Contribution of macrophages to intracochlear tissue remodeling responses following cochlear implantation and neural survival

Introduction: Cochlear implants (CIs) restore hearing to deafened patients. The foreign body response (FBR) following cochlear implantation (post-CI) comprises an infiltration of macrophages, other immune and non-immune cells, and fibrosis into the scala tympani; a space that is normally devoid of cells. This FBR is associated with negative effects on CI outcomes including increased electrode impedances and loss of residual acoustic hearing. This study investigates the extent to which macrophage depletion by an orally administered CSF-1R specific kinase (c-FMS) inhibitor, PLX-5622, modulates the tissue response to CI and neural health. Materials and methods: 10–12-week-old CX3CR1+/GFP Thy1+/YFP mice on C57Bl6 background with normal hearing were fed chow containing 1200 mg/kg PLX5622 or control chow for the duration of the study. 7-days after starting the diet, 3-channel cochlear implants were implanted ear via the round window. Serial impedance and neural response telemetry (NRT) measurements were acquired throughout the study. Electric stimulation began 7 days post-CI until 28- days post-CI for 5 hrs/day, 5 days/week, with programming guided by NRT and behavioral responses. Cochleae harvested at 10-, 28- or 56-days post-CI were cryosectioned and labeled with antibody against α-smooth muscle actin (α-SMA) to identify myofibroblasts and quantify the fibrotic response. Using IMARIS image analysis software, the outlines of scala tympani, Rosenthal canal, modiolus and lateral wall for each turn were traced manually to measure region volume. Density of nuclei, CX3CR1+ macrophages, Thy1+ spiral ganglion neuron (SGN) numbers and ratio of volume of α-SMA+ space/volume of scala tympani were calculated. Results: Cochlear implantation in control diet subjects caused infiltration of cells, including macrophages, into the cochlea: this response was initially diffuse throughout the cochlea and later localized to the scala tympani of the basal turn by 56-days post-CI. Fibrosis was evident in the scala tympani adjacent to the electrode array. Mice fed PLX5622 chow showed reduced macrophage infiltration throughout the implanted cochleae across all timepoints. However, scala tympani fibrosis was not reduced relative to control diet subjects. Further, mice treated with PLX5622 showed increased electrode impedances compared to controls. Finally, treatment with PLX5622 decreased SGN survival in implanted and contralateral cochleae. Discussion: The data suggest that macrophages play an important role in modulating the intracochlear tissue response following CI and neural survival.


Introduction
Cochlear implants (CIs) provide auditory rehabilitation to individuals with moderate to severe sensorineural hearing loss. The device has undergone tremendous technological advancements to broaden the range of candidacy including those at the extremes of age, individuals with residual hearing or unilateral hearing loss, and patients with auditory neuropathy spectrum disorder. (Roche & Hansen, 2015) Improved surgical techniques have led to a reduction in insertion trauma and electrode array translocation. (Ishiyama, Ishiyama, Lopez, & Linthicum, 2019). While modern surgical techniques and biocompatible implant materials enable a high rate of long-term device function, an intracochlear tissue response in the form of a foreign body response (FBR) or hypersensitivity reaction to the CI electrode array has been widely documented (Nadol, & Nadol, 2007) This tissue response largely occurs within the scala tympani, a space that is normally devoid of cells making it unique to most other instances of FBR that occur in cellular tissues. Rahman et al. recently reviewed the current understanding of the in ammatory FBR following cochlear implantation, its impact on implant function in human subjects and animal models, and the emerging mitigation strategies for this deleterious response following cochlear implantation. (Rahman et al., 2022) Histopathologic studies on human cochleae from CI recipients reveal an intrascalar tissue response comprised of densely organized brotic tissue and new bone growth (Seyyedi & Nadol, 2014) with variable severity occurring in a majority of cases. (Nadol et al., 2008). This post-CI FBR is associated with signi cant detrimental consequences including loss of spiral ganglion neurons (SGNs), poorer auditory function (Ishiyama et al., 2019), late onset loss of residual low frequency acoustic hearing (Quesnel et al., 2016), increased electrode impedances (Shaul et al., 2019;Tykocinski, Cohen, & Cowan, 2005), loss of acoustic hearing (Scheperle et al., 2017), poor word recognition scores (Kamakura & Nadol, 2016), and, in rare cases, late onset device failure (Nadol et al., 2008). While the FBR to electrode array biomaterials appears universal in human cochleae, it is exacerbated by traumatic insertion (e.g., electrode array translocation from scala tympani into scala vestibuli or media or damage to the lateral wall of the scala tympani).
