Cochlear outer hair cells (OHCs) are electromotile and implicated in amplification of responses to sound that enhance sound sensitivity and frequency tuning. They send afferent information through glutamatergic synapses onto type II spiral ganglion neurons (SGNs). These synapses are weaker than those from cochlear inner hair cells (IHC) onto type I SGN, suggesting that type II SGNs respond only to intense sound levels. OHCs also receive efferent innervation from medial olivocochlear (MOC) neurons. MOC neurons are cholinergic yet inhibit OHCs due to the functional coupling of alpha9/alpha10 nicotinic acetylcholine receptors (nAChRs) to calcium-activated SK potassium channels. The resulting hyperpolarization reduces OHC activity-evoked electromotility and is implicated in cochlear gain control, protection against acoustic trauma, and attention. MOC neurons also label for markers of GABA and GABA synthesis. GABA_B autoreceptor (GABA_BR) activation on MOC terminals has been demonstrated to reduce ACh release, confirming important negative feedback roles. However, the full complement of GABAergic activity in the cochlea is not currently understood, including mechanisms of GABA release from MOC axons, whether GABA diffuses from MOC axons to other postsynaptic cells, and the location and function of GABA_A receptors (GABA_ARs). We used optical neurotransmitter detection, immunohistochemistry, and patch-clamp electrophysiology to demonstrate that in addition to presynaptic GABA_B autoreceptor activation, MOC efferent terminals release GABA onto type II SGN afferent dendrites with postsynaptic activity mediated by GABA_ARs. This synapse may have roles including developmental regulation of cochlear innervation, fine tuning of OHC activity, or providing feedback to the brain about MOC and OHC activity.

Mice and ethical approval

Animal procedures followed National Institutes of Health guidelines, as approved by the National Institute of Neurological Disorders and Stroke/National Institute on Deafness and Other Communication Disorders Animal Care and Use Committee. Male and female mice were used in experiments. Mouse lines included wildtype (WT) C57BL/6J (Jax Cat No: 000664), ChAT-IRES-Cre (Jax Cat No: 028861), ChAT-IRES-Cre Δneo (Jax Cat No: 031661), Ngn-Cre^ERT2 (Jax Cat No: 008529), Ai14 tdTomato reporter mice (Jax Cat No: 007914), and Bhlhb5-Cre (on Sv/129, C57BL/6J mixed background). Mice were housed on a 12/12 hr light/dark cycle, with continuous availability of food and water. Mice were anesthetized by carbon dioxide inhalation at a rate of 20% of anesthesia chamber volume per minute and then killed by decapitation.

Immunohistochemistry

Mice aged postnatal day (P)5- 20 were euthanized as above and used for immunohistochemical experiments in cochlear whole mount preparations. Temporal bones were immediately removed and fixed via perfusion of 1 ml cold 4% PFA (Electron Microscopy Science) in 1X PBS through the oval and round windows, with 20-30 min post-fix in cold 4% PFA. Cochleae were washed with 1X PBS perfused through the round and oval window, then washed 3x 10-15 min in 1X PBS. Cochleae were either fully or partially dissected, followed by blocking non-specific labeling (10% normal goat/donkey serum in 0.5% Triton X-100/1X PBS) for 1-2 hr, and washed for 15-20 min in 1X PBS. For experiments using an anti-GAD antibody (antibodies listed in Table M1), cochleae underwent an additional block in M.O.M. buffer (3-4 drops in 2.5 ml 1X PBS; Vector MKB-2213-1) at room temperature (RT) for 1 hr. After a 15–20-min wash, cochlear turns were incubated in primary antibody in blocking solution (Table M1) at 4ºC for 2-48 hr. Tissues were washed 3x 20 min in 0.25% TX-100/1X PBS, then incubated in secondary antibody in 5% normal goat/donkey serum and 0.25% TX-100/1X PBS for 1-2 hr at RT. Tissues were washed for 15 min in 0.25% TX-100/1X PBS with DAPI (1:5000), washed 2x 10 min in 0.25% TX-100/1X PBS, then washed 15-20 min in 1X PBS. 

Antibody-labeled cochlear whole mounts were imaged on a Nikon A1R inverted microscope with 40X or 60X objectives, using Nikon Elements software v5.30.

Table M1. Antibodies used in immunohistochemistry experiments.

