradients along
the pillar–modiolar IHC axis. AZs of the pillar face, on average,
were smaller but activated at more hyperpolarized potentials and,
hence, are likely presynaptic to high-spontaneous-rate, low-threshold
SGNs. The number of Ca
V
1.3 Ca
2+
channels and maximal synaptic
Ca
2+
influx, on average, were greater for modiolar AZs, which also
were larger in size. We propose that these AZs, given their more
depolarized operation range, are recruited by stronger sounds and
likely drive low-spontaneous-rate, high-threshold SGNs. Disruption
of Gipc3 reversed the normal modiolar–pillar gradient of maximal
AZ Ca
2+
influx, shifted Ca
2+
channel activation to more hyper-
polarized potentials, and increased the fraction of SGNs with high-
spontaneous firing rate.
Dynamic Range Fractionation Through Heterogeneity of Synaptic
Voltage Dependence. Each presynaptic AZ in a given IHC is con-
trolled by a common potential and provides the sole excitatory
input to “its”postsynaptic SGN. How the operating range of Ca
2+
influx at an AZ matches the IHC resting and receptor potential
critically determines its transmitter release and hence the spon-
taneous and sound-evoked firing of the postsynaptic SGN. Be-
cause technically challenging in vivo tight-seal patch-clamp
recordings from IHCs have not yet been achieved there is un-
certainty about the resting (in quiescence) and the receptor
(during sound stimulation) potential. Pioneering recordings with
sharp electrodes from guinea pig IHCs indicated a resting poten-
tial of −40 mV and maximal receptor potentials between ∼10 and
30 mV (1), and a patch-clamp study trying to emulate physiological
ionic conditions in the explanted gerbil organ of Corti estimated a
resting potential of −55 mV (55). Regardless of the precise resting
potential it is evident that Ca
2+
influx is partially active at least at a
subset of IHC AZs (Fig. 3) and the relatively depolarized IHC
resting potential is expected to facilitate Ca
2+
influx and transmitter
release (56, 57). Moreover, the range covered by Ca
2+
activation
at the various AZs (V
0.5
spanning from −38 mV to −18 mV) matches
the reported range of IHC receptor potentials very well (Fig. 3).
The SGNs with identical frequency tuning but different sound-
response properties presumably receive input from the same IHC
and collectively convey acoustic information across the entire audible
range of sound pressures to the brain. Candidate mechanisms
underlying the diversity of SGN response properties include pre-
synaptic (7, 11, 12), postsynaptic (13), and efferent (17, 18)
mechanisms. Here, we asked how IHCs might fractionate the
auditory signal for different parallel outputs through heteroge-
neous Ca
2+
influx among AZs. Fast 3D live imaging enabled us to
analyze functional and morphological properties of most, if not all,
AZs as a function of position within an individual IHC. In parallel,
we performed semiquantitative confocal microscopy of immuno-
labeled IHC synapses. AZs differed considerably in size, number,
and voltage dependence of Ca
2+
channels whereby differences
C
A
050 100 150
0
1
Spontaneous rate (spikes/s)
Cumulative fraction
B
WT
Dynamic range (dB SPL)
**
WT Gipc3
Gipc3
300
250
200
150
100
50
100806040200-20
0
Sound pressure level (dB SPL)
WT Gipc3
Spike rate (spikes/s)
WT
Onset rate
(spikes/s)
Steady state
rate (spikes/s)
Gipc3
0
10
20
30
*
0.5
500
400
300
200
100
0
806040200
Time (ms)
Spike rate (spikes/s)
WT
Gipc3 600
400
200
100 150 200
Fig. 8. Gipc3 disruption enhances spontaneous SGN firing. (A) Cumulative
histogram of spontaneous rate distribution: no major overlap of the data. WT
(C57BL/6) data were taken from ref. 19. The spontaneous rate of Gipc3 SGNs
showed a shift of mean about 21.5 Hz in comparison with that of WT SGNs.
The null hypothesis that the spontaneous discharge rates Gipc3 and WT SGNs
came from populations with the same distribution was rejected, and the al-
ternative hypothesis that the cumulative distribution function of WT was
larger than that of Gipc3 SGNs was favored (P<0.05, one-sided Kolmogorov–
Smirnov test). (B) Peristimulus time histogram of 3-wk-old Gipc3 and WT SGNs
to 50-ms suprathreshold tone bursts at the characteristic frequency. The onset
firing rates of Gipc3 were elevated (451.4 ±27.0 Hz for Gipc3 mutant, 25 SGNs
vs. 347.1 ±26.1 Hz for WT, 24 SGNs, P<0.05). The steady-state firing rates did
not differ significantly (159.5 ±8.6 Hz for Gipc3 mutant, 25 SGNs vs. 154.3 ±
10.4 Hz for WT, 24 SGNs, not significant by Wilcoxon rank sum test). (C)Rate-
intensity functions (Left) and dynamic range (Right) of sound encoding in
Gipc3 and WT SGNs. Gipc3 SGNs had a narrower dynamic range than WT SGNs
(9.9 ±0.6 dB for Gipc3 p14–25, 19 SGNs vs. 13.7 ±1.1 dB for WT p14–21, 20
SGNs, P<0.01, Wilcoxon rank sum test).
8of10
|
www.pnas.org/cgi/doi/10.1073/pnas.1605737113 Ohn et al.

among AZs within individual IHCs explained most of the variance
of the entire AZ population. The estimates for Ca
2+
-channel
number were positively correlated with our proxy of AZ size in
both live-imaging and immunohistochemistry experiments, even at
the level of the single IHC.
Modiolar AZs were larger, had more Ca
2+
channels, and
tended to show stronger maximal Ca
2+
influx. The hypothesis
that these AZs drive SGNs with lower spontaneous firing rate
and higher sound threshold, at first sight, seems counterintuitive
because a greater number of Ca
2+
channels is expected to elicit
more spontaneous and evoked release. However, we found that
modiolar AZs, on average, required more depolarization for
their activation (Fig. 4), consistent with the idea that modiolar
synapses drive high-threshold SGNs (2, 7, 8). Modiolar AZs that
hold more Ca
2+
channels may support higher maximal rates of
transmitter release, which could explain the higher susceptibility
of low-spontaneous-rate, high-threshold SGNs to excitotoxic
insult during acoustic overexposure (58). The operation of pillar
AZ at relatively hyperpolarized potentials may contribute to
high-spontaneous firing rates and responses to soft sounds found
for SGNs innervating the pillar side (2, 7). Based on the spatial
distribution of the voltage dependence of presynaptic Ca
2+
influx
found here for mice, and the spatial distribution of postsynaptic
spontaneous rates in cats, we hypothesize that the voltage de-
pendence of presynaptic Ca
2+
influx is a major determinant of
spontaneous and sound-driven SGN firing. This hypothesis is
supported by the more hyperpolarized V
0.5
of AZ Ca
2+
influx
in IHCs (Fig. 7) and the increase in SGN spontaneous firing rates
in Gipc3 mutant mice (Fig. 8), but not in in Ca
V
1.3
HA/HA
mice, for
which, too, we found more AZ Ca
2+
influx but no change in voltage
dependence. However, we note that different from classic ex-
periments in mature cats (2, 7) the present data were obtained
from mice and we cannot rule out some immaturity of the afferent
synaptic connectivity because we recorded soon after the onset of
hearing. Moreover, the variability of voltage dependence of pre-
synaptic Ca
2+
influx was high in both the modiolar and pillar
halves, whereby the differences between V
0.5
among modiolar
or pillar synapses could exceed that between the average pillar and
modiolar synapse. In addition, the spatial segregation of func-
tionally distinct classes of type SGNs at the level of IHC in-
nervation in mice might deviate from that described for cats.
