Neurod1 Is Essential for the Primary Tonotopic Organization and Related Auditory Information Processing in the Midbrain

Neurod1-deficient mice retain many SG neurons

Neurod1 was eliminated specifically in the inner ear by crossing Neurod1^loxP/loxP mice (Goebbels et al., 2005) with Islet1^cre mice (Dvorakova et al., 2016; Neurod1cKO). Neurod1cKO mice are viable without any obvious abnormal motor activity behavior that would indicate major defects in the vestibular system. To evaluate the cochlear phenotype of Neurod1cKO mutants, we stained cochlear whole mounts with antibody to Myo7a, a hair cell marker. The organization of the organ of Corti of mutants was comparable to controls, with three rows of outer hair cells (OHCs) and one row of inner hair cells (IHCs; Fig. 1A–D). In contrast to previous work showing that all (Liu et al., 2000) or nearly all SG neurons degenerate in different Neurod1 mutants (Kim et al., 2001; Jahan et al., 2010a), we detected dense radial fibers and SG neurons in the cochlea of Neurod1cKO (Fig. 1A–D,E–H,E′–H′). However, there were noticeable abnormalities associated with aberrant innervation and migration of neurons in the cochlea, e.g., elongated and reduced radial fibers, increased spacing between radial fiber bundles, no intraganglionic spiral bundle formed by efferent axons, spread out and missing SG neurons, disorganized central axons, and complete absence of SG at the apical end. In the apex of Neurod1cKO, radial fiber bundles were noticeably disarranged with crossing fibers (Fig. 1D) and the epithelium was disorganized with missing OHCs, trans-differentiated OHCs into IHCs, and disarranged β-tubulin⁺ supporting pillar cells (Fig. 1I–L). This sensory epithelium phenotype of trans-differentiation of some OHCs into IHCs was comparable to previous reports on different Neurod1 deletion mutants (Liu et al., 2000; Kim et al., 2001; Jahan et al., 2010a). We also directly assessed the synapses between auditory nerve terminals and IHCs by immunostaining the cochlear sensory epithelium for CtBP2, a major component of presynaptic ribbons in the adult animals (Chumak et al., 2016). Ribbon counts showed an average reduction of 58% in Neurod1cKO compared with controls, indicating a reduction of IHC afferent ribbon synapses (Fig. 1M–O). The total number of SG neurons positioned inside the Neurod1cKO cochlea was reduced at P0 by 80% compared with control littermates but the number of surviving SG neurons was maintained during postnatal development up to adulthood (Fig. 1P–R).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1.

Morphology of the Neurod1cKO inner ear is altered. A–D, Whole-mount immunostaining with anti-Myo7a [a marker of hair cells (HCs)] and anti-acetylated α-tubulin (nerve fibers) antibodies shows subtle changes in a number of radial fibers (RFs), lack of the intraganglionic spiral bundle (IGSB; arrow), noticeable disorganization of RFs, and overshooting fibers in the apex of Neurod1cKO cochlea (C, D). Scale bar, 50 μm. E–H, Whole-mount immunostaining of SG neurons (anti-NeuN, a neuronal soma marker) and innervation shows reduction and disorganization of SG neurons in mutants. Scale bar, 50 μm. E′–H′, Higher-magnification images demonstrate a decreased number and altered distribution of SG neurons. Scale bar, 20 μm. I, J, Immunohistochemistry for Myo7a and for β tubulin (a marker of inner pillar supporting cells) shows the disorganization of OHCs (red) and inner pillar cells (IPs; green) in the apex in Neurod1 cKO compared with the organ of Corti of control littermates at P7. Scale bar, 20 μm. HS, Hoechst nuclear staining. K, L, Immunohistochemistry for prestin (a marker for OHC) and calretinin (a marker for IHCs) shows trans-differentiated IHCs instead of OHCs (arrows) in the apex of the mutant cochlea at 1 month of age (M1). Scale bar, 20 μm. M, N, Confocal analysis of immunostaining for a synaptic ribbon protein (CtBP2) in the IHC area shows a reduction of ribbons in mutant adult mice compared with the controls. Phalloidin labels F-actin in stereocilia of hair cells. HS, Nuclear staining Hoechst. Scale bar, 10 μm. O, Quantification of ribbon synapses per IHC area along the tonotopic axis. Data are expressed as the mean ± SEM; n = 3; two-way ANOVA. ****p < 0.0001. P, Q, Immunohistochemistry for NeuN in SG neurons at M1, the dotted line indicates the boundaries of SG in the vibratome sections of the cochlea. Scale bar, 50 μm. R, Quantification of SG neurons in the control and Neurod1cKO cochlea at P0 and M1. Data are represented as the mean ± SEM; n = 3/genotype/age. Two-tailed unpaired t test. ***p < 0.001.

Neurod1cKO mice have a shortened cochlea, smaller cochlear nuclei, normal size of the IC, and altered ABR

The dissected cochleae of control and Neurod1cKO mice were mounted, imaged in a confocal microscope and the length of the organ of Corti was measured (Fig. 2A) and found to be on average ∼40% shorter in the adult mutant (Fig. 2B). Note that the density of radial fibers in Neurod1cKO is reduced compared with littermate control. Although their density was evidently lower than in controls, radial fibers were preserved in adult mutant cochlea (Fig. 2G). We next measured the volume of the CN at the entry of the auditory nerve in coronally sectioned adult brains. The volume of the CN was reduced by ∼39% (Fig. 2C,D). Because Islet1^cre is not expressed in the CN (Fig. 2H), the size reduction is not likely because of Neurod1 deletion in the CN (Fritzsch et al., 2006) but is exclusively a secondary effect of reduced afferent input consistent with the effects of neonatal cochlear ablation previously reported (Rubel and Fritzsch, 2002). In contrast to the CN, sections of the IC showed no significant reduction in the adult mutant mice, indicating that the reduction of the CN does not result in a matching reduction of the IC (Fig. 2E,F).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2.

