**Changes in microRNA Expression in the Cochlear Nucleus and Inferior Colliculus after Acute Noise-Induced Hearing Loss**

**Sohyeon Park 1,2, Seung Hee Han <sup>2</sup> , Byeong-Gon Kim 2,3, Myung-Whan Suh <sup>2</sup> , Jun Ho Lee 2,3 , Seung Ha Oh 2,3 and Moo Kyun Park 2,3,4,\***


Received: 27 October 2020; Accepted: 19 November 2020; Published: 20 November 2020

**Abstract:** Noise-induced hearing loss (NIHL) can lead to secondary changes that induce neural plasticity in the central auditory pathway. These changes include decreases in the number of synapses, the degeneration of auditory nerve fibers, and reorganization of the cochlear nucleus (CN) and inferior colliculus (IC) in the brain. This study investigated the role of microRNAs (miRNAs) in the neural plasticity of the central auditory pathway after acute NIHL. Male Sprague–Dawley rats were exposed to white band noise at 115 dB for 2 h, and the auditory brainstem response (ABR) and morphology of the organ of Corti were evaluated on days 1 and 3. Following noise exposure, the ABR threshold shift was significantly smaller in the day 3 group, while wave II amplitudes were significantly larger in the day 3 group compared to the day 1 group. The organ of Corti on the basal turn showed evidence of damage and the number of surviving outer hair cells was significantly lower in the basal and middle turn areas of the hearing loss groups relative to controls. Five and three candidate miRNAs for each CN and IC were selected based on microarray analysis and quantitative reverse transcription PCR (RT-qPCR). The data confirmed that even short-term acoustic stimulation can lead to changes in neuroplasticity. Further studies are needed to validate the role of these candidate miRNAs. Such miRNAs may be used in the early diagnosis and treatment of neural plasticity of the central auditory pathway after acute NIHL.

**Keywords:** noise-induced hearing loss; microRNAs; cochlear nucleus; inferior colliculus; neuroplasticity

### **1. Introduction**

According to the World Health Organization, 466 million people—more than 5% of the world's population—are affected by hearing loss [1]. Loss of hearing makes communication difficult and can lead to psychosocial problems, such as depression and feelings of loneliness [2]. Therefore, losing the ability to hear can affect people's lives in many ways.

NIHL is a type of acquired hearing loss that is due to a sudden excessively loud sound or continuous moderately loud sounds. Depending on the duration and intensity of the sounds, hearing loss can be temporary or permanent [3]. Threshold sensitivity loss after exposure to a loud noise can be recovered to baseline levels after a few hours to weeks and that is called temporary threshold shift (TTS). Permanent threshold shift (PTS) occurs when the noise is too loud to recover to its baseline level or the noise takes place repeatedly or continuously, impeding recovery.

Even though hearing thresholds may return to normal, the changes in the synapse may affect auditory processing and hearing upon noise [4]. Exposure to noise not only targets the hair cells (HCs) in the cochlea, but also the neuronal cells in the auditory pathway. Noise exposure can also alter the synaptic transmission and ion channel function, distort sensory maps, and cause abnormal neuronal firing patterns. These cellular and physiological changes can cause another disorder, such as loudness recruitment or tinnitus [5].

NIHL usually progresses slowly over a period of years, making it difficult to recognize. Initial symptoms include difficulty in differentiating conversations against background noise. Later, listening becomes difficult, even in ordinary circumstances. Since NIHL progresses gradually and prevention is so crucial, it is important to diagnose NIHL at an early stage [6].

MicroRNAs (MiRNAs) are small non-coding single-stranded endogenous RNAs that regulate posttranscriptional gene expression. Mature miRNAs are 18–22 nucleotides in length and can repress and degrade target mRNAs by binding to their 30 or 50 untranslated regions. MiRNAs are involved in a number of physiological processes and play key roles in neural development and plasticity. MiRNAs are abundant in the brain, due to its complexity [7,8], and are also found in body fluids such as the cerebrospinal fluid, whole blood, plasma, and serum [9]. Consequently, miRNAs can be detected and analyzed non-surgically and may provide the basis for a new generation of therapeutic agents [10]. In fact, there are some miRNA-based therapeutics for cancer and other diseases and several miRNAs are currently being studied in Phase I/II clinical trials, with promising outcomes [11].

Electrical signals from the HCs in the inner ear flow to primary sensory neurons, which are called spiral ganglion neurons. These cochlear nerves combine with vestibular nerves to become branch VIII of the cranial nerve [12], which transfers the converted acoustic information to the cochlear nucleus (CN), developing the neural pathway [13]. The superior olivary complex (SOC) in the ventral auditory brainstem receives auditory information from the CN and sends the signal to the inferior colliculus (IC). The IC plays a crucial role in the auditory pathway and is the first location where parallel auditory signals from both sides of the cochlea are integrated. Most of the auditory information is processed in the IC and sent to the auditory cortex through the medial geniculate body [14].

Noise exposure is one of the most common causes of hearing loss. Exposure to loud noise leads to secondary changes, such as decreases in the number of synapses, the degeneration of auditory nerve fibers, and reorganization of the CN and IC, which may induce neural plasticity in the central auditory pathway. However, little is known about the role of miRNAs in the central auditory pathway. Therefore, this study investigated the role of miRNAs in the neural plasticity of the central auditory pathway after NIHL.

### **2. Results**

### *2.1. Hearing Changes after Noise Exposure*

Tone-burst acoustic stimuli were measured at three different frequencies of 4, 8, and 16 kHz. Our noise-control model (*n* = 4) showed that the auditory brainstem response (ABR) threshold started to decrease at day 1 after the noise exposure. By day 3 after the exposure to noise at 4 and 8 kHz, the ABR thresholds had returned to normal (Figure 1a). The mean ABR thresholds observed for the day 1 control group (4 kHz, 20.6 ± 2.2 dB; 8 kHz, 21.3 ± 2.7 dB; 16 kHz, 25.4 ± 3.6 dB), day 1 group (4 kHz, 20.0 ± 0.0 dB; 8 kHz, 20.8 ± 1.9 dB; 16 kHz, 27.1 ± 3.6 dB), day 3 control group (4 kHz, 20.8 ± 2.4 dB; 8 kHz, 21.9 ± 3.2 dB; 16 kHz, 25.6 ± 4.3 dB), and day 3 group (4 kHz, 24.2 ± 4.8 dB; 8 kHz, 25.8 ± 4.6 dB; 16 kHz, 26.9 ± 3.6 dB) confirmed that hearing levels at all three frequencies were normal before the experiment (Figure 1b–d). After 2 h of noise exposure, both the day 1 group (4 kHz, 81.9 ± 11.6 dB; 8 kHz, 87.1 ± 3.3 dB; 16 kHz, 88.3 ± 2.4 dB) and the day 3 group (4 kHz, 78.8 ± 11.8 dB; 8 kHz, 84.6 ± 2.9 dB; 16 kHz, 86.7 ± 3.8 dB) exhibited significant increases in ABR thresholds compared

to the day 1 control group (4 kHz, 20.2 ± 1.0 dB; 8 kHz, 20.4 ± 1.4 dB; 16 kHz, 23.5 ± 3.8 dB) and the day 3 control group (4 kHz, 20.4 ± 1.4 dB; 8 kHz, 21.0 ± 2.5 dB; 16 kHz, 24.4 ± 3.7 dB), respectively (Figure 1b–d). At 4 kHz, the mean ABR threshold of the day 1 group (65.6 ± 19.5 dB) was slightly lower than that of the day 1 control group (81.9 ± 11.6 dB). Moreover, the mean ABR threshold of the day 3 group (42.7 ± 17.1 dB) was lower than that of the day 1 group. The differences between these groups were significant (*p* < 0.001; Figure 1b,e). At 8 kHz, the mean ABR thresholds of both the day 1 (73.3 ± 10.7 dB) and day 3 (51.9 ± 13.7 dB) groups were significantly lower than those of the control groups (87.1 ± 3.3 dB; *p* < 0.001; Figure 1c,f). The mean ABR thresholds of the day 1 (79.4 ± 8.5 dB) and day 3 (63.5 ± 12.6 dB) groups were also significantly lower than those of the control groups (88.3 ± 2.4 dB) at 16 kHz (*p* < 0.001; Figure 1d,g). The day 3 group exhibited significantly better hearing than the day 1 group at all three frequencies (*p* < 0.001; Figure 1e–g).

**Figure 1.** Hearing changes after noise exposure. (**a**) Long-term auditory brainstem response (ABR) data of the noise-control model obtained over a period of 3 weeks. (**b**–**d**) Line graphs showing differences between control and treatment groups at frequencies of 4, 8, and 16 kHz. \*\*\* *p* < 0.001. (**e**–**g**) Line graphs comparing ABR thresholds at days 1 and 3 after noise exposure at frequencies of 4, 8, and 16 kHz. \*\*\* *p* < 0.001.

### *2.2. ABR Amplitudes*

The amplitudes of waves II and IV were calculated based on ABR waveforms. The wave II amplitudes observed in the day 1 treatment group (4 kHz, 0.61 ± 0.28 µV; 8 kHz, 0.62 ± 0.25 µV; 16 kHz, 0.39 ± 0.23 µV) were significantly smaller than those observed in the day 3 treatment group, at all frequencies (4 kHz, 1.56 ± 0.90 µV; 8 kHz, 1.45 ± 0.84 µV; 16 kHz, 1.08 ± 0.0 µV; *p* < 0.001; Figure 2a). At 4 kHz, the mean wave IV amplitude observed in the day 1 treatment group (0.44 ± 0.39 µV) was significantly smaller than that in the day 3 treatment group (0.93 ± 0.22 µV). However, at the other frequencies, the amplitudes observed in the day 1 treatment group (8 kHz, 0.36 ± 0.22 µV; 16 kHz, 0.26 ± 0.18 µV) did not differ significantly from those observed on day 3 (8 kHz, 0.43 ± 0.40 µV; 16 kHz, 0.26 ± 0.26; Figure 2b).

1

1 **Figure 2.** Amplitudes of waves II and IV and latencies of waves IV–II. (**a**) Comparison of the amplitude of wave II at different frequencies. The wave II amplitude for rats assayed at day 3 after noise exposure was significantly larger than that for rats assayed at day 1 after noise exposure, at all frequencies. \*\*\* *p* < 0.001. (**b**) Comparison of the amplitude of wave IV among different frequencies. The wave IV amplitude for rats assayed at day 3 after noise exposure was significantly larger than that for rats assayed at day 1 after noise exposure at 4 kHz. \*\*\* *p* < 0.001. No significant differences were observed at the other frequencies. (**c**) The latencies of waves IV–II at each frequency. No significant differences were observed at any frequency.

### *2.3. ABR Latencies*

Latencies between waves IV and II were calculated based on ABR waveforms. At 4 kHz, there was no difference in latencies between the day 1 (2.54 ± 0.50 ms) and day 3 treatment groups (2.64 ± 0.77 ms). Similar results were seen at 16 kHz, with no difference in latencies being evident between the day 1 (2.77 ± 0.15 ms) and day 3 treatment groups (2.80 ± 0.41 ms). Conversely, at 8 kHz, the latency observed in the day 3 treatment group (2.38 ± 0.54 ms) was slightly reduced compared to that of the day 1 treatment group (2.68 ± 0.71 ms); however, this difference was not statistically significant (Figure 2c).

### *2.4. Histology of the Organ of Corti*

A series of sagittal sections from decalcified cochleae were stained with hematoxylin and eosin (H&E) to investigate the structure of the organ of Corti. These structures were intact in both the day 1 and day 3 control groups. The apical turn and middle turn sections of the organ of Corti from the day 1 and day 3 treatment groups were also normal, with intact HCs and other non-sensory cells, such as Deiter and pillar cells. However, the basal turn sections of the organs of Corti exhibited abnormalities. The degree of damage differed among cochleae, with some exhibiting only a loss of HCs, and others also exhibiting a loss of supporting cells (Figure 3).

**Figure 3.** Hematoxylin and eosin (H&E) staining of the basal, middle, and apical turn sections of the organ of Corti. Damage (indicated by black arrows) was only observed in the basal turn section in **Figure 3.** Hematoxylin and eosin (H&E) staining of the basal, middle, and apical turn sections of the organ of Corti. Damage (indicated by black arrows) was only observed in the basal turn section in rats assayed at days 1 and 3 after noise exposure. All scale bars represent 50 µm.

rats assayed at days 1 and 3 after noise exposure. All scale bars represent 50 µm.

### *2.5. Phalloidin Staining of Outer HCs*

*2.5. Phalloidin Staining of Outer HCs*  Following completion of the whole mount surface preparation procedure, phalloidin was used to stain the outer HCs. Two 200-µm-long segments were selected from each turn section of the organ of Corti, and the mean number of surviving outer HCs was determined. All three rows of outer HCs from all control samples were normal, with no evidence of missing HCs. However, significant HC loss was observed in the basal turn sections in both the day 1 (85% ± 7%; *p* = 0.001) and day 3 (93% ± 6%; *p* = 0.019) treatment groups. Some HCs were also lost from the middle turn sections of samples from the day 1 (99% ± 1%; *p* = 0.008) and day 3 (99% ± 1%; *p* < 0.001) treatment groups. In contrast, outer HCs in the apical turn sections of samples from both the day 1 and day 3 treatment groups were largely intact (both, 100% ± 0%). For the basal turn sections, significantly fewer HCs were lost from the day 3 treatment group compared to the day 1 treatment group (*p* = 0.039). No statistically significant difference in HC loss was observed between the day 1 and day 3 groups for the middle or apical turn sections of the organ of Corti (Figure 4). Following completion of the whole mount surface preparation procedure, phalloidin was used to stain the outer HCs. Two 200-µm-long segments were selected from each turn section of the organ of Corti, and the mean number of surviving outer HCs was determined. All three rows of outer HCs from all control samples were normal, with no evidence of missing HCs. However, significant HC loss was observed in the basal turn sections in both the day 1 (85% ± 7%; *p* = 0.001) and day 3 (93% ± 6%; *p* = 0.019) treatment groups. Some HCs were also lost from the middle turn sections of samples from the day 1 (99% ± 1%; *p* = 0.008) and day 3 (99% ± 1%; *p* < 0.001) treatment groups. In contrast, outer HCs in the apical turn sections of samples from both the day 1 and day 3 treatment groups were largely intact (both, 100% ± 0%). For the basal turn sections, significantly fewer HCs were lost from the day 3 treatment group compared to the day 1 treatment group (*p* = 0.039). No statistically significant difference in HC loss was observed between the day 1 and day 3 groups for the middle or apical turn sections of the organ of Corti (Figure 4).

