Efferent neurons control hearing sensitivity and protect hearing from noise through the regulation of gap junctions between cochlear supporting cells

Abstract

It is critical for hearing that the descending cochlear efferent system provides a negative feedback to hair cells to regulate hearing sensitivity and protect hearing from noise. The medial olivocochlear (MOC) efferent nerves project to outer hair cells (OHCs) to regulate OHC electromotility, which is an active cochlear amplifier and can increase hearing sensitivity. Here, we report that the MOC efferent nerves also could innervate supporting cells (SCs) in the vicinity of OHCs to regulate hearing sensitivity. MOC nerve fibers are cholinergic, and acetylcholine (ACh) is a primary neurotransmitter. Immunofluorescent staining showed that MOC nerve endings, presynaptic vesicular acetylcholine transporters (VAChTs), and postsynaptic ACh receptors were visible at SCs and in the SC area. Application of ACh in SCs could evoke a typical inward current and reduce gap junctions (GJs) between them, which consequently enhanced the direct effect of ACh on OHCs to shift but not eliminate OHC electromotility. This indirect, GJ-mediated inhibition had a long-lasting influence. In vivo experiments further demonstrated that deficiency of this GJ-mediated efferent pathway decreased the regulation of active cochlear amplification and compromised the protection against noise. In particular, distortion product otoacoustic emission (DPOAE) showed a delayed reduction after noise exposure. Our findings reveal a new pathway for the MOC efferent system via innervating SCs to control active cochlear amplification and hearing sensitivity. These data also suggest that this SC GJ-mediated efferent pathway may play a critical role in long-term efferent inhibition and is required for protection of hearing from noise trauma.

NEW & NOTEWORTHY The cochlear efferent system provides a negative feedback to control hair cell activity and hearing sensitivity and plays a critical role in noise protection. We reveal a new efferent control pathway in which medial olivocochlear efferent fibers have innervations with cochlear supporting cells to control their gap junctions, therefore regulating outer hair cell electromotility and hearing sensitivity. This supporting cell gap junction-mediated efferent control pathway is required for the protection of hearing from noise.

INTRODUCTION

The cochlea is the auditory sensory organ in mammals. In addition to ascending afferent auditory nerves (ANs) projecting to the brain, the cochlea also receives descending efferent nerve fibers to hair cells, forming a negative feedback loop to control hair cell activity (1). The cochlear efferent nerves are composed of the medial olivocochlear (MOC) nerve fibers, which cross the cochlear tunnel projecting to outer hair cells (OHCs), and the lateral olivocochlear (LOC) nerve fibers, which project to inner hair cells (IHCs) and form synapses with the dendrites of type I afferent ANs under IHCs (1–3). MOC nerve fibers are cholinergic and acetylcholine (ACh) is a primary neurotransmitter, whereas LOC nerves have cholinergic and dopaminergic fibers, releasing ACh, dopamine, and other neurotransmitters (2, 4, 5). It is well known that the cochlear efferent system plays an important role in many aspects of hearing, such as the regulation of hearing sensitivity, enhancement of the discrimination of sounds in background noise, and the protection of hearing from acoustic trauma (1, 2, 6, 7). However, the detailed mechanisms for how the efferent system controls hearing sensitivity and the protection from noise still remain largely unclear.

The cochlea contains sensory hair cells and nonsensory supporting cells (SCs). SCs in the cochlea provide supporting function to hair cell activity. For example, Deiters cells (DCs) and pillar cells (PCs) in the vicinity of OHCs act as a scaffold to support OHCs standing on the basilar membrane, allowing OHC motility amplifying sound stimulation-induced basilar membrane vibration to increase hearing sensitivity and frequency selectivity (8–10). This prestin-based active cochlear amplification or mechanics (11) is required for mammalian hearing. Deficiency of this active cochlear amplification can induce hearing loss (12). In anatomy, cochlear SCs are extensively coupled by gap junctions (GJs) (13–15). Connexin (Cx)26 and Cx30 are predominant GJ isoforms in the cochlea (13, 15, 16). However, there is neither connexin expression in hair cells nor GJs between hair cells or between hair cells and SCs (9, 14, 15, 17). Cx26 mutations can cause hearing loss, responsible for >50% of nonsyndromic hearing loss (18–20). Our previous studies also demonstrated that SCs and GJs between them can modulate OHC electromotility and participate in active cochlear amplification (17, 21, 22). These data suggest that SCs in the cochlea have important roles in hearing.

SC function in hearing, however, still remains largely undetermined. For example, it was reported that the cochlear SCs have nerve innervations (23–28). However, the source and function of these nerve innervations in the SCs remain unclear or are under debate (29, 30). In this study, we found that MOC efferent nerves had innervation with SCs to inhibit GJs between them and enhanced the direct effect of ACh on OHC electromotility. Diminution of this efferent SC GJ-mediated control could compromise the regulation of active cochlear amplification and the protection of hearing against noise.

MATERIALS AND METHODS

Animal Selection and Preparation

In this study, adult mice (1–6 mo old) and adult guinea pigs (200–450 g) of both sexes were used. Most electrophysiological recordings were performed in guinea pigs, since the whole cochlear sensory epithelium (including the basal turn) could be easily isolated to obtain large numbers of cochlear SCs and OHCs, in particular OHC-DC pairs. Most histological and in vivo examinations were performed in mice, since most of the transgenic animal models are mice. We also performed electrophysiological recording in mice for further verification.

Adult Hartley guinea pigs were purchased from Charles River Laboratories (United States) and used in experiments. For mouse experiments, adult CBA/CaJ mice (stock no. 000654, The Jackson Lab, United States; RRID:IMSR_JAX:000654) were used. Peripherin-eGFP transgenic mice (31), which were gifted by Dr. Ebenezer Yamoah at the University of Nevada, were also used in morphological experiments. For targeted deletion of Cx26 in outer supporting cells, a Cx26-Prox1 conditional knockout (cKO) mouse line, which was established in our previous studies (21, 22) by crossing Cx26^loxP/loxP transgenic mice (EM00245, European Mouse Mutant Archive; IMSR Cat no. EM_00245, RRID:IMSR_EM:00245) with the Prox1-CreER^T2 Cre line (stock no. 022075, Jackson Laboratory, United States), was used. As we previously reported (21, 22), tamoxifen (T5648, Sigma-Aldrich, St. Louis, MO) was administered to all litters at postnatal day 0 (P0) by intraperitoneal injection (0.5 mg/10 g × 3 days). Wild-type (WT) littermates were used as controls. All experimental procedures were approved by the University of Kentucky Animal Care & Use Committee.