Patch-clamp recordings

For patch-clamp recordings from outer hair cells (OHCs) and type II spiral ganglion neuron (SGN) afferent dendrites,  the cochleae of P11-13 (OHC) or P5-10 (type II SGN) mice were dissected and placed under an insect pin affixed to a round glass coverslip with Sylgard (Dow Chemicals, Midland, MI, USA). Tissue was initially visualized using a Nikon Eclipse NI-E microscope with a QI-Click camera or FN-1 microscope with a pco.edge camera, with a 4X air objective for positioning of stimulation electrodes, drug application pipettes, or suction electrodes. Recordings were then performed using 40X or 60X water immersion objectives with DIC optics and an additional 1.25-2X magnification. For experiments with visualization of fluorescent cells, tissue was visualized under epifluorescent illumination using a Sola or Aura II lamp (Lumencor, Beaverton, OR, USA).

For OHC recordings, evoked inhibitory postsynaptic currents (IPSCs) were obtained from the first row of OHCs in the whole-cell voltage-clamp recording configuration in response to extracellular electrical stimulation of the MOC fibers ^54,55. For type II SGN dendrite recordings, fibers under OHCs were exposed by removal of 4-6 OHCs using a glass pipette with a 10-15 µm diameter-tip. Patch-clamp recordings were performed using 1 mm diameter glass micropipettes (WPI: World Precision Instruments, Sarasota, FL, USA) pulled (Sutter P1000, Sutter Instrument Company, Novato, CA, USA) to resistances of 3-6 MΩ (OHC recordings) or 6-10 MΩ (type II SGN dendrite recordings). Dissection and recording ‘extracellular’ solution contained (in mM): 5.8 KCl, 155 (OHC recordings) or 150 (all else) NaCl, 0.9 MgCl_2, 1.3 CaCl_2, 0.7 NaH₂PO_4, 5.6 glucose, and 10 HEPES. The pH was 7.4, adjusted with 1N NaOH, and osmolarity = ~315 mOsm. Extracellular solution at RT was perfused through the recording chamber at a rate of ~1-2 ml/min. Internal solutions for type II SGN recordings contained (in mM): 50 KCl, 80 K-methanesulfonate, 5 MgCl_2, 0.1 CaCl_2, 5 EGTA, 5 HEPES, 5 Na₂ATP, 0.3 Na₂GTP, and 5 Na₂-phosphocreatine. The pH was adjusted to 7.2 with KOH. 0.01 mM AlexaFluor hydrazide 488 or 594 was added during some experiments to allow visualization of cell morphology to confirm neuron identity. The osmolarity = ~290 and the liquid junction potential = -9 mV. Internal solutions for OHC recordings contained (in mM): 140 KCl, 3.5 MgCl₂, 0.1 CaCl₂, 5 EGTA, 5 HEPES, and 2.5 Na₂ATP. The pH was adjusted to 7.2 with KOH and osmolarity = 283-290 mOsm. Drug application was performed using either a large bore gravity-fed glass pipette positioned close to the cochlear tissue (GABA or ACh application) or via addition to the re-circulating bath solution. Drugs were obtained from Sigma-Aldrich, Inc. (St. Louis, MO, USA), Alomone Labs (Jerusalem, Israel), and Thermo Fisher Scientific (Waltham, MA, USA).

Patch-clamp recordings were performed in voltage-clamp using a Multiclamp 700B amplifier with 1550B or 1322A Digidata controlled by Multiclamp Commander v11.2 and pClamp v11.2 (Molecular Devices, Silicon Valley, CA, USA) or a HEKA EPC 10 amplifier controlled by PatchMaster v2.91 (HEKA Instruments Inc., Holliston, MA, USA). Recordings were sampled at 50 kHz and lowpass filtered at 10 kHz. OHCs were held at -40 mV for recordings of MOC to OHC synaptic activity. Type II SGN dendrites were held at -80 mV for most recordings unless stated otherwise. To test the reversal potential of GABAergic currents, the type II SGN dendrite membrane potential was stepped to holding voltages from -80 to +20 mV in 20 mV increments for 10 seconds prior to GABA application. In all patch-clamp experiments, GABA application was performed once per 10 min.

Electrical stimulation of MOC axons

Endogenous neurotransmitter release from MOC axons was evoked using electrical axon stimulation via ~10-20 µm diameter-glass micropipettes (Sutter) containing extracellular solution described above and placed ~20-30 µm modiolar to the inner spiral bundle (ISB). For OHC recordings, the stimulation electrode was positioned below the IHC aligned with the OHC under study. For optimal stimulation, cochlear supporting cells were gently removed by applying negative pressure through a glass pipette with a broken tip. The MOC axon stimulation was applied using an electrical stimulus isolation unit (Iso Stim, AMPI or SIU isolation unit A-M System). Stimulation timing and rate (for imaging experiments: 50 Hz, 1 second train duration) was controlled by the PClamp software during both patch-clamp and optical experiments.

