IntroductionMilitary personnel, mostly those in combat, often develop noise-induced hearing loss (NIHL) (Cave et al., 2007; Helfer et al., 2011), a condition exacerbated in some individuals by tinnitus (a phantom buzzing or ringing sensation) and hyperacusis (sounds perceived as intolerably loud or even painful) (Gilles et al., 2012; Henry et al., 2014; Sun, 2011). Recent evidence suggests that tinnitus and hyperacusis arise from maladaptive neuroplastic changes in a distributed neural network that involves portions of the central auditory pathway plus direct and indirect neural connections with other brain regions associated with arousal, stress, anxiety and attention (Auerbach et al., 2014; Baguley et al., 2013; Dornhoffer et al., 2006; Leaver et al., 2016; Lockwood et al., 1999; Moller, 2003). A common feature of intense noise or ototoxic drug exposure is that they reduce the neural output from the cochlea (hypoactivity). To adapt to this altered acoustic input homeostatic mechanisms in the central nervous system “kick in” and increase the gain at successively higher levels of the auditory pathway (Auerbach et al., 2014; Turrigiano, 1999). By the time the neural activity reaches the medial geniculate body (MGB) and auditory cortex (AC) sound-evoked responses are generally larger than normal, a phenomenon referred to as enhanced central gain (Brotherton et al., 2015; Chambers et al., 2016; Chen et al., 2016; Qiu et al., 2000; Salvi et al., 1990). In the case of salicylate-induced hearing loss where there is consistent evidence of tinnitus and hyperacusis, our functional magnetic resonance imaging (fMRI) studies revealed neural hyperactivity in an auditory network consisting of AC, MGB and inferior colliculus (IC) (Figure 1A) plus an emotional center, the amygdala (AMY), that is linked to the AC (Antunes and Moita, 2010; Chen et al., 2015; Newton et al., 2004) (Figure 1). Unexpectedly, salicylate-induced hyperactivity was also observed in the reticular formation (RF), an area of the brain involved in arousal and sleep, and the paraflocculus (PFL), a part of the cerebellum involved in head-eye motor control and which receives some auditory inputs (Azizi and Woodward, 1990; Horikawa and Suga, 1986). When a functional connectivity analysis was performed on our fMRI data, the AC was found to be strongly coupled to both the RF and PFL.
Paraflocculus and Reticular FormationAlthough the PFL is best known for its role in coordinating eye and head movements (Nagao, 1992; Rambold et al., 2002), some neurons in this region respond to sound through direct and indirect connections with neurons in the cochlear nucleus (CN), IC and AC (Aitkin and Boyd, 1978; Azizi et al., 1990; Azizi et al., 1985; Azizi et al., 1981; Horikawa et al., 1986; Huang and Liu, 1990; Huang et al., 1982; Kawamura, 1975; Lockwood et al., 1999; Misrahy et al., 1961; Morest et al., 1997; Snider, 1950; Snider, 1948). The cerebellum in turn can influence the auditory system (Rossi et al., 1967; Velluti and Crispino, 1979). Importantly, recent studies suggest that the PFL is involved in gating or regulating tinnitus and hyperacusis (Bauer et al., 2013; Chen et al., 2015). In the case of chronic noise-induced tinnitus, manganese enhanced magnetic resonance imaging (MEMRI) revealed enhanced spontaneous activity in the PFL. Moreover, behavioral evidence of chronic noise-induced tinnitus was abolished by lesioning or inactivating the PFL (Bauer et al., 2013; Brozoski et al., 2007; Brozoski et al., 2013). Collectively, the imaging and behavioral results suggest that the PFL is involved in gating or modulating tinnitus and hyperacusis. In the case of salicylate-induced tinnitus, fMRI revealed enhanced spontaneous activity in the PFL and stronger functional coupling between the PFL and the AC as well (Chen et al., 2015) (Figure 1B). Salicylate also enhances spontaneous activity in the RF and increased the functional coupling between the RF and the AC (Figure 1B). The RF plays an important role in generating the acoustic startle reflex, a strong reflex movement of the head, neck and eyes elicited by brief sound at intensities above ~75 dB SPL. Acoustic signals for eliciting the startle reflex are routed through the cochlear nerve root and the PnC to facial and spinal cord motor neurons that produce abrupt motor responses linked to the acoustic startle reflex (Davis et al., 1982; Lee et al., 1996). Interestingly, high doses of sodium salicylate (SS) greatly enhance the amplitude of the acoustic startle reflex (Lu et al., 2011). Recent research suggests that the preceding effects could be mediated locally or modulated by descending effects from the IC, AC or AMY (Bowen et al., 2003; Chen et al., 2012; Du et al., 2011; Yeomans et al., 2006).
