A total of eighty-five C57BL/6J mice were used in this study. Auditory brainstem response (ABR) experiments were conducted on a cohort of mice with an average age of postnatal day 37 (P37, range = 34–40; female n = 17, male n = 13), and a cohort with an average age of P60 (range = P57–P73; female n = 10, male n = 13). Throughout the remainder of the study, these mice will be referred to as the P37 and P60 cohorts, respectively. For in situ hybridization (female n = 4, male n = 4) and RT-qPCR (female n = 12, male n = 12), mice were aged P34–50. C57BL/6J mice in these age ranges are sexually mature (Bell, 2018). ABR data were collected on two sexually mature age cohorts to ensure that the sex difference in ABR wave I amplitude was maintained in the age ranges used for in situ hybridization and RT-qPCR. C57BL/6J mice were used in this study because they are the most commonly used strain for genetic modifications. We wanted to establish this sex difference in WT C57BL/6J mice for potential future experiments examining the mechanisms of this sex difference in genetically modified mice. The age range of the C57BL/6J mice used in this study is before the age where early-onset age-related hearing loss has been previously documented in this strain, which was P100 for high frequencies beyond the age range used in this study (Henry, 2002), and P150 for a broadband threshold shift (Henry, 2004). Mice were housed in a room with ambient noise levels averaging 15–20 dB SPL (unweighted mean SPL, 3 kHz–90 kHz, Sensory Sentinel, Turner Scientific), and a 12-h light/12-h dark daily photoperiod. Mice were fed ad libitum. All procedures were carried out in accordance with the National Institute of Health Guide for the Care and Use of Laboratory Animals and were approved by the University of Pittsburgh Institutional Animal Care and Use Committee.

Auditory brainstem responses (ABRs) were recorded and analyzed in a similar protocol to that used by Clarkson et al. (2016) and García-Hernández et al. (2017). However, the sound was presented in the open field in this study. Recordings were done in a Faraday cage in a sound attenuation chamber. During ABR experiments, mice were continuously anesthetized with 1.5% isoflurane in oxygen, and body temperature was kept constant via isothermal heat pads. The hydration and internal temperature of the mice were not measured; however isothermal heat pads were heated under similar conditions for each experiment. Thus, the external environment was controlled between mice. The speaker was calibrated in the chamber using a microphone (PCB electronics, model no. 377C01, Depew, NY) placed 10 cm from the multi-field magnetic speaker (MF1, Tucker-Davis Technologies (TDT), Alachua, FL), which was the same distance as the mouse’s external ear to the speaker during experiments. Broadband click and pure tone burst stimuli (4, 8, 12, 16, 24, and 32 kHz) were presented in the open field. For clicks, stimuli were presented for 0.1 ms starting at 90 dB and decreased to 10 dB in 5 dB steps, while pure tone stimuli were presented for 5 ms (2.5 ms linear rise/fall time with no plateau) at 90 dB and decreased to 20 dB in 10 dB steps. All stimuli were presented at 21 sweeps/s (i.e., stimulus period = 47.619 ms). Subdermal electrodes were placed on the vertex of the skull (recording electrode), ipsilateral (right) ear mastoid through the cheek (reference electrode), and contralateral (left) ear mastoid through the cheek (ground electrode). All ABRs were recorded in 10 ms increments. Incoming signals were digitized and recorded using a RA4LI headstage, RA4PA preamplifier/digitizer, and RZ6 signal processor (TDT). Data were collected in real-time using PC interface BioSigRZ software (TDT). All recordings were averaged over 512 sweeps and were band-pass filtered from 300 to 3,000 Hz. ABR click and pure tone threshold were operationally defined as the lowest stimulus level at which ABR waveforms could be visualized. Wave I latencies and amplitudes were calculated from click ABR recordings. Latency of wave I was defined as the time from the onset of the stimulus to the peak of wave I in ms, and wave I amplitude was defined as the peak of wave I to the subsequent trough in μV.