Future experiments simultaneously addressing presynaptic
properties and postsynaptic firing will be required to further
test our hypothesis.
Regulation of the Number and Voltage Dependence of Ca
2+
Channels
at IHC AZs. Which mechanisms determine the number and volt-
age dependence of Ca
2+
channels at an AZ? The number of Ca
2+
channels at the presynaptic AZ is governed by the expression of
the specific Ca
2+
-channel subunits and splice variants (33, 36, 59)
as well as by that of scaffold proteins that tether Ca
2+
channels to
the AZ (15, 42, 43, 60, 61) and/or regulate their turnover (44) by
direct or indirect interaction with the channels. Moreover, sub-
unit composition as well as splice variants and interacting pro-
teins also modulate functional properties of the specific Ca
2+
channel (62). The Ca
2+
-channel complex at IHC AZs is likely
composed of splice variants of the pore-forming subunit Ca
V
1.3α
containing exons 43S or 43L (36), of the Ca
V
β2 (33), and less
likely of other Ca
V
βsubunits (32), and of a yet-to-be-identified
Ca
V
α2δsubunit. Ca
V
β2, the synaptic ribbon, and the presynaptic
scaffolds bassoon and RIM2αand βwere previously shown to
promote the abundance of Ca
V
1.3atIHCAZs(15,33,43),
whereas harmonin seems to reduce the number of synaptic Ca
V
1.3
(44). Work in zebrafish hair cells has revealed an intriguing mo-
lecular interplay between RIBEYE and Ca
V
1.3 channels, whereby
RIBEYE overexpression promotes the formation of Ca
V
1.3-positive
AZ-like specializations and synaptic Ca
2+
influx negatively regulates
RIBEYE abundance at the AZ (63, 64). The present study cor-
roborates the notion that the Ca
2+
-channel number is proportional
to ribbon size in cochlear hair cells (11, 30). We speculate that re-
lease rate for a given open probability scales linearly with the
number of Ca
2+
channels, assuming nanodomain coupling (29, 65,
66), regardless of the AZ size. Future simultaneous measurements
of AZ Ca
2+
influx and exocytosis or SGN firing will be required to
analyze the consequences of presynaptic heterogeneity for trans-
mitter release at IHC AZs.
How do IHCs form opposing gradients of AZ size, Ca
V
1.3
abundance, and voltage dependence of activation? The modiolar–
pillar gradient of AZ size was lost upon lesion of the lateral oli-
vocochlear efferents (18), and it is tempting to speculate that they
provide an instructive influence on the postsynaptic SGN terminal
and/or the IHC AZ. Here, we found that the intracellular gradient
of maximal Ca
2+
influx was altered in Gipc3 mutant mice. In
addition, Gipc3 disruption increased Ca
2+
influx and caused a
hyperpolarizing shift in its activation. These presynaptic changes
likely underlie the enhanced spontaneous and sound-onset firing
in Gipc3 mutant mice, although alternative explanations exist.
How the Gipc3 protein might be involved directly or indirectly for
establishing a modiolar–pillar gradient remains to be elucidated.
Considering analogy to Gipc1 (50, 52), it is tempting to speculate
that Gipc3 serves to adapt components of the Ca
2+
-channel
complex to motor proteins and thereby assists their trafficking to
AZs with a preference for the modiolar face. Signals instructing
polarized trafficking might originate from lateral olivocochlear
fibers, SGNs themselves, and/or the planar cell polarity that sets
the orientation of the apical hair bundle (67, 68) and involves
Gipc1 signaling (50). Future work should address the mechanism
by which Gipc3 operates to traffic and/or regulate Ca
2+
channels.
Methods
Research followed national animal care guidelines and was approved by the
University of Göttingen board for animal welfare and the animal welfare
office of the state of Lower Saxony. For details of patch-clamp and confocal
Ca
2+
imaging, immunohistochemistry and confocal imaging, extracellular
recordings from auditory nerve fibers, and data analysis see SI Methods.
ACKNOWLEDGMENTS. We thank S. Gerke, C. Senger-Freitag, and
N. Herrmann for expert technical assistance and Dr. Konrad Noben-Trauth
for providing us with Black Swiss mice (Gipc3 mutant). This work wassupported
by a scholarship from the German Academic Exchange Service (T.-L.O.), by a
fellowship from the Alexander von Humboldt Foundation (M.A.R.), funding
from MED-EL, an international project grant from Action on Hearing Loss,
and the Department of Otolaryngology at Washington University in
St. Louis. This work was also supported by the German Federal Ministry of
Education and Research through Bernstein Focus for Neurotechnology Grant
01GQ0810 (to T.M.) and the German Research Foundation through the Col-
laborative Research Center 889 [Projects A2 (to T.M.) and A6 (to N.S.)], Cen-
ter for Nanoscale Microscopy and Molecular Physiology of the Brain Grants
ECX 101 and FZT 103 (to T.M.), the Leibniz Program (T.M.), and Austrian
Science Fund Grant FWF F44020 (to Jörg Striessnig).
1. Russell IJ, Sellick PM (1983) Low-frequency characteristics of intracellularly re-
corded receptor potentials in guinea-pig cochlear hair cells. J Physiol 338(1):
179–206.
2. Liberman MC (1978) Auditory-nerve response from cats raised in a low-noise cham-
ber. J Acoust Soc Am 63(2):442–455.
3. Sachs MB, Abbas PJ (1974) Rate versus level functions for auditory-nerve fibers in cats:
Tone-burst stimuli. J Acoust Soc Am 56(6):1835–1847.
4. Taberner AM, Liberman MC (2005) Response properties of single auditory nerve fibers
in the mouse. J Neurophysiol 93(1):557–569.
5. Winter IM, Robertson D, Yates GK (1990) Diversity of characteristic frequency rate-
intensity functions in guinea pig auditory nerve fibres. Hear Res 45(3):191–202.
6. Zagaeski M, Cody AR, Russell IJ, Mountain DC (1994) Transfer characteristic of the
inner hair cell synapse: steady-state analysis. J Acoust Soc Am 95(6):3430–3434.
7. Merchan-Perez A, Liberman MC (1996) Ultrastructural differences among afferent
synapses on cochlear hair cells: Correlations with spontaneous discharge rate. J Comp
Neurol 371(2):208–221.
8. Liberman MC (1982) Single-neuron labeling in the cat auditory nerve. Science
216(4551):1239–1241.
Ohn et al. PNAS Early Edition
|
9of10
NEUROSCIENCE PNAS PLUS

9. Kiang NY, Pfeiffer RR, Warr WB, Backus AS (1965) Stimulus coding in the cochlear
nucleus. Trans Am Otol Soc 53:35–58.