*Int. J. Mol. Sci.* **2020**, *21*, x FOR PEER REVIEW 6 of 20

**Figure 4.** Phalloidin staining of outer hair cells (HCs) and HC survival. (**a**) Fluorescence staining of outer HCs from each turn section of the cochlea. Scale bars represent 50 µm. Asterisks indicate the positions of lost HCs. The blue line along the hair cell line indicates the length of 200 µm. (**b**) Survival rates of outer HCs in each turn section. The surviving HCs per 200 µm along the length of the cochlea in the basal, middle, and apical turn sections were counted. \*\* *p* < 0.05 and \*\*\* *p* < 0.001. **Figure 4.** Phalloidin staining of outer hair cells (HCs) and HC survival. (**a**) Fluorescence staining of outer HCs from each turn section of the cochlea. Scale bars represent 50 µm. Asterisks indicate the positions of lost HCs. The blue line along the hair cell line indicates the length of 200 µm. (**b**) Survival rates of outer HCs in each turn section. The surviving HCs per 200 µm along the length of the cochlea in the basal, middle, and apical turn sections were counted. \*\* *p* < 0.05 and \*\*\* *p* < 0.001.

#### *2.6. Selection of Candidate miRNAs 2.6. Selection of Candidate miRNAs*

*miR-200b-3p* (Table 1).

#### 2.6.1. The CN 2.6.1. The CN

Microarray analysis of the CN identified 1228 candidate miRNAs. Changes in miRNA expression were assessed via three pair-wise comparisons: Between the day 1 treatment and control groups; between the day 3 treatment and control groups; and between the day 1 and day 3 groups. For each comparison, miRNAs with normalized expression changes ≥ 1.5-fold (*p* < 0.1) were excluded. Then, the miRNAs from each set were combined and those found in more than one dataset were eliminated. miRNAs that exhibited differences in expression between the day 1 and day 3 groups were also excluded. A hierarchical clustering heat map was created using Multiple Experiment Viewer (http://mev.tm4.org) to visualize the remaining 33 miRNAs (Figure 5a). Then, miRNAs with ≥1.5-fold differences in expression between the day 1 and day 3 control groups were excluded. Of the 21 miRNAs that remained, only those expressed in humans were considered for further analysis, resulting in a final list of 10 candidate miRNAs, including *miR-411-3p, miR-183-5p, miR-377-3p, miR-20b-5p, miR-137-5p, miR-211-3p, miR-483-5p, miR-92a-1-5p, miR-187-5p,* and Microarray analysis of the CN identified 1228 candidate miRNAs. Changes in miRNA expression were assessed via three pair-wise comparisons: Between the day 1 treatment and control groups; between the day 3 treatment and control groups; and between the day 1 and day 3 groups. For each comparison, miRNAs with normalized expression changes ≥1.5-fold (*p* < 0.1) were excluded. Then, the miRNAs from each set were combined and those found in more than one dataset were eliminated. miRNAs that exhibited differences in expression between the day 1 and day 3 groups were also excluded. A hierarchical clustering heat map was created using Multiple Experiment Viewer (http://mev.tm4.org) to visualize the remaining 33 miRNAs (Figure 5a). Then, miRNAs with ≥1.5-fold differences in expression between the day 1 and day 3 control groups were excluded. Of the 21 miRNAs that remained, only those expressed in humans were considered for further analysis, resulting in a final list of 10 candidate miRNAs, including *miR-411-3p*, *miR-183-5p*, *miR-377-3p*, *miR-20b-5p*, *miR-137-5p*, *miR-211-3p*, *miR-483-5p*, *miR-92a-1-5p*, *miR-187-5p*, and *miR-200b-3p* (Table 1).

*Int. J. Mol. Sci.* **2020**, *21*, x FOR PEER REVIEW 7 of 20

**Figure 5.** Heat maps of the CN and IC. Heat maps of (**a**) 33 miRNAs from the CN and (**b**) 27 miRNAs from the IC selected based on ≥1.5-fold changes in normalized expression (*p* < 0.1). **Figure 5.** Heat maps of the CN and IC. Heat maps of (**a**) 33 miRNAs from the CN and (**b**) 27 miRNAs from the IC selected based on ≥1.5-fold changes in normalized expression (*p* < 0.1).



*rno-miR-211-3p* 1 20 GGCAAGGACAGCAAAGGGGG 0.649 1.420 1.334 0.610 *rno-miR-483-5p* 1 22 AAGACGGGAGAAGAGAAGGGAG 1.072 2.025 1.393 0.737 *rno-miR-92a-1-5p* 15 23 AGGUUGGGAUUUGUCGCAAUGCU 0.588 1.194 1.468 0.724 <sup>1</sup> Fold change of day 1 treatment vs. control, <sup>2</sup> fold change of the day 3 treatment group vs. control, <sup>3</sup> fold change of the day 1 and day 3 treatment groups, and <sup>4</sup> fold change of the day 1 vs. day 3 control groups. A color index chart for of the fold change data is provided in Supplementary Figure S1.

#### *rno-miR-187-5p* 18 18 AGGCUACAACACAGGACC 0.529 1.404 1.827 0.688 *rno-miR-200b-3p* 5 23 UAAUACUGCCUGGUAAUGAUGAC 1.826 3.587 2.895 1.474 2.6.2. The IC

1 Fold change of day 1 treatment vs. control, 2 fold change of the day 3 treatment group vs. control, 3 fold change of the day 1 and day 3 treatment groups, and 4 fold change of the day 1 vs. day 3 control groups. A color index chart for of the fold change data is provided in Supplementary Figure S1. 2.6.2. The IC Microarray analysis of the IC identified 1200 miRNAs. Changes in miRNA expression were assessed in three pair-wise comparisons: Between the day 1 treatment and control groups; between the day 3 treatment and control groups; and between the day 1 and day 3 groups. For each comparison, miRNAs with ≥1.5-fold normalized expression changes (*p* < 0.1) were excluded. Then, the miRNAs from each set were combined, and those found in more than one dataset were eliminated. miRNAs that exhibited differences in expression between the day 1 and day 3 groups Microarray analysis of the IC identified 1200 miRNAs. Changes in miRNA expression were assessed in three pair-wise comparisons: Between the day 1 treatment and control groups; between the day 3 treatment and control groups; and between the day 1 and day 3 groups. For each comparison, miRNAs with ≥1.5-fold normalized expression changes (*p* < 0.1) were excluded. Then, the miRNAs from each set were combined, and those found in more than one dataset were eliminated. miRNAs that exhibited differences in expression between the day 1 and day 3 groups were also excluded. A hierarchical clustering heat map was created using Multiple Experiment Viewer to assess the remaining 27 miRNAs (Figure 5b). Then, miRNAs with expression differences changes >1.5-fold or <0.5-fold between the day 3 and day 1 control groups were excluded. Of the 26 miRNAs that remained, only those expressed in humans were considered for further analysis, resulting in a final list of 13 candidate miRNAs, including *miR-204-5p*, *miR-376b-5p*, *miR-26b-5p*, *miR-136-3p*,

were also excluded. A hierarchical clustering heat map was created using Multiple Experiment

*miR-132-5p*, *miR-128-2-5p*, *miR-132-3p*, *miR-377-5p*, *miR-210-3p*, *miR-92a-1-5p*, *miR-425-3p*, *miR-362-5p*, and *miR-150-3p* (Table 2).



<sup>1</sup> Fold change of day 1 treatment group vs. control, <sup>2</sup> fold change of the day 3 treatment group vs. control, <sup>3</sup> fold change of the day 1 vs. day 3 treatment groups, and <sup>4</sup> fold change of the day 1 vs. day 3 control groups. A color index chart for the fold change data is provided in Supplementary Figure S1.

### *2.7. Validation of Candidate miRNAs Using qRT-PCR*

### 2.7.1. The CN

Based on the results of the microarray analysis of the CN, 10 candidate miRNAs were selected for further analysis, including *miR-411-3p*, *miR-183-5p*, *miR-377-3p*, *miR-20b-5p*, *miR-137-5p*, *miR-211-3p*, *miR-483-5p*, *miR-92a-1-5p*, *miR-187-5p*, and *miR-200b-3p* (Figure 6a). Microarray data of the day 3 and day 1 treatment groups are compared in Figure 6b, along with the accompanying qRT-PCR results. Based on these data, five miRNAs were selected due to their consistent expression patterns, including *miR-411-3p*, *miR-183-5p*, *miR-377-3p*, *miR-20b-5p*, and *miR-200b-3p*. The expression of *miR-200b-3p* increased after noise exposure, whereas that of the other miRNAs decreased.

**Figure 6.** A Cytoscape map of the CN and validation of candidate miRNAs. Cytoscape was used to visualize networks among candidate miRNAs. Only miRNAs that were connected to other miRNAs were selected for validation by quantitative reverse transcription polymerase chain reaction (qRT-PCR). (**a**) A total of 10 miRNAs were selected as CN candidate miRNAs. (**b**) Ratio of expression of each candidate miRNA in the CN between the day 3 and day 1 treatment groups. Expression levels were measured using microarray analysis (open bars) and qRT-PCR (filled bars). Crosses indicate validated candidate miRNAs.

### 2.7.2. The IC

Based on the results of the microarray analysis of the IC, 13 candidate miRNAs were selected for further analysis, including *miR-204-5p*, *miR-376b-5p*, *miR-26b-5p*, *miR-136-3p*, *miR-132-5p*, *miR-128-2-5p*, *miR-132-3p*, *miR-377-5p*, *miR-210-3p*, *miR-92a-1-5p*, *miR-425-3p*, *miR-362-5p*, and *miR-150-3p* (Figure 7a). Microarray data of the day 3 and day 1 treatment groups are compared in Figure 7b, along with the accompanying qRT-PCR results. Three miRNAs were selected due to their consistent expression patterns, including *miR-92a-1-5p*, *miR-136-3p*, and *miR-26b-5p*. The expression of *miR-92a-1-5p* increased after noise exposure, whereas that of the other miRNAs decreased.

2

**Figure 7.** A Cytoscape map of the IC and validation of candidate miRNAs. Cytoscape was used to visualize networks among the candidate miRNAs. Only miRNAs that were connected to other miRNAs were selected for validation by qRT-PCR. (**a**) A total of 13 miRNAs were selected as IC candidate miRNAs. (**b**) Ratio of expression of each candidate miRNA in the IC between the day 3 and day 1 treatment groups. Expression levels were measured using microarray analysis (open bars) and qRT-PCR (filled bars). Crosses indicate validated candidate miRNAs.

### *2.8. Target Pathway Analysis of Candidate miRNAs*

### 2.8.1. The CN

2 Five candidate miRNAs expressed in the CN were validated using qRT-PCR, including *miR-411-3p*, *miR-183-5p*, *miR-377-3p*, *miR-20b-5p*, and *miR-200b-3p*. DIANA-miRPath software (ver. 3.0; http://www. microrna.gr/miRPathv3) was used to investigate the regulation of biological pathways by miRNAs in the CN. A Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis identified 12 significantly overrepresented pathways. The most relevant pathways for these miRNAs involved mitogen-activated protein kinase (MAPK) signaling, axon guidance, and transforming growth factor-beta (TGF-β) signaling (Figure 8a).

transforming growth factor-beta (TGF-β) signaling (Figure 8a).

2.8.1. The CN

2.8.2. The IC

Five candidate miRNAs expressed in the CN were validated using qRT-PCR, including *miR-411-3p, miR-183-5p, miR-377-3p, miR-20b-5p,* and *miR-200b-3p*. DIANA-miRPath software (ver. 3.0; http://www. microrna. gr/miRPathv3) was used to investigate the regulation of biological pathways by miRNAs in the CN. A Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis identified 12 significantly overrepresented pathways. The most relevant pathways for these miRNAs involved mitogen-activated protein kinase (MAPK) signaling, axon guidance, and

Three candidate miRNAs expressed in the IC were validated using qRT-PCR, including *miR-92a-1-5p, miR-136-3p*, and *miR-26b-5p*. DIANA-miRPath software (ver. 3.0; DIANA TOOLS, http://www. microrna. gr/miRPathv3) was used to investigate the regulation of biological pathways

most relevant pathway for these three miRNAs was the MAPK signaling pathway (Figure 8b).

**Figure 8.** Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis of candidate miRNAs. (**a**) KEGG pathway analysis for the CN. Of the 12 possible in silico pathways identified, eight (marked in red) were highlighted based on their relevance to the five candidate miRNAs. (**b**) KEGG pathway analysis for the IC. Of the 14 possible in silico pathways identified, only the mitogen-activated protein kinase (MAPK) signaling pathway (marked in red) was identified as a relevant target for the three candidate miRNAs. Abbreviations: mTOR, mammalian target of rapamycin; TGF, transforming growth factor; PPAR, peroxisome proliferator-activated receptor. **Figure 8.** Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis of candidate miRNAs. (**a**) KEGG pathway analysis for the CN. Of the 12 possible in silico pathways identified, eight (marked in red) were highlighted based on their relevance to the five candidate miRNAs. (**b**) KEGG pathway analysis for the IC. Of the 14 possible in silico pathways identified, only the mitogen-activated protein kinase (MAPK) signaling pathway (marked in red) was identified as a relevant target for the three candidate miRNAs. Abbreviations: mTOR, mammalian target of rapamycin; TGF, transforming growth factor; PPAR, peroxisome proliferator-activated receptor.

