Experiments were designed using 20 adult male Sprague-Dawley rats (weighing 300 g) with normal eardrums. All procedures were approved by the Institutional Animal Care and Use Committee (CNU IACUC-H-2018-55). The animals were randomly divided into either a control group (n=10) or a study group (n=10) for GPIAS, NBPIAS, and ABR evaluations. The control group was used to evaluate salicylate-induced tinnitus, and the study group to assess the therapeutic effect of memantine on salicylate-induced tinnitus. The control group was injected intraperitoneally (IP) with 400 mg·kg⁻¹·day⁻¹ of sodium salicylate, whereas the study group was injected IP with sodium salicylate combined with 5 mg·kg⁻¹·day⁻¹ of memantine (Sigma, USA) for 7 consecutive days.

A startle response measurement system was used to obtain the GPIAS and NBPIAS values. As shown in Fig. 1A, the system was composed of a mesh cage with a vibration sensor, a noise box with an anechoic inner wall, an acoustic stimulator (PM-5004 amplifier; Marantz, Kawasaki, Japan, with a full-range loud speaker), a reference microphone (40PH; GRAS, Holte, Denmark), sensor signal acquisition hardware (PC and NI PCIe-6321; National Instruments, Austin, TX, USA), and LabVIEW-based custom graphical user interface (GUI) software. An accelerometer sensor module (LIS344ALH; STMicroelectronics, Geneva, Switzerland) for sensing the startling vibration in rats was attached at the bottom plate of the mesh cage. Our implemented GUI software controlled all of the processes for acquiring startle responses through acoustic stimulation, and performed the analysis for obtaining GPIAS and NBPIAS values. All the measurements for the startle responses were obtained using the integrated system in the soundproof box (Fig. 1A). For the GPIAS measurement, a session with each rat involved 15 startle responses evoked by gap-conditioned stimuli and another 15 startle responses evoked by a stimulus with no gap. In addition, two kinds of acoustic stimuli (a gap stimulus and a nogap stimulus) were presented to rats in randomly ordered pairs. The intervals of the acoustic stimuli were changed randomly from 17 s to 23 s. Before beginning the first session, the rat was placed in the cage for 2 min to acclimatize it to the measurement environment. The acoustic stimulation for the GPIAS consisted of narrowband background noise (1 kHz bandwidth, 16 kHz center frequency, and 60 dB sound pressure level [SPL]), with a short high-level stimulus for startling (broadband noise burst, 105 dB SPL, 50 ms length) and a prepulse gap inserted before the startle stimulus (50 ms length, onset on 100 ms prior to the onset of startle stimulus), as shown in Fig. 1B.

After GPIAS measurement, NBPIAS test was also measured. The main object of NBPIAS is to assess the animal hearing function. Because the hearing loss is the potential factor which affects the result of GPIAS measurement. If the hearing loss caused by salicylate treatment, it can make the background sound and the embedded gaps less audible. Moreover, hearing loss can also attenuate startle reflex magnitude, therefore GPIAS cannot be used for tinnitus assessment. NBPIAS recording was performed same as GPIAS measurement equipment.

For NBPIAS, a narrowband noise burst prepulse (50 ms length, 1 kHz bandwidth, 16 kHz center frequency, and 60 dB SPL) was used instead of a prepulse gap and without background noise (Fig. 1B). In addition, 15 startle responses to the prepulse condition stimulus and another 15 responses to the noprepulse stimulus were obtained with the same recording configuration as used for GPIAS. The outlying startle responses measured in each stimulation condition group were eliminated using Grubbs’ test. Root mean square (RMS) value is still the effective for noise or signal bursts. It is commonly used to compute a single value that represents the composite of different sine wave components. GPIAS and NBPIAS were defined as the inhibition percentages, calculated as (1–RMS of gap startle response/RMS of no gap startle response)×100 and (1–RMS of noise burst response/RMS of no noise burst response)×100, respectively. A high GPIAS or NBPIAS percentage value meant that a gap or a noise burst prepulse had inhibited the startle response. A rat with tinnitus at approximately 16 kHz typically shows a significantly lower GPIAS value because the animal cannot distinguish between the gap prepulse and the background sound owing to the tinnitus. A low NBPIAS value indicates that the rat could not hear the noise burst prepulse of the same SPL and frequency as the background noise used in the GPIAS measurement. Therefore, a rat with a low GPIAS and a normal NBPIAS value can be considered as being a model of induced tinnitus without hearing loss [17]. All the animals were tested for GPIAS and NBPIAS before sodium salicylate injection and 2 h afterwards. GPIAS measurements were performed continuously on a daily basis from 0 to day 8 and NBPIAS measurements were taken at day 0, 1, 3, 5, 7, and 8 (Fig. 2).

