Skip to main content
PMC home page
eLife logoLink to eLife
. 2017 Mar 21;6:e21452. doi: 10.7554/eLife.21452

Fast-spiking GABA circuit dynamics in the auditory cortex predict recovery of sensory processing following peripheral nerve damage

Jennifer Resnik 1,2,*, Daniel B Polley 1,2,*
Editor: Dwight E Bergles3
PMCID: PMC5378474  PMID: 28323619

Abstract

Cortical neurons remap their receptive fields and rescale sensitivity to spared peripheral inputs following sensory nerve damage. To address how these plasticity processes are coordinated over the course of functional recovery, we tracked receptive field reorganization, spontaneous activity, and response gain from individual principal neurons in the adult mouse auditory cortex over a 50-day period surrounding either moderate or massive auditory nerve damage. We related the day-by-day recovery of sound processing to dynamic changes in the strength of intracortical inhibition from parvalbumin-expressing (PV) inhibitory neurons. Whereas the status of brainstem-evoked potentials did not predict the recovery of sensory responses to surviving nerve fibers, homeostatic adjustments in PV-mediated inhibition during the first days following injury could predict the eventual recovery of cortical sound processing weeks later. These findings underscore the potential importance of self-regulated inhibitory dynamics for the restoration of sensory processing in excitatory neurons following peripheral nerve injuries.

DOI: http://dx.doi.org/10.7554/eLife.21452.001

Research Organism: Mouse

Introduction

The enduring plasticity of the adult brain supports a remarkable recovery of perceptual and motor capabilities following peripheral nerve injury. In sensory cortex, reorganization following injury involves a rapid unmasking of inputs from adjacent, undamaged regions of the sensory periphery and, in cases of incomplete injury, a slower, progressive increase in responsiveness to surviving nerve fibers within the damaged region (Rasmusson, 1982; Merzenich et al., 1983; Gilbert and Wiesel, 1992; Kaas et al., 1990; Robertson and Irvine, 1989; Qiu et al., 2000; Chambers et al., 2016; Calford and Tweedale, 1988). The initial stages of this reorganization may be enabled by homeostatic processes that compensate for dramatic swings in afferent input to maintain neural excitability around a set point (Nahmani and Turrigiano, 2014). Recordings from cortical pyramidal neurons in culture or acute brain slices following sudden shifts in excitatory input have revealed a coordinated sequence of changes at glutamatergic and GABAergic synapses that normalize firing rates and rebalance network activity (Kilman et al., 2002; O'Brien et al., 1998; Turrigiano et al., 1998; Xue et al., 2014; D'amour and Froemke, 2015; Li et al., 2014). Whether and how synaptic and extra-synaptic modifications underlie reorganized sensory processing in deafferented zones of the adult cortex remains to be determined, but accumulating evidence from intact preparations suggests that the stabilization of new synaptic inputs from spared regions of the sensory periphery may emerge through a combination of rapid disinhibition and increased structural motility (Li et al., 2014; Marik et al., 2014; Yamahachi et al., 2009; Darian-Smith and Gilbert, 1994; Keck et al., 2011, 2013; Garraghty et al., 1991; Florence et al., 1998).

In the auditory system, a loss of cochlear hair cells and/or cochlear nerve afferent synapses is associated with reduced GABA signaling and cortical hyperexcitability, as manifest in elevated spontaneous firing rates, enhanced central gain and unmasking of sound-evoked responses from undamaged regions of the cochlea (Qiu et al., 2000; Chambers et al., 2016; Seki and Eggermont, 2003; Yang et al., 2012; Scholl and Wehr, 2008). However, these biomarkers have not been compared with one another over the full course of functional recovery with fine-grain temporal resolution at the level of individual neurons. As a result, it is unclear whether dynamic changes in intracortical inhibition precede, follow, outlast or recede ahead of functional changes in receptive field mapping and neural hyperexcitability. Nor is it clear whether the decline in GABA markers following peripheral injury arise from a reduced influence of PV-expressing fast-spiking interneurons, which synapse onto pyramidal neuron somata and have a well-recognized role in gating activity-dependent cortical plasticity, or instead from other GABA neuron subtypes, which can synapse onto other GABA neurons and exert a net excitatory effect on network activity (Isaacson and Scanziani, 2011). Here, we address these unanswered questions by measuring changes in PV-mediated intracortical inhibition alongside hyperexcitability and receptive field plasticity from individual regular spiking (RS) putative pyramidal neurons over a several month period surrounding varying degrees of auditory nerve damage. These findings highlight a rapid loss and recovery in PV-mediated inhibition that may compensate for a sudden drop in afferent drive following cochlear afferent loss to support a progressive recovery of sensory processing from spared nerve fibers.

Results

We implanted optetrode assemblies into the primary auditory cortex (A1) of adult PV-Cre:Ai32 mice (Figure 1a) to isolate regular spiking (RS) putative pyramidal neurons based on their spike shape and estimate the strength of local inhibition on RS units during optogenetic activation of PV–expressing GABA neurons (Figure 1b–c). The mechanical stability of the optetrode assembly combined with the thin, flexible tetrode wires allowed us to isolate single units (Figure 1d) and hold them for many weeks, according to a conservative, objective statistical standard (Figure 1e–f). With this approach, we could measure auditory responsiveness and local inhibitory tone from individual RS neurons in awake, head-fixed adult mice over a 7–8 week period surrounding varying degrees of auditory nerve injury (Figure 1g). Although we could occasionally isolate PV-expressing FS interneurons on our tetrodes (Figure 1b–c), we could rarely hold these neurons for more than a single recording session (refer to Figure 1—figure supplement 1). Therefore, all descriptions of long-term changes in single unit response properties were limited to RS, putative pyramidal neurons.

Figure 1. Approach for chronic single unit recordings and optogenetic activation in mouse A1.

(A) Immunolabeling of PV neurons in A1 with co-localization of EYFP reporter in Pv-Cre:Ai32 transgenic mice. DAPI labels cell nuclei. Scale bar = 15 µm. (B) Spike raster plots illustrating that optogenetic activation of fast spiking (FS) PV+ units (black, top) inhibits regular spiking (RS) units (gray, bottom). Right, spike waveforms for the RS and FS units. Arrowheads denote spike peak and trough. Scale bars, 0.5 ms and 50 µV. (C) Scatter plot showing the bimodal distribution of peak-tough amplitude and timing differences across all RS (gray) and FS (black) units. (D) A random sub-sampling of spike waveforms recorded over 53 days from a single wire of a tetrode projected down into the first three principal components (PC). (E) Spike waveforms from the two units identified in (D) across all recording sessions, color-coded and superimposed chronologically. Waveforms for random units were selected at random from all simultaneously recorded units. (F) The variability in the actual unit waveforms, estimated as the sum of squared errors (SSE), is significantly less than randomly shuffled units (p<0.001, f(1)=814.73, mean ± SEM). (G) Experimental design.

DOI: http://dx.doi.org/10.7554/eLife.21452.002

Figure 1.

Figure 1—figure supplement 1. Long-term tetrode recordings from isolated single units is feasible with RS neurons, but not FS neurons.

Figure 1—figure supplement 1.

(A) As per Figure 1E, single unit spike waveforms across all recording sessions, are shown for a representative RS unit as well as two representative FS units. The RS unit waveform is consistent across the 7–8-week recording period, while the FS unit waveforms are lost after a few recording sessions. (B) Cumulative probability functions depict the number of consecutive recording sessions before each RS and FS unit was lost in noise-exposed and ouabain-treated mice. (C) For the analysis provided in the main text, we only analyzed units that were present in at least 75% of all recording sessions for a given mouse. None of the FS units we recorded met this criterion. Thus, we were unable to contrast changes in the FS neurons themselves with changes reported in the main text for RS neurons or PV-mediated inhibition.
DOI: http://dx.doi.org/10.7554/eLife.21452.003
Open in a new tab

Most forms of cochlear injury induce an intractable set of changes to the auditory nerve as well as sensory and non-sensory cells within the cochlea. In cases of widespread cochlear damage, it is impossible to attribute abnormal cortical tuning to a central plasticity versus abnormal cochlear filtering or amplification. We circumvented this problem by selectively eliminating afferent nerve fibers without permanently affecting cochlear mechanics using a well-characterized noise exposure protocol that lesions approximately 50% of high-frequency afferent nerve synapses in the 16–45 kHz region of the cochlea without damaging hair cells (n = 4, Figure 2a) (Wan et al., 2014; Kujawa and Liberman, 2009). This protocol induces only a temporary elevation of the auditory brainstem response (ABR) and distortion product otoacoustic emission (DPOAE) thresholds, but a permanent reduction in ABR wave 1 amplitude, which has been shown to reflect the permanent elimination of high-frequency cochlear afferent synapses (Figure 2b, the reader is referred to the figure legends and Supplementary file 1 for most statistical reporting) (Liberman and Kujawa, 2017).

Figure 2. Moderate loss of high-frequency auditory nerve fibers induces striking – but partially reversible - receptive field reorganization and inhibition.

Figure 2.

Open in a new tab

(A) Adult mice (16 weeks) were exposed to octave-wide noise at 98 dB SPL for 2 hr, which has been shown to eliminate approximately 50% of auditory nerve synapses onto inner hair cells in the 16–45 kHz region of the cochlea (Wan et al., 2014; Kujawa and Liberman, 2009). (B) Mean ± SEM ABR (left) and DPOAE thresholds (middle) were temporarily elevated 2 days after noise exposure but recover fully by 30 days. ABR wave 1b amplitude reflects the permanent loss of Type-I spiral ganglion nerve fibers (right). (C) A frequency response area (FRA) from a single unit recorded in the high-frequency zone of the A1 tonotopic map rapidly assumed a low-frequency FRA hours after noise exposure but shifted back towards baseline tuning over the ensuing month. (D–E) Rasters depict spiking from the same RS unit evoked by broadband noise at varying sound levels (D) or inhibited by optogenetic activation of neighboring PV+ neurons (E). Note temporary threshold shift and change in inhibition strength. (F) Quantification approach illustrated from a different single unit recorded before (black) and after (red) noise exposure. (G) Mean ± SEM (n = 208 units) values for each response property illustrated in (F). Data for each unit are normalized to the mean value measured during the 3-day period prior to noise exposure. Red symbols are used to visually highlight the rapid plasticity occurring during the first 5 days following nerve damage. Asterisks = post-hoc pairwise comparison between identified groups (B) or to baseline value (G) p<0.05 after correcting for multiple comparisons.

DOI: http://dx.doi.org/10.7554/eLife.21452.004

We implanted optetrode assemblies into the 32 kHz region of the A1 tonotopic map and recorded RS units with high-frequency receptive fields, low-intensity thresholds to broadband noise stimulation and strong feedforward inhibition from PV neurons (Figure 2c–e, top row). Several hours after noise exposure that affects high-frequency regions of the cochlea, the same unit exhibited elevated response thresholds for a broadband noise burst, slightly reduced PV-mediated inhibition and the appearance of a new, well-tuned low-frequency receptive field that was absent just hours earlier (Figure 2c–e, second row). We quantified the invasion of spared, low-frequency inputs and the eventual restoration of the original receptive field by measuring daily changes in frequency response area (FRA) overlap with the baseline FRA as well as tone-evoked firing rates from the spared (4–6 kHz) and partially denervated (32–48 kHz) cochlear regions (Figure 2f, left). We compared these metrics of receptive field reorganization against the typical biomarkers of hyperexcitability in the auditory pathway: decreased response thresholds, increased spontaneous firing rate and increased gain, defined here as the slope of the stimulus input-output function (Figure 2f, middle). Finally, we related these markers of RS unit hyperexcitability and receptive field plasticity to changes in the strength of PV-mediated inhibition (Figure 2f, right; n = 208).