Immunofluorescent Staining and Confocal Microscopy

Immunofluorescent staining was performed as described in our previous reports (15, 16). The mouse cochlea was freshly isolated, and the temporal bone was removed. The otic capsule was opened. The round window membrane was broken by a needle, and a small hole was also made at the apical tip of the cochlea. Then, the isolated cochlea was fixed with 4% paraformaldehyde for 30 min. After washout with PBS, the cochlea was decalcified by 10% EDTA for 5–8 h and the cochlear sensory epithelium was isolated after removal of the bone and stria vascularis. After washout with PBS, the isolated epithelia were incubated in a blocking solution (10% goat serum and 1% BSA in PBS) with 0.5% Triton X-100 and then incubated with polyclonal chicken anti-neurofilament (1:500, Cat. No. AB5539, Millipore Corp, California), monoclonal mouse anti-Tuj1 (1:500, Cat No. MMS-435P, BioLegend, California), monoclonal chicken anti-GFP (1:500, Cat. No. ab13970, Abcam, Massachusetts), monoclonal mouse anti-AChRα9 (1:100, Cat. No. sc-293282, Santa Cruz Biotech Inc, California), monoclonal rabbit anti-VAChT (1:500, Cat. No. ab279710, Abcam, Massachusetts), monoclonal mouse anti-Sox2 (1:200, Cat No. sc-365823, Santa Cruz Biotech Inc, California), polyclonal goat anti-prestin (1:50, Cat. No. sc-22694, Santa Cruz Biotech Inc, California), or monoclonal mouse anti-Cx26 (1: 400, Cat. No. 33-5800, Invitrogen), at 4°C overnight. After being washed with PBS three times, the epithelia were incubated with corresponding Alexa Fluor-conjugated second antibodies (1:500, Molecular Probes) at room temperature (23°C) for 1 h. At the last 15 min, 4′, 6-diamidino-2-phenylindole (DAPI, 0.1 mg/ml, D1306; Molecular Probes) was added to visualize the cell nuclei.

After washing out and mounting on the glass slide, the stained epithelia were observed under a Nikon A1R confocal microscope system with a Nikon ×60 or ×100 Plan Apo oil objective. Serial sections were scanned along the z-axis from the bottom to the apical surface of the epithelium with a 0.25-μm step. NIS Elements AR Analysis software (Nikon) was used for constructing three-dimensional (3-D) images from z-stack sections.

Cochlear Cell Isolation for Electrophysiological Recording

As we previously reported (16, 17, 32–34), guinea pigs or mice were decapitated and the temporal bone was removed. The otic capsule was dissected in normal extracellular solution (NES; in mM: 142 NaCl, 5.37 KCl, 1.47 MgCl₂, 2 CaCl₂, 10 HEPES; 300 mosM and pH 7.2). After removal of the bone and stria vascularis, the sensory epithelium (organ of Corti) was exposed and picked away along the cochlea with a sharpened needle. The isolated sensory epithelia were further dissociated with trypsin (0.5 mg/mL) for 2–3 min with shaking. Then, the dissociated cells were transferred to the recording chamber. The cochlear OHCs and SCs could be unequivocally identified under microscope by their morphological shapes (33, 34).

Patch-Clamp Recording

The isolated cells were continuously perfused with the NES (0.5 mL/min). The selected OHC or SCs were recorded in the whole cell configuration with an Axopatch 200B patch-clamp amplifier (Molecular Devices, California) (17, 33). The patch pipette was filled with an intracellular solution (in mM: 140 KCl, 5 EGTA, 2 MgCl₂, and 10 HEPES; pH 7.2, 300 mosM) with initial resistance of 2.5–3.5 MΩ in the bath solution. Data were collected by jClamp software (SciSoft, New Haven, CT). The signal was filtered by a four-pole low-pass Bessel filter with a cutoff frequency of 2 kHz and digitized with a Digidata 1322 A (Molecular Devices, California). The recording was stopped when the seal resistance was <500 MΩ in the SC recording. All recordings were performed at room temperature (23°C) and finished within 4–6 h after isolation to ensure that cells were in good condition.

Input capacitance (C_in) was continually recorded online at 1–3 Hz from the transient charge induced by small (−10 mV) test pulses with duration of 18 times the time constant at the holding potential (34, 35). The transient charge was calculated from the integration of capacitance current with time. Membrane potential (V_m) was corrected for pipette series resistance (R_s).

OHC electromotility-associated nonlinear capacitance (NLC) was measured with a two-sinusoidal wave voltage stimulus in jClamp (17, 32). This voltage stimulus was composed of a ramp command (−150 mV to +150 mV) summed with two sinusoidal commands (f₁ = 390.6 Hz, f₂ = 781.3 Hz, 25 mV peak to peak). The signal was filtered by a four-pole low-pass Bessel filter with a cutoff frequency of 10 kHz. The capacitance was calculated by admittance analysis of the current response. In some cases, the peak of NLC and the voltage corresponding to the peak capacitance (V_pk) were continuously recorded by a tracking technique (sampling rate: 4/s, tracking step: 0.25 mV) (17, 32, 36).

Data analysis was performed with jClamp or MATLAB (RRID:SCR_001622) (9, 17, 32). The voltage-dependent NLC was fitted to the first derivative of a two-state Boltzmann function:

$$C_{\mathrm{m}}=\text{NLC}+C_{\mathrm{lin}}=Q_{\text{max}}\frac{ze}{k\mathrm{T}}\frac{\exp \left[\frac{-ze(V\mathrm{m}-V\mathrm{pk})}{k\mathrm{T}}\right]}{{\left\{1+\exp \left[\frac{-ze(V\mathrm{m}-V\mathrm{pk})}{k\mathrm{T}}\right]\right\}}^2}+C_{\mathrm{lin}}$$ (1)

where Q_max is the maximum charge transferred, V_pk is the potential that corresponds to the peak of NLC and also has an equal charge distribution, z is the number of elementary charge (e), k is Boltzmann’s constant, T is the absolute temperature, and C_lin is the cell membrane capacitance. Curve fitting and figure plotting were performed with SigmaPlot software (SPSS Inc. Chicago, IL; RRID:SCR_003210). Membrane potential (V_m) was corrected for pipette series resistance (R_s).

For double patch-clamp recording in DC-OHC pairs, a pair of two DCs connected to one or two OHCs isolated from the guinea pig cochlea was selected. One pipette was patched at the DC and another patch pipette was patched at the basal nuclear pole of the OHC (see Fig. 9A) in a whole cell configuration with Axopatch 700A (Molecular Devices, California). The command and data recording in each patch clamp were separately controlled by jClamp (17).

Figure 9.

The influence of changes in membrane current in Deiters cells (DCs) on outer hair cell (OHC) electromotility-associated nonlinear capacitance (NLC) in a DC-OHC pair. A: a captured image of dual-patch-clamp recording at OHC and DC in a pair of DCs and OHC. One patch pipette was recording at an OHC, and another pipette was patched at DCs. B: change of OHC electromotility by alternation of holding current (I_h) in DCs. DCs were held at different holding currents, and OHC electromotility-associated NLC was simultaneously recorded by a voltage ramp (V_c) with sinusoidal voltage. OHC NLC was left-shifted with injection of negative current in DCs. Voltage at the peak of NLC (V_pk) was −86.0, −81.0, −77.5, −74.8, and −72.5 mV for DCs holding at −4, −2, 0, 2, and 4 nA, respectively.

