Metalloproteinases and Their Associated Genes Contribute to the Functional Integrity and Noise-Induced Damage in the Cochlear Sensory Epithelium

Acoustic overstimulation causes a time-dependent change in cochlear MMP activity

To determine the role of MMPs in noise-induced cochlear pathogenesis, we examined the MMP activity in normal and noise-traumatized rat cochleae. In normal cochleae, we detected strong MMP activity (Fig. 1A). To provide a context for the assessment of MMP activity in the cochlea, we also examined the MMP activity in the brain and kidney. We found that the MMP activity in the cochlea was comparable with that observed in the kidney. In both tissues, the activity was stronger than that in brain (one-way ANOVA, F = 17.62, df = 2,12, p = 0.0003; Tukey's multiple comparison test: cochlea vs brain, p < 0.05; kidney vs brain, p < 0.05; cochlea vs kidney, p > 0.05). The detection of stronger MMP activity in the kidney than in the brain is consistent with the observations of a previous study in rodents (Pagenstecher et al., 2000).

We then compared the MMP activity between the normal and the noise-traumatized cochleae collected at 2 h or 1 d after noise exposure. We found that the samples of the 2 h post-noise exposure group had a similar activity level to that in the samples of normal cochleae (Fig. 1B). In contrast, the MMP activity was reduced in the samples of the 1 d post-noise exposure group (Student's t test, p = 0.035; Fig. 1C). This observation suggests that acoustic trauma causes a time-dependent change in cochlear MMP activity.

Exposure to intense noise causes transcriptional changes in MMPs and their related genes

To investigate the molecular basis underlying the change in MMP enzymatic activity, we examined the transcriptional changes in the genes related to the MMP family in normal and noise-traumatized cochlear sensory epithelia.

Expression changes in Timp1 and Timp2 after acoustic trauma

MMP activity is regulated in part by endogenous inhibitors, including TIMPs. To determine whether acoustic trauma causes changes in TIMP expression, we examined the changes in the protein expression of two TIMPs (Timp1 and Timp2) at 2 h after noise exposure. Timp1 was selected for analysis because it was upregulated at the mRNA level in both the apical and basal partitions of the sensory epithelium at 2 h after noise exposure (Table 3). Timp2 was selected because it was highly expressed in the normal cochlea (Table 2) and was downregulated at both 1 d and 28 d after noise exposure. An immunolabeling assay was used to assess the spatial distribution of the changes in the organ of Corti.

In the normal cochlea, Timp1 immunoreactivity was weakly present in the cytoplasm of the sensory cells and the supporting cells (Fig. 5B, see A for the nuclear morphology of the cells in the same area). In noise-traumatized cochleae, we observed a cell-state-dependent change in Timp1 immunoreactivity. To define the progression of sensory cell degeneration, we used nuclear morphology as a marker because this phenotype of sensory cell degeneration is an early event of cell damage that progresses in a step-by-step manner (Yang et al., 2004; Hu et al., 2006). Compared with the staining intensity of Timp1 in sensory cells with normal nuclear morphology, sensory cells that had slight nuclear condensation (an increase in propidium iodide fluorescence but a decrease in the nuclear size) exhibited reduced Timp1 immunoreactivity (Fig. 5D–F). In contrast, Timp1 immunoreactivity was increased in sensory cells with advanced nuclear condensation as evidenced by the appearance of small nuclear fragments (Fig. 5G–I). Noticeably, not all cells with advanced nuclear condensation exhibited increased Timp1 immunoreactivity, suggesting that the increase in Timp1 immunoreactivity is a transient event of cell degeneration or that different cells undergo different molecular changes. We quantified the number of sensory cells that exhibited nuclear malformation and the numbers of cells that showed increased Timp1 immunoreactivity and found the ratio to be ∼3:1 (Fig. 5C). Because only a few sensory cells exhibited the early signs of nuclear condensation, we did not quantify their number. Together, this observation suggests the presence of a biphasic expression change in the Timp1 protein with an early downregulation and a subsequent upregulation during sensory cell degeneration.

