Altered cochlear innervation in developing and mature naked and Damaraland mole rats

Abstract Compared to many other rodent species, naked mole rats (Heterocephalus glaber) have elevated auditory thresholds, poor frequency selectivity, and limited ability to localize sound. Because the cochlea is responsible for encoding and relaying auditory signals to the brain, we used immunofluorescence and quantitative image analysis to examine cochlear innervation in mature and developing naked mole rats compared to mice (Mus musculus), gerbils (Meriones unguiculatus), and Damaraland mole rats (Fukomys damarensis), another subterranean rodent. In comparison to mice and gerbils, we observed alterations in afferent and efferent innervation as well as their patterns of developmental refinement in naked and Damaraland mole rats. These alterations were, however, not always shared similarly between naked and Damaraland mole rats. Most conspicuously, in both naked and Damaraland mole rats, inner hair cell (IHC) afferent ribbon density was reduced, whereas outer hair cell afferent ribbon density was increased. Naked and Damaraland mole rats also showed reduced lateral and medial efferent terminal density. Developmentally, naked mole rats showed reduced and prolonged postnatal reorganization of afferent and efferent innervation. Damaraland mole rats showed no evidence of postnatal reorganization. Differences in cochlear innervation specifically between the two subterranean rodents and more broadly among rodents provides insight into the cochlear mechanisms that enhance frequency sensitivity and sound localization, maturation of the auditory system, and the evolutionary adaptations occurring in response to subterranean environments.


| INTRODUCTION
Naked mole rats (Heterocephalus glaber) are long-lived rodents that live in large eusocial colonies in narrow, underground burrows. Their subterranean environment restricts oxygen levels, light exposure, and acoustic stimulation and has led to various physiological adaptations (Hetling et al., 2005;Park et al., 2017). Some of these adaptations might involve neotenous retention of immature characteristics by prolonged and/or arrested development (Larson & Park, 2009;Skulachev et al., 2017). The unique life history of naked mole rats has made them of enormous interest to comparative biologists and biomedical researchers. Nonetheless, their auditory system has been relatively uninvestigated. Further investigation of the structure and function of the auditory system in this species would provide comparative insight into the anatomical and physiological mechanisms that shape auditory responses throughout the lifespan of an organism.
Compared to many other rodent species, naked mole rats have relatively high auditory thresholds, poor high frequency hearing, and a limited ability to localize sound. For comparison, mice (Mus musculus) hear frequencies extending from 1 kHz to about 100 kHz, with the most sensitive frequency in the audiogram being approximately 16 kHz (Heffner & Heffner, 1993). Gerbils (Meriones unguiculatus) hear frequencies ranging from 125 Hz to about 60 kHz, with the most sensitive frequency being approximately 2 kHz (Müller, 1996;Ryan, 1976). Both mice and gerbils localize sound effectively, with mice using primarily interaural level differences and pinna cues (Allen & Ison, 2010;Heffner, Koay, & Heffner, 2001) and gerbils taking additional advantage of interaural time differences (Tolnai, Beutelmann, & Klump, 2017). In contrast, naked mole rats hear within a more restricted range of frequencies, from 125 Hz to 8 kHz, with only slightly more sensitive hearing at 4 kHz (Heffner & Heffner, 1993;Okanoya et al., 2018), and, moreover, have very limited ability to localize sound (Heffner & Heffner, 1993). Although hearing in Damaraland mole rats, another subterranean mole rat, has not been investigated, other Fukomys species show reduced cochlear tuning (Kössl, Frank, Burda, & Muller, 1996), elevated auditory thresholds and frequency ranges restricted to between 125 Hz and 4 kHz (five octaves), with best hearing within 0.8 and 1.4 kHz (Gerhardt, Henning, Begall, & Malkemper, 2017).
