Lack of Fractalkine Receptor on Macrophages Impairs Spontaneous Recovery of Ribbon Synapses After Moderate Noise Trauma in C57BL/6 Mice

Noise trauma causes loss of synaptic connections between cochlear inner hair cells (IHCs) and the spiral ganglion neurons (SGNs). Such synaptic loss can trigger slow and progressive degeneration of SGNs. Macrophage fractalkine signaling is critical for neuron survival in the injured cochlea, but its role in cochlear synaptopathy is unknown. Fractalkine, a chemokine, is constitutively expressed by SGNs and signals via its receptor CX3CR1 that is expressed on macrophages. The present study characterized the immune response and examined the function of fractalkine signaling in degeneration and repair of cochlear synapses following noise trauma. Adult mice wild type, heterozygous and knockout for CX3CR1 on a C57BL/6 background were exposed for 2 h to an octave band noise at 90 dB SPL. Noise exposure caused temporary shifts in hearing thresholds without any evident loss of hair cells in CX3CR1 heterozygous mice that have intact fractalkine signaling. Enhanced macrophage migration toward the IHC-synaptic region was observed immediately after exposure in all genotypes. Synaptic immunolabeling revealed a rapid loss of ribbon synapses throughout the basal turn of the cochlea of all genotypes. The damaged synapses spontaneously recovered in mice with intact CX3CR1. However, CX3CR1 knockout (KO) animals displayed enhanced synaptic degeneration that correlated with attenuated suprathreshold neural responses at higher frequencies. Exposed CX3CR1 KO mice also exhibited increased loss of IHCs and SGN cell bodies compared to exposed heterozygous mice. These results indicate that macrophages can promote repair of damaged synapses after moderate noise trauma and that repair requires fractalkine signaling.


INTRODUCTION
The inner hair cell-spiral ganglion neuron (IHC-SGN) synaptic transmission is excitatory and glutamatergic (Puel, 1995). Glutamate is released from IHCs (Ruel et al., 2008;Seal et al., 2008) and both NMDA-and AMPA-type glutamate receptors are present on SGNs (Eybalin, 1993). The IHC-SGN synapses (a.k.a ribbon synapses) are vulnerable to degeneration due to noise trauma (Puel et al., 1998). Such degeneration has been attributed to glutamate excitotoxicity. Excessive glutamate release due to acoustic trauma can overstimulate postsynaptic glutamate receptors on afferent nerve fiber (ANF) terminals resulting in their swelling, disruption of postsynaptic structures, degeneration of terminals and loss of function. Such synaptic damage can be blocked by glutamate receptor antagonists (Pujol et al., 1985(Pujol et al., , 1993Gil-Loyzaga and Pujol, 1990;Puel et al., 1994Puel et al., , 1998Ruel et al., 2007), while inhibiting glutamate reuptake exacerbates damage (Hakuba et al., 2000). Synaptic degeneration can precede both hair cell loss and threshold elevation and can trigger gradual loss of SGNs (Kujawa and Liberman, 2009). Such primary neural degeneration can affect sound localization and understanding of speech in noisy environments (Liberman, 2017). Moreover, lack of SGNs can limit the effectiveness of primary therapies for hearing loss such as hearing aids and cochlear implants. The denervated IHCs can be partially reinnervated by SGNs after excitotoxic or acoustic trauma; (Pujol et al., 1985;Puel et al., 1995Puel et al., , 1998Pujol and Puel, 1999;Wang and Green, 2011) however, the underlying mechanisms remain unclear.
We previously demonstrated that macrophages promote the survival of SGNs via fractalkine signaling after loss of their target hair cells (Kaur et al., 2015(Kaur et al., , 2018. Notably, the role of macrophages and fractalkine signaling in degeneration and repair of synapses are unknown. The present study characterized the immune response and examined the contribution of fractalkine signaling toward degeneration and repair of damaged synapses after synaptopathic noise trauma. We report that moderate noise trauma caused rapid degeneration of ribbon synapses and immediate macrophage migration into the damaged synaptic region without any evident hair cell loss. The damaged synapses undergo post-exposure spontaneous recovery in animals with intact fractalkine signaling. Notably, disruption of fractalkine signaling diminished synaptic recovery and increased neuronal loss after noise trauma. To our knowledge this is the first evidence of a protective role for macrophages and fractalkine signaling in noise-induced cochlear synaptopathy.

Animals
The study used young adult (6 weeks of age) mice of both sexes on a C57BL/6 (B6) background. To define the role of fractalkine signaling in cochlear excitotoxicity CX 3 CR1 +/+ , CX 3 CR1 GFP/+ , and CX 3 CR1 GFP/GFP mice were employed (Jung et al., 2000). The mice were obtained from Dr. Hirose, Washington University in St. Louis, Missouri, originally obtained from Dan Littmann, New York University, New York. A targeted deletion of CX 3 CR1 and replacement with the gene encoding green fluorescent protein (GFP) rendered all monocytes and macrophages endogenously fluorescent (Jung et al., 2000), which facilitates their visualization through confocal microscope. Cochlear macrophages express CX 3 CR1 (Hirose et al., 2005). The CX 3 CR1 GFP/+ mice (denoted as CX 3 CR1 +/− in the manuscript) with one copy of CX 3 CR1 retain fractalkine signaling, while CX 3 CR1 GFP/GFP mice (denoted as CX 3 CR1 −/− in the manuscript) lack fractalkine signaling. Identification of CX 3 CR1 +/− and CX 3 CR1 −/− followed previously described methods (Jung et al., 2000). Mice were housed in the animal facility at the Central Institute for the Deaf (Washington University School of Medicine) and were maintained on a 12 h/daynight light cycle with open access to food and water. All experimental protocols were approved by the Animal Studies Committee of the Washington University School of Medicine (St. Louis, MO, United States).