Timp2 immunoreactivity was only weakly present in the sensory cells of the normal sensory epithelium (Fig. 6A–C). After acoustic trauma, an increase in Timp2 immunoreactivity was found in the circumferential ring of outer hair cells, in which the cells form junctions with Deiters cells (Fig. 6E). The double labeling of the tissue with propidium iodide revealed that all of the cells with increased immunoreactivity in the circumferential rings exhibited malformed nuclei (Fig. 6D–F). Again, not all sensory cells with nuclear condensation exhibited increased Timp2 immunoreactivity. The quantitative analysis of the number of sensory cells with nuclear malformation and the number of cells with increased Timp2 immunoreactivity revealed a ratio of 3.5:1 (Fig. 6G), similar to the ratio observed for Timp1 staining.

Nuclear condensation and fragmentation are morphological signs of apoptosis. To confirm that the cells with these nuclear changes were dying by apoptosis, we stained the tissues with an antibody against Bax, a proapoptotic protein (Lalier et al., 2007). We found that all of the cells showing condensed or fragmented nuclei displayed strongly increased Bax immunoreactivity (Fig. 6H,I). This observation is consistent with our previous observations using other apoptosis markers (the activation of caspase-3 and the terminal deoxynucleotidyl transferase-mediated dUTP-digoxigenin nick end labeling) in sensory cells showing nuclear condensation (Hu et al., 2009; Cai et al., 2012). Together, the observations suggest that Timp1 and Timp2 are involved in the apoptotic process in noise-induced cochlear damage.

To verify the specificity of the antibodies used in the current investigation, we confirmed the molecular weights of the proteins targeted by the Timp1, Timp2, and Bax antibodies using a Western blotting assay. We detected a clear 24 kDa band for Timp2 and a 21 kDa band for Bax, consistent with the molecular weights of these two proteins. In the cochlear tissue, we did not identify the Timp1 band, possibly because of insufficient protein level for the Western blotting assay. We therefore used HEK293 cell lysate (sc-110547; Santa Cruz Biotechnology) for Timp1-positive control and found a strong 30 kDa band, which is consistent with the molecular weight of this protein. In addition to Western blotting for primary antibody assessment, we examined the nonspecific staining of the secondary antibodies by omitting the primary antibody and detected no clear fluorescence in the tissues.

MMP inhibition compromises cochlear function

Given the current finding that acoustic overstimulation altered MMP activity, we sought to determine whether the change in MMP activity affected cochlear function. To this end, we inhibited MMP activity with doxycycline, a broad-spectrum inhibitor of MMPs that has been used in both clinical and basic research for the inhibition of MMP activity (Clark et al., 1997; Hanemaaijer et al., 1998; Lee et al., 2001; Shlopov et al., 2001; Bedi et al., 2010). We used a dose of 50 mg · kg⁻¹ · d⁻¹ for 7 d, a treatment paradigm that has been shown to be effective for MMP inhibition in rats (Curci et al., 1998; Nogueira et al., 2011).

The ABR thresholds were measured before and after the treatment. A two-way ANOVA (frequency × time) revealed a significant time effect (F = 34, df = 2,110, p < 0.0001; Fig. 7A). Compared with the pretreatment thresholds, the average threshold shifts were 15.0 ± 18.9 to 25.8 ± 21.3 dB (mean ± SD) at the time of the completion of the doxycycline treatment (Bonferroni's post hoc test, p < 0.05–0.001, depending on testing frequency). These threshold shifts partially recovered 7 d after the completion of the drug treatment, resulting in reduced threshold shifts ranging from 7.5 ± 21.8 to 17.1 ± 19.9 dB (mean ± SD). At this time point, the threshold shifts were significant at only two tested frequencies (5 and 10 kHz, Bonferroni's post hoc test, p < 0.05). The control animals, which received a saline treatment with an identical application paradigm, exhibited no significant threshold shifts at the time of the completion of the treatment (F = 3.76, df = 1,45, p > 0.05; Fig. 7B). Together, these results indicate that the inhibition of MMP activity compromises cochlear function.

