Diseases of the inner ear that lead to hearing loss (HL) may also result in subjective tinnitus (Ahlf et al., 2012), an auditory phantom sensation that is experienced, although no physical sound is present. Tinnitus occurs with surprisingly high prevalence, affecting about 35% (Shargorodsky et al., 2010) of the general population, with 10–15% of individuals experiencing prolonged periods of tinnitus that require medical evaluation. For 10% of the population, tinnitus has a significant impact on their quality of life (Chao et al., 2014). Despite this high prevalence and the tinnitus-associated distress of affected patients, which, in severe cases, may experience insomnia, psychological disorders like depression, the inability to work, or even commit suicide (Coles, 1984; Lewis et al., 1994; Langguth et al., 2011), the cause(s) and pathophysiology of tinnitus are still controversially discussed, making prevention and treatments difficult to develop (Turner et al., 2006). More than 20 years ago, tinnitus was thought to result from aberrant neural activity generated in the periphery of the auditory system (Jastreboff, 1990). In particular, tinnitus was proposed to result from increased activity in the cochlear nerve. More recently, due to the developments in basic neuroscience, a central origin of tinnitus-related activity seems to have replaced the former peripheral hypothesis (Noreña and Farley, 2013). The main reason for this shift is based on evidence that surgical neurotomy of the cochlear nerve, which should suppress tinnitus if the activity driving the percept originated in the cochlear nerve, has hardly ever had this intended effect (House and Brackman, 1981; Barrs and Brackmann, 1984; Silverstein et al., 1986; Pulec, 1995; Baguley et al., 2002). An effective cure for tinnitus still does not exist, and the main reason is that the neurophysiological mechanism that leads to the development of tinnitus is still not fully understood. Until today, several tinnitus models are being debated, although, due to developments in basic neuroscience, a central origin of tinnitus-related activity seems to have replaced the former peripheral hypothesis (Noreña and Farley, 2013). In particular, three main models, which are based on altered lateral inhibition, homeostatic plasticity, or stochastic resonance (SR) (Gerken, 1996; Eggermont and Roberts, 2004; Schaette and McAlpine, 2011; Ahlf et al., 2012; Tziridis et al., 2015; Leaver et al., 2016; Vanneste and De Ridder, 2016; Schilling et al., 2020), propose a central origin of tinnitus, resulting from damage to the cochlea as the initial step in tinnitus development.

In our recent work, Krauss and colleagues (Krauss et al., 2016) have suggested a central origin of tinnitus-related development of neuronal hyperactivity based on SR, which refers to the phenomenon that weak signals, which are subthreshold for a given sensor (or synapse), can still be detected and transmitted by that sensor if (neuronal) noise is added to the sensor input. We assumed that SR at the level of the dorsal cochlear nucleus constantly optimizes information transmission into the auditory system and, thereby, may, e.g., compensate for hearing loss. In this view, the noise necessary for SR is then the neurophysiological source of tinnitus-related enhanced neuronal activity.

The two main tinnitus inducers in humans are noise trauma (Chermak and Dengerink, 1987; Metternich and Brusis, 1999; Temmel et al., 1999; Stankiewicz et al., 2000; Mrena et al., 2002; Langguth et al., 2011) and high dose of salicylate (Myers and Bernstein, 1965; McFadden et al., 1984; Day et al., 1989; Cazals, 2000; Baguley et al., 2002; Langguth et al., 2011). In this study, the effect of salicylate is tested in Mongolian gerbils, and the results are compared to data of noise trauma-induced tinnitus (Ahlf et al., 2012; Tziridis et al., 2015) to investigate if those two types of tinnitus are based on the same neurophysiological mechanism or not.

We induce tinnitus in Mongolian gerbils, because, in contrast to mice and rats, the hearing of the gerbil up to 20 kHz closely resembles the human audiogram (best hearing around 4 kHz) (Ryan, 1976). We, here, compare neurophysiological and behavioral markers of tinnitus in animals receiving a noise trauma at 2 kHz and 115 dB SPL for 75 min and animals receiving subcutaneous injections of a high dose of salicylate (300 mg/kg), as it has been demonstrated before that salicylate doses between 150 and 350 mg/kg induce tinnitus in rodents (Sheppard et al., 2014). Behavioral estimates of salicylate-and noise trauma-induced tinnitus were obtained, using gap prepulse inhibition of the acoustic startle reflex (GPIAS, cf. Schilling et al., 2017), and auditory brainstem responses (ABR) were recorded to monitor changes in central auditory activity. We tested frequency-specific differences in GPIAS after noise trauma and salicylate treatment, and we correlated these data with possible ABR threshold changes in the same animals to evaluate similarities and differences of both tinnitus models. Taken together, the aim of the study was the characterization of neurophysiological and behavioral effects of salicylate-induced tinnitus and its comparison with data obtained with the noise trauma-induced tinnitus in the context of our model of SR-based tinnitus development.