Noise Exposures
The study employed mice on B6 background, which is a strain susceptible to noise-and age-related hearing loss (Henry and Chole, 1980;Hequembourg and Liberman, 2001). To induce cochlear synaptopathy, young mice (6 weeks of age) were exposed for 2 h to an octave 8-16 kHz band noise at 90 dB SPL. Noise exposures were performed in a foamlined, single-walled soundproof room from Industrial Acoustics Company (IAC). Fully awake and unrestrained animals were placed singly or in pairs in modified cages (food, water, bedding removed) positioned up to two cages at once directly under an exponential horn. All noise was octave band (8-16 kHz), generated digitally using custom Labview routines (running on a PC) in conjunction with a Tucker-Davis Technologies Rz6 signal processor, and a Crown D-150A power amplifier that drove the speaker.

Auditory Brainstem Response
All auditory brainstem responses (ABRs) were performed by a "blinded" observer. ABRs were analyzed both prior to noise exposure (pre-NE, baseline) and after noise exposure (post-NE) at time 0 h (immediately after noise exposure), 2 and 8 weeks (2 months) recovery. Mice were anesthetized via i.p. injections of ketamine (100 mg/kg) and xylazine (20 mg/kg). Subcutaneous electrodes were placed behind the right pinna (inverting) and vertex (active). A ground electrode was placed near the trail of the mouse. Stimuli were 5-ms tone pips (0.5 ms cos2 rise-fall), delivered at 21/s with alternating stimulus polarity. Recorded electrical responses were amplified (∼10,000X), filtered (300 Hz to 3 kHz) and averaged using BioSig software (Tucker-Davis Technologies, Alachua, FL). The sound level was decreased in 5-dB steps from 99 dB SPL down to 15 dB SPL. At each sound level, 1,024 responses were averaged, and response waveforms were discarded as artifacts if the peak-to-peak voltage exceeded 15 µV. Thresholds at 5, 10, 20, 28.3, 40, and 56.6 kHz were determined by a single observer who noted the lowest sound level at which a recognizable waveform could be obtained. Waveforms were confirmed as auditory-evoked responses by their increasing latency and decreasing amplitude as the intensity of the stimulus was lowered. These threshold values (actual or assigned) were then used to calculate the mean ABR thresholds at each stimulus frequency.

Input/Output Function
For neural response, ABR wave 1 component was identified and the peak to trough amplitudes were computed by off-line analysis of stored ABR waveforms. ABR Wave I amplitudeversus-stimulus level (ABR I/O) data were obtained at 10 and 28.3 kHz. To minimize fatigue, repetition numbers varied from 100 at high sound levels to 1,000 near threshold. Stimuli were ordered from high to low sound levels [from max sound pressure available (∼100 dB) to 5 dB below visual detection of Wave I] in 5 dB steps. Wave I amplitude was measured from the estimated baseline prior to the response to the positive peak of Wave I. ABR wave 1 amplitudes were analyzed both prior to noise exposure (pre-NE, baseline), and after noise exposure (post-NE) at time 0 h (immediately after noise exposure), 2 and 8 weeks (2 months) recovery.

Distortion Product Otoacoustic Emissions
Mice were anesthetized as described above. Stimuli were presented at 5-40 kHz and delivered to the right ear by a customcoupling insert. Distortion product (DP) grams were obtained for f2 ranging from 5 to 40 kHz, with a frequency ratio of f2/f1 of 1.2 and L1-L2 = 10 dB. Recordings were performed using EMAV software (Boys Town National Research Hospital).
DPOAEs were performed by genotype-blinded person prior to noise exposure (pre-NE, baseline) and after noise exposure (post-NE) at time 0 h (immediately after noise exposure), 2 and 8 weeks (2 months) recovery.

Hair Cell Counts
Both inner and outer hair cells were identified by their immunoreactivity for Myosin VIIa. Hair cells were counted from the apical (5, 8, and 11 kHz), middle (16, 22, 28 kHz), basal (32 and 45 kHz) and hook (56 and 64 kHz) region of the cochlea unless otherwise stated. Data are expressed as percentage hair cell survival along the cochlear length.

Macrophage Counts
To assess macrophages per 100 µm of sensory epithelium, GFP-labeled macrophages were counted in the organ of Corti from maximum intensity projections taken from the apical (5, 8, and 11 kHz), middle (16, 22, 28 kHz) and basal (32 and 45 kHz) region of cochlear whole mounts. Macrophages were also counted in the IHC basal region above the habenula using yz optical sectioning and 3D slice features on Volocity 3D image analysis software (version 6.1.1, PerkinElmer) and reported as macrophages in the IHC-basal region per 100 µm of sensory epithelium. Macrophages in spiral ganglia were counted from at least 5-6 mid-modiolar sections per cochlea and normalized to the cross-sectional area of the Rosenthal's canal of the respective cochlear turn and averaged as number per 1,000 µm 2 .

Spiral Ganglion Neuron Counts
Spiral ganglion counts were analyzed at 8 weeks (2 months), 16 weeks (4 months), and 24 weeks (6 months) after noise exposure. To assess the numbers of spiral ganglion cell bodies, NF165 and Tuj-1 labeled somata within Rosenthal's canal were counted from the maximum intensity projections of each section. Cell bodies counted from 5 to 6 sections per cochlea were normalized to the cross-sectional area of Rosenthal's canal per cochlear turn and averaged and reported as SGN density (per 1,000 µm 2 ).