Many ototoxic agents affect cochlear function by affecting sensory cell viability. To determine whether doxycycline treatment at the current dosage compromised sensory cell viability, we examined nuclear morphology, an indicator of cell damage, 14 d after doxycycline treatment. Compared with the normal control cochleae that received only a saline treatment, the doxycycline-treated ears exhibited no significant increase in the number of missing cells (Fig. 7C, one-way ANOVA, Kruskal–Wallis test, p > 0.05). This result indicates that, at the currently used dosage, the doxycycline treatment did not cause significant sensory cell death. Because MMP inhibition has been shown to cause sensory cell death in cultured cochlear sensory tissues (Setz et al., 2011), we wanted to determine whether the lack of cell death after the doxycycline treatment was attributable to a dosage that was insufficient to induce cell death. We therefore increased the dose from 50 to 100 mg · kg⁻¹ · d⁻¹ for 7 d. With this high-dose treatment, we observed a significant difference in the numbers of missing cells between the treated and the control ears (one-way ANOVA, Kruskal–Wallis test, p < 0.05), indicating that a high dose of doxycycline is able to compromise cell viability.

To confirm that the doxycycline treatment inhibited MMP activity, we examined the cochlear MMP activity immediately after the completion of the doxycycline treatment (50 mg · kg⁻¹ · d⁻¹ for 7 d). We observed a decrease in MMP activity compared with the saline-treated control ears (Fig. 8A), although the difference was not statistically significant (Student's t test, p = 0.13) because of large individual variability. We also examined the mRNA expression levels of four MMP genes (Mmp9, Mmp11, Mmp13, and Mmp14) to determine whether the doxycycline treatment would affect MMP transcription. We selected these MMP mRNAs for analysis because their expression or activity has been found to be inhibited by doxycycline treatment (Greenwald et al., 1998; Hanemaaijer et al., 1998; Shlopov et al., 2001; Lee et al., 2004). We observed no significant changes in the expression levels of these genes after the doxycycline treatment (Student's t test, p > 0.05; Fig. 8B).

MMP inhibition alters cochlear responses to acoustic trauma

We proceeded to investigate whether the inhibition of MMP activity affects the cochlear responses to acoustic overstimulation (120 dB SPL for 2 h) by examining the effect of a 7 d application of doxycycline treatment (50 mg · kg⁻¹ · d⁻¹). The drug was administered 1 d before noise exposure, and the same daily dosage was maintained for 6 d after the noise exposure. This time period covers the acute phases and the beginning of the chronic phases of cochlear pathogenesis.

We first examined the effect of the drug effect during the treatment period. The ABRs were measured before, during, and immediately after the completion of the drug treatment. To simplify the analysis, we averaged the thresholds tested at different frequencies for each animal and conducted a two-way ANOVA (time × treatment) to examine the time-dependent changes in the ABR thresholds. We observed a time-dependent difference in the ABR thresholds between the doxycycline-treated and saline-treated groups (F = 3.84, df = 2,154, p < 0.05; Fig. 9A). There was no significant threshold difference between the two groups before doxycycline treatment (Tukey's test, p > 0.05). There was also no significant difference after three doses of the doxycycline treatment (just before the noise exposure; Tukey's test, p > 0.05), suggesting that the doxycycline-treated and the control groups had similar hearing sensitivities at the time of noise exposure. At 2 h after noise exposure, the doxycycline-treated animals exhibited a slightly greater hearing loss than the saline-treated animals, but the difference was not statistically significant (4.3 dB; Tukey's test, p > 0.05). By the end of the 7 d doxycycline treatment, the animals exhibited a greater threshold shift than the saline-treated animals (8.8 dB; Tukey's test, p < 0.01). The difference appeared relatively homogenous across the five tested frequencies (Fig. 9B). This observation suggests that the doxycycline treatment potentiates noise-induced hearing loss during the period of the drug treatment.

To determine whether the 7 d doxycycline treatment affected noise-induced permanent hearing loss, we repeated the hearing test at 4 weeks after the noise exposure (3 weeks after the completion of the doxycycline treatment). A two-way ANOVA (treatment × frequency) revealed that the greater hearing loss in the doxycycline-treated animals observed at the end of the drug treatment was no longer present. Instead, the doxycycline-treated animals exhibited a slightly better hearing sensitivity than the saline-treated animals, although the difference was not statistically significant (Tukey's test, p > 0.05; Fig. 9C).