Fluorescence Recovery after Photobleaching

The freshly isolated guinea pig cochlear epithelia or cells were incubated in 10 µM carboxy SNARF-1 AM (C-1271, Molecular Probes) at room temperature for 15–30 min, protected from light. Then, the incubated tissues or cells were continuously perfused with the NES for 30–45 min to remove the residual dye, allowing completion of the hydrolysis of AM ester before measurement.

Fluorescence recovery after photobleaching (FRAP) measurement was performed with the use of a laser scanning confocal system (Zeiss LSM 510 plus META confocal system, Thornwood, NY). Prior to laser bleaching, the fluorescence image of the selected area in the cochlear sensory epithelium was scanned and saved. Then, a laser beam (the laser power was nominally ∼50 µW) illuminated a selected cell for 20 s to bleach the fluorescence. After bleaching, the fluorescent images were taken at 0, 5, 15, 30, 60, 120, 210, and 330 s to measure the recovery. The intensity of fluorescence of the bleached cell was measured off-line with ImageJ software (NIH, Bethesda, MD; RRID:SCR_003070) and normalized to the fluorescent intensity in the same cell before bleaching. The fluorescence recovery data were fitted by

$$\mathrm{F}(t)=\mathrm{F}(\infty )-\Delta \mathrm{F}(0)e^{-t/\tau }$$ (2)

where F(∞) is the fluorescence signal of the bleached cell at t = ∞, ΔF(0) is the initial fluorescence change due to the bleaching, and τ is the recovery time constant.

Noise Exposure

Mice were awake and freely moving in a small cage under loudspeakers in a soundproof chamber and were exposed to white noise (96 dB SPL) one time for 2 h. Sound pressure level and spectrum in the cage were measured before placement of the animal (37).

Auditory Brainstem Response and Distortion Product Otoacoustic Emission Recording

As we previously reported (21, 38), the auditory brainstem response (ABR) was recorded with the use of a Tucker-Davis ABR & DPOAE workstation with ES-1 high-frequency speaker (Tucker-Davis Tech., Alachua, FL). Mice were anesthetized by intraperitoneal injection of a mixture of ketamine and xylazine (8.5 mL saline + 1 mL ketamine + 0.55 mL xylazine, 0.1 mL/10 g). Body temperature was maintained at 37–38°C. ABR was measured by clicks and tone bursts (8–40 kHz) from 80 to 10 dB SPL in a 5-dB step. The ABR threshold was determined by the lowest level at which an ABR could be recognized. If mice had severe hearing loss, the ABR test from 100 to 70 dB SPL was added (21, 38).

Distortion product otoacoustic emission (DPOAE) was recorded as described in our previous reports (21, 38). Two pure tones (f₁ and f₂) were simultaneously delivered into the ear through two plastic tubes coupled to two high-frequency speakers (EC-1, Tucker-Davis Tech., Alachua, FL). The test frequencies were presented by a geometric mean of f₁ and f₂ [f₀ = (f₁ × f₂)^1/2] from f₀ = 4 to 20 kHz. The ratio of f₂ versus f₁ (f₂/f₁) was 1.22. The intensity of f₁ was set at 5 dB SPL higher than that of f₂. One hundred fifty responses were averaged. A cubic distortion component of 2f₁ − f₂ in DPOAEs was measured.

Chemicals and Data Processing

All chemicals were purchased from Sigma Chemical Company (St. Louis, MO). Chemicals in patch-clamp recording were delivered by a Y-tube perfusion system (17, 32, 33). Patch-clamp data analyses were performed by jClamp or MATLAB (9, 17, 32). FRAP data analyses were performed with MATLAB. Data are expressed as means ± SE unless otherwise indicated in text and were plotted by SigmaPlot.

Statistical Analysis and Reproducibility

The statistical analyses were performed by SPSS v.18.0 (SPSS Inc. Chicago, IL; RRID:SCR_002865). Parametric and nonparametric data comparisons were performed by one-way ANOVA or Student t tests after assessment of normality and variance. The threshold for significance was P = 0.05. Bonferroni post hoc test was used in ANOVA.

The numbers of recording cells in each experiment are indicated in the figures. For morphological experiments, the staining was repeated at least three times. For the in vivo noise exposure experiment, only one or two mice were exposed to noise at each time.

RESULTS

Innervation of MOC in the Cochlear SCs

Efferent MOC nerves are well known to project to OHCs in the cochlea (1, 3). However, we found that the MOC fibers also innervate the cochlear SCs. Figure 1 shows that the MOC fibers, which were labeled by Tuj1 or neurofilament (NF), passed through the cochlear tunnel projecting to SCs in addition to OHCs. The MOC nerves had branches projecting to Deiters cells (DCs), outer pillar cells (OPCs), and Hensen cells (HCs) (Fig. 1, A–I). The nerve branch projected from the first row of OHCs to the second and third rows of DCs (indicated by white arrowheads in Fig. 1C). The MOC nerves are cholinergic fibers, and ACh is a primary neurotransmitter. Intensive labeling for ACh receptors (AChRs) was visible at the SCs with the MOC nerves (indicated by white arrowheads in Fig. 1, D–G), besides the positive AChR labeling at the basal pole of OHCs (indicated by green triangles in Fig. 1, F and G). However, the labeling was not colocated with the outer spiral bundle (OSB) of type II AN fibers (indicated by red triangles in Fig. 1, B and D). To further immunostain the synaptic terminals, we also used antibody to presynaptic vesicular acetylcholine transporter (VAChT) to show presynaptic endings (3, 39). Immunofluorescent staining for VAChT shows that small puncta of labeling associated with MOC nerves were visible at SCs (indicated by white arrowheads in Fig. 1, H and I), beside large puncta of labeling at the OHC basal pole. However, such VAChT labeling was not associated with enhanced green fluorescent protein (eGFP)-targeted type II AN fibers (Fig. 1, H and I). In addition, the staining showed different locations for MOC nerves and type II AN fibers crossing the cochlear tunnel. Tuj1-labeled MOC fibers crossed the cochlear tunnel at the middle level of the tunnel projecting to OHCs and SCs, whereas eGFP-targeted type II AN fibers from OHCs went down to the bottom at the outer range of the epithelium and along the bottom of the cochlear tunnel passing through it (Fig. 1, H and I).

Figure 1.