Previous work in naked mole rats has documented anatomical differences in their peripheral and central auditory structures, but how these differences contribute to comparative differences in their auditory function is unclear. Most conspicuously, naked and Damaraland mole rats lack pinnae, external structures of the outer ear that funnel sound and provide localization cues. Structural differences in the middle and inner ear of naked mole rats were also recently characterized by Mason, Cornwall, and Smith (2016) using microcomputed tomography and dissections. The authors of this study concluded that these structural differences were not by themselves sufficient to explain poor high frequency hearing in naked mole rats. Moreover, naked mole rats did not show specializations (most notably, hypertrophy of the malleus) that support low frequency hearing by inertial bone conduction (i.e., seismic hearing). Centrally, naked mole rats possess all major auditory brainstem nuclei (Gessele, Garcia-Pino, Omerbasic, Park, & Koch, 2016;Heffner & Heffner, 1993) except the superior paraolivary nucleus (Gessele et al., 2016). However, the structures involved in binaural processing of sound and sound localization are comparatively smaller (Gessele et al., 2016;Heffner & Heffner, 1993). The molecular organization of these structures has been investigated only more recently and only centrally. Perineuronal nets, specialized extracellular matrix structures responsible for synaptic stabilization in the adult brain, are observed in the auditory brainstem nuclei in naked mole rats and show no differences compared to other rodents (Beebe & Schofield, 2018). Nonetheless, molecular differences were reported in another study: specifically, the entire superior olivary complex lacked membrane-bound expression of the hyperpolarization-activated channel HCN1 (Gessele et al., 2016), an ion channel that greatly contributes to binaural temporal precision in the superior olivary complex of other rodents (Khurana et al., 2012;Koch, Braun, Kapfer, & Grothe, 2004).
In this study, we were particularly interested in examining the anatomical and molecular organization of cochlear innervation in the auditory sensory epithelium (organ of Corti) of naked mole rats because this innervation is responsible for encoding and relaying auditory signals to the brain and, moreover, shapes the development of auditory brainstem structures (O'Neil, Connelly, Limb, & Ryugo, 2011).
We hypothesized that differences in the innervation of the organ of Corti, perhaps resulting from reduced postnatal maturation, contribute to differences in auditory function in naked mole rats in comparison to other rodents. To begin to test this hypothesis, we used immunofluorescence to examine patterns of afferent and efferent innervation in excised organs of Corti from mature and developing naked mole rats compared to mice, gerbils, and also Damaraland mole rats (Fukomys damarensis). This selection of species allowed comparison of naked mole rats to rodents with sensitive high frequency hearing (mouse), sensitive low frequency hearing (gerbils) and another more closely related, subterranean, eusocial mole rate (Damaraland mole rats). We observed alterations in afferent and efferent innervation as well as their patterns of developmental refinement in naked mole rats compared to both mice and gerbils and also compared to Damaraland mole rats. Differences in cochlear innervation specifically between the two subterranean rodents and more broadly among rodents provides insight into the evolutionary adaptations to subterranean environments as well as the cochlear mechanisms that contribute to hearing sensitivity and sound localization.

| Animals
Animal protocols conformed to the respective legislation at the University of Illinois at Chicago, the University Medical Center Groningen (the Netherlands), the University of Groningen (the Netherlands), and the University of Oldenburg (Germany) outlined specifically in the Ages of animals are indicated in Table 1. For mice and gerbils exact ages were known and given in the text. For naked and Damaraland mole rats exact ages are given when known. Adult Damaraland mole rats were approximately 5 years of age and referred to as "mature" in the text.
For comparison, the onset of hearing in mice and gerbils occurs around 2 weeks of age but auditory maturation continues until weaning (Ehret, 1976;Finck, Schneck, & Hartman, 1972). Mice wean at 3-4 weeks of age and begin reproductive activity between 6 and 8 weeks of age (Prichett & Taft, 2007). Gerbils also wean between 3 and 5 weeks of age and reach reproductive activity at 4 months of age (Elwood, 1975). Although the exact age at the onset of hearing is not known in either naked or Damaraland mole rats, naked mole rats do hear by 1 year of age (Heffner & Heffner, 1993). Naked and Damaraland mole rats wean at 4 weeks of age (Jarvis, O'Riain, Bennett, & Sherman, 1994) and reach reproductive activity between 12 and 18 months of age (Edrey, Hanes, Pinto, Mele, & Buffenstein, 2011;Jarvis et al., 1994).