Several hours after noise exposure, firing rates evoked by stimulation of spared, low-frequency cochlear regions had nearly doubled (increased by 96.7%, p=0.04) with a small, but significant decrease in inhibition (−9.8%, p=0.02) and no significant change in gain or spontaneous rate (Figure 2g). Six to ten days after noise exposure, inhibition had decreased by 40% (p<0.0001), receptive fields remained focused on spared regions of the cochlea (p<0.01) and spontaneous firing rate was significantly elevated (p<0.05). Central gain, spontaneous firing rate and frequency tuning all shifted back towards baseline values as response thresholds in the cochlea, ABR and cortical neurons recovered to pre-exposure levels. By the time of the final recording session, approximately 7 weeks after noise exposure, reduced PV-mediated inhibition (−19%, p<0.001) was the only enduring compensatory plasticity marker for the putative loss of high-frequency auditory nerve synapses.

Exposure to ototoxic drugs can cause a sudden and extreme loss of afferent signaling from the cochlea. Would homeostatic adjustments support as complete a recovery of function following a massive loss of auditory nerve fibers? We addressed this question by applying ouabain bilaterally to the cochlear round window membrane (Figure 3a). In mice, ouabain selectively and evenly eliminates >95% of Type-I spiral ganglion afferent neurons across the frequency map, without affecting other types of sensory and non-sensory cells in the cochlea or auditory nerve (Yuan et al., 2014). Bilateral ouabain treatment virtually eliminated the ABR without any adverse DPOAE effects in all mice (n = 6, Figure 3b), but, interestingly, was associated with a slow, partial recovery of sound-evoked cortical spiking in half the mice (Figure 3c) but an unremitting loss of auditory responsiveness in the other half (Figure 3d).

Figure 3. Gradual and variable recovery of auditory processing following a massive, bilateral loss of cochlear afferent nerve fibers can be predicted from early changes in inhibitory strength.

Plotting conventions follow Figure 2. (A) Ouabain was applied bilaterally to the round window membrane. (B) Mean ± SEM ABR thresholds were substantially elevated 30 days after ouabain application (left) but DPOAE thresholds were unaffected, indicating normal outer hair cell function (middle). ABR wave1b was virtually eliminated after ouabain (right). Darker and lighter shading represent measurements from mice that recovered cortical sound thresholds versus mice that did not, respectively. (C–D) Rasters document changes in noise-evoked spiking and laser-induced inhibition, respectively, from single A1 RS units recorded over a 53-day period from a mouse that eventually recovered function (C) and a mouse that did not recover function (D). (E–H) Mean ± SEM noise-evoked thresholds (E), rate-level function gain (F), spontaneous firing rate (G) and inhibition strength (H) for all units recorded from three mice that recovered (top row, n = 156), three mice that did not recover (middle row, n = 169) or three mice that underwent a sham surgery (bottom row, n = 156). Asterisks = post-hoc pairwise comparison between identified groups (B) or to baseline value (E–H) p<0.05 after correcting for multiple comparisons.

DOI: http://dx.doi.org/10.7554/eLife.21452.005

Figure 3.

Figure 3—figure supplement 1. ABR wave 1 amplitude scales with the degree of auditory nerve damage, but does not vary systematically between mice that recover cortical sound thresholds versus those that do not.

Figure 3—figure supplement 1.

The amplitude of ABR wave 1b arises directly from the auditory nerve compound action potential and, provided DPOAE thresholds are normal (as is the case here), serves as a useful proxy for the surviving number of afferent nerve fibers onto inner hair cells. The ABR wave 1 amplitude is presented alongside the mean auditory response threshold measured in A1 single units for each mouse subjected to auditory nerve damage. ABR wave 1 amplitude does not predict the eventual recovery of sound processing in mice subjected to massive auditory nerve damage with ouabain.
DOI: http://dx.doi.org/10.7554/eLife.21452.006
Open in a new tab

When DPOAE thresholds are normal, ABR wave 1b amplitude is an accurate proxy for the number of cochlear afferent synapses (Wan et al., 2014; Kujawa and Liberman, 2009; Liberman and Kujawa, 2017). However, the ABR wave 1b amplitude was not significantly different between mice that recovered auditory processing following ouabain treatment versus those that did not (unpaired t-test, p>0.4; see also Figure 3—figure supplement 1). Whereas our estimate of peripheral denervation bore no relationship to the mode of cortical recovery, early dynamics in inhibitory tone and spontaneous firing rate were linked to the recovery of cortical processing. RS units that eventually recovered auditory thresholds exhibited a transient spike in spontaneous firing rate, a sustained increase in central gain and a steep drop in PV-mediated inhibition that began to recover as high-threshold sound-evoked activity returned (Figure 3e–h, top row, n = 156). By contrast, in RS units that never recovered auditory sensitivity, changes in spontaneous rate were comparatively sluggish, while inhibitory strength declined gradually and monotonically (Figure 3e–h, middle row, n = 169). This stark dichotomy in the cortical response to nerve injury was shared among all units recorded from a given mouse; if one unit recovered auditory responsiveness, all units recovered auditory responsiveness (Figure 4). As a negative control in a separate group of mice (n = 3), we applied sterile water to the round window membrane rather than ouabain and confirmed that cortical response properties were stable over the recording period in mice with intact auditory nerves (Figure 3e–h, bottom row, n = 156).

Figure 4. Early changes in PV-mediated inhibitory strength predict eventual recovery of cortical sound processing after nerve damage.

(A–B) PV-mediated inhibition strength (A) and change in spontaneous firing rate (B) timelines for sound exposed, ouabain recovered and ouabain non-recovered mice (red, blue and teal accordingly, plotted for visualization purposes as a fourth order polynomial fit to the data presented Figures 2 and 3). (Middle and Right) Correlation between broadband noise threshold at day 50 and PV-mediated inhibition (A) or spontaneous firing rate change (B) at days 6–10 for ouabain-treated (middle) and sound-exposed mice (right, Spearman’s correlation coefficient, Rs). Each symbol represents a single RS unit. The shape and color correspond to a particular group, as described above. The shading corresponds to a particular mouse within that group.

DOI: http://dx.doi.org/10.7554/eLife.21452.007

Figure 4.

Figure 4—figure supplement 1. Auditory cortex units show robust PV-mediated inhibition and low-threshold sensory responses prior to nerve damage, both of which are unrelated to the eventual response thresholds after nerve damage.

Figure 4—figure supplement 1.

(A) Distribution of best frequency values for primary auditory cortex units recorded before nerve damage. RS units recorded in noise-exposed mice were targeted to a high-frequency region of the tonotopic map, in the zone of putative cochlear denervation. (B–C) Histogram of PV-mediated inhibition strength (B) and response threshold (C) distributions prior to auditory nerve damage. (D) The strength of PV-mediated inhibition and auditory response thresholds are not correlated with each other before nerve damage (Pearson’s correlation coefficient). (E–F) PV-mediated inhibition strength and auditory response thresholds after nerve damage cannot be predicted from the strength of PV-mediated inhibition before nerve damage (Pearson’s correlation coefficient).
DOI: http://dx.doi.org/10.7554/eLife.21452.008
Open in a new tab

These findings highlight the coordination of short- and long-term changes in adult A1 following peripheral nerve injury. Spontaneous rate, central gain and inhibition all scaled with the degree of nerve damage (Figure 3f–h versus Figure 2g, Figure 4), but varied substantially both over time and between mice. Restored auditory sensitivity was associated with a sharp loss and partial recovery of inhibition from PV interneurons, where the degree of eventual recovery could be predicted from changes in inhibition ~1 week after nerve damage (Figure 4a). Not only was PV-mediated inhibition a predictor of sensory recovery for both mild and severe peripheral damage, but also it was a better predictor than were modifications in spontaneous activity (Figure 4b) or central gain, the biomarkers most commonly employed to assess central auditory plasticity following sensorineural hearing loss (refer to Figure 4—figure supplement 1 for an expanded analysis of predictors of recovery).

Discussion

In this study, we tracked a plasticity in PV-mediated inhibition and single neuron responsiveness in the primary auditory cortex of adult mice following a sudden loss of input from the auditory nerve. Just hours after noise exposure, we observed the appearance of new, low-threshold receptive fields tuned to undamaged regions of the cochlea, accompanied by a small, significant drop in PV-mediated inhibition. Over the ensuing weeks, as cochlear thresholds and cortical receptive field tuning returned to baseline conditions, we noted only a brief spike in traditional biomarkers of central auditory hyperactivity, such as increased spontaneous firing rate and increased central gain. By contrast, PV-mediated inhibition remained reduced through our final recording session and was the sole physiological indicator for the permanent, presumed loss of high-frequency cochlear synapses. Importantly, this is not to say that auditory response normalization was mediated by disinhibition; our observations are correlative and do not speak to the necessity nor sufficiency of changes in PV networks for functional recovery. Compensation for peripheral damage could have arisen from post-synaptic changes in GABA receptors in the RS principal neurons recorded here (Sarro et al., 2008) or from presynaptic changes in the GABA neurons themselves (Marik et al., 2014; Keck et al., 2011; Hengen et al., 2013), as well as many other homeostatic plasticity mechanisms in the RS neurons including upward scaling of glutamatergic synapses (O'Brien et al., 1998; Turrigiano et al., 1998), changes in extrasynaptic GABA or glutamate receptors that mediate tonic currents (Chen et al., 2010; Fleming et al., 2011; Sametsky et al., 2015), or changes in intrinsic membrane response properties (Turrigiano et al., 1994; Karmarkar and Buonomano, 2006; Li et al., 2015).

In classic and contemporary studies of adult cortical plasticity following deafferentation, fine-scale reorganization varied widely between local subnetworks of RS neurons and the completeness of global remapping was negatively correlated with the spatial extent of peripheral lesions (Robertson and Irvine, 1989; Merzenich et al., 1984; Barnes et al., 2015; Keck et al., 2008). Here, we report that the mode of cortical sensory recovery (complete, partial or none) was roughly shared between all RS units recorded in a given mouse and did not strictly depend on the estimated auditory nerve damage. Unlike focal ablations of the retina, basilar membrane or skin surface, cochlear ouabain treatment eliminates 95% of primary afferent synapses without affecting sensory transduction mechanisms (Chambers et al., 2016; Yuan et al., 2014). This provided us with an uncommon opportunity to track the restoration of sensory responses transmitted through the small fraction of surviving nerve fibers, rather than the typical approach of describing competitive reorganization from ectopic, spared inputs neighboring the lesion. As compared to the rapid unmasking of excitatory inputs from neighboring undamaged regions of the sensory organ, which occurs within hours (Figure 2, see also [Gilbert and Wiesel, 1992; Calford and Tweedale, 1988]), the recovery of function for spared afferent inputs interspersed within a damaged portion of the nerve, if it occurred at all, unfolded over several weeks. The variable modes of recovery between mice with carefully matched nerve damage may relate to earlier observations that cortical sensory reorganization is not only determined by the pattern and density of peripheral innervation, but may also be directed by activity-dependent differences that are orchestrated through behavioral use (Clark et al., 1988; Xerri et al., 1998; Polley et al., 1999).