Innervations of medial olivocochlear (MOC) fibers in the cochlear outer hair cells (OHCs) and supporting cells (SCs). A: confocal image of immunofluorescent staining of the cochlear sensory epithelium at the apical-middle turn for neurofilament (NF, green), prestin (red), and Sox2 (purple) in whole mount preparation. The MOC fibers, OHCs, and SCs are labeled by NF, prestin, and Sox2, respectively. MOC fibers are clearly visible in OHC and SC areas. B: the surface view at the SC layer after removal of the OHC layer. Innervations of MOC fibers with SCs are clearly visible. Red triangles indicate the outer spiral bundle (OSB) of type II auditory nerves (ANs) running in the outer range of SCs. C: cross-sectional view of 3-dimensional (3-D) images constructed from z stack of confocal scanning images at the apical turn. White arrowheads indicate the MOC branch projecting to the Deiters cell (DC), Hensen cell (HC), and outer pillar cell (OPC), respectively. D–G: ACh receptor (AChR) expression at the OHCs and SCs. AChR and NF are labeled green and red, respectively. Green triangles indicate AChR expression at the OHC basal pole, and white arrowheads indicate expression of AChR at DCs and HCs. Red triangles in D indicate the OSB of type II ANs in outer range of the epithelium. H and I: immunofluorescent staining of the cochlear sensory epithelium of peripherin-eGFP transgenic mice (1 mo old) for enhanced green fluorescent protein (eGFP, green), Tuj1 (red), and vesicular acetylcholine transporter (VAChT, white). Type II ANs in peripherin-eGFP transgenic mice were labeled by eGFP, which was further enhanced by immunofluorescent staining for eGFP. The images are the cross-sectional view of 3-D images constructed from z stack of confocal scanning images at the apical-middle turn. White arrowheads indicate VAChT labeling at DCs with MOC fibers. CT, cochlear tunnel; IHC, inner hair cell. J: schematic drawing of innervations of MOC and type II AN fibers in OHC and SC areas in the cochlea. Scale bars: 20 µm in A, B, D, and E and 10 µm in C and F–I.

Current Responses of OHCs and SCs to ACh

To further test functional innervation, we recorded responses of OHCs and SCs to the MOC neurotransmitter ACh. ACh could evoke a typical inward current in OHCs and SCs (Fig. 2, B and D). At holding at −80 mV, 0.1 mM ACh evoked −0.31 ± 0.07 nA (n = 12) and −0.25 ± 0.07 nA (n = 5) inward currents in OHCs and DCs, respectively, in guinea pigs (Fig. 2F). There was no significant difference between them (P = 0.60, t test, 2-tailed). As ACh was repeatedly applied, the evoked inward current was repeatedly recordable (Fig. 2, B and D). We measured the ACh dose curve in single DCs in the guinea pig (Fig. 2E). The EC₅₀ was 72.6 µM, and Hill’s coefficient was 1.64 (Fig. 2E). The ACh-evoked inward current was also visible in mouse SCs (Fig. 3). However, compared with the responses in guinea pigs, the ACh-evoked current in mouse SCs appeared small. At −80 mV, 0.5 mM ACh-evoked currents in DCs of guinea pigs and mice were −0.38 ± 0.07 nA (n = 12) and −0.14 ± 0.03 nA (n = 7), respectively (Fig. 3B). The evoked current in mouse DCs was significantly smaller than that in guinea pig DCs (P = 0.006, t test, 2-tailed).

Figure 2.

ACh-evoked inward current in outer hair cells (OHCs) and cochlear supporting cells (SCs) in guinea pigs. A and B: ACh-evoked inward current in OHCs. A is a captured image of patch-clamp recording in a single OHC. Horizontal bars in B represent application of 0.1 mM ACh. The membrane potential was clamped at −80 mV. C and D: ACh-evoked inward current in a Deiters cell (DC). Horizontal bars represent the application of ACh. The membrane potential was clamped at −80 mV. E: dose curve of ACh-evoked current (I) in single DCs. Solid line represents data fitting to a Hill’s function: I = a × Cⁿ/(Kⁿ + Cⁿ), where Hill coefficient (n) = 1.64 and K = 72.6 µM (EC₅₀) for ACh. F: inward currents measured in OHCs and DCs at –80 mV evoked by 0.1 mM ACh. Green lines in boxes represent mean levels, which are –0.31 ± 0.07 and –0.25 ± 0.07 nA in OHCs and DCs, respectively (P = 0.60, t test, 2-tailed).

Figure 3.

ACh-evoked responses in mouse and guinea pig supporting cells (SCs). A: ACh-evoked current in a single mouse Deiters cell (DC). B: comparison of ACh-evoked current (I) in DCs in guinea pigs and mice. Green lines in boxes represent the mean levels. The average currents evoked by 0.5 mM ACh in guinea pig and mouse DCs are –0.38 ± 0.07 and –0.14 ± 0.03 nA, respectively. The evoked current in mice is significantly smaller than that in guinea pigs (P = 0.006, t test, 2-tailed). C: percentage of ACh-evoked responses in the recorded SCs in guinea pigs and mice. Numbers within each bar represent the number of cells with ACh-evoked responses vs. total of recorded cells. In both guinea pigs and mice, ACh-evoked responses could be observed in 91–100% of recorded DCs and 75–83% of recorded Hensen cells (HCs). However, only a few Claudius cells (CCs, 1/6) in guinea pigs and none of the recorded mouse CCs (0/4) had responses to 0.5 mM ACh.

In the experiment, we found that almost all recorded DCs (91–100%, i.e., 20/22 and 11/11 in guinea pigs and mice, respectively) and most of the recorded HCs (75–83%, i.e., 6/8 and 10/12 in guinea pigs and mice, respectively) had responses to ACh (Fig. 3C). However, only a few (1/6 and 0/4 in guinea pigs and mice, respectively) recorded Claudius cells (CCs) had response to ACh (Fig. 3C). This corresponded well with the observation by immunofluorescent staining that MOC fibers mainly innervated with DCs and HCs and rarely with CCs in the cochlea (Fig. 1).

Amplification of ACh Response in SCs by GJs

SCs in the cochlea are extensively coupled by GJs (13–15), which provide an intracellular electrical conduit between cells to synchronize or amplify electrical responses in a cell group. Figure 4 shows the ACh-evoked currents in two-coupled DCs (2DCs). At holding at −80 mV, the current evoked by 0.5 mM ACh in 2DCs in guinea pigs was −1.11 ± 0.45 nA (n = 6), three times larger than that (−0.38 ± 0.07 nA, n = 12) in single DCs (Fig. 4B). In mice (Fig. 4D), the current evoked by 0.5 mM ACh in 2DCs was −0.29 ± 0.14 nA (n = 4), two times larger than that (−0.14 ± 0.03 nA, n = 7) in the single DCs.

Figure 4.

Amplification of ACh-evoked inward currents in supporting cells (SCs) by gap junction (GJ) coupling. A and B: ACh-evoked current (I) in 2-coupled Deiters cells (2DCs) in guinea pigs. Green lines in boxes represent the mean levels. In comparison with a single DC (1DC), the current evoked by 0.5 mM ACh in 2DCs is −1.11 ± 0.45 nA, 3 times larger than that (−0.38 ± 0.07 nA) in single guinea pig DCs. C and D: ACh-evoked current in mouse DCs. Red lines in boxes represent the mean levels. In comparison with ACh-evoked current (−0.14 ± 0.03 nA) in 1DCs in mice, the average current evoked by 0.5 mM ACh in mouse 2DCs is −0.29 ± 0.14 nA, increased 2 times.