10. Wen B, Wang GI, Dean I, Delgutte B (2009) Dynamic range adaptation to sound level
statistics in the auditory nerve. J Neurosci 29(44):13797–13808.
11. Frank T, Khimich D, Neef A, Moser T (2009) Mechanisms contributing to synaptic Ca
2+
signals and their heterogeneity in hair cells. Proc Natl Acad Sci USA 106(11):
4483–4488.
12. Grant L, Yi E, Glowatzki E (2010) Two modes of release shape the postsynaptic re-
sponse at the inner hair cell ribbon synapse. J Neurosci 30(12):4210–4220.
13. Liberman LD, Wang H, Liberman MC (2011) Opposing gradients of ribbon size and
AMPA receptor expression underlie sensitivity differences among cochlear-nerve/hair-
cell synapses. J Neurosci 31(3):801–808.
14. Kantardzhieva A, Liberman MC, Sewell WF (2013) Quantitative analysis of ribbons,
vesicles, and cisterns at the cat inner hair cell synapse: Correlations with spontaneous
rate. J Comp Neurol 521(14):3260–3271.
15. Frank T, et al. (2010) Bassoon and the synaptic ribbon organize Ca
2+
channels and
vesicles to add release sites and promote refilling. Neuron 68(4):724–738.
16. Meyer AC, et al. (2009) Tuning of synapse number, structure and function in the
cochlea. Nat Neurosci 12(4):444–453.
17. Ruel J, et al. (2001) Dopamine inhibition of auditory nerve activity in the adult
mammalian cochlea. Eur J Neurosci 14(6):977–986.
18. Yin Y, Liberman LD, Maison SF, Liberman MC (2014) Olivocochlear innervation
maintains the normal modiolar-pillar and habenular-cuticular gradients in cochlear
synaptic morphology. J Assoc Res Otolaryngol 15(4):571–583.
19. Wong AB, et al. (2013) Concurrent maturation of inner hair cell synaptic Ca
2+
influx
and auditory nerve spontaneous activity around hearing onset in mice. J Neurosci
33(26):10661–10666.
20. Ariel P, Hoppa MB, Ryan TA (2013) Intrinsic variability in Pv, RRP size, Ca
2+
channel
repertoire, and presynaptic potentiation in individual synaptic boutons. Front
Synaptic Neurosci 4:9.
21. Holderith N, et al. (2012) Release probability of hippocampal glutamatergic terminals
scales with the size of the active zone. Nat Neurosci 15(7):988–997.
22. Sheng J, et al. (2012) Calcium-channel number critically influences synaptic strength
and plasticity at the active zone. Nat Neurosci 15(7):998–1006.
23. Roberts WM, Jacobs RA, Hudspeth AJ (1990) Colocalization of ion channels involved
in frequency selectivity and synaptic transmission at presynaptic active zones of hair
cells. J Neurosci 10(11):3664–3684.
24. Issa NP, Hudspeth AJ (1996) The entry and clearance of Ca
2+
at individual presynaptic
active zones of hair cells from the bullfrog’s sacculus. Proc Natl Acad Sci USA 93(18):
9527–9532.
25. Tucker T, Fettiplace R (1995) Confocal imaging of calcium microdomains and calcium
extrusion in turtle hair cells. Neuron 15(6):1323–1335.
26. Zenisek D, Horst NK, Merrifield C, Sterling P, Matthews G (2004) Visualizing synaptic
ribbons in the living cell. J Neurosci 24(44):9752–9759.
27. Brandt A, Striessnig J, Moser T (2003) CaV1.3 channels are essential for development
and presynaptic activity of cochlear inner hair cells. J Neurosci 23(34):10832–10840.
28. Platzer J, et al. (2000) Congenital deafness and sinoatrial node dysfunction in mice
lacking class D L-type Ca
2+
channels. Cell 102(1):89–97.
29. Wong AB, et al. (2014) Developmental refinement of hair cell synapses tightens the
coupling of Ca
2+
influx to exocytosis. EMBO J 33(3):247–264.
30. Martinez-Dunst C, Michaels RL, Fuchs PA (1997) Release sites and calcium channels in
hair cells of the chick’s cochlea. J Neurosci 17(23):9133–9144.
31. Moreno CM, et al. (2016) Ca
2+
entry into neurons is facilitated by cooperative gating
of clustered CaV1.3 channels. eLife 5:e15744.
32. Kuhn S, et al. (2009) Ba
2+
currents in inner and outer hair cells of mice lacking the
voltage-dependent Ca
2+
channel subunits beta3 or beta4. Channels (Austin) 3(5):
366–376.
33. Neef J, et al. (2009) The Ca
2+
channel subunit beta2 regulates Ca
2+
channel abun-
dance and function in inner hair cells and is required for hearing. J Neurosci 29(34):
10730–10740.
34. Bock G, et al. (2011) Functional properties of a newly identified C-terminal splice
variant of Cav1.3 L-type Ca
2+
channels. J Biol Chem 286(49):42736–42748.
35. Tan BZ, et al. (2011) Functional characterization of alternative splicingin the C terminus
of L-type CaV1.3 channels. JBiolChem286(49):42725–42735.
36. Scharinger A, et al. (2015) Cell-type-specific tuning of Cav1.3 Ca
2+
-channels by a
C-terminal automodulatory domain. Front Cell Neurosci 9:309.
37. Cui G, et al. (2007) Ca
2+
-binding proteins tune Ca
2+
-feedback to Cav1.3 channels in
mouse auditory hair cells. J Physiol 585(Pt 3):791–803.
38. Schrauwen I, et al. (2012) A mutation in CABP2, expressed in cochlear hair cells, causes
autosomal-recessive hearing impairment. Am J Hum Genet 91(4):636–645.
39. Yang PS, et al. (2006) Switching of Ca
2+
-dependent inactivation of Ca(v)1.3 channels
by calcium binding proteins of auditory hair cells. J Neurosci 26(42):10677–10689.
40. Davydova D, et al. (2014) Bassoon specifically controls presynaptic P/Q-type Ca
2+
channels via RIM-binding protein. Neuron 82(1):181–194.
41. Gebhart M, et al. (2010) Modulation of Cav1.3 Ca
2+
channel gating by Rab3 inter-
acting molecule. Mol Cell Neurosci 44(3):246–259.
42. Kaeser PS, et al. (2011) RIM proteins tether Ca
2+
channels to presynaptic active zones
via a direct PDZ-domain interaction. Cell 144(2):282–295.
43. Jung S, et al. (2015) Rab3-interacting molecules 2αand 2βpromote the abundance of
voltage-gated Ca
V
1.3 Ca
2+
channels at hair cell active zones. Proc Natl Acad Sci USA
112(24):E3141–E3149.
44. Gregory FD, et al. (2011) Harmonin inhibits presynaptic Cav1.3 Ca²⁺channels in mouse
inner hair cells. Nat Neurosci 14(9):1109–1111.
45. Gregory FD, Pangrsic T, Calin-Jageman IE, Moser T, Lee A (2013) Harmonin enhances
voltage-dependent facilitation of Cav1.3 channels and synchronous exocytosis in
mouse inner hair cells. J Physiol 591(13):3253–3269.
46. Johnson KR, et al. (2003) Mouse models of USH1C and DFNB18: Phenotypic and
molecular analyses of two new spontaneous mutations of the Ush1c gene. Hum Mol
Genet 12(23):3075–3086.
