Impact of noise exposure on the circadian clock in the auditory system

The circadian clock system is present in all cells whose prime function is to regulate behavioral and physiological processes in order to maintain an internal temporal framework, even in the absence of external stimuli. Having a tightly-coupled circadian regulation over all bodily functions, including the sleep-wake cycle, metabolism, feeding, immune responses, blood pressure, urine production, cell division, hormone release, and detoxification (Gachon et al., 2004a) increases the survival and reproduction of the organism. The circadian clocks in all tissues, referred to as peripheral clocks, generate self-sustained oscillations in a tissue-specific manner, allowing distinct biological pathways to act at specific times of the day to optimize functional responses (Albrecht, 2012). The suprachiasmatic nucleus (SCN) of the hypothalamus, which is highly conserved and hierarchically organized, acts as the master clock and synchronizes the activity of peripheral clocks. For instance, the SCN regulates in a circadian manner the glucose metabolism in the liver and glucocorticoid secretion in the adrenal glands by influencing the activity of the paraventricular nucleus in the hypothalamus (Kalsbeek et al., 2004; Ishida et al., 2005). Communication between the SCN and the peripheral clocks is mediated via the autonomic nervous system or through humoral signals, such as glucocorticoids. In addition to the SCN-controlled synchronization, there are external cues that can entrain circadian rhythms. The light-dark cycle entrains the SCN by photic input from the intrinsically photosensitive retinal ganglion neurons that express melanopsin (Berson, 2003). Other environmental factors that have a strong influence on circadian rhythms are temperature and feeding (Weinert and Waterhouse, 1998; Albrecht and Oster, 2001; Ruiter et al., 2003; Roedel et al., 2006). Diurnal species are active and consume energy during the day and sleep during the night when energy-conserving processes occur. Nocturnal species are active during the night and sleep during the day. Both species have similar circadian clock systems that control both the circadian period (i.e., the length of the intrinsic clock) and the circadian phase (i.e., the clock timing).

Rest and activity cycles drive feeding and body temperature rhythms. For instance, feeding cycles are important Zeitgebers (time keepers) in the liver, pancreas, heart, and kidneys (Dibner et al., 2010), possibly through glucose sensing and Sirtuin signaling pathways (Gachon et al., 2004b). Changes in body temperature (1–4 °C) can reset peripheral clocks via heat shock factor 1 (HSF1), which regulates core clock gene expression through the binding to heat shock response elements (Reinke et al., 2008). Imposed feeding schedules in mice during resting phases can completely invert the circadian rhythms of peripheral tissues, while the SCN remains unaffected, showing that feeding cues can dominate hormonal signals (Abraham et al., 2010). Once normal feeding schemes are provided, the phase of the tissues resets in 2–3 days, showing that the SCN can rapidly take over its role as a master clock.

The importance of the SCN in synchronizing peripheral rhythms has been evidenced in lesion experiments where, in absence of the SCN, peripheral tissues become desynchronized with time (Yoo et al., 2004; Tahara et al., 2012). It is well-established that the SCN acts as a master clock instructing the peripheral tissues to maintain a circadian control over physiological responses. As demonstrated in SCN lesioned animals, circadian rhythms from peripheral clocks were found to be autonomous and self-sustained but their phase is desynchronized in a tissue-specific manner, highlighting their strong dependence on SCN-input (Yoo et al., 2004). Similarly, after a period in constant dim light, the SCN adjusts itself relatively rapidly when put back on a normal light/dark cycle, while the peripheral tissues take a longer time to reset in a manner that is tissue-specific (Mohawk et al., 2012; Sellix et al., 2012).