Mice were anesthetized via intraperitoneal injection with a mixture of ketamine (60 mg/kg) and xylazine (6.5 mg/kg) and transcardially perfused with 0.1 M phosphate buffer (PB) followed by 4% paraformaldehyde (PFA) in 0.1 M phosphate-buffered saline (PBS) in RNAse-free water. Cochleae were carefully removed and perilymphatically perfused with 4% PFA in 0.1 M PBS and then immersion fixed for 1–3 h on ice. Immediately following immersion-fixation, cochleae were moved to 10% EDTA in 0.1 M PBS and gently agitated overnight at 4°C for decalcification. The next morning, cochleae were washed in 0.1 M PBS, serially dehydrated in 10%, 20%, and 30% sucrose in 0.1 M PBS, and frozen on dry ice in molds with tissue freezing medium (Electron Microscopy Sciences, Hatfield, PA), and stored at −20°C for one month. The right cochlea of each mouse was sectioned at 12 μm thickness using a cryostat. Four sections were adhered per slide in series and stored at −80°C for 6 months prior to in situ hybridization.

Each in situ hybridization experiment used slides from at least one male and one female. Slides were chosen for each experiment in an attempt to control for equal numbers of SGNs per mouse at each cochlear region. In situ hybridization was carried out using probes, amplifiers, and fluorescent dyes from the RNAscope multiplex fluorescent assay developed by ACDBio (Newark, CA). This included probes for Gria2 (cat. no. 416091-C2), Gria 3 (cat. no. 426251-C3), and Gria4 (cat. no.422801-C1) as well as a kit necessary for amplification and labeling steps (RNAscope Multiplex Fluorescent Reagent Kit v2, cat. no. 323100). The probes do not differentiate between the flip and flop isoforms of each subunit. Probe specificity was validated in each experiment by including a slide that was incubated with a positive control probe mixture (3-Plex positive control probe, cat no. 320881), and a slide that was incubated with a negative control probe mixture (3-Plex negative control probe, cat no. 320871) in each experiment. Due to the limitations of the assay, only two RNA probes were targeted in each experiment. Therefore, in half of the experiments, Gria2 and Gria3 were targeted in tandem, while in the other half, Gria2 and Gria4 were targeted in tandem. Gria2 was targeted in every experiment both to serve as a control between experiments and to allow for calculating the index of Gria3 and Gria4 relative to Gria2 mRNA in individual SGNs (see Section “Statistical analysis” below). In all experiments, the Gria2 probe was labeled with Opal 690 dye, while Gria3 and Gria4 were labeled with Opal 570 dye (Akoya Biosciences, Marlborough. MA, cat. no. FP1497001KT and FP1488001KT, respectively). Following hybridization, amplification, and labeling, slides were mounted in ProLong Gold with DAPI (Invitrogen, Eugene, OR, cat. no. P36931), cover-slipped, and stored at 4°C away from light for up to one month prior to confocal imaging.

All slides were imaged with a Nikon A1 confocal inverted stage microscope (Nikon Instruments, Inc.) at the University of Pittsburgh Center for Biologic Imaging. DAPI, Opal 570, and Opal 690 were imaged simultaneously from lasers with excitation/emission spectra filters set at 405/425–475, 561/570–620, and 640/663–738 nm, respectively with a 40× 1.3 NA oil immersion lens and pinhole set to 1 airy unit. Images were collected in z-stacks of 0.25 μm increments using NIS Elements software (Nikon). Image stacks were combined into one z-projection containing all three fluorescence channels using Image-J¹. Opal 690 (Gria2) puncta were pseudo-colored to green, Opal 570 (Gria3 or Gria4) puncta were pseudo-colored to magenta, and DAPI remained blue (see Figure 1). ROIs (approximately 12 μm in diameter) for puncta counts were defined as separate type I-like SGNs, which were identified in each image by the presence of opal 690 (green) and opal 570 (magenta) puncta clustered around large DAPI-stained nuclei (see Figure 1). To ensure that clusters of transcript puncta were segregated to one SGN, we spot-checked multiple z-stacks for each cochlea to ensure that there was no overlap of SGNs in the z-plane. Amounts of Gria3 or Gria4 and Gria2 puncta were manually counted in individual SGNs semi-quantitatively: green and magenta puncta were counted separately using the Image-J point tool. On rare occasions, puncta were observed in large clusters where individual puncta could not be distinguished. Puncta with diameters exceeding 14 pixels were considered clustered, and these SGNs were excluded from the analysis, similar to the methods in Balaram et al. (2019). Puncta with diameters less than 2 pixels were considered noise and were not counted within SGNs. Gria2 and Gria3 counts are likely under-reported because these subunits tended to cluster the most. However, only up to four SGNs per image (out of over 30 in most images) were excluded from the analysis, and many of those images spot-checked (18 out of 30) had no clustering. See red arrowheads in Figures 1B, 3A for examples of SGNs that were excluded due to clustering. The intensity of puncta fluorescence was not considered in the analysis. On average, puncta from 30 SGNs were quantified per image. Throughout all experiments, four-six slides containing four cochlear sections each were probed per mouse. All fluorescence analysis was performed by observers blind to the sex of the animal.