Effect of ACh on GJs between SCs

Cochlear SCs are well coupled by GJs (13–15). GJ channels are sensitive to intracellular Ca²⁺; elevation of intracellular Ca²⁺ can close GJ channels, i.e., uncouple GJs (40, 41). ACh receptors are permeable to Ca²⁺ ions and can elevate intracellular Ca²⁺ (42–44). We further assessed the effect of ACh on GJs between SCs. As reported in our previous studies (34, 35), C_in was recorded to assess GJ coupling between the cochlear SCs. C_in in the 2DCs and single DCs in guinea pigs was 69.9 ± 3.62 pF (n = 8) and 32.6 ± 1.54 pF (n = 15), respectively (Fig. 6A). C_in of single DCs was half (50%) of C_in in 2DCs. Application of 0.5 mM ACh reduced C_in in pairs of DCs to a half value (Fig. 5, A and C), indicating that two-coupled DCs were uncoupled, i.e., GJ channels became closed. In both guinea pigs and mice, C_in of 2DCs was reduced to 69.3 ± 7.94% (n = 8) and 64.0 ± 10.7% (n = 6), respectively, under application of 0.5 mM ACh (Fig. 6B). The uncoupling effect of ACh was reversible. After stop of perfusion of ACh, C_in was increased and returned to the coupled levels (Fig. 5, A and C). However, long-term application of ACh on the order of minutes usually caused irreversible uncoupling (Fig. 5, B and D). Figure 5D shows uncoupling effect of ACh in a HC group. During the long-term (∼8 min) treatment of 0.5 mM ACh, C_in showed a step reduction; a step was equal to a change of one-cell capacitance. However, C_in in a single cell had no apparent change to application of ACh (Fig. 5, E and F), although ACh could evoke current response in the same recording cells (Fig. 5, E and F, insets). This further indicated that C_in measurement was not dependent on changes in membrane current.

Figure 6.

Uncoupling of gap junctions (GJs) between cochlear supporting cells by ACh stimulation. A: input capacitance (C_in) measured in 2-coupled Deiters cells (2DCs) and single DC (1DC) in guinea pigs. C_in of 2DCs and 1DC in guinea pigs is 69.9 ± 3.62 pF and 32.6 ± 1.54 pF, respectively. C_in of 1DC is about half of C_in of 2DCs. B: uncoupling of GJs by ACh in 2DCs. Application of 0.5 mM ACh reduced C_in of 2DCs in guinea pigs and mice to 69.3 ± 7.94% and 64.0 ± 10.7%, respectively. The gray dashed line represents 50% of reduction in C_in after 2-coupled cells completely uncoupled to 1 cell. There is no significant difference in ACh-induced reduction of C_in, i.e., uncoupling effect, between guinea pig and mouse DCs (P = 0.70, t test, 2-tailed).

Figure 5.

ACh-induced uncoupling effect on gap junctions (GJs) between the cochlear supporting cells (SCs). The GJ coupling between the SCs was measured as input capacitance (C_in). A and B: ACh-induced uncoupling effect on GJs between Deiters cells (DCs) in guinea pigs. A shows the reversible uncoupling of GPs between 2-coupled DCs (2DCs) by ACh. After application of 0.5 mM ACh, C_in was reduced to the half value, to 1-cell level. B shows uncoupling of GJs in long-term application (∼16 min) of ACh. C and D: ACh-induced uncoupling effect on GJs between SCs in mice. D shows step changes in C_in for uncoupling of GJs among 3-coupled Hensen cells (3HCs) in long-term application (∼8 min) of ACh. E and F: no apparent changes in C_in in single DC (1DC) for the same ACh (0.5 mM) application in both guinea pigs and mice. Insets: the ACh-evoked current (I) in the same 1DC.

To further assess the effect of ACh on GJ coupling, we also used fluorescence recovery after photobleaching (FRAP) to measure GJ permeability (Fig. 7). The fluorescence in one cell in the outer supporting cell (DC and HC) area in the cochlear sensory epithelium was bleached by laser zapping (Fig. 7A). After bleaching, fluorescence in the bleached cell gradually recovered as fluorescent dye diffused back from neighboring cells through GJs (Fig. 7, A and B). The speed of the FRAP is inversely proportional to GJ permeability. Figure 7, B and C, show that after application of 0.5 mM ACh the recovery time constant of FRAP was increased to 67.3 ± 4.66 s (n = 48; Fig. 7C). In comparison with the time constant (44.7 ± 4.21 s, n = 62) in the control group without ACh treatment, ACh significantly increased the recovery time constant of FRAP > 50% (P < 0.001, one-way ANOVA with a Bonferroni correction). That is, ACh reduced GJ permeability between cochlear SCs. Moreover, the effect of ACh on GJ permeability showed slow development on the order of minutes (Fig. 7D). The time constant of the effect of ACh on GJ permeability was ∼11.0 min. Glutamate is a major excitatory neurotransmitter in the cochlea and is also thought to be the putative neurotransmitter of the synapses between OHCs and type II ANs (45, 46). Figure 7, B and C, show that application of glutamate (0.2 mM) had no significant effect on GJ permeability between SCs. The time constant of recovery was 42.8 ± 4.65 s (n = 31) and had no significant difference from that in the control group (P = 0.33, one-way ANOVA).

Figure 7.

The effect of neurotransmitters ACh and glutamate on gap junction (GJ) permeability between outer supporting cells [Deiters cells (DCs) and Hensen cells (HCs)] in guinea pigs measured by fluorescence recovery after photobleaching (FRAP). A: fluorescent images of cochlear sensory epithelium in the HC area and fluorescence recovery after photobleaching by laser zapping (indicated by red arrow). Scale bar: 10 µm. B: fluorescence recovery of outer supporting cells at ACh (0.5 mM) or glutamate (0.2 mM) treatment. The data points were averaged from different cells measured at 10–20 min after treatment with ACh or glutamate. Solid lines represent exponential fitting to data. C: the recovery time constant (τ) of FRAP. Green lines in boxes represent the mean levels. ACh but not glutamate significantly increased the recovery time constant of FRAP, i.e., reduced GJ permeability. **P < 0.001, one-way ANOVA with a Bonferroni correction. D: dynamic changes of GJ permeability in cochlear supporting cells by application of ACh. The time constants of FRAP were measured before and after application of 0.5 mM ACh. The black circles and solid line represent the average value from 3 measurements. The red line represents exponential fitting. The time constant of the fitting is 11.0 min.

Effects of ACh and Uncoupling of GJs between SCs on OHC Electromotility

OHCs have electromotility (8, 11), which plays a critical role in mammalian hearing (12). Figure 8, A–D, show the direct effect of ACh on OHC electromotility. After application of 0.1 mM ACh, OHC electromotility-associated NLC was shifted to the left negative voltage direction. The peak voltage (V_pk) of NLC was shifted from −22.8 ± 2.50 mV at control level to −30.7 ± 2.92 mV (n = 16, P = 2.68E−07, paired t test, 2-tailed) (Fig. 8C). The shift is reversible and repeatable. After stop of application of ACh, V_pk returned to the preapplication level (Fig. 8B). However, the peak capacitance (C_pk) had no significant changes or reduction. C_pk values for control and application of 0.1 mM ACh were 48.2 ± 1.50 pF and 48.4 ± 1.72 pF (n = 16), respectively (Fig. 8D). There was no significant difference between them (P = 0.71, paired t test, 2-tailed).