Synaptic Counts
Confocal z-stacks were obtained using a high-resolution oilimmersion objective (63×) from 5, 8,11,16,22,28,32, and 45 kHz regions in each case. Each stack spanned the entire synaptic pole of the hair cells in the z-dimensions, with z-step-size of 0.3 µm, from apical portion of the IHC to nerve terminal in the habenula perforata region. Maximum intensity projection images were exported to ImageJ (Fiji) 1.47b software and converted to black and white images. Juxtaposed pre-and post-synaptic punctae were counted manually from row of 10-12 IHCs along the lengths of the organ of Corti in the apical (5, 8, and 11 kHz), mid-apical and mid-basal (16, 22, 28 kHz), and basal (32 and 45 kHz) regions of each cochlea. Synaptic counts were divided by total number of surviving IHCs in the image and reported as synaptic ribbons per IHC.

Statistical Analyses
All the data analyses and statistics were performed using Prism version 7.0a (GraphPad). Data are presented as mean ± SD. t-test, one way or two-way ANOVA was applied as appropriate. Significance main effects or interactions were followed by appropriate post hoc tests. Details on error bars, statistical analysis, degree of freedom, number of animals, experimental replicates can be found in results and figure legends section of the manuscript. Results were considered statistically significant when probability (p-values) of the appropriate statistical test was less than or equal to the significance level, alpha (α) = 0.05. F values for ANOVAs are reported as F (degree of freedom numerator, degree of freedom denominator) .

Moderate Noise Trauma Induces Temporary Hearing Loss Without Any Hair Cell Death
CX 3 CR1 +/− mice (intact fractalkine signaling) were exposed for 2 h to moderate noise levels at 90 dB SPL. To verify whether such moderate noise exposure caused temporary or permanent hearing loss ABRs and DPOAEs were performed both prior to noise exposure (baseline) and after noise exposure (post-NE) at time 0 h (immediately after noise exposure), 2 and 8 weeks recovery. Noise exposure produced a ∼20-30 dB elevation of neural response thresholds at stimulus frequencies of 20, 28.3, 40, and 56.6 kHz at time 0 h recovery ( Figure 1A). By 2 weeks post-exposure, response thresholds recovered to baseline thresholds and remained stable 8 weeks later (data not shown here). ABR thresholds at time 0 h recovery were significantly different at 20 and 28.3 kHz (p < 0.0001), and at 40 and 56.6 kHz (p = 0.01, F (10,172) = 4.89, two-way ANOVA) when compared to thresholds at baseline and 2 weeks recovery. Elevation in ABR thresholds was coupled with attenuation of DPOAE levels at time 0 h recovery ( Figure 1B) that nearly recovered to baseline DPOAE levels by 2 weeks recovery. DPOAE levels at time 0 h recovery were significantly different from baseline levels at frequencies above 6 kHz (p < 0.0001, F (48,350) = 2.60, two-way ANOVA). DPOAE levels at 2 weeks were significantly different from baseline levels only at highest tested frequency (40 kHz) (p = 0.0046, two-way ANOVA). Moderate noise trauma did not result in loss of outer and IHCs throughout the cochlea at 0 h and 2 weeks after exposure, however, minimal degree of OHC loss was observed in the basal turn of the cochlea at 8 weeks after exposure ( Figure 1C).

Noise Trauma Induces Immediate Macrophage Recruitment Into the IHC-Synaptic Region Despite No Hair Cell Loss
To determine whether moderate noise exposure that does not cause hair cell death is able to activate and increase macrophage numbers in the damaged epithelium, we examined GFP positive macrophages in the sensory epithelium as well as in the inner FIGURE 2 | Macrophage migrate into the epithelium after noise trauma. Representative cochlear whole mount micrographs from CX 3 CR1 +/− mice (A) not exposed to noise (no exposure), (B) at 0 h, and (C) 2 weeks after noise exposure immunolabeled for macrophages (green, GFP) and peripheral nerve fibers (red, NF165 and Tuj-1). (D) Macrophage numbers per 100 µm of sensory epithelium. No exposure (n = 5), 0 h (n = 11) and 2 weeks (n = 7). * p < 0.05, * * p < 0.01, control vs. spiral plexus of CX 3 CR1 +/− mice not exposed to noise (no exposure) and at time 0 h and 2 weeks after TTS-like noise trauma. In unexposed CX 3 CR1 +/− mice, GFP-macrophages are occasionally observed below or in the basilar membrane and also in the osseous spiral lamina away from the inner spiral plexus (Figures 2A,D,E,H). Moderate noise trauma led to an immediate (0 h) increase in macrophage numbers in the apical and middle region of the epithelium ( Figure 2B) and also in the basal region of the IHCs above the habenula (Figures 2F-J). The numbers in the epithelium remained elevated at 2 weeks recovery and also increased in base of the cochlea (Figures 2C,D,H). The macrophage numbers in the epithelium at 0 h recovery were insignificant from no exposure control (p > 0.05, two-way ANOVA) but the numbers in the epithelium at 2 weeks recovery were significantly different from no exposure control in the apex (p = 0.044), middle (p = 0.006) and base (p = 0.045) cochlear region (F (2,55) = 21.93, two-way ANOVA). Macrophage numbers counted in the apex (5-11 kHz), middle (16-28 kHz) and base (32 and 45 kHz) IHC-ANF synaptic region (above the habenula) per 100 µm of the epithelium at 0 h were 2.8 ± 1.8, 4.2 ± 1.3 and FIGURE 3 | Auditory brainstem responses. (A-C) Audiograms of CX 3 CR1 +/+ (black), CX 3 CR1 +/− (blue) and CX 3 CR1 −/− (red) mice not exposed to noise at 6 weeks (A), 8 weeks (B), and 14 weeks (C) of age. n = 5-6 mice per genotype. (D-F) Audiograms of CX 3 CR1 +/+ (black), CX 3 CR1 +/− (blue), and CX 3 CR1 −/− (red) mice exposed to noise at time 0 h (D), 2 weeks (E), and 8 weeks (F) recovery. Filled circles with solid lines represent thresholds prior to noise-exposure (Pre-NE) and unfilled circles with hyphenated lines represents thresholds post-noise exposure (Post-NE). CX 3 CR1 +/+ (n = 7), CX 3 CR1 +/− (n = 9), CX 3 CR1 −/− (n = 15). Gray bar in the graph represents the frequency band (8-16 kHz) of noise exposure.
1.1 ± 0.75, respectively. Macrophage numbers in the IHC-ANF synaptic region at time 0 h recovery was significantly different from no exposure control only in the middle region (p = 0.0002, two-way ANOVA). By 2 weeks macrophage numbers in the apex, middle and base IHC synaptic region were 5.72 ± 0.6, 7.92 ± 1.6, and 3.42 ± 1.1, respectively. Those numbers were significantly different from no exposure control and 0 h recovery in all cochlear regions (p < 0.0001, F (4,58) = 4.67, two-way ANOVA). These results indicate that moderate noise trauma that does not cause hair cell death is sufficient to induce a rapid and localized immune response into the vulnerable and damaged IHC synaptic region.