Mice were anesthetized with isoflurane and euthanized via cervical dislocation and decapitation. Immediately following decapitation, the cranium was opened. The inner ears were removed, flash-frozen in liquid nitrogen, and stored at −80°C for up to one month until reverse transcription-polymerase chain reaction (RT-PCR). In preparation for RT-PCR, both inner ears from each mouse were homogenized by hand with mortar and pestle and RNA was extracted with Trizol (Ambion by life technology). The pellet was resuspended and the supernatant containing RNA from each inner ear (including auditory and vestibular tissue) was prepared for RT- PCR using the SuperScript Strand Synthesis System kit (Invitrogen, cat. no. 11904018). The resulting cDNA was stored at −20°C for 1 week or less prior to quantitative PCR (qPCR). qPCR was performed at the Genomics Research Core at the University of Pittsburgh using EvaGreen qPCR kit (MidSci, Valley Park, MO, cat. no. BEQPCR_R) and DNA primers for Gria2, Gria3, and Gria4 flip and flop, which were the same primers used successfully in a previous RT-qPCR experiment by Hagino et al. (2004); see Table 1 in this reference for sequence information. In a 25 μl PCR reaction mixture, 2 μl cDNA samples were amplified in a Chromo 4 detector (MJ Research, Waltham, MA). GAPDH and 18S rRNA were used as housekeeping genes. Each sample (consisting of cDNA product of both cochleae from each mouse) was run in triplicate, and average cycle thresholds (CTs) were used for quantification. Relative abundances of each splice isoform for males compared to females were calculated using the following equation: 2^−ΔΔCT, where ^ΔΔCT = (CT_Male − CT_GAPDH or CT_{18S rRNA}) − (CT_Female − CT_GAPDH or CT_{18S rRNA}). For a more in-depth explanation of this equation see Schmittgen and Livak (2008).

All statistical tests were calculated using GraphPad Prism or IBM SPSS software. ABR thresholds, wave I latency and wave I amplitude were compared between males and females using separate two-way mixed ANOVAs for P37 and P60 mice. For comparisons of click wave I latency and wave I amplitude, the within-subjects independent variable (IV) was sound level (90–50 dB in 5 dB increments). In contrast, for ABR thresholds the within-subjects IV was sound frequency (click, 4, 8, 12, 16, 24, and 32 kHz). For these two-way mixed ANOVAs, the between-subjects IV was sex. For latency and amplitude, linear regressions were calculated to compare the slope of ABR wave I latency or amplitude as a function of sound level between males and females of P37 and P60. For wave I amplitude slopes, Tukey pairwise comparisons were calculated between each group.

For RNAscope mRNA quantification data, three separate two-way ANOVAs were conducted to compare Gria2, 3, and 4 mRNA quantities between males and females (IV1, sex) in the basal, middle, and apical cochlear turns (IV2, region). To quantify possible differences in the expression of Gria2 vs. Gria3 and Gria2 vs. Gria4 in individual SGNs, we calculated a Gria3/2 and Gria4/2 index with the following equation, derived from Balaram et al. (2019): (Gria3 or Gria4 − Gria2)/(Gria3 or Gria4 + Gria2). A positive-value index means that a SGN has more Gria3 or Gria4 relative to Gria2; a value of zero means that Gria3 or Gria4 and Gria2 puncta levels in a SGN are equal; and a negative value means that a SGN has more Gria2 puncta relative to Gria3 or Gria4. The indices were then compared across cochlear regions and sex using separate two-way ANOVAs for each index. Due to the nature of our experimental design, we planned to investigate sex differences of Gria2, 3, and 4 mRNA counts and Gria3/2 and Gria4/2 indices in each separate cochlear region. Therefore, we conducted independent samples t-tests for comparisons of these measures between the sexes in each cochlear region a priori.

For RT-qPCR data, separate independent samples t-tests were conducted between males and females to compare flip and flop relative abundances for Gria2, 3, and 4.