Estimation of the quantal content of transmitter release.

Under the assumption that evoked synaptic events in synapses with a low release probability follow a Poisson distribution, the quantal content of transmitter release (m) upon electrical stimulation of the MOC fibers was estimated by the “method of failures”: m = ln (N/N₀)⁵⁶, where N is the total number of stimuli and N₀ the number of failures. Protocols of 100 stimuli at a frequency of 1 Hz were used to estimate m. At this stimulation frequency, neither the kinetics nor the amplitudes of the evoked responses changed throughout the protocol. To test the effect of drugs or toxins on the quantal content of evoked ACh release from the MOC synaptic terminals, m was first assessed 2-3 times in order to establish a mean control value (m_c) before incubating the tissue with the drug or toxin to evaluate changes in neurotransmitter release.  The tissue was then incubated for the time necessary to reach a plateau in the compound’s effect (ranging from 5 to 15 minutes, depending on the compound), after which the stimulation protocol was repeated (2-3 times).

Posterior semi-circular canal (PSC) adeno-associated virus (AAV) injections

Posterior semicircular canal (PSC) injections to introduce adeno-associated virus (AAV) particles into the cochlea were performed as described in ⁵⁷, using aseptic procedures. In brief, neonatal pups (P0-2) were hypothermia-anaesthetized for ~5 min until they did not respond to stimulation, and then remained on an ice pack for the duration of the procedure. A postauricular incision was made using micro-scissors and the skin was retracted. The PSC was identified under a surgical microscope and a glass micropipette pulled to a fine point was positioned using a micro-injector (WPI). For each mouse only one ear was injected with ~1.2 µL AAV solution containing gene sequences encoding optical GABA indicator (iGABASnFR) variants: iGABASnFR.F102G or FLEX.iGABASnFR2.0. The incision was closed using a drop of surgical glue. About 4-5 pups per litter were injected. Pups were recovered to normal body temperature on a warming pad, while receiving manual stimulation to aid recovery. To increase the likelihood of the dam accepting the pups post-surgery, each pup was gently rubbed with bedding from the home cage and, if possible, urine that was collected from the dam using a cotton-tipped applicator, before returning to their home cage. In addition, mineral oil was applied to the dam’s nose to block detection of any surgical or human odors on the pups prior to being returned to the home cage.

iGABASnFR imaging

For fluorescence imaging of iGABABSnFR-transduced cells in acutely dissected cochlear preparations, the euthanasia, dissections, extracellular solutions, drug application, and electrical stimulation of MOC axons were as above for patch-clamp recordings. iGABASnFR.F102G was obtained as plasmids from Addgene.org (Watertown, MA, USA). The plasmids were then packaged by SignaGen (Frederick, MD, USA) into AAV particles with the PHP.eB serotype and human synapsin (hSyn) promotor, the combination of which results in expression specifically in neurons, although a few hair cells also exhibited expression. iGABASnFR2-containing plasmids were kindly gifted by the GENIE Project Team at the Janelia Research Campus, Howard Hughes Medical Institute (Ashburn, VA, USA), and similarly packaged into AAV particles.

iGABASnFR imaging experiments were carried out using non-Cre-dependent virus injected into C57BL/6J mice (P14-15) or Cre-dependent (FLEX) virus injected into Ngn1-CreERT2; tdTomato or Bhlhb5-Cre; tdTomato mice (P8-11), which result in iGABASnFR expression specifically in neurons. There was no difference in the baseline fluorescence of FLEX.iGABASnFR2.0 between Bhlhb5-Cre or Ngn1CreERT2 mice (Bhlhb5-Cre mean baseline fluorescence: 36.4 ± 16.4 arbitrary units (AU), n = 14, Ngn1CreERT2 mean baseline fluorescence 40.6 ± 11.3 A.U., n = 3 preparations, One-way ANOVA p = 0.92262), so results were pooled between these two genotypes for the P8-11 mouse experiments. The iGABASnFR or tdTomato fluorescence was localized to cells with the clear morphology of type II SGN dendrites. In experiments in which OHCs were also labeled, the focal plane of imaging was set to image only type II SGNs, or in the case of curved tissue that had focal planes containing both OHCs and type II SGN fibers, only the type II SGN-containing regions-of-interest (ROIs) were analyzed (for analysis see Image Processing, below). Although unlikely, we cannot completely exclude the possibility that iGABASnFR was aberrantly expressed in MOC neurons due to spread of AAV particles via cerebrospinal fluid (CSF). However, analysis was restricted to regions with clear type II SGN morphology, and we did not detect expression in MOC axon terminals.