Although doxycycline has been widely used as a broad-spectrum MMP inhibitor, this drug has other cellular effects, including anti-inflammatory effects (Krakauer and Buckley, 2003). To confirm that the modulation of MMPs could alter the cochlear responses to acoustic trauma, we used an Mmp7 knock-out mouse model (B6.129-Mmp7^tm1Lmm/J) to investigate the effect of targeted deletion of Mmp7 on the changes in cochlear function after acoustic overstimulation. Mmp7 was selected for additional analysis because it had the greatest change in expression after acoustic trauma compared with the changes in expression of the other examined MMP genes.

Under normal physiological conditions, Mmp7^−/− mice and age-matched wild-type controls (C57BL/6J) exhibited similar baseline ABR thresholds (two-way ANOVA, F = 1.9, df = 1,100, p = 0.19; Fig. 10A), suggesting that knocking out Mmp7 does not affect the hearing sensitivity of the mice. However, after noise exposure, the Mmp7^−/− mice exhibited greater hearing loss (Fig. 10B). A two-way ANOVA (treatment × frequency) revealed an average of 11.6 ± 8.6 dB difference in the ABR thresholds between the knock-out and the wild-type mice (F = 14.0, df = 1,100, p < 0.01). The difference is not dependent on frequency (F = 0.071, df = 4,100, p > 0.05). This observation suggests that the interruption of Mmp7 potentiates noise-induced hearing loss. To confirm that the altered physiological response to acoustic overstimulation was attributable to a lack of Mmp7 expression, we measured the transcriptional expression levels of Mmp7 in the Mmp7^−/− and wild-type mice in quiet control conditions and after noise exposure. Without exposure to noise, Mmp7 transcripts were undetectable in the cochlear sensory epithelium of both the wild-type and the Mmp7^−/− mice. After the noise exposure, the expression became detectable in the wild-type mice, whereas it remained undetectable in the Mmp7^−/− mice, suggesting that the knock-out mice lacked the Mmp7 response to acoustic overstimulation. Together, these observations further support the finding that the MMPs participate in the cochlear responses to acoustic trauma.

Short-term application of doxycycline reduces cochlear damage after acoustic trauma

Considering that the treatment with doxycycline for 7 d compromised hearing sensitivity in normal animals (Fig. 7A), we wondered whether shortening the doxycycline treatment would reduce the impact of the doxycycline treatment on acute hearing loss and enhance the hearing recovery during the late phase of cochlear pathogenesis. To address this question, we administered three doses of doxycycline (50 mg · kg⁻¹ · d⁻¹) before the noise exposure (twice on 1 d before the noise exposure and once on the day of the noise exposure). Again, a two-way ANOVA revealed a time-dependent difference in the ABR thresholds between the doxycycline-treated and the saline-treated animals (F = 4.7, df = 1,83, p < 0.03; Fig. 11). We observed no significant differences in the ABR thresholds between the doxycycline- and saline-treated animals after the completion of the doxycycline treatment (just before the noise exposure; Tukey's test, p > 0.05). At 2 h after the noise exposure, the doxycycline-treated animals exhibited slightly greater hearing loss than the saline-treated noise-control animals (5.5 dB; Tukey's test, p < 0.001). However, at 14 and 28 d after the noise exposure, the doxycycline-treated animals exhibited a much greater hearing recovery than the saline-treated noise control animals (9.8 and 12.9 dB for 14 and 28 d after noise exposure, respectively; Tukey's test, p < 0.001). This result suggests that a short-term doxycycline treatment before noise exposure reduces permanent hearing loss after the noise exposure. Together, the current results suggest that the inhibition of MMP activity by doxycycline treatment modulates the cochlear responses to acoustic overstimulation.