CX 3 CR1 −/− Mice Exhibit Nearly Complete Recovery of Hearing Thresholds and OHC Function Following Moderate Noise Trauma
We next sought to determine if cochlear macrophages influence auditory function, synaptic degeneration and repair after moderate noise trauma via fractalkine receptor, CX 3 CR1. Unexposed CX 3 CR1 +/+ , CX 3 CR1 +/− , and CX 3 CR1 −/− mice had normal hearing thresholds at 6 weeks of age ( Figure 3A). ABR thresholds were elevated at stimulus frequencies of 40 and 56.6 kHz in all genotypes at 8 weeks ( Figure 3B) and at 14 weeks of age ( Figure 3C) which is likely due to presbycusis in B6 mice. Therefore, any functional or structural data analysis at and above 40 kHz was excluded from the present study. ABR measurements immediately after exposure demonstrated a significant ∼20-30 dB elevation in hearing thresholds at stimulus frequencies above 20 kHz when compared to baseline thresholds in all the three genotypes ( Figure 3D, p < 0.0001, F (25,431) = 3.75, two-way ANOVA). The thresholds at 0 h were not different between the genotypes. The elevated thresholds returned to baseline by 2 weeks post-exposure ( Figure 3E) and remained stable at 8 weeks ( Figure 3F) except at higher frequencies. CX 3 CR1 −/− mice exhibit incomplete recovery of thresholds at 28.3 kHz region (Figures 3E,F) and were significantly elevated from baseline thresholds (p = 0.0056 at 2 weeks post-NE, p < 0.0001 at 8 weeks post-exposure, two-way ANOVA, Tukey's post hoc multiple comparison test).
OHC function assessed by DPOAEs indicated no significant difference among the three genotypes prior to noise exposure ( Figure 4A, filled circles with solid lines, p > 0.999, two-way ANOVA). Noise exposure immediately reduced DPOAE levels in all the genotypes (Figure 4A, unfilled circles with hyphenated lines, p < 0.0001, F (120,1150) = 1.649, two-way ANOVA), when compared to baseline levels, with no significant difference among the genotypes. At 2 weeks after noise exposure, DPOAE levels returned to normal with no difference between the genotypes (Figure 4B). At 8 weeks, the DPOAE levels were reduced in all genotypes at frequencies above 20 kHz, however, the differences were not statistically significant ( Figure 4C). These results demonstrate that lack of CX 3 CR1 does not influence hearing thresholds and OHC function after moderate noise trauma. FIGURE 4 | Distortion product otoacoustic emissions. DP grams of CX 3 CR1 +/+ (black), CX 3 CR1 +/− (blue), and CX 3 CR1 −/− (red) mice exposed to noise at time 0 h (A), 2 weeks (B), and 8 weeks (C) recovery. Filled circles with solid lines represent DP grams prior to noise-exposure (Pre-NE) and unfilled circles with hyphenated lines represents DP grams post-noise exposure (Post-NE). CX 3 CR1 +/+ (n = 8), CX 3 CR1 +/− (n = 9), CX 3 CR1 −/− (n = 9). Gray bar in the graph represents the frequency band (8-16 kHz) of noise exposure.
CX 3 CR1 −/− Mice Display Attenuated ABR Wave 1 Amplitudes Following Noise Trauma ABR wave 1 amplitudes-versus-stimulus level (ABR I/O) analysis at 28.3 kHz frequency region ( Figure 5A) were significantly attenuated in all three genotypes immediately after noise exposure (0 h) compared to genotype matched pre-noise amplitudes (Figures 5B-B , p < 0.01, two-way ANOVA). The attenuated wave 1 amplitudes recovered in CX 3 CR1 +/+ and CX 3 CR1 +/− animals by 2 weeks after exposure and remained stable at 8 weeks (Figures 5C,C ,D,D ). However, CX 3 CR1 −/− mice displayed persistent attenuation in amplitudes (53%) at both thresholds as well as suprathresholds levels (Figures 5C ,D , p = 0.015 and p = 0.0083 at 2 and 8 weeks postexposure, respectively, two-way ANOVA). The raw amplitudes in CX 3 CR1 +/+ , CX 3 CR1 +/− , and CX 3 CR1 −/− mice at 28.3 kHz are provided in Supplementary Table S1 (see Supplementary Material). ABR wave 1 amplitudes-versus-stimulus level (ABR I/O) analysis at 10 kHz frequency region revealed no difference in wave 1 amplitudes with respect to either genotype or recovery time post-noise exposure (data not shown). Together, the results show that exposed mice lacking CX 3 CR1 have attenuated amplitudes of ABR wave 1 when compared to exposed mice with intact fractalkine signaling.