Deletion of Neurod1 affects the size of the components of the auditory pathways. A, B, Representative image of a whole-mount of control and Neurod1cKO cochlea at P3. Hair cells labeled with anti-Myo7a (green) and neuronal fibers with anti-β-tubulin (magenta). Note a reduced density of radial fibers in Neurod1cKO cochlea. Scale bar, 100 μm. Quantification of cochlear length at 4 weeks of age (n = 3). C, Sections of the CN immunostained with anti-NeuN (red) of adult control and Neurod1cKO, showing the DCN and PVCN. The line indicates the boundaries of the CN. HS, Nuclear staining Hoechst. Scale bar, 200 μm. D, Quantification of the volume of the adult CN normalized to body weight (n = 5). E, F, Immunostaining of coronal brain sections for NeuN and quantification of the volume of the adult control and Neurod1cKO IC, normalized to body weight (n = 5). Scale bar, 200 μm. Data (B, D, F) are the mean ± SEM. Two-tailed unpaired t test. ***p < 0.001, ****p < 0.0001. G, In the adult cochlea, decreased density and increased length of radial fibers (RFs) stained by OsO4 in Neurod1cKO are noticeable compared with the control. Scale bar, 100 μm. H, Using a LacZ reporter (R26R mouse line), Isl1-cre-mediated β-galactosidase reporter expression shows no Cre recombination in the CN in the brain sections. The dotted line indicates the boundaries of the CN. Cre recombination is detected by the LacZ reporter in the facial motor nucleus (FMN). Scale bar, 200 μm.

We evaluated DPOAE, as an objective measure of the function of the cochlear OHCs and cochlear amplification, using frequency range from 4 to 38 kHz (Fig. 3A). Based on the base-apical gradient from high to low frequencies (Müller et al., 2005), we detected significant DPOAE changes in the frequency range between 4 and 18 kHz corresponding to the locations of the OHCs in the mid-apex and the apex of the cochlea. These data indicate that OHC dysfunction mostly correlates with the morphological disorganization of the epithelium detected in the apex (Fig. 1I–L). Hearing of mice was assessed with ABRs, which measure electrical activity associated with the propagation of acoustic information through auditory nerve fibers to higher auditory centers. The ABR thresholds of mutant mice were elevated compared with the thresholds of control animals throughout the entire measured frequency range, with a relatively even threshold shift in all frequencies averaging ∼35 dB of SPL (Fig. 3B). Using click-evoked ABR, we evaluated waveform characteristics (Fig. 3C). ABR wave amplitudes were five times lower, but the absolute latency of peak I was shorter in Neurod1cKO mice. Wave I reflects the synchronous firing of the auditory nerve, whereas waves II–IV are attributed to the electrical activity of downstream circuits in the CN, superior olivary complex, and IC, respectively (Martin and Rickets, 1981). We expected that CN responses should be more affected compared with IC responses because of the overall size reductions. The shorter wave I latency in Neurod1cKO indicates altered properties of SG neurons in information processing. The relative interpeak latency between peaks I and II was retained within normal limits, whereas the interval between peaks II and III was prolonged, most likely because of the reduced size of the CN of Neurod1cKO. The timing and distribution of ABR III and IV peaks were comparable between control and mutant mice. After normalization and direct comparison of wave I with wave IV, we found that the 3:1 ratio of controls changed to a 1:1 ratio in the Neurod1cKO mice (Fig. 3D). Moreover, normalization showed that although attenuated, the overall intensity function of wave IV was near normal, whereas wave I was offset even more with higher intensities. These data indicate that morphological changes in the cochlea and CN translate to ABR differences.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3.

DPOAEs and ABRs are altered in Neurod1cKO. A, OHC function assessed by DPOAE shows significantly reduced levels in the low-frequency range. Data are the mean ± SEM; n = 12/control, n = 8/Neurod1cKO; two-way ANOVA with Bonferroni post hoc test. *p < 0.05, **p < 0.01, ****p < 0.0001. B, The average ABR thresholds of control (n = 12) and Neurod1cKO mice (n = 8). Five of the 8 Neurod1cKO animals did not have any response to the highest measured intensity at 2 or 40 kHz, indicated by the dotted line. Data are the mean ± SD; two-way ANOVA. ****p < 0.0001. C, Averaged ABR response curves evoked by an 80 dB SPL click; control in black, Neurod1cKO in red. The solid line shows the average and the shading indicates ±SEM. The ABR wave amplitudes of the mutant were five times lower than the control. Averaged individual ABR wave latency values are shown by the corresponding peaks. D, Averaged click-evoked ABR amplitude-intensity functions normalized to the wave amplitude at 90 dB SPL for ABR waves I and IV, and the ratio of ABR wave IV amplitude to wave I amplitude. Data are represented as the mean ± SD; two-way ANOVA. **p < 0.01, ***p < 0.001.