Following either type of nerve damage, the cortical networks that recovered function were those that matched the sudden drop in bottom-up excitatory drive with a rapid dip in PV-mediated inhibition. Precisely balanced excitation and inhibition enables a wider dynamic range of sensory information coding (Isaacson and Scanziani, 2011; Shadlen and Newsome, 1998; Pouille et al., 2009; Zhou et al., 2014) and supports adaptive plasticity without forsaking network stability (Vogels et al., 2011; Froemke et al., 2013; Hellyer et al., 2016; Barrett et al., 2016). These findings suggest that self-regulating PV circuits may not only play an important role in rebalancing network activity in response to reduced afferent drive during developmental critical periods (Xue et al., 2014; Yazaki-Sugiyama et al., 2009), but in the adult cortex following sensory nerve damage as well.

Materials and methods

Animals and cochlear denervation

All procedures were approved by the Animal Care and Use Committee at the Massachusetts Eye and Ear Infirmary and followed guidelines established by the National Institutes of Health for the care and use of laboratory animals. Subjects included 13 PV-Cre:Ai32 mice of either sex (a cross between B6;129P2-Pvalbtm1(cre)Arbr/J and Ai32 (RCL-ChR2(H134R)/EYFP), Jackson Laboratory), aged 16 weeks at the time of optetrode implantation.

Bilateral auditory nerve damage

Acoustic exposure

The acoustic over-exposure stimulus was an octave band of noise (8–16 kHz) presented at 98 dB sound pressure level (SPL) for 2 hr. During exposures, animals were awake and unrestrained within a 12 × 10 cm, acoustically transparent cage. The cage was suspended directly below the horn of the sound-delivery loudspeaker in a small, reverberant chamber. Noise calibration to target SPL was performed immediately before each exposure session.

Ouabain

Selective elimination of Type-I spiral ganglion neurons was achieved by applying a 1 mM solution of ouabain octahydrate (Sigma) and sterile water to the left and right round window niche, as described previously (Yuan et al., 2014). Animals were anesthetized with ketamine (120 mg/kg) and xylazine (12 mg/kg), with half the initial ketamine dose given as a booster when required. The connective tissue and underlying muscle were blunt dissected and held away from the bulla with retractors. A small opening was made in the bulla with the tip of a 28.5-gauge needle. The exposed round window niche was filled with 1–2 µL of the ouabain solution using a blunted needle. Ouabain was reapplied five more times at 15 min intervals, wicking the remaining solution away with absorbent paper points before each application. For control experiments, sterile water was placed on the cochlear round window using an identical procedure. Measurements of the auditory brainstem response (ABR) and distortion product otoacoustic emission (DPOAE, see subsequent Experimental Procedures) were made during the ouabain application to confirm functionality and ABR threshold shift without changes in DPOAE thresholds or amplitudes. Additional ouabain was applied, as necessary, until the ABR threshold at 16 kHz was greater than 55 dB sound pressure level (SPL). The incision was sutured and the mouse was given an analgesic (Buprenex, 0.05 mg/kg) before being transferred to a heated recovery chamber.

Cochlear function tests

Mice were anesthetized with ketamine and xylazine (as above), and placed on a homeothermic heating blanket during testing. ABR stimuli were 5 ms tone pips (8, 16 and 32 kHz pure tones) with a 0.5 ms rise-fall time delivered at 30 Hz. Intensity was incremented in 5 dB steps, from 20–80 dB SPL. ABR threshold was defined as the lowest stimulus level at which a repeatable waveform could be identified. DPOAEs were measured in the ear canal using primary tones with a frequency ratio of 1.2, with the level of the f2 primary set to be 10 dB less than f1 level, incremented together in 5 dB steps. The 2f1-f2 DPOAE amplitude and surrounding noise floor were extracted. DPOAE threshold was defined as the lowest of at least two continuous f2 levels, for which the DPOAE amplitude was at least two standard deviations greater than the noise floor. All treated animals underwent rounds of DPOAE and ABR testing before, 2 days and approximately 30 days after the procedure.

Immunohistochemistry

Mice were perfused with 0.01 M phosphate buffered saline (PBS) (pH = 7.4) followed by 4% paraformaldehyde in 0.01 M PBS. Brains were removed and stored in 4% paraformaldehyde for 12 hr before transferring to cryoprotectant (30% sucrose) for 48 hr. Sections (40 µm) were cut using a cryostat (Leica CM3050S). Sections were washed in PBS containing 0.1% Triton X-100 (three washes, 5 min each), incubated at room temperature in blocking solution (Super Block) for 6 min at room temperature and then incubated in primary antibody (PV 27 rabbit anti-parvalbumin, Swant,1:1000 dilution) at four degrees C overnight. The next day, slices were washed and incubated in secondary antibody (Alexa 647 goat anti-rabbit immunoglobulin G, Invitrogen, 1:200 dilution) for 1.5 hr at room temperature. Sections were mounted on gelatin-subbed glass slides (BBC Biochemical) and coverslipped. Fluorescence images were obtained with a confocal microscope (Leica).

Chronic optetrode implantation

Mice were brought to a surgical plane of anesthesia with ketamine/xylazine, as described above. Body temperature was maintained at 36.5°C with a homeothermic blanket system. Using a scalpel, a small craniotomy was centered over the right auditory cortex leaving the dura mater intact. The brain surface was covered with sterile ointment. Chronic implants consisted of a modified microdrive array (VersaDrive 4, Neuralynx Inc) containing 4 independently moveable, closely spaced tetrodes and an optic fiber (0.2 mm, Thorlabs, Inc.). The optic fiber was mounted into the microdrive assembly so that it rested several millimeters above brain the surface. Tetrodes were four twisted nichrome wires (12.5 μm in diameter, Stablohm 650 wire, California Fine Wire Company), electroplated with gold to reach an impedance of 0.5–1 MΩ (nanoZ and ADPT-NZ-VERSA adapter, Neuralynx Inc). Microdrive arrays were positioned over the right primary auditory cortex based on an initial mapping of tonotopic gradient orientation using independently moveable tungsten microelectrodes (Hackett et al., 2011). Tetrodes were lowered into the middle cortical layers to match the approximate depth of tungsten microelectrodes recordings. Silver wire was fixed in place atop the left frontal cortex to serve as a ground. A titanium head plate was affixed to the skull once the microdrive assembly was firmly in place with dental cement (C&B Metabond).

Neurophysiology data collection

Single unit recordings were made from awake, head-fixed mice. Mice were continuously video monitored during recording. Raw neural signals were digitized at 32-bit, 24.4 kHz (RZ5 BioAmp Processor; Tucker-Davis Technologies) and stored in binary format for offline analysis. The signal was bandpass filtered at 300–3000 Hz with a second-order Butterworth filter and movement artifacts were minimized through common mode rejection. Spike waveforms were extracted based on a threshold and sorted offline into single units using Wave_clus, a semiautomatic clustering algorithm (Quiroga et al., 2004). Sorts were performed on concatenated data files from all recording sessions. Neurons were considered continuously recorded if spike sorting from data across the entire experiment yielded a single unit by the following criteria: (1) Waveforms constituted a statistically isolated cluster when all waveforms from all recording sessions were considered; (2) Biophysical properties consistent with single units, such as an absolute refractory period, were met; (3) High uniformity in spike shape was maintained, as assessed by a comparison of the sum of squared errors (SSE) across days (fractional difference in amplitude squared); and (4) Consistently high signal to noise ratios.

The first recordings after auditory nerve injury were made approximately 2 hr after the end of noise exposure and approximately 14 hr after ouabain application (to allow for extra recovery from injectable anesthetics). Recordings were made every 1–4 days thereafter. We only included units that were active in at least 75% of recording sessions and whose firing rate was stable for the two baseline recording sessions prior to peripheral damage. After offline sorting the mean waveform of neurons was calculated, the tetrode wire on which the amplitude of the spikes was largest was used. The trough-to-peak interval and the peak trough ratio were calculated and units were classified as fast-spiking PV+ units or RS-putative excitatory units. The classification was confirmed by laser response in the pre-condition. All subsequent analyses were performed in MATLAB 2015a (MathWorks).

Acoustic stimuli

Stimuli were generated with a 24-bit digital-to-analog converter (National Instruments model PXI-4461). For DPOAE and ABR tests, as well as during electrode implant surgery, stimuli were presented via in-ear acoustic assemblies consisting of two miniature dynamic earphones (CUI CDMG15008–03A) and an electret condenser microphone (Knowles FG-23339-PO7) coupled to a probe tube. Stimuli were calibrated in the ear canal in each mouse before recording. In awake recordings, stimuli were presented via a free-field electrostatic speakers (Tucker-Davis Technologies) facing the left (contralateral) ear. Free-field stimuli were calibrated before recording with a wide-band ultrasonic acoustic sensor (Knowles Acoustics, model SPM0204UD5).

Neurophysiology data analysis

Rate-level functions

Broadband noise tokens (4–64 kHz, 0.1 s duration, 4 ms raised cosine onset/offset ramps) were presented at 0–80 dB SPL in 5 dB increments. Threshold was defined as the lowest of at least three continuous stimulus levels for which the response to sound was significantly higher than the spontaneous activity. Gain was defined as the relationship between sound level (input) and firing rate (output). The gain was measured as the slope of the linear fit of the initial rising phase from the rate-level function.

Frequency response areas

FRAs were delineated using pseudorandomly presented tone pips (50 ms duration, 4 ms raised cosine onset/offset ramps, 0.5–1 s intertrial interval) of variable frequency (4–48 kHz in 0.1 octave increments) and level (0–75 dB SPL in 5 dB increments). Each tone pip was repeated three times and responses to each iteration were averaged. The start of the spike collection window was set to be the point when the firing rate began to consistently exceed the spontaneous rate by at least 3 SD. The offset of the spike collection window was the first bin after the response decreased to less than 4 SD above the spontaneous rate. Additional details on spike windowing and FRA boundary determination are described elsewhere (Guo et al., 2012). Receptive field overlap was defined as the number of frequency–intensity combinations in the intersection of the reference and comparator FRA divided by the number of points contained within their union. The reference FRA was the average of the pre-damage FRAs for a given unit and the comparator was selected from each individual measurement time.

Optogenetic activation

Collimated blue light (473 nm, 500 ms duration, 12 mW) was generated by a laser (DPSS, LaserGlow Co.) and delivered to the brain surface via an optic fiber coupled to the tetrode assembly. Precise timing of the light pulse was controlled via a custom shutter system. The intensity of the blue laser was calibrated for each implant prior to the start of recording with a photodetector (Thorlabs, Inc.). Percentage of inhibition in excitatory cells was calculated as the percentage of 20 ms bins, during the 500 ms laser duration, that were significantly suppressed when compared to spontaneous firing rate (paired t-test corrected for multiple comparisons).

Statistical analysis

A one-way ANOVA or mixed design ANOVA was employed for determining statistical significance (Matlab). Post-hoc pairwise comparisons were corrected for multiple comparisons using the Bonferroni correction. All error bars are mean ± SEM. Pearson and Spearman correlation coefficients were used for correlations of normally and not normally distributed variables accordingly.

Acknowledgements

We thank Prof. Nao Uchida for assistance with optetrode fabrication and Dr. Ken Hancock for support with data collection software. This work was supported by an EMBO postdoctoral fellowship (JR), R01 DC009836 (DBP), a research grant for Autifony Therapeutics (DBP) and the Lauer Tinnitus Research Center (DBP).