To obtain cochleae, animals were deeply anesthetized by either isoflurane inhalation (University of Groningen and University of Illinois at Chicago) or injection of a lethal dose of pentobarbital (University of Oldenburg) and then rapidly decapitated. Procedures published previously were used to isolate intact organs of Corti from cochleae for immunofluorescence McLean, Smith, Glowatzki, & Pyott, 2009). To prevent differences in antigen detection due to differences in processing, identical fixation and immunostaining conditions were used for all specimens. Specifically, whole cochleae were isolated from the temporal bone in ice cold phosphate-buffered saline (PBS) and immediately placed into ice cold 4% paraformaldehyde (PFA) diluted in PBS. Cochleae were fixed in 4% PFA/PBS for exactly 1 hr at 4 C. Organs of Corti were then dissected from the cochleae in ice cold PBS and treated with a blocking buffer (PBS with 5% normal goat serum, 4% Triton X-100, and 1% saponin) for 1 hr at room temperature. Turns were incubated in the primary antibody diluted in blocking buffer overnight at room temperature and then rinsed three times for 10 min in PBS with 0.2% Triton X-100 (PBT). After rinsing, the turns were then incubated in the secondary antibody diluted in blocking buffer for 4 hr at room temperature. Samples were then rinsed three times for 10 min in PBT, one time in PBS, and mounted on glass slides in Vectashield mounting medium (Vector Labs). All incubations and rinses were performed on a rocking table.
Unless specifically stated, data are collected from the 16 kHz region in mice and from the 2 kHz region in gerbils. These regions represent the frequencies of best hearing and were identified using previously determined cochlear place-frequency maps (Müller, 1996;Müller, von Hunerbein, Hoidis, & Smolders, 2005). For both mice and gerbils, these regions are approximately 50% along the length of the basilar membrane. The total basilar membrane (cochlear canal) length is approximately 6.8 mm in mice and 12.1 mm in gerbils (Kirk & Gosselin-Ildari, 2009). Because similar cochlear place-frequency maps have not been determined for either naked or Damaraland mole rats, we compared to equivalent positions (50%) along the length of the basilar membrane. The total basilar membrane length is approximately 5.8 mm in naked mole rats (Mason et al., 2016) and 10.9 mm in Fukomys micklemi (Mason et al., 2016) and 11.1 mm in Fukomys anselli (Kirk & Gosselin-Ildari, 2009), two relatives of the Damaraland mole rat.

P365
(1 year) 13.6 ± 1.2 (N = 4) 7.1 ± 0.7 (N = 4) Damaraland mole rat synaptic elements per hair cell, the Imaris spots function was used to detect immunopuncta within a given field of view. This value was then divided by the total number of hair cells within that field of view. Hair cell counts were obtained from counts of immunofluorescently detected hair cell nuclei (for CTBP2 immunolabeling). For organs of Corti from naked and Damaraland mole rats, CTBP2 immunolabeling was not restricted to the nuclei of the hair cells but also apparent in the nuclei of other cell types. In 3D reconstructions, the hair cell nuclei were identified by their generally larger size and proximity to the CTBP2-immunolabeled afferent ribbons. In addition, differential interference contrast images obtained in parallel aided identification of the hair cells. Volumes (μm 3 ) of immunopuncta were determined using the Imaris surfaces function. All numerical values and their 3D locations were exported from Imaris for further statistical analyses.
Custom script developed in R was used to calculate Euclidean distances between immunopuncta (as described previously in Sadeghi, Pyott, Yu, & Glowatzki, 2014;Ye, Goutman, Pyott, & Glowatzki, 2017) and to classify inner hair cell (IHC) afferent ribbons (CTBP2 immunopuncta) as either pillar or modiolar. Briefly, for this later analysis, x, y, were then used to classify ribbons as belonging to either the pillar or modiolar side. To allow comparison across z-stacks, ribbon volumes, determined using the Imaris surfaces function, from a given stack were normalized to the median volume from that stack.