We evaluated hearing function using an ABR measuring system (Tucker-Davis Technologies, Miami, FL, USA). ABR thresholds were obtained in all animals before the first drug treatment and on day 8 following it. The procedure used for recording the ABR was the same as described in our previous publication [18]. In brief, after placement of the subdermal needle electrodes at the vertex and the reference electrode at the occiput, the hearing thresholds were assessed using tone bursts (2 ms rise/decay, 1 ms plateau), presented at a rate of 50/s in a decreasing intensity series, beginning with levels that elicited distinct evoked potentials. The stimulus was started at 90 dB SPL and repeated at intervals of 10 dB in a downward progression until no response was identifiable.

Rats were sacrificed 7 days after the injection of each drug or vehicle for histological examination of the brain tissue. The animals were anesthetized and perfused with 4% (w/v) paraformaldehyde (PFA) in PBS (pH 7.4), following which the brains were removed immediately and stored in 4% (w/v) PFA in PBS for 2 days at 4°C. The brains were suspended in 30% (w/v) sucrose for 4 days and then embedded in optimum cutting temperature compound (Miles Inc., Elkhart, IN, USA). Using a sliding microtome (SM2010R; Leica Microsystems, Wetzlar, Germany), the hemispheres were coronally sectioned at approximately 4.30~5.30 mm caudal to the bregma for the auditory cortex. Free-floating serial sections (30 μm thick) were collected into 10 wells filled with PBS.

The free-floating sections were incubated with 0.3% (v/v) hydrogen peroxide in distilled water for 20 min, and then blocked with 5% (v/v) normal goat serum in PBS with 0.3% (v/v) Triton X-100 (PBS-T) for 1 h at RT. The sections were incubated with rabbit anti-NR2B (1:500 dilution) primary antibody in antibody dilution buffer (Invitrogen, USA) for 1 day at 4°C. After washing with PBS-T, the sections were reacted with biotinylated goat anti-rabbit IgG (Vector ABC Elite Kit; Vector Laboratories, Burlingame, CA, USA) for 1 h at RT, washed with PBS-T, and incubated for 1 h at RT with an avidin–biotin peroxidase complex (Vector ABC Elite Kit). After washing with PBS-T, the peroxidase reaction was performed using the diaminobenzidine substrate (contained in the DAB kit; Vector Laboratories). Immunohistochemically stained specimens were examined using a microscope (Olympus, Tokyo, Japan) fitted with an eXcope X3 digital camera (DIXI Optics, Daejeon, South Korea).

Analysis of variance followed by Bonferroni post hoc test was performed. The data represent the mean±SEM.

During the baseline sessions, the mean GPIAS value was 46.3% in the control group and 47.2% in the study group. As shown in Fig. 5A, the mean GPIAS value was reduced in the control group, indicating that tinnitus had been induced in the rats. As compared with that in the group treated with salicylate alone, the GPIAS value attenuated to a significantly lesser extent in the group treated with a combination of salicylate and memantine (p=0.01 at day 1, p=0.01 at day 3, p=0.003 at day 5, and p=0.03 at day 7). The mean NBPIAS values in both groups were not significantly different throughout the entire testing period (Fig. 5B). This was supported by the experimental results using GPIAS and audiological measurements. The GPIAS recovery rate in the study group was similar to that reported by Ralli et al [19].

To determine whether salicylate could induce a temporary auditory threshold shift, ABR threshold shifts were observed before the drug treatment and on days 7 and 8 (1 day after the end of treatment). The salicylate group showed an elevated threshold compared with that before treatment (the asterisk indicates statistical significance, p=0.26), whereas the salicylate plus memantine group showed an attenuation of the threshold shift on day 7. The mean ABR threshold in both groups was not significant before and 1 day after the end of treatment (p=0.17) (Fig. 6A and 6B). These results indicated that memantine attenuated the salicylate-induced threshold shift at 8 and 16 kHz.

In the salicylate group, NR2B protein expression was markedly increased in the auditory cortex, whereas it was reduced in the memantine-treated group (Fig. 7).