Eighteen 8-week-old Sprague–Dawley rats were separated into four groups as follows: sham ovariectomy (OP; n=5), OP only (n=3), OP with low-dose estrogen replacement therapy (ET1; n=5), and OP with high-dose estrogen replacement therapy (ET2; n=5). All rats were anesthetized before auditory brainstem response (ABR) measurements were recorded and before surgery. After the baseline ABR measurements were recorded, surgery was performed on all rats. All animals were exposed to noise at 2 weeks after surgery and two different doses of estrogen therapy were administered to the treatment groups. To maintain estrogen levels during the noise stress and recovery periods, estrogen injections were given 1 day before and 3 days after noise exposure. ABR measurements were recorded again 1 day and 2 weeks after noise exposure (Fig. 1).

Rats were anesthetized using 400 µL/kg intramuscular injections of zolazepam (Zoletil; Virbac, Carros, France) and xylazine (Rompun; Bayer AG, Leverkusen, Germany). The surgical procedures were performed in accordance with methods described previously [10,11]. Briefly, animals were placed in a prone position to facilitate access to both dorsolateral sides. Incisions (approximately 1.5–2 cm) were made halfway between the middle of the back and the base of the tail and the ovarian fat pads were gently exposed using artery forceps. The ovaries were identified and gently removed (Supplementary Fig. 1). After both ovaries had been excised, the fat pads were repositioned inside the abdomen and muscles and skin were stitched up. After surgery, the animals were transferred to a heated pad for recovery. Sham OP group animals underwent a similar procedure but the ovaries were not removed.

The rats were separated in different chambers to ensure they were equally exposed to noise. The chambers were placed in a sound reverberating acrylic noise box and the noise was introduced from a CP800Ti speaker in the ceiling (Beyma, Valencia, Spain). Narrow band noise (centered at 16 kHz with 1 kHz of band width) was generated using a type 1024 sine random generator (Bruel and Kjaer, Copenhagen, Denmark) and amplifier (Inter-M, Seoul, Korea). The rats were exposed to 120 dB sound pressure level (SPL) for 6 hours continuously. The noise level was monitored using a type 2250 frequency-specific sound level meter (Bruel and Kjaer).

β-estradiol powder (estradiol, E2758; Sigma-Aldrich, St. Louis, MO, USA) was dissolved in ethanol. The estradiol solution was diluted to the two working concentrations using distilled water. A total volume of 1 mL of estradiol was injected into the peritonea of rats in each of the estrogen replacement therapy groups (10 µg/kg for ET1 and 100 µg/kg for ET2) 1 day before and 3 days after noise exposure. For OP-only and sham OP groups, only distilled water with same concentration of ethanol was injected.

ABR thresholds were measured at baseline (before noise exposure) and 1 day and 2 weeks after noise exposure. Each animal was anesthetized and three subdermal needle electrodes were positioned as follows: one at the vertex (active) and one behind each pinna (reference/ground). Tone-burst stimuli were delivered through electrostatic speakers using an EC-1 insert tube (Tucker Davis Technologies, Alachua, FL, USA) and the response was measured at 8, 16, and 32 kHz in 5 dB steps from 90 to 10 dB SPL using a System III evoked response signal-processing system (Tucker Davis Technologies). A total of 512 responses were amplified, filtered, and averaged to obtain the hearing thresholds. The stimulus intensity above the flat waveform was considered the threshold.

After the final ABR measurements were recorded, the animals were sacrificed and the cochleae were carefully harvested. The cochleae were fixed in 10% neutral buffered formalin for 2 hours, decalcified in 0.1 M EDTA (ethylenediaminetetraacetic acid) for 1 week, and gently microdissected as whole mount preparations. The samples were immunostained with antibodies against myosin VIIa (rabbit anti-myosin VIIa, 1:200; MilliporeSigma, Burlington, MA, USA) to identify hair cells and neurofilament-heavy (1:1,000, MilliporeSigma) to identify the auditory nerves. The appropriate secondary antibodies were coupled with the corresponding primary antibodies (anti-rabbit Alexa 568, 1:1,000 and anti-chicken Alexa 488, 1:1,000). After staining, the samples were mounted on slides using Vectashield H-1000 crystal mounting solution (Vector Laboratories, Burlingame, CA, USA).

An FV-3000 confocal microscope (Olympus, Tokyo, Japan) with a 40× objective was used for imaging. The z-dimension was acquired at 1 µm for each image stack. Images of the basilar membranes were obtained for 8, 16, and 32 kHz [12]. Anti-myosin VIIa-positive inner (IHCs) and outer hair cells (OHCs) were counted within a square area 200 µm in length.

All data were analyzed using IBM SPSS ver. 19.0 (IBM Corp., Armonk, NY, USA) or GraphPad Prism software (GraphPad Software, San Diego, CA, USA). Kolmogorov-Smirnov tests were used to determine data equivalence. Kruskal-Wallis tests and multivariate analysis of variance (MANOVA) were used to determine differences in hearing thresholds, threshold shifts, and hair cell loss among the groups. Univariate ANOVA and Tukey’s test were used for post-hoc analyses.