| Statistics
Statistical analyses were performed using SPSS 24. Data for independent samples were analyzed using a general linear model ANOVA. In case of repeated measurements from the same individual subject (e.g., when obtaining data from multiple hair cells in a subject) a mixed model ANOVA was applied taking the subject into account. Post hoc pairwise tests used the Bonferroni correction, which was not applied in planned comparisons. Differences were considered significant when P < 0.05 (*) and highly significant when P < 0.01 (**). Group results are reported as the mean ± SEM. In most cases, n represents the number of individuals.

| RESULTS
3.1 | Afferent innervation in the organs of Corti from mature mice, gerbils, naked, and Damaraland mole rats In the mammalian organ of Corti, there are two classes of sensory hair cells (Goutman, Elgoyhen, & Gomez-Casati, 2015). The IHCs are responsible for relaying information to the brain for the perception of sound and make afferent connections to the type I spiral ganglion neurons. The OHCs are responsible for generating the cochlear amplifier and make afferent connections to the type II spiral ganglion neurons.
We quantified the numbers of afferent synapses in these two types of sensory hair cells in the mature organs of Corti from mice (C57BL6, 6 weeks old) gerbils (6 weeks old), naked mole rats (1 year old), and Damaraland mole rats (mature) by immunolabeling with an antibody against CTBP2, a protein enriched in the presynaptic afferent ribbons.
In all animals, both classes of hair cells show CTBP2-immunoreactivity indicative of presynaptic afferent ribbons (green, Figure 1a).
To compare afferent ribbon numbers per hair cell both within and across these four species, we quantified the number of afferent ribbons (CTBP2 immunopunta) per IHC (gray bars, Figure 1b Table 1.
Although not quantified, there were no apparent systematic differences in the numbers of afferent ribbons per OHC row in any of the four species investigated.
We also investigated postsynaptic afferent glutamate receptors by immunolabeling with a rabbit polyclonal antibody against GluR2/3.
In mice and gerbils, GluR2/3 immunoreactivity is punctate and juxtaposed to presynaptic afferent ribbons. In contrast to mice and gerbils, we observed no GluR2/3 immunoreactivity in naked or Damaraland mole rats (data not shown). Postsynaptic glutamate receptor labeling was not characterized further.
We were also interested in the distribution of the IHC afferent ribbons and ribbon sizes along the pillar-modiolar axis of the IHCs ( Figure 1c). In cat, pillar-modiolar differences in IHC afferent ribbon sizes correlate with pillar-modiolar differences in afferent fiber thresh- Liberman, 1996). Pillar-modiolar differences in IHC afferent ribbon sizes have also been documented in mice (Liberman, Wang, & Liberman, 2011) and gerbils (Zhang, Engler, Koepcke, Steenken, & Koppl, 2018). To compare differences between the numbers and volumes of pillar and modiolar IHC afferent ribbons, we localized and quantified the volumes of ribbons (CTBP2 immunopunta) as described in Section 2. In mice, there were fewer pillar ribbons (44%) than modiolar ribbons (56%). In contrast, in gerbils, naked mole rats, and Damaraland mole rats, the fractions of pillar and modiolar ribbons were equivalent. When comparing the mean normalized volumes of the pillar (gray circles, Figure 1c) versus modiolar (white circles, Figure 1c) IHC afferent ribbons, values within species were significantly different in mice, gerbils, and naked mole rats but not Damaraland mole rats. The steepness of this gradient (i.e., the ratio between the mean volumes of the pillar versus modiolar afferent ribbons) also appeared to vary across species, with the steepest gradients observed in mice. Values are provided in Table 2.
To summarize, Damaraland mole rats are particularly unique among the rodent species examined: unlike mice, gerbils, and naked mole rats, Damaraland mole rats show the same number of afferent ribbons per hair cell in both IHCs and OHCs and lack the pillar-modiolar gradient in IHC afferent ribbons sizes observed in other rodents.