Spiral ganglion neurons form an aberrant “spiro-vestibular” ganglion

The application of different colored lipophilic dyes into the apex (green) and base of the cochlea (red), and anterior vestibular end organs (utricle, anterior and horizontal canal cristae; magenta) showed an aberrant distribution of spiral and vestibular ganglion neurons into a “spiro-vestibular” ganglion (SVG) complex in Neurod1cKO (Fig. 4B,D) in contrast to the vestibular ganglion of control mice with neurons exclusively labeled by anterior vestibular dye applications (Fig. 4A,C). Additionally, apex-dye application labeled fibers were detected in the base of the Neurod1cKO cochlea (Fig. 4B, arrows). In controls, the segregated central axons of neurons labeled by dyes injected into the base and apex in the auditory nerve (AN) were separated from the vestibular nerve (labeled by the dye injection into vestibular end organs; Fig. 4A,C). In contrast, the segregation of central axons was lost as basal and apical afferents completely overlapped with vestibular axons in the auditory-vestibular nerve (AVN) in Neurod1cKO (Fig. 4B, in detail D,E). Note that double-labeled neurons (from base/apex or base/utricular injections) were detected only in the Neurod1cKO at E18.5 and P7 (Fig. 4D–D‴,E–E‴, arrowheads), indicating multiple branches of some single SG neurons and abnormal innervation within the cochlea and between the cochlea and vestibular organs. Soma size of SVG neurons labeled from dye injections into the apex of the cochlea (green) was noticeable smaller in mutants (Fig. 4F; area: 253 ± 80 μm²) compared with adjacent vestibular neurons labeled from anterior vestibular organ applications (magenta; area: 364 ± 74 μm²; n = 80 neurons/2 mutants, t test, p < 0.001). We verified the tracing data using PROX1 immunocytochemistry that labels only SG neurons in wild-type (Fritzsch et al., 2010). In Neurod1cKO, PROX1⁺ translocated SG neurons were detected in the area of the vestibular ganglion, thus forming a mammalian SVG (Fig. 4H) instead of the normally distinct SG in Rosenthal's canal and a vestibular ganglion between the ear and the brain of controls (Fig. 4G). The adult SVG of Neurod1cKO mice was enlarged and deformed compared with control animals (Fig. 4I,J). Our data confirm previous suggestions of such unusual migration of some SG neurons to colocalize with vestibular neurons in Pax2^creNeurod1^loxP/loxP (Jahan et al., 2010b) or Wnt1^creSox10^loxP/loxP mice (Mao et al., 2014). We showed not only translocated SG neurons and the formation of the aberrant SVG but also single ganglion neurons having multiple branches within the cochlea (double-labeled by base/apex injections) and between the cochlea and vestibular end organs (double-labeled by cochlear/vestibular injections) in Neurod1cKO.

• Download figure
• Open in new tab
• Download powerpoint

Figure 4.

Translocated SG neurons form an aberrant, unsegregated spiro-vestibular ganglion in Neurod1cKO. A–F, Representative images of triple-dye labeling from the anterior vestibular end organs [labeled as utricle (U); magenta], apex (green), and base (red) of the cochlea shows the distribution of vestibular and SG neurons in whole-mounted collapsed stacks (A–D) and optical confocal sections (E, F). Dye injection sites are indicated by arrowheads and labeled correspondingly “apex”, “base”, “U” (A, B). In the control (A, C), neurons located in the vestibular ganglion (VG) of the control are only labeled by utricular dye applications (A, C, magenta; basal turn, red; apex, green) injections label only fibers in the AN and utricular injection labels fibers of the vestibular nerve (VN). In Neurod1cKO (B–F), neurons labeled by dyes injected into the cochlear base and apex, and the utricle combine to form an aberrant enlarged ganglion, the SVG, as shown in whole mounts (B, D) or optical sections (E, F). In mutants, there is a tendency of fibers labeled from the apex (green) to be more anterior in the combined SVG and the combined AVN (B), whereas apical fibers run in the most posterior part of the AN in control animals (A, C). Overlapping fibers (yellow) labeled by the basal and apical dye applications are present in the mutant cochlear base (B, arrows). Note that double-labeled SVG neurons can be seen after basal and apical injections (D, E, yellow cells). The double-labeled cells after basal and anterior vestibular end organ injections can best be visualized when the dye is false colored in blue (E, double-labeled cells in magenta). D′–D‴, E′–E‴, Images of individual colors of the separate channels shown as a merged image in D and E (arrowheads indicate double-labeled cells from merged images). F, A merged image shows only apical turn dye labeled SG neurons (green, dotted circles) near magenta labeled vestibular neurons in the SVG. Note that green cells are substantially smaller compared with the VG neurons labeled from anterior vestibular end organ dye application. Scale bars, 100 μm; dorsal is up and posterior is to the right in all dye-tracing images. G, H, Representative vibratome sections show translocated spiral ganglion PROX1⁺ neurons in the SVG as detected by immunohistochemistry for PROX1 (a marker for SG neurons) and NeuN (a marker for differentiated neurons) in Neurod1cKO at P0. Scale bar, 100 μm. HS, Nuclear staining Hoechst. I, J, Immunohistochemistry for NeuN and β-tubulin in the adult control VG and Neurod1cKO SVG. Scale bar, 50 μm. Dotted lines indicate the boundaries of the SG, VG, and SVG.

Spiral ganglion neurons are miswired in the cochlea of Neurod1cKO

Having recognized the major effects of Neurod1 deletion on the cochlea and auditory nucleus size, overall physiology, and abnormal arrangement of SG neurons together with vestibular ganglion neurons in the SVG complex (Fig. 4), we next wanted to establish the details of the SG connection between the organ of Corti and vestibular organs as well as connection between the cochlea and the CN. We first investigated whether SG neuron projections are restricted to the CN by backfilling afferents in the ear through dye tracing injections into the CN (red) and cerebellum (green). These dye-tracing data showed that afferents in the cochlea of controls were strictly labeled through dye injected into the CN (Fig. 5A, red). In contrast, some SG neurons in the developing cochlea of Neurod1cKO mutants projected to the cerebellum, because their afferents were colabeled through dye injected into the cerebellum (Fig. 5B, overlapping yellow fibers in the Neurod1cKO cochlea). Note that the location of the vestibular ganglion (green dye tracing injection into the cerebellum) was anterior to the auditory nerve in controls (Fig. 5A) but posterior to the aberrant SVG ganglion in Neurod1cKO (Fig. 5B) consistent with the reorganization of anterior posterior fibers and cell distributions in the SVG (Fig. 4B,D). We next wanted to establish whether the branches of cochlear projecting neurons also reach vestibular organs in the ear by injecting dye into the cochlea, posterior canal crista and utricle (n = 9 at E14.5, E16.5, E18.5). Applying dye to the posterior vertical canal crista consistently labeled fibers in the cochlea in a pattern reminiscent of backfilling from the cerebellum in Neurod1cKO (Fig. 5C,D). Neurons labeled with posterior canal injections had multiple branches toward the developing organ of Corti (Fig. 5D′, in detail D″) that overlapped with fibers to the base and to the posterior canal crista after dye was applied to the apex (Fig. 5 D′, inset, green), indicating the presence of neurons projecting to both the cochlea and the posterior canal crista. No such labeled intertwined innervation was found in the control cochlea at E14.5 nor were fibers labeled from the cerebellum (Fig. 5A). Later at E16.5, posterior vertical canal injections labeled at most a few efferent collaterals in control animals (Fig. 5E), similar to previous descriptions of branching of the efferent axons in the ear (Simmons et al., 2011; Sienknecht et al., 2014). In contrast, posterior vertical canal injection labeled SG fibers in the base, middle, and apex of the shortened cochlea of Neurod1cKO (Fig. 5F,F′). Not only translocated SG neurons but also vestibular ganglion neurons in the SVG innervated the cochlear hair cells, as indicated by an unusual innervation pattern near inner hair cells in Neurod1cKO (Fig. 5F′, arrow). These fiber terminations are comparable to rerouted vestibular fibers in neurotrophin mutants (Tessarollo et al., 2004; Fig. 5F′, inset), implying that some neurons reach both the organ of Corti as well as vestibular organs.