Figure 8.

Effects of ACh and uncoupling gap junctions (GJs) between Deiters cells (DCs) on outer hair cell (OHC) electromotility-associated nonlinear capacitance (NLC) measured in guinea pigs. A and B: ACh-induced changes in OHC NLC. After application of 0.1 mM ACh, NLC was shifted to left negative voltage direction. V_c, voltage command. B is the continuously tracked voltage at the peak of NLC (V_pk). Application of ACh caused V_pk to be shifted to negative voltage. The shift is reversible and repeatable. C and D: ACh-induced V_pk and peak capacitance (C_pk) changes. Dashed lines connecting 2 identical symbols represent the change of V_pk or C_pk in the same OHC before and after application of 0.1 mM ACh. Pink lines in boxes represent the mean levels. There is significant change in V_pk for application of 0.1 mM ACh (**P = 2.68E−7, paired t test, 2-tailed). However, there is no significant change in C_pk (P = 0.71, paired t test, 2-tailed). E: influence of uncoupling GJs between DCs on OHC NLC. Uncoupling GJs between DCs shifted OHC NLC to left negative voltage direction. Smooth lines represent NLC fitted by Boltzmann function. The parameters of fitting are maximum charge transferred (Q_max) = 4.20 and 3.29 pC, number of elementary charge (z) = 0.42 and 0.51, V_pk = −23.3 and −40.8 mV, and cell membrane capacitance (C_lin) = 16.6 and 19.7 pF for coupling and uncoupling of GJs between DCs, respectively. V_m, membrane potential. Inset: a captured image of recording at an OHC in a DC-OHC pair. F and G: effect of uncoupling of GJs between DCs on OHC NLC. Dashed lines connecting 2 identical symbols represent the change of V_pk or C_pk in the same OHC before and after uncoupled GJs between DCs. Green lines in boxes represent the mean levels. There is significant change in OHC V_pk for uncoupling GJs between DCs (**P = 0.0001, paired t test, 2-tailed). However, there is no significant change in C_pk of NLC in the OHC after uncoupled GJs between DCs (P = 0.06, paired t test, 2-tailed). H: shift of V_pk in single OHC and OHC in OHC-DC pair at the application of ACh and uncoupling of GJs between DCs. In OHC-DC pairs, direct effect of ACh on OHC electromotility and indirect effect of ACh-induced uncoupling between DCs on OHC electromotility are superimposed to shift OHC NLC to left negative voltage direction.

Our previous study demonstrated that electronic stimulation in DCs and/or changes in GJs between DCs could influence OHC electromotility even though there is no GJ between DCs and OHCs (17). Since ACh could evoke inward currents in DCs and uncouple GJs between DCs (Figs. 2–6), which could change membrane current in DCs, we further investigated the effect of uncoupling of GJs between DCs on OHC electromotility. Figure 8, E–G, show that uncoupling of GJs between the DCs in OHC-DC pairs could shift OHC NLC to the left negative voltage direction as observed for the direct effect of ACh on OHC electromotility (Fig. 8, A–C). V_pk of NLC was shifted from −34.9 ± 3.09 mV to −46.1 ± 2.63 mV (n = 11) after GJs between DCs uncoupled by using patch pipette breaking in one DC in two-coupled DCs (Fig. 8F). The change was significant (P = 0.0001, paired t test, 2-tailed). However, the C_pk of OHCs was not significantly changed (from 44.0 ± 1.65 pF to 45.4 ± 1.76 pF; n = 11) after DCs uncoupled (Fig. 8G) (P = 0.06, paired t test, 2-tailed). Both ACh and uncoupling of GJs between DCs produced the same left shift in OHC NLC (Fig. 8, C and F). Application of 0.1 mM ACh shifted V_pk of NLC by −7.85 ± 0.89 mV (n = 16) (Fig. 8C), and the uncoupling of GJs between DCs shifted OHC NLC about −11.2 ± 1.80 mV (n = 11) (Fig. 8F). Thus, both effects could produce an approximately −20-mV shift of V_pk in OHC NLC in OHC-DC pair configuration (Fig. 8H).

ACh could evoke large inward currents in DCs (Figs. 2–4). Changes in membrane current in DCs also can influence OHC electromotility (17). To assess the influence of such current changes on OHC electromotility, we further did double patch-clamp recording to simulate the effect of this current change in DCs on OHC electromotility. Figure 9 shows that the OHC and DC in a pair of OHC and DCs were simultaneously recorded by double patch clamps. When the holding current in the DC was changed from −4 to 4 nA, OHC NLC showed a left shift (Fig. 9B), the same as the observed effect of ACh and uncoupling of GJs between DCs on OHC NLC (Fig. 8). Thus, all ACh-evoked responses in SCs and OHCs (i.e., evoked inward current in SCs, uncoupling effect on GJs between SCs, and direct effect of ACh on OHCs) produced the same left shifting in OHC electromotility.

The Function of This SC-Mediated Efferent Pathway in Vivo

To assess the function of this SC GJ-mediated efferent pathway in vivo, we selectively deleted Cx26 expression in DCs by crossing with the Prox1-Cre line (21, 22). As described in previous reports (21, 22), Cx26 expression in the DCs was selectively deleted and there was no apparent hair cell degeneration in this Cx26 cKO mouse line (Fig. 10A). However, active cochlear amplification measured as DPOAE was reduced (Fig. 10, B and C). The gain of amplification, which was measured by amplitude of 2f₁ − f₂ re: f₁ amplitude in DPOAE, was reduced and the gain regulation assessed by the input/output (I/O) function was impaired (Fig. 10D).

Figure 10.

Deficiency of the gap junction (GJ)-mediated efferent pathway by targeted deletion of connexin 26 (Cx26) expression in Deiters cells (DCs) decreases active cochlear amplification and regulation. Wild-type (WT) littermates served as controls. Mice were at postnatal day 45. A: targeted deletion of Cx26 in DCs. A white arrow indicates lack of Cx26 labeling in the DC area in Cx26 conditional knockout (cKO) mice in immunofluorescent staining for Cx26 (green). Outer hair cells (OHCs) were visualized by prestin staining (red). Images at bottom show whole mounting view of targeted deletion of Cx26 in DCs by immunofluorescent labeling for Cx26. Scale bars: 50 µm. B: spectrum of acoustic emission recorded from Cx26 cKO mice and WT mice. Insets: large-scale plotting of 2f₁ − f₂ and f₁ peaks. The peak of distortion product otoacoustic emission (DPOAE) (2f₁ − f₂) in Cx26 cKO mice was reduced, but f₁ and f₂ peaks remained the same as those in WT mice. f₀ = 20 kHz, I₁/I₂ = 60/55 dB SPL. C: reduction of DPOAE in Cx26 cKO mice in input/output (I/O) plot. D: I/O function of amplification (DP) gain (2f₁ − f₂ re: f₁) in Cx26 cKO mice and WT mice. DP gain in WT mice was almost flat, whereas the DP gain in Cx26 cKO mice decreased as sound intensity increased. **P < 0.01, t test, 2-tailed.