With this study, we aimed to investigate the neurophysiological mechanisms underlying salicylate and trauma-induced tinnitus. To this end, the effect of salicylate on hearing thresholds measured by ABR and behavioral correlates of a tinnitus percept assessed by GPIAS was tested in Mongolian gerbils, and the results were compared to the same variables in animals with noise trauma-induced tinnitus. For the latter, we hypothesize the underlying neurophysiological mechanism to be based on auditory information optimization processes based on an SR mechanism (cf. Krauss et al., 2016). We found that salicylate induced behavioral changes associated with a possible tinnitus percept. However, due to the observed shifts in hearing threshold and correlation analyses of tinnitus strength and HT, this percept is most probably not produced by an SR-induced increase of neuronal activity but must be based on a different neuronal mechanism.

The model of SR that has been recently proposed in our group (Krauss et al., 2016, 2019) predicts that HT should be improved within the frequency range of the tinnitus percept. In accordance with the SR model, the audiometric data of almost 40,000 patients from the ENT clinic in Erlangen, tinnitus patients had significantly better HT than patients without tinnitus in the low-frequency range up to about 3 kHz, that is, in the speech-relevant frequency range (Krauss et al., 2016, 2019). Additionally, utilizing this mechanism, we proposed a therapeutic approach to tinnitus suppression, using external acoustic noise to replace the internal neuronal noise. In a pilot study, this approach was successful in patients with a hearing loss not exceeding 40 dB (Ahlf et al., 2012).

In the present study, we could demonstrate in an animal model that salicylate-induced transient tinnitus is most probably based on a different mechanism. Salicylate has been shown to act on the outer hair cell (OHC) lateral wall stiffness (Lue and Brownell, 1999), increasing the membrane conductance of the OHCs (Stypulkowski, 1990) probably via acting on the voltage sensitivity of the motor protein prestin (Oliver et al., 2001; Grosh et al., 2004; Yu et al., 2008; Yang et al., 2009). Most likely, salicylate primarily influences electromotility and OHC non-linear capacitance via a direct interaction with prestin (Greeson and Raphael, 2009). We speculate that this mechanism provides an explanation for the hearing loss induced by salicylate, but the mechanism and the site of the generation of the tinnitus percept still remain unclear (Guitton et al., 2003). The effects of salicylate are not only limited to the periphery, as a direct central effect of salicylate has also been demonstrated (Basta et al., 2008; Chen et al., 2013). Indeed, salicylate can easily get through the blood–brain barrier (Jastreboff et al., 1986) and change the delicate balance between the excitatory and inhibitory circuits in the central auditory system (Xu et al., 2005). In fact, synaptic inhibition of the auditory cortex is predominantly GABAergic (Prieto et al., 1994a,b) and an alteration of these circuits can greatly change the response properties of auditory neurons (Rajan, 1998; Wang et al., 2000, 2002) and could consequently cause tinnitus (Eggermont and Roberts, 2004; Eggermont, 2005). Instead, noise trauma-induced tinnitus could be caused by central changes arising from the noise-induced reduction of cochlear input (Norena et al., 2002; Eggermont, 2007; Moffat et al., 2009). In other words, noise trauma-induced tinnitus has to have a central origin as well. However, we hypothesize that it develops due to an indirect effect triggered by damage in the cochlea and not due to a direct effect in the brain. One could speculate that the direct central effect of the salicylate might also contribute to the difference in the frequency range between the broad tinnitus percept induced by salicylate and the narrow phantom percept induced by sound exposure (Norena et al., 2010; Ahlf et al., 2012).

For assessing any change in hearing sensitivity, we used ABR measurements. In our data, the HT of the group C (Figure 2A) decreased over time in a frequency-independent manner. In that context, Ruebhausen and colleagues (Ruebhausen et al., 2012) noted that ABR generators were, primarily, in the central nervous system, and that interaction between general anesthesia and signal processing in the auditory brain stem would be expected. They conclude that, although both isoflurane and ketamine/xylazine were glutamatergic NMDA receptor antagonists, their global effect on neural systems would be complex and not known with sufficient precision to predict how each might affect auditory processing at a threshold. In our data, the HT before any kind of treatment (salicylate or saline) is comparable in groups C and S (Figure 3A). In group C, we see a reduction of HT over time, while, in group S, we do not observe this reduction, and we even find a frequency-dependent increase of the HT at 4 kHz 2 h post-injection (Figure 3B). This indicates that the effect of salicylate not only counteracts the reduction of HT but even increases the threshold in a frequency-specific manner. The reason why we see an effect only at 4 kHz is probably due to the effect of the salicylate in the OHC, which seems to be most prominent in the range of the best hearing. This means that the increase in the membrane conductance of the OHCs due to the effect of the salicylate would generate a stronger effect on the HT in the middle frequency region. Strikingly, this is not only a group effect (as demonstrated in the HT changed) but also an effect on an individual animal, as demonstrated by the HL at 4 kHz (Figures 4A,B). Why this effect is so specific in the frequency range of best hearing, one can only speculate. It may be due to basilar membrane thickness that peaks around this frequency range (Plassmann et al., 1987), making it more stiff and, therefore, reducing the effectiveness of the weaker pull from the affected OHCs further, as force production, along the cochlea, seems to be similar (Mahendrasingam et al., 2010) and also the numbers and innervation of these cells (Wilson et al., 1991) do not seem to change over the course of cochlear frequency locations.