47. Grillet N, et al. (2009) Harmonin mutations cause mechanotransduction defects in
cochlear hair cells. Neuron 62(3):375–387.
48. Charizopoulou N, et al. (2011) Gipc3 mutations associated with audiogenic seizures
and sensorineural hearing loss in mouse and human. Nat Commun 2:201.
49. Rehman AU, et al. (2011) Mutations of GIPC3 cause nonsyndromic hearing loss
DFNB72 but not DFNB81 that also maps to chromosome 19p. Hum Genet 130(6):
759–765.
50. Giese AP, et al. (2012) Gipc1 has a dual role in Vangl2 trafficking and hair bundle
integrity in the inner ear. Development 139(20):3775–3785.
51. Naccache SN, Hasson T, Horowitz A (2006) Binding of internalized receptors to the
PDZ domain of GIPC/synectin recruits myosin VI to endocytic vesicles. Proc Natl Acad
Sci USA 103(34):12735–12740.
52. Yano H, et al. (2006) BDNF-mediated neurotransmission relies upon a myosin VI
motor complex. Nat Neurosci 9(8):1009–1018.
53. Robertson D, Paki B (2002) Role of L-type Ca
2+
channels in transmitter release from
mammalian inner hair cells. II. Single-neuron activity. J Neurophysiol 87(6):2734–2740.
54. Kros CJ, Ruppersberg JP, Rüsch A (1998) Expression of a potassium current in inner
hair cells during development of hearing in mice. Nature 394(6690):281–284.
55. Johnson SL, Beurg M, Marcotti W, Fettiplace R (2011) Prestin-driven cochlear ampli-
fication is not limited by the outer hair cell membrane time constant. Neuron 70(6):
1143–1154.
56. Goutman JD, Glowatzki E (2011) Short-term facilitation modulates size and timing of
the synaptic response at the inner hair cell ribbon synapse. J Neurosci 31(22):
7974–7981.
57. Cho S, Li G-L, von Gersdorff H (2011) Recovery from short-term depression and fa-
cilitation is ultrafast and Ca
2+
dependent at auditory hair cell synapses. J Neurosci
31(15):5682–5692.
58. Furman AC , Kujawa SG, Liberman MC (2013) Noise-induced cochlear neuropathy is
selective for fibers with low spontaneous rates. J Neurophysiol 110(3):577–586.
59. Hoppa MB, Lana B, Margas W, Dolphin AC, Ryan TA (2012) α2δexpression sets pre-
synaptic calcium channel abundance and release probability. Nature 486(7401):
122–125.
60. Han Y, Kaeser PS, Südhof TC, Schneggenburger R (2011) RIM determines Ca
2+
channel
density and vesicle docking at the presynaptic active zone. Neuron 69(2):304–316.
61. Kiyonaka S, et al. (2007) RIM1 confers sustained activity and neurotransmitter vesicle
anchoring to presynaptic Ca
2+
channels. Nat Neurosci 10(6):691–701.
62. Zamponi GW, Striessnig J, Koschak A, Dolphin AC (2015) The physiology, pathology,
and pharmacology of voltage-gated calcium channels and their future therapeutic
potential. Pharmacol Rev 67(4):821–870.
63. Sheets L, Trapani JG, Mo W, Obholzer N, Nicolson T (2011) Ribeye is required for
presynaptic Ca(V)1.3a channel localization and afferent innervation of sensory hair
cells. Development 138(7):1309–1319.
64. Sheets L, Kindt KS, Nicolson T (2012) Presynaptic CaV1.3 channels regulate synaptic
ribbon size and are required for synaptic maintenance in sensory hair cells. J Neurosci
32(48):17273–17286.
65. Brandt A, Khimich D, Moser T (2005) Few CaV1.3 channels regulate the exocytosis of a
synaptic vesicle at the hair cell ribbon synapse. J Neurosci 25(50):11577–11585.
66. Pangr
siˇc T, et al. (2015) EF-hand protein Ca
2+
buffers regulate Ca
2+
influx and exo-
cytosis in sensory hair cells. Proc Natl Acad Sci USA 112(9):E1028–E1037.
67. Ezan J, Montcouquiol M (2013) Revisiting planar cell polarity in the inner ear. Semin
Cell Dev Biol 24(5):499–506.
68. Sienknecht UJ, Köppl C, Fritzsch B (2014) Evolution and development of hair cell
polarity and efferent function in the inner ear. Brain Behav Evol 83(2):150–161.
69. Moser T, Beutner D (2000) Kinetics of exocytosis and endocytosis at the cochlear inner
hair cell afferent synapse of the mouse. Proc Natl Acad Sci USA 97(2):883–888.
70. Jing Z, et al. (2013) Disruption of the presynaptic cytomatrix protein bassoon de-
grades ribbon anchorage, multiquantal release, and sound encoding at the hair cell
afferent synapse. J Neurosci 33(10):4456–4467.
71. Staudt T, Lang MC, Medda R, Engelhardt J, Hell SW (2007) 2,2′-thiodiethanol: A new
water soluble mounting medium for high resolution optical microscopy. Microsc
Res Tech 70(1):1–9.
10 of 10
|
www.pnas.org/cgi/doi/10.1073/pnas.1605737113 Ohn et al.

Supporting Information
Ohn et al. 10.1073/pnas.1605737113
SI Methods
Animals. C57BL/6 mice of either sex were used for hair cell
physiology and morphology at p14–20 (Figs. 1–4). Ca
V
1.3
HA/HA
mice (36) and littermate controls were provided by Jörg Striessnig,
University of Innsbruck, Innsbruck, Austria, and used at p14–18
(patch clamp) and 6–8 wk (in vivo). Dfcr mice (46), Jax (Balb/C
background), and littermates were used at p14–17. Gipc3 mutants
(48) were compared with age-matched C57BL/6 at p14–27 (19).
Patch Clamp and Confocal Ca
2+
Imaging. Apical cochleae (first 2/3
turns) were freshly dissected. Inner border cells and phalangeal
cells were gently removed by a suction pipette, exposing an IHC
basolateral membrane. The patch pipette solution contained (in
millimolar) 123 Cs-glutamate, 1 MgCl
2
, 1 CaCl
2
, 10 EGTA, 13
tetraethylammonium (TEA)-Cl, 20 Hepes, 2 MgATP, and 0.3
NaGTP (pH 7.3). For live imaging, it contained the Ca
2+
indicator
Fluo-8FF (0.8 mM; AAT Bioquest) and the TAMRA-conjugated
CtBP2/RIBEYE-binding dimer peptide (20 μM; Biosynthan). For
Fig. S5, we used 111 mM L-glutamate, 4 mM MgATP, 1 mM
glutathione, 0.8 mM Fluo-4FF penta-K
+
salt (Life Technologies),
and 10 μM TAMRA-conjugated CtBP2/RIBEYE-binding dimer
peptide. IHCs were continuously superfused with an extracellular
solution containing (in millimolar) 102.2 NaCl, 2.8 KCl, 1 MgCl
2
,
5 CaCl
2
, 35 TEA-Cl, 10 Hepes, and 2 g/L glucose (pH 7.2). An
EPC-10 amplifier and Patchmaster software (HEKA Elektronik)
were used and Ca
2+
influx was evoked by step depolarizations
from −87 mV to −7 mV or ramp depolarizations from −87 mV to
63 mV. All voltages were corrected for liquid junction potential
(17 mV) and voltage drops across series resistance (R
s
). IHCs
were held at −87 mV. Recordings were discarded when holding
current exceeded 50 pA, R
s
exceeded 10 MΩwithin 4 min after
break-in, or rundown of Ca
2+
current exceeded 25%. Ca
2+
imaging
was performed with a spinning disk confocal scanner (CSU22;
Yokogawa) mounted on an upright microscope (Axio Examiner;
Zeiss) with 63×, 1.0 N.A. objective (W Plan-Apochromat; Zeiss).