3.2 | Tonotopic variations in afferent ribbon density and BK channel expression in the IHCs from mature mice, gerbils, and naked mole rats In both mice and gerbils, IHC afferent ribbon density peaks where the cochlea is most sensitive to sound (i.e., the most sensitive frequency in the audiogram; Meyer et al., 2009). Thus, reduced tonotopic variation in IHC afferent ribbon density could contribute to the poor sensitivity of hearing across frequencies observed in naked mole rats (Heffner & Heffner, 1993). Therefore, we analyzed IHC afferent ribbon density along the length of the cochlear spiral. In parallel, we investigated tonotopic changes in BK channel expression. BK channels are large conductance, voltage-, and calcium-activated potassium channels that are present in both inner and OHCs in the mammalian cochlea (reviewed in Pyott & Duncan, 2016). BK channel expression in both types of hair cells shows increasing expression from apical (low frequency) to basal (high frequency) regions following the onset of hearing in a variety of animals investigated (Pyott & Duncan, 2016). Therefore, the presence of BK channels in mammalian hair cells is associated with both frequency tuning of the cochlea and functional maturation of the cochlea. For these  reasons, we were interested in whether BK channels were also present in IHCs from naked more rats, and, if so, if BK channel expression together with afferent ribbon density varied tonotopically.
To examine afferent ribbon density and BK channel expression tonotopically in the IHCs, we immunolabeled organs of Corti isolated from mice (6 weeks old), gerbils (6 weeks old), and naked mole rats (6 months) with a mouse monoclonal antibody against CTBP2 (green, Figure 2a) and a rabbit polyclonal antibody against the BK channel (red, Figure 2a  In the mature mammalian cochlea, there are two olivocochlear efferent  Table 3 systems (Guinan, 2006).      Values are provided in Table 1.
For comparison across all four species, we additionally quantified the number of afferent ribbons (CTBP2 immunopunta) per hair cell for immature and mature gerbils (P9 and P42/6 weeks) and Damaraland mole rats (P5 and mature) compared to immature and mature mice (P7 and P21) and immature and mature naked mole rats (P7 and 1 year). When comparing within species for these two ages (and, hence, without Bonferroni correction for multiple comparisons as in Figure 4c), mice, gerbils, and naked mole rats showed However, because only one P5 Damaraland mole rat sample was available, no statistical comparisons were made between immature and mature Damaraland mole rats. Values are provided in Table 1. However, because only one P5 Damaraland mole rat sample was available, no statistical comparisons were made between immature and mature Damaraland mole rats. Values are provided in Table 5.
The normalization of ribbon volume used throughout our study precludes any comparison across different ages. Therefore, absolute ribbon volumes were used only for this specific developmental question.

| Developmental changes in postnatal efferent innervation in the organs of Corti from mice, gerbils, naked, and Damaraland mole rats
Observations in mice and gerbils (as well as rats and hamsters) indicate that efferent innervation of the mammalian cochlea undergoes extensive modification shortly before the onset of hearing, generally around the second postnatal week (Simmons, 2002). As part of this maturation, efferent terminals retract from the IHCs as medial efferent contacts onto the IHCs are lost and lateral efferent contacts onto the IHC afferent terminals consolidate. To assess postnatal maturation of efferent innervation in the naked mole rat and Damaraland mole rat, we immunolabeled organs of Corti isolated from  Table 4 P9 and P42 (6 week old) gerbils, P11 and 1 year old naked mole rats, and P5 and mature Damaraland mole rats with a rabbit polyclonal antibody against synapsin to label presynaptic efferent terminals (red, Figure 5a) and a mouse monoclonal antibody against CTBP2 to label IHC afferent ribbons and nuclei (green, Figure 5a). In both gerbils and naked mole rats, but less obviously in Damaraland mole rats, efferent terminals underwent extensive developmental reorganization. In young gerbils and naked mole rats, efferent terminals could be seen in the regions both above and below the IHC afferent ribbons and nuclei, consistent with efferent terminals contacting the IHCs. In contrast, in older gerbils and naked mole rats, efferent terminals were largely consolidated to the region below the IHC afferent ribbons and nuclei at the base of the IHCs. (This separation is not always apparent in the 2D projections through the 3D z-stacks shown in Figure 5a.) In contrast, in Damaraland mole rats, efferent terminals were largely consolidated to the region below the IHC afferent ribbons and nuclei at the base of the IHC even at the younger age.