Lack of CX 3 CR1 Impairs Synaptic Repair and Influences IHC Survival After Noise Trauma
To determine whether macrophages influence degeneration and repair of synapses via fractalkine signaling, CX 3 CR1 +/− and CX 3 CR1 −/− mice received a TTS-inducing noise exposure and temporal bones were collected at 24 h-, 2 weeksand 8 weeks-recovery. Cochleae were fixed and processed for immunolabeling of presynaptic zones with an antibody against CtBP2, postsynaptic AMPA receptor subunit GluA3 antibody and for hair cells with anti-Myosin VIIa antibody (Figures 6A-D). There was no difference in ribbon synapses per surviving IHC across the cochlear length in unexposed (control) CX 3 CR1 +/− and CX 3 CR1 −/− mice (Figure 6E). At 24 h after exposure there was a rapid ∼50% loss of synaptic ribbons in both CX 3 CR1 +/− and CX 3 CR1 −/− mice throughout the basal turn of the cochlea (Figures 6B,E), compared to unexposed mice (Figures 6A,E). Two-way ANOVA followed by Tukey's post hoc test revealed that synaptic counts were significantly reduced in exposed mice at the 22.6, and 32.0 kHz regions compared to unexposed control mice (p < 0.01, F (25,208) = 1.79). There was no difference in synaptic counts between CX 3 CR1 +/− and CX 3 CR1 −/− mice at 24 h after exposure ( Figure 6E, p = 0.99, two-way ANOVA). In exposed CX 3 CR1 +/− mice, damaged synapses recovered by 2 weeks postexposure (data not shown) and remained stable at 8 weeks (Figures 6C,E). However, CX 3 CR1 −/− mice displayed persistent synaptic degeneration at the 22-32 kHz region at 2 weeks (data not shown) and 8 weeks after exposure (Figures 6D,E). Synaptic counts per surviving IHCs in CX 3 CR1 −/− mice at 8 weeks post-exposure were significantly different from unexposed CX 3 CR1 +/− and CX 3 CR1 −/− mice (p = 0.013, twoway ANOVA) but not significant from CX 3 CR1 +/− mice at 8 weeks recovery (p = 0.056). Lack of CX 3 CR1 did not affect macrophage numbers in the IHC-ANF synaptic region after exposure. The average macrophage numbers in the IHC synaptic region in exposed CX 3 CR1 +/− mice was 6.55 ± 0.88 and in exposed CX 3 CR1 −/− mice was 6.22 ± 1.98 (Figure 6F, p = 0.654, t = 0.46, Df = 11.4, two-tailed t-test). CX 3 CR1 deficiency also resulted in damage and loss of IHCs from the base of the cochlea of 40% mice (Figure 6H). IHC loss in exposed CX 3 CR1 −/− mice was significantly different from exposed CX 3 CR1 +/− and unexposed control mice (p = 0.0081, F (6,108) = 3.08, two-way ANOVA). Noise-exposed mice had OHC loss in the base and hook regions in all three genotypes (Figure 6G), however the loss was statistically insignificant when compared to unexposed mice (p = 0.144, F (6,108) = 1.67, two-way ANOVA). These results demonstrate that disruption of fractalkine signaling impairs the spontaneous recovery of damaged synapses after moderate noise trauma.

CX 3 CR1 Deficiency Leads to Diminished Survival of SGNs Following Moderate Noise Trauma
Immunolabeling for macrophages using anti-GFP antibody (to enhance the fluorescence signal) in mid-modiolar frozen sections demonstrate that such moderate noise exposure for 2 h to an octave band (8-16 kHz) noise at 90 dB SPL did not result in an increase in macrophage numbers in the spiral ganglion (SG) at any recovery time point (Figures 7A-E). Two-way ANOVA followed by Tukey's post hoc test revealed that macrophage density in the SG after noise exposure at any recovery time point was not significantly different from control or other recovery time points (p = 0.46, F (8,78) = 0.977). To determine whether fractalkine signaling influenced SGN survival after synaptopathic TTS-like noise trauma, cochlear mid-modiolar frozen sections from CX 3 CR1 +/− and CX 3 CR1 −/− mice were immunolabeled for SGNs using NF165 and Tuj-1 antibodies. SGN density in exposed CX 3 CR1 −/− mice was indistinguishable to unexposed CX 3 CR1 −/− and exposed CX 3 CR1 +/− mice at 8 and 16 weeks post-exposure (data not shown). At 24 weeks after exposure, CX 3 CR1 −/− mice displayed increased loss of SGN cell bodies from the basal turn of the cochlea (Figures 7I,J) compared to unexposed CX 3 CR1 +/− and CX 3 CR1 −/− (Figures 7F,G,J) and exposed CX 3 CR1 +/− mice (Figures 7H,J). Two-way ANOVA followed by Tukey's post hoc test revealed that the SGN density in the basal cochlear region of CX 3 CR1 −/− mice was significantly different from unexposed age-matched control mice (p = 0.0034, F (10,57) = 6.22). The SGN density in the exposed CX 3 CR1 +/− mice was not statistically significant from the unexposed agematched controls (p = 0.98). The macrophage density in the SG of CX 3 CR1 −/− mice was not different from that of CX 3 CR1 +/− mice at all recovery time point including 24 weeks after noise trauma ( Figure 7K, p = 0.93, two-way ANOVA). These results demonstrate that disruption of fractalkine signaling due to genetic loss of CX 3 CR1 results in an increase in neuronal loss after moderate synaptopathic-noise trauma.