We further tested the protective function against noise exposure in this mouse line. Figure 11 shows that after exposure to ∼96-dB white noise for 2 h, DPOAE and ABR thresholds in WT mice had a quick recovery in 3 days and completely recovered at postexposure day 28 (Fig. 11, C and D). However, DPOAE in Cx26 cKO mice had no recovery after noise exposure (Fig. 11A) and was continuously reduced, reaching the maximum reduction at postexposure day 3 (indicated by a blue arrow in Fig. 11A). ABR thresholds in Cx26 cKO mice also had significant increases after noise exposure (Fig. 11B). Increases in ABR thresholds in Cx26 cKO mice and WT mice at postexposure day 1 were 32.9 ± 3.66 dB SPL and 23.8 ± 2.27 dB SPL (P = 0.046, t test, 2-tailed), respectively, for the same noise exposure (Fig. 11, B and D); there was an ∼10-dB increment of ABR threshold in Cx26 cKO mice (Fig. 11D, inset). Moreover, unlike WT mice, whose ABR thresholds were completely recovered and returned to the control level at postexposure day 28 (Fig. 11D), there was an ∼10-dB permanent threshold shift (PTS) in Cx26 cKO mice (Fig. 11B). It is worthy to be noted that the Cx26 cKO mice without noise exposure had a progressive hearing loss and DPOAE reduction with age increased (Fig. 11, A and B), consistent with our previous report that this Cx26 cKO mouse line has progressive hearing loss (22).

Figure 11.

Deficiency of gap junction (GJ)-mediated control pathway increases susceptibility to noise in mice. Mice were exposed to 96 dB SPL white noise 1 time for 2 h. Black vertical arrows indicate the noise exposure day, which was defined as postexposure day 0. Control mice were not exposed to noise. Distortion product otoacoustic emission (DPOAE) (2f₁ − f₂) was measured at f₀ = 20 kHz, I₁/I₂ = 60/55 dB SPL. Auditory brainstem response (ABR) thresholds were measured by 16-kHz tone bursts and normalized to the pre-noise exposure level. A and B: noise-induced changes of DPOAE and ABR thresholds in connexin 26 (Cx26) conditional knockout (cKO) mice. Blue arrow in A indicates that DPOAE was reduced to the minimum level at postexposure day 3 in Cx26 cKO mice. DPOAE and ABR thresholds in Cx26 cKO mice were not completely recovered after noise exposure. C and D: changes of DPOAE and ABR thresholds in wild-type (WT) mice for noise exposure. DPOAE and the thresholds of ABR were completely recovered at postexposure day 28. Inset: changes of ABR thresholds in Cx26 cKO mice and WT mice at day 1 after noise exposure. Blue lines in boxes represent the mean levels. The increase in ABR thresholds in Cx26 cKO mice at postexposure day 1 was significantly larger than that in WT mice for the same level of noise exposure. *P = 0.046, t test, 2-tailed.

DISCUSSION

In this experiment, we found that MOC efferent nerves had functional innervations with SCs in the cochlea (Figs. 1–3). Application of the MOC neurotransmitter ACh could evoke inward currents in SCs and reduced GJs between them (Figs. 2–7). This ACh-induced uncoupling effect on GJs between SCs had a long-lasting influence (Figs. 4 and 7) and enhanced the direct effect of ACh on OHC electromotility, i.e., both indirect effect via SCs and direct effect on OHCs shifted NLC to the left negative voltage direction but did not reduce the NLC (Figs. 8 and 9). Deficiency of this SC GJ-mediated control via targeted deletion of Cx26 between DCs compromised the regulation of active cochlear amplification and the protection of hearing against noise trauma (Figs. 10 and 11). Taken together, these data demonstrate that the MOC efferent nerves have functional innervations with the cochlear SCs to inhibit GJs between them, thereby modulating OHC electromotility. These data also suggest that this SC GJ-mediated efferent pathway has a critical role in control of hearing sensitivity and hearing protection against noise.

Previous studies reported that there are nerve innervations in the cochlear SCs, which could form chemical synapses with outer supporting cells (DCs and HCs) (23–28). However, the source, function, and significance of such neural innervations in the cochlear SCs remain unclear or are under debate (29, 30). In this study, we found that the MOC fibers had branches projecting to the cochlear SCs (Fig. 1). Presynaptic VAChTs associated with MOC nerves were visible in SCs (Fig. 1, H and I). Postsynaptic AChRs also had expression at the SCs (Fig. 1). Finally, application of the principal MOC neurotransmitter ACh could evoke current responses in the cochlear SCs (Figs. 2–4). Taken together, these data suggest that cholinergic MOC fibers have functional innervations with the cochlear SCs.

It has been reported that the cochlear SCs also have innervations of the branches from the type II AN fibers under OHCs (29, 30, 47–49). However, type II ANs are not cholinergic fibers. In the experiments, we found that VAChT and AChR expressed in the SCs and were associated with MOC fibers (Fig. 1, D–I). Application of ACh could evoke the current responses in SCs (Figs. 2–4). Thus, type II AN fibers are unlikely to be responsible for our observed cholinergic nerve fiber innervations and cholinergic responses in SCs.

The dose dependence of ACh-evoked inward currents in the cochlear SCs (Fig. 2E) provides further evidence for MOC fibers functionally innervating the cochlear SCs. The Hill’s coefficient of ACh in the DC was 1.64 (Fig. 2E), which is comparable with ACh-evoked responses in OHCs in previous studies, e.g., 1.6 in the report of Shigemoto and Ohmori in 1991 (50), 1.7 in the report of McNiven et al. in 1996 (51), and 1.6 in the report of Dallos et al. in 1997 (52). The EC₅₀ in single DCs was 72.6 µM (Fig. 2E), which is higher than that (10–20 µM) recorded in OHCs (50–52). However, considering that DCs are electrically coupled by GJs, the electronic response in the DC group could be amplified by GJ coupling between them. As shown in Fig. 4, the responses in two-coupled DCs were amplified by two to three times. Thus, the EC₅₀ in the group DCs could be reduced by two to three times to 20–30 µM or even more, close to the EC₅₀ recorded in OHCs.

