A delayed pharmacological treatment strategy attenuates noise-induced tinnitus in rats

Abstract

It is widely accepted that tinnitus arises from maladaptive neuroplasticity in the central auditory system, which compensates for reduced peripheral input; notably, the loss of inner hair cell (IHC) synapses is considered a key underlying factor. IHC synapses can immediately and permanently degenerate after noise exposure, independent of hair cell loss, whereas the cell bodies and central projections of cochlear neurons can survive for months to years, providing an extended therapeutic window for improving afferent input through synapse regeneration. Using NHPN-1010 (a combination of 2,4-disulfophenyl-N-tert-butylnitrone and N-acetylcysteine), we propose a delayed treatment strategy (≥ 4 weeks after trauma) for noise-induced tinnitus (NIT) in rats. NHPN-1010 enhanced gap-prepulse inhibition of acoustic startle responses—indicative of diminished tinnitus-like deficits—coupled with a concomitant upregulation of GABAergic inhibition in the dorsal cochlear nucleus, relative to vehicle alone (saline). Furthermore, NHPN-1010, compared to saline, led to increased wave-I amplitudes of auditory brainstem responses, the effect accompanied by a corresponding rise in the number of presynaptic ribbons in IHCs. Our findings suggest that NHPN-1010 plays a therapeutic role in NIT by promoting cochlear synaptogenesis and normalizing central auditory signaling.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-026-40960-2.

Introduction

Tinnitus has been linked to the common sequela of acoustic trauma^1–3. Despite the high prevalence and the potential adverse impact on quality of life, there are currently no FDA-approved drugs specifically for treatment of tinnitus. A clear understanding of the underlying pathophysiology of tinnitus is of fundamental importance to the development of a drug-based approach for addressing this disorder. While the prevailing view has shifted toward the belief that tinnitus originates within the central auditory system (CAS), peripheral auditory damage has consistently been identified as the primary risk factor^4–8. This perspective is reinforced by the “central gain model of tinnitus,” which suggests that tinnitus results from an abnormal compensatory increase in gain (i.e., neural amplification) within the CAS to compensate for a loss of sensory input from the cochlea^9–11. Although, in the majority of cases, tinnitus is associated with audiometrically-detected hearing loss, it can also occur without measurable hearing decline and independently of outer hair cell (OHC) loss^12,13. This finding is consistent with clinical observations that some patients with tinnitus exhibit clinically normal audiograms^14,15. Animal studies have revealed that mild to moderate acoustic overexposure, even when causing only a temporary threshold shift (TTS), can immediately and permanently induce “cochlear synaptopathy”—characterized by the loss of synaptic connections between cochlear hair cells and auditory nerve fibers, without any loss of the hair cells themselves^16–18. This type of cochlear synaptopathy, referred to as “hidden hearing loss” because the deficits can hide behind a normal audiogram¹¹, is posited to trigger neural hyperactivity that underlies tinnitus^19–21, potentially offering a valuable drug target for the development of novel tinnitus treatments. These deficits, even when localized to higher frequencies, can induce compensatory plasticity within central auditory circuits extending beyond the original site of damage, producing percepts at dissociated frequencies^7,10,11. Although the primary degeneration of cochlear nerve synapses often occurs within hours to days after the onset of acoustic overexposure, the cell bodies and central projections of these cochlear neurons can survive for months to years^22–24. This suggests a prolonged therapeutic window wherein a treatment to elicit neurite outgrowth could potentially reconnect silenced cochlear ganglion cells with hair cells, restore synaptic functions, and improve peripheral sensory input to the CAS.

NHPN-1010 is a fixed combination product consisting of two antioxidants, i.e., N-acetylcysteine (NAC, a glutathione prodrug) and 2,4-disulfophenyl-N-tert-butylnitrone (HPN-07, a nitrone compound). They have overlapping and unique mechanisms of action, resulting in synergistic efficacy. In addition to serving as a safe antidote for glutathione deficiency, NAC reduces neuroinflammation, mitochondrial injury, and cell death^25–27. HPN-07 scavenges free radicals, inhibits iNOS, COX-2 and NF-κB, and reduces excitotoxicity^28,29. Together with their antioxidant roles against reactive oxygen species (ROS) and reactive nitrogen species (RNS), NAC and HPN-07—whose structural analog is α-phenyl-N-tert-butylnitrone (PBN)—have each been reported to promote neurogenesis through proposed mechanisms involving the activation of the Ras–extracellular signal-regulated kinase (ERK) and protein kinase C (PKC) pathways^30–32. In a chinchilla model of acute acoustic trauma, HPN-07 itself enhanced auditory/cochlear functional recovery, and reduced loss of cochlear HCs, afferent nerve fibers in the spiral lamina and fibrocytes within the spiral ligament³³. In a rat model, our previous research on early NHPN-1010 treatment for blast-induced tinnitus—specifically targeting cochlear synaptopathy—demonstrated that (1) blast overpressure caused a permanent loss of cochlear nerve ribbon synapses (RSs), and (2) administration of NHPN-1010 within 3 days of blast trauma, compared to placebo, significantly rescued RSs, as evidenced by ribbon counts in cochlear histological analyses and corresponding functional changes in suprathreshold wave-I amplitudes of the auditory brainstem response (ABR)³⁴. These findings indicated that early intervention with NHPN-1010 confers treatment-specific effectiveness in promoting cochlear synaptogenesis.