### 2.8.2. The IC

Three candidate miRNAs expressed in the IC were validated using qRT-PCR, including *miR-92a-1-5p, miR-136-3p*, and *miR-26b-5p*. DIANA-miRPath software (ver. 3.0; DIANA TOOLS, http://www.microrna. gr/miRPathv3) was used to investigate the regulation of biological pathways by miRNAs in the IC. A KEGG analysis identified 14 significantly overrepresented pathways. The most relevant pathway for these three miRNAs was the MAPK signaling pathway (Figure 8b).

### **3. Discussion**

Our results demonstrated that even short-term acoustic stimulation can cause changes in miRNA expression in the CN and IC, and that these changes may also induce plasticity in the central auditory pathway. Microarray analysis and qRT-PCR suggested that miRNAs play a key role in neural plasticity after noise exposure in both the CN (i.e., *miR-411-3p*, *miR-183-5p*, *miR-377-3p*, *miR-20b-5p*, and *miR-200b-3p*) and IC (i.e., *miR-92a-1-5p*, *miR-136-3p*, and *miR-26b-5p*). Further research is necessary to understand the specific roles of these candidate miRNAs, with preliminary evidence suggesting that they may be involved in regulating the MAPK signaling pathway, axon guidance, and the TGF-β signaling pathway. Numerous studies have investigated NIHL without identifying a reliable tool for

early diagnosis or a treatment that results in complete recovery. Currently, hearing aids and cochlear implants are used to treat patients with severe hearing loss, but more effective methods for diagnosing and treating NIHL are required. Since miRNAs are stable over long periods of time and can be detected in the blood, as well as in brain tissue, these sequences may represent a viable diagnostic target for blood tests, enabling earlier diagnosis of NIHL and potentially protecting people against tinnitus. Moreover, gene therapy involving the transfer of miRNAs to target cells using viral vectors or siRNAs could be used to treat NIHL in the future.

Consecutive ABR tests were performed in this study, with evidence of ABR threshold recovery beginning as early as day 1 after noise exposure. The shift in the ABR threshold decreased significantly on days 2 and 3 after noise exposure, and was insignificant after day 3. Based on these observations, we hypothesized that significant changes may occur within this time frame. We therefore chose day 1 and 3 after noise exposure as our two major time points.

We focused on the CN and IC, because the latter is the origin of the central auditory pathway and the IC forms its core. The central auditory pathway receives the bilateral auditory signal, with waves II and IV corresponding to the CN and IC, respectively [15]. Therefore, we evaluated the latency between waves IV and II, and the height of these waves relative to their resting points. The latency between waves IV and II was slightly reduced at 8 kHz in the day 3 treatment group; however, the difference was not statistically significant. Changes in the latency of wave I might be expected because the auditory nerve is the primary region affected by HC loss [16]. An increased latency of wave I may indicate dysfunction in action potential propagation along the auditory nerve [17]. However, it is difficult to measure wave I using the ABR [18]. Moreover, short-term noise exposure may affect auditory nerve fibers, but not the CN or IC.

We observed a statistically significant difference in the amplitude of wave II between the day 1 and day 3 treatment groups. The amplitude of wave II was greater in the day 3 treatment group at all frequencies. In general, the ABR amplitude of wave I is reduced by overexpressed sound stimuli [19,20], while the amplitudes of later ABR waves (i.e., II–V) should increase due to compensatory hyperactivity of the central auditory pathway. A reduced sensory input from both ears triggers an increase in excitatory activity and/or a decrease in inhibitory activity [21]. Regarding interpretation of the ABR wave, latency represents the speed of transmission and amplitude represents the number of neurons that fire together [22]. One possible explanation for our observations is that the number of neurons in the CN decreased. However, either the damaged neurons recovered, or axonal sprouting from the original neurons must have occurred, because the measured amplitudes were large. Moreover, with the exception of the 4 kHz frequency measurements, there was no difference in the amplitude of wave IV between the two noise exposure groups. This suggests that the levels of noise used in this study were insufficient for affecting the IC, or that neuronal activity in the CN was able to compensate for the damage.

All three rows of outer HCs in the control groups were normal, and there were no missing HCs. Only a few HCs were missing in the middle and basal turn sections in the day 1 and day 3 treatment groups. The minimum survival rates of outer HCs in the basal turn and apical turn sections were 78% and 99%, respectively. This suggests that outer HC loss was mild in general. The confirmation of outer HC loss indicates that our noise-exposure protocol can cause a PTS, while also providing evidence that as little as 2 h of noise exposure can permanently damage the cochlea [23]. When a TTS occurs, the ABR threshold returns to its normal level, but long-term damage to the synapses may exist, even in the absence of HC loss. The disruption of signaling between the inner HCs and type-1 afferent auditory nerve fibers causes degeneration of the auditory nerve fibers and spiral ganglia. In particular, if synapses connecting low-spontaneous-rate auditory nerve fibers deteriorate, communication may be disrupted due to the signals being less distinct from background noise [24]. Therefore, it is important to protect the auditory canals from noise, regardless of its intensity and duration.

Loss of HCs after noise exposure can lead to secondary damage, including auditory synaptopathy. Noise exposure not only damages the cochlea, but also triggers extensive changes in the central auditory

pathway. Sensory deprivation due to a decrease in the number of HCs can induce a reduction in the cell density in the upper auditory structures. For example, after overstimulation, the cell population in the ventral CN (VCN) has been shown to decrease due to apoptosis [25].

It is important to understand the molecular events that occur after acoustic trauma, in order to minimize damage to the central auditory pathway. A bioinformatics analysis by Alagramam et al. showed that exposure to both 116 and 110 dB noise could induce genes related to the MAPK signaling pathway For example, the Fos gene, which has a putative role in neuronal apoptosis and cell death, was induced in response to both treatments [26].

Acoustic overstimulation, unlike ablation, can lead to widespread axon degeneration and death. A study of cats with acoustic trauma demonstrated that new axons are able to grow in the VCN. Following noise exposure, cochlear nerve endings in the VCN degenerated over a period of several months and disappeared completely after 3 years. However, other small axons subsequently began to appear throughout the VCN, suggesting that degenerated neurons can reorganize the structure of the CN by generating new axons [27].

The CN is the primary point of convergence between auditory and somatosensory inputs. The balance of auditory sensory inputs can be disrupted by a decrease therein due to peripheral hearing loss. This imbalance, caused by increased excitatory activity and/or decreased inhibitory activity, enhances central neural receptivity and leads to hyperexcitability [28]. This change in the plasticity of the CN occurs in the form of axonal sprouting, which is regulated by the TGF-β signaling pathway [29]. However, such axonal sprouting can trigger tinnitus, which may itself be problematic for some patients [30]. Therefore, many researchers are trying to develop treatments that prevent axonal sprouting by inhibiting the TGF-β signaling pathway.

Among the potential genes targeted by the miRNAs described here, dual specificity protein phosphatase 10 (DUSP 10; Gene ID, 11221) is co-regulated by *miR-411-3p* and *miR-183-5p*. The expression levels of both *miR-411-3p* and *miR-183-5p* were shown to decrease after acoustic trauma, leading to an increase in the DUSP 10 level. The by-products of DUSP 10 inactivate p38 and stress-activated protein kinase/c-Jun NH(2)-terminal kinase (SAPK/JNK), thereby inhibiting the JNK pathway. As the JNK pathway serves as a major driver of apoptosis, inhibition thereof is likely to protect cells against apoptosis [31].

Profilin2 (PFN2) is also co-regulated by *miR-411-3p* and *miR-183-5p*. The expression levels of both *miR-411-3p* and *miR-183-5p* were shown to decrease after acoustic trauma, leading to an increase in PFN2. PFN2 is an actin binding protein that plays an important role in maintaining the structure of synapses in neural tissues [32]. In the case of TTS, afferent synapses are damaged and an increase in PFN2 expression may promote structural recovery of the damaged synapses [33].

### **4. Materials and Methods**

### *4.1. Study Design*

Noise-induced hearing loss (NIHL) was achieved by exposing subjects to 2 h of noise at a 115 dB sound pressure level (SPL). Tests of the auditory brainstem response (ABR) and histological examinations of the cochleae confirmed loss of hearing. Microarray analysis of the CN and IC tissues was used to identify candidate microRNAs (miRNAs). These miRNAs were validated using quantitative reverse transcription polymerase chain reaction (qRT-PCR) and target pathway analysis.

### *4.2. Animal Subjects*

All of the animal experiments described were approved (8 February 2018) by the Institutional Animal Care and Use Committee of Seoul National University Hospital (Seoul, Korea; 18-0025-C1A0), which is endorsed by the Association for the Assessment and Accreditation of Laboratory Animal Care International. The animals used in these experiments were kept under 12-h/12-h day/light cycle conditions, with free access to food and water. They were acclimated to laboratory conditions 1 week

prior to the initiation of these experiments. A total of 48 male Sprague–Dawley rats, aged 6 weeks, were randomly separated into four groups (all, *n* = 12). One group was assayed 1 day after noise exposure (day 1), one group was assayed 3 days after noise exposure (day 3), and the other two groups were used as the day 1 and day 3 controls.

### *4.3. Noise-Exposure Protocol*

Animals were anesthetized using a mixture of 40 mg/kg Zoletil (Zoletil 50; Virbac, Bogotá, Colombia) and 10 mg/kg xylazine (Rumpun; Bayer-Korea, Seoul, Korea) via an intramuscular injection before noise exposure. Each animal was placed in a separate wire cage to avoid unequal noise exposure, and each experiment was performed in a customized acrylic box in a sound-attenuating laboratory booth (900 mm × 900 mm× 1720 mm) with an electromagnetic shield. The animals were exposed to 2 h of broadband white noise at 115 dB SPL using a 2446-J compression driver (JBL Professional, Los Angeles, CA, USA) with an MA-620 power amplifier (Inkel, Incheon, Korea), in order to create the bilateral NIHL animal model (Supplementary Figure S2a,b). The sound intensity within the acrylic box was measured every hour using a CR152B sound level meter (Cirrus Research plc, Hunmanby, UK) to confirm that there were no alterations in the sound level during the noise-exposure treatments. The control animals were injected with the same dose of anesthetic and kept in the sound attenuating booth for the same period of time, without noise exposure [34]. Audiometry was performed at 4 h after noise-exposure treatments, to allow stable measurements to be recorded.

### *4.4. Auditory Brainstem Response (ABR) Recordings*

The hearing function of all animals was evaluated before noise exposure using the ABR. Animals were anesthetized and placed in sound-attenuating booths. Subdermal needle electrodes were positioned at the nape of the neck as the vertex, the ipsilateral mastoid as the negative, and the contralateral mastoid as the ground (Supplementary Figure S2c) [35]. Sound stimuli tone-bursts of 4, 8, and 16 kHz (duration, 1562 µm; CoS shaping, 21 Hz) were applied. High-frequency software (ver. 3.30; Intelligent Hearing Systems, Miami, FL, USA) and high-frequency transducers (HFT9911-20-0035; Intelligent Hearing Systems) were used to measure the ABR. Before obtaining the electroencephalography signal, the impedance between the electrodes was assessed to establish whether this was less than 2 kΩ. Responses to the signal were amplified approximately 100,000-fold and band-pass filtered (100–1500 Hz). The intensity of the stimuli ranged from 90 to 20 dB SPL in 5 dB increments. A total of 512 sweeps were averaged at each intensity level. Additional ABR measurements were recorded at 4 h, and on days 1 and 3 after noise exposure. The ABR threshold was defined as the smallest stimulus intensity level that produced a visible waveform for wave II or IV.

### *4.5. Cochlear Whole-Mount Surface Preparation*

Both control (*n* = 8) and noise-exposed (*n* = 8) animals were sacrificed under anesthesia. For each sample, the cochlea was detached from the temporal bone and fixed in 4% paraformaldehyde solution for 24 h at 4 ◦C. Fixed ears were washed three times in 1× phosphate-buffered saline (PBS) [36]. The thin layer of laminar bone covering the cochlea was trimmed under a SZ2-ILST stereomicroscope (Olympus Corporation, Tokyo, Japan) using a drill (Strong 90; Saeshin Precision, Daegu, Korea) with a 2 mm-diameter diamond burr attachment (Supplementary Figure S3a). A hole was created by breaking the bone between the oval and round windows using very fine forceps (Supplementary Figure S3b,c). The laminar bone was removed using a conventional 1-mm syringe needle (Supplementary Figure S3d–f). The cochlear nerve was cut and the spiral structure of the organ of Corti was isolated. Next, the stria vascularis and Reissner's membrane were removed (Supplementary Figure S3g,h). The first turn from the top of the organ of Corti was removed using scissors. This was called the 'apical turn' section. A second turn was removed and called the 'middle turn' section and a final half-turn section was removed and called the 'basal turn' section

(Supplementary Figure S3i). The sections were placed in 1× PBS solution to prevent them from drying out.

### *4.6. Outer HC Staining*

Phalloidin was used to stain F-actin, and the photostable orange fluorescent Alexa Fluor 546 dye was used to visualize the cuticular plate and stereocilia within the HCs. After surface preparation, the isolated spiral structure of the organ of Corti was incubated in a mixed solution of 0.3% Triton X-100 and Alexa Fluor 546 phalloidin (1:100 dilution; Invitrogen, Carlsbad, CA, USA) for 45 min at room temperature in a lightproof box [37]. The sample was washed three times in 1× PBS and separated into three segments using Vannas capsulotomy scissors (E-3386; Karl Storz SE & Co. KG, Tuttlingen, Germany), consisting of the apical, middle, and basal turn sections. The first complete turn from the apex was the apical turn, the next complete turn was the middle turn, and the final half-turn was the basal turn. Each turn section was mounted on a slide using ProLong™ Gold Antifade mountant (P36930; Invitrogen) to prevent the fluorescent dyes from fading. Images were generated using a STED CW confocal laser scanning system (Leica, Wetzlar, Germany) and HCs within the images were counted.