Funding Statement

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Funding Information

This paper was supported by the following grants:

  • European Molecular Biology Organization Long term postdoctoral fellowship to Jennifer Resnik.

  • National Institute on Deafness and Other Communication Disorders RO1 DC009836 to Daniel B Polley.

Additional information

Competing interests

The authors declare that no competing interests exist.

Author contributions

JR, Conceptualization, Data curation, Formal analysis, Funding acquisition, Methodology, Writing—original draft, Writing—review and editing.

DBP, Conceptualization, Funding acquisition, Methodology, Writing—original draft, Project administration, Writing—review and editing.

Ethics

Animal experimentation: All procedures were approved by the Animal Care and Use Committee at the Massachusetts Eye and Ear Infirmary (protocol number 10-03-006) and followed guidelines established by the National Institutes of Health for the care and use of laboratory animals. All surgeries were performed under ketamine and xylazine, and every effort was made to minimize suffering.

Additional files

Supplementary file 1. Statistical reporting.

Detailed description of statistical analysis for all figures and figure supplements.

DOI: http://dx.doi.org/10.7554/eLife.21452.009

elife-21452-supp1.xlsx (19.4KB, xlsx)
DOI: 10.7554/eLife.21452.009

References

  1. Barnes SJ, Sammons RP, Jacobsen RI, Mackie J, Keller GB, Keck T. Subnetwork-Specific homeostatic plasticity in mouse visual cortex in vivo. Neuron. 2015;86:1290–1303. doi: 10.1016/j.neuron.2015.05.010. [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Barrett DG, Denève S, Machens CK. Optimal compensation for neuron loss. eLife. 2016;5:12454. doi: 10.7554/eLife.12454. [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Calford MB, Tweedale R. Immediate and chronic changes in responses of somatosensory cortex in adult flying-fox after digit amputation. Nature. 1988;332:446–448. doi: 10.1038/332446a0. [DOI] [PubMed] [Google Scholar]
  4. Chambers AR, Resnik J, Yuan Y, Whitton JP, Edge AS, Liberman MC, Polley DB. Central gain restores auditory processing following Near-Complete cochlear denervation. Neuron. 2016;89:867–879. doi: 10.1016/j.neuron.2015.12.041. [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Chen X, Shu S, Schwartz LC, Sun C, Kapur J, Bayliss DA. Homeostatic regulation of synaptic excitability: tonic GABA(A) receptor currents replace I(h) in cortical pyramidal neurons of HCN1 knock-out mice. Journal of Neuroscience. 2010;30:2611–2622. doi: 10.1523/JNEUROSCI.3771-09.2010. [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Clark SA, Allard T, Jenkins WM, Merzenich MM. Receptive fields in the body-surface map in adult cortex defined by temporally correlated inputs. Nature. 1988;332:444–445. doi: 10.1038/332444a0. [DOI] [PubMed] [Google Scholar]
  7. D'amour JA, Froemke RC. Inhibitory and excitatory spike-timing-dependent plasticity in the auditory cortex. Neuron. 2015;86:514–528. doi: 10.1016/j.neuron.2015.03.014. [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Darian-Smith C, Gilbert CD. Axonal sprouting accompanies functional reorganization in adult cat striate cortex. Nature. 1994;368:737–740. doi: 10.1038/368737a0. [DOI] [PubMed] [Google Scholar]
  9. Fleming TM, Scott V, Naskar K, Joe N, Brown CH, Stern JE. State-dependent changes in astrocyte regulation of extrasynaptic NMDA receptor signalling in neurosecretory neurons. The Journal of Physiology. 2011;589:3929–3941. doi: 10.1113/jphysiol.2011.207340. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Florence SL, Taub HB, Kaas JH. Large-scale sprouting of cortical connections after peripheral injury in adult macaque monkeys. Science. 1998;282:1117–1121. doi: 10.1126/science.282.5391.1117. [DOI] [PubMed] [Google Scholar]
  11. Froemke RC, Carcea I, Barker AJ, Yuan K, Seybold BA, Martins AR, Zaika N, Bernstein H, Wachs M, Levis PA, Polley DB, Merzenich MM, Schreiner CE. Long-term modification of cortical synapses improves sensory perception. Nature Neuroscience. 2013;16:79–88. doi: 10.1038/nn.3274. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Garraghty PE, LaChica EA, Kaas JH. Injury-induced reorganization of somatosensory cortex is accompanied by reductions in GABA staining. Somatosensory & Motor Research. 1991;8:347–354. doi: 10.3109/08990229109144757. [DOI] [PubMed] [Google Scholar]
  13. Gilbert CD, Wiesel TN. Receptive field dynamics in adult primary visual cortex. Nature. 1992;356:150–152. doi: 10.1038/356150a0. [DOI] [PubMed] [Google Scholar]
  14. Guo W, Chambers AR, Darrow KN, Hancock KE, Shinn-Cunningham BG, Polley DB. Robustness of cortical topography across fields, laminae, anesthetic states, and neurophysiological signal types. Journal of Neuroscience. 2012;32:9159–9172. doi: 10.1523/JNEUROSCI.0065-12.2012. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Hackett TA, Barkat TR, O'Brien BM, Hensch TK, Polley DB. Linking topography to tonotopy in the mouse auditory thalamocortical circuit. Journal of Neuroscience. 2011;31:2983–2995. doi: 10.1523/JNEUROSCI.5333-10.2011. [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Hellyer PJ, Jachs B, Clopath C, Leech R. Local inhibitory plasticity tunes macroscopic brain dynamics and allows the emergence of functional brain networks. NeuroImage. 2016;124:85–95. doi: 10.1016/j.neuroimage.2015.08.069. [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Hengen KB, Lambo ME, Van Hooser SD, Katz DB, Turrigiano GG. Firing rate homeostasis in visual cortex of freely behaving rodents. Neuron. 2013;80:335–342. doi: 10.1016/j.neuron.2013.08.038. [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Isaacson JS, Scanziani M. How inhibition shapes cortical activity. Neuron. 2011;72:231–243. doi: 10.1016/j.neuron.2011.09.027. [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Kaas JH, Krubitzer LA, Chino YM, Langston AL, Polley EH, Blair N. Reorganization of retinotopic cortical maps in adult mammals after lesions of the retina. Science. 1990;248:229–231. doi: 10.1126/science.2326637. [DOI] [PubMed] [Google Scholar]
  20. Karmarkar UR, Buonomano DV. Different forms of homeostatic plasticity are engaged with distinct temporal profiles. European Journal of Neuroscience. 2006;23:1575–1584. doi: 10.1111/j.1460-9568.2006.04692.x. [DOI] [PubMed] [Google Scholar]
  21. Keck T, Mrsic-Flogel TD, Vaz Afonso M, Eysel UT, Bonhoeffer T, Hübener M. Massive restructuring of neuronal circuits during functional reorganization of adult visual cortex. Nature Neuroscience. 2008;11:1162–1167. doi: 10.1038/nn.2181. [DOI] [PubMed] [Google Scholar]
  22. Keck T, Scheuss V, Jacobsen RI, Wierenga CJ, Eysel UT, Bonhoeffer T, Hübener M. Loss of sensory input causes rapid structural changes of inhibitory neurons in adult mouse visual cortex. Neuron. 2011;71:869–882. doi: 10.1016/j.neuron.2011.06.034. [DOI] [PubMed] [Google Scholar]
  23. Keck T, Keller GB, Jacobsen RI, Eysel UT, Bonhoeffer T, Hübener M. Synaptic scaling and homeostatic plasticity in the mouse visual cortex in vivo. Neuron. 2013;80:327–334. doi: 10.1016/j.neuron.2013.08.018. [DOI] [PubMed] [Google Scholar]
  24. Kilman V, van Rossum MC, Turrigiano GG. Activity deprivation reduces miniature IPSC amplitude by decreasing the number of postsynaptic GABA(A) receptors clustered at neocortical synapses. Journal of Neuroscience. 2002;22:1328–1337. doi: 10.1523/JNEUROSCI.22-04-01328.2002. [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Kujawa SG, Liberman MC. Adding insult to injury: cochlear nerve degeneration after "temporary" noise-induced hearing loss. Journal of Neuroscience. 2009;29:14077–14085. doi: 10.1523/JNEUROSCI.2845-09.2009. [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Li L, Gainey MA, Goldbeck JE, Feldman DE. Rapid homeostasis by disinhibition during whisker map plasticity. PNAS. 2014;111:1616–1621. doi: 10.1073/pnas.1312455111. [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Li S, Kalappa BI, Tzounopoulos T. Noise-induced plasticity of KCNQ2/3 and HCN channels underlies vulnerability and resilience to tinnitus. eLife. 2015;4:07242. doi: 10.7554/eLife.07242. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Liberman MC, Kujawa SG. Cochlear synaptopathy in acquired sensorineural hearing loss: manifestations and mechanisms. Hearing Research. 2017 doi: 10.1016/j.heares.2017.01.003. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Marik SA, Yamahachi H, Meyer zum Alten Borgloh S, Gilbert CD, Borgloh MzumA. Large-scale axonal reorganization of inhibitory neurons following retinal lesions. Journal of Neuroscience. 2014;34:1625–1632. doi: 10.1523/JNEUROSCI.4345-13.2014. [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Merzenich MM, Kaas JH, Wall J, Nelson RJ, Sur M, Felleman D. Topographic reorganization of somatosensory cortical areas 3b and 1 in adult monkeys following restricted deafferentation. Neuroscience. 1983;8:33–55. doi: 10.1016/0306-4522(83)90024-6. [DOI] [PubMed] [Google Scholar]
  31. Merzenich MM, Nelson RJ, Stryker MP, Cynader MS, Schoppmann A, Zook JM. Somatosensory cortical map changes following digit amputation in adult monkeys. The Journal of Comparative Neurology. 1984;224:591–605. doi: 10.1002/cne.902240408. [DOI] [PubMed] [Google Scholar]
  32. Nahmani M, Turrigiano GG. Adult cortical plasticity following injury: recapitulation of critical period mechanisms? Neuroscience. 2014;283:4–16. doi: 10.1016/j.neuroscience.2014.04.029. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. O'Brien RJ, Kamboj S, Ehlers MD, Rosen KR, Fischbach GD, Huganir RL. Activity-dependent modulation of synaptic AMPA receptor accumulation. Neuron. 1998;21:1067–1078. doi: 10.1016/S0896-6273(00)80624-8. [DOI] [PubMed] [Google Scholar]
  34. Polley DB, Chen-Bee CH, Frostig RD. Two directions of plasticity in the sensory-deprived adult cortex. Neuron. 1999;24:623–637. doi: 10.1016/S0896-6273(00)81117-4. [DOI] [PubMed] [Google Scholar]
  35. Pouille F, Marin-Burgin A, Adesnik H, Atallah BV, Scanziani M. Input normalization by global feedforward inhibition expands cortical dynamic range. Nature Neuroscience. 2009;12:1577–1585. doi: 10.1038/nn.2441. [DOI] [PubMed] [Google Scholar]
  36. Qiu C, Salvi R, Ding D, Burkard R. Inner hair cell loss leads to enhanced response amplitudes in auditory cortex of unanesthetized chinchillas: evidence for increased system gain. Hearing Research. 2000;139:153–171. doi: 10.1016/S0378-5955(99)00171-9. [DOI] [PubMed] [Google Scholar]
  37. Quiroga RQ, Nadasdy Z, Ben-Shaul Y. Unsupervised spike detection and sorting with wavelets and superparamagnetic clustering. Neural Computation. 2004;16:1661–1687. doi: 10.1162/089976604774201631. [DOI] [PubMed] [Google Scholar]
  38. Rasmusson DD. Reorganization of raccoon somatosensory cortex following removal of the fifth digit. The Journal of Comparative Neurology. 1982;205:313–326. doi: 10.1002/cne.902050402. [DOI] [PubMed] [Google Scholar]
  39. Robertson D, Irvine DR. Plasticity of frequency organization in auditory cortex of guinea pigs with partial unilateral deafness. The Journal of Comparative Neurology. 1989;282:456–471. doi: 10.1002/cne.902820311. [DOI] [PubMed] [Google Scholar]
  40. Sametsky EA, Turner JG, Larsen D, Ling L, Caspary DM. Enhanced GABAA-Mediated tonic inhibition in auditory thalamus of rats with behavioral evidence of tinnitus. Journal of Neuroscience. 2015;35:9369–9380. doi: 10.1523/JNEUROSCI.5054-14.2015. [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Sarro EC, Kotak VC, Sanes DH, Aoki C. Hearing loss alters the subcellular distribution of presynaptic GAD and postsynaptic GABAA receptors in the auditory cortex. Cerebral Cortex. 2008;18:2855–2867. doi: 10.1093/cercor/bhn044. [DOI] [PMC free article] [PubMed] [Google Scholar]
  42. Scholl B, Wehr M. Disruption of balanced cortical excitation and inhibition by acoustic trauma. Journal of Neurophysiology. 2008;100:646–656. doi: 10.1152/jn.90406.2008. [DOI] [PubMed] [Google Scholar]
  43. Seki S, Eggermont JJ. Changes in spontaneous firing rate and neural synchrony in cat primary auditory cortex after localized tone-induced hearing loss. Hearing Research. 2003;180:28–38. doi: 10.1016/S0378-5955(03)00074-1. [DOI] [PubMed] [Google Scholar]
  44. Shadlen MN, Newsome WT. The variable discharge of cortical neurons: implications for connectivity, computation, and information coding. Journal of Neuroscience. 1998;18:3870–3896. doi: 10.1523/JNEUROSCI.18-10-03870.1998. [DOI] [PMC free article] [PubMed] [Google Scholar]
  45. Turrigiano G, Abbott LF, Marder E. Activity-dependent changes in the intrinsic properties of cultured neurons. Science. 1994;264:974–977. doi: 10.1126/science.8178157. [DOI] [PubMed] [Google Scholar]
  46. Turrigiano GG, Leslie KR, Desai NS, Rutherford LC, Nelson SB. Activity-dependent scaling of quantal amplitude in neocortical neurons. Nature. 1998;391:892–896. doi: 10.1038/36103. [DOI] [PubMed] [Google Scholar]
  47. Vogels TP, Sprekeler H, Zenke F, Clopath C, Gerstner W. Inhibitory plasticity balances excitation and inhibition in sensory pathways and memory networks. Science. 2011;334:1569–1573. doi: 10.1126/science.1211095. [DOI] [PubMed] [Google Scholar]
  48. Wan G, Gómez-Casati ME, Gigliello AR, Liberman MC, Corfas G. Neurotrophin-3 regulates ribbon synapse density in the cochlea and induces synapse regeneration after acoustic trauma. eLife. 2014;3:03564. doi: 10.7554/eLife.03564. [DOI] [PMC free article] [PubMed] [Google Scholar]
  49. Xerri C, Merzenich MM, Peterson BE, Jenkins W. Plasticity of primary somatosensory cortex paralleling sensorimotor skill recovery from stroke in adult monkeys. Journal of Neurophysiology. 1998;79:2119–2148. doi: 10.1152/jn.1998.79.4.2119. [DOI] [PubMed] [Google Scholar]
  50. Xue M, Atallah BV, Scanziani M. Equalizing excitation-inhibition ratios across visual cortical neurons. Nature. 2014;511:596–600. doi: 10.1038/nature13321. [DOI] [PMC free article] [PubMed] [Google Scholar]
  51. Yamahachi H, Marik SA, McManus JN, Denk W, Gilbert CD. Rapid axonal sprouting and pruning accompany functional reorganization in primary visual cortex. Neuron. 2009;64:719–729. doi: 10.1016/j.neuron.2009.11.026. [DOI] [PMC free article] [PubMed] [Google Scholar]
  52. Yang S, Su W, Bao S. Long-term, but not transient, threshold shifts alter the morphology and increase the excitability of cortical pyramidal neurons. Journal of Neurophysiology. 2012;108:1567–1574. doi: 10.1152/jn.00371.2012. [DOI] [PMC free article] [PubMed] [Google Scholar]
  53. Yazaki-Sugiyama Y, Kang S, Câteau H, Fukai T, Hensch TK. Bidirectional plasticity in fast-spiking GABA circuits by visual experience. Nature. 2009;462:218–221. doi: 10.1038/nature08485. [DOI] [PubMed] [Google Scholar]
  54. Yuan Y, Shi F, Yin Y, Tong M, Lang H, Polley DB, Liberman MC, Edge AS. Ouabain-induced cochlear nerve degeneration: synaptic loss and plasticity in a mouse model of auditory neuropathy. Journal of the Association for Research in Otolaryngology. 2014;15:31–43. doi: 10.1007/s10162-013-0419-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
  55. Zhou M, Liang F, Xiong XR, Li L, Li H, Xiao Z, Tao HW, Zhang LI. Scaling down of balanced excitation and inhibition by active behavioral states in auditory cortex. Nature Neuroscience. 2014;17:841–850. doi: 10.1038/nn.3701. [DOI] [PMC free article] [PubMed] [Google Scholar]
eLife. 2017 Mar 21;6:e21452. doi: 10.7554/eLife.21452.012