Images were acquired by a back-illuminated CCD camera with
80 ×80 pixels within a 19.2- ×19.2-mm CCD chip (NeuroCCD;
Redshirt Imaging) or a scientific complementary metal–oxide–
semiconductor camera (Neo; Andor). Ca
2+
indicators and TAMRA-
conjugated peptide were excited by diode-pumped solid-state la-
sers with 491-nm and 561-nm wavelength, respectively (Cobolt
AB). The spinning disk was set to 2,000 rpm to synchronize with
the 10-ms acquisition time of the camera. Using a piezo positioner
for the objective (Piezosystem), focal planes were acquired in an
order randomized in each cell to avoid systematic effects of flu-
orophore bleaching. C
m
measurements were performed as pre-
viously described (69) and exocytic C
m
changes calculate from the
difference of C
m
before and after the depolarization (400 ms,
skipping the initial 40 ms).
Immunohistochemistry and Confocal Imaging. The organ of Corti
apical 2/3 turn was prepared for “whole-mount”imaging as de-
scribed in ref. 70, but the protocol was modified after the sec-
ondary antibody rinse. The following primary antibodies were
used: goat anti-CtBP2 (1:150; Santa Cruz Biotechnology), mouse
anti-GluA2 (1:75; Millipore), rabbit anti-Ca
V
1.3 (1:75; Alomone
Labs), mouse anti-bassoon SAP7F407 (Abcam), rabbit anti-
BK
Ca
(1:200; Alomone Labs), and mouse anti-otoferlin (1:300;
Abcam). Secondary antibodies used were Alexa Fluor 488-con-
jugated antimouse, Alexa Fluor 596-conjugated anti-goat, and
Alexa Fluor 647-conjugated anti-rabbit IgG antibodies (1:200;
Invitrogen).
For improved structural preservation, 50-μm-thick plastic film
was taped to the glass surface on either side of the sample to act
as spacers, preventing overflattening of the whole-mount. To
minimize differences in attenuation at different focal depths,
tissue was embedded in 2,2′-thiodiethanol (71). To control for
artifactual differences between synapses due to bleaching and
attenuation of the optical signal resulting from order of acqui-
sition and synapse position, respectively, we mounted the tissue
between two coverslips. This allowed for imaging the same field
of IHCs with either the apical surface or the basilar membrane
surface oriented toward the objective of the Leica SP5 confocal
microscope (Fig. S3). Data pooled from multiple volumes were
acquired with identical settings. For display in Fig. 1Cthe pro-
jections were made from optical sections deconvolved with a
model of the optical point spread function using the Richardson–
Lucy algorithm in ImageJ software, with default settings and the
pixel dimensions, emission wavelength, and numerical aperture
of the experiments.
Extracellular Recordings from Auditory Nerve Fibers. Single unit
recordings from mouse SGNs were performed as described (4,
70). In brief, mice were anesthetized by i.p. injection of urethane
(1.32 μg/g), xylazine (5 μg/g), and buprenorphine (0.1 μg/g), and
parts of the occipital bone and cerebellum were removed to
expose the anteroventral cochlear nucleus (AVCN). Sound-
responsive single neurons were identified based on spontaneous
and noise-induced action potential firing and a basic character-
ization was performed by measuring their spontaneous rate,
tuning curve, and poststimulus time histograms. SGNs were
discriminated from primary neurons of the AVCN by their dis-
charge pattern, first spike latency, and stereotaxic position. Re-
cordings and offline analysis using waveform-based spike
detection were performed using custom-written MATLAB soft-
ware (The MathWorks, Inc.) and TDT system III hardware
(Tucker-Davis Technologies) and an ELC-03XS amplifier (NPI
Electronics).
Data Analysis. Live-imaging and IHC patch-clamp data were ana-
lyzed by custom programs in Igor Pro-6.2 (Wavemetrics). The
change of Ca
2+
-indicator fluorescence, our proxy of AZ Ca
2+
in-
flux, was estimated as ΔF/F
0
, where F
0
is the fluorescence intensity
at −87 mV and ΔF is the difference when depolarized to −7mV
(steps) or variable potentials (voltage ramps). Ca
2+
-indicator in-
tensity was calculated by the average of nine pixels centered on the
pixel showing the greatest fluorescence increase. Maximal ΔF/F
0
was the average of last 40 ms during step depolarizations and
five points at the peak ΔF/F
0
during the voltage ramp. To re-
fine the noisy FV traces, we fit the raw traces to the equation
FðVÞ=F0+½GmaxðVr−VÞ=ð1+e½ðVh−VÞ=kÞ. The fit was used
to calculate the fractional activation curve, dividing it by the ex-
trapolated maximal ΔF/ΔV estimated by linear fitting of the decay
of fluorescence with voltage. Then, this trace was fitted by a
Boltzman function to estimate the V
h
and slope factor of the Ca
2+
-
indicator fluorescence at a given AZ. Synaptic ribbon fluores-
cence was estimated as the ratio of TAMRA fluorescence to that
of the nearby IHC cytoplasm typically eight or nine pixels away
(F
ribbon
/F
nearby
), measuring the pixel with strongest intensity. IHC
cytoplasmic volumes were displayed with the plugin “3D viewer”
in Fiji software.
To analyze the intensity and position of synapses in confocal
images of fixed organ of Corti whole mounts, a customized al-
gorithm was developed as a program in MATLAB software. The
Ohn et al. www.pnas.org/cgi/content/short/1605737113 1of9

locations of synapses were defined as the centers of mass of
fluorescent spots after thresholding by a subjective intensity
criterion for each of the three channels. The average voxel in-
tensity in the volume, excluding voxels that exceeded the
threshold value, was subtracted as background. Gaussian func-
tions were fitted in three dimensions to determine the center of
mass of each cluster. Immunofluorescence intensities were
measured on each channel as the sum of the voxel values within a
defined region of interest (±500, 500, and 835 nm or ±10, 10,
and 2 pixels in X, Y, and Z) with origin at the center of mass of
each GluA2-positive cluster. After marking the center of each
IHC nucleus, observing the relative orientation of the pre-
synaptic ribbon and the postsynaptic receptor array allowed us to
assign each synapse to an IHC. The Cartesian coordinates of
synapses were transformed to cell-centric cylindrical coordinates
to define the cellular axes and to adjust differences in cellular
orientation relative to the XYZ axes of the microscope. Keeping
one end of the line centered on the nucleus, the angle of the axis
was adjusted once in the XY plane and once in the YZ plane to
compensate for the pitch and yaw of each hair cell (Fig. S3).