### *4.7. Cochlear Histology*

Cochleae from both control (*n* = 8) and noise-exposed (*n* = 8) rats were fixed and washed. Samples were decalcified using 10% (*w*/*v*) ethylenediaminetetraacetic acid (Santa Cruz Biotechnology, Dallas, TX, USA) in 1× PBS for 4 weeks. A histological examination was performed weekly to determine when the samples were ready for the embedding procedure. The tissues were dehydrated using a series of ethanol washes, and the ethanol was then removed using xylene. After the ethanol was removed, tissues were infiltrated with paraffin wax [38] in a PELORIS II tissue processing system (Leica). The processed cochleae were embedded in a stainless mold and trimmed into 4-µm-thick sagittal sections using a RM2255 microtome (Leica). The sections were deparaffinized for 1 h at 60 ◦C in a dry oven and cleaned using a series of ethanol washes. Next, the nuclei were stained for 7 min with hematoxylin (DAKO, Jena, Germany) and the cytoplasm was stained for 30 s with eosin Y (Sigma-Aldrich, St. Louis, MO, USA). The stained slides were dehydrated and preserved in 70% ethanol [39]. After mounting, the organ of Corti was examined using a light microscope (ECLIPSE Ci-L; Nikon, Tokyo, Japan).

### *4.8. RNA Extraction*

Whole brain tissue was harvested, and the bilateral CN and IC were dissected out using a brain matrix. The locations of the CN (−9.30 to −11.30 mm from the bregma) and the IC (–8.30 to –9.30 mm from the bregma) were determined according to the rat brain atlas [40] (Supplementary Figure S4a–c). Tissues were frozen in liquid nitrogen immediately after removal and stored at −80 ◦C. The harvested tissue was lysed in 1 mL QIAzol solution using a TissueLyzerII (Qiagen, Hilden, Germany) and incubated at room temperature for 5 min. The samples were placed on a vortex mixer after adding 200 µL chloroform to each and then incubated at room temperature for 3 min. Next, the samples were centrifuged at 12,000× *g* for 15 min at 4 ◦C and the upper aqueous phase containing the RNA was removed to a fresh tube. A total of 500 µL isopropyl alcohol was added to each tube. The tubes were then inverted and incubated at room temperature for 10 min. Thereafter, the tubes were centrifuged at 7500× *g* for 5 min at 4 ◦C and the RNA pellets were washed twice with 1 mL 75% ethanol. The pellets were dried for approximately 5 min and redissolved in RNase-free water.

### *4.9. Analysis of miRNA Arrays*

The analysis of miRNAs was performed by Ebiogen Inc. (Seoul, Korea) using the Affymetrix GeneChip miRNA 4.0 array (Affymetrix, Santa Clara, CA, USA). A total of 24 animals were randomly separated into two treatment groups and two corresponding control groups. Treated rats were exposed

to noise and assayed after 1 or 3 days. After hearing loss was confirmed, the CN and IC from two animals from the same group were combined and treated as a single sample, and three samples from each group were used for the analysis. Extracted total RNA was assessed for quality and quantity using a Bioanalyzer 2100 system (Agilent, Santa Clara, CA, USA). A total of 250 ng of RNA was analyzed. After ligating biotin-labeled 3DNA dendrimers, each RNA strand was labeled using poly-A polymerase. The biotinylated RNA strands were hybridized for 18 h at 48 ◦C on an Affymetrix GeneChip miRNA 4.0 array. The hybridized GeneChip was washed and stained using an Affymetrix 450 Fluidics station. Fluorescence signals from the 3DNA dendrimers were detected using an Affymetrix GeneChip 3000 7G scanner.

### *4.10. Quantitative Reverse Transcription Polymerase Chain Reaction (qRT-PCR)*

Using an miScript® II RT kit (Qiagen, Hilden, Germany), 2 µg of RNA was mixed with reverse-transcription master mix and incubated for 60 min at 37 ◦C. To inactivate the miScript reverse transcriptase, the mixture was incubated for 5 min at 95 ◦C and then placed on ice. A total of 20 µL of cDNA was diluted to 1:16 and used as template cDNA. The miScript SYBR® Green PCR kit (Qiagen) was used with miScript Primer Assay reagents (Qiagen) for qRT-PCR. U6 small nuclear RNA was used as an endogenous control gene [41]. The miScript Primer Assay reagents and the reaction mix were dispensed into wells containing template cDNA. The PCR plate was sealed with film and centrifuged at 1000× *g* for 1 min at room temperature. Initial activation was performed for 15 min at 95 ◦C. The reactions consisted of 40 cycles of denaturation, annealing, and extension, and fluorescence data were collected during the extension phase. The reactions were performed using an ABI 7500 real-time PCR system (Applied Biosystems, Foster City, CA, USA). Relative quantification values were obtained for each of the target genes using the observed cycle threshold (Ct) results and the 2 <sup>−</sup>∆∆*C*<sup>t</sup> method.

### *4.11. Pathway Analysis of Candidate miRNAs*

For the CN and IC, a total of 10 and 13 candidate miRNAs were selected from the microarray analysis based on 1.5-fold changes in normalized intensity values (*p* < 0.1) respectively. Of these, five and three miRNAs, respectively, were selected following qRT-PCR validation. Using DIANA-miRPath software (ver. 3.0), a KEGG pathway analysis was performed using DIANA-microT-CDS (ver. 5.0; DIANA TOOLS, http://diana.imis.athena-innovation.gr/DianaTools/index.php?r=microT\_CDS/index) with a threshold of 0.8 and a false discovery rate correction [42,43]. A total of 12 and 14 KEGG pathways were identified for the CN and IC, respectively, using a gene union module and a *p*-value threshold of 0.05 and 0.3.

### *4.12. Statistical Analyses*

All data are expressed as the means ± standard error of the mean, and all data were analyzed using SPSS software (ver. 25; IBM, Armonk, NY, USA). An F-test was performed to determine whether the levels of variation within the groups were equal. After the F-test, data were analyzed using Student's t-tests to identify significant differences between groups. A *p*-value of <0.05 was considered statistically significant.

### **5. Conclusions**

Using a noise exposure animal model, we were able to show that even acute short-term noise exposure can lead to hearing loss. Changes in the ABR amplitude of wave II suggest an alteration in either synaptic transmission or the number of neuronal cells. To investigate the role of miRNAs in the central auditory pathway, CN and IC were compared in both the treatment and control groups, with microarray analysis and qRT-PCR results suggesting that *miR-200b-3p*, *miR-183-5p*, *miR-411-3p*, *miR-20b-5p*, *miR-377-3p*, *miR-92a-1-5p*, *miR-136-3p*, and *miR-26b-5p* may play key roles in the neuroplasticity of the central auditory pathway. Using the KEGG database, we found that

five of these candidate miRNAs may be involved in the MAPK signaling pathway, axon guidance, and the neurotrophin signaling pathway in the CN, while an additional three candidate miRNAs may influence the MAPK signaling pathway in the IC. Further validation of these candidate miRNAs will be achieved using miRNA oligomers such as mimics and inhibitors, in order to better refine the specific signaling pathways underlying these processes. These target miRNAs, which play crucial roles in the central auditory pathway, can be used for diagnosis in the early stage of NIHL, and for treatment of the damage caused by the cellular and physiological changes after NIHL.

**Supplementary Materials:** Supplementary Materials can be found at http://www.mdpi.com/1422-0067/21/22/ 8792/s1.

**Author Contributions:** Conceptualization, J.H.L. and S.H.O.; methodology, B.-G.K.; validation, B.-G.K. and S.H.H.; formal analysis, B.-G.K. and S.H.H.; investigation, M.-W.S. and S.H.O.; resources, M.-W.S. and J.H.L.; data curation, S.H.H.; writing—original draft preparation, S.P. and M.K.P.; writing—review and editing, S.P. and M.K.P.; visualization, S.P.; supervision, J.H.L. and S.H.O.; project administration, S.H.H. and S.P.; funding acquisition, S.P. and M.K.P. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was supported by the Basic Science Research Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Science, ICT & Future Planning (NRF-2017R1D1A 1B03034832) and the SNUH Research Fund (03-2020-0240).

**Conflicts of Interest:** The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript; or in the decision to publish the results.

### **Abbreviations**


### **References**


**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

## *Article* **Regulator of G Protein Signalling 4 (RGS4) as a Novel Target for the Treatment of Sensorineural Hearing Loss**

**Christine Fok, Milan Bogosanovic, Madhavi Pandya, Ravindra Telang, Peter R. Thorne and Srdjan M. Vlajkovic \***

Department of Physiology and The Eisdell Moore Centre, Faculty of Medical and Health Sciences, The University of Auckland, Private Bag 92019, Auckland 1142, New Zealand; c.fok@auckland.ac.nz (C.F.); milan.bogosanovic@uq.net.au (M.B.); m.pandya@auckland.ac.nz (M.P.); r.telang@auckland.ac.nz (R.T.); pr.thorne@auckland.ac.nz (P.R.T.)

**\*** Correspondence: s.vlajkovic@auckland.ac.nz; Tel.: +64-9-9239782

**Abstract:** We and others have previously identified signalling pathways associated with the adenosine A<sup>1</sup> receptor (A1R) as important regulators of cellular responses to injury in the cochlea. We have shown that the "post-exposure" treatment with adenosine A1R agonists confers partial protection against acoustic trauma and other forms of sensorineural hearing loss (SNHL). The aim of this study was to determine if increasing A1R responsiveness to endogenous adenosine would have the same otoprotective effect. This was achieved by pharmacological targeting of the Regulator of G protein Signalling 4 (RGS4). RGS proteins inhibit signal transduction pathways initiated by G protein-coupled receptors (GPCR) by enhancing GPCR deactivation and receptor desensitisation. A molecular complex between RGS4 and neurabin, an intracellular scaffolding protein expressed in neural and cochlear tissues, is the key negative regulator of A1R activity in the brain. In this study, Wistar rats (6–8 weeks) were exposed to traumatic noise (110 dBSPL, 8–16 kHz) for 2 h and a small molecule RGS4 inhibitor CCG-4986 was delivered intratympanically in a Poloxamer-407 gel formulation for sustained drug release 24 or 48 h after noise exposure. Intratympanic administration of CCG-4986 48 h after noise exposure attenuated noise-induced permanent auditory threshold shifts by up to 19 dB, whilst the earlier drug administration (24 h) led to even better preservation of auditory thresholds (up to 32 dB). Significant improvement of auditory thresholds and suprathreshold responses was linked to improved survival of sensorineural tissues and afferent synapses in the cochlea. Our studies thus demonstrate that intratympanic administration of CCG-4986 can rescue cochlear injury and hearing loss induced by acoustic overexposure. This research represents a novel paradigm for the treatment of various forms of SNHL based on regulation of GPCR.

**Keywords:** sensorineural hearing loss; noise-induced cochlear injury; cochlear rescue; otoprotection; adenosine A<sup>1</sup> receptor; regulator of G protein signalling 4; CCG-4986; intratympanic drug delivery

### **1. Introduction**

Hearing loss is the most prevalent form of sensory impairment, affecting about 466 million people worldwide including 34 million children [1]. Most of the hearing loss is sensorineural due to disease, degeneration, or trauma to the cochlea of the inner ear [2]. Treatment options for sensorineural hearing loss (SNHL) are currently limited to prosthetic devices such as hearing aids and cochlear implants. Both devices can partly restore auditory function, but these have limitations because the ear remains damaged. There is thus a significant need for the development of effective therapies to prevent cochlear damage and hearing loss or restore cochlear sensorineural structure and hearing. We have identified that signalling pathways activated by adenosine receptors are important regulators of cellular responses to injury in cochlear tissues. Animal studies reveal that stimulation of the A<sup>1</sup> adenosine receptor (A1R) is particularly promising for the treatment of acute noise-induced cochlear injury [3,4] and other forms of SNHL such as from cytotoxic drugs, including cisplatin and aminoglycoside antibiotics [5,6]. The principal advantage of this approach

**Citation:** Fok, C.; Bogosanovic, M.; Pandya, M.; Telang, R.; Thorne, P.R.; Vlajkovic, S.M. Regulator of G Protein Signalling 4 (RGS4) as a Novel Target for the Treatment of Sensorineural Hearing Loss. *Int. J. Mol. Sci.* **2021**, *22*, 3. https://dx.doi.org/ 10.3390/ijms22010003

Received: 27 November 2020 Accepted: 19 December 2020 Published: 22 December 2020

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/ licenses/by/4.0/).

is that the A1R stimulation affects multiple mechanisms of cochlear injury (e.g., oxidative stress, glutamate excitotoxicity, activation of apoptotic pathways), thus providing comprehensive protection from SNHL [7]. One of the issues with advancing these approaches to clinical applicability is the delivery of receptor agonists to the inner ear and their suitability for long-term therapy.

As an alternative to the use of exogenous A1R agonists, we have previously considered other strategies that would regulate the action of endogenous adenosine, for example by manipulating intracellular adenosine metabolism [8]. Recently, we have identified a novel otoprotective paradigm based on increasing A1R responsiveness to endogenous adenosine, which can be achieved by inhibiting the Regulator of G protein Signalling 4 (RGS4). RGS is a large family of proteins that inhibit signal transduction pathways initiated by G protein-coupled receptors (GPCR) including A1R [9–11]. RGS increase the intrinsic GTPase activity of G proteins and thus enhance G protein inactivation and promote receptor desensitisation [9].