Altogether, the composite results reviewed above suggest that the perceptual features or salience of SS-induced tinnitus may arise from enhanced spontaneous activity, decreased inhibition, increased excitation and/or increased functional connectivity in a distributed network involving the AC, AMY, RF and PFL. While both MEMRI and fMRI imaging techniques revealed enhanced activity in the PFL and RF in animals with noise-induced and salicylate-induced tinnitus, these imaging techniques cannot identify the exact nature of the electrophysiological changes such as distinguishing between hyperactivities resulting from decreased inhibition or increased excitation. To provide relevant physiological measurements relevant to these imaging studies, we carried out electrophysiological recordings from the PFL and the PnC in rats before and after treatment with a high-dose of salicylate where hyperactivity was observed by fMRI measurement. The rationale for recording from the PnC is based on the fact that giant neurons in this region play a critical role in generating the acoustic startle reflex (Lingenhohl and Friauf, 1992), which has been shown to become hyperactive after high-dose SS (Lu et al., 2011). While hearing loss may be a prerequisite for developing tinnitus and hyperacusis, it may be necessary but not a sufficient condition for developing these disorders because many individuals with hearing loss do not develop tinnitus or hyperacusis. A number of studies suggest that stress combined with auditory impairment may interact synergistically to promote the development of tinnitus and/or hyperacusis or alternatively to contribute to the severity of these conditions such as becoming more bothersome (Alpini and Cesarani, 2006; Hasson et al., 2011; Kim et al., 2012; Mazurek et al., 2012). Stress increases corticosterone stress hormones and enhances sound-evoked activity in the auditory cortex (Ma et al., 2015). High doses of salicylate also increased c-fos expression (a transcription factor that regulates gene expression) in areas of the brain associated with stress (Wallhausser-Franke, 1997; Wallhausser-Franke et al., 2003). Salicylate also induces cochlear hearing loss and enhances sound evoked activity in the central auditory system; the combination of these effects may help to explain why salicylate reliably induces tinnitus and hyperacusis. To determine if salicylate-induced tinnitus and hyperacusis are correlated and linked to stress, we treated rats with different doses of SS and measured serum corticosterone stress hormone levels at various times following the salicylate treatment.
DiscussionPrevious studies have shown that SS causes hyperactivity in selected regions of the central auditory pathway and the AMY (Figure 1) (Chambers et al., 2016; Chen et al., 2013; Lu et al., 2011; Norena et al., 2010; Qiu et al., 2000; Stolzberg et al., 2011; Syka and Rybalko, 2000; Zhang et al., 2011). Our more recent fMRI studies identified two additional areas outside the classical auditory pathway where SS induced hyperactivity, named the PFL and RF (Chen et al., 2015). To confirm and characterize the electrophysiological changes that occur in these non-auditory regions, we recorded LFP and MUC firing patterns from the PFL and PnC before and after SS treatment and measured the changes in serum corticosterone to evaluate the potential contribution of this stress hormone. The three new findings of the current study are that: (1) high-dose SS causes a subset of neurons in the PFL and PnC to become hyperactive to suprathreshold acoustic stimuli, (2) the firing rate and SS-induced hyperactivity in the PnC increases significantly with increasing ISI and (3) high doses of SS (≥ 150 mg/kg) known to induce tinnitus and hyperacusis cause a large increase in corticosterone stress hormone levels. Our results suggest that SS-induced hyperactivity observed in selected parts of the central nervous system may be linked to the upregulation of circulating corticosterone stress hormone or the receptors to which it binds (Ma et al., 2015).
Auditory PFLSeveral parts of the cerebellum respond to auditory stimulation including the PFL (Altman et al., 1976; Azizi et al., 1990; Azizi et al., 1985; Azizi et al., 1981; Buser and Franchel, 1960; Huang and Liu, 1985; Huang and Burkard, 1986; Huang and Liu, 1991; Lockwood et al., 1999; Lorenzo et al., 1977; Mihailoff et al., 1981; Misrahy et al., 1961; Petacchi et al., 2005; Snider and Stowell, 1944; Snider et al., 1964; Stowell, 1942; Wolfe, 1972; Woody et al., 1999). Noise bursts evoked distinct LFPs in ~60% (86/144) of our recording sites in the PFL (Figure 2). The latency of the LFPs evoked by 100 dB noise burst presented to the contralateral ear and the onset response of most PSTH MUC were relatively long on the order of 10 ms; these results are consistent with the long latency responses seen in the paraflocculus of the mustache bat (Horikawa et al., 1986). The long-latency PFL-responses may reflect a long signaling pathway through high auditory levels like the AC, which showed SS-enhancing effect, whereas the short-latency PFL-responses (Fig. 5C) may reflect a short signaling pathway through lower levels of the auditory pathway such as IC and even more peripheral parts of the auditory pathway showing no SS-enhancing effect. In addition, the PSTH profiles and the peaks in the LFP waveforms were relatively broad. Taken together, these results suggest that the inputs to the contralateral PFL were relayed to it from one or more higher auditory centers such as the IC or AC (Azizi et al., 1985; Brodal, 1968). Moreover, relatively high intensity noise bursts (>60 dB) were needed to elicit a robust LFP (Figure 3). PFL MUC were most responsive to low frequency tone bursts and their tuning was relatively broad (1–8 kHz, see Figure 6), consistent with results from mustache bat (Horikawa et al., 1986).