DISCUSSION
Cochlear synaptopathy can occur due to noise trauma (Kujawa and Liberman, 2009), ageing (Kujawa and Liberman, 2015), FIGURE 5 | ABR Wave 1 amplitudes against level series. (A) Representative ABR trace (28.3kHz, 80 dB SPL) showing placement of arrows at peak and trough of ABR wave 1 for amplitude measurements. ABR wave 1 amplitudes at 28.3 kHz from CX 3 CR1 +/+ (black), CX 3 CR1 +/− (blue), and CX 3 CR1 −/− (red) mice exposed to noise at time 0 h (B-B ), 2 weeks (C-C ), and 8 weeks (D-D ) recovery. Filled circles with solid lines represent amplitudes prior to noise-exposure (Pre-NE) and unfilled circles with hyphenated lines represents amplitudes post-noise exposure (Post-NE). CX 3 CR1 +/+ (n = 7), CX 3 CR1 +/− (n = 9), CX 3 CR1 −/− (n = 15). and aminoglycoside-induced ototoxicity (Ruan et al., 2014). Synaptic loss can precede threshold elevation and can trigger gradual degeneration of SGNs (Kujawa and Liberman, 2009). The mechanisms underlying synaptic damage and subsequent neurodegeneration have not been elucidated beyond the studies of glutamate excitotoxicity (Pujol et al., 1985(Pujol et al., , 1993. It is notable, however, that the ANFs can undergo spontaneous regeneration and can partially recover their synaptic connections with IHCs after excitotoxic insults Pujol and Puel, 1999) or following acoustic trauma (Puel et al., 1998). Whether this represents transient protein down-and up-regulation or actual degeneration and regeneration of synaptic elements remains unclear. The mechanisms of such spontaneous synaptic repair are also unclear. In the present study young B6 mice exposed to 90 dB SPL for 2 h produced a TTS. The elevated thresholds, incomplete recovery of DPOAE levels and some degree of loss of OHCs observed at higher cochlear frequencies at 2 months postexposure recovery is attributed to the early onset of hearing loss in B6 mice (Henry and Chole, 1980;Hequembourg and Liberman, 2001). We report that young B6 mice exposed to 90 dB SPL produces a rapid loss of up to 50% of synapses at cochlear regions tuned to frequencies higher than the exposure band (8-16 kHz). At 2 weeks post-exposure recovery, the damaged synapses undergo nearly complete spontaneous repair. This is in contrast to previous studies that reported permanent loss of synapses following moderate noise trauma that causes reversible threshold elevation (Kujawa and Liberman, 2009). Those studies utilized mice of CBA/CaJ strain at 16 weeks of age. Previous studies in guinea pigs demonstrated post-exposure regeneration of cochlear nerve terminals in the IHC region at the ultrastructural level (Puel et al., 1998). On the contrary, Lin et al., 2011 reported the presence of the irreversible primary neural degeneration in noise-exposed guinea pigs by immunolabeling pre-and postsynaptic markers and applying high power confocal analysis. The spontaneous synaptic recovery that has been reported in the present study may be attributed to the different strain, and age of mice as well as to the noise levels. We have found that young B6 mice exposed to 100 dB SPL causes permanent threshold shifts (PTS) and partial synapse recovery (data not shown). Similar post-exposure synapse recovery in B6 mice have also been demonstrated by Shi et al. (2015) and Kim et al. (2019). Further understanding of the mechanisms of spontaneous synaptic repair in B6 mice would lead to the identification of novel targets to elicit synapse regeneration.
Moderate noise trauma induced a rapid and focal recruitment of macrophages into the damaged synaptic region despite any evident hair cell death. The source and phenotype of macrophages as well as the mechanisms that regulate their recruitment into the synaptic region remain unknown. Studies from the CNS have reported that excitotoxity and/or neuronal injury can lead to microglial activation and chemotaxis toward the site of injury, and various candidate signals that n = 6-7 mice per genotype, 24 h after exposure; n = 5-6 mice per genotype, 8 weeks after exposure; n = 9-15 mice per genotype. * * p < 0.01 (CX 3 CR1 +/− and CX 3 CR1 −/− , no exposure vs. 24 h), #p < 0.01 (CX 3 CR1 −/− , 8 weeks vs. no exposure). Gray bar in the graphs represents the frequency band (8-16 kHz) of noise exposure. (F) Average macrophage numbers in IHC synaptic region per 100 µm of sensory epithelium of exposed CX 3 CR1 +/− and CX 3 CR1 −/− mice, n = 9 per genotype, ns, not significant. Percentage outer hair cell survival (G) and inner hair cell survival (H). Mice per genotype not exposed to noise (n = 6), CX 3 CR1 +/− (n = 9), CX 3 CR1 −/− (n = 15) at 8 weeks after exposure. * * p < 0.01. Scale, 17 µm (A-D).