The aim of the present study was to evaluate the efficacy of NHPN-1010 in a validated rat TTS model of noise-induced tinnitus (NIT), using a delayed treatment strategy. Here, prepulse inhibition (PPI) of the acoustic startle response (ASR) was implemented as a behavioral paradigm to provide evidence for the potential induction of tinnitus following sound exposure. Individual animals with PPI-based evidence of NIT were subsequently treated with either NHPN-1010 or saline, wherein intervention was deliberately initiated four weeks (wks) after exposure. Following treatment, characteristic differences were detected through either a within-subjects comparison (i.e., the before/after difference) or a between-subjects comparison (i.e., the treatment/no-treatment difference). Alongside the use of startle reflex Inhibition to assess tinnitus-like behavior, additional outcome measures included thresholds and wave-I amplitudes of the ABR, ribbon synapse counts in inner hair cells (IHCs) of the cochlea, and expression levels of GABA type A receptor (GABA_AR) α1 subunits in the dorsal cochlear nucleus (DCN). The current work demonstrates that NHPN-1010 significantly improves all assessed outcome measures relative to vehicle, suggesting that the delayed therapeutic approach is highly effective in addressing NIT in this model, including its associated cochlear synaptopathy and central auditory processing deficits. We elucidated the mechanism underlying the efficacy of NHPN-1010 in the treatment of NIT, highlighting its suppression of noise-enhanced gain in the central auditory pathways and its reversal of noise-induced primary deafferentation in the cochlea.

Results

Experimental design and timeline

Figure 1 depicts the experimental procedures and timeline for this study. Rats with normal hearing and PPI behavior were divided into experimental and control groups. Four weeks after noise exposure, tinnitus identification was conducted via behavioral tests alongside ABR measures of cochlear function and auditory sensitivity. Experimental animals exhibiting NIT were then randomly assigned to either a treatment group (receiving NHPN-1010) or a vehicle treated group (receiving saline), while noise-exposed animals that did not develop NIT were excluded from the study. Notably, NIT treatment was initiated 4 wks from the onset of trauma as a delayed intervention. At 12 wks post-exposure (corresponding to 8 wks after commencement of treatment), final behavioral, electrophysiological, and histological evaluations were performed. Naive control rats (unexposed and untreated) underwent identical measurements concurrently with the experimental groups and served as companion controls.

Fig. 1.

Schematic overview of the design and timing of the experimental protocol. Baseline behavioral and electrophysiological measurements were conducted the day before noise exposure. Tinnitus treatment (NHPN-1010 or saline vehicle) was initiated 4 wks post-exposure and continued for 14 consecutive days (d1–d14). Follow-up behavioral and electrophysiological assessments were conducted at 4 and 12 wks after noise exposure, with histological analysis performed at 12 wks post-exposure (8 wks after treatment). ABRs: auditory brainstem responses; T + : tinnitus-positive; T-: tinnitus-negative.

Noise exposure induced a chronic tinnitus percept as measured by startle reflex inhibition

In the use of the PPI paradigm as a behavioral assay, we collected data from 34 companion controls to construct a 95% confidence interval (CI) for the mean tinnitus index score, which served as the normative reference for estimating tinnitus. Sixty-six rats were exposed to octave-band noise (8–16 kHz) for 2 h at 108 dB SPL and subsequently identified as with or without tinnitus 4 wks later. This noise exposure paradigm has been shown to produce only TTS but significant cochlear synaptopathy³⁵. Gap-PPI deficits in noise-exposed animals (defined as an index score exceeding the 95% CI) indicated that 38 of the 66 animals developed tinnitus, resulting in an overall incidence rate of 58%. When categorized by tinnitus tone (or pitch), the frequency-specific incidence of a tinnitus percept exhibited substantial variability, with the highest percentage observed at 20 kHz (Fig. 2).

Fig. 2.

In noise-exposed rats, the incidence of a tinnitus percept displayed a frequency-dependent distribution. Tinnitus tones (or pitches) are defined as test frequencies at which tinnitus index scores exceed the 95% CI of the non-exposed companion control group. The perception of noise-induced tinnitus (NIT) spans a broad frequency range, with a peak incidence at 20 kHz. Dot plot inset shows the number of test frequencies at which tinnitus was identified in 66 subjects, where each dot above a number represents one subject with that corresponding number of tinnitus-associated frequencies. Data were obtained 4 wks post-exposure from n = 66 animals.

NHPN-1010 induced the attenuation of tinnitus as evidenced by behavioral measures

To test for tinnitus, the basic hypothesis is twofold: (1) a reduction in the amplitude of the startle response when it is preceded by a brief gap in a background sound indicates that gap can reduce the normal startle response to a loud sound, and (2) the startle stimulus is preceded by the perceived gap with the idea that an animal’s tinnitus will make the gap less salient and therefore less effective in reducing the startle response. The extent to which an animal suffers from tinnitus is believed to be reflected in its tinnitus index scores, as higher scores are associated with a stronger startle response, indicating a reduced effectiveness of gap inhibition. Following PPI testing at 4 wks post-noise exposure, 36 of the 38 rats that tested positive for chronic NIT were randomly assigned to either the NHPN-1010 treatment group or the vehicle (saline) group, each with n = 18 cases, where two animals were excluded from the treatment phase due to health-related concerns. Each cohort was then treated for 14 consecutive days with twice-daily treatments with either 300 mg/kg of NHPN-1010 or an equivalent volume of saline.

To evaluate the therapeutic effect of NHPN-1010 on tinnitus, PPI tests were repeated 8 wks after NHPN-1010 or saline treatment were initiated (corresponding to 12 wks after noise trauma) in the 36 rats that had exhibited behavioral evidence of tinnitus at 4 wks post-exposure. Here, a within-subjects comparison was performed using a two-way repeated measures ANOVA to assess mean differences in index scores between pre- and post-treatment conditions in the same animals. For the NHPN-1010 group (Fig. 3A), the two-way interaction between time and frequency did not yield a significant statistic [F(2.57,43.61) = 0.62, P = 0.58], while, of note, there was a significant main effect of time [F(1,17) = 4.68, P < 0.05]. Thus, the overall statistical outcome indicated that the average index score, collapsed across frequency, was significantly reduced following NHPN-1010 treatment. For the saline group (Fig. 3B), however, the main effect of time was not statistically significant in the ANOVA [F(1,17) = 0.09, P = 0.76]. Taken together, these analyses suggest that NHPN-1010 can reverse or attenuate noise-induced tinnitus-like deficits in the rat model.