• Download figure
• Open in new tab
• Download powerpoint

Figure 5.

Loss of Neurod1 causes projection defects of SG neurons. A, B, Only afferents to the cochlea are labeled by the lipophilic dye injected into the CN (red) but not by dye injected into the cerebellum (green) in the control at E13.5. In Neurod1cKO littermates, some SG neurons and fibers are labeled from cerebellar injections (B, yellow, representing green overlapping with red). Note that afferents in the cochlea are distributed irregularly in mutants compared with controls (A, B, red). C–F′, The cyan arrowheads show the approximate application site of the dye [posterior canal crista (PC)] that results in labeling of multiple cells and branches projecting to the developing organ of Corti in Neurod1cKO. Injection of three different colored dyes [utricle, magenta; PC, cyan; apex, green] into the E14.5 Neurod1cKO ear shows overlap of spiral and vestibular ganglion neurons in the spiro-vestibular ganglion and magenta fibers to the cochlea (C, D, D′). Higher-magnification of fibers in the cochlea labeled from PC injections shows multiple neuronal branches toward the developing organ of Corti hair cells (D″). The dye injected into the apex of Neurod1cKO (D′, inset, green) labels fibers reaching the base as well as fibers to the posterior canal crista in a pattern reminiscent of fibers labeled from PC injections. These neurons have branched axons that reach not only the brain but also the PC nerve. E, F, F′, After PC dye application at E16.5, only a few efferent collaterals are labeled in controls that contribute to the intraganglionic spiral bundle (IGSB). In Neurod1cKO at E16.5, PC dye application labels SG neurons near the apex that terminate in an unusual pattern along IHC (F′, arrow). This pattern is comparable to the reported pattern of vestibular fibers in a replacement model of neurotrophins (F′, inset; data from Tessarollo et al., 2004), indicating that some SG neurons project like vestibular neurons in the brain, reach vestibular end organs (such as the canal crista) and terminate at the organ of Corti in a pattern comparable to rerouted vestibular neurons. Scale bars, 100 μm. PC, Posterior crista; U, utricle; VG, vestibular ganglion.

We next investigated in more detail the branching of SG neurons within the cochlea and between the cochlea and vestibular organs. In controls, dye injections into the cochlear apex (green) and base (red) resulted in a spatially restricted labeling of SG neurons and fibers in the cochlea according to the injection site without any labeling detected from injections to anterior vestibular organs (Fig. 6A). In addition to the projection of afferents from the ear to the brain, the ear is also innervated by efferents that are rerouted facial branchial motor neurons that end on hair cells instead of muscle fibers (Simmons et al., 2011; Fritzsch and Elliott, 2017). Typically, dye applications into the cochlea label efferent branches as the dye diffuse close to the radial fibers into the organ of Corti and efferent fibers can be distinguished by a recognizable intraganglionic spiral bundle, formed exclusively by efferents. Accordingly, in our controls, most efferent fibers were strictly either labeled by the dye applications to the apex or base in controls (Fig. 6A′, red or green). Note that fibers formed evenly spaced radial bundles. Overlapping (yellow) fibers represented efferents forming the intraganglionic spiral bundle and branching to reach multiple points along the cochlea in controls (Fig. 6A′). In contrast similar dye applications (into the base, apex, and anterior vestibular organs) labeled neurons and branches throughout the inner ear in Neurod1cKO, showing mingled bundles of fibers from the apex and base (Fig. 6B). Tracing data revealed at least three different groups of “SG neurons” within the cochlea: those restricted to base or apex, those overlapping between base and apex and those that have in addition branches to vestibular organs (Fig. 6B–B‴). A few SG neurons labeled from vestibular injections revealed multiple branches to reach different parts of the cochlea (Fig. 6B′). Thus, our tracing experiments show that both the branching properties and migration properties of inner ear sensory neurons are affected by the deletion of Neurod1, generating neurons that are colocalized with vestibular neurons (Fig. 4) and reach multiple parts of the organ of Corti as well as different vestibular organs (Figs. 5, 6).

• Download figure
• Open in new tab
• Download powerpoint

Figure 6.

Distribution and multiple branches of SG neurons reveal disorganization in the Neurod1cKO cochlea. A, B, The application of different colored dyes into the apex (green), base (red), and utricle (magenta) shows a spatially restricted labeling of SG neurons and fibers based on the injection site in controls (A), whereas both fibers and neurons are labeled throughout the Neurod1cKO cochlea by the utricle, base and apex injections (B). A′, In controls, overlapping (yellow) efferent fibers superimpose in the intraganglionic bundles (IGSB). Some efferent fibers are labeled either by the dye application to the apex or base in controls (red or green) and together with afferents form evenly spaced radial bundles. B′–B‴, images of individual colors of the separate channels shown as a merged image in B. Note that SG neurons near the apex are double labeled with magenta (utricle) and green (apex) as indicated by arrowheads, and show in addition two or more local branches to different parts of the cochlea (B′). Scale bars, 100 μm.