In the past two decades, RGS proteins have received increasing interest as potential drug targets in cardiovascular disease, CNS disorders and several types of cancer [12–15]. Targeted inhibition of RGS proteins could potentially provide a way to fine-tune GPCR signalling by potentiating or prolonging the effect of receptor agonists. This approach could also be used to enhance endogenous ligand effects on GPCR. A few small molecule RGS inhibitors have been identified, particularly in the well-studied RGS4 family [16,17]. The selectivity of RGS proteins is mediated either through direct interaction with target receptors, or through selective interactions with accessory proteins [16]. For example, the neurabin-RGS4 molecular complex regulates A1R signalling events in the brain. Neurabin is an intracellular scaffolding protein (protein phosphatase 1 regulatory inhibitor subunit 9a, PPP1R9A) expressed in neural tissues which facilitates interactions of RGS4 with the A1R [18]. After A1R stimulation by endogenous or exogenous ligands, neurabin forms a complex with RGS4 and recruits it to the cell surface to the A1R [18]. Disruption of the neurabin-RGS4 complex, either by genetic deletion of neurabin or by selective inhibitors of RGS4, enhances A1R signalling even without administration of exogenous A1R ligands. Mice with genetic deletion of neurabin are protected against kainate-induced seizures, evidenced by reduced severity and occurrence of seizures, improved neuronal survival and overall lower mortality [18]. Similarly, this anticonvulsant and neuroprotective effect is also conferred by CCG-4986, a small molecule inhibitor of RGS4 [18]. The neurabin/RGS4 complex thus appears to be the key regulator of A1R activity in the brain and a promising neuroprotective target.

Here, we investigated an otoprotective strategy based on inhibition of the neurabin/RGS4 complex in the cochlea, with the aim to increase A1R responsiveness to endogenous adenosine released from cochlear tissues during acoustic stress. This novel otoprotective strategy is based on intratympanic injection of a small molecule RGS4 inhibitor to the round window membrane (RWM) of the cochlea.

### **2. Results**

### *2.1. Expression and Immunolocalisation of RGS4 and Neurabin I in the Cochlea*

RT-PCR demonstrated the expression of neurabin isoforms I and II in the rat cochlea (Figure 1A). Generated PCR products corresponded to the predicted sizes of DNA fragments (Table 1). Omitting reverse transcriptase (-RT) in control reactions resulted in the absence of reaction products (Figure 1A). As only the neurabin I isoform makes complexes with RGS4 and the A1R, immunolocalisation studies were only performed for this isoform. Neurabin I was immunolocalised in the inner and outer hair cells in the organ of Corti and cell bodies of the spiral ganglion neurons (Figure 1B). Neurabin I distribution in sensory hair cells and spiral ganglion neurons coincided with the RGS4 immunofluorescence pattern (Figure 1C). In addition, RGS4 immunofluorescence was observed in the auditory nerve fibres in the osseous spiral lamina and blood vessels in the spiral limbus (Figure 1C).

Neurabin I (NM\_053473)

Neurabin II

No immunofluorescence was detected when the primary antibody was replaced with control mouse IgG (Figure 1D) or control rabbit IgG (not shown). (NM\_053474) position: 1976–1995 position: 2536–2517 Expected amplicon length: Neurabin I—520bp, Neurabin II—561bp.

**Primer Forward (Sense) Reverse (Antisense)** 

5′-GGAGCCGTTAGAAGATGCTG-3′ position: 1247–1266

5′-GAGTGGAGAGGTTGGAGCTG-3′

*Int. J. Mol. Sci.* **2021**, *22*, x FOR PEER REVIEW 3 of 20

control mouse IgG (Figure 1D) or control rabbit IgG (not shown).

pattern (Figure 1C). In addition, RGS4 immunofluorescence was observed in the auditory nerve fibres in the osseous spiral lamina and blood vessels in the spiral limbus (Figure 1C). No immunofluorescence was detected when the primary antibody was replaced with

**Table 1.** Primer sequences and positions for neurabin I and II isoforms*.* 

5′-CCCATCCTCATCTTTCTCCA-3′ position: 1766–1747

5′-GGAGCTCCTTGAACTTGTGC-3′

**Figure 1.** (**A**) Expression of Neurabin I and II isoforms in the rat cochlea. PCR products (shown in duplicates) correspond to the neurabin I (520 bp) and neurabin II (561 bp) isoforms, respectively. In the absence of reverse transcriptase (-RT controls) no amplification was observed. (**B**) The most prominent neurabin I immunostaining was observed in sensory inner hair cells (IHC) and outer hair cells (OHC) in the organ of Corti (o/C) and spiral ganglion neurons (SGN). (**C**) The RGS4 antibody also demonstrated predilection for sensory hair cells (IHC and OHC) and SGN. RGS4 immunofluorescence was also observed in the auditory nerve fibres in the osseous spiral lamina (OSL) and blood vessels in the spiral limbus (SL). (**D**) Control section where the primary antibody was replaced by IgG isotype control. Abbreviations: DC, Deiters' cells; RM, Reissner's membrane; SV, stria vascularis. Scale bars, 50 μM. **Figure 1.** (**A**) Expression of Neurabin I and II isoforms in the rat cochlea. PCR products (shown in duplicates) correspond to the neurabin I (520 bp) and neurabin II (561 bp) isoforms, respectively. In the absence of reverse transcriptase (-RT controls) no amplification was observed. (**B**) The most prominent neurabin I immunostaining was observed in sensory inner hair cells (IHC) and outer hair cells (OHC) in the organ of Corti (o/C) and spiral ganglion neurons (SGN). (**C**) The RGS4 antibody also demonstrated predilection for sensory hair cells (IHC and OHC) and SGN. RGS4 immunofluorescence was also observed in the auditory nerve fibres in the osseous spiral lamina (OSL) and blood vessels in the spiral limbus (SL). (**D**) Control section where the primary antibody was replaced by IgG isotype control. Abbreviations: DC, Deiters' cells; RM, Reissner's membrane; SV, stria vascularis. Scale bars, 50 µM.

**Table 1.** Primer sequences and positions for neurabin I and II isoforms.


Expected amplicon length: Neurabin I—520 bp, Neurabin II—561 bp.

#### *2.2. Auditory Brainstem Responses and the Effect of Treatment 48 Hours after Noise Exposure* Auditory brainstem responses (ABR) were used to measure auditory thresholds prior to noise exposure (baseline), and 16 days after noise exposure (final) to determine noise-

*Int. J. Mol. Sci.* **2021**, *22*, x FOR PEER REVIEW 4 of 20

Auditory brainstem responses (ABR) were used to measure auditory thresholds prior to noise exposure (baseline), and 16 days after noise exposure (final) to determine noiseinduced threshold shifts. induced threshold shifts. The baseline auditory thresholds were similar in noise-exposed and non-exposed animals (Figure 2A–C). The control group exposed to ambient noise levels in the animal

*2.2. Auditory Brainstem Responses and the Effect of Treatment 48 Hours after Noise Exposure* 

The baseline auditory thresholds were similar in noise-exposed and non-exposed animals (Figure 2A–C). The control group exposed to ambient noise levels in the animal facility showed no change in ABR thresholds 16 days after initial ABR measurement (Figure 2A). In contrast, exposure to octave band noise (8–16 kHz at 110 dB SPL) for two hours induced a permanent threshold shift (PTS) in both drug- and vehicle-treated animals (Figure 2B–D). In the control vehicle-treated group, the average threshold shift was between 40 and 50 dB at frequencies above 4 kHz (Figure 2B,D). ABR thresholds were still elevated after treatment with CCG-4986 (100 µM), but to a lesser degree than in control drug vehicle-treated animals (Figure 2C). At mid-to-high frequencies (8–28 kHz), CCG-4986 treatment reduced noise-induced PTS by 10–19 dB compared to the vehicletreated animals (Figure 2D). The greatest improvement of ABR thresholds was observed at 8–16 kHz (Figure 2D), with average threshold shift reductions of 19 dB at 8 kHz (*p* = 0.029), 19 dB at 12 kHz (*p* = 0.018) and 17 dB at 16 kHz (*p* = 0.013). facility showed no change in ABR thresholds 16 days after initial ABR measurement (Figure 2A). In contrast, exposure to octave band noise (8–16 kHz at 110 dB SPL) for two hours induced a permanent threshold shift (PTS) in both drug- and vehicle-treated animals (Figure 2B–D). In the control vehicle-treated group, the average threshold shift was between 40 and 50 dB at frequencies above 4 kHz (Figure 2B,D). ABR thresholds were still elevated after treatment with CCG-4986 (100 μM), but to a lesser degree than in control drug vehicle-treated animals (Figure 2C). At mid-to-high frequencies (8–28 kHz), CCG-4986 treatment reduced noise-induced PTS by 10–19 dB compared to the vehicle-treated animals (Figure 2D). The greatest improvement of ABR thresholds was observed at 8–16 kHz (Figure 2D), with average threshold shift reductions of 19 dB at 8 kHz (*p* = 0.029), 19 dB at 12 kHz (*p* = 0.018) and 17 dB at 16 kHz (*p* = 0.013).

**Figure 2.** Baseline and final auditory brainstem responses (ABR) thresholds for 4–28 kHz tone pips in animals exposed to octave band noise (8–16 kHz, 110 dB SPL) for 2 h and non-exposed animals. Grey area denotes noise band. (**A**) Controls exposed to ambient noise (55–65 dB SPL) in the animal facility. (**B**) Noise-exposed vehicle-treated control group. (**C**) Animals treated with CCG-4986 (100 μM) 48 h post-exposure. (**D**) Comparison of permanent threshold shifts 16 days post-exposure in drug- and vehicle-treated animals. Data presented as mean ± SEM. No noise control, *n* = 10; Vehi-**Figure 2.** Baseline and final auditory brainstem responses (ABR) thresholds for 4–28 kHz tone pips in animals exposed to octave band noise (8–16 kHz, 110 dB SPL) for 2 h and non-exposed animals. Grey area denotes noise band. (**A**) Controls exposed to ambient noise (55–65 dB SPL) in the animal facility. (**B**) Noise-exposed vehicle-treated control group. (**C**) Animals treated with CCG-4986 (100 µM) 48 h post-exposure. (**D**) Comparison of permanent threshold shifts 16 days post-exposure in drug- and vehicle-treated animals. Data presented as mean ± SEM. No noise control, *n* = 10; Vehicle control, *n* = 16; CCG-4986, *n* = 17. \* *p* < 0.05; Two-way ANOVA followed by Holm–Sidak post-hoc test.

#### cle control, *n* = 16; CCG-4986, *n* = 17. \* *p* < 0.05; Two-way ANOVA followed by Holm–Sidak post-*2.3. Input–Output Functions*

hoc test. *2.3. Input–Output Functions*  To further investigate the otoprotective effects of CCG-4986 treatment, three frequencies (4 kHz, 16 kHz and 28 kHz), were selected as representative of the low, mid, and high frequency regions of the cochlea, respectively. The amplitudes and latencies of wave I were analysed at suprathreshold intensities (80, 85, and 90 dB; Figure 3A).

To further investigate the otoprotective effects of CCG-4986 treatment, three frequencies (4 kHz, 16 kHz and 28 kHz), were selected as representative of the low, mid, and high frequency regions of the cochlea, respectively. The amplitudes and latencies of wave I

were analysed at suprathreshold intensities (80, 85, and 90 dB; Figure 3A).

**Figure 3.** (**A**) Representative traces of ABR Wave I prior to noise exposure (green) and post-exposure (red). P<sup>1</sup> represents a positive peak and N<sup>1</sup> a negative peak. ABR responses were recorded at 16 kHz at suprathreshold intensities (80–90 dB SPL). Average baseline and final ABR wave I amplitudes (**B**) and latencies (**C**) at suprathreshold intensities in noise-exposed animals treated with CCG-4986 (100 µM) or drug vehicle solution 48 h after noise exposure. Solid lines represent baseline amplitudes/latencies and dashed lines represent final amplitudes/latencies. Data presented as mean ± SEM, Vehicle control, *n* = 16; CCG-4986, *n* = 17. \* *p* < 0.05; Multivariate ANOVA followed by planned contrast comparisons.

Prior to noise exposure, ABR Wave I amplitudes and latencies were similar in drugand vehicle-treated rats at each test frequency. As expected, wave I amplitudes decreased, and latencies increased after noise exposure in all animals (Figure 3B,C).

The vehicle-treated control animals showed an average 60–70% reduction in wave I amplitude at 90 dB, with 16 kHz as the most affected frequency (70% reduction). CCG-4986 treatment improved wave I amplitudes compared to vehicle-treated rats, which was significant for 4 kHz (*p* = 0.03) and 16 kHz (*p* = 0.05) at 90 dB (Figure 3B). At 28 kHz the effect of CCG-4986 treatment was not significant (*p* = 0.052).

After noise exposure, wave I latency increased by 10–22% in vehicle-treated animals, with the greatest increase at 16 kHz (Figure 3C). CCG-4986 treatment significantly (*p* < 0.05) reduced wave I latencies 4 kHz and 16 kHz compared to vehicle-treated animals, although the latencies remained elevated relative to pre-noise levels (Figure 3C).

### *2.4. Hair Cell Survival*

As expected from exposure to the octave band noise (8–16 kHz), turn-related differences in hair cell loss were observed in both vehicle- and drug-treated animals. The loss of outer hair cells (OHC; Figure 4A) exceeded the loss of the inner hair cells (IHC; Figure 4C) in both the middle and the basal segment of the cochlea (Figure 4B). The most heavily affected region was the basal segment of the cochlea (Figure 4B), whilst there was virtually no hair cell loss in the apical segment (data not shown).

The average OHC loss in vehicle-treated animals was 38% ± 10.1% and 85% ± 5.7% in the middle and basal segments, respectively (Figure 4B). Treatment with CCG-4986 reduced OHC loss to 8.7% ± 4% (*p* = 0.0027) in the middle segment and to 25% ± 5.4% (*p* < 0.001) in the basal segment (Figure 4B).

The average IHC loss in the middle turn was low (~2%), and there was no difference between the drug- and vehicle-treated animals (Figure 4B). In contrast, the survival of IHC in the basal turn was significantly improved with CCG-4986 treatment (12.0% ± 3.5%) compared to vehicle-treated controls (32% ± 6.4%; *p* = 0.0008).