LFP amplitudes and MUC discharge rates increased significantly after SS treatment. The nature of the SS-effects was linked to the latency of the response. The early response (<15 ms) was typically reduced whereas the late response was enhanced post-SS (Figure 2C, 5C, 9A and B, 10A). The early PFL-response may reflect direct input from the cochlea and/or cochlear nucleus, regions where the sound-evoked responses are reduced post-SS. The increased responsivity of the late component post-SS could be due to diminished inhibition or increased excitation within the PFL itself or to the AC or IC that relay information to the PFL. It has to be noted that this facilitative effect occurred only in half of the recordings. Because SS induces robust hyperactivity in the AC, it is possible that the ensuing hyperactivity in the PFL is influenced by the AC neuronal activity. However, this interpretation is tempered by the fact that electrical stimulation of the AC and IC produces a mixture of excitatory and inhibitory responses in the PFL (Azizi et al., 1985). In the AC and AMY, SS-treatment caused the frequency receptive fields of low frequency neurons to shift upward toward the mid-frequencies (Chen et al., 2012; Lobarinas, 2006; Stolzberg et al., 2011; Yamamoto et al., 2007). However, there was little evidence of an upward shift in the PFL (Figure 6).
At a systems level, the PFL receives multiple types of sensory information, including inputs from vestibular and auditory centers. The prevailing view is that this information is integrated and processed in order to fine tune head and eye movements such as those used to direct the head and eyes toward or away from a sound source, such as the echo orientation system in bats (Horikawa et al., 1986). Interestingly, some vestibular schwannoma patients develop gaze-evoked tinnitus, a condition in which deviation of the eyes from straight ahead alters the loudness or pitch of tinnitus (Cacace et al., 1996; Coad et al., 2001; Lockwood et al., 2001; van Gendt et al., 2012). Eye movements have also been shown to modulate tinnitus in non-tumor patients (Simmons et al., 2008). Taken together, these results suggest that the eye-movement circuitry of the PFL may be involved in modulating tinnitus.
One consequence of the SS-induced hyperactivity in the PFL is that it could speed up and/or facilitate such motor activity. Consistent with this view, we have found using an operant-conditioning paradigm that assesses loudness growth, that SS induces a reduction in subjects’ reaction times (shorter latency) in response to sounds of moderate to high intensity (Chen et al., 2015). Functional imaging studies have also suggested that the PFL is involved in gating or modulating tinnitus (Brozoski et al., 2007; Chen et al., 2015). This interpretation was supported by behavioral studies showing that tinnitus was suppressed or enhanced by inactivating or activating the PFL (Bauer et al., 2013; Brozoski et al., 2007; Brozoski et al., 2013). Noise-induced spontaneous hyperactivity in the IC, considered a neural correlate of tinnitus, is enhanced by removal of the PFL whereas PFL removal had no effect on spontaneous activity in normal animals (Vogler et al., 2016). If spontaneous hyperactivity in the IC is a neural indicator of tinnitus (Manzoor et al., 2013), then these results suggest that PFL removal would enhance tinnitus, an interpretation at odds with behavioral data showing that PFL inactivation suppresses tinnitus (Bauer et al., 2013).