To fully evaluate efferent innervation in the cochlea, we applied lipophilic dyes into the olivocochlear bundle (green) to label efferents and into the CN (red) to label SG neurons and afferent radial fibers reaching the organ of Corti. In control animals, efferents formed the intraganglionic spiral bundle along SG neurons and together with afferents formed evenly spaced radial bundles (Fig. 7A, in detail A′). In contrast, Neurod1cKO efferents reached outer hair cells like in control littermates but did not form the intraganglionic spiral bundle (Fig. 7B,B′). Accordingly, the function of outer hair cells of control and Neurod1cKO was comparable to the exception of the mid-apex, as shown by DPOAE tests (Fig. 3A). These data support the notion that much of the intracochlear trajectory of efferents depends on the distribution of SG neurons but efferents can reach hair cells no matter the deviation from their normal trajectory.

• Download figure
• Open in new tab
• Download powerpoint

Figure 7.

Disorganized inner ear efferents reach the organ of Corti in the Neurod1cKO cochlea. A, A′, In control animals, the injections of lipophilic dye into the olivocochlear bundle (OCB; green) labels inner ear efferent fibers forming the intraganglionic bundle (IGSB) and dye application into the CN (red) labels SG neurons and afferent radial fibers reaching the organ of Corti (OC). B, B′, In Neurod1cKO, the same dye applications label the scattered SG neurons (red dye-labeled; arrowheads) in the cochlea and show strong efferent innervation that lacks an IGSB formation. Note also the profound innervation of the OHC region of the OC only in the apex of the Neurod1cKO mutant (B, arrow). Scale bars, 100 μm.

Neurod1 deletion results in a loss of the tonotopic organization of the CN

Next, we evaluated the auditory nerve fiber projections to the brainstem cochlear nucleus complex, composed of the anterior ventral CN (AVCN), posterior ventral CN (PVCN), and dorsal CN (DCN; Muniak et al., 2013, 2016). In control animals, application of different colored lipophilic dyes into the apex (green) and base of the cochlea (red), and vestibular end organs (magenta) showed the segregation of central axons of SG neurons from the base and apex in the auditory nerve and that these discrete, non-overlapping bundles were separated from the vestibular nerve (Fig. 8A,E). In contrast, the segregation of central axons of SG neurons was lost as basal and apical afferents completely overlapped in the AVN (Fig. 8B, F). Dye applications in the control cochlea showed the tonotopic organization of the CN subdivisions, with low-frequency-encoding fibers from the apex terminating dorsally and high-frequency-encoding fibers from the base terminating ventrally in both the VCN and DCN (Fig. 8A,E). In the CN of Neurod1cKO, fibers labeled with lipophilic dye tracing from the entire cochlea were restricted to the VCN with just a few fibers occasionally expanding to the DCN (Fig. 8B,F–F‴). Very large dye injections covering the basal half and apical half resulted in segregated projections with almost no space between apical and basal afferents in control animals (Fig. 8G). In contrast, comparable injections in Neurod1cKO mutants showed the entering fibers overlapped in the AVN and were mostly restricted to the VCN with just a few fibers occasionally expanding to the DCN (Fig. 8H,I). In essence, the DCN remained virtually non-innervated by cochlear afferents and the dorsal part of both AVCN and PVCN likewise received only sporadic and variable individual fibers, indicating a relocation of high-frequency-encoding basal turn fibers. In controls, cochlear afferents formed parallel fibers, the isofrequency bands (Fig. 8C,G, inset), whereas Neurod1cKO afferents neither expanded across the entire CN nor showed a parallel fiber organization (Fig. 8D,H,I). As expected from overlapping peripheral connections of neurons in the SVG (Fig. 4) as well as neurons in cochlea reaching vestibular organs (Figs. 5, 6), we also found fibers in the CN of Neurod1cKO labeled by the application of lipophilic dye into vestibular end organs (Fig. 8F″). Thus, Neurod1cKO mutants have both the peripheral and central connections of spiral and vestibular neurons incompletely segregated and miswired with overlapping central axons of SG neurons in the auditory-vestibular nerve and limited projections of widely ramifying fibers to the ventral part of CN that showed variable and inconsistent segregation (Fig. 8J, summary).

• Download figure
• Open in new tab
• Download powerpoint

Figure 8.

Neurod1cKO SG neurons project unsegregated widely ramifying central axons to the CN. A, C, E, Injections of different colored dyes (magenta, vestibular organs; red, base; green, apex) label distinct bundles of neuronal fibers (AN) projecting to the CN and vestibular nuclei in controls. The tonotopic organization of the CN subdivisions in controls is shown in the AVCN and the DCN with low-frequency fibers labeled from the apex and high-frequency from the base and no labeling between (the middle cochlea; A, E, G). The only mixed bundle (yellow) are inner ear efferents (IEEs; A). In Neurod1cKO, neuronal fibers completely overlap forming a mixed-labeled AVN (yellow) and are restricted to the ventral part of the CN with just a few fibers occasionally expanding to the DCN (B, D, F). The overlapping fibers in the AVCN indicate a loss of the tonotopic organization of the central projections (B, yellow fibers). C, D, Higher-magnification shows central afferents as parallel fibers (isofrequency bands) in control AVCN and DCN, whereas central projections of afferents in mutants branch in a restricted and erratic pattern in the CN (green dye injection into the apex). E, F, The tonotopic organization of the CN subdivisions with low-frequency-encoding fibers from the apex (green) terminating ventrally and high-frequency-encoding fibers from the base (red) terminating dorsally in the control AVCN and DCN. In contrast, disorganized fibers labeled by dye injected in the apex of Neurod1cKO project into the ventral part of the AVCN, whereas virtually only sporadic fibers labeled by basal dye injection reach only PVCN subdivisions. F′–F‴, Single color images show the incomplete cochlear projection segregation as well as some “vestibular” fibers projecting to the CN of Neurod1cKO (F″) that never happens in control animals (A, C, E). G, Very large lipophilic dye injections result in barely segregated projection with almost no space between apical and basal afferents in controls. Inset, Cochlear afferents organized as parallel fibers in isofrequency bands. H, I, Comparable large dye injections label sporadic and divergent fibers mostly in the ventral part of the CN of Neurod1cKO at P7 and E18.5. Note that single afferents project widely ramifying across the CN complex (H, I, arrowheads). Scale bars, 100 μm. J, Schematics of changes in Neurod1cKO: (1) SG neurons are miswired and misplaced into an aberrant SVG, (2) overlapped central auditory and vestibular afferents form the AVN, and 3) the CN has reduced projections without tonotopic organization both in terms of branch distribution within the cochlea and to the CN. The triangles indicate the application sites of the dyes in the base, apex, utricle (U), and posterior canal crista (PC). AC, Anterior crista; HC, horizontal crista; VG, vestibular ganglion; VN, vestibular nerve.