Decision letter

Editor: Dwight E Bergles1

In the interests of transparency, eLife includes the editorial decision letter and accompanying author responses. A lightly edited version of the letter sent to the authors after peer review is shown, indicating the most substantive concerns; minor comments are not usually included.

Thank you for submitting your article "Coordinated homeostatic modifications enable cortical compensation for sensory nerve damage" for consideration by eLife. Your article has been favorably evaluated by Andrew King (Senior Editor) and three reviewers, one of whom is a member of our Board of Reviewing Editors. The reviewers have opted to remain anonymous.

The reviewers have discussed the reviews with one another and the Reviewing Editor has drafted this decision to help you prepare a revised submission.

Summary:

This manuscript describes an analysis of the behavior of regular spiking (RS) neurons in primary auditory cortex following two forms of peripheral injury to the cochlea. In a recent report from the same group (Chambers et al., 2016), they showed that the central auditory system can compensate for loss of sensory input, enabling cortical neurons to respond relatively normally to sounds after ~95% of Type I afferents were destroyed by applying ouabain to the round window. Here they examined the time course of changes in RS cell spontaneous activity, responsiveness and degree of inhibition by PV interneurons (using ChR2-mediated stimulation in mice engineered to express ChR2 only in PV interneurons: PV-Cre x Ai32 mice) following two forms of cochlear injury. They find that PV-mediated inhibition of RS cells is transiently reduced following noise-induced trauma, which recovers along with normal response properties of the neurons. In response to bilateral ouabain application, which induces more extensive damage to afferents, they find that some cells were able to partially recover responsiveness to sound. These cells exhibited a greater initial reduction in PV mediated inhibition (75-80% versus 30% inhibition) and correspondingly higher rates of spontaneous activity than RS cells that did not recover, suggesting that homeostatic changes in the strength of PV interneuron-mediated inhibition controls the recovery from these injuries. The design and analyses are thoughtful and well-executed, and the results are of importance to our general understanding of CNS plasticity. However, insight into the mechanisms responsible for the reduced strength of PV inhibition is limited, as hearing loss-induced changes in excitatory currents or voltage-gated properties of RS neurons, which were not assessed, could explain the apparent reduction of inhibitory gain. In addition, there is an uncertain relationship between the concept of homeostasis and the measurements of inhibitory strength which display complicated and dissimilar changes in recovered and non-recovered animals.

Essential revisions:

Correlation versus causation:

1) The main conclusion of the studies is that the initial reduction in PV interneuron-mediated inhibition predicts whether the cell will recover from the trauma. The authors provide correlative data in support of this argument, but do not show causation. The primary question is whether this effect is just one of a large number of changes in cortical circuits or is key to the recovery process. Several observations raise some concerns about this hypothesis. First, the amount of reduction in PV inhibition in the non-recovering RS cells following ouabain exposure (~35%) is similar (~40%) to that observed in the FS cells that recovered from noise-induced trauma. Second, on a cell-by-cell basis the inhibition strength in non-recovered animals does not appear to predict recovery – the data presented in Figure 3D show that there is a similar loss of inhibition in both cases and yet one cell partially recovered and one did not. Of note, the summary data for PV-mediated inhibition doesn't appear to match data illustrated in the figures (see panel 2E, day-3 and day 50, summarized in 2F, and compare to bottom panel of Figure 2G). Day 0 recordings were obtained 2-14 hours after noise exposure or ouabain application. Moreover, the example rasters suggest that GABAergic inhibition was completely unchanged on Day 0, whereas the summary data show a significant, but incomplete, reduction (except for the non-recovery group). It would be more intuitive for the reader to view examples that matched the summary data. More importantly, the time course of reduced inhibitory gain after each of these hearing loss protocols is not shown with great enough temporal resolution. In addition, it is not clear why the time course of inhibitory change differed between recovered and non-recovered groups. If all changes are homeostatic, then the kinetics should be consistent across all groups. Third, the authors note that all cells within an animal showed the same behavior (e.g. all partially recover or all fail to recover). Given the intrinsic variability among RS cells with regard to PV inhibition and stochastic changes following injury, one might predict that some cells would show recovery and others would not. These findings suggest that the rapid reduction in PV inhibition and increase in firing rate may not be key to inducing compensatory changes in cortical circuits. Indeed, the dramatic shift in responsiveness to low frequency sounds after noise exposure appears to occur before there is much change in inhibition, suggesting that there are other important sites of regulation. Without some direct manipulations of PV interneurons, the role of this decrease in PV inhibition remains speculative.

2) The interpretation seems to be that GABAergic changes emerge over the first 5 days. In contrast, there is a rapid unmasking of low frequency responsiveness (Figure 2C, top two panels) on Day 0. Since data were collected during a 12 hour interval on Day 0, and data were merged for the subsequent 5 days, the actual time course over which inhibitory strength declines is uncertain. By way of comparison to nerve transection experiments in NHPs, short latency (hours) changes to GABAergic transmission are observed, raising the possibility that there is a very rapid response to the manipulations that may not be revealed by the sampling resolution used in this study. Since the study seeks to characterize the temporal relationship between inhibitory homeostatic mechanisms and other measures of recovery following hearing loss, a better temporal resolution could be important to the overall interpretation. A similar question arises for the data presented in Figure 3.

3) Concerning the difference between recovered and unrecovered animals, perhaps there is a critical amount of survival that is required to promote reorganization, and a greater number of spiral ganglion cells survived in the animals that recovered. Are there within animal comparisons that could establish whether the level of damage correlates with the neurophysiological findings (e.g., recovered versus non-recovered)? Notably, the authors do not provide histological evidence that the degree of synapse loss induced by these manipulations in this strain of mice is similar to that reported by Kujawa and Lieberman (C57BL/6 versus CBA/Cj, which were not examined by Kujawa). This issue is important, as different strains of mice exhibit different sensitivities for noise induced hearing loss. Again, with reference to the NHP somatosensory literature, when animals receive amputations or dorsal rhizotomies that deafferent a whole region of sensory cortex, there are large areas that fail to reorganize. However, when a sufficient number of nerve fibers remain, then that is enough for the neuraxis to reorganize.