Fig. 3.

NHPN-1010 reversed noise-induced increases in tinnitus index scores, whereas saline did not. (A, B) Mean scores (± SEMs) for the NHPN-1010 (A) and saline (B) groups, measured immediately before and at 8 wks after treatment (corresponding to 4 and 12 wks, respectively, post-exposure). Each group consisted of 18 animals, all of which showed behavioral evidence of tinnitus at 4 wks after exposure. P-values represent the main effect of the time factor in a two-way repeated measures ANOVA. Color key in B applies to both panels.

It should be noted that, upon review of group averages, the tinnitus scores were lower in the saline-treated group relative to the NHPN-1010 group prior to treatment (Fig. 3A and B). However, all animals included in this data set had behavioral evidence of noise-induced tinnitus at four weeks post-noise exposure and were randomly assigned to the two experimental groups at that time. The apparent difference in pre-treatment tinnitus scores observed in Fig. 3 emerged during post hoc analysis of the raw data, rather than being an intentional group difference. The values labeled “prior to treatment” represent measurements taken 4 weeks after noise exposure and do not correspond to “pre-exposure” baseline levels. The cut-off values based on the upper limit of the 95% CI from normal controls were: 0.52 at 9.3 kHz, 0.55 at 16 kHz, 0.51 at 20 kHz, and 0.61 at 24 kHz.

As illustrated in Fig. 4, a trend indicating either an upward or downward shift in tinnitus incidence across frequencies was observed when comparing pre- and post-treatment data in both experimental groups. Tinnitus occurred at 16–20 kHz but not at 24 kHz. While our 8–16 kHz noise exposure can induce damage beyond the exposure band, tinnitus in rats is most commonly observed around 16–20 kHz³⁶, aligning with their most sensitive hearing range. Above 20 kHz, hearing sensitivity declines³⁷, and tinnitus percepts are less reliably observed, likely because they tend to align with the edge or center of the damaged region rather than frequencies far above it. Compared to the pre-treatment condition, NHPN-1010 led to a decreased incidence of tinnitus, as opposed to an increased incidence following saline treatment, where noticeable elevations occurred at the 16- and 20-kHz frequencies although the changes were not statistically significant (P > 0.05). This observation further supports the therapeutic efficacy of NHPN-1010 in reducing tinnitus. It is possible that, in the treated group, tinnitus frequency just shifted after treatment. However, in the treated group, the reduction in tinnitus incidence at 16 kHz and 20 kHz was substantially greater than the slight increase observed at 9.3 kHz and 24 kHz following NHPN-1010 treatment. Therefore, it is unlikely that the tinnitus frequency simply shifted. The incidence showed minimal change over time at 24 kHz, suggesting that this frequency may reflect a consistent deficit in the animals. The overall incidence of noise-induced tinnitus was 100% before treatment and decreased to 78% after treatment.

Fig. 4.

Tinnitus incidence showed a downward trend at 16 and 20 kHz in the NHPN-1010 group in contrast to an upward trend in the saline group. The incidence of tinnitus decreased from 58.5 to 42.0% (averaged across 16 and 20 kHz) following NHPN-1010 treatment (lower panel), whereas an increase from 42.0 to 64.0% was recorded following saline administration (upper panel). Data in the lower and upper panels correspond to the same animals shown in panels A and B, respectively, of Fig. 3. Color key in the upper panel applies to both panels.

NHPN-1010 promoted functional normalization from noise-induced hearing loss as reflected in electrophysiological measures

With pre-exposure measurements serving as the baseline, auditory sensitivity was evaluated by ABR recordings following noise exposure and subsequent treatment in the chronic phase. On an individual basis, octave-band noise caused a temporary shift of up to 15 dB in ABR thresholds at certain frequencies—detectable 4 wks post-noise exposure—with thresholds gradually returning to near-baseline levels over time. For omnibus statistics, a two-way repeated measures ANOVA revealed a significant main effect of time in both the saline [Fig. 5B; F(1.16,19.77) = 16.75, P < 0.001] and NHPN-1010 [Fig. 5A; F(1.80,30.56) = 27.29, P < 0.001] groups, while the interaction of time and frequency was not significant in either cohort. In this case, the global ANOVA indicated that the effect of temporal condition on ABR thresholds was consistent across all test frequencies. For post hoc statistics, Dunnett’s test demonstrated that, at 8 wks post-intervention, threshold elevations (averaged across 4–32 kHz) remained statistically significant in the saline group (Fig. 5B; P < 0.01) but not in the NHPN-1010 group (Fig. 5A; P = 0.19), relative to the pre-exposure baseline. Despite small mean differences and overlapping SEMs in Fig. 5B, the within-subjects design and large sample size (18 rats, 36 ears) provided sufficient statistical power to detect consistent threshold differences, resulting in a significant effect of p < 0.01. These statistical results suggest that, following NHPN-1010 treatment, sensorineural hearing loss induced by the acoustic trauma could be fully reversible with respect to hair cells—possibly also including the subgroup of cochlear neurons characterized by low thresholds and high spontaneous discharge rates, although the two-hour exposure at 108 dB SPL caused a mild inner ear injury in vehicle-treated animals.

Fig. 5.

In noise-exposed rats, ABR thresholds fully recovered in the NHPN-1010 group by 8 wks after treatment, while a slight permanent threshold elevation remained in the saline group. (A, B) Mean thresholds (± SEMs) are shown at each of the four test frequencies for three time points: baseline (pre-exposure), prior to treatment (4 wks post-exposure), and 8 wks after treatment (12 wks post-exposure). Data in panels A and B were obtained from the same animals as in Fig. 3A and B, respectively. P-values represent the main effect of time in a two-way repeated measures ANOVA. Color key in B applies to both panels. Average thresholds ± SEM across frequencies are as follows—in A: 17.5 ± 0.327 (baseline), 19.4 ± 0.403 (prior to treatment), and 18.0 ± 0.390 (8 wks after treatment); in B: 17.5 ± 0.386 (baseline), 19.3 ± 0.545 (prior to treatment), and 18.0 ± 0.436 (8 wks after treatment). Statistical significance by Dunnett’s test: ***P < 0.001, prior to treatment different from baseline (A, B); **P < 0.01, 8 wks after treatment different from baseline (B). The asterisks in the figure refer to the average threshold.