could participate in microglial activation have been identified. Cytokines such as TNF-α, IL-1β, IFN-γ released from damaged neurons and from reactive astrocytes following excitotoxic events can activate the surrounding microglia (Zhang and Zhu, 2011). Lipid peroxidation (an oxidative stress indicator) plays an important role in neuronal degeneration. 4-hydroxynonenal, a product of lipid peroxidation, can activate and increase the phagocytic activity of microglia and peripheral macrophages through their association with scavenger receptors (Bruce-Keller, 1999). Molecules such as fractalkine have received much attention as microglial chemoattractants (Haskell et al., 1999;Hermand et al., 2008). CNS excitotoxicity can cause rapid cleavage of fractalkine from the neurons, increasing the levels of soluble fractalkine, which is a known chemoattractant (Chapman et al., 2000;Limatola et al., 2005). Nevertheless, our data show that the lack of CX 3 CR1 did not influence macrophage density in the IHC-synaptic region or in the spiral ganglion following moderate noise trauma, suggesting that fractalkine may not be a necessary chemotactic factor. Other molecules such as adenosine triphosphate (ATP), are released from injured cells, and evoke microglial chemotaxis to the injured site, acting via microglia P2Y12 purinergic receptors (Honda et al., 2001;Haynes et al., 2006;Eyo et al., 2014;Kato et al., 2016). Noise is known to stimulate local ATP release in the cochlea (Telang et al., 2010), however, whether ATP evokes macrophage recruitment into the synaptic region is unknown. Macrophages expresses glutamate receptors (Biber et al., 1999), hence it is possible that the excessive glutamate released from the IHCs due to noise trauma can directly induce macrophage chemotaxis toward the synaptic region. Glutamate can directly induce microglial chemotaxis which is mediated by AMPA and metabotropic glutamate receptors on the microglia (Liu et al., 2009). Future studies investigating the source and phenotype of macrophages and mechanisms that regulate their migration FIGURE 7 | Macrophage and Spiral ganglion neuron density in the spiral ganglion of CX 3 CR1 +/− and CX 3 CR1 −/− mice after exposure. Representative micrographs of cochlear mid-modiolar sections from CX 3 CR1 +/− mice not exposed to noise (no exposure) (A) and from CX 3 CR1 +/− mice at 24 h (B), 2 weeks (C), and 8 weeks (D) after exposure immunolabeled for macrophages (green, GFP) and spiral ganglion neurons (red, NF165 and Tuj-1). (E) Macrophage numbers in spiral ganglion (SG) (per 1,000 µm 2 ), n = 4-8 CX 3 CR1 +/− mice per recovery time point. Representative micrographs showing fewer SGN cell bodies immunolabeled for NF165 and Tuj-1 (red) in CX 3 CR1 −/− mice (I) compared to CX 3 CR1 +/− mice (H) at 24 weeks post-exposure and to unexposed age-matched control mice (F,G). (J) Quantitative data on SGN density per 1,000 µm 2 from age-matched unexposed and exposed CX 3 CR1 +/− and CX 3 CR1 −/− mice at 24 weeks post-noise trauma. n = 4 mice per group, * * p < 0.01 CX 3 CR1 −/− mice 24 weeks post-NE vs. no exposure CX 3 CR1 +/− and CX 3 CR1 −/− mice. (K) Average macrophage density in SG of CX 3 CR1 +/− and CX 3 CR1 −/− mice at time 24 h, 2, 8, 16, and 24 weeks recovery after exposure. n = 4-9 per genotype per recovery time point. Scale bar, 63 µm.
into the damaged synaptic region after noise trauma would be valuable to better understand the inflammatory cells dynamics in cochlear pathology.
Our data demonstrate that the genetic disruption of fractalkine signaling impairs the spontaneous recovery of damaged synapses and leads to enhanced degeneration of auditory neurons following synaptopathic noise trauma. The enhanced neurodegeneration in the damaged cochlea of CX 3 CR1 knockout animals corroborates our previous work (Kaur et al., 2015(Kaur et al., , 2018. The lack of synaptic recovery in the exposed CX 3 CR1 −/− mice is the first evidence for the role of macrophages and fractalkine in synaptic plasticity in the damaged cochlea. Increased synaptic degeneration in CX 3 CR1 −/− mice correlates with attenuated suprathreshold neural responses at higher frequencies without affecting hearing thresholds and DPOAE levels. Lack of CX 3 CR1 did not affect macrophage density in the IHC-ANF synaptic region or in the ganglion after exposure suggesting that the increased synapse degeneration and neuronal loss in exposed CX 3 CR1 −/− mice is not due to reduced macrophage numbers. The underlying mechanisms of synaptopathy in the CX 3 CR1 knockout animals is unclear, but could be driven, at least in part, by IHC pathology that is observed in a few (40%) CX 3 CR1 −/− mice. The exact contribution of macrophages in synaptic recovery is also not yet clear. An intriguing possibility is that macrophage migration into the damaged synaptic region is a protective mechanism designed to limit neurodegeneration and improve synaptic recovery following noise trauma. In support of this idea, serval studies in the CNS have suggested that microglial activation can attenuate excitotoxic injury and limit neurodegeneration (Simard and Rivest, 2007;Lauro et al., 2010;Vinet et al., 2012;Eyo et al., 2014;Kato et al., 2016) and improve synaptic recovery (Lazarov-Spiegler et al., 1996;Prewitt et al., 1997;Batchelor et al., 1999). Microglia-mediated protection against CNS excitotoxicity has been attributed to multiple mechanisms such as promoting neurite outgrowth (Prewitt et al., 1997;Batchelor et al., 1999), phagocytosing degenerating neurons and synapses (Abiega et al., 2016), and by promoting the release of neurotrophic factors such as basic fibroblast growth factor (bFGF) and nerve growth factor (NGF) (Heumann et al., 1987;Araujo and Cotman, 1993).