The timing of the circadian oscillations is determined by a transcriptional-translational feedback loop of circadian genes that has a periodicity of approximately 24 h. There are two interconnected transcription loops composed of the Clock-Bmal1 heterodimers and the other loop consisting of the Period (Per) and Cryptochrome (Cry) families that maintain the intrinsic circadian rhythms. The CLOCK/BMAL1 dimer stimulates the expression of Per and Cry gene transcription that in turn suppresses the activity of the CLOCK/BMAL1 dimer. Therefore, the Bmal1 transcription is in anti-phase with that of Per and Cry. In addition, the CLOCK/BMAL1 dimers activate retinoic orphan receptor (ROR) and REV-ERB (REV-ERBα and REV-ERBβ) which control Bmal1 and Clock gene transcription. The REV-ERB is in antiphase with BMAL1, i.e., downregulates Bmal1 transcription (Fig. 1). In addition, there are clock-controlled genes that act downstream to the core clock genes (e.g., Dbp, Hlf, Tef) and modulate physiological functions in a tissue-specific manner. As will be described below, these clock genes and clock-controlled genes are expressed in the cochlea and inferior colliculus (IC).

Circadian misalignment imposes strain on many physiological processes that can cause central and peripheral disorders. Disruptions of circadian regulation are associated with increased risk for sleep disorders, cognitive difficulties, premature death, cancer, metabolic syndrome, cardiovascular dysfunction, immune dysregulation, hormonal problems, mood disorders, and learning deficits (Evans and Davidson, 2013). It is well documented that circadian related metabolic disturbances lead to diabetes and neurological disorders, such as depression (Li et al., 2013) and schizophrenia (Johansson et al., 2016). Interestingly, shift workers are at a high risk for developing circadian-related disorders due to the discrepancies between day and night activities causing conflict with their circadian biology. These shifts in timing are primarily the result of a misalignment between the clock system and the light-dark cycle, but other cues such as temperature and feeding can also contribute to the circadian shift. Evidence shows that night-shift workers have an increased susceptibility to type 2 diabetes (Pan et al., 2011). Immune responses are also regulated in a circadian manner (Scheiermann et al., 2013) and when compared to day workers, night shift workers display a higher risk for infections (Mohren et al., 2002) and autoimmune disorders (Magrini et al., 2006).

The impact of circadian disturbances on physiology was also evidenced in animal studies. For instance, when Syrian hamsters were housed in inverted light-dark cycles on a weekly basis their life span decreased by 11% (Penev et al., 1998). A near 40% decrease in life span was found in the tau mutant hamster when maintained on a static 24 h light-dark cycle whose inherent period length is 22 h (Hurd and Ralph, 1998). The pathologies found in these hamsters were an increase in cardiomyopathy, hypotension and interstitial fibrosis. Likewise, Bmal1 mutant mice have an increased mortality and accelerated age-related pathology occurring as early as 26 weeks of age (Bunger et al., 2005). Even immune responses are susceptible to alteration in light-dark cycles. Mice with mutations in core clock genes show circadian misalignment and reduced immune function. For example, mutations in Per2 change the rhythmicity of cytokine release and the effector functions of immune cells (Arjona and Sarkar, 2006a,b; Logan et al., 2013) while mutants of Bmal1 have progressive corneal inflammation and decreased immune cell expression (Kondratov et al., 2006; Sun et al., 2006a; Sun et al., 2006b).

In this review, we summarize the recent evidence that the murine cochlea and the IC harbor a circadian machinery, and exposure to noise at nighttime is more damaging than at daytime. Interestingly, noise exposure differentially impacts the expression of clock genes in the auditory periphery and the IC. Since circadian dysregulations affect metabolism, hormone regulation and immune responses, all of which are required for maintaining normal homeostasis in the auditory system, it is likely that such circadian disturbances will have an impact on auditory function (Basinou et al., 2016). Such knowledge would help to better understand the mechanisms and risks that humans working in noisy environments, shift workers, flight crew that frequently travel across time zones may endure. Furthermore, as the majority of research performed on rodents is during daytime (their inactive period), it may not provide meaningful data for its translation to humans. This review offers a summarized basis to understand the known interactions between the circadian and the auditory system, in view of encouraging the consideration of circadian aspects in auditory research.