Since the size of the CN depends on the full complement of afferents and the survival of CN neurons depends on innervation (Rubel and Fritzsch, 2002; Syka, 2002), we next investigated the survival of one class of neurons that receive input from auditory nerve fibers through the large end bulbs of Held, the bushy cells in the AVCN (Muniak et al., 2016). Although the size of the AVCN in Neurod1cKO was reduced, SG afferents of Neurod1cKO and control formed morphologically comparable clusters of boutons that wrap the somas of their targets with the end bulbs of Held, as shown by excitatory VGLUT1 synaptic marker labeling (Fig. 9A–B′). However, innervation patterns of afferents outside the overlapping area in AVCN of Neurod1cKO (Fig. 8D,H,I) suggest an unusual distribution that does not conform to the regular pattern of stratified afferent projections in controls (the isofrequency bands in Fig. 8C,G, inset), which represent the tonotopic input of the organ of Corti onto the CN (Muniak et al., 2013, 2016).

• Download figure
• Open in new tab
• Download powerpoint

Figure 9.

The formation of isofrequency bands in the anteroventral CN of Neurod1cKO is altered. A, A′, B, B′, Auditory nerve synaptic terminals of control and mutant mice are comparable, as anti-VGLUT1 labels the endbulb of Held synapses and anti-parvalbumin labels the bushy cell soma in the AVCN of the adult control and Neurod1cKO. C, Activated neurons are labeled by anti-c-Fos antibody in response to 15 kHz pure tone stimulation in the AVCN of 2 months old controls and Neurod1cKO. Dotted line indicates boundaries of the AVCN. A, Anterior; L, lateral; M, medial; P, posterior. D, The total number of c-Fos⁺ neurons determined in the sections of the AVCN of control (n = 4) and Neurod1cKO mice (n = 4). Two sections were evaluated per each individual mouse. Values are mean ± SEM. **p = 0.0099, t test. Scale bars: 200 μm; A′, B′, 10 μm. HS, Nuclear staining Hoechst.

c-Fos shows a distorted frequency map representation in the CN

We next wanted to investigate whether SG afferents formed functional synapses onto AVCN bushy cells and how distinct frequencies are mapped throughout the system in terms of the activation of the immediate early response of c-Fos, a technique shown to allow such mapping (Tomková et al., 2015; Karmakar et al., 2017). We subjected freely moving adult mice (2 months old) to pure tone auditory stimulation with 15 kHz pips at 75 dB for 90 min in a sound-proof chamber. We used a 15 kHz tone that drives cochlear activation near the middle in controls and likely also in Neurod1cKO mutants, assuming the apex and base shorten equally. Consistent with previous reports (Karmakar et al., 2017), in control mice, we found a narrow band of c-Fos⁺ cells near the center of the AVCN, corresponding with the known isofrequency representation (Fig. 9C). In contrast, in Neurod1cKO, 15 kHz-stimulation activated AVCN neurons but the c-Fos activation area was spread all over the AVCN and the number of activated neurons was doubled (Neurod1cKO 20.7 ± 2.5; control 9.8 ± 1.4, p = 0.0099; Fig. 9D). These data indicate that the pure tone stimulation of 15 kHz activated AVCN neurons but the tonotopic precision of SG axon targeting was degraded with a spread of activation to many surrounding frequencies.

Frequency tuning, intensity, and temporal coding properties of IC units are distorted in Neurod1cKO

With this background on neuroanatomical, quantitative and frequency related c-Fos activity changes in mind, we next investigated the tuning properties of IC neurons at various frequencies as well as other parameters of their function i.e., two-tone suppression, intensity, and temporal resolution as the IC is the first level of auditory space map projection and shows some amelioration of primary afferent dysfunction (Buran et al., 2010; Pelgrim et al., 2018).

The study of the tuning characteristics of IC neurons revealed striking differences in the shape of the excitatory receptive fields between the Neurod1cKO and control mice. Instead of simple narrow V-shape receptive fields (a mono-peak response) seen in the controls, we recorded mostly wide receptive fields with two or more peaks in clusters of IC neurons of Neurod1cKO mice (Fig. 10A), suggesting multiple inputs from the lower levels of the auditory system. Two-tone stimulation, used to detect inhibitory areas surrounding the excitatory tuning curves, showed the presence of low- and high-frequency sideband inhibitory areas in controls, and small and disorganized inhibitory areas in Neurod1cKO (Fig. 10B). Because we mostly recorded the multiple-unit activity of neuronal clusters in the IC, we wanted to establish whether the multipeaked broad tuning curves in mutants are formed by the integration of responses of several neurons with different best frequencies or whether they reflect a miswiring of auditory fibers occurring below the IC. We found that the multipeaked broad tuning curves in the responses of neuronal clusters were also observed in the tuning curves of isolated single units recorded at the same electrode (Fig. 10C–H).

• Download figure
• Open in new tab
• Download powerpoint

Figure 10.