To functionally assess noise-induced cochlear neuropathy, central neural hyperactivity, and subsequent post-treatment changes, peak-to-peak amplitudes of ABR waves I and V were recorded in response to tone bursts of 80 dB SPL at frequencies 4–32 kHz. Except at 4 kHz, there was no reduction in ABR amplitudes caused by the noise (Fig. 6A). This paradoxical finding may indicate that noise exposure overdrives cochlear neurons and induces glutamate excitotoxicity—an excessive release of glutamate by the IHCs and possibly linked to functional changes in lateral olivocochlear (LOC) efferent neurons during exposure—which temporarily enhances wave I before the eventual silencing of auditory nerve fibers and synapses.

Fig. 6.

NHPN-1010 rescued wave-I amplitudes following synaptopathic noise exposure while maintaining wave-V amplitudes in ABR suprathreshold responses. (A, B, C, D) Amplitudes of ABR waves I (A, B) and V (C, D) at a stimulus level of 80 dB SPL were normalized with respect to mean values in age-matched unexposed naïve controls. Means (± SEMs) are shown. The control group size was n = 34; experimental group sizes were as indicated in Fig. 3. P-values represent the main effect of the group factor in a two-way mixed AN(C)OVA. Dashed lines (–) indicate the normalized control mean (1.0). Color key in D applies to all panels.

For analysis of differences in mean values between the two experimental groups, a between-subjects comparison was conducted prior to treatment (4 wks post-exposure) using a mixed ANOVA, and at 8 wks after treatment (12 wks post-exposure), using a mixed-design ANCOVA. As shown in Fig. 6 and Supplemental Fig. 1, the ANOVA did not reveal a significant main effect of group on either wave-I amplitude [Fig. 6A; F(1,70) = 2.27, P = 0.14] or wave-V amplitude [Fig. 6C; F(1,70) = 0.36, P = 0.55]. Of note, after adjusting for pre-treatment measurements (see Fig. 6A, C) as a covariate, the ANCOVA showed a stronger significance in main effect of group on wave-I amplitude [Fig. 6B; F(1,279) = 7.064, P < 0.01], but insignificance on wave-V amplitude [Fig. 6D; F(1, 279) = 1.75, P = 0.19]. Overall, the results indicated that NHPN-1010 enhanced wave-I amplitudes while maintaining wave-V amplitudes compared to saline. This contrasts with the phenomenon of compensatory central gain, where a loss of cochlear input (a reduced wave-I amplitude) leads to a maladaptive increase in neural gain within the CAS that can preserve central auditory signals (an unchanged wave-V amplitude) but amplifies central spontaneous neural noise, contributing to tinnitus. Here, no significant two-way interaction between frequency and group was found in either the ANOVA or ANCOVA. Accordingly, in the animal NIT model, NHPN-1010 treatment, relative to saline, increased wave-I amplitudes while maintaining stable wave-V amplitudes, leading to a reduced wave-V/I amplitude ratio that is consistent with attenuation of tinnitus-related compensatory central gain and supports the therapeutic potential of NHPN-1010 (see further discussion in the Discussion section).

NHPN-1010 upregulated GABA_AR-α1 subunit expression in the DCN of rats with NIT, as demonstrated by immunohistochemical analysis

For the assessment of NHPN-1010’s effect on GABAergic function in the DCN within the NIT model, seven non-exposed companion naïve controls, as well as seven tinnitus-positive rats (identified 4 wks post-exposure) from both treatment groups, were randomly selected for immunohistochemical analysis conducted 8 wks post-treatment (Fig. 7A-C). Previous work from our laboratory using this NIT model system included an evaluation of potential tinnitus-related biomarkers in the CAS³⁵. Herein, quantitative histological analyses were repeated for GluR2, T22, and GABA_AR-α1 in the DCN, auditory cortex, and hippocampus. No significant treatment-specific effects in GluR2 or T22 levels were measured in any of these regions of the CAS (all p > 0.05). However, we did denote significant NHPN-1010-mediated effects for GABA_AR-α1 expression in the DCN. Quantification of GABA_AR-α1 subunit cluster density was conducted in three structural regions of the DCN: medial, middle, and lateral to cover the tonotopic axis. GABA_AR-α1 positive cells varied in size and shape, indicating they are different types of cells. One of our recent studies demonstrated that the most common GABA_AR-α1 positive cells are inhibitor cells (Golgi, vertical and cartwheel cells) although all eight types of DCN cells could be GABA_AR-α1 positive³⁸. NHPN-1010 treatment significantly upregulated GABA_AR-α1 subunit expression compared to saline intervention (Fig. 7D; F(1,12) = 5.21, P < 0.05, two-way mixed ANOVA) in all three regions, suggesting that α1 subunits of GABA_ARs in the DCN may represent a therapeutic target of NHPN-1010. Given that GABA functions to reduce neuronal excitability in the CNS, the NHPN-1010–induced enhancement of α1 subunit expression in the DCN could indicate a reversal of central neural hyperactivity.

Fig. 7.

NHPN-1010 treatment upregulated GABA_AR-α1 subunit expression in the DCN. A–C are examples of GABA_AR-α1 immunostaining in the DCN of rats in the normal control, saline and NHPN-1010 groups. Images were taken from the fusiform cell layer in the middle region of the DCN. Numerous GABA_AR-α1 positive cells were observed in the DCN of rats of all conditions (arrows in A–C), especially in the DCN of NHPN-1010 rats (arrows in C). The positive cells had different sizes and shapes (arrows in A–C). D indicates that the number of GABA_AR-α1 positive cells was quantified in the medial, middle, and lateral regions of the DCN. Cell density (cells/mm²) was calculated and normalized to the mean of the corresponding control region. Both the NHPN-1010 and saline groups consisted of 7 tinnitus animals each, while the control group comprised 7 age-matched unexposed naïve control animals. Immunohistochemical analysis was performed 8 wks after treatment (12 wks post-noise exposure). Means (± SEMs) are shown. A two-way mixed ANOVA yielded a significant main effect of group, but no significant interaction between group and region. An omnibus p-value represents the main effect of the group factor in a two-way mixed ANOVA. The scale bar in C = 20 μm for A–C.