After adjustment, points along the central axis were defined as
zero on the tonotopic and modiolar–pillar axes. Intensities of
synaptic puncta as a function of position could be analyzed for
multiple cells by overlaying their central axes with alignment to
the center of each nucleus.
Jitter is presented as SD or SEM, as noted. Two-tailed ttests
or, if data were not normally distributed and/or variance was
unequal between samples, the Wilcoxon rank sum test were used
for statistical comparisons between two samples. Box plots show
10, 25, 50, 75, and 90% percentiles; *P<0.05, **P<0.01, and
***P<0.001.
Ohn et al. www.pnas.org/cgi/content/short/1605737113 2of9

A
B
1.5
1.0
0.5
0.0
Fluorescence intensity (x 10
6
a.u.)
# of cells
8
6
4
2
0
-0.5 0.5 1.00
r
rmean: 0.49
0
3
2
1
4
5
0 0.2 0.80.4 0.6 1.0
# of cells
r
individual cells
rmean: 0.61
C
2.5
2.0
1.5
1.0
0.5
0.0
mean of remainders
winners
E
G
D
F
all
cells
F
ribbon
/F
nearby
5
4
3
2
1
6
all
AZs
modiolar
pillar
1.5
1.0
0.5
0.0
2.0
2.5
0.83 0.17 0.55 0.33 0.43 0.42 0.09 0.44 0.58 0.2 0.36 0.27 0.17 0.5
Fpillar
Cav1.3 GluA2 CtBP2
individual cells
-40
-30
-20
-10
V
0.5
(mV)
individual cells
20
15
10
5
0
2520151050
r= 0.625, n= 77, p< 0.001
CaV1.3 fluorescence intensity
(x103 a.u.)
Basoon fluorescence intensity (x103 a.u.)
∆F-7mV / F0
∆F-7mV / F0
Fig. S1. Quantification of functional and molecular markers of the strength of IHC synapses. (A) Box plots (10, 25, 50, 75, and 90% values) show the statistics of
maximal AZ Ca
2+
influx (ΔF
−7mV
/F
0
, green) intensity and of RIBEYE-peptide fluorescence (red) of AZs in individual cells (left), of mean ΔF
−7mV
/F
0
and mean RIBEYE-
peptide fluorescence of all cells (right), and of ΔF
−7mV
/F
0
and of RIBEYE-peptide fluorescence all AZs (rightmost) as recorded in the live-imaging experiments.
Individual data points are superimposed on the box plots of the individual IHCs, with color coding for modiolar (blue) or pillar (amber) position of the AZ. The
number underneath each IHC represents the fraction of pillar AZs. (B) Statistics of the immunofluorescence for all labeled synapses of individual IHCs in the organ
of Corti. Box plots display the distributions of the immunofluorescence intensity of presynaptic Ca
v
1.3 clusters (green) and RIBEYE/CtBP2-labeled ribbons (red) as
well as GluA2-labeled glutamate receptor clusters (blue). Individual data points are superimposed on the box plots of the individual IHCs. (C) Scatter plot of the
immunofluorescence intensity of presynaptic Ca
v
1.3 clusters and corresponding bassoon immunofluorescence. (D) Correlation of maximal AZ Ca
2+
influx and
RIBEYE-peptide fluorescence within individual IHCs. Histogram displays the distribution of the correlation coefficients of individual IHCs in the live-imaging ex-
periment: The average correlation coefficient among cells in the live-imaging experiment was 0.49. (E) Correlation of Ca
V
1.3 and RIBEYE/CtBP2 immunofluores-
cence within individual IHCs. Histogram displays the distribution of the correlation coefficients of individual IHCs in the immunolabeled organ of Corti. The
averaged correlation coefficient among cells was 0.61 in immunohistochemistry. (F)AZCa
2+
influx between the strongest AZ (winners) and the mean of the
remaining AZs in individual IHCs. Gray: mean ±SEM. (G) Box plots and individual values of V
0.5
of the AZs within individual IHCs.
Ohn et al. www.pnas.org/cgi/content/short/1605737113 3of9

Fig. S2. The method and definition for the reconstruction of hair cell in cylindrical-like coordination system. (A) Three-dimensional images of an IHC from
different viewpoints. (Left) The viewpoint from modiolar (tonotopically) basal side. (Right) The viewpoint from modiolar side. The plane containing yellow dots
depicts the plane of symmetry of the IHC and the green arrow represents the normal vector of the plane of symmetry (V
sym
). (B) The plane of symmetry: The
white dot line is drawn along the pillar edge of the IHC. The original 3D stack image was sectioned along this line. The yellow disk is the center of the new
coordination system, which is the center of mass of the section containing the largest area of cell fluorescence. The central axis V
z
connects the origin to the
center of mass of the lowest section. (C,Right) The distance of a synapse (the red ball) to the center of mass of the plane containing this synapse. H, distance in
height between the plane containing origin to the plane containing the synapse; Θ, angle of R to V
sym
. We identified the plane of symmetry by maximizing
mirror symmetry along the orthogonal tonotopic axis (A). Then we sectioned the IHC in a straight line (white dashed line, B) which fits to the pillar edge on the
plane of symmetry. The center of mass for each section was estimated and we connected the center of the bottom-most one to the center of the largest section
to form the central axis (V
z
,Band C). The cylindrical coordinate of each synapse/microdomain was transformed based on the central axis V
z
, vector of plane of
symmetry V
sym
, and their cross-product V
mp
.
Ohn et al. www.pnas.org/cgi/content/short/1605737113 4of9

Fig. S3. Analysis of AZ properties by synaptic position on the IHC. (Aand B) Multicellular alignment via 3D transformation from Cartesian coordinates of the
confocal stack to a cell-centric cylindrical coordinate system, using custom MATLAB routines. For each cell, the nuclear center (white asterisks) was registered in
three dimensions using the nuclear immunoreactivity of the CtBP2 antibody. The tonotopic dimension for each image was approximated as the confocal
microscope’s X dimension (white line in A) by adjusting the rotation of the region of interest. Synapses were sorted by cell, based on proximity to nuclei and
orientation of each juxtaposed pre- and postsynaptic pair of CtBP2- and GluA2-labeled puncta. Aand Bshow the CtBP2 channel only. Presynaptic ribbons are
marked with dots colored by cell. The center of the nucleus defined one point on a line chosen as the cell’s central longitudinal axis. The cell on the left of A(XY
plane) has a blue circle around the nucleus. IHCs were visualized individually in the ZY plane by projecting a volume limited along the X axis, within the width
of the cell nucleus (yellow vertical lines in A). Synapses belonging to this cell, but positioned outside of the subvolume displayed in the ZY plane, were included
as colored markers (red dots without accompanying fluorescence in B). The center of each nucleus was defined as zero on the central longitudinal axis, which
extended positively toward the basal pole of the IHC from the direction of the cuticular plate toward the habenula perforata. The second point definingthis
axis was set by adjusting the axis line in the XY plane (green line in A) and in the ZY plane (green line in B). In the YZ plane, visualizing only the synapses
belonging to the cell being analyzed, the axis was chosen based on a user-defined balance of synapses on either side of a line through the nuclear center and
through the cluster of synapses at the basal pole. The IHC central longitudinal axis (green) is perpendicular to the defined modiolar–pillar axis in the ZY plane
(orange line in B). Using the cell-centric cylindrical coordinates, data from all six cells in the image were overlaid on a projection along the longitudinal axis
(polar plot, B, lower right). The longitudinal axis is orthogonal to the plane defined by the perpendicular tonotopic axis (0–180°) and modiolar–pillar axis.