To quantify the degree of developmental retraction, the distance between each afferent ribbon (CTBP2 immunopunctum) and its nearest efferent terminal (synapsin immunopunctum) was quantified as described previously (Sadeghi et al., 2014;Ye et al., 2017). We predicted that this distance would increase with development.
Cumulative distributions plotting the Euclidean distance between each IHC afferent ribbon and its nearest efferent terminal at the younger (gray line) and older (black line) ages showed a rightward shift with development for both gerbils (Figure 5b) and naked mole rats (Figure 5c). This rightward shift indicates an increase in the distance between nearest neighbor afferent ribbon and efferent terminal pairs, consistent with the loss of medial efferent contacts onto IHCs and consolidation of lateral efferent contacts onto the IHC afferent terminals below the IHCs. In contrast, this increase in the distance between nearest neighbor afferent ribbon and efferent terminal pairs was not observed in Damaraland mole rats (Figure 5d).
When comparing within species, in gerbils and naked mole rats, the mean Euclidean distances were significantly different between the two ages examined (Figure 5e-f). In contrast, the mean Euclidean distances were not obviously different between the two ages in Damaraland mole rats (Figure 5g). However, because only one P5 Damaraland mole rat sample was available, no statistical comparisons were made between immature and mature Damaraland mole rats. Values are provided in Table 6.

| Overview
In this study, we used immunofluorescence and quantitative image analysis to examine cochlear innervation in mature and developing naked and Damaraland mole rats compared to mice and gerbils. This work was motivated by previous findings that indicate relatively poor hearing in naked mole rats compared to other rodents (Heffner & Heffner, 1993) despite the absence of overt differences in the anatomy of the middle and inner ear in naked mole rats (Mason et al., 2016). is widely accepted that, in mammals, the type I auditory neurons are more abundant and contact single IHCs, whereas the much less abundant type II auditory neurons contact multiple OHCs (Spoendlin, 1972).
Thus, the altered ratio of inner and OHC afferent innervation in naked and Damaraland mole rats could reflect a greater proportion of type II auditory neurons and perhaps also a greater number of contacts between the type II auditory neurons and OHCs compared to other mammals. Future work examining the relative abundance of types I and

II auditory neurons and their contacts with the hair cells in naked and
Damaraland mole rats would provide insight into the generalizability of the pattern of types I and II afferent innervation across mammals.
In addition to quantifying the number of afferent ribbons per hair cell, we also examined differences in the distributions and sizes (volumes) of afferent ribbons between the pillar and modiolar sides of the IHCs. When comparing the relative numbers of modiolar versus pillar afferent ribbons, we observed equivalent numbers of modiolar and pillar synapses in gerbils as well as naked and Damaraland mole rats. In contrast, we observed more modiolar than pillar synapses (56% compared to 44%) in mice, similar to previous reports . However, these differences are minor and similar to the differences reported in different studies in the same species. For example, a previous study investigating gerbils reported 57% modiolar ribbons (Zhang et al., 2018) compared to our observation of 50%. Thus, these differences could easily result from subtle differences in the methods of defining the spatial coordinates of ribbons relative to the hair cells, and, therefore, further significance to these differences is not attributed.