PnC HyperactivityA loud, abrupt and unexpected sound induces the acoustic startle reflex. The acoustic startle circuit consists of cochlear root neurons, neurons in the PnC, and motor neurons. Repetitive sound stimuli induce startle reflexes with long ISI, but not short ISI. This gating effect appeared to result from the PnC response which increases with increase of ISI (see Fig. 9). In addition, only high level sounds above 70 dB SPL induce startle reflexes. This is consistent with the high thresholds (>65–70 dB SPL) observed in our PnC responses (Fig. 7). High doses of SS significantly enhance the amplitude of the acoustic startle reflex (Chen et al., 2014; Du et al., 2011). Since SS greatly reduces the neural output of the cochlea at suprathreshold levels, the mechanisms for enhancing acoustic startle amplitudes must occur beyond the cochlear nerve root either in the PnC or the facial and spinal motor neurons that drive the musculature responsible for the startle reflex (Davis et al., 1982; Lee et al., 1996). Our results show that SS significantly increased the discharge rate of PnC MUC to high-intensity noise bursts (Figure 8A–B), intensities that are typically needed to induce the startle reflex. SS shifted the low-intensity portion of our discharge rate-intensity functions to the right 20–30 dB. This threshold shift is consistent with and likely due to the cochlear hearing loss induced by SS (Chen et al., 2013; Stolzberg et al., 2011). Since we are unaware of any evidence indicating that SS enhances motor neuron output, our results suggest that the SS-induced increase in startle amplitude has its origins in the PnC or hyperactivity from descending inputs to the PnC from the AC, AMY or IC (Chen et al., 2015). This interpretation is consistent with several features of PnC discharge patterns. First, the neural output of the PnC and the amplitude of the startle reflex continue to increase with ISI extending out tens of seconds or more (Hoffman and Fleshler, 1963; Ison and Hammond, 1971). Second, PnC MUC responded most robustly to tone bursts below 10 kHz (Figure 10) consistent with large startle responses to lower frequency tone bursts (Blaszczyk and Tajchert, 1997). Because SS mainly enhanced the discharge to tone bursts below 10 kHz, one might expect that SS would selectively increase startle amplitude to low (<10 kHz), but not to high frequency tone bursts. The effects of SS on PnC firing patterns were duration dependent. SS mainly enhanced PnC firing during the late portion of the PSTH and this effect was ISI dependent. Based on these results, SS would presumably cause the greatest increase in startle amplitude at long ISI and long duration noise bursts. The effects of SS on startle amplitude would likely be minimal for short duration noise bursts and short ISIs (Figure 9).
Corticosterone UpsurgeBecause high doses of SS produced strong c-fos immunolabeling in brain areas associated with stress, anxiety and pain (Wallhausser-Franke, 1997; Wallhausser-Franke et al., 2003), we hypothesized that SS would significantly increase corticosterone levels. Our results are generally consistent with these predictions. High doses of SS sufficient to induce tinnitus (Guitton et al., 2003; Jastreboff et al., 1988; Lobarinas et al., 2004; Ruttiger et al., 2003) caused a large increase in serum corticosterone levels whereas low doses incapable of inducing tinnitus had almost no effect on corticosterone levels. The 250 mg/kg SS dose caused ~18-fold increase in corticosterone over baseline (Figure 11) whereas 150 mg/kg induced ~8 fold increase. The large variability associated with the 150 mg/kg dose of SS may be related to the fact that this dose lies near the threshold for inducing tinnitus (Lobarinas et al., 2004). Twenty-four hours after the cessation of salicylate, corticosterone levels decreased to near control levels consistent with previous behavioral studies showing the disappearance of tinnitus one day after discontinuing SS (Guitton et al., 2003; Lobarinas et al., 2004). Corticosterone levels in response to a stressor normally have a half-life of 25–60 minutes (Haemisch et al., 1999; Vachon and Moreau, 2001); however, our results show that corticosterone levels did not approach baseline levels until 24 h post-treatment. These results indicate that high doses of salicylate can disrupt the hypothalamic-pituitary-adrenal (HPA) system for a prolonged period of time resulting in abnormally high corticosterone levels. Corticosterone is normally bound to carrier proteins and is inactive as it is transported through the circulatory system. Corticosterone binding globulin (CBG), the dominant carrier protein, binds approximately 90% of circulating corticosterone. When corticosterone is released from CBG it can bind to glucocorticoid receptors and exert numerous biological effects (Ekins, 1992). Interestingly, in fluorometric assays, salicylate is used to free corticosterone from CBG (Erkens et al., 1998; Kane, 1979; Shrivastav, 2004). This raises the possibility that high levels of serum SS promotes the release of corticosterone from CBG resulting in a large unbound pool of serum corticosterone consistent with our results (Figure 11A–B). The SS-induced increases in corticosterone may also contribute to sound-induced hyperactivity in the AC, MGB and AMY (Chen et al., 2013; Lu et al., 2011; Yang et al., 2007), the same regions where SS increased c-fos expression (Wallhausser-Franke, 1997). Interestingly, rodent repellent stressors that increase serum corticosterone significantly increased the amplitude of auditory evoked responses (Mazurek et al., 2010). Moreover, exogenous corticosterone enhances auditory evoked potential amplitudes (Maxwell et al., 2006). Taken together, these results suggest that the hyper or altered activity is due to increases in corticosterone levels.