(Cand D) Synapses with greater Ca
V
1.3 and CtBP2 fluorescence tended to be positioned toward the modiolar side of the modiolar–pillar axis. In C, the scatter
plot is for the six cells shown in A, imaged with the modiolar side toward the objective in a top-down sequence, imaging the modiolar-side synapses first. In D,
the scatter plot is for the same six cells imaged with opposite orientation of the specimen, to control for potential artifacts due to optical attenuation and
bleaching. The modiolar-side synapses closer to the coverslip and imaged first in Cwere further from the coverslip and imaged last in D. A similar relationship
was observed in each case, showing that modiolar synapses tended to have more immunofluorescence for Ca
V
1.3 and CtBP2.
Ohn et al. www.pnas.org/cgi/content/short/1605737113 5of9

apical
basal
pillar
modiolar
012
∆F-7mV / F0
6
i
apical
basal
pillar
modiolar
V0.5 (mV)
-40 -30 -20
6
0.4
0.8
1.2
1.6
2.0
0
∆F-7mV / F0
pillar
(n= 42)
modiolar
(n= 73)
*
pillar
(n= 31)
modiolar
(n= 33)
ii
*
-40
-30
-20
V0.5 (mV)
A
-40-35-30-25-20-15
-12
-10
-8
-6
-4
-2
0
2
H (μm)
-14
V0.5 (mV)
iii
6
ii
12345
Fribbon / Fnearby
-12
-10
-8
-6
-4
-2
0
2
H (μm)
24 6810
12
k (mV)
iv
-12
-10
-8
-6
-4
-2
0
2
-14
H (μm)
3
i
-12
-10
-8
-6
-4
-2
0
2
120
H (μm)
∆F-7mV / F0
C
20
10
0
3210
Apical (n= 92)
Basal (n= 110)
# of active zones
∆F-7mV / F0
15
10
5
0
840
# of active zones
Fribbon / Fnearby
Apical (n= 67)
Basal (n= 80)
40
30
20
10
0
-40 -35 -30 -25 -20 -15
V0.5 (mV)
# of active zones
Apical (n= 97)
Basal (n= 103)
30
20
10
0
26
10 14
k (mV)
# of active zones
Apical (n= 97)
Basal (n= 103)
iii
iii iv
B
iii iv
μm
μm
Fig. S4. AZ properties as a function of position within the IHC. (A) Spatial distribution of maximal AZ Ca
2+
influx (ΔF
−7mV
/F
0
) without the basal cap. (i) Polar
chart displays ΔF
−7mV
/F
0
as a function in position, when AZ were projected along the central axis. In contrast to Fig. 2A, we removed AZs with a radius smaller
than 3 μm, which are mostly located on the basal end of IHC (n=115) and pose a challenge to assign to one of the sectors. (ii) Box plots describe the dis-
tribution of ΔF
−7mV
/F
0
of modiolar and pillar halves. The AZs on the modiolar half had significantly stronger ΔF
−7mV
/F
0
than those on the pillar half. (iii) Polar
chart displays V
0.5
as a function in position, when AZ were projected along the central axis. In contrast to Fig. 4A, we removed AZs with a radius smaller than
3μm, which are mostly located on the basal end of IHC and pose a challenge to assign to one of the sectors. (iv) Box plots describe the distribution of V
0.5
on
modiolar and pillar halves. The AZs of the pillar half had significantly more hyperpolarized V
0.5
than those of the modiolar side. (B) No obvious gradients of
synaptic properties along the tonotopic axis. (i) The distributions of maximal AZ Ca
2+
influx (ΔF
−7mV
/F
0
), (ii) intensity of RIBEYE-peptide fluorescence, (iii)V
0.5
,
and (iv) k in both tonotopical apical (low-frequency) and basal (higher-frequency) sides of IHCs. There are no significant statistical differences observed between
these two sides in all four measured parameters (Wilcoxon rank sum test). (C) No obvious gradients of synaptic properties along the longitudinal axis. This
figure shows the spatial arrangements of (i) maximal AZ Ca
2+
influx (ΔF
−7mV
/F
0
), (ii) the intensity of RIBEYE-peptide fluorescence, (iii )V
0.5
, and (iv) k along the
longitudinal axis of the IHC. The 0 layer of the H position corresponds to the plane with the largest IHC cross-section on the axis V
z
(Fig. S2). Positive H values
indicate an AZ position closer to cuticular plate, and negative closer to the basal pole. Red bars in this figure represent the mean and SD of the corresponding
quantity in their specified longitudinal positions.
Ohn et al. www.pnas.org/cgi/content/short/1605737113 6of9

A
BK Parvalbumin
dfcr/dfcr
FE
dfcr/dfcr
CaV1.3 RIBEYE
+/+
10
8
6
4
2
0
∆F
max
/ F
0
(a.u.)
WT dfcr
D
-200
-150
-100
-50
50
604020-20-40-60
Vm (mV)
ICa (pA)
WT
dfcr
-35
-30
-25
-20
-15
V0.5 (mV)
*
WT dfcr
*
C
Pactivation, GCa
Vm (mV)
1
0.8
0.6
0.4
0.2
0
-60 -40 -20 0
WT
dfcr
-40
-30
-20
-10
0
V0.5 (mV)
WT dfcr
B
Pactivation, FCa
1
0.8
0.6
0.4
0.2
0
-80 -60 -40 -20 020 40
Vm (mV)
WT
dfcr
WT dfcr
7
6
5
4
3
2
1
*
F
ribbon
/ F
nearby
(a.u.)
*
F
ribbon
/ F
nearby
(a.u.)
50
40
30
20
10
0
# of active zones
65432
WT
dfcr
1
Fig. S5. Genetic disruption of harmonin function leads to decreased ribbon size and hyperpolarizing shift of the activation of Ca
2+
influx but leaves de-
velopmental confinement of Ca
2+
and expression of large-conductance Ca
2+
-activated K
+
channels unaltered. (A) Distribution of the intensities of AZ TAMRA-
peptide fluorescence in dfcr (red, 105 AZs of 15 IHCs, c.v. =0.26) and in WT (BALB/c, black, 107 AZs of 14 IHCs, c.v. =0.35) IHCs (Left) and box plot (Right)
depicting a significant decrease in the fluorescence intensity of labeled ribbons in dfcr IHCs (Left,P<0.05, Wilcoxon rank sum test). (B) Fractional activation of
AZ Ca
2+
influx shows comparable heterogeneity of dfcr (red) and WT (black) IHCs, but a more hyperpolarized mean V
0.5
in dfcr IHCs (Right,P<0.05, ttest).