Similar to humans, histological evidence for post-CI FBR occurs in animal models and is accompanied by loss of residual hearing, hair cells and SGNs. (Rahman et al., 2022) Hearing loss occurs early after implantation in guinea pigs and is followed by a partial recovery, which may be limited by the FBR. (Zhang, Stark, & Reiss, 2015) (Shepherd, Matsushima, Millard, & Clark, 1991). A correlation between brous tissue growth and sensory hair cell loss has been shown in cat (Clark, Shute, Shepherd, & Carter, 1995), guinea pigs (O' Leary et al., 2013), and macaques (Shepherd, Clark, Xu, & Pyman, 1995). Likewise, SGN loss is associated with electrode insertion trauma and in ammation in hearing cats (Xu, Shepherd, Millard, & Clark, 1997). The electrically evoked compound action potential (ECAP) re ects the synchronous response of auditory nerve bers upon electrical stimulation; ECAP amplitude growth function is used to assess the SGN population health. CI insertion trauma and the FBR appear to cause SGN dysfunction (P ngst et al., 2015; Schvartz-Leyzac et al., 2020) and, correspondingly, brosis and new bone formation are correlated with elevated ECAP thresholds in guinea pigs. (Simoni et al., 2020) Further, a correlation between the extent of the FBR after cochlear implantation and electrical impedance changes has been documented in guinea pig (Wilk et al., 2016), kitten (Ni et al., 1992), cat Xu et al., 1997), and macaque models .
Macrophages have been identi ed in implanted human cochleae using antibodies to the markers CD163, Iba1, and CD68 (O'Malley, Nadol, & McKenna, 2016). These have been shown to phagocytize platinum and silicone from the electrode arrays (Nadol, O'Malley, Burgess, & Galler, 2014). 'Activated' macrophage are present within the brotic sheath surrounding the electrode arrays (Okayasu, Quesnel, O'Malley, Kamakura, & Nadol, 2020) and demonstrate increased responses in cases of translocation of the electrode array and damage to lateral wall (Noonan et al., 2020). In the mouse model, monocyte/macrophage (F4/80 positive cells) in ltrate in the cochlea in an apparent biphasic pattern: an early (3 days post implantation) and late (14-28 days post-implantation) peaks. (Bas et al., 2015;Claussen et al., 2022) Macrophages can drive pro-in ammatory and pro-healing responses; their contribution to the unique FBR that develops within the cochlea remains unkown. Beyond cellular in ltration, a brotic tissue response develops within the scala tympani after CI, evident by deposition of alpha-SMA positive cells and type I collagen. (Bas et al., 2015) Given the deleterious consequences of the FBR, various strategies have been explored to mitigate this response. One such approach is the use of systemic or locally delivered glucocorticoids such as dexamethasone. Compared with standard CIs, electrode arrays that elute dexamethasone decrease brosis, bone growth and electrical impedances, protect hair cells, and help preserve auditory function after implantation without affecting SGN density in guinea pigs. ( One limitation of these approaches is that they use non-speci c immunosuppressive agents that impact on a variety of immune cell types and cytokines. While macrophages comprise a major immune cell type involved in FBR post CI (Claussen et al., 2022) other immune cell types including T and B lymphocytes, cytokines (CXCL1, IL-1β, TNF-α) cell adhesion molecules (ICAM-1), connective tissue growth factor (CTGF), tissue remodeling proteins (TGF-β, MMP2, MMP9) are also involved in the FBR post-CI. (Rahman et al., 2022) While non-selective anti-in ammatory drugs help mitigate the FBR post-CI and improve functional outcomes, these agents preclude investigation of the contribution of speci c cells or cytokines to the FBR.