Deletion of Neurod1 affects the IC tuning curves of multiple and isolated single units. A, Representative examples of tuning curves recorded in the IC in control and Neurod1cKO mice. B, Response map to two-tone stimulation (fixed tone of 13 kHz), shows low- and high-frequency sideband inhibitory areas in control (the areas are outlined by white lines) and small and disorganized inhibitory areas in Neurod1cKO mice. C, Feature space window (in which each action potential is represented by a single point) with a delineated cluster of action potentials (purple dashed box) belonging to isolated single units recorded at the electrode channel. Action potentials corresponding to multiple unit activity are within the larger box (indicated by solid blue line). D, E, Examples of the tuning curves representing multiple unit activity and the activity of a single isolated unit from five different recordings in Neurod1cKO mice. F, Tuning curve of a multiple unit in Neurod1cKO mice; for the same recording, frequency profile of the same response to tonal stimulation at different frequencies with bimodal frequency tuning (at 30 dB sound attenuation) is calculated from single-unit activity (G). H, Action potential curves, recorded at two different points marked with the corresponding color in the frequency profile at (E), demonstrating very similar spike shapes and therefore classified as generated by the same neuron (single unit).

The investigation of responsiveness of IC units of the mutant mice to different sound frequencies revealed a frequency range with a limited presence of CFs from 9 up to 28 kHz (Fig. 11A). These features are consistent with the observed shortening of the cochlea, and the reduced amplitude of the ABR. In addition, the excitatory thresholds of IC units were higher in mutants than in control animals in all measured frequencies (Fig. 11A). One commonly used metric of auditory tuning is the “quality factor”, or Q, defined as the CF divided by the bandwidth, typically measured 10 dB above the minimum threshold. Comparing the data of control and mutant mice, the tuning curves recorded at individual electrodes in the IC of mutant mice had a significantly lower quality factor Q₁₀, indicating that their frequency selectivity was worse (Fig. 11B).

• Download figure
• Open in new tab
• Download powerpoint

Figure 11.

Characteristics of IC neurons are affected by acoustic information processing in the periphery. A, The dependence of IC excitatory thresholds on the CF of neuronal clusters (multiunits) in control and Neurod1cKO mice. Excitatory thresholds are shown as individual units and as averages in 0.3 octave bins. Data are mean ± SEM. Two-way ANOVA with Bonferroni post hoc test. ****p < 0.0001. B, Sharpness of the neuronal tuning expressed by quality factor Q₁₀ (the ratio between the CF and bandwidth at 10 dB above the minimum threshold) averaged in 0.3 octave bins. Data are represented as the mean ± SEM; two-way ANOVA with Bonferroni post hoc test. ****p < 0.0001, **p < 0.01. C, Percentages of strictly monotonic, saturating and non-monotonic RIFs. Fisher's exact test. ***p < 0.001, ****p < 0.0001. D, Comparison of the RIF parameters in control and Neurod1cKO mice: dynamic range, relative response at the RIFs point R10 (10% of the RIF response magnitude), maximum response magnitude and spontaneous activity. Data are the mean ± SD; unpaired t test. *p < 0.05, ****p < 0.0001. For all extracellular recordings Neurod1cKO (n = 9; 432 units from the IC) and control mice (n = 10; 480 units) were used. E, The average probability of a response to individual clicks in the series of clicks. Inset, Diagram of the clicks train, where the rate continuously speeds up and then slows down again (mean ± SEM; n = 10/control, n = 9/Neurod1cKO; ****p < 0.0001, unpaired t test). F, Degree of synchronization for different interstimulus intervals (ISIs) expressed using the vector strength computed for responses to click trains with different interclick intervals (mean ± SD; n = 10/control, n = 9/Neurod1cKO; ****p < 0.0001, two-way ANOVA with Bonferroni post hoc test).

To assess the responsiveness of neurons to sound intensity, we recorded the responses of IC multiunits to WN bursts of variable intensity and used them to construct RIFs. The percentage of multiunits with non-monotonic responses (i.e., encoding the increasing intensity initially by increasing spike rate, and then decreasing spike rate as sound intensity is further increased) was doubled in the IC of Neurod1cKO (35%) compared with controls (16%; p < 0.0001; Neurod1cKO n = 9; 432 IC recording sites and control n = 10; 480 IC recording sites; Fig. 11C). Saturated rate intensity functions were present only in 7% of the measured IC multiunits in Neurod1cKO compared with 37% in the controls (p < 0.0001), whereas monotonic rate-intensity functions represented 55% of the recorded neuronal clusters in Neurod1cKO compared with 47% in the controls (p = 0.0007). As a result, the IC multiunits in Neurod1cKO had a narrower dynamic range (23.38 ± 0.31 dB vs control 24.23 ± 0.29 dB, p < 0.05) as well as significantly lower maximum response magnitudes (Neurod1cKO 16.96 ± 0.44 spikes/s; control 25.26 ± 0.51 spikes/s, p < 0.0001; Fig. 11D). Additionally, it was necessary to use significantly higher WN intensity in Neurod1cKO to reach the R10, representing 10% of the RIF response magnitude (53.8 ± 0.5 dB SPL; control 34.3 ± 0.4 dB SPL, p < 0.0001; Fig. 11D), as a consequence of a higher excitatory threshold. However, at this low suprathreshold noise level, Neurod1cKO IC units showed facilitated evoked activity (0.15 ± 0.0029 vs control 0.12 ± 0.00091, p < 0.0001). We also observed significantly higher spontaneous activity in the Neurod1cKO IC units than in controls (Neurod1cKO 6.21 ± 0.34 spikes/s; control 1.5 ± 0.11 spikes/s, p < 0.0001; Fig. 11D).

We performed acoustical stimulation of the IC units with two types of click trains, a long one with changing frequency of clicks in the train and a train of five clicks with different interclick intervals from 100 up to 5 ms. In Neurod1cKO, this revealed a significantly lower efficiency in recognizing individual clicks in the complex train (Neurod1cKO 23.96 ± 1.07%; control 49.86 ± 1.47%, p < 0.0001; Fig. 11E); and a lower degree of response synchronization with clicks in the train for the whole range of interclick intervals, suggesting a reduction in the precise response timing in the Neurod1cKO (Fig. 11F).