In this study we used ABR thresholds to assess hearing in each animal. Significant threshold shifts in hearing function occurred after noise exposure, and were unaffected by the OP procedure. Different doses of estrogen in replacement therapy resulted in differences in hearing threshold recovery. The magnitude of the threshold shifts after noise exposure was significantly reduced at both estrogen concentrations. However, no dose-dependent effects of estrogen replacement therapy were observed. To investigate the histological basis of these observations, the number of IHCs and OHCs were counted in the regions responsible for different hearing frequencies. In contrast to the functional data, which revealed significant differences at all the frequencies tested, the histological results showed significant differences in the number of hair cells of both types only at the highest frequency. The estrogen replacement therapy groups had significantly more hair cells of both types than the OP group, and the OP group had fewer hair cells of both types than the sham OP control group. This differs from the functional data, which showed no differences between the OP and control groups. Possible reasons for differences between the functional and histological data are discussed below. Nevertheless, both the ABR and histological data suggest that estrogen replacement therapy, regardless of estrogen concentration, improved hearing and increased hair cell density compared to the OP-only animals, which reflected menopausal conditions.

The noise frequency used in this study was centered at 16 kHz. However, the disruption of OHCs at 32 kHz was more severe than at 16 kHz. This may have been due to differences in vulnerability between the apex and base of the cochlea. Different concentrations of glutathione generate differences in the susceptibility of hair and supporting cells to free radicals, resulting in different levels of damage in the apex and base when the cochlea is exposed to aminoglycosides or traumatic noise [13].

The relationship between hearing and estrogen levels has been studied extensively in humans and animal models. Previous studies in humans reported that ABR wave latencies were delayed and thresholds were higher in postmenopausal subjects and those with congenitally low levels of estrogen [14,15]. Previous studies in animals also showed that ABR latencies were delayed after OP [8]. Naturally estrogen-deficient mice had prolonged ABR latencies compared to those of control mice [10]. These studies suggest that hearing thresholds and ABR peak latencies may be affected by internal estrogen levels. The effect of OP on hearing can be altered by external stimuli. Willot et al. [11] reported that ovariectomized mice showed smaller threshold shifts than control mice after noise exposure. In this study, we found no differences between sham OP and OP-only rats in hearing threshold or threshold shifts after noise exposure.

A limitation of this study is that we used an acute animal model to represent menopause. Furthermore, other organs, such as adipose tissue, may be able to compensate for estrogen depletion [12]. We did not analyze ABR wave latency in this study because the noise level used was sufficient to cause permanent threshold shifts, and the thresholds in both groups approached the level of deafness. Nevertheless, this limitation could have been responsible for the nonsignificant hearing threshold results.

In this study, both the ET1 and ET2 groups showed better functional and histological responses to noise exposure than the sham OP and OP-only groups. Several previous studies have shown that estrogen is protective against noise exposure and glutamate toxicity [16,17]. Estrogen can increase the production of the powerful antioxidants superoxide dismutase 2 and hydrogen sulfide, conferring protection against oxidative stress and decreasing the production of reactive oxygen species [18,19]. An estrogenstimulated increase in the production of brain-derived neurotrophic factor may also confer protective effects [16]. Although we did not investigate the pathways through which estrogen exerted its effects in this study, mechanisms similar to those reported in previous studies may have been involved.

Our functional and histological analyses found no differences between the ET1 and ET2 groups. This might have been because the low dose of estrogen (10 µg/kg) used in this study may already be high enough to produce the maximum effect possible. Coleman et al. [8] compared the effects of various doses of estrogen therapy in an ovariectomized rat model and found no dose-dependent effects on ABR. Another study demonstrated that different doses of estrogen did not increase the therapeutic effect if the lowest dose was already sufficient [20].

We found two mismatches in the results of this study. First, hearing thresholds in all groups increased significantly, whereas improved histological results were observed at all the frequencies tested. Second, there was no significant hearing threshold difference between the sham OP and OP-only groups, whereas the histological results showed marked differences in hair cell survival at 32 kHz.

Loss of function in the remaining hair cells is a possible explanation for the first mismatch. The synapses between IHCs and nerve fibers are the most vulnerable part of the organ of Corti [21], and these synapses may be damaged even if all the hair cells remain intact. Furthermore, in cases of severe noise exposure, at least 2 months may be necessary for hair cell recovery [21]. Therefore, it is possible that further studies should record observations for more than 2 months to understand the mismatch between the histological and functional results described in this study.

Regarding the second mismatch, the level of noise may have been too high to investigate cochlear vulnerability due to OP. Because the thresholds in both the sham OP and OP-only groups approached the level of deafness, we surmise that the noise intensity might have offset any differences in hearing thresholds. Experimenting with lower levels of noise exposure may make it possible to identify differences in cochlear vulnerability between these groups.

Although the number of animals in each group was small, our histological results and images showed clear differences among the groups. Internal hormone levels can fluctuate after OP and estrogen replacement therapy and were not measured here. Nevertheless, the discrepancies between the functional and histological data may have occurred due to the factors described above. A larger sample size and comprehensive functional and histological analyses, such as peak 1 amplitude ABR, ribbon synapse analysis, and nerve structure studies are needed to confirm the findings presented in this study.