### *2.5. Synaptic Ribbon Counts*

Confocal imaging of afferent synapses in controls exposed to ambient noise showed turn-related differences in the average number of paired synapses per IHC (Figure 5A–C). The greatest number of synapses was found in the middle turn (Figure 5B), with an average of 24.3 ± 0.4 paired synapses per IHC. In comparison, apical and basal turns had an average of 19.8 ± 0.2 and 22.4 ± 0.6 synapses per IHC respectively (Figure 5A,C). To identify and quantify afferent synapses, whole mounts of the organ of Corti were immunolabelled with antibodies to CtBP2 (component of the presynaptic ribbon), GluA2 subunit (postsynaptic glutamate receptor) and myosin VIIa (IHC) (Figure 5E–H).

Under ambient noise levels (no-noise controls), vast majority of synaptic ribbons were paired with the post-synaptic GluA2 receptor (Figure 5F). Only about 1% of synapses were characterised as orphan synapses (unpaired pre-synaptic ribbon or post-synaptic glutamate receptor). The percentage of orphan synapses, however increases with noise exposure (Figure 5D,E,G,H).

The number of synapses in the apical turn was similar in non-exposed and noiseexposed animals, regardless of the treatment (Figure 5A). At higher frequencies, noise exposure significantly reduced the number of paired synapses. In animals treated with drug vehicle solution, the number of paired synapses decreased to 15.9 ± 2.0 and 17.0 ± 1.8 in the middle and basal turns respectively (Figure 5B,C). There was also a significant increase in the proportion of orphan synapses in these regions (10% and 8% respectively; Figure 5D).

**Figure 4.** (**A**) Representative images of OHC in the middle and the basal segment of the cochlea in noise-exposed vehicle-treated and CCG-4986 treated animals. (**B**) Percentage of hair cell loss (OHC and IHC) in the middle and basal cochlear segments of vehicle- and CCG-4986 treated animals. Data presented as mean ± SEM, vehicle control *n* = 16, CCG-4986 *n* = 17. \*\* *p* <0.01, \*\*\* *p* < 0.0001; Two-way ANOVA followed by Holm–Sidak post-hoc test. (**C**) Representative images of IHC in the middle and the basal segment of the cochlea of vehicle-treated and CCG-4986 treated animals. **Figure 4.** (**A**) Representative images of OHC in the middle and the basal segment of the cochlea in noise-exposed vehicletreated and CCG-4986 treated animals. (**B**) Percentage of hair cell loss (OHC and IHC) in the middle and basal cochlear segments of vehicle- and CCG-4986 treated animals. Data presented as mean ± SEM, vehicle control *n* = 16, CCG-4986 *n* = 17. \*\* *p* <0.01, \*\*\* *p* < 0.0001; Two-way ANOVA followed by Holm–Sidak post-hoc test. (**C**) Representative images of IHC in the middle and the basal segment of the cochlea of vehicle-treated and CCG-4986 treated animals. White arrows point at the sensory hair cells and red arrows at spaces with missing hair cells.

White arrows point at the sensory hair cells and red arrows at spaces with missing hair cells. The average OHC loss in vehicle-treated animals was 38% ± 10.1% and 85% ± 5.7% in the middle and basal segments, respectively (Figure 4B). Treatment with CCG-4986 reduced OHC loss to 8.7% ± 4% (*p* = 0.0027) in the middle segment and to 25% ± 5.4% (*p* < 0.001) in the basal segment (Figure 4B). Treatment with CCG-4986 significantly (*p* = 0.0085) reduced the loss of synapses in the middle turn (Figure 5B) but did not improve the survival of afferent synapses in the basal turn (Figure 5C). In the middle turn, the number of paired synapses improved to 21.1 ± 1.3 synapses per IHC in CCG-4986-treated animals, which was similar to the number of synapses in non-noise exposed controls (Figure 5B).

The average IHC loss in the middle turn was low (~2%), and there was no difference between the drug- and vehicle-treated animals (Figure 4B). In contrast, the survival of IHC in the basal turn was significantly improved with CCG-4986 treatment (12.0% ± 3.5%)

compared to vehicle-treated controls (32% ± 6.4%; *p* = 0.0008).

identify and quantify afferent synapses, whole mounts of the organ of Corti were immunolabelled with antibodies to CtBP2 (component of the presynaptic ribbon), GluA2 subu-

nit (postsynaptic glutamate receptor) and myosin VIIa (IHC) (Figure 5E–H).

**Figure 5.** The average number of synapses per IHC at various frequency regions of the cochlea. (**A**) Number of paired synapses in the apical turn (low frequency region), (**B**) Middle turn (mid frequency region) and (**C**) Basal turn (high frequency region). (**D**) Number of orphaned synapses in all turns expressed as a percentage of total synapses per IHC. Data presented as mean ± SEM. No-noise control, *n* = 10, vehicle-treated, *n* = 16, CCG-4986 treated, *n* = 17. \* *p* < 0.05, \*\* *p* < 0.01, n.s. not significant. One-way ANOVA followed by Holm–Sidak post-hoc test. (**E**) Representative images of IHC-auditory nerve synapses in the middle turn for controls exposed to traumatic or ambient noise. Labelling shows IHCs (grey), post-synaptic glutamate receptors (green) and pre-synaptic ribbons (red). (**F**–**H**) High power projection of IHC synapses in (**F**) No-noise controls, (**G**) Noise-exposed vehicle-treated and (**H**) Noise-exposed CCG-4986-treated animals. White arrows indicate paired ribbon synapses, green arrows orphaned post-synaptic glutamate receptors, and **Figure 5.** The average number of synapses per IHC at various frequency regions of the cochlea. (**A**) Number of paired synapses in the apical turn (low frequency region), (**B**) Middle turn (mid frequency region) and (**C**) Basal turn (high frequency region). (**D**) Number of orphaned synapses in all turns expressed as a percentage of total synapses per IHC. Data presented as mean ± SEM. No-noise control, *n* = 10, vehicle-treated, *n* = 16, CCG-4986 treated, *n* = 17. \* *p* < 0.05, \*\* *p* < 0.01, n.s. not significant. One-way ANOVA followed by Holm–Sidak post-hoc test. (**E**) Representative images of IHC-auditory nerve synapses in the middle turn for controls exposed to traumatic or ambient noise. Labelling shows IHCs (grey), post-synaptic glutamate receptors (green) and pre-synaptic ribbons (red). (**F**–**H**) High power projection of IHC synapses in (**F**) No-noise controls, (**G**) Noise-exposed vehicle-treated and (**H**) Noise-exposed CCG-4986-treated animals. White arrows indicate paired ribbon synapses, green arrows orphaned post-synaptic glutamate receptors, and red arrows orphaned pre-synaptic ribbons.

#### Under ambient noise levels (no-noise controls), vast majority of synaptic ribbons *2.6. Spiral Ganglion Neuron Counts*

red arrows orphaned pre-synaptic ribbons.

were paired with the post-synaptic GluA2 receptor (Figure 5F). Only about 1% of synapses were characterised as orphan synapses (unpaired pre-synaptic ribbon or post-synaptic glutamate receptor). The percentage of orphan synapses, however increases with noise exposure (Figure 5D,E,G,H). The number of synapses in the apical turn was similar in non-exposed and noiseexposed animals, regardless of the treatment (Figure 5A). At higher frequencies, noise ex-At ambient sound levels (non-exposed controls), the average SGN density in the middle turn of the cochlea was 2530 <sup>±</sup> 72 cells/mm<sup>2</sup> (Figure 6A,D). Noise exposure induced a significant (*p* < 0.0001) loss of SGNs in both vehicle- and drug-treated animals (Figure 6B–D). CCG-4986 treatment did not reduce the loss of SGN, and the average cell densities for CCG-4986 treated (2209 <sup>±</sup> 97 cells/mm<sup>2</sup> ) and vehicle controls (2183 ± 161 cells/mm<sup>2</sup> ) were similar (Figure 6D).

posure significantly reduced the number of paired synapses. In animals treated with drug vehicle solution, the number of paired synapses decreased to 15.9 ± 2.0 and 17.0 ± 1.8 in the middle and basal turns respectively (Figure 5B,C). There was also a significant increase

5D).

ber of synapses in non-noise exposed controls (Figure 5B).

*2.6. Spiral Ganglion Neuron Counts* 

similar (Figure 6D).

in the proportion of orphan synapses in these regions (10% and 8% respectively; Figure

Treatment with CCG-4986 significantly (*p* = 0.0085) reduced the loss of synapses in the middle turn (Figure 5B) but did not improve the survival of afferent synapses in the basal turn (Figure 5C). In the middle turn, the number of paired synapses improved to 21.1 ± 1.3 synapses per IHC in CCG-4986-treated animals, which was similar to the num-

At ambient sound levels (non-exposed controls), the average SGN density in the middle turn of the cochlea was 2530 ± 72 cells/mm2 (Figure 6A,D). Noise exposure induced a significant (*p* < 0.0001) loss of SGNs in both vehicle- and drug-treated animals (Figure 6B– D). CCG-4986 treatment did not reduce the loss of SGN, and the average cell densities for CCG-4986 treated (2209 ± 97 cells/mm2) and vehicle controls (2183 ± 161 cells/mm2) were

**Figure 6.** Loss of spiral ganglion neurons (SGN) after noise exposure. (**A**–**C**) Representative images of Rosenthal's canal in the middle turn of the cochlea. (**A**) Control non-exposed animals, (**B**) Noise-exposed vehicle-treated, (**C**) Noise exposed CCG-4986 treated. SGN were immunostained with the neurofilament antibody (cytoplasm, green) and Hoescht (nucleus, blue). White arrows point at spaces with missing SGN. Abbreviations: SC, satellite cell; SGN, Spiral ganglion neuron. (**D**) Average SGN densities in the middle turn of the cochlea in noise-exposed vs. non-exposed **Figure 6.** Loss of spiral ganglion neurons (SGN) after noise exposure. (**A**–**C**) Representative images of Rosenthal's canal in the middle turn of the cochlea. (**A**) Control non-exposed animals, (**B**) Noise-exposed vehicle-treated, (**C**) Noise exposed CCG-4986 treated. SGN were immunostained with the neurofilament antibody (cytoplasm, green) and Hoescht (nucleus, blue). White arrows point at spaces with missing SGN. Abbreviations: SC, satellite cell; SGN, Spiral ganglion neuron. (**D**) Average SGN densities in the middle turn of the cochlea in noise-exposed vs. non-exposed animals. Data presented as mean ± SEM. No noise control, *n* = 10; Vehicle control, *n* = 16; CCG-4986, *n* = 17. \*\*\* *p* < 0.001; One-way ANOVA followed by Holm–Sidak post hoc test.

### 4986, *n* = 17. \*\*\* *p* < 0.001; One-way ANOVA followed by Holm–Sidak post hoc test. *2.7. Auditory Brainstem Responses and the Effect of Treatment 24 Hours after Noise Exposure*

animals. Data presented as mean ± SEM. No noise control, *n* = 10; Vehicle control, *n* = 16; CCG-

We have also investigated the effect of earlier CCG-4986 treatment (24 h after noise exposure) on ABR thresholds and suprathreshold responses. Like the previous study, exposure to octave band noise (8–16 kHz) for 2 h resulted in 40–50 dB PTS in vehicle-treated (control) animals at frequencies above 4 kHz (Figure 7A,C). However, the administration of a small molecule RGS4 inhibitor, CCG-4986 (100 µM), 24 h after noise exposure reduced PTS by up to 32 dB (Figure 7B,C). The greatest effect was observed at mid-frequencies (12–20 kHz), representing the region of the cochlea most damaged by noise exposure. Average threshold shift was reduced by 32 dB at 12 kHz (*p* = 0.0025), 23 dB at 16 kHz (*p* = 0.01), and 28 dB at 20 kHz (*p* = 0.001). A significant improvement (12–18 dB; *p* < 0.05) was also observed at other test frequencies (Figure 7C).

**Figure 7.** Baseline and final ABR thresholds for 4–28 kHz tone pips in animals exposed to octave band noise (8–16 kHz, 110 dB SPL) for 2 h and non-exposed animals. (**A**) Noise-exposed vehicletreated animals, (**B**) Animals treated with CCG-4986 (100 μM) 24 h post-exposure. (**C**) Comparison of permanent threshold shifts 15 days after exposure in drug- and vehicle-treated animals. Average baseline and final ABR wave I amplitudes (**D**) and latencies (**E**) at suprathreshold intensities in noise-exposed animals treated with CCG-4986 (100 μM) or drug vehicle solution 24 h after noise exposure. Data presented as mean ± SEM. Vehicle control, *n* = 10; CCG-4986, *n* = 9. \* *p* < 0.05 \*\* *p* < **Figure 7.** Baseline and final ABR thresholds for 4–28 kHz tone pips in animals exposed to octave band noise (8–16 kHz, 110 dB SPL) for 2 h and non-exposed animals. (**A**) Noise-exposed vehicle-treated animals, (**B**) Animals treated with CCG-4986 (100 µM) 24 h post-exposure. (**C**) Comparison of permanent threshold shifts 15 days after exposure in drugand vehicle-treated animals. Average baseline and final ABR wave I amplitudes (**D**) and latencies (**E**) at suprathreshold intensities in noise-exposed animals treated with CCG-4986 (100 µM) or drug vehicle solution 24 h after noise exposure. Data presented as mean ± SEM. Vehicle control, *n* = 10; CCG-4986, *n* = 9. \* *p* < 0.05 \*\* *p* < 0.01; \*\*\* *p* < 0.001. Two-way ANOVA followed by Holm–Sidak post-hoc test (ABR thresholds), Multivariate ANOVA followed by planned contrast comparisons (ABR suprathreshold responses).

We have also observed improved suprathreshold responses (ABR Wave 1 amplitudes) after treatment with CCG-4986 24 h post-exposure (Figure 7D,E). The amplitudes and latencies of ABR Wave I were analysed at suprathreshold levels (80–90 dB SPL) in different frequency regions of the cochlea (4 kHz, 16 kHz, and 28 kHz) to assess auditory nerve function. Noise exposure reduced ABR Wave I amplitudes by 60–70% and increased latencies by up to 25% (Figure 7D,E).