We first tested for potential sex differences in ABR threshold and wave I latency and amplitude in our colony of C57BL/6J mice (Figure 2). These parameters are physiological measures of hair cell health (threshold), SGN speed of conduction (wave I latency), and SGN health and synchronous activity (wave I amplitude; see Liberman and Kujawa, 2017; Young et al., 2021). Our data showed that ABR clicks- and tone-thresholds varied by frequency (i.e., broadband click, 4, 8, 12, 16, 24, and 32 kHz pure tones), but not by sex for both P37 and P60 aged cohorts (two-way mixed ANOVA with a main effect of frequency on ABR threshold, no main effect of sex, nor an interaction of sex and frequency; Figures 2C,D; see Table 1 for ANOVA statistics). We next analyzed wave I latencies. Data showed that click wave I latency decreased with increasing sound levels (50–90 dB), but did not vary by sex (two-way mixed ANOVA with main effect for sound level, no main effect of sex, nor an interaction of sex and sound level; Figures 2E,F). There was no difference in slopes of click wave I latency as a function of stimulus level between the sexes of either age cohort (Figure 2J; Table 1). In contrast, we found that click wave I amplitude differed depending on sex and increased with sound levels (50–90 dB) for both age cohorts (two-way mixed ANOVA with main effect of sex, the main effect of sound level, and an interaction between sex and sound level; Table 1). In P37 mice, pairwise comparisons revealed that there was a sex difference at all tested sound levels with a visualized ABR (50–90 dB SPL; p < 0.05). In P60 mice, females had significantly larger wave I amplitude than males from 80 to 90 dB SPL (p < 0.05; Figures 2G,H). Similarly, analysis of the slope of ABR wave I amplitude as a function of stimulus level revealed that females had higher slopes than males at P37 and P60 (Figure 2I). Wave I amplitudes were also examined for all pure tone stimuli (4, 8, 12, 16, 24, and 32 kHz) for P60 mice. Though not significant, females had slightly higher ABR wave I amplitudes at all frequencies except for 4 kHz (data not shown). Thus, the significant difference for click wave I amplitudes is likely due to a cumulative effect from cochlear regions above 4 kHz. Together, our ABR analysis of sexually mature female and male mice showed no sex differences in click and pure tones thresholds and wave I latency. However, a sex difference in wave I amplitude suggests greater SGN synchrony in females at supra-threshold sound levels, and variability begins to increase at P60, particularly in males.

In both females and males, puncta of the RNAscope signal for Gria2, Gria3, and Gria4 were only observed within and along the Rosenthal canal of the cochlea and around large DAPI nuclei of SGNs (Figures 1, 3A, 4A, 5A). We compared total mRNA levels for Gria2, Gria3, and Gria4 between females and males. In addition, for determining whether mRNA levels for Gria2–4 followed a tonotopic organization and whether there were sex-specific differences in their distribution, we quantified the puncta for each of the genes in SGNs of the basal (high-frequency sounds), middle (middle frequency sounds), and apical (low-frequency sounds) cochlea turns.

In our RNAscope procedure, each SGN had mRNA expression for two genes, either Gria2 and Gria3 or Gria2 and Gria4. Our initial semiquantitative analysis of the puncta showed in both female and male mice that SGNs differed in the ratios of the Gria3/Gria2 and Gria4/Gria2 puncta (Figure 1). For example, some SGNs had more Gria3 than Gria2 puncta and vice versa; other SGNs had roughly equal amounts of Gria3 and Gria2 puncta. We used an index calculation to classify the putative existence of a heterogenous population of SGNs based on Gria2, 3, and 4 mRNA levels (Balaram et al., 2019). The Gria3/Gria2 and the Gria4/Gria2 index calculations revealed a gradient of SGNs based on these indices in males and females throughout the cochlea (Figure 6; see Section “Materials and Methods” for details). We found a population of SGNs with a positive index value (> 0), meaning that these neurons had higher mRNA levels for Gria3 or Gria4 than Gria2. Data also showed another population of SGNs with an index value of 0, meaning that they had equal mRNA levels for each of the paired genes analyzed. Finally, data showed a third population of SGNs with a negative index value (< 0), which indicates that Gria2 mRNA is higher than Gria3 or Gria4. We found that the percentage of each population of SGN differed along the cochlea turns and by sex.