In two sequential studies, we evidenced that the mouse cochlea and the IC harbor an autonomous and self-sustained circadian system, as well as expressing, in a circadian manner, the core clock genes, Per1, Per2, Bmal1, and Rev-Erbα (Meltser et al., 2014; Park et al., 2016). RNA expression analyses show that the cochlea displays a very ample Per2 and Bmal1 expression, similar to that of the liver, while Per1 and Rev-Erbα were smaller [Fig. 2(a)]. The difference in peak and trough expression of Per2 in the cochlea could reach sixfold, while in the IC it would be near twofold.

Using real-time bioluminescence, rhythms of clocks in adult cochlear explants and in sections or whole IC were demonstrated using PERIOD2::LUCIFERASE (PER2::LUC) mice [Fig. 2(b)]. These mice have a Luc gene fused in frame to the endogenous mouse Per2 gene, allowing the monitoring of rhythmicity in any tissue harboring a functional clock (Yoo et al., 2004). The amplitude of oscillations of PER2 in the cochlea was near that found in the liver (Meltser et al., 2014). In contrast, oscillations in the IC dampened faster, only 30% of adult samples showed oscillations (Park et al., 2016). This low number of samples showing oscillations could relate to the culture conditions, that may not be optimal for the IC or due to the high metabolic activity of the IC (Canlon and Schacht, 1983), which may be disrupted when put into culture. For those samples that initially show oscillations they progressively dampen over time and treatment with the synchronizers, dexamethasone or forskolin, restores rhythmicity by resetting the oscillators through activating glucocorticoid receptors (Balsalobre et al., 2000) or the cAMP/PKA pathway (Yagita and Okamura, 2000), respectively. The loss of synchronicity could be due to the culture conditions or a lower number of rhythmic cells in the IC as described in Park et al. (2016). The expression of PER2 was detected primarily in the spiral ganglion neurons (SGN) and in hair cells, whereas it was homogeneous in the IC (rostral to caudal).

To what degree the circadian oscillations observed in cochlear or IC cultures can be extrapolated to the in vivo models remains to be established. It is well known in the circadian field that simply the fact of placing an organ on a culture membrane will reset its clock even if it was dysfunctional in vivo (e.g., in organs from SCN ablated animals) (Tahara et al., 2012). Thus, in vitro manipulations can impact on the read-out. Furthermore, the amplitude of PER2 will differ depending on the time when the organ is placed in culture. Organs collected at nighttime show greater PER2 rhythms than those collected at daytime (Yoshikawa et al., 2005), which is also observed for the cochlea (Meltser et al., 2014). An experiment that would demonstrate the relationship of the in vivo vs in vitro PER2 oscillations would be to monitor PER2::LUC rhythms from the cochlea in the living animals, for instance, by delivering luciferin transtympanically as previously described (Kanzaki et al., 2012), and compare interventions that can be performed both in vivo and in vitro (e.g., drug administration).

A unique property of the cochlea is the tonotopic arrangement, from the base (high frequencies) to the apex (low frequencies), allowing for frequency selectivity (Rosenblatt et al., 1997; Davis, 2003; Ricci et al., 2003). When observing PER2::LUC oscillations in the cochlea at the cellular level, it was found that multi-phased cellular clocks were arranged in a tonotopic manner (Park et al., 2017). The oscillations started at the apex (low-frequency region) and traveled toward the base (high-frequency region), with a 3 h phase difference between the cellular oscillators in the apical and middle regions. It was hypothesized that the cochlea could have differentially phased cellular networks that integrate into a single cochlear rhythm. To test this, cochlea were dissected into apical, middle and basal turns and oscillations measured separately. The phase of the dissected middle region was delayed by 3.49 h compared to the whole cochlea. The basal part also showed differences in phase and period when compared to the whole cochlea (Park et al., 2017). Thus, multi-phased cellular oscillators are organized in region-specific networks that have different degrees of synchrony along the cochlea. These findings illustrate coupling between the different regions is required for an integrated and coherent circadian rhythm in the cochlea.