Mechanisms responsible for the reduced PV neuron mediated suppression of RS cell firing

1) The apparent reduction of PV-mediated inhibition (i.e., failure of a light-evoked PV cell activity to block spontaneous discharge) could be explained by an increase in tonic excitatory currents. That is, if tonic excitatory currents increased following hearing loss, then a normal level of inhibition would appear to be less efficacious at suppressing discharge. The interpretation of reduced inhibitory gain would be more convincing if there were data or analyses to confirm that excitatory gain did not change, at least during the first 5 days. However, in vitro and in vivo experiments in the context of homeostasis suggest that excitatory and voltage-gated currents do change with similar manipulations, and the increased spontaneous discharge rate observed in the current study is consistent with this idea.

2) The interpretation that inhibitory gain declines (Figure 2C and 2F) is based on the suppression of responses of RS neurons and could be explained by reduced FS cell excitability, reduced GABA release, or reduced GABAA receptors located postsynaptically. Are there observations that support any of these specific mechanisms? For example, were a sufficient number of FS cell waveforms recovered through spike sorting to determine how these cells responded to blue light pulses following hearing loss? If there are multiple mechanisms, do they occur over a similar time course? What happens to FS cell spontaneous firing rates following hearing loss? Since they are recurrently connected to pyramidal cells which, themselves, display increased spontaneous rates, one might expect an increase. Does the shift of RF after hearing loss apply for PV neurons as well?

Relationship to prior studies

1) The authors make the following statement in the Introduction: "It is unclear how these distinct biomarkers of cortical reorganization following injury -receptive field remapping, spontaneous rate increases, and central gain increases – are coordinated over the days and weeks following varying degrees of sensory nerve injury. Nor is it certain whether these plasticity processes precede or follow homeostatic regulation of intracortical inhibition."

There are many relevant publications investigating how similar outcome measures are coordinated during the somatosensory reorganization that accompanies peripheral and central nerve injury. Changes to inhibition and concomitant reorganization of receptive fields are generally evident as soon as recordings can be obtained (e.g., Turnbull and Rasmusson, Somatosens Mot Res. 7:365, 1990; Calford and Tweedale, Somatosens Mot Res. 8:249, 1991; Rasmusson et al., Somatosens Mot Res. 10:69, 1993). More recently, a brief period of inactivity was demonstrated to induce a short latency reduction of IPSP amplitude in cortex (Lu et al., PNAS 111:1616, 2014). In addition, the rapid unmasking of subthreshold receptive fields is known to be kinetically distinct from the NMDA receptor-dependent reorganization that takes place in the weeks that follow sensory dennervation, especially in NHP models of peripheral dennervation. Thus, there are likely to be many mechanisms that operate across different time courses. In fact, homeostatic synaptic responses (reduced inhibitory or increased excitatory gain) to deprivation may permit "silent" inputs to become suprathreshold through NMDAR-dependent plasticity. It would be helpful to place the current findings in the context of a very well-developed literature.

2) Concerning the title. The results show plasticity that begins soon after the manipulation and lasts for a long time. The question is whether all of this plasticity can be termed "homeostatic." Homeostasis covers many specific cellular mechanisms that take place after sensory deprivation or during learning. It is important to make it clear exactly what, where, and when you are referring to homeostasis (e.g., Are all of the observations, from day 0-50, a single homeostatic process?). Since the changes seem to occur proximal to the hearing loss, this issue could be addressed by specifying what is meant by the word "compensatory" and specifying what is meant by the term "homeostatic." The Abstract does refer to homeostasis as occurring "during the first days," but it is not clear that homeostatic mechanisms emerge over that long a period of time.

Additional issues

1) The authors state that noise exposure induces permanent loss of the ABR, assessed through wave 1 amplitude; however, Figure 2 shows almost 50% recovery by 30 days. Please revise or provide further justification for this statement.

2) The authors do not provide histological evidence that the degree of synapse loss induced by these manipulations in this strain of mice is similar to that reported by Kujawa and Lieberman (C57BL/6 versus CBA/Cj, which were not examined by Kujawa). This issue is important, as different strains of mice exhibit different sensitivities for noise induced hearing loss.

3) The authors indicate that "optetrode assemblies into the 32 kHz region of the A1 tonotopic map and recorded RS units with high-frequency receptive fields, low-intensity thresholds to broadband noise stimulation and strong feedforward inhibition from PV neurons." It would be helpful if the authors indicated what proportion of sampled cells exhibited these properties. Was there a correlation between the degree of PV-mediated inhibition and the response of the neurons to peripheral injury?

eLife. 2017 Mar 21;6:e21452. doi: 10.7554/eLife.21452.013

Author response

Essential revisions:

Correlation versus causation:

1) The main conclusion of the studies is that the initial reduction in PV interneuron-mediated inhibition predicts whether the cell will recover from the trauma. The authors provide correlative data in support of this argument, but do not show causation.

It seems there was a misunderstanding here. We never intended to claim that changes in PV-mediated inhibition were necessary or sufficient(i.e., causal) for the changes in RS firing properties. The primary motivation and innovation of our study was to apply a new approach to monitor day-by-day changes in PV-mediated inhibition while tracking long-term changes in functional recovery following sensory nerve damage. Our work covers new ground by combining 1) recordings from isolated principal neurons over a period of several months, 2) with single day resolution 3) in unanesthetized animals 4) before and after 5) varying degrees of 6) selective primary afferent lesions, 7) while concurrently measuring dynamics in intracortical inhibition from 8) a genetically isolated subtype of GABA neuron. Another important, novel element of this work is that it tracks the restoration of cortical processing for spared afferent fibers in the damaged portion of the nerve. The typical approach is to completely lesion a restricted region of the retina, skin surface or basilar membrane or else entirely transect a peripheral nerve. With these methods, cortical reorganization can only be studied from the perspective of a competitive remapping of viable regions of the peripheral receptor into deafferented zones of cortex. This has been described extensively in the literature. With our approach, we can contrast this competitive invasion (e.g., low-frequency unmasking in Figure 2) against the less-studied but clinically important question of how cortical neurons recover sensitivity to intact peripheral inputs within a damaged region of the nerve.

We submitted this work in a Short Report format (bearing in mind all the associated space limitations) because we thought it leveraged a few clever contemporary technologies to make a succinct and surprising correlative observation about cortical network recovery from varying degrees of peripheral injury. Indeed, the major advance of this study isn’t the technological innovation described above, but rather the striking, dynamic and variable expression of cortical plasticity that it allowed us to document. Of course, we also see the value of a full-length report that delves into the underlying mechanisms. We are pursuing this line of questioning in separate studies. We regret leaving the reviewers with the impression that we were claiming a mechanism and have carefully revised the manuscript to alleviate this confusion. For example, we now write, “Importantly, this is not to say that auditory response normalization was mediated by disinhibition; our observations are correlative and do not speak to the necessity nor sufficiency of changes in PV networks for functional recovery.”

We want to be unequivocal that we would have carried out our experiments differently, had we set out to test whether changes in PV-mediated inhibition caused varying patterns of functional recovery. Adding the experiments to test causality to this revision would have essentially amounted to adding an entirely new project that, as we argue below, would not have had a high chance of yielding interpretable results.

For starters, testing the contribution of other possible mechanisms raised by the reviewers such as changes in extra-synaptic (tonic) excitation or inhibition, changes in intrinsic properties or even sourcing inhibitory changes to the PV neurons themselves (i.e., presynaptic) versus putative pyramidal neurons (i.e., postsynaptic) would be best addressed with whole cell recordings, not extracellular recordings of action potentials. Of course, this would be very challenging with our approach to using adult (16 weeks) mice with complex past histories of nerve damage and unestablished trajectories of recovery. Providing a strong test of these mechanisms by measuring spikes alone is not feasible, and we recognized this going into the study. On the other hand, there is no way to record from single neurons for 50 days with a whole cell recording (50 minutes is hard enough). This study was focused on describing long-term changes with single day (and single cell) resolution.

As a second point, proving that changes in PV-mediated inhibition was necessary and sufficient for functional recovery would have required us to block and/or artificially introduce a rapid swing in PV signaling following injury on the timescale reported here (several days). We considered these experiments, but decided that they were unlikely to yield interpretable results. Technically, conventional opsins are optimally used on time scales of tens or hundreds of milliseconds, while step opsins or DREADDS are effective on a scale of tens of minutes. But no available method can continuously activate or silence a genetically targeted cell type for days, as would be required to recreate or block the PV dynamics described here without directly engaging homeostatic mechanisms or, worse yet, drastically changing intracellular pH and all the unwanted complications that come therein. Conceptually, those experiments would be likely to fail because PV neurons are reciprocally connected with recurrently interconnected networks of principal neurons. Any approach that purported to replace or introduce PV signaling by uniformly activating or silencing PV neurons would have to contend with the heterogeneous, non-linear effects that arise from not only removing/adding inhibition from principal neurons, but also the downstream effects of changing principal neuron outputs onto other principal neurons and back onto PV neurons. It just wouldn’t work for a densely interconnected local circuit like those found in primary sensory cortex (and the literature repeatedly bears this out). Such an approach might be possible, for example, if one were testing the necessity or sufficiency of a long-range modulatory input for a plasticity in the local cortical network. But that’s a separate issue.

We did state (and continue to state) that early changes in PV-mediated inhibition are predictive of later changes in functional recovery, but that, in itself, does not imply that these changes are causal. Knowing the depth of PV modulation in the early days following injury allows an observer to make an accurate prediction of how/whether functional networks will eventually recover sensory processing (e.g., Figure 4).

The primary question is whether this effect is just one of a large number of changes in cortical circuits or is key to the recovery process. Several observations raise some concerns about this hypothesis. First, the amount of reduction in PV inhibition in the non-recovering RS cells following ouabain exposure (~35%) is similar (~40%) to that observed in the FS cells that recovered from noise-induced trauma.

Without belaboring the points raised above, we agree with the reviewers that changes in PV circuits cannot be directly equated with changes in functional response properties. Other mechanisms are likely to contribute. The reviewers’ critiques made it clear to us that we did not do a very good job of communicating this point in the original version.

We have made extensive changes to the text and figures to ensure that we are communicating a clear and consistent message on this point in the revised manuscript. We have replaced all of our example units to ensure that they better represent the group trends, have reanalyzed all data to include the key time intervals shortly following injury and have completely overhauled the Introduction and Discussion sections. Yes, the reviewers are absolutely right that the degree of PV-mediated inhibition, measured at an arbitrary time point, cannot be equated with the degree of nerve damage. We emphasize that it is the depth of change (i.e., the first derivative) of PV-mediated inhibition shortly following nerve damage, and not the absolute change, that distinguishes mice that recover sound processing from those that do not. With that said, the changes in PV-mediated inhibition are distinct for all three groups, as shown in Figure 4.