iGABASnFR imaging was performed on a Nikon A1R upright confocal microscope using resonant scanning in both red (568 nm, for tdTomato imaging in cochlear neurons from transgenic mice) and green (488 nm, for iGABASnFR variants) channels. In a subset of experiments with exogenous application of GABA or ACh, but not in experiments with electrical stimulation of MOC axons, the Nikon ‘denoise’ function was utilized during imaging, which improves image quality by estimating and removing the Poisson-distributed shot noise, without changing signal intensity. In some experiments, a DIC-like image was simultaneously collected using the transmitted light detector, which converts the laser signal into a greyscale 3D image. Imaging settings included line averaging of 4-16 lines, bi-directional scanning, 512-1024 resolution, and frame rates of ~7-15 frames per second.

Image processing

Fluorescence intensity changes were measured in iGABASnFR-transduced neurons using ImageJ (NIH, Bethesda, MD, USA). A maximum intensity projection of the green (iGABASnFR) image stack was generated and then thresholded to set the regions to be used for ROI selection. Thresholds were set to 121, but manually adjusted in the case of tissue with brighter or dimmer fluorescence background (mean = 119 ± 20). The ‘analyze particles’ function was used to automatically draw ROIs around the iGABASnFR-expressing structures of interest, and each ROI was classified by eye as type I or type II SGNs from either fluorescence or transmitted detector images. It was not possible to identify individual neurons from either tdTomato or iGABASnFR images, so ROIs likely contain multiple neuronal segments. These ROIs were then used to measure fluorescence intensities in the original image stack for each frame. Fluorescence intensity values per ROI and per frame were imported into Origin v2021 (Origin Lab, Northhampton, MA, USA). The fluorescence baseline was measured for 1 second prior to stimulus onset (electrical MOC axon stimulation) or 10 seconds prior to drug application (exogenous GABA or ACh application). The iGABASnFR responses were determined for the 30 seconds of exogenous ACh or GABA application compared to the 10 seconds prior to drug application. For experiments utilizing electrical stimulation of neurotransmitter release from MOC axons, we made a binary determination of whether each ROI had a positive ‘response’ to the axon stimulation. First, we measured the mean and standard deviation of the baseline fluorescence (1 second prior to MOC axon stimulation). Then, we measured the maximum fluorescence following MOC axon stimulation. To prevent a noisy fluorescence signal from giving an artificially high ‘maximum’ intensity, we used a rolling average (5 frames before, 5 frames after) to smooth the trace, and determined the fluorescence maximum from this rolling average trace. An ROI had a positive ‘response’ if the maximum fluorescence following axon stimulation was greater than two standard deviations above the baseline mean (mean +2SDs). Next, to determine the magnitude of the response to MOC axon stimulation we calculated the ΔF/F of the fluorescence. This was defined as the maximum fluorescence following MOC axon stimulation minus the mean baseline fluorescence, divided by the mean baseline fluorescence.

Quantification and Statistical Analyses

Statistics were performed in Origin v2021-2023 or GraphPad Prism 6 (RRID:SCR_002798, Boston, MA, USA). Data were tested for normality using a Shapiro-Wilk test. Statistical tests included a one-way ANOVA for parametric tests and one-way repeated measures ANOVA with post-hoc pairwise Scheffe tests. For non-parametric statistical analyses, with two groups, a Wilcoxon rank sum was applied, and for non-parametric comparisons of more than two groups, a Friedman test followed by a Dunn multiple-comparison test was used. Figures were prepared in Origin, ImageJ, and Adobe Illustrator (Adobe, San Jose, CA, USA).

Methods References

           Ballestero, J. et al. Short-term synaptic plasticity regulates the level of olivocochlear inhibition to auditory hair cells. J. Neurosci. 31, 14763–74 (2011).

           Vattino, L. G., Wedemeyer, C., Elgoyhen, A. B. & Katz, E. Functional Postnatal Maturation of the Medial Olivocochlear Efferent–Outer Hair Cell Synapse. J. Neurosci. 40, 4842–4857 (2020).

           DEL CASTILLO, J. & KATZ, B. Quantal components of the end-plate potential. J. Physiol. 124, 560–73 (1954).

           Isgrig, K. & Chien, W. W. Posterior Semicircular Canal Approach for Inner Ear Gene Delivery in Neonatal Mouse. J. Vis. Exp. (2018) doi:10.3791/56648.