NHPN-1010 regenerated cochlear synapses following acoustic trauma, as documented by confocal microscopy of immunolabeled cochlear sensory epithelia

In examining NHPN-1010–induced synaptogenesis in the IHC area, presynaptic ribbons in IHCs were counted 8 wks after either NHPN-1010 or saline intervention in the 22.6 kHz and 32 kHz regions. Data were collected from seven tinnitus-afflicted rats in the saline group, eight in the NHPN-1010 group, and nine normal companion controls. As presented in Fig. 8, a two-way mixed ANOVA revealed significance both in main effect of group [F(1,13) = 7.32, P < 0.05] and in interaction of frequency and group [F(1,13) = 6.08, P < 0.05], where intergroup comparisons were significant at 32 kHz (P < 0.01) but not at 22.6 kHz (P = 0.11) in post hoc t-tests. Specifically, ribbon counts showed a large and significant increase for the 32 kHz region but only a small and nonsignificant increase for the 22.6 kHz region in the NHPN-1010 group with respect to the saline group. Previous studies have demonstrated that permanent loss of ribbon synapses can occur immediately following noise exposure, even in cases where threshold elevations are transient and hair cells survive^16,18,39. Our prior work in this TTS noise-exposure model demonstrated that ribbon synapse loss across these tonotopic frequency regions is both rapid (i.e. within the first 24 h) and permanent³⁵. As such, our current and previous findings highlight the potential therapeutic effect of NHPN-1010 on ribbon synapses^34,40.

Fig. 8.

NHPN-1010 rescued noise-induced loss of IHC synapses. Synaptic ribbon counts per IHC at 22.6 kHz and 32 kHz were normalized to the mean values of age-matched, unexposed controls. Group sizes were as follows: normal, n = 7; saline, n = 7; NHPN-1010, n = 8. Data are presented as mean (± SEM) values. An omnibus p-value represents the main effect of group in a two-way mixed ANOVA. Statistical significance by t-test: **P < 0.01, NHPN-1010 group different from saline group. The inset bar graph displays absolute synaptic counts (mean ± SEM) per IHC at 22.6 kHz and 32kHz for all three groups. Color key in the inset applies to both the main and inset panels.

Discussion

Although there is no broad consensus regarding the specific mechanisms or the exact loci generating tinnitus, the central origin is now widely acknowledged. It is increasingly clear that the development and perpetuation of stimulus-independent “noise” within the CAS can arise from compensation for a loss of peripheral cochlear sensory input⁷. Animal studies suggest that cochlear synaptopathy—defined as the loss of synapses between IHCs and auditory nerve fibers—may contribute specifically to central auditory hyperactivity and tinnitus following noise exposure⁴¹. This selective synaptic disruption predominantly affects low-spontaneous-rate, high-threshold auditory nerve fibers, effectively silencing a subset of the afferent input¹⁷. Such targeted deafferentation may preferentially engage maladaptive neuroplastic mechanisms, which are not equivalently triggered by more extensive cochlear damage, such as IHC loss [³⁴,⁴²,⁴³]. Taken together, the association between cochlear neuronal degeneration and central neural hyperactivity may underlie the perceptual manifestation of NIT. The ABR wave-I amplitude, which represents cochlear nerve activity, can provide an objective measure of cochlear synaptopathy when cochlear thresholds and associated hair cell function remain normal¹⁶, while the wave V, generated in the lateral lemniscus and inferior colliculus⁴⁴, is commonly used to evaluate central auditory gain. The combination of these two measures can reveal a link between cochlear neuropathy and central neural hyperactivity, providing characteristic evidence that cochlear synaptopathy—seen as a reduction in wave-I amplitude—is followed by compensatory central gain, reflected in the relatively stable amplitude of wave V^11,45. This can be viewed as a form of homeostatic plasticity, where the CAS enhances its activity to maintain a relatively constant level of neural firing in response to diminished peripheral sensory input. Precisely at the core of the central gain model is the hypothesis that extensive spontaneous and sound-evoked neuronal hyperactivity in higher auditory structures (e.g., DCN) may constitute a key neural correlate underlying the sound percepts of tinnitus^46–49. The increase in central neural activity is believed to be a consequence of an alteration in the normal balance between excitation and inhibition in nerve transmission, potentially due to a diminution of GABAergic inhibition^50–52. GABA is the primary inhibitory neurotransmitter in the mammalian brain where it acts predominantly through GABA_ARs. Downregulation of these receptors, as indicated by decreased receptor density, can cause disinhibition in neurons.

In the present study, the characteristic changes in outcome measures observed in the NHPN-1010 and placebo groups—with audiometrically-matched hearing profiles (Fig. 5A, B)—are consistent with the central gain-based predictions when evaluated 8 wks post-treatment. Specifically, in addition to a reduction in both tinnitus index scores (Fig. 3A) and tinnitus incidence rates (Fig. 4, lower panel), which indicated diminished tinnitus-like deficits, the efficacy of NHPN-1010 versus vehicle treatment was demonstrated by two key findings: (1) NHPN-1010 treatment resulted in an increase in IHC ribbon counts (Fig. 8) with a commensurate enhancement in ABR wave-I amplitudes (Fig. 6B), suggestive of reduced cochlear deafferentation; and (2) an elevation in GABA_AR-α1-positive cell densities in the DCN (Fig. 7) with a corresponding reduction in ABR wave-V/I ratios is indicative of increased central inhibition. This therapeutic ratio change was reflected by stable wave-V amplitudes (Fig. 6D) in the presence of enhanced wave-I amplitudes (Fig. 6B). These findings suggest that NHPN-1010 could be a beneficial agent for the treatment of NIT through reversal of cochlear synapthopathy in the IHC area and suppression of neuronal hyperactivity in the CAS.