In this study we focused on determining the role of macrophages and the innate immune response to the FBR post-CI and to neural survival. To this end, we used the colony stimulating factor 1 receptor (CSF-1R) inhibitor, PLX-5622, to deplete macrophages in a CX3CR1 +/GFP reporter mouse model of cochlear implantation. CSF1R is activated by 2 ligands, colony-stimulating factor-1 (CSF-1) and interleukin-34 (IL-34), and plays a critical role in development of microglia and most tissue macrophages. (Stanley & Chitu, 2014). In selecting a CSF1R inhibitor, we considered several factors. First, the inhibitor needs to penetrate blood-labyrinth barrier and second, post-CI FBR is a chronic in ammatory condition where macrophages are involved for extended period. Considering these factors, a highly selective CSF1R inhibitor, PLX5622 (PLX), that can cross the blood brain barrier and allows extended elimination of macrophages was employed in this study (Spangenberg et  . Therefore, we tested whether this dose (1200ppm) and duration (7days) is su cient to eliminate cochlear macrophages. At day 7, all mice were euthanized for histopathologic examination and confocal imaging. The timeline for this experiment is shown in Fig. 1 (Fig. 1a).
After con rmation of cochlear macrophage depletion with PLX5622 chow in nonimplanted mouse cochlear implantation on 2 groups of CX3CR1 +/GFP Thy1 +/ YFP mice. In PLX-5622 group (PLX), cochlear macrophages were depleted with oral administration of PLX5622 (1200ppm for 7 days). In 'control' (No PLX) group, CX3CR1 +/GFP Thy1 +/ YFP mice were fed with 'control chow' for 7 days before CI surgery. On day 7, CI surgery was performed on both groups. Following recovery from surgery, mice from both groups were continued with respective chow (PLX or No PLX) until they were sacri ced at 10-, 28-and 56-days post implantation. The timeline for this experiment is presented in Fig. 2 (Fig. 2a Immediately following surgery, impedance and NRT thresholds were measured and repeated at least weekly thereafter. Electrodes within compliance limits (de ned as having electrical impedance ≤ 35kOhms) were shorted together during electrical stimulation using a software patch. The hardware system for electrical stimulation consisted of a modi ed rodent housing with a sliding tether connected to a CI emulator (CIC4 implant emulator, Cochlear Ltd., AUS) which was activated by interfacing the receiver coil with a commercial CI sound processor (Cochlear Ltd., AUS). Starting on post-operative day 7, mice placed within and connected to the aforementioned system were stimulated for 5 h per day, 5 days a week, programmed to 30CL below NRT threshold with a dynamic range of 1CL between threshold and comfort levels. and it becomes di cult to provide consistent stimulation across each group.

Immunohistochemistry
Under anesthesia with ketamine (80mg/kg) and xylazine (10mg/kg), mice were exsanguinated with transcardial perfusion of ice-cold Phosphate Buffer Solution (PBS) followed by transcardial perfusion of 4% paraformaldehyde (PFA) at the nal respective timepoint for each subject. Cochleae were harvested and were xed overnight with 4% paraformaldehyde at 4°C in dark. Excess PFA was removed with PBS in a rotator overnight. Decalci cation of cochlea was done in 0.1M EDTA (pH 7.5) solution on a rotator that is changed every day for 3-5 days. After decalci cation, cochlea washed with PBS 3 times, 10 minutes each time. Decalci ed cochleae were cryoprotected using serial concentration of sucrose solutions starting at 10% sucrose solutions and increasing concentration by 10% every hour, nally reaching 30%. Cryoprotected mice cochleae were stored at -20 C until sectioned. Cochlea infused with O.C.T. embedding medium (Tissue-TEK) were then mounted to the stage of sliding block microtome (American Optical 860) stage with O.C.T. and dry ice. Mounted cochleae were sectioned parallel to the mid-modiolar plane at 30µm thickness, sections were placed on Fisher Superfrost slides, and stored at − 20°C until immunolabeling was performed. For immunolabeling, slides were rst warmed to room temperature (~ 20-22°C), washed (3 x 5 min each wash) with 'washing buffer' containing PBS, 0.03% Triton X-100 and 0.1% Tween-20. Then samples were permeabilized and blocked in 'blocking buffer' (0.03% Triton X-100 and 0.1% Tween-20, 1% bovine serum albumin (RPI, CAS#9048-46-8) and 0.02% sodium azide (Sigma, catalog #S2002) in PBS for 2 hours. Following blocking, sections on slides were incubated in primary antibody (Alpha-smooth muscle actin monoclonal antibody, 1A4, eBioscience ™ , Catalog# 14-9760-82 in 'blocking buffer' overnight (~ 16 h) at 4°C. After primary antibody application, sections were washed (3 x 5 min) in 'washing buffer', then incubated in blocking buffer containing secondary antibodies (Alexa Fluor conjugates, 1:400, Invitrogen) for 2-hrs at room temperature. Sections were then washed 3 x 5 min in 'washing buffer'. Nuclei were stained with Hoechst 3342 (10 µg/ml in PBS, Sigma) for 20 min at room temperature, followed by washing with 'washing buffer' (3 x 5 min) and cover slipped with Fluoro-Gel Mounting Medium with Tris Buffer (catalog #17985-10, Electron Microscopy Sciences).