How might fractalkine signaling influence excitotoxicity and repair of cochlear synapses? One hypothesis is that fractalkine signaling regulates macrophage inflammatory and toxic behavior in damaged cochlea. Studies in the CNS have shown that fractalkine signaling exert neuroprotective actions in many neuroinflammatory and neurodegenerative disease models and can also prevent neuronal damage during glutamate excitotoxity (Cardona et al., 2006;Limatola and Ransohoff, 2014). The protective effect of fractalkine against glutamate excitotoxicity appears to involve a reduction in NMDA-or glutamate-mediated rise in intracellular calcium levels (Deiva et al., 2004;Sheridan et al., 2014), reduced AMPA type mediated currents (Meucci et al., 1998;Limatola et al., 2005;Ragozzino et al., 2006), increased glutamate removal from synaptic cleft by enhancing the expression of function of glutamate transporter-1 (GLT-1) on astrocytes (Catalano et al., 2013), and increased microglial adenosine release (Lauro et al., 2008;Cipriani et al., 2011). In addition, CX 3 CL1 can also modulate the clearance and phagocytic activity of microglia (Noda et al., 2011) that could be involved in the neuroprotective effects of CX 3 CL1 in CNS disorders. Mice lacking CX 3 CR1 develop increased disease severity in animal models of experimental autoimmune encephalomyelitis (EAE), low-endotoxemia, Parkinson's disease, Amyotrophic lateral sclerosis, and Diabetic retinopathy (Cardona et al., 2006(Cardona et al., , 2015Mendiola et al., 2017). To a certain degree, such pathology is attributed to enhanced microglial expression of inflammatory cytokines like IL-1β, TNF-α, and IL-6. Our previous work reported an increased expression of IL-1β in the injured cochlea of mice lacking CX 3 CR1 (Kaur et al., 2018). Future efforts toward understanding the mechanisms of fractalkine-mediated synaptic repair and neuroprotection may lead to therapeutic strategies for regeneration of synapses and neuroprotection in the injured cochlea.
Several animal studies have shown that neurotrophin therapies can partially regenerate damaged synapses and restore hearing after noise trauma (Wan et al., 2014;Suzuki et al., 2016;Chen et al., 2018). However, such partial effectiveness clearly demonstrates an additional need to understand the cellular and molecular mechanisms of neurite outgrowth and synapse regeneration to develop therapies to fully restore hearing. Our study has identified a novel endogenous immune pathway that may promote synaptic repair and neuron survival during hearing loss. Future studies investigating the contribution of activation of fractalkine signaling in degeneration and repair of cochlear synapses and determining the function of each fractalkine isoform (membrane or soluble) are necessary to completely understand the role of fractalkine in cochlear synaptopathy and neuropathy. Our research work also poses important clinical relevance for humans carrying the polymorphic variant CX 3 CR1 I249/M280 that is estimated to be present in about 20-30% of the population. These changes in human CX 3 CR1 decrease CX 3 CL1 affinity (McDermott et al., 2001) and have been associated with multiple neurodegenerative disorders such as age-related macular degeneration (Chan et al., 2005;Schaumberg et al., 2014), Alzheimer's disease (Lopez-Lopez et al., 2017), and multiple sclerosis (Arli et al., 2013;Cardona et al., 2018). Based on the evidence of enhanced synaptic and neuronal damage in the injured cochlea of CX 3 CR1deficient mice, it would be of clinical relevance to dissect the effects of human CX 3 CR1 V249/T280 receptor and its polymorphic variant CX 3 CR1 I249/M280 in deaf ears. Such studies may merit evaluation as risk factors for susceptibility to sensorineural hearing loss. Together, our data suggest a pivotal role of CX 3 CL1-CX 3 CR1 signaling in repair of damaged synapses and neuron survival after noise trauma. Thus, modulating fractalkine signaling may be a relevant approach to mitigate cochlear synaptopathy and neuropathy.

DATA AVAILABILITY
The datasets generated for this study are available on request to the corresponding author.

ETHICS STATEMENT
This study was carried out in accordance with the recommendations of the Animal Welfare Act and Animal Welfare Regulations issued by the United States Department of Agriculture (USDA). All experimental protocols were approved by the Institutional Animal Care and Use Committee of the Washington University School of Medicine (St. Louis, MO, United States).

AUTHOR CONTRIBUTIONS
TK and KO designed and planned the study. TK, AC, AN, and KO performed the experiments and analyzed the data. AS performed the mouse genotyping. TK, MW, and KO contributed to the writing and editing of the manuscript. All authors read and approved the submitted version of the manuscript.

FUNDING
This work was supported by the National Institutes of Health and the National Institute on Deafness and Other Communication Disorders Grants R03DC015320 to TK and R01DC006283 to MW. This work was a part of the R03 grant funded to TK. Funds from R01DC006283 (MW) partly supported mouse breeding, maintenance, and genotyping.