Previous animal studies have revealed that noise exposure can induce immediate and permanent cochlear synaptopathy¹⁸, as evidenced by the fact that synaptic counts measured at 0 h post-exposure are similar to those seen at 24 h, 1 wk and 8 wks thereafter¹⁶. Accordingly, the effectiveness of NHPN-1010 in treating cochlear neuropathy in the absence of hair cell loss likely reflects the regeneration of missing peripheral terminals and the formation of new synaptic connections, rather than simply enhancing recovery of damaged neuronal terminals. NHPN-1010 comprises two key active components: NAC, a glutathione precursor that acts as a free radical scavenger, and HPN-07, a member of the PBN family of nitrones used as spin-trapping agents to neutralize free radicals. Known for their role in preventing ROS/RNS-induced oxidative stress^53–55, NAC and PBN have also been shown to promote neuronal regeneration. Several mechanisms have been proposed for NAC- or HPN-07-induced neurogenesis, potentially involving activation of the Ras-ERK cascade and the PKC pathway^30–32; clearance of myelin debris, which otherwise prevents remyelinization⁵⁶; attenuation of retrograde neuronal cell death⁵⁷; and reduction of neurite adhesion to the substratum, which can impede neurite extension⁵⁸. Overall, the involvement of NHPN-1010 in IHC synaptogenesis may be a critical factor in alleviating phantom sounds caused by noise exposure.

Considering that spiral ganglion neurons can survive for months to years in the absence of hair cell afferent synapses¹⁶, Suzuki et al. suggest a prolonged therapeutic window within which to regenerate cochlear synapses in cases where the hair cells remain intact⁵⁹. In our prior study with this TTS model of NIT, we found that acoustic trauma did not result in significant hair cell loss—either IHCs or OHCs—at any tonotopic frequency position³⁵. As documented in the present study, delayed treatment with NHPN-1010 (≥ 4 wks after exposure onset) has proven to be a promising therapeutic approach for NIT.

As further consideration of translation of this therapeutic to the clinic, cochlear synaptopathy is implicated as a potential contributor to human tinnitus, although the published evidence is emerging and, so far, is largely indirect^11,60. Several studies using non-invasive physiological markers, such as reduced ABR wave-I amplitude, suggest an association between cochlear synaptopathy and tinnitus, particularly in individuals with normal audiograms^11,43,61,62. Meta-analytic data indicate that tinnitus patients with normal hearing thresholds often exhibit significantly reduced ABR wave-I amplitudes, consistent with synaptic loss, and an increased wave-V/I ratio, suggesting central gain enhancement, although changes in wave-V amplitude are inconsistent across studies^43,63. Computational modeling and physiological data further support that lower predicted synapse counts are associated with increased likelihood of tinnitus^9,11,64,65.

Of note, except at 4 kHz, there was no reduction in ABR amplitudes at 4 wks after noise exposure (Fig. 6A). This transient, noise-induced paradoxical phenomenon, which precedes the eventual silencing of auditory nerve fibers and synapses, may be characterized by (1) excessive glutamate release from IHCs, resulting in overactivation of receptors on auditory nerve fibers⁶⁶, and (2) functional compensation by LOC efferent neurons, ensuring that auditory nerve activity is enhanced during periods of increased demand⁶⁷, consistent with evidence suggesting that a primary function of LOC efferents is to reduce auditory neuropathy⁶⁸.

Although our behavioral results show tinnitus-like responses at frequencies without detectable cochlear synaptopathy, this pattern is consistent with central-gain models of tinnitus. Localized peripheral deficits, often at higher frequencies, can induce compensatory plasticity within central auditory circuits that extends beyond the original site of damage, producing percepts at dissociated frequencies^7,10,11. The observed frequency mismatch therefore does not contradict the role of peripheral damage in initiating central hyperactivity; rather, it highlights how localized cochlear injury can drive broad, frequency-nonspecific central adaptations, supporting a central-gain–based interpretation of our behavioral findings^16,69.

Materials and methods

Animals

Male Sprague Dawley rats, aged 1–3 months, were used in behavioral, physiological, and histological studies of noise-induced pathology and tinnitus. Rats weighing 250–300 g were purchased from Harlan Laboratories and were maintained on a normal day/night cycle at 21°C with ad libitum access to food and water. Following standard protocol, the animals were given a 1-wk acclimation period prior to experimental procedures. For this study, rats were divided into three groups: (1) a normal hearing group (control), which was not exposed to noise; (2) a non-treatment group (saline), in which animals were exposed to noise and received a placebo saline solution; and (3) a treatment group (NHPN-1010), in which animals were exposed to noise and received the active compound NHPN-1010. This study is reported in accordance with the ARRIVE (Animal Research: Reporting of In Vivo Experiments) guidelines to ensure transparent and comprehensive reporting of animal research. All animal procedures were approved by the Institutional Animal Care and Use Committee at the University of Oklahoma Health Sciences Center, and were conducted in compliance with the U.S. Public Health Service Policy on Humane Care and Use of Laboratory Animals and the Guide for the Care and Use of Laboratory Animals (National Research Council, 8th edition, 2011).