In the presence of either the voltage-gated potassium channel blocker (TEA) or the extracellular calcium chelator (BAPTA) these cochlear oscillations were inhibited. These findings indicate that multiple ion channels contribute to the maintenance of coherent rhythms. These findings highlight a dynamic regulation and longitudinal distribution of cellular clocks in the cochlea, but function remains to be uncovered. Interestingly, this type of spatiotemporal organization has been demonstrated in the SCN, where the expression of PER1 rhythms peaks earlier in the dorsomedial part compared to the ventrolateral part (Yamaguchi et al., 2003; Nakamura et al., 2005). Yet, the synchronization of the ventral oscillators was not affected when the dorsal part of the SCN was removed (Yamaguchi et al., 2003). As for the cochlea, an integration of signals from the different cochlear regions is needed to generate a coherent synchrony of rhythms at the organ level. The physiological significance of having cellular clocks in the cochlea with different peak expression times may be related to a time-dependent regulation of frequency sensitivity, as found for spontaneous otoacoustic emissions in humans (Haggerty et al., 1993). Another reason could be to distribute, in a time dependent manner, energy expenses throughout the organ in order to avoid metabolic exhaustion.

Consistent with a circadian regulation of cochlear function, the same noise trauma applied during the day or during the night yielded different outcomes. When mice are exposed to a 1 h noise [100 dB sound pressure level (SPL), 6–12 kHz band noise] during the night (active phase), a greater threshold shift of the auditory brainstem response is found two weeks post exposure when compared to those exposed during the day (inactive phase) (Meltser et al., 2014). A potential mechanism to explain the day and night differences in response to noise trauma included neurotrophic signaling in the cochlea (Meltser et al., 2014). Neurotrophins are important regulators of synaptogenesis and synaptic plasticity in the cochlea. Two important neutrophins, namely, neurotrophin-3 (NT-3) and brain derived neurotrophic factor (BDNF), play an important role during cochlear development and in adult auditory physiology. The study from Meltser et al. revealed that the induction of Bdnf transcription in response to noise trauma differed after day or night exposure. Day-time noise exposure (inactive phase) causes a 30-fold increase in Bdnf mRNA transcript levels, whereas no increase after nighttime noise (inactive phase) is found. These findings indicate that the cochlea is incapable of triggering a BDNF-dependent protective response after noise trauma delivered at night, which could explain the subsequent increased vulnerability. Treatment before nighttime noise exposure with dihydroxyflavone (DHF), a selective agonist of TrkB (Jang et al., 2010), was sufficient to restore the recovery of hearing thresholds to a level comparable to day-time noise exposure (inactive phase). Interestingly, nighttime noise exposure caused a two-fold reduction in the number of synaptic ribbons two weeks after noise exposure and DHF pre-treatment partially protected them (Meltser et al., 2014). These findings provide evidence of the involvement of neurotrophins in the circadian recovery from noise trauma.

In contrast to the cochlea, Bdnf did not show any differential induction in the IC after daytime or nighttime noise exposure (Table I). Both daytime and nighttime noise exposure caused a 15-fold increase in Bdnf, suggesting that neurotrophic signaling in the cochlea and IC respond differently to noise trauma. The increase in Bdnf mRNA in the IC could contribute to increased plasticity and neuroprotection against cell death known to occur in the IC after noise trauma (Coordes et al., 2012). Thus, the plastic responses to noise exposure in the IC should not be differentially altered at different times of the day. These findings also highlight differences in basal levels of Bdnf expression in the cochlea and the IC. Bdnf expression shows a circadian pattern in the cochlea, but not the IC, according to statistical tests using CircWave (Hut, 2013). In the cochlea, Bdnf expression peaks at night and is low in the day, allowing for a greater dynamic range to trigger its expression in response to daytime noise exposure. In contrast, in the IC, the expression is low throughout a 24 h cycle, which may underlie the ability of the IC to induce Bdnf expression in response to both daytime and nighttime noise exposure.