Statistical analyses
Statistical analyses for impedance measurements, counts of immune cells, nuclei, neurons, and volume of brotic tissue within scala tympani were performed using R version 4.3.0 (2023-04-21)("R Core Team," 2021). Speci c comparisons that were made are described in respective gure legends. General linear models were t to assess main effects of group, day and their interaction, except in impedance data where non-linear trends were expected respective to time and a linear mixed model was t. Shapiro-Wilk or D'Agostino-Pearson test was used to determine normality of data. If model assumptions were not met, simpler models that excluded outlier groups were t for parametric tests. For parametric data, pairwise comparisons of the least square means were made with Tukey adjustment for multiple comparisons. Additionally, non-parametric tests (e.g., Kruskal-Wallis Test) were performed Using Wilcoxon Rank scores to compare all groups in data that were not normally distributed. Results of the statistical analysis are included in the text; as a regression analysis approach was used where comparisons are made on the regression models, no statistical annotations are included on the gures. Signi cance was de ned as p < 0.05).
Orally administered PLX5622 effectively reduced the resident macrophage population throughout the cochlea for the study duration in unimplanted cochleae compared to No PLX (pairwise comparisons of No PLX and PLX unimplanted groups on least squares mean effect of group using Tukey adjustment, p < 0.05 in all areas except scala tympani of base, p = 0.4295, where little to no macrophage in ltration was seen in either group). Likewise, oral administration of PLX5622 reduced the in ltration of CX3CR1positive cells throughout implanted cochlea at all timepoints compared to No PLX implanted (Kruskal-Wallis Test on Wilcoxon Rank scores of non-parametric data, p < 0.0001 in all cases, a parametric model could not be t to include No PLX implanted) (Fig. 3).  3.6. PLX5622 causes SGN degeneration Figure 7 shows mean density of Thy1 YFP + SGNs for the 'No PLX' and 'PLX' groups overtime. First, we considered the change in SGN density over time in the base (signi cant effect of day in parametric model, p < 0.0001). In the spiral ganglion of the base of the cochlea, SGN density at day 28 is signi cantly lower than that of day 10 (pairwise comparison on least squares means for effect of day with Tukey adjustment, p < .0001). The difference between day 10 and 56 was also signi cant (pairwise comparison on least squares means for effect of day with Tukey adjustment, p = .0011). The difference between day 28 and day 56 was non-signi cant. Similarly, SGN degeneration is observed in middle turn of cochlea from day 10 to day 28 (signi cant effect of day in parametric model, p = 0.0004; pairwise comparison on least squares means for effect of day with Tukey adjustment, p = .0003). In the apical turn, density at day 28 is signi cantly lower from day 10 (signi cant effect of day in parametric model, p = 0.0120; pairwise comparison on least squares means for effect of day with Tukey adjustment, p = 0.0148) The difference between day 10 and 56 in apical neuron density was marginally signi cant (p = 0.0503), but the difference between day 28 and day 56 was non-signi cant (p = 0.96). These experiments are done on mice with B6 background and suggest that signi cant SGN degeneration happens in these mice between 3 and 4 months of age. Next, we considered the effect of cochlear implantation on SGN density. Following cochlear implantation, we did not observe evidence for SGN degeneration in cochleae compared to respective PLX and No PLX unimplanted groups (signi cant effect of group in parametric model of apical, middle and basal neuron density, p = 0.0001-0.0051; pairwise comparison on least squares means for effect of day with Tukey adjustment, p > 0.05).