CCG-4986 treatment slightly improved Wave I amplitudes compared to vehicle-treated rats in all three test frequencies (90 dB; *p* < 0.05; Figure 7D). In addition, Wave I latencies were reduced after treatment with CCG-4986 (Figure 7E), suggesting partial recovery of neural function.

### **3. Discussion**

Our study demonstrates the cochlear rescue effect of CCG-4986 treatment in rats up to 48 h after traumatic noise exposure. This novel treatment is based on enhanced endogenous adenosine A1R activity in the cochlea. CCG-4986 is a small molecule RGS4 inhibitor which disrupts the signalling complex (RGS4/Neurabin) that regulates A1R activation [18]. We have shown that immunolocalisation of this molecular complex in sensory hair cells and spiral ganglion neurons corresponds to A1R distribution in the rat cochlea [19,20]. Local intratympanic administration of CCG-4986 48 h after acoustic overexposure mitigated noise-induced threshold shifts by 10–19 dB, which is considered clinically significant. This otoprotective effect was greater when the CCG-4986 treatment was delivered earlier (24 h after noise exposure), effectively reducing moderate-severe hearing loss to mild hearing loss. CCG-4986 administration also improved suprathreshold responses in noise-exposed animals (increased amplitudes and reduced latencies of ABR wave I), suggesting partial recovery of auditory nerve function. The treatment enhanced the survival of sensory hair cells in the noise-exposed cochlea and mitigated noise-induced loss of afferent synapses in the frequency-specific regions. Noise-induced loss of spiral ganglion neurons was, however, irreversible. The present study thus demonstrates that the RGS4 inhibition is the promising strategy for the treatment of noise-induced cochlear injury and introduces a novel paradigm for the treatment of NIHL and other forms of SNHL based on regulation of GPCR.

### *3.1. Drug Delivery to the Inner Ear*

The intratympanic method of drug delivery to the round window membrane has two principle advantages. Firstly, it precludes off-target effects of A1R activation in cardiovascular and other tissues, which is an important caveat for systemic administration. Secondly, poloxamer-407 is liquid at low temperatures, but becomes a gel at body temperature which allows slow drug release to the cochlear perilymph [21]. The small size of CCG-4986 (375 g/mol) and its apparent otoprotective effect suggest that this drug readily crosses into cochlear perilymph through the round window membrane. However, further studies are required to establish CCG-4986 concentration in cochlear fluid compartments after intratympanic injection and its pharmacokinetic properties.

### *3.2. ABR Threshold Shifts and Suprathreshold Responses are Mitigated by CCG-4986*

CCG-4986 treatment 48 h post-exposure produced a cochlear rescue effect by reducing PTS at all test frequencies above 4 kHz. The PTS reduction of >10 dB is considered clinically significant [22]. As expected, the earlier treatment 24 h post-exposure enhanced the rescue effect of CCG-4986 (up to 32 dB at 12 kHz). The window of opportunity to treat NIHL was thus reminiscent of our previous study using an A1R agonist adenosine amine congener [23].

Whilst auditory thresholds are considered a good metric of hair cell function, they are poor indicator of neuronal damage in the cochlea [24]. In this study, the changes in wave I amplitude and latency were used as indicators of afferent neural fibre (ANF) integrity. Noise overstimulation leads to a decrease in suprathreshold Wave 1 amplitudes [25] and

prolonged latencies [26] due to reductions in synchronous firing, lower discharge rates, and decreased recruitment of the high threshold ANF [27]. Recent studies postulate that the synaptopathy and the loss of ANF is a primary and mostly irreversible event in noiseinduced hearing loss [28]. This loss of cochlear afferents is functionally measured by reduced suprathreshold responses, due to preferential vulnerability of high threshold, low spontaneous rate ANF [29]. The diffuse afferent denervation that cannot be detected by measuring auditory thresholds is thought to contribute to poor performance in complex auditory tasks such as speech discrimination in a noisy environment, and thus has been termed "hidden hearing loss" [30]. In this study, treatment with CCG-4986 24 and 48 h after noise exposure reduced the ABR wave I latencies at all test frequencies, but only a minor improvement of wave I amplitudes was observed, suggesting partial recovery of neural injury.

### *3.3. CCG-4986 Improves the Survival of Sensory Hair Cells*

Auditory threshold shifts are largely determined by the integrity of sensory hair cells; hence, the quantitative histological analysis of hair cell survival was carried out in the apical, middle, and basal turns of the noise-exposed cochleae. The hair cell population in the apical segment was virtually unaffected by noise exposure, but a significant loss of OHC was observed in the middle and basal segments of the cochlea, the latter being most severely affected. There was very little IHC loss in the apical and middle turns, but almost one third of IHC was missing in the basal turn of the cochlea.

The increased vulnerability of OHC to noise, particularly in the basal turn, have been well documented in the past. "Inappropriate" loss of high frequency hair cells has also been observed in previous studies where the noise exposures targeted lower frequencies [31]. Basal OHC are particularly vulnerable to acoustic insult, likely due to their lower antioxidant buffering capacity leading to increased susceptibility to oxidative stress [32].

CCG-4986 treatment conferred a significant protection from noise-induced OHC loss in the middle segment and both IHC and OHC loss in the basal segment of the cochlea. Hair cell death is primarily mediated by oxidative stress and calcium overload leading to caspase-dependent cell death pathways [33]. We postulate that the RGS4 inhibition by CCG-4986 enhanced endogenous adenosine A1R signalling, which in turn improved antioxidant defences and restored calcium homeostasis [3,18].

### *3.4. CCG-4986 Partly Restores Afferent Synapses but does not Prevent Neuronal Loss*

In the absence of noise exposure, the vast majority of IHC synapses contain a presynaptic ribbon paired with a post-synaptic terminal from a single ANF, and only around 1% of synapses appear as orphaned pre-synaptic ribbons or post-synaptic terminals [25]. The survival of IHC-auditory nerve synapses is considered a sensitive metric for quantitative analysis of afferent innervation in the cochlea [25].

Noise exposure affected only synaptic ribbons in the mid and high frequency regions, whilst IHC synapses in the low frequency region were mostly intact. Treatment with CCG-4986 yielded a robust neuroprotective effect in the middle turn, to such extent that the average number of functional (paired) synapses in drug-treated animals was not significantly different from control animals exposed to ambient noise.

IHC synaptic loss is an acute event that is usually complete within 24 h after noise exposure. It is generally considered that the loss of afferent synapses is irreversible [25,34,35] but, more recently, post-exposure regeneration of afferent synapses has also been reported after intratympanic administration of neurotrophin-3 [36]. Given that CCG-4986 was administered 48 h after acoustic trauma, it is unclear how it prevents synaptic loss or regenerates afferent synapses. Further studies are required to investigate the timeline of IHC synaptopathic injury in the rat cochlea and establish the underlying mechanism of the rescue effect by CCG-4986.

CCG-4986 treatment 48 h post-insult, however, did not protect against SGN loss. SGN loss usually progresses at a much slower rate than the loss of afferent synapses after

acoustic overexposure [25]. Since SGN loss was measured only at one time point (16 days after insult), further studies investigating SGN survival months after acoustic insult are required to fully assess the neuroprotective effect of CCG-4986.

### *3.5. Putative Mechanisms of Otoprotection by CCG-4986*

CCG-4986 is a small molecule inhibitor of RGS4 which enhances adenosine A1R signalling by disrupting the molecular complex (Neurabin/RGS4) that terminates A1R signalling [18]. RGS4 is a GTP-activating protein (GAP) that selectively terminates A1R signalling at the level of G protein activation, by accelerating the intrinsic GTPase selfmediated hydrolysis and thus reverting Gα and Gβγ subunits to an inactive state [37]. This effectively terminates the intracellular response following adenosine A1R activation.

CCG-4986 blocks RGS4 activity by a dual mode of inhibition, through covalent modification of two different surface-exposed cysteine residues. Firstly, CCG-4986 modifies the Cys132 residue near the binding site of Gα protein and competitively prevents Gα protein binding to RGS4, which moderately reduces the binding affinity of Gα protein to RGS4 [37,38]. The second, more dominant mechanism, involves modification of the Cys148 residue located in an allosteric site that causes a conformational change in RGS4 that prevents Gα from interacting with RGS4, subsequently inhibiting GAP activity and thus extending adenosine A1R signalling [37].

Chen et al. [18] demonstrated that the scaffolding protein neurabin is required to facilitate the interaction between RGS4 and A1R in the brain. In the absence of A1R stimulation, RGS4 is localised to the cytosol. After A1R activation, RGS4 is recruited by neurabin to the cell surface, forming the A1R/neurabin/RGS4 complex that specifically regulates A1R signalling.

Our study shows that CCG-4986 can partially rescue the cochlea from noise-induced injury, most likely by enhancing A1R signalling. Noise exposure can induce the release of adenosine into the cochlear fluids and lead to up-regulation of A1R expression in cochlear tissues [3,39]. Adenosine A1R signalling reduces oxidative stress and lipid peroxidation, likely by boosting endogenous antioxidant defences, including superoxide dismutase and glutathione peroxidase activity [3,40]. A<sup>1</sup> receptors in the central nervous system are known to exhibit an inhibitory tone by preventing neuronal excitability and synaptic transmission [41]. A1R activation thus has a potential to directly counteract the main mechanisms of noise-induced cochlear injury, including oxidative stress, calcium overload, and glutamate excitotoxicity [23,33].

There is a certain advantage of CCG-4986 treatment over adenosine A1R agonists such as ADAC and R-PIA, which activate A1R with greater selectivity than adenosine. Our previous study [23] demonstrated a biphasic dose-response relationship effect of ADAC after systemic administration, otoprotective at doses 100–200 µg/kg and less effective at higher doses. This "effect inversion" might be due to overstimulation of A1R causing their desensitisation and internalisation [42]. In contrast, inhibition of the neurabin/RGS4 complex bolsters the A1R signalling without causing a change in A1R number or affinity [18].

### **4. Materials and Methods**

### *4.1. Animals*

For this study, male Wistar rats (6–8 weeks) were sourced from the animal facility at the University of Auckland. Animals were housed in standard cages with ad libitum access to food and water, under controlled conditions (constant humidity and temperature, 12-h light/dark cycle). A minimum of two animals were housed together, up to a maximum of four per cage. Animal welfare was continuously assessed during the study to ensure that animal health was maintained at the highest standard. Noise-exposed animals were randomly assigned into drug treatment or drug vehicle-treated (control) group. All experimental procedures were carried out with the approval of the University of Auckland Animal Ethics Committee (approval # 1631, 8 September 2015), in agreement with the Animal Welfare Act (1999).

### *4.2. Auditory Brainstem Responses*

Auditory brainstem responses (ABR) were used to determine auditory thresholds in rats prior to noise exposure (baseline) and 14 days after intratympanic drug or vehicle injection (final). The ABR is an auditory evoked potential that represents the synchronised summation of neuronal activity in response to auditory clicks and tone pip stimuli. ABR recordings were made in a custom-built sound isolating chamber (Shelberg Acoustics, Sydney, Australia), equipped with internal ventilation and a light source. Animals were anesthetised with a mixture of Ketamine (25 mg/kg) and Domitor (0.5 mg/kg) given intraperitoneally. The tympanic membrane was checked for signs of infection, physical trauma, or scarring before ABR recording. Only the left ear was used for assessment of ABR thresholds in each animal. Animals were placed on a thermostatically controlled electric blanket during recordings to maintain body temperature at 37 ◦C. The Tucker-Davis Technologies (TDT) System 3 and BioSig digital signal processing software (Alachua, FL, USA) were used to generate the auditory stimuli. A multi-field magnetic speaker (MF1, TDT) with 10 cm plastic tubing was used to deliver auditory stimuli into the ear. Three subdermal needle electrodes connected to a Medusa RA4LI headstage amplifier (×20 gain) were used to measure ABR responses. The active electrode was placed at the vertex of the scalp, the reference electrode at the mastoid region of the ear of interest, and the ground electrode at the mastoid region of the contralateral ear. Tone pips (5 ms duration, 2 ms rise/fall, presented at a 21/s rate) were used to elicit ABR responses at intensities between 90 dB SPL to 5 dB SPL, presented in decremental 5 dB steps. Tone pip responses were acquired at an alternating polarity sampling rate of 512 and averaged for each sound intensity. A bandpass filter (300–3000 Hz, 50 Hz notch) was applied to all responses. The ABR wave I was used to determine auditory thresholds defined as the lowest sound intensity level capable of eliciting a reproducible waveform. The cut off amplitude was set at ≥120 nV, as consistently reproducible waveforms were obtained at and above this amplitude. ABR recordings were repeated at sound intensities 10 dB above and 5 dB below the threshold in 5 dB decrements to confirm threshold intensity. In cases where the threshold ceiling was exceeded (above 90 dB SPL), these thresholds were arbitrarily assigned a value of 95 dB SPL. ABR assessments were carried out one day prior to noise exposure (baseline) and 14 days after intratympanic injection (final).

The amplitudes and latencies of wave I were assessed at suprathreshold intensities for selected frequencies (4, 16, and 28 kHz) to investigate the effect of CCG-4986 treatment on noise-induced neuronal injury. Animals with final auditory thresholds of 80 dB SPL or lower were included for input-output functional analysis. The amplitude of wave I (peak to trough) was measured at 90, 85, and 80 dB SPL intensities, and latency was measured as the time taken to reach the peak (including 0.3 ms signal transduction time from the speaker to the ear).