Other reports have attempted to consider circadian aspects in noise exposure but without evidencing clear day/night noise differences in noise sensitivity (Sheppard et al., 2018). Of note, the study from Sheppard et al. performed a lower intensity noise exposure (10–20 kHz, 75 dB SPL) that encompassed 12 h (either during daytime or nighttime) instead of 1 h exposure at specific times of the day as performed by Meltser et al. (near peak and trough of PER2 expression). Such noise exposure did not cause any immediate threshold shifts after day or night exposure and thus, presumably, no peripheral damage had occurred. It is thus possible that the circadian effects of noise do not apply to low- or extremely high intensity levels and that such circadian differences may occur within a dynamic range that remains to be defined. In addition, Sheppard et al. investigated Sprague Dawley rats, which circadian biology may differ from mice. For instance, the pattern of glucocorticoid secretion varies considerably within and between rat and mouse strains and this can have a clear impact on noise-induced stress responses (Lightman and Conway-Campbell, 2010). Given the clear implications of glucocorticoids in circadian rhythms and in auditory physiology, extrapolation of the Meltser et al. results to other strains has to be carefully performed.

A gene expression analysis of circadian genes from samples collected around the clock show that there are several differences between the response to daytime (inactive phase) and nighttime (active phase) noise trauma between the cochlea and the IC (Table I). When analyzing circadian responses to nighttime noise, the IC often showed an inversed circadian response compared with the cochlea. For example, the amplitude response of Per1 mRNA transcripts levels were greater after night noise exposure, whereas it was not changed in the cochlea. Per2 mRNA transcripts levels was also increased in the IC after night noise and decreased in the cochlea. These findings were also reinforced by the ex vivo results showing the greater amplitude of PER2::LUC rhythms in the IC after night noise, while night noise decreased the oscillations in the cochlea (Meltser et al., 2014). Conversely, night noise exposure suppressed the Rev-erbα circadian profile in the cochlea, whereas it was not affected in the IC. After daytime noise exposure (inactive phase), there were also several differences in the clock gene expression between the IC and the cochlea. The transcript levels of Per1, Per2, and Bmal1 mRNA did not show any changes in amplitude after daytime noise, while the cochlea showed an increase in Per1 and a decrease in Per2 and Bmal1 expression. Both the IC and the cochlea showed a decrease in Rev-erbα amplitude.

Compensatory mechanisms could be responsible for the inversed circadian response of the IC to nighttime noise trauma, making it difficult to determine if the IC is dependent or independent of the cochlea. Interestingly, hyperactivity in the IC and suppressed auditory nerve responses after noise trauma were found in several studies (Salvi et al., 2000; Niu et al., 2013), reinforcing the notion that central and peripheral structures respond differentially to noise exposure. Together, the electrophysiological responses (Salvi et al., 2000; Niu et al., 2013) and our mRNA data suggest that peripheral and central structures do not respond in parallel to noise exposure. Such findings illustrate the complex differences in molecular responses operating in the two organs in response to noise trauma at different times of the day.

The cochlea and the IC both demonstrate circadian oscillations evident at both the mRNA and protein level. The circadian oscillations from the cochlea and the IC differentially respond to daytime (inactive phase) and nighttime (active phase) noise exposure. However, the manners by which the clock genes in the cochlea and IC react to noise trauma are often inverted to each other. These differences between the cochlea and IC agree with electrophysiological findings that demonstrate central and peripheral differences in response to noise trauma. These differences will be relevant for understanding the molecular mechanisms underlying noise-induced trauma to the peripheral and central auditory system.

Research in the Canlon lab is funded by Communication Disorders of the National Institutes of Health Grant Nos. R21DC013172 and 1R56DC016415-01, the Swedish Medical Research Council Grant No. K2014-99X-22478-01-3, Karolinska Institutet, Tysta Skolan, Hörselforskningsfonden, Magnus Bergvalls, and the EU Grant No. H2020-MSCA-ITN, ESIT, C.R.C - project # 722046. B.C. and C.C. also received funding from the Office of the Assistant Secretary of Defense for Health Affairs through the Neurosensory and Rehabilitation under Award No. W81XWH-16-1-0032. Opinions, interpretations, conclusions, and recommendations are those of the authors and are not necessarily endorsed by the Department of Defense. All authors declare that there are no competing financial interests.