AMPAR subunit composition determines channel permeability to calcium and sodium, and the gating kinetics of the channel is determined by the flip-and-flop splice variants (Mosbacher et al., 1994). Thus, we next determined Gria2, 3, and 4 flip and flop splice variant mRNA levels in the female and male cochlea with RT-qPCR. Our data showed that the relative abundances of Gria2 flip and flop mRNA were not different between females and males (t = 1.198, p = 0.26 and t = 1.631, p = 0.13, respectively). However, males had roughly 2.5-fold more Gria3 flip mRNA than females, whereas females had roughly 2-fold more Gria3 flop mRNA than males (t = 3.32, p = 0.007, and t = 3.986, p = 0.002, respectively). Additionally, females had roughly 2-fold more Gria4 flip and flop compared to males (t = 4.27, p = 0.002 and t = 5.56, p = 0.0005, respectively; Figure 7). These data suggest that females have more post-transcriptional processing of Gria3 into the faster flop isoform and Gria4 into the slower flip and faster flop isoform than males throughout the entire cochlea. In contrast, males had more of the slower gating Gria3 flip isoform. We note that these results were obtained from the entire inner ear that included SGN and vestibular neurons.

Relative levels of each AMPAR subunit (GluA2, 3, and 4) are significant contributors to synaptic speed in SGNs because they affect the channel kinetics of AMPARs differently. AMPARs that lack GluA2 are calcium permeable and are known to have extremely fast channel kinetics (Cull-Candy et al., 2006; Traynelis et al., 2010). For example, cochlear nucleus principal neurons that receive excitatory synaptic input from auditory nerve fibers where fast synaptic responses are essential for reliably encoding auditory information, AMPARs are mainly composed of GluA3 and GluA4 subunits (Rubio and Wenthold, 1997; Wang et al., 1998; Gardner et al., 1999, 2001; Rubio et al., 2017). While many of these studies focused on mature AMPAR subunit proteins at the synapse, importantly others have shown that subunit mRNA levels correlate with channel kinetics. Indeed, studies in the CNS have shown that faster AMPAR gating kinetics correlate with Gria4 mRNA levels, but not with Gria2 mRNA levels (Geiger et al., 1995). Furthermore, relative levels of alternatively spliced AMPAR subunit flip and flop isoforms more accurately describe each subunit’s contribution to gating kinetics (Sommer et al., 1990; Mosbacher et al., 1994; Lambolez et al., 1996; but see Angulo et al., 1997). In fact, in cells expressing recombinant AMPARs, homomeric AMPAR channels containing Gria3 and Gria4 flip are 3.4 and 4x slower gating, respectively, than those containing the flop isoforms of each (Mosbacher et al., 1994). The Gria2 isoform has a weaker effect on AMPAR desensitization, though Gria2 flop increases the speed of desensitization relative to the flip isoform in both homomeric Gria2 and heteromeric Gria2/3 and Gria2/4 channels (Mosbacher et al., 1994; Koike et al., 2000). Thus, SGNs enriched with flop isoforms of the GluA3 or GluA4 subunits will always have AMPAR channels with faster kinetics and therefore faster synaptic responses than SGNs enriched with the flip isoform. Based on this prior evidence, we propose that SGNs enriched in Gria3 and/or Gria4 mRNA, especially in the flop isoform, will likely have increased temporal precision in response to sound, thus contributing to greater synchrony and higher ABR wave I amplitudes.

Our analysis of AMPAR subunit mRNA showed that the relative expression of the three AMPAR subunits (Gria2, 3, and 4) are not equal in individual SGNs, as evidenced by the differences in Gria3/2 and Gria4/2 indices. This result indicates the existence of a heterogenous population of SGN that could differ in AMPAR channel kinetics and permeability to calcium and sodium. For example, AMPARs lacking GluA2 are calcium-permeable and recently have been proposed as mediators of SGN excitotoxicity (Sebe et al., 2017; Hu et al., 2020). Thus, SGNs with lower levels of GluA2 relative to GluA3 and GluA4 will not only likely have faster synaptic responses once depolarized but will be more vulnerable to excitotoxicity due to AMPARs with higher permeability to calcium and sodium (Cull-Candy et al., 2006; Traynelis et al., 2010; Sebe et al., 2017; Hu et al., 2020). Interestingly, women are better protected from sensorineural hearing loss than men (Cruickshanks et al., 1998; Agrawal et al., 2008), and estrogen has been shown to have oto-protective effects in mice (Meltser et al., 2008; Simonoska et al., 2009; Williamson et al., 2019). Thus, the mechanism underlying sex differences in auditory processing may differ from the protective mechanism of estrogens. Whether SGNs can also be classified by differences in the AMPAR subunits’ flip/flop ratio, is still undetermined, because our analysis was only performed from whole inner ear samples rather than on single SGNs. Future studies that induce, for example, excitotoxicity in SGNs coupled with RNAscope, electrophysiology, and single-cell RT-qPCR for Gria2, 3, and 4 and their splice variants (flip and flop) will increase our understanding of whether these different SGNs differ in their fast AMPAR synaptic responses and gating, and calcium permeability.