Experiments were performed as in Fig. 3. Thin lines present the fractional activation of individual AZs of dfcr (light red, 105 synapses) and WT (gray, 107
synapses) and thick red and black lines the mean fractional activation, respectively. (C) Fractional activation of the whole-cell Ca
2+
conductance (GCa). (Left)
The activation of single IHCs (thinner and lighter traces) from dfcr (red and light red, 14 IHCs) and WT (black and gray, 15 IHCs) and mean fractional activation
of all IHCs (thick and dark traces), respectively. In agreement with the single AZ data (B), the mean V
0.5
of the whole-cell Ca
2+
conductance was also shifted
toward more hyperpolarized potentials in the dfcr cells (Right,P<0.05, Wilcoxon rank sum test). (D) Current–voltage relationship for I
Ca
evoked by 20-ms
voltage steps in whole-cell patch-clamp recordings from dfcr (red, 14 IHCs) and WT (black, 15 IHCs) cells showing an increased Ca
2+
current peak in the dfcr mice
(Left,mean±SEM, P<0.05, Wilcoxon rank sum test). However, the maximal AZ Ca
2+
influx dfcr (red, 105 AZs, c.v. =0.61) was not significantly different from
that of WT (black, 107 AZs, c.v. =0.83), suggesting an increased extrasynaptic Ca
2+
influx (Right). (E) Maximum projection of confocal sections of organs of
Corti immunolabeled for Ca
V
1.3 (green) and RIBEYE/CtBP2 (red). Most Ca
V
1.3 immunofluorescence was confined to presynaptic AZs (labeled by the RIBEYE
immunofluorescence) in both dfcr (Right) and WT (Left) IHCs. Immaturity would have expected to show more ribbons and many more Ca
V
1.3 immunofluo-
rescence spots (29). (F) Maximum projection of confocal sections of organs of Corti immunolabeled for large-conductance Ca
2+
-activated K
+
channels (BK,
green) and the EF-hand Ca
2+
binding protein parvalbumin (red). Spot-like BK immunofluorescence, most likely reflecting BK channel clusters was found at the
‟neck”of the IHC in both dfcr (Right) and WT (Left) IHCs. Immaturity would have expected to cause a lack of BK immunofluorescence. (Scale bar: 10 μm.)
Ohn et al. www.pnas.org/cgi/content/short/1605737113 7of9

AB
Gipc3 p14-18, p21-25
WT p14-15, p20-21
1.0
0.8
0.6
0.4
0.2
0.0
)
%
(y
rev
oc
e
r
t
n
ec
r
eP
4 5 6 7 8 9
10
2 3 4 5 6 7 8 9
100
2 3 4 5
Interval (ms)
100 ms
6
4
2
0
ABR amplitude (μV)
12010080604020
Click dB SPL
Gipc3 p14 (n= 11)
Gipc3 p21 (n= 10)
WT p20-27 (n= 9)
WT p13 (n= 4)
120
100
80
60
40
20
0
4 6 8 12 16 24 32 Click
Gipc3 p14 (n= 9)
Gipc3 p21 (n= 10)
WT p13 (n= 3-4)
WT p20-24 (n= 7)
Frequency (kHz)
Gipc3C57Bl/6
**** **
0
50
100
12
10
8
6
4
2
0
510 20 40
CF (kHz)
Gipc3 p14-p18
Gipc3 p21-p25
WT p14-p15
WT p20-p21
0
510 20 40
Gipc3 p14-p18
Gipc3 p21-p25
WT p1 4-p15
WT p2 0 -p 21
CDE
p14-15
p20-21
p14-18
p21-25
F
Threshold (dB SPL)
Threshold (dB SPL)
Threshold (db SPL)
Q10 dB
120
100
80
60
40
20
CF (kHz)
Fig. S6. Auditory population responses, single unit tuning curves, and assessing recovery from adaptation using a forward masking paradigm in Gipc3 mutant
mice. (A) ABR thresholds for tone burst stimuli and click were elevated in both p14 and p21 Gipc3 mutant mice compared with WT (C57BL/6). (B) Average ABR
wave I amplitude as a function of click sound pressure level. At p14, higher click levels were needed to evoke a reproducible ABR wave I, but the amplitudeof
this wave can grow very steeply to a value that was comparable to WT controls. However, at p21, the wave I amplitude of Gipc3 mutant mice did not reach the
level at p14 and of control. (C) Representative tuning curves of Gipc3 mutant SGNs. (D)Gipc3 mutant SGNs had elevated thresholds at the characteristic
frequency in comparison with WT SGNs at comparable ages (66.6 ±3.1, n =13 for Gipc3 mutant p14–18 vs. 38.5 ±3.9, n=12 for WT p14–15, P<0.00001; 77.8 ±
1.8, n=12 for Gipc3 mutant p21–25 vs. 5.4 ±3.7, n=12 for WT p20–21, P<0.0001). (E) Sharpness of tuning was significantly broader in Gipc3 mutant SGNs
(2.7 ±0.5, n=12 for Gipc3 mutant p14–18 vs. 4.3 ±0.4, n=12 for WT p14–15, P<0.05; 2.1 ±0.5, n=7forGipc3 mutant p21–25 vs. 7.1 ±0.6, n=12 for WT p20–21,
P<0.0001). (F) Assessing recovery from adaptation using forward masking paradigm. (Left) Representative responses of a Gipc3 mutant SGN to the exper-
imental paradigm that consisted of a 100-ms-long masker stimulus and a 15-ms probe tone (both tone bursts at the characteristic frequency, 30 dB above
threshold). (Right) The time course of recovery was derived by normalizing the response to 15-ms probe stimulus to the last 15 ms (0% recovery) and first 15 ms
of masker response (100% recovery). Gipc3 mutant SGNs (n=9) showed a recovery time course similar to that of WT SGNs (n=8).
Ohn et al. www.pnas.org/cgi/content/short/1605737113 8of9

B
D
WT
Gipc3
Otoferlin BK
WT Gipc3
0
-70
-50
-30
-10
n.s.
RMP (mV)
20
15
10
0
5
Current (nA)
Time (ms)
50 150 50 150
Gipc3
WT
n.s.
2
4
6
8
10
12
0
WT Gipc3
Synapses per IHC
C
A
Fig. S7. Increased whole-cell Ca
2+
current but largely maintained IHC K
+
currents and synapses in IHCs of Gipc3 mutant mice. (A) The resting membrane
potential as recorded in the current-clamp configuration was normal in Gipc3 mutant IHCs: 71.7 ±0.7 mV, n=12 for Gipc3 mutant IHCs vs. −72.8 ±0.5 mV;
n=19, for WT IHCs (P=0.21, both p14–16). (B) Maximum projection of confocal section of immunolabeled organs of Corti: Gipc3 mutant IHCs (visualized by
otoferlin immunofluorescence) show clusters of large-conductance Ca
2+
-activated K
+
channels (magenta) at their “neck”region. (C) Representative potassium
currents in Gipc3 mutant and WT IHCs: no obvious reduction due to Gipc3 disruption. (D) Normal number of ribbon synapses in Gipc3 mutant IHCs of 2-wk-old
mice (12.09 ±0.9, n=28 for WT, 12.6 ±0.5, n=44 for Gipc3,P=0.48). Analysis was performed on confocal stacks of organs of Corti following immunolabeling
for RIBEYE/CtBP2 and postsynaptic GluA2/3.
Ohn et al. www.pnas.org/cgi/content/short/1605737113 9of9