When comparing the mean normalized volumes of modiolar versus pillar afferent ribbons, we observed statistically significant differences in mice, gerbils, and naked but not Damaraland mole rats. A difference  Tables 1 and 5 in size in the same direction, relatively larger modiolar afferent ribbons, has been reported previously in cats (Merchan-Perez & Liberman, 1996), mice , and gerbils (Zhang et al., 2018). Furthermore, in cats, the type I afferent fibers with high thresholds and low spontaneous rates tended to associate with relatively larger ribbons on the modiolar side of the IHC whereas afferent fibers with low thresholds and high spontaneous rates tended to associate with relatively smaller ribbons on the pillar side of the IHC (Merchan-Perez & Liberman, 1996). The presence of the same spatial gradient in ribbon volumes in naked mole rats suggests that this species may also display a similar distribution of type I afferent subtypes.  (Weisz, Glowatzki, & Fuchs, 2009), consistent with earlier in vivo recordings suggest that the type II auditory neurons respond only to the loudest sounds or not at all (Brown, 1994;Robertson, 1984;Robertson, Sellick, & Patuzzi, 1999 (Brown, Berglund, Kiang, & Ryugo, 1988). Alternatively, or in addition, enhanced activation of type II auditory neurons might contribute peripherally to modulate the local neuronal circuitry of the OHCs, for example, via reciprocal synapses (Thiers, Nadol, & Liberman, 2008) that may play a role in regulating OHC electromotility. Previous observations of poor cochlear tuning in a related mole rat F. anselli (Kössl et al., 1996), suggest that the increased OHC afferent innervation (observed in this work) in naked and Damaraland mole rats would not contribute to enhanced cochlear tuning but might even serve to broaden it. Future work examining the structural and functional differences in OHC afferent innervation between naked and Damaraland mole rats and other mammals would help clarify the as yet undetermined role of OHC afferent innervation by type II auditory neurons.

| BK channel expression in naked mole rats
Because across their range of audible frequencies (Heffner & Heffner, 1993).
These findings suggest that BK channel density in the IHCs may contribute to differential sensitivity across frequencies in the cochlea in ways that require further investigation. Finally, BK channels are also expressed in the OHCs of mice (Engel et al., 2006;Maison, Pyott, Meredith, & Liberman, 2013;Rohmann et al., 2015;Wersinger et al., 2010). In this study, BK channels were not detected in the OHCs of naked mole rats. However, we have previously reported that BK channel immunoreactivity in OHCs is sensitive to the duration of fixation (Wersinger et al., 2010). Therefore, the absence of BK channel expression in OHCs in naked mole rats should be investigated more carefully in future experiments.

| Efferent innervation in mature naked and Damaraland mole rats
In addition to examining patterns of afferent innervation, we also exam-  (Darrow, Maison, & Liberman, 2006;Irving, Moore, Liberman, & Sumner, 2011), modulate afferent excitability (Groff & Liberman, 2003), and protect the cochlea from noise-induced injury (Darrow, Maison, & Liberman, 2007). Finally, circuits within the lateral efferent system employ various neurotransmitters and show differences in their patterns of origination within the brain and termination within the cochlea (reviewed in Reijntjes & Pyott, 2016). Therefore, future work should additionally examine possible differences in the neurotransmitters and patterns of innervation employed by the lateral efferent system in naked and Damaraland mole rats. Differences in these patterns may permit a comparative approach to identify the functional contributions of neurotransmitter-specific lateral efferent circuits.
In addition to lateral efferent innervation, all species examined showed medial efferent innervation of the OHCs (Figure 3 lower panels). Although we did observe medial efferent terminals in naked and Damaraland mole rats, these terminals were smaller and less distinct than the medial efferent terminals observed in mice and gerbils.
The diffuse pattern of innervation made quantification of the efferent density difficult; however, we did detect significantly reduced medial efferent innervation density in naked mole rats compared to all other species, including Damaraland mole rats. In comparison to lateral efferent innervation, the role of medial efferent innervation is much better understood. Specifically, medial efferent innervation provides inhibitory input to the OHCs and, thereby, regulates the cochlear amplifier. In this way, medial efferent innervation is believed to improve signal transduction, enhance signal detection, and protect the cochlea from noiseinduced injury (reviewed in both Guinan, 2018; Lopez-Poveda, 2018).