Since females have higher ABR wave I amplitude (Figure 2), we hypothesized that they would have more overall AMPAR mRNA that could contribute to fast channel kinetics compared to males. Three findings from our mRNA quantification experiments suggest that Gria3 is the most likely AMPAR subunit to potentially mediate increased SGN synchrony in females. For one, females have more overall Gria3 mRNA than males (Figure 4). Second, females have a significantly higher Gria3/2 index overall than males, meaning the gradient of SGNs with Gria3 relative to Gria2 is shifted higher in females (Figures 6A–C). Last, Gria3 was post-transcriptionally processed into the flop isoform to a greater level in females than males who had more Gria3 flip (Figure 7). Thus, SGNs of the female cochlea have more overall Gria3 mRNA in the flop isoform, which was previously shown to lead to faster AMPAR gating kinetics than Gria3 flip whether it was dimerized to Gria2 flip or flop (Mosbacher et al., 1994). Also, more Gria3 relative to Gria2 suggests that females may have more heteromeric AMPARS containing Gria3 and Gria4, which are known to comprise AMPARs in fast excitatory synapses in the auditory CNS (Rubio and Wenthold, 1997; Gardner et al., 2001; Rubio et al., 2017). Our study shows that SGNs of female and male mice differ in their mRNA expression levels for Gria2, 3, and 4 AMPAR subunits, thus potentially contributing to our understanding of basic molecular mechanisms of sex differences in normal hearing mice.

Glutamate-induced calcium excitotoxicity via AMPARs does not only occur in the cochlea (Sebe et al., 2017; Hu et al., 2020), but also in other brain regions. In fact, calcium-permeable (GluA2 lacking) AMPARs are implicated in neuronal cell death and in neurological disorders in the CNS (Liu and Zukin, 2007; Guo and Ma, 2021). The sex differences in AMPAR subunits gene expression we describe here may not be specific to SGNs in the inner ear but present in other glutamatergic neurons in the CNS. For example, in human hippocampal neurons, schizophrenia patients had less overall Gria2 than controls (Eastwood et al., 1997), suggesting that differences in AMPAR subunit transcription could contribute to the disease. Interestingly, women and men differ in presenting schizophrenic symptoms and response to antipsychotic medication. There is some evidence that both sex hormones and sex chromosomes are related to schizophrenia in women and men (Li et al., 2016). Thus, probing sex differences in AMPAR subunit mRNA and the alternatively spliced flip and flop isoforms may aid in understanding sex differences in neurological disorders involving glutamatergic signaling through AMPARs.

The mechanism for the sex difference in AMPAR subunit mRNA expression described in this study is still unknown. Estrogen receptors (ER) and androgen receptors (AR) are both transcription factors (Levin and Hammes, 2016), and therefore estrogens and androgens could regulate the expression of AMPAR subunits either directly or indirectly via the recruitment of other transcription factors. Evidence that ERα and ERβ are found in SGNs (Stenberg et al., 1999; Meltser et al., 2008; Shuster et al., 2021) together with data from knockout mouse models have demonstrated that ERβ is important for SGN health (Simonoska et al., 2009). On the other hand, while the Gria2 and Gria4 genes are located on autosomes, the Gria3 gene is located on the X chromosome (Mahadevaiah et al., 2009). Therefore, it is also possible that females have more Gria3 mRNA due to incomplete X-inactivation in some or all SGNs leading to increased gene dosage (Carrel and Willard, 2005; Berletch et al., 2011). This would suggest that this sex difference is not due to hormonal differences but rather intrinsic biological sex due to the XX/XY sex-determination system. It is essential to study potential anatomical and molecular sex differences in the nervous system to better tailor treatment options for optimal auditory processing in both sexes.