For the assessment of a possible tinnitus percept, we used the behavioral approach of the GPIAS paradigm. GPIAS is the most common method for tinnitus assessment because it does not require any training, avoids conditioning-related plasticity, and saves time (Turner et al., 2006). However, it is still controversial if the method is appropriate for tinnitus screening, as the “filling-in” interpretation has been questioned (Campolo et al., 2013; Radziwon et al., 2015). Furthermore, a wide range of criteria for positive tinnitus detection has been used across different laboratories, and there, still, is no consensus on a “best practice” for statistical evaluation of GPIAS results, as it exists for other behavioral paradigms (Hinkle et al., 2003). The method has also been strongly criticized for being not reliable and does not rule out the possibility detecting hearing loss rather than tinnitus. In order to overcome these limitations, Schilling and coworkers (Schilling et al., 2017) developed a new statistical approach based on the effect size of the behavioral response, used as a normalized measure for the PPI change. The method is robust and does not require any removal of outliers [which, otherwise, is a common practice (Longenecker and Galazyuk, 2011)]. The negative values of the effect size are easy to interpret and indicate less effect of the gap relative to the response of the startle pulse after treatment, which is considered to indicate a “filling” of the gap by a tinnitus percept in that frequency range and cannot be appointed to hearing loss alone. The method is, among others, applicable for salicylate or mono- or binaural noise trauma-induced tinnitus studies. We, here, see a clear effect of the salicylate 2 h but not 20 min post-injection. This is supported by the results of Jastreboff and coworkers (Jastreboff et al., 1986), who found that following i.p. injection of salicylate, the maximum levels in blood serum occurred after 1.5 h, while the levels in the perilymph and spinal fluid reached their maximum within 2–4 h. Figures 5, 6 show the effect size, a normalized measure for the PPI change in the GPIAS (Hedges, 1982). If an animal has a stronger response to the gap (lower startle amplitude) during the post-recording compared to the precondition, the effect size will be positive. This may be due to proposed cortical learning effects that lead to increased responses to the startle (Moreno-Paublete et al., 2017). This phenomenon is always present, either when treating the animals with saline (Figure 5) or without treating the animals at all (unpublished data from our lab), or even when treating the animals with salicylate. On the other hand, the absolute values of the negative effect sizes can be interpreted as tinnitus severity, as it results from a smaller response to the gap, i.e., stronger startle amplitude due to potential “filling” of the gap by the tinnitus percept in the appropriate frequency range. The learning effects in the group S can be seen in those frequency ranges where tinnitus is not so strongly perceived (Figure 6B, 1 kHz) while negative effect sizes dominate the 4 kHz range. The consequence of the proposed cortical learning effect would be that the tinnitus percept must be strong enough to overcome the learning. In other words, we probably always underestimate the tinnitus percept, but, still, at 2 h post-injection, animals show a significant negative effect size compared to the control group at 4 kHz (Figure 6B). This matches perfectly with our data of HT shift, indicating that the salicylate-induced HL and the tinnitus percept both lie in the best hearing frequency range.

In data of trauma animals (Figure 7), we found clear effects of HL dependency on the effect size of the behavioral measurements, i.e., on a significant tinnitus percept in a least one frequency or the lack thereof. Interestingly, while the maximum HL was centered around the frequency range of best hearing, comparable with the effect of salicylate, salicylate-induced tinnitus increases HL while trauma-induced tinnitus decreases the effect of the trauma on hearing thresholds.

Comparable between both tinnitus induction methods is the maximum effect size change at exactly the frequency of best hearing/maximal HL in tinnitus animals only. This indicates that the behavioral outcome of both methods is comparable.

To further demonstrate that the underlying neurophysiological mechanism – independent if our hypothesis is correct or not – is different for trauma and salicylate-induced tinnitus, we correlated the effect size with the HL for all the given frequencies after 2 h post the injection in the groups C and S. As expected, in the group C, no correlation between the two variables can be found (Figure 8A, blue line) as, in this group, no tinnitus was induced. In the group S, we also found no correlation between the tinnitus strength and the HL (Figure 8A, red line). In contrast, in noise trauma-induced tinnitus, a significant positive correlation could be found, indicating that, for strongest tinnitus percepts (negative effect size), hearing thresholds are improved (negative HL). We can, therefore, conclude that, in noise trauma/hearing loss-induced tinnitus in rodents, the tinnitus percept is most probably based on the neurophysiological mechanism of SR, but salicylate-induced tinnitus is not based on that same mechanism. The exact neurophysiological differences between both models of tinnitus induction have to be investigated in further studies.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The animal study was reviewed and approved by the Regierungspräsidium Unterfranken, Würzburg, Germany.

VL collected the data. VL and KT analyzed the data. VL, KT, and HS wrote the manuscript. All authors contributed to the article and approved the submitted version.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.