Noise exposure

Noise exposures were conducted as described previously³⁵. Briefly, under anesthesia induced by intraperitoneal (i.p.) injection of ketamine (80 mg/kg) and xylazine (10 mg/kg), rats were exposed to octave-band noise (8–16 kHz) at 108 dB SPL for 2 h. The exposure was performed in a small reverberant chamber, within which rats were placed in an acoustically transparent wire platform. The noise was digitally generated using RPvdsEx software 5.4 running on an RP2.1 Enhanced Real-time Processor (Tucker-Davis Technologies, Inc., or TDT), amplified by a PLX-3402 Power Amplifier (QSC Audio Products Inc.), and delivered through a JBL 2446H driver coupled to a JBL 3860A Bi-Radial Horn (JBL, Harman International), both mounted in the chamber ceiling. Noise intensity was monitored at the ear canal using a Brüel & Kjær 3560-L PULSE Lite pocket analyzer through a Brüel & Kjær 4133 ½-inch microphone and 2619 preamplifier. Sound levels varied by < 1 dB on the surface of the platform.

Therapeutic administration

NHPN-1010, an antioxidant compound consisting of HPN-07 and NAC, was prepared in a saline solution that was sterilely-filtered to achieve a final concentration of 60 mg/mL for each component. The medication was administered i.p. at 5 mL/kg (i.e. 300 mg/kg) in a twice-daily dosing regimen to animals with behavioral evidence of tinnitus, starting at 4 wks after noise exposure and continuing for 14 consecutive days. Animals in the placebo group received i.p. injections of saline at the same volumetric dose as the active ingredient formulation, following the same dosing schedule as the treatment group.

Cochlear functional testing

ABRs were measured bilaterally, testing each ear separately in a sound booth (Industrial Acoustics Company). To ensure monaural stimulation, the contralateral ear was occluded with an earplug during testing. For testing, animals were anesthetized with ketamine (80 mg/kg) and xylazine (10 mg/kg) via i.p. injection, while basal body temperature was maintained using a servo heating pad. The ABR approaches were based on those used in our previous research³⁴. Stimulus generation and data acquisition were accomplished through a TDT RX6 workstation running TDT BioSigRP software. Acoustic stimuli were tone bursts of alternating polarity with a 5-ms plateau and 0.5-ms cos² rise-fall envelope at frequencies of 4, 8, 16, and 32 kHz. Tone bursts were presented monaurally in free-field mode of a TDT MF1 speaker, positioned 10 cm lateral to the pinna and driven by a TDT stereo power amplifier. ABRs were recorded using subdermal needle electrodes inserted at the vertex (active), ipsilateral mastoid (reference), and contralateral mastoid (ground), where the responses were amplified 10,000 × , band-pass filtered from 0.3 to 3 kHz, and averaged across 512 repetitions at a rate of 21/s. Sound intensities were varied in 5-dB steps up and down to identify ABR thresholds. The threshold was defined as the lowest dB stimulus level that yielded a repeatable, clearly discernible waveform by visual analysis of stacked waveforms from the highest to lowest SPL. Plus, suprathreshold amplitudes of waves I and V were measured at a stimulus level of 80 dB SPL. ABR peak-to-peak amplitudes were defined as the voltage difference, in microvolts, from the peak to the subsequent negative trough.

Behavioral testing

Gap and acoustic PPI test sessions, adapted from Lu et al.³⁴, were based on the ability of the ASR to be reduced when preceded either by a silent gap in a constant acoustic background or by an acoustic burst in a quiet background. Briefly, behavioral testing was conducted using the StartleMonitor/AuxAmp II system (Kinder Scientific, LLC), where animals were placed in a sound-attenuating box equipped with two ceiling-mounted loudspeakers and a floor-mounted piezoelectric transducer platform. Acoustic prepulses and background noise were delivered through one speaker (XT25TG30-04, Vifa), while startle stimuli were presented via the other speaker (CTS KSN-1005, Powerline). The piezoelectric transducer, attached to the animal holder, provided a measure of the startle force applied by the animal. The startle stimuli were 50-ms broadband noise bursts (0 ms rise-fall) presented at 107 dB SPL. Two prepulse modalities were used in the PPI tests: a bandpass noise burst for acoustic PPI and a silent gap in an otherwise continuous carrier for gap-PPI. With an envelope of 50 ms featuring 1.0-ms ramps and occurring 100 ms before the startle stimulus, the acoustic prepulse and the gap carrier were both filtered noise (one-third octave bandpass, 48 dB/octave roll-off) centered at 9.3, 16, 20, and 24 kHz, each at an intensity of 60 dB SPL. Each test session began with a 2-min acclimation period followed by 3 startle-only trials in a quiet background to habituate the startle response to a more stable baseline. For the remainder of each session, a mix of startle-only trials (both in quiet and in continuous frequency-specific noise backgrounds), gap-PPI trials, and acoustic PPI trials were conducted in a counterbalanced fashion. The startle amplitude was determined by the peak force in newtons following the onset of the startle stimulus. The PPI was defined as a fractional reduction of startle, i.e., the ratio of startle amplitude with vs. without prepulse. Thus, a value of 1 means no effect of the prepulse, and a value of 0 means complete prepulse inhibition of the startle response. In this study, we used the tinnitus index score as a metric, calculated by subtracting the acoustic PPI from the gap PPI, to assess the presence of a tinnitus percept in animals. Higher scores are indicative of tinnitus-like deficits in processing silent gap cues. The threshold values (i.e., upper CI limits) were: 0.52 at 9.3 kHz, 0.55 at 16 kHz, 0.51 at 20 kHz, and 0.61 at 24 kHz. To minimize intra-individual variability (behavioral fluctuations within the same individuals), the testing process spanned four consecutive days with data collection occurring during the final two days. Animals were excluded from the behavioral dataset if the ASR in response to a 107-dB SPL stimulus in quiet was less than 3 standard deviations above the average baseline activity recorded prior to the startle stimulus. Additionally, extreme values in the dataset were removed when identified as outliers through Grubb’s method.