Thus, our findings of reduced medial efferent innervation density are consistent with previous reports of reduced cochlear tuning in a related FIGURE 5 Comparison of efferent development in gerbils, naked mole rats, and Damaraland mole rats. (a) Projections through z-stacks of confocal sections spanning the regions of the organ of Corti containing the inner hair cell (IHC) and outer hair cell (OHC) afferent ribbons immunolabeled with an antibody against CTBP2 (green) and efferent terminals immunolabled with an antibody against synapsin (red). Developmental ages are indicated. Note that medial efferent terminals on OHC are not included within the z-range shown. (b-d) Cumulative histograms plotting the Euclidean distance between each CTBP2 immunopunctum and its nearest synapsin immunopunctum in immature (gray lines) and mature (black lines) gerbils, naked mole rats, and Damaraland mole rats. (e-g) Mean Euclidean distance between each CTBP2 immunopunctum and its nearest synapsin immunopunctum in immature (gray bars) and mature (white bars) gerbils, naked mole rats, and Damaraland mole rats. Asterisks indicate significantly different values (**P < 0.01). Values are provided in Table 6 mole rat species, F. anselli (Kössl et al., 1996). The observation of medial efferent innervation in the two mole rats examined in this study, however, contrasts previous ultrastructural observations in another mole rat species, Spalax ehrenbergi, in which no efferent innervation of the OHCs was observed (Raphael, Lenoir, Wroblewski, & Pujol, 1991). Thus, the patterns of medial efferent innervation appear to show considerable variation among subterranean rodents.
Finally, although the organization of the hair cells was not specifically investigated, the distribution of IHC afferent ribbons (Figure 1a upper panels) and lateral efferent terminals ( (Sobkowicz et al., 1982;Wong et al., 2014), which have led to the suggestion that smaller and more numerous ribbons in immature cells merge to form larger and less numerous ribbons in mature cells. Our findings in Damaraland mole rats, in which there is a developmental increase in ribbon size without a change in total ribbon number, suggest that the developmental increase in IHC afferent ribbon volumes occurs independently from the developmental pruning of IHC afferent ribbons.
We also examined postnatal pruning of OHC afferent ribbons. As in mice and gerbils, OHC afferent ribbons were significantly reduced in naked mole rats. In contrast, Damaraland mole rats showed similar numbers of OHC afferent ribbons at both the immature and mature ages. In our data, OHC afferent ribbon pruning is much more prolonged in naked mole rats but nevertheless certainly complete between P28 and 6 months of age. This prolonged period of maturation compared to mice and gerbils may reflect the prolonged retention of an immature pattern of afferent innervation into adulthood, paralleling observations in the naked mole rat brain (Penz et al., 2015).
As a second indicator of postnatal maturation, we examined retraction of efferent terminals from the IHCs ( Figure 5). This retraction results from the loss of medial efferent terminals directly contacting the IHCs and consolidation of lateral efferent terminals beneath the IHCs.
We quantified retraction as a developmental increase in the distance between the IHC afferent ribbons (located at the base of the IHCs) and their nearest neighboring efferent terminals. As validation of this methodological approach, we observed significantly greater distances between the IHC afferent ribbons and nearest neighboring efferent terminals in mature compared to immature gerbil. The same was also observed in mature compared to immature naked mole rats. In contrast, there was no evidence of retraction in Damaraland mole rats. In fact, the mean distances suggest that, in Damaraland mole rats, efferent innervation is already consolidated early in postnatal development or that there is simply no postnatal axosomatic efferent innervation of the IHCs. Again, because only one immature (P5) Damaraland mole rat specimen was available, these conclusions are made cautiously. Transient efferent innervation of the IHCs likely serves to shape activitydependent maturation of the auditory system (Clause et al., 2014).
Future work would be necessary to determine how changes in the time course or pattern of this efferent innervation contribute to maturation of the auditory system, and specifically maturation of afferent innervation, in naked and Damaraland mole rats.

| CONCLUSION
For comparative biologists, this work establishes naked and Damaraland mole rats as species worthy of future investigation to examine cochlear mechanisms that enhance frequency sensitivity and sound localization, maturation of the auditory system, and also evolutionary adaptations occurring in response to subterranean environments. For biomedical researchers, naked mole rats, which are extremely long-lived and highly resistant to various diseases of aging, may also prove particularly valuable to identify mechanisms that prevent or reduce age-related hearing loss, which involves loss of cochlear function and is one of the most prevalent chronic conditions of aging in humans (Gates & Mills, 2005).

ACKNOWLEDGMENTS
The project was supported by grants from the University of Groningen (to S. J. P) and the National Science Foundation (Grant 1655494 to T. J. P). We acknowledge the assistance of Suzanne Bezema with synaptic counts. We thank Dr. David Ryugo for critically reading the manuscript.