The fact that ACh could evoke inward currents in the SCs indicates that AChRs in the SCs are nicotinic. It has been reported that nicotinic α9 and α10 subunits are predominant isoforms in the cochlea and are permeable to Ca²⁺ (2, 42–44, 53–57). We found that SCs had positive labeling to AChRα9 (Fig. 1, D–G). This is also consistent with a previous report that ACh could elevate intracellular Ca²⁺ concentration in DCs, which could be eliminated by α9 AChR antagonists (58). Thus, the SCs may have α9 nicotinic AChR expression. However, the detailed information about composition and formation of AChRs in the SCs currently is unclear and needs to be further investigated in the future.

In this study, we found that ACh could close GJs between the cochlear SCs (Figs. 5–7). This may result from ACh-induced Ca²⁺ influx to close GJ channels, since GJ channels, including the inner ear GJ channels, are sensitive to intracellular Ca²⁺; elevation of intracellular Ca²⁺ can close GJ channels (40, 41). We previously reported that ATP can activate purinergic P2x receptors, leading to Ca²⁺ influx in the cochlear SCs to close GJs between them (34). ACh may take the same mechanism to close GJs between SCs. As discussed above, ACh could activate AChRs in the SCs and influx Ca²⁺ to elevate intracellular Ca²⁺, thereby leading to closed GJ channels.

In the experiments, we further found that uncoupling of GJs between DCs could shift OHC electromotility to the left hyperpolarization (negative voltage) direction and enhanced the direct effect of ACh on OHC electromotility (Fig. 8). Our previous studies demonstrate that there is no electronic and GJ coupling between DCs and OHCs (9, 59). However, OHC electromotility is membrane tension or loading dependent (10, 60, 61). In situ, phalangeal processes of DCs attach to the apex of an OHC at the reticular lamina, and their basal ends extend to the basal of another OHC (Fig. 8E, inset, and Fig. 9A). Thus, the mechanical changes in DCs can alter OHC loading or tension, thereby modulating OHC electromotility. DCs have microfilaments along their phalangeal processes. It has been reported that Ca²⁺ influx and electronic stimulation can induce a small movement of the head of the DC phalangeal process and increase its stiffness (62, 63), which could consequently alter OHC loading or tension to modulate OHC electromotility. Indeed, our previous study demonstrated that the membrane current changes in DCs and uncoupling of GJs between them could modulate OHC electromotility (17); the modulation could be eliminated by destruction of cytoskeleton in DCs (17). In this study, we found that ACh could evoke inward current in DCs (Figs. 2–4); ACh also could uncouple GJs between them, which could produce large current changes as well (Fig. 4). As described above, these current changes in DCs could modulate OHC electromotility through the same mechanism, i.e., causing DC phalangeal process contraction or changing its tension to alter OHC loading, thereby modulating OHC electromotility.

It is intriguing that we found that ACh shifted OHC electromotility-associated NLC to the left hyperpolarization direction (Fig. 8, A–C) but did not reduce NLC (Fig. 8D). However, this is consistent with previous reports that ACh did not inhibit electromotility measured in the isolated OHC (52, 64). This result seems contrary to a common concept that the MOC system reduces OHC electromotility to inhibit the cochlear responses in vivo. At present, the detailed mechanisms for how the MOC efferent system inhibits active cochlear mechanics or plays an inhibitory effect in vivo are still unclear. In this study, we found that MOC fibers had innervations with SCs (Fig. 1) and that ACh could evoke inward current in DCs (Figs. 2–4) and uncoupled GJs between them (Figs. 5–7), which could shift OHC electromotility and enhanced the direct shifting effect of ACh on OHC electromotility (Fig. 8). ACh also could evoke intracellular Ca²⁺ elevation in DCs (58). As discussed above, all of these ACh-evoked responses in DCs could cause the DC phalangeal process contraction and increase its stiffness (62, 63), thereby altering OHC loading to shift OHC electromotility (17). This could eventually reduce active cochlear amplification (65). In particular, the cochlear SCs are extensively coupled by GJs, and these ACh-evoked SC contractions could increase the whole organ of Corti stiffness to reduce active cochlear amplification in vivo. Indeed, it has been reported that the whole organ of Corti including DCs and HCs in vivo was contracted to reduce sensitivity under acoustic overstimulation (66, 67). Moreover, such contraction could still exist and last a long time (>30 min) after the stimulation was terminated, suggesting that SCs may take an active part in the protection (66, 67). Taken together, these data suggest that this SC-mediated efferent pathway may play an important role in the efferent inhibition of active cochlear amplification in vivo by changing the stiffness of organ of Corti and shifting the operation point of OHC electromotility and active cochlear amplification.

These data may also have an important implication that this SC GJ-mediated efferent pathway may have a role in the slow MOC effect. Besides the fast effect, the MOC efferent system has a slow effect on the order of minutes (1, 68). At present, the mechanism underlying this slow MOC effect still remains unclear. In this study, we found that, different from the fast dynamic of the direct effect of ACh on OHC electromotility (Fig. 8B), the effect of ACh on GJs between SCs was relatively slow and had a long-lasting influence (Figs. 4, 5, and 7). This is consistent with the observation that contraction of SCs lasted a longer time in vivo even if the stimulation was terminated (66, 67). Moreover, it has been reported that the slow MOC effect plays a critical role in the protection of hearing from noise (1). In the experiment, we found that impairment of this SC GJ-mediated pathway could increase susceptibility to noise and caused a delayed impairment in the active cochlear amplification (Figs. 10 and 11). These data further demonstrate that this indirect SC-mediated efferent control pathway has a critical role in the protection of hearing from noise trauma. These data also suggest that this SC-mediated efferent control pathway may have an important role in the slow efferent effect.

We previously reported that GJs have a critical role in active cochlear amplification (21, 22). In this study, we further demonstrated that GJs also play an important role in the regulation of hearing sensitivity (Figs. 8 and 10). In particular, we found that Cx26 deficiency could increase susceptibility to noise (Fig. 11). This indicates that the GJs in the cochlea also play an important role in the protection of hearing from noise trauma. It is well known that GJ deficiency can induce a high incidence of nonsyndromic hearing loss (18–20). However, as mentioned above, GJs only exist in the cochlear SCs and not hair cells in the cochlea (9, 13–15, 17). Our new findings demonstrate that GJs between the SCs in the cochlea have broader functions than simple intracellular transfer and communication between cells. This study also provides valuable cues for understanding the deafness mechanisms underlying GJ deficiency-induced hearing loss.

GRANTS

This work was supported by NIH Grants R01 DC 017025, R01 DC 019687, and R56 DC 016585 to H.-B.Z.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

H.-B.Z. conceived and designed research; H.-B.Z., L.-M.L., N.Y., Y.Z., L.M., J.C., and C.L. performed experiments; H.-B.Z., L.-M.L., N.Y., Y.Z., L.M., J.C., and C.L. analyzed data; H.-B.Z. and L.-M.L. interpreted results of experiments; H.-B.Z., L-M.L., and N.Y. prepared figures; H.-B.Z. drafted manuscript; H.-B.Z. and L.-M.L. edited and revised manuscript; H.-B.Z., L.-M.L., N.Y., Y.Z., L.M., J.C., and C.L. approved final version of manuscript.