Immunostaining of GABA_AR-α1 subunits in the DCN

Animals used for brainstem sectioning and subsequent immunohistological evaluations in each experimental group (6–8 rats/group) were evaluated for sufficient level of sedation after ABR and if necessary given additional ketamine at half the initial dose, then euthanized with i.p. injection of approximately 400 mg/kg pentobarbital sodium. They were then intracardially perfused with saline, followed by 4% paraformaldehyde in 0.1 M phosphate-buffered saline (PBS, pH 7.2), at the final phase—following terminal physiological and behavioral assessments. Brains were removed, post-fixed in the same fixative for 1 wk, washed 3 times with PBS, and then stored in PBS at 4˚C. The brainstem from each animal was cryoprotected in 30% sucrose in PBS at 4˚˚C until the tissue settled to the bottom of the container. Brainstem tissues were then embedded in Tissue-Tek (Sakura Finetek USA Inc. Torrance, CA) and serially sectioned in a coronal plane with a Thermo Cryotome (Thermo Fisher Scientific, Inc., Waltham, MA) at 20 μm. One in every 10 sections from each brainstem was mounted onto a gelatin-coated slide, for a total of 10 slides per brainstem. The distance between two adjacent sections on each slide was about 200 μm. For immunohistological staining, brainstem sections were blocked with 1% bovine serum albumin (BSA, fraction V) and 1% normal goat serum in 0.2% Triton X-100 in PBS (PBS/T). Blocked and permeabilized sections were then incubated overnight at room temperature with rabbit anti-GABA_AR-α1 (1:500, EMD Millipore, Billerica, MA, catalog# 06–868). After washing with PBS/T, biotinylated goat anti-rabbit IgG (1:200, Vector Laboratories, Inc. Burlingame, CA) was applied to the slides for one hour at room temperature. Immunolabeling visualization was performed using the Vectastain ABC and DAB kits (Vector Laboratories, Inc., Burlingame, CA). Immuno-positive cells displayed a brown reaction product at the sites of the target epitopes. Methyl green was used for nuclear counter-staining. Negative controls were prepared by omitting the primary antibodies. For quantitative evaluations, images were collected with a BX51 Olympus microscope. DCN images were collected from the medial third (medial), middle third (middle), and lateral third (lateral) of the sections⁷⁰. A modified two-dimensional quantification method was employed to count immuno-positive cells in the DCN^71,72. The total number of immune-positive cells in each image was quantified using ImageJ software (National Institutes of Health) by a technician who was blinded to the identity of the samples on each slide. Only dark, brown-stained cells were counted. The density of biomarker-positive cells (number of positive cells per mm²) was calculated and subjected to statistical analysis.

Immunolabeling of ribbon synapses in the cochlea

Collection of cochlear tissue and immunolabeling of ribbon synapses have been described previously^35,40. In brief, animals were decapitated under deep anesthesia at the final phase of the experiment. Cochleae were quickly removed from the temporal bones and placed in cold PBS. The round and oval windows were opened, and a hole was created in the apical portion. Cochleae were perfused with a 4% formaldehyde solution in PBS and then post-fixed for an additional 10 min at 4˚C. Cochleae were then dissected in PBS and blocked with 1% Triton X-100 and 5% normal horse serum in PBS for 1 h. Cochlear tissues were immunolabeled with a combination of rabbit anti-GluR2/3 (Millipore Bioscience, catalog # AB1506, 1:100) and mouse anti-C-terminal binding protein antibody (CtBP2, BD Transduction Laboratories, catalog # 612044, 1:200) for 20 h at 37˚C. The tissues were rinsed with PBS and incubated with Alexa Fluor488 chicken anti-rabbit (1:1000, Life Technologies) for one hour at 37 °C and then rinsed with PBS and incubated with Alexa Fluor488 goat anti-chicken antibody (1:1000, Life Technologies) and Alexa Fluor568 goat anti-mouse antibody (1:1000, Life Technologies) for another hour at 37 °C. The tissues were counterstained with DAPI (4’,6-diamidino-2-phenylindole) for 10 min at room temperature to label the nuclei and then mounted on slides with anti-fade medium (ProLong Gold, Thermo Fisher Scientific Inc., Rockford, IL). The entire cochlea was photographed using an epifluorescence microscope. Cochlear length was measured, and a frequency map was generated using a custom ImageJ plug-in (http://www.masseyeandear.org/research/ent/eaton-peabody/epl-histologyresources/). Confocal z-stacks at frequency locations of 22.6 and 32 kHz were obtained. Images were acquired using a Zeiss LSM-710 confocal microscope (Carl Zeiss Microimaging, LLC) with a 1024 × 1024-pixel frame and a z step of 0.5 μm. From an endolymphatic surface view of the organ of Corti, each stack contained ribbon synapses from six to nine IHCs. Image stacks were processed using Amira 3D software (FEI). All quantitative analyses were performed with raw image stacks. The presynaptic ribbons (red channel) were identified through segmentation, then quantified and tracked in the z-dimension to avoid superimposition ambiguity or overestimation in each stack. Individual ribbons were isolated, counted without knowing the experimental group, and expressed as synaptic ribbons per IHC in the stack.

Statistical analysis

To determine the perception of tinnitus, a 95% CI for the average tinnitus index score was constructed for each test frequency using the dataset from the companion control group. Noise-exposed subjects with tinnitus index scores exceeding the 95% CI at any given frequency were deemed different from non-exposed controls and classified as having spectrally restricted tinnitus at that frequency. To examine group differences in the effect of pharmacotherapy on tinnitus, an ANOVA (or ANCOVA when controlling for a pre-treatment measure as a covariate) was employed to test this specific hypothesis. When a global AN(C)OVA F-test revealed a significant overall result, Tukey’s HSD for all pairwise comparisons or Dunnett’s test for multiple-to-one comparisons was used as a post hoc test. Statistical analyses, including data visualization, were conducted via the R programming language and environment (version 4.3.2).

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

J.L., X.D., R.D.K. and M.B.W. designed research; J.L., X.D., Z.Y., Q.C., W.C., D.N., W.L. and X.H. performed research; J.L. analyzed data; J.L., X.D. and M.B.W. wrote the paper.

Data availability

All data are available from the corresponding author.

Declarations

Competing interests

Richard D. Kopke has a financial interest in Otologic Pharmaceutics, Inc. All other authors declare no competing interests.