Auditory behavior of Neurod1cKO is altered

Previous work has demonstrated that even minor frequency presentation mistakes in the CN can lead to system effects, revealed by altered auditory behavior (Karmakar et al., 2017; Cruces-Solís et al., 2018). The thresholds and growth of the ASR were measured for tonal and WN startle stimuli in a continuous-noise and in a noiseless background. In mutants, ASR thresholds were significantly reduced for startle tone stimuli of 8 kHz (Neurod1cKO 72.14 ± 2.67 dB SPL; control 81.43 ± 3.78 dB SPL, p < 0.001; n = 8/group) and 16 kHz (Neurod1cKO 66.43 ± 6.27 dB SPL; control 83.57 ± 3.78 dB SPL, p < 0.0001), suggesting that instead of perceiving a pure tone, mutants perceive noise-like stimuli. This is further supported by the fact that the startle thresholds of mutants in response to pure tone stimulation were similar to those for WN (Fig. 12A). The differences in thresholds in response to WN between control and mutant mice were not significant. Consistent with the lower dynamic range, lower maximum response and higher percentage of non-monotonic responses of IC multiunits (Fig. 11C,D), we found that Neurod1cKO mice had a reduced amplitude of the startle response at higher intensities of noise startle stimuli (Fig. 12B). However, the 16 kHz stimulation resulted in significantly larger amplitudes of the response at low and middle intensities in mutants (Fig. 12C). To further assess complex auditory discrimination behavior in Neurod1cKO mutants, we subjected control and mutant adult mice to a PPI paradigm, which is an operational measure of sensory-motor gating. We used either WN or pure tone pips at increasing intensities as a non-startling acoustic stimulus (prepulse) that preceded the startle stimulus in a quiet background. PPI with WN as the prepulse resulted in mutants with a larger inhibition of the startle response than in controls (Fig. 12D). PPI growth functions demonstrated that the prepulse WN bursts can significantly modulate the startle response at sound pressures as low as 10 dB SPL in mutant mice, indicating heightened sensitivity of the central auditory system. Similarly, subsequent comparisons revealed that the PPI of Neurod1cKO for 16 kHz prepulses (a preserved audible frequency in the mutant IC) was enhanced starting at 40 dB SPL compared with control mice (Fig. 12E). Additionally, mutant mice showed higher sensitivity to increasing intensities of the continuous background WN than controls (Fig. 12F). Thus, Neurod1cKO displayed hyperacusis-like behavioral ASR and PPI responses parallel to animal models with noise-induced cochlear neuropathy, which is also associated with increased behavioral salience of still-audible sounds (Hickox and Liberman, 2014).

• Download figure
• Open in new tab
• Download powerpoint

Figure 12.

ASR and PPI responses are altered in Neurod1cKO. A, Thresholds of ASRs for WN, and tone pips at 8 and 16 kHz in control and Neurod1cKO mice. Data are the mean ± SEM, n = 8/group. Statistical significance determined by the Holm–Sidak method, t test. *p < 0.05, **p < 0.01, ***p < 0.001. Amplitude-intensity ASR functions for WN (B), and for tone pips of 16 kHz (C) in control and Neurod1cKO mice. Data are the mean ± SEM, n = 8/group. D, Efficacy of the WN and (E) 16 kHz tone prepulse intensity on the relative ASR amplitude; 100% corresponds to the ASR amplitude without a prepulse (uninhibited ASR). Data are the mean ± SEM, n = 8/group. F, Impact of a continuous background WN with increasing intensity on the relative ASR amplitude; 100% corresponds to the amplitude of uninhibited ASR (response in silence). Data are the mean ± SEM, n = 8/group. Two-way ANOVA with Bonferroni post hoc tests for B–F. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. G, Ratio of the orientation (orient) and transition (trans) time in the presence of background WN (BWN) or series of clicks to the time in silence on rods of 20 or 15 mm diameters. The green line depicts the point when the ratio is 1 (no change in measured times with/without sound). Data are the mean ± SEM (control, n = 4; Neurod1cKO, n = 3). Statistical significance determined using two-way ANOVA with Sidak's multiple-comparison test. *p < 0.05, **p < 0.01.

Neurod1 deletion results in aberrant vestibular function

Finally, we investigated the possibility of interference of vestibular function of Neurod1cKO with auditory stimuli. This would be expected given the distribution of SG neurons in the aberrant SVG, shared central projections after labeling vestibular and cochlear afferents, and labeled neurons in the SG by dye injected into vestibular organs (Figs. 4, 6, 8). Thus, these results demonstrate interconnections of afferents between cochlear and vestibular sensory epithelia previously only shown in certain neurotrophin mutants (Tessarollo et al., 2004) and Foxg1 mutant mice (Pauley et al., 2006). We tested whether the motor coordination of Neurod1cKO can be affected by acoustic stimulation. We conducted two different experiments, one with continuous WN at an 80 dB SPL intensity, and another with a 600 ms series of clicks with a changing frequency of spike occurrence at an intensity of 70 dB SPL. We measured turning time (time taken to perform an 180° turn from the starting position) and transition time on static rods of two different diameters. WN had a negative effect on the transition time of Neurod1cKO mice on both rods (Fig. 12G). However, the train of clicks with changing frequency of occurrence had a lower impact on the mutant mice than on the controls. These results demonstrate that Neurod1 deletion affected the processing of vestibular information during sound exposure. Although we cannot exclude the possibility that the altered motor behavior of Neurod1cKO mutants was the result of their increased sound sensitivity to WN exposure, our data indicate that the vestibular and auditory systems may be functionally interconnected. The possibility of an interconnection is supported by the observed shared central projections and distribution of SG neurons in the vestibular ganglion and their projection to both the cochlea and vestibular organs.