Second, on a cell-by-cell basis the inhibition strength in non-recovered animals does not appear to predict recovery – the data presented in Figure 3D show that there is a similar loss of inhibition in both cases and yet one cell partially recovered and one did not. Of note, the summary data for PV-mediated inhibition doesn't appear to match data illustrated in the figures (see panel 2E, day-3 and day 50, summarized in 2F, and compare to bottom panel of Figure 2G). Day 0 recordings were obtained 2-14 hours after noise exposure or ouabain application. Moreover, the example rasters suggest that GABAergic inhibition was completely unchanged on Day 0, whereas the summary data show a significant, but incomplete, reduction (except for the non-recovery group). It would be more intuitive for the reader to view examples that matched the summary data.

The reviewers were correct to point out that our example units conveyed a different message than the statistical analysis of the group data. This was an error that we have remedied in the revised manuscript. In terms of the group data, our main findings remain unchanged: RS units in mice that recover auditory thresholds following bilateral ouabain show a transient loss and recovery of inhibition, whereas units from mice that do not recover show only a monotonic loss of inhibition (Figure 3H top vs. middle row). Further to this, on a cell-by-cell basis, the loss of inhibition 6-10 days after ouabain predicts whether the animal ultimately recovers or does not (Figure 4A).

We have selected all new example units that better represent these findings from a noise-exposed mouse (Figure 2C-E), a ouabain-treated mouse that recovered (Figure 3C) and a ouabain-treated mouse that did not recover auditory thresholds (Figure 3D). The data in panel 2F was chosen to illustrate our analysis strategy from an additional unit; those data do not come from the unit above. The figure legend has been revised to make this clear.

More importantly, the time course of reduced inhibitory gain after each of these hearing loss protocols is not shown with great enough temporal resolution.

We addressed this comment by reanalyzing all of our data and updating all statistical analyses to reflect a finer-grained analysis of changes over the first days following auditory nerve injury. All summary data in Figures 2 and 3 now present changes with day-by-day resolution within the first 5 days following noise exposure (Figure 2G) or ouabain/sham treatment (Figure 3E-H). Data points from the first five days are plotted in red, to enhance the visual contrast with the preceding and following recordings. The new analyses and plots show a small (~10%) but significant decrease in inhibition at the first time point after noise exposure (Figure 2G), a large, significant decrease in inhibition at the first time point after ouabain treatment in mice that recover function (Figure 3H, top) and no change in inhibition at the earliest time points in ouabain-treated mice that do not recover function (Figure 3H, middle) or in control mice treated with saline (Figure 3H, bottom).

In addition, it is not clear why the time course of inhibitory change differed between recovered and non-recovered groups. If all changes are homeostatic, then the kinetics should be consistent across all groups.

Yes, this is one of the more intriguing observations in the study. Though we have yet to identify the factor that enables the cortex to recover auditory processing (or, alternatively, the factor that thwarts this process), we have identified predictors for the eventual mode of recovery, as illustrated in Figure 4.

As suggested by the reviewers and described below, we performed due diligence and test all plausible sources available to us in our dataset to explain the status of recovery. Nearly all of these findings are negative. For example, we cannot explain the qualitative difference in inhibitory change between the Recovered and Non-Recovered group based on the completeness of auditory nerve damage as the reduction in ABR wave 1 amplitude is massive, selective to Type-I spiral ganglion neurons and equivalent between groups. Nor can we attribute the difference to changes in outer hair cell-based cochlear amplification, because the DPOAE thresholds are at equivalent and normative levels in both groups. The data described in Figures 2–4 suggest that individual differences in the PV circuit dynamics are associated with the mode of recovery following a similar injury. As the reviewers note and we now make patently clear in the revised manuscript, these observations are correlative but they set the stage for future studies that would to optogenetically commandeer long-range neuromodulatory inputs to “steer” the mode of recovery towards one or another functional endpoint.

We addressed this point in our revised Discussion and took care not to use the term “homeostatic” when specifically referring to changes in the Non-Recovered group.

Third, the authors note that all cells within an animal showed the same behavior (e.g. all partially recover or all fail to recover). Given the intrinsic variability among RS cells with regard to PV inhibition and stochastic changes following injury, one might predict that some cells would show recovery and others would not.

We made sure to cite the work in the revised Discussion that might lead one to this expectation. Beyond that, it is not clear how we can further address this comment. The data are the data. The observation that neurons from the same mouse change in a similar way would be less likely to occur by chance than the finding that changes were variable between neurons of a given mouse. The variability in RS unit response properties before and after nerve damage is now provided as a new figure, Figure 4—figure supplement 1. There is some variance in RS response properties but it does not predict the function outcome and RS units.

These findings suggest that the rapid reduction in PV inhibition and increase in firing rate may not be key to inducing compensatory changes in cortical circuits. Indeed, the dramatic shift in responsiveness to low frequency sounds after noise exposure appears to occur before there is much change in inhibition, suggesting that there are other important sites of regulation. Without some direct manipulations of PV interneurons, the role of this decrease in PV inhibition remains speculative.

This comment overlaps with many of the points raised above. We agree that the causal role of PV inhibition is speculative and make this limitation clear in the revised Discussion. The revised analysis shows that PV-mediated inhibition is significantly reduced (albeit only by ~10%) at the first time point, when the receptive field has shifted to low frequencies.

2) The interpretation seems to be that GABAergic changes emerge over the first 5 days. In contrast, there is a rapid unmasking of low frequency responsiveness (Figure 2C, top two panels) on Day 0. Since data were collected during a 12 hour interval on Day 0, and data were merged for the subsequent 5 days, the actual time course over which inhibitory strength declines is uncertain. By way of comparison to nerve transection experiments in NHPs, short latency (hours) changes to GABAergic transmission are observed, raising the possibility that there is a very rapid response to the manipulations that may not be revealed by the sampling resolution used in this study. Since the study seeks to characterize the temporal relationship between inhibitory homeostatic mechanisms and other measures of recovery following hearing loss, a better temporal resolution could be important to the overall interpretation. A similar question arises for the data presented in Figure 3.

Our previous plots did not show the finest temporal resolution possible during the early period of recovery. We now plot changes in all functional markers with single day resolution for all groups in Figures 2 and 3. The revised analysis reveals that feedforward inhibition from local PV networks is significantly reduced at the first time point in both of groups that recover sound processing. Importantly, a significant drop in PV-mediated inhibition is not found for another week in mice that fail to recover after massive nerve damage (and don’t change at all in sham-treated mice).

3) Concerning the difference between recovered and unrecovered animals, perhaps there is a critical amount of survival that is required to promote reorganization, and a greater number of spiral ganglion cells survived in the animals that recovered.

We explicitly compared cochlear afferent synapse survival with the degree of functional recovery in our prior study (Chambers et al., Neuron 2016, Figure 7D) and concluded that there was no obvious relationship. In an earlier study, we have explicitly compared ABR wave 1b amplitude to the number of surviving cochlear afferent synapses and determined that it can be used as a reliable proxy for afferent innervation of the inner hair cell both in normal and ouabain-treated ears (Yuan et al., JARO 2013, Figure 9). Here, we compare mean ABR wave 1b amplitude in ouabain-treated mice that recover versus those that do not and did not see any indication of a difference between Recovered and Non-Recovered mice (Figure 3B). Because of our prior study that explicitly compared cochlear synapse survival to cortical recovery and our original work demonstrating that wave 1b could be a proxy for cochlear afferent synapse innervation, we reasoned that there was not a strong rationale to process cochleaes in ouabain-treated mice.

However, to address the reviewers’ comments as completely as possible we have reanalyzed the data linking ABR wave 1b and refer the reader to the newly added Figure 3—figure supplement 1. This revised, more granular, analysis describes markers of cochlear denervation and functional recovery of auditory threshold from each mouse tested and the lack of correspondence between the auditory nerve compound action potential amplitude (i.e., wave 1b) and recovery of function at the level of the cortex is made more apparent.

Are there within animal comparisons that could establish whether the level of damage correlates with the neurophysiological findings (e.g., recovered versus non-recovered)?

Other than the correlations provided in Figure 4, no. However, we agree that it is important to show the features that fail to predict recovery, and not just the factors that do show an association. The peripheral predictors are provided in Figure 3—figure supplement 1, described above. The within-neuron predictors are now provided in the newly added Figure 4—figure supplement 1.

Notably, the authors do not provide histological evidence that the degree of synapse loss induced by these manipulations in this strain of mice is similar to that reported by Kujawa and Lieberman (C57BL/6 versus CBA/Cj, which were not examined by Kujawa). This issue is important, as different strains of mice exhibit different sensitivities for noise induced hearing loss.

That’s true, we did not dissect the cochleaes from these mice and perform quantitative immunolabeling on primary afferent synapses. To be fair, there are exceedingly few studies of cortical plasticity following peripheral denervation that recover and directly analyze the sensory receptor epithelium. We go further than most by quantifying the compound action potential from the partially denervated peripheral nerve (i.e., ABR wave 1b) and establishing that the damage is neural, rather than affecting the transduction apparatus itself (i.e., by measuring otoacoustic emissions). As described above, we have performed the cochlear anatomy studies in prior work with ouabain-based denervation and this work provides a justification to use ABR wave 1b amplitude as a proxy for the survival of cochlear afferent synapses onto inner hair cells.

Regarding cochlear synaptopathy with the moderate noise exposure protocol, we naturally had many discussions with Kujawa and Liberman (my neighbors at Mass. Eye and Ear) about the cochlear synaptopathy in other mouse strains. Though the reviewers are correct in stating that many published studies on selective cochlear synaptopathy with noise exposures that cause only a temporary threshold shift (TTS) have been performed on the CBA mouse strain, TTS and cochlear synaptopathy have, in fact, been described using C57BL6 and C67BL6 hybrid strains (e.g., Wan et al., eLife 2014). We include a subset of the data published in that paper (Author response image 1). The published work from our lab and our colleagues’ labs provide the requisite proof that ABR wave 1b amplitude can be used as a proxy for Type-I ganglion cell innervation of the inner hair cell both after ouabain-treatment or noise exposures that cause only a temporary threshold shift.

Author response image 1. Additional published data demonstrating the link between temporary ABR threshold shift, permanent wave 1 ABR amplitude loss and permanent loss of cochlear afferent synapses in a C57BL6 hybrid mouse strain.

Author response image 1.

Open in a new tab

Cochlear afferent synapses are quantified by measuring GluR2+ puncta on peripheral auditory nerve processes that oppose synaptic ribbons, a presynaptic release protein complex at the base of the inner hair cell.

DOI: http://dx.doi.org/10.7554/eLife.21452.010

Again, with reference to the NHP somatosensory literature, when animals receive amputations or dorsal rhizotomies that deafferent a whole region of sensory cortex, there are large areas that fail to reorganize. However, when a sufficient number of nerve fibers remain, then that is enough for the neuraxis to reorganize.

This is an important point that we now address our revised Discussion. For starters, one of the main points of our paper is that cortical sound processing recovers more completely, with less drastic short-term compensatory changes, after moderate auditory nerve injury (noise exposure) than for extreme nerve injury. This is consistent with the larger literature on cortical plasticity following deafferentation mentioned by the reviewers.

As a second point, as described above, there is no evidence in our dataset (namely, ABR wave1b amplitude and DPOAE threshold) or in our previous study (Chambers et al., Neuron 2016) that the degree of peripheral injury can explain how well ouabain-treated mice recover auditory processing or not. This isn’t to say that a peripheral factor cannot explain the mode of recovery, just that we looked at the most obvious markers and nothing fits the pattern of cortical recovery.