### *4.3. Noise Exposure*

Twenty-four hours after baseline ABR measurements, animals were exposed to an octave band noise (8–16 kHz) for 2 h at 110 dB SPL. Acoustic overstimulation was carried out in a custom-built sound-attenuating chamber (Shelburg Acoustics, Sydney, Australia), equipped with internal ventilation, light source, and speakers suspended from the ceiling. Frequency and intensity of sound were adjusted by external controls. The speakers were calibrated using a sound level meter (Precision Sound level Meter Type 2235, Brüel & Kjær; Nærum, Denmark) prior to each noise exposure session, with the average sound pressure intensity measuring 110 ± 1.5 dB SPL across the cage floor. Animals exposed to noise were placed inside a conventional rat cage, positioned with a 30 cm distance underneath the speakers, and allowed to acclimatise for 5 min. Sound intensity was gradually increased over a period of 5 min. Control animals were placed in the sound isolating chamber for two hours to control for relocation stress. Afterwards, animals were returned to the animal housing facility and kept at ambient noise levels (55–65 dB SPL) for the remainder of the experimental timeline.

### *4.4. Intratympanic Injections*

RGS4 inhibitor CCG-4986 or drug vehicle solution (control) was delivered by intratympanic injection into the middle ear cavity 24 or 48 h after noise exposure. Drug solution was made by dissolving CCG-4986 (ChemBridge™; San Diego, CA, USA) in 1% DMSO and 0.9% saline with the final 100 µM working dilution for intratympanic injection. Control animals were treated with the drug vehicle solution (1% DMSO in 0.9% saline). Drug and vehicle solutions were mixed with 17% *w/w* poloxamer-407 (Sigma-Aldrich) and placed on ice until fully dissolved. Solutions were then aliquoted and stored at −20 ◦C for later use. Poloxamer-407 allows for slow drug delivery to the cochlea as the solution becomes a gel at body temperature [21].

Prior to drug treatment, animals were anaesthetised with a mixture of Ketamine (25 mg/kg) and Domitor (0.5 mg/kg) injected intraperitoneally and administered one dose of Temgesic (Buprenorphine; 0.05 mg/kg, subcutaneously) for analgesia. Intratympanic injections were carried out using a Hamilton syringe mounted on a micromanipulator arm. The needle was inserted through the posterior-superior quadrant of the tympanic membrane to deliver injection solution into the vicinity of the round window membrane of the cochlea. A total solution volume of 27 µL was slowly injected into the middle ear. The animal was then returned to a recovery cage and left on its side for 30 min to allow the solution to settle onto the RWM in a gel form. Then the procedure was repeated for the contralateral ear. Animals were then given a subcutaneous dose of Antisedan (1 mg/kg) to reverse the effects of ketamine/domitor anaesthesia.

### *4.5. Cochlear Tissue Preparation for Histology and Immunohistochemistry*

After the final ABR assessment, animals were euthanized by an anaesthetic overdose (pentobarbitone, 90–100 mg/kg intraperitoneally). The animals were perfused with the flush solution (0.9% NaCl containing 10% NaNO2) and then tissue fixative (4% paraformaldehyde (PFA) in 0.1 M phosphate buffer (PB, pH 7.4) overnight at 4 ◦C. Cochleae were then decalcified using 5% EDTA in 0.1 M PB for 9 days at 4 ◦C, cryoprotected overnight in 30% sucrose in 0.1 M PB and embedded in optimal cutting temperature compound (OCT). Cochleae were then snap frozen using N-pentane and stored at −80 ◦C for further processing.

### *4.6. Hair Cell and Ribbon Synapse Counting*

The extracted cochleae were decapsulated and micro-dissected into the apical, middle, and basal segments (turns), after removal of the lateral wall, Reissner's membrane and tectorial membrane. Cochlear turns were transferred into a 48 well plate containing 0.1 M PBS (pH 7.4), permeabilised with 1% Triton X-100 and blocked with 10% Normal Goat Serum (NGS) for 2 h at room temperature (RT). Whole mount tissues were then incubated overnight at RT with the following primary antibodies: rabbit polyclonal anti-Myosin-VIIa (Proteus Biosciences, 1:500), mouse anti-C-terminal binding protein 2 (CtBP2; IgG1; BD Biosciences, 1:500) and mouse anti-Glutamate receptor 2 (GluA2; IgG2; Merck Millipore, 1:500) in antibody solution containing 0.1% Triton X-100 in 0.1 M PBS with 5% NGS. The following day, sections were washed three times for 60 min with 0.1 M PBS and then incubated at RT with the following secondary antibodies: goat anti-rabbit (Alexa 568, 1:500; Invitrogen), goat anti-mouse IgG1 (Alexa 647, 1:500; Invitrogen), and goat anti-mouse IgG2 (Alexa 488, 1:500; Invitrogen) in antibody solution containing 0.1% Triton X-100 and 5% NGS in 0.1 M PBS. Whole mounts were then washed for 60 min with 0.1 M PBS and then mounted on glass slides using Citifluor AFI mounting medium, cover slipped, sealed with nail polish, and stored in the dark at 4 ◦C. Inner hair cell-auditory nerve synapses were imaged and analysed at three frequency regions (4, 16, and 28 kHz) of the cochlea, based on the distance from the cochlear apex [43]. Immunostained synapses were imaged using confocal microscopy (Olympus FV1000 Live Cell System, Tokyo, Japan) with oil immersion 60× objective (1.35 NA) and 2.6× digital zoom. Images were captured as a z-stack, with the *z* dimension sampled in 0.2 µm steps, imaging frame in the *xy* dimension capturing 10 adjacent inner hair cells. Z-stacks were processed to remove nuclear CtBP2

staining and compiled into a colour composite stack: inner hair cells labelled with Myosin-VIIa (grey), post-synaptic terminals labelled with GluA2 (green), pre-synaptic ribbons labelled with CtBP2 (red). As a max z-projection could potentially obscure juxtaposed paired synapses in the z-dimension, synapses were counted frame by frame in the zdimension using the cell counter plugin in ImageJ, with markers displayed through the stack to avoid duplicate counts. A paired synapse was defined as a post-synaptic terminal (GluA2) immediately adjacent to a presynaptic ribbon (CtBP2). Any unpaired pre-synaptic ribbons or post-synaptic terminals were defined as orphan synapses. The total number of synapses were counted for 10 inner hair cells and averaged to get the mean number of synapses per IHC. The number of unpaired (orphan) synapses was expressed as a percentage of the total synapse count per IHC.

For hair cell counting, each cochlea was divided into three segments covering the entire length of the cochlea and representing different frequency regions. The apical segment occupied approximately 0–30% from the apex, middle segment 30–75% from the apex, and basal segment 75–100% from the apex. The organ of Corti was imaged with a Zeiss Axioplan 2 epifluorescence microscope (Carl Zeiss, Jena, Germany) with 20× objective (0.6 NA), and captured with a Photometrics Prime sCMOS monochrome camera. Inner and outer hair cells were counted in ImageJ using the CellCounter to mark intact and missing hair cells, with the number of missing cells expressed as a percentage of the total number of hair cells counted. For regions with complete OHC loss, but intact IHC, one IHC was approximated to three missing OHC (one in each row). For regions of absolute hair cell loss, an adjacent length of IHC was measured (3–4 cells) to calculate pixel width per IHC. The distance of the lesion was measured and the number of missing IHC and corresponding missing OHC was estimated.

### *4.7. Spiral Ganglion Neuron Counting*

Cochleae designated for spiral ganglion neuron counts were cryosectioned at 12 µm, and every second mid-modiolar section was placed into 0.1 M PBS (pH 7.4) in a 24 well plate. Cryosections were washed with 0.1 M PBS, permeabilised with 1% Triton X-100 in 0.1 M PBS and blocked with 10% normal donkey serum (NDS) for 1 h at RT. Cochlear sections were then incubated overnight at 4 ◦C with goat polyclonal Neurofilament (NF-L) antibody (2 µg/mL; Santa Cruz Biotechnology, Inc., Dallas, TX, USA) in antibody solution containing 0.1% Triton X-100 and 10% NDS in 0.1 M PBS. The next day sections were washed (3 × 10 min) with 0.1 M PBS and incubated with donkey anti-goat secondary antibody (Alexa Fluor 488, 1:600 dilution; Invitrogen) for 2 h at RT. Sections were then washed with 0.1 M PBS (10 min), incubated with Hoechst 33342 nuclear stain (1 µg/mL in 0.1 M PBS, pH 7.4; Thermo Fisher Scientific) for 15 min, then washed with 0.1 M PBS for 10 min. Sections were mounted on a glass slide in Citifluor AF1 mounting medium, covered with a coverslip and stored in the dark at 4 ◦C. Tissues were imaged with a Zeiss Axioplan 2 epifluorescence microscope (Carl Zeiss, Jena, Germany) with 40× objective. Images of the spiral ganglion neurons located in the middle turn were captured at two different wavelengths (UV and 488) using a Photometrics Prime sCMOS monochrome camera and merged to identify individual spiral ganglion neurons. The spiral ganglion area, represented by the bony edge of Rosenthal's canal, was selected using the "free-hand selection tool" in ImageJ. Individual neurons with unambiguous round nuclei in the middle turn of the cochlea were counted in each section and then averaged to determine SGN density for each animal as described previously [44].

### *4.8. Characterisation of Neurabin Expression in the Rat Cochlea*

To determine mRNA expression of the two neurabin isoforms (Neurabin I and II), four intact rat cochleae were extracted, decapsulated, and placed into separate Eppendorf tubes containing cold lysis buffer (100 mM TRIS-HCl pH 7.5, 500 mM LiCl, 10 mM EDTA pH 8.0, 1% LiDS, 5 mM dithiothreitol (DTT) and RNase inhibitors) pre-chilled to 4 ◦C. Cochleae were then homogenised using sterile Teflon mini-pestles. Polyadenylated RNA

(mRNA) was extracted using the magnetic Dynabeads® (Oligo(dT)25 (5 mg/mL)) mRNA DIRECT kit (Invitrogen). First-strand cDNA synthesis was carried out in a 20-µl reverse transcription (RT) reaction with random hexamers, dNTPs and Superscript III reverse transcriptase (Invitrogen). The complementary DNA was amplified by PCR with ratspecific primers for neurabin isoforms (Table 1) designed using OligoPerfect™ (Invitrogen). Negative control without reverse transcriptase was included in each PCR run. RT-PCR with a 40 cycle profile was performed as follows: 94 ◦C denaturation (1 min), 60 ◦C annealing (1.5 min), 72 ◦C extension (2 min) steps using PTC-100™Programmable Thermal Controller (MJ Research Inc., Waltham, MA, USA). PCR amplicons were separated by agarose gel electrophoresis, and visualised using SYBR safe DNA gel stain (Invitrogen). PCR products were purified by PureLink™ PCR Purification Kit (Invitrogen) and the identity of the amplicons confirmed by DNA sequencing (Centre for Genomics & Proteomics, School of Biological Sciences, the University of Auckland, Auckland, New Zealand).

### *4.9. RGS4 and Neurabin Immunohistochemistry*

The immunolocalisation of RGS4 and neurabin I isoform which forms molecular complexes with RGS4 was demonstrated in the rat cochleae using immunofluorescence. Briefly, rat cochlear tissues were cryosectioned at 30 µm using a CM3050 S cryostat (Leica, Germany) and mid-modiolar sections were placed in a 24 well plate containing 0.1 M PBS (pH 7.4) and washed twice for 20 min. Sections were then incubated in blocking and permeabilisation solution (10% NGS, 1% Triton X-100 in 0.1 M PBS) for 1 h at RT, followed by incubation with a mouse monoclonal RGS4 antibody (Santa Cruz Biotechnologies, sc-398658; 2 µg/mL) or rabbit polyclonal neurabin I antibody (Santa Cruz Biotechnologies; 2 µg/mL) overnight at 4 ◦C. Control sections were incubated with the normal mouse or rabbit IgG (2 µg/mL) instead of the primary antibody. The next day, sections were washed with 0.1 M PBS three times (60 min), followed by the incubation with the secondary antibody (Alexa Fluor 488 goat anti-mouse or anti-rabbit IgG, dilution 1:500) for 2 h at RT. After a washout with 0.1 M PBS (3 × 10 min), cryosections were mounted on glass slides using Citifluor AF1 Mounting Medium, covered with a coverslip and sealed with nail polish. Images of mid-modiolar cochlear cryosections were acquired using a confocal microscope (Olympus FluoView FV1000) and processed with FluoView software (version 2.0c, Olympus, Tokyo, Japan).

### *4.10. Data Analysis*

The researchers were blinded for all ABR assessments, tissue collections, and histological analyses. Animals were assigned a subject ID by an independent researcher and allocated into the treatment or vehicle control group using a randomly generated number list (https://www.randomizer.org). Aliquoted injection solutions were labelled only by subject ID. All data were tested for normality using the Shapiro–Wilk Test. Auditory thresholds, ribbon synapse and hair cell counts were analysed using a two-way ANOVA with a post-hoc Holm–Sidak test. Spiral ganglion counts were analysed with one-way ANOVA. Suprathreshold data were analysed using multi-level factorial ANOVA with planned orthogonal contrasts to determine differences between groups. Data are presented as mean ± SEM.

### **5. Conclusions**

Intratympanic administration of a small molecule RGS4 inhibitor presents a novel therapeutic strategy that precludes systemic side effects associated with systemic administration of adenosine A1R agonists, while demonstrating a strong rescue effect against noise-induced cochlear injury. Translational studies are required to determine clinical potential of this treatment for NIHL and other forms of SNHL. Future studies will need to determine pharmacokinetic and pharmacodynamic CCG-4986 profile in the cochlea after intratympanic administration, its effect on adenosine concentrations in cochlear perilymph, metabolism, and toxicity profile, before considering clinical trials.

**Author Contributions:** S.M.V. designed the experiments and coordinated the study; C.F., M.B., M.P., and R.T. performed the experiments; S.M.V. and C.F. prepared the manuscript; P.R.T. contributed to study design, data interpretation and critically revised the manuscript. All authors have read and agreed to the published version of the manuscript.

**Funding:** This study was supported by a research grant from the Auckland Medical Research Foundation (New Zealand).

**Institutional Review Board Statement:** All experimental procedures were carried out with the approval of the University of Auckland Animal Ethics Committee (approval # 1631, 8 September 2015), in agreement with the Animal Welfare Act (1999).

**Conflicts of Interest:** The authors declare no conflict of interest.

### **References**