The reduction in permanent hearing loss with doxycycline treatment is a promising result. We wanted to determine whether the treatment could also reduce the level of sensory cell damage. To this end, we compared the differences in the numbers of missing cells between the doxycycline–noise-treated (50 mg · kg⁻¹ · d⁻¹ for 7 d) and the saline–noise-treated (Fig. 12) cochleae. The doxycycline-treated animals exhibited, on average, a 41.0% reduction in the total number of missing cells compared with the saline–noise control animals, and the difference was statistically significant (Student's t test, p < 0.01). To determine whether the differences between the treated and control ears occurred in both the apical and basal sections of the cochleae, we performed a two-way ANOVA (treatment × location) and found no significant interaction between the two factors (F = 0.6, df = 1,92, p = 0.441), suggesting that the decrease in sensory cell damage after doxycycline treatment occurs in both the apical and basal sections of the cochlear sensory epithelium.

Together, these results suggest that doxycycline treatment potentiates hearing loss during the acute phase of noise-induced cochlear damage but reduces permanent hearing loss by conferring protection during the chronic phase of cochlear damage.

Constitutive expression of MMPs and their related genes in the cochlear sensory epithelium

The expression of MMPs and their related genes has been documented in a variety of tissue structures in the PNS and CNS (Wells et al., 2003; Weaver et al., 2005). In cochlear tissues, the expression of several MMP family members, including Mmp2, Mmp9, and Mmp14, have been identified (Reuter et al., 1998; Cho et al., 2002; Setz et al., 2011). Here, we used RNA-seq and qRT-PCR array techniques to profile the expression patterns and document the constitutive expression of a set of MMP genes, many of which have not been reported previously in the cochlear sensory epithelium. The detection of the strong, constitutive expression of multiple MMP genes in the cochlea provides the molecular basis for the prominent cochlear MMP activity observed in the current investigation. Interestingly, the TIMP genes are also highly expressed in the cochlea, suggesting that MMP activity is regulated by their intrinsic inhibitors under physiological conditions.

Our correlation analysis reveals a strong association between the expression levels of the MMPs and their related genes in the apical section and the basal section of the cochlear sensory epithelium. This result was anticipated, considering the fact that the two sections of the cochlear sensory epithelium contain the same populations of cells. However, comparison of the expression levels reveals that many of the examined genes were more highly expressed in the apical section than in the basal section. This type of difference was not observed for the other gene families in previous observations (Sato et al., 2009; Cai et al., 2012). Interestingly, the differences in the constitutive expression coincided with greater expression changes in the basal section of the cochlear sensory epithelium after acoustic trauma. It has been shown that the apical and basal sections of the cochlea differ in their frequency responses to acoustic stimuli (Robles and Ruggero, 2001). The two sections also exhibit different susceptibilities to various common pathological insults, including acoustic trauma, ototoxicity, and aging-related cochlear degeneration (Thorne et al., 1984; Bohne et al., 1987; Forge and Schacht, 2000; Sha et al., 2001; Hu et al., 2002; Pouyatos et al., 2009). Given the finding that the gene knock-out can lead to frequency-dependent hearing loss (Cheatham et al., 2004), we suspect that the site-specific differences in the expression levels of MMPs and their related genes may serve as a contributing factor to the tonotopic differences in the function and biology of the two cochlear sections.

MMPs and their related genes contribute to noise-induced cochlear degeneration

Acoustic overstimulation causes cochlear dysfunction and sensory cell damage. Here, we provide strong evidence for the contribution of MMPs and their related gene products to acoustic injury. First, acoustic overstimulation resulted in time- and site-dependent transcriptional changes in MMPs and their related genes. Temporally, we observed that the changes in expression during the early phases of cochlear pathogenesis were dominated by upregulation and that the changes in expression during the later phase of cochlear pathogenesis were dominated by downregulation. This finding is consistent with the early increase in MMP expression that was observed in other traumatic injuries in the nervous system, including injury to the spinal cord (Wells et al., 2003; Veeravalli et al., 2009). Regarding the spatial pattern of the expression changes, our study shows that a greater change in expression occurs in the basal section of the cochlear sensory epithelium. This expression pattern is accompanied by greater sensory cell damage in the basal section of the cochlea (Cai et al., 2012), highlighting the spatial correlation between these two events in the cochlear response to acoustic trauma.