Thirdly, it is important to point out that dorsal rhizotomy is a complete deafferentation, whereas both modes of injury employed here spare a subset of afferent synapses within the zone of injury. This is a key difference that distinguishes our model of auditory denervation from focal retinal lesions or nerve transection in primate models of somatosensory and visual cortex plasticity.

Finally, while the reviewers are correct to point out that the degree of injury can explain the completeness of ‘filling in’ from spared regions of the skin surface, other studies in NHP and rodent models, also make the point that factors unrelated to the degree of injury, such as the manner of use of skin surface can also exert a profound influence on adult somatotopic maps (Clark et al., 1988 Nature, Allard et al., 1991 J. Neurophys, Xerri et al., 1996 J Neurophys; Merzenich & Jenkins 1993 J Hand Therapy; Wang et al., 1995 Nature; Xerri et al., 1998, J Physiol., Polley et al., 1999, Neuron). Therefore, it is not unreasonable to surmise that activity-dependent features unrelated to the number of spared peripheral inputs could bias, if not explain outright, whether mice recovery auditory thresholds or not. We refer the reviewers to the revised Discussion where we address this issue more completely.

Mechanisms responsible for the reduced PV neuron mediated suppression of RS cell firing

1) The apparent reduction of PV-mediated inhibition (i.e., failure of a light-evoked PV cell activity to block spontaneous discharge) could be explained by an increase in tonic excitatory currents. That is, if tonic excitatory currents increased following hearing loss, then a normal level of inhibition would appear to be less efficacious at suppressing discharge. The interpretation of reduced inhibitory gain would be more convincing if there were data or analyses to confirm that excitatory gain did not change, at least during the first 5 days. However, in vitro and in vivo experiments in the context of homeostasis suggest that excitatory and voltage-gated currents do change with similar manipulations, and the increased spontaneous discharge rate observed in the current study is consistent with this idea.

We agree. There are good reasons to question whether voltage-gated currents, excitatory synaptic currents and/or extrasynaptic currents might also be contributing to the functional recovery described here. Every approach has its strength and limitations. Ours is the first approach to track changes in receptive fields, central gain, spontaneous firing rate and PV-mediated inhibition in the same neuron over a period of months surrounding a controlled injury. This approach has many merits but we could not measure EPSCs or tonic currents from the neurons we have isolated and therefore cannot identify their contribution one way or the other. On the other hand, we can monitor the strength of PV-mediated inhibition over time and relate these dynamics to the progressive recovery of function. This is something that no previous study has been able to do, not at the level of an individual neuron at least, and this proves to be a useful measurement that might have been missed with acute measurements that offer more penetration insight into possible mechanisms.

As described above, we have reworked our Discussion paragraphs to focus on possible mechanisms and caveats in the interpretation of this work. We now cite work related to each of the mechanisms raised by the reviewers. We make it clear that changes in PV circuits described here are likely not the sole mechanism. Indeed, enhanced sensitivity to glutamatergic inputs may work synergistically with dynamic changes in PV input to bring about enhanced responsiveness following peripheral denervation. As an example, we refer the reviewers to preliminary fluorescent in situ hybridization data from an ongoing study in our lab that provides single cell labeling of mRNA for an AMPA and GABAAreceptor subunit (Author response image2). One can appreciate that compared to a sham-treated age-matched control mouse, the cell in the ouabain-treated mouse has less message for the GABAA receptor and more message for the AMPA receptor.

Author response image 2. Triple fluorescence in situ hybridization of an example neuron from the auditory cortex of a ouabain-treated adult mouse or an age-matched sham-treated control.

Author response image 2.

Open in a new tab

Transcripts for Gria2, which encodes an AMPA receptor subunit, and GABRA1, which encodes a GABAA receptor subunit, are visualized in the peri-nuclear compartment. Compared to the sham control, glutamate receptor transcripts are increased and GABAA receptor transcripts are decreased. These data are part of a separate study that examines the interplay between glutamate sensitization and GABA disinhibition after auditory nerve damage.

DOI: http://dx.doi.org/10.7554/eLife.21452.011

2) The interpretation that inhibitory gain declines (Figure 2C and 2F) is based on the suppression of responses of RS neurons and could be explained by reduced FS cell excitability, reduced GABA release, or reduced GABAA receptors located postsynaptically. Are there observations that support any of these specific mechanisms? For example, were a sufficient number of FS cell waveforms recovered through spike sorting to determine how these cells responded to blue light pulses following hearing loss? If there are multiple mechanisms, do they occur over a similar time course? What happens to FS cell spontaneous firing rates following hearing loss? Since they are recurrently connected to pyramidal cells which, themselves, display increased spontaneous rates, one might expect an increase. Does the shift of RF after hearing loss apply for PV neurons as well?

All good questions. We sought to address some of these points by quantifying laser-evoked, sound-evoked and spontaneous spiking in fast-spiking, PV-expressing interneurons. We provide our analysis in the newly added Figure 1—figure supplement 1 where we show, unfortunately, that it was essentially impossible to hold a FS neuron for two or more consecutive days. Not only were these waveforms less commonly encountered but, presumably because of their comparatively smaller size, the recordings were far less stable than RS units. We concluded that we just didn’t have the basis to perform the complementary analysis on FS units without a robust sample of stable chronic recordings. However, we revised the Discussion to touch on these important points. We made this limitation clear in the revised Introduction and by including the new figure.

Relationship to prior studies

1) The authors make the following statement in the Introduction: "It is unclear how these distinct biomarkers of cortical reorganization following injury -receptive field remapping, spontaneous rate increases, and central gain increases – are coordinated over the days and weeks following varying degrees of sensory nerve injury. Nor is it certain whether these plasticity processes precede or follow homeostatic regulation of intracortical inhibition."

There are many relevant publications investigating how similar outcome measures are coordinated during the somatosensory reorganization that accompanies peripheral and central nerve injury. Changes to inhibition and concomitant reorganization of receptive fields are generally evident as soon as recordings can be obtained (e.g., Turnbull and Rasmusson, Somatosens Mot Res. 7:365, 1990; Calford and Tweedale, Somatosens Mot Res. 8:249, 1991; Rasmusson et al., Somatosens Mot Res. 10:69, 1993). More recently, a brief period of inactivity was demonstrated to induce a short latency reduction of IPSP amplitude in cortex (Lu et al., PNAS 111:1616, 2014). In addition, the rapid unmasking of subthreshold receptive fields is known to be kinetically distinct from the NMDA receptor-dependent reorganization that takes place in the weeks that follow sensory dennervation, especially in NHP models of peripheral dennervation. Thus, there are likely to be many mechanisms that operate across different time courses. In fact, homeostatic synaptic responses (reduced inhibitory or increased excitatory gain) to deprivation may permit "silent" inputs to become suprathreshold through NMDAR-dependent plasticity. It would be helpful to place the current findings in the context of a very well-developed literature.

We cited several seminal studies relating to somatosensory deafferentation in NHP in our original manuscript and cite additional work, including some of those mentioned above, in the revised work. The revised manuscript features an overhauled Introduction and Discussion with many new references that hopefully do a better job of putting our findings into context, as the reviewers have suggested. Space is limited in the Short Report format and the literature on cochlear denervation and auditory system plasticity is also well-developed and important to cite in this context. We have done our best to cite the seminal and most relevant work from the auditory, somatosensory and visual literature.

At the same time, it is also important to note the differences between the earlier studies and the findings described here. Most work in the existing literature was performed with acute measurements of unit activity in anesthetized animals at one or two times post-injury, as compared to separate control animals. If measurements of intracortical inhibition were performed, they were almost invariably at single time points and are not specific to any subtype of GABA neuron.

2) Concerning the title. The results show plasticity that begins soon after the manipulation and lasts for a long time. The question is whether all of this plasticity can be termed "homeostatic." Homeostasis covers many specific cellular mechanisms that take place after sensory deprivation or during learning. It is important to make it clear exactly what, where, and when you are referring to homeostasis (e.g., Are all of the observations, from day 0-50, a single homeostatic process?). Since the changes seem to occur proximal to the hearing loss, this issue could be addressed by specifying what is meant by the word "compensatory" and specifying what is meant by the term "homeostatic." The Abstract does refer to homeostasis as occurring "during the first days," but it is not clear that homeostatic mechanisms emerge over that long a period of time.

We revised the title so that it no longer includes the term “homeostatic” because it is unclear whether/how some of longer term changes we describe (or lack of changes in the Non-Recover group) reflect homeostatic mechanisms.

Additional issues

1) The authors state that noise exposure induces permanent loss of the ABR, assessed through wave 1 amplitude; however, Figure 2 shows almost 50% recovery by 30 days. Please revise or provide further justification for this statement.

The change in ABR amplitude during the early period of temporary threshold shift (e.g., at the day 2 measurement time in Figure 2B) reflects and immediate and permanent loss of cochlear synapses, but an additional transient changes to stereocilia tip links or cochlear supporting cells that affect mechanoelectric transduction. With TTS, the changes in the organ of Corti reverse and thresholds go back to normal and part of the ABR amplitude loss is recovered. The persistent decrease in ABR amplitude when unaccompanied by a difference in DPOAE or ABR threshold arises from the cochlear afferent synaptopathy. We revised the manuscript to make this clear in the sixth paragraph of the Results. The same pattern is observed in row one of Author responses image 1, taken from Wan et al., eLife 2014.

2) The authors do not provide histological evidence that the degree of synapse loss induced by these manipulations in this strain of mice is similar to that reported by Kujawa and Lieberman (C57BL/6 versus CBA/Cj, which were not examined by Kujawa). This issue is important, as different strains of mice exhibit different sensitivities for noise induced hearing loss.

The issue has already been raised and addressed in Point #3 of the Essential revisions, above.

3) The authors indicate that "optetrode assemblies into the 32 kHz region of the A1 tonotopic map and recorded RS units with high-frequency receptive fields, low-intensity thresholds to broadband noise stimulation and strong feedforward inhibition from PV neurons." It would be helpful if the authors indicated what proportion of sampled cells exhibited these properties. Was there a correlation between the degree of PV-mediated inhibition and the response of the neurons to peripheral injury?

The requested analyses are now provided in the top left panel of Figure 4—figure supplement 1. We show that most single units, particularly for the noise exposure group for which the above claim was made, have best frequencies between 16-32 kHz (Figure 4—figure supplement 1A). We further show that the PV-mediated inhibition and response thresholds for broadband noise bursts before damage was > 85% and less than 40 dB SPL, respectively, was fairly uniform across our sample (Figure 4—figure supplement 1B-C), and were not correlated with each other (r = 0.04, Figure 4—figure supplement 1D). Finally, we show that neither the strength of inhibition prior to auditory nerve damage, nor auditory response thresholds prior to nerve damage could predict the eventual recovery of each marker (r = 0.05 and 0.06, respectively, Figure 4—figure supplement 1E-F). Beyond the specific requests made here, we mined our dataset for features that predict the eventual mode of recovery. The best candidates are described in Figure 4.

Associated Data

    This section collects any data citations, data availability statements, or supplementary materials included in this article.

    Supplementary Materials

    Supplementary file 1. Statistical reporting.

    Detailed description of statistical analysis for all figures and figure supplements.

    DOI: http://dx.doi.org/10.7554/eLife.21452.009

    elife-21452-supp1.xlsx (19.4KB, xlsx)
    DOI: 10.7554/eLife.21452.009

    Articles from eLife are provided here courtesy of eLife Sciences Publications, Ltd

    RESOURCES