The second piece of evidence linking the MMPs and their related genes to acoustic injury comes from the analysis of the protein expression pattern of TIMPs, the endogenous inhibitors of MMPs. We found an increase in the immunoreactivity of Timp1 and Timp2. Interestingly, these changes were spatially related to the cells that exhibited condensed or fragmented nuclei. Given that all of the cells that had condensed or fragmented nuclei exhibited strong immunoreactivity for Bax, a biological marker of apoptosis (Lalier et al., 2007), we believe that the alteration in TIMP protein expression is related to apoptotic cell damage and to permanent hearing loss. Here, we found that the numbers of Timp1/Timp2-positive cells are limited at the time of observation. Because the apoptotic process is a rapid cell death process, the presence of a limited number of positive cells at a given time point after noise exposure is expected. With the continuous growth of cochlear lesions after noise exposure, more cells will enter the apoptotic cell death pathway. Therefore, the total number of TIMP-positive cells is expected to be much higher than that observed. It should be noted that the change in Timp protein expression does not necessarily reflect a change in their enzymatic activity. Additional research on the changes in the activity of Timps is warranted.

The third piece of evidence for MMP participation in acoustic trauma involves changes in the enzymatic activity of MMPs. We found that the reduction in MMP activity was time-dependent and occurred at 1 d after injury. This type of dynamic change in MMP activity has been observed previously in non-cochlear tissues after traumatic injury and has been linked to the progression/recovery of tissue pathology (Wells et al., 2003; Kolar et al., 2008; Veeravalli et al., 2009). MMP activity is controlled at multiple levels, including transcriptional regulation, posttranslational modification, intrinsic inhibition, proenzyme activation, and MMP compartmentalization. In the current investigation, we examined two major regulatory mechanisms: the transcriptional expression of the MMP genes and the protein expression of the Timps. We found time-dependent changes in the expression levels of these proteins. To date, the molecular mechanisms responsible for the changes in expression of the MMPs are not clear. Studies using non-cochlear tissues have linked the regulation of MMP expression with nuclear binding factors, such as nuclear factor κB and activator protein-1 (Overall et al., 2002). Both proteins have been linked to acoustic trauma (Ogita et al., 2000; Masuda et al., 2006). The presence of the complex regulatory mechanisms suggests that noise-induced cochlear damage is a multifactorial disorder.

Finally, we found that the inhibition of MMP activity compromised normal cochlear function, suggesting the importance of MMP activity in the maintenance of cochlear function. Moreover, the modulation of MMP expression and activity alters cochlear responses to acoustic overstimulation. The detailed molecular mechanisms responsible for this impact of MMPs are not clear. Recent studies have also shown that the MMPs have a variety of intracellular substrates (Sternlicht et al., 2001; Rodriguez et al., 2010). All of the MMPs that were found to be upregulated after noise exposure in the current study have implications for the range of intracellular substrates accessible to these MMPs (Golubkov et al., 2005; Choi et al., 2008; Cuadrado et al., 2009). Given the detection of expression changes in multiple MMPs and their related genes, we suspect that the observed effects are orchestrated by the integrated action of MMP molecules. Together, the observation of changes in the levels of enzymatic activity, protein expression, and transcriptional expression indicates that the MMPs and their related genes are involved in the regulation of the cochlear response to acoustic overstimulation. This observation is the first to link the MMPs and their related genes to acute acoustic trauma.

Therapeutic implications of MMP inhibition

The regulation of acoustic trauma through MMPs raises the possibility of using MMP intervention as a therapeutic strategy to reduce noise-induced cochlear damage. Here, we reveal two features associated with doxycycline treatment. First, a short-term treatment before acoustic trauma is protective, whereas prolonged treatment potentiates noise-induced hearing loss. This finding is supported by the data from the experiment using Mmp7^−/− mice. Second, the protection of hearing induced by the doxycycline treatment manifested after the completion of the drug treatment rather than during the treatment. These observations suggest that the timing of MMP intervention is an important factor that contributes to the ultimate outcome of the intervention. It should be noted that the biological effects of doxycycline are complex. Although this drug has been widely used for modulating MMP activity, it can exert other biological impacts, including anti-inflammation (Krakauer et al., 2003). To this end, future investigations using more specific MMP inhibitors are expected to provide more insights into the effects of MMP intervention on noise-induced cochlear damage.