PLX5622 increases electrode impedance
We also considered whether PLX treatment causes degeneration of SGNs in the unimplanted condition. Compared to 'No PLX mice, in 'PLX mice, SGN density was signi cantly lower in contralateral (unimplanted) cochlea at the basal turn (p = .0053), middle turn (p = .0002) and apical turn (p = .0148) (signi cant effect of group in parametric model of apical, middle and basal neuron density, p = 0.0001-0.0051; pairwise comparison on least squares means for effect of day with Tukey adjustment). We further analyzed the impact of PLX treatment in implanted cochlea. In implanted cochlea, PLX treatment is associated with SGN degeneration in the base only (p < .0001 signi cant effect of group in parametric model apical, middle and basal neuron density, p = 0.0001-0.0051; pairwise comparison on least squares means for effect of day with Tukey adjustment). After looking at pairwise comparisons in the group x day interaction, we found that the only signi cant differences between treatment groups are in day 10 and 28.
At day 10 post-CI, the mean for PLX-CI is signi cantly lower than no-PLX CI (p = 0.0202) and at day 28, PLX CI is signi cantly lower than No PLX CI (p = .0487); pairwise comparison on least squares means for day x group interaction with Tukey adjustment. Thus, PLX5622 administration was associated with SGN degeneration irrespective of cochlear implantation status.

Discussion
Our data suggest that PLX5622, a speci c inhibitor of CSF1R, at a dose of 1200 mg/kg for 7 days eliminated almost all cochlear mononuclear phagocytes. Following cochlear implantation, a cellular in ltration, including macrophages, with brotic tissue deposition occurs adjacent to the electrode array in the basal scala tympani and was associated with increased electrode impedance. When cochlear implantation was performed in mice with ongoing PLX 5622 macrophage depletion, cellular in ltration (including macrophage in ltration) was inhibited but the volume of brotic response was not. Electrical impedance following cochlear implantation trended higher in PLX5622 treated group. Moreover, PLX5622 treatment was associated with degeneration of SGNs in base of the cochlea independent of cochlear implantation.
With short term (7 days  PLX5622 not only depleted resident CX3CR1 + cells prior to cochlear implantation, it also caused sustained depletion of the in ltrating CX3CR1 + cells after placement of the electrode array. To the best of our knowledge, this is the rst study to explore the role of CSF1R inhibition in a cochlear implant model. Studies on brain implants demonstrated similar effects on brain microglial population. (Sharon, Jankowski, Shmoel, Erez, & Spira, 2021) They have shown that although PLX5622 treatment depletes microglia from the rat brain, astrocytes encapsulate the neuro-implant suggesting that microglia are redundant for this FBR in the brain. The reduction in cellular density in the scala tympani of PLX treated mice following CI might be a direct effect of elimination of resident and in ltrating macrophage population. Also, in the spiral ganglia, we observed degeneration of SGNs that can contribute to the decline in cellular density within spiral ganglia. Moreover, macrophages also secrete growth and angiogenic factors. One signi cant nding of these experiments is that the brotic response, as measured by anti-αSMA immunolabeling, was not signi cantly reduced by PLX5622 treatment. These results mirror other studies that explored the role of CSF1R inhibition with PLX5622 on the FBR to neuro-implants in the brain. Following CI, there is a gradual rise in electrode impedances consistent with an evolving tissue response. PLX 5622 treatment lead to a more rapid rise in electrical impedance compared to No PLX. As PLX5622 treatment reduces cellular in ltration into cochlea, it appears that reducing cellular in ltration alone is not su cient to prevent the rise in electrical impedance associated with the FBR. Further the extent of brosis, as measured by anti-αSMA immunolabeling, is not affected by PLX5622 treatment suggesting that the brotic tissue might be the factor maintaining the high electrical impedance in PLX5622 treated implanted cochlea. Moreover, electrode impedances in mice treated with PLX5622 rose more rapidly than the impedances in control mice raising the possibility that there are functional differences in the nature of the brotic response in the absence of macrophages. Post-implantation cochlear brosis is often accompanied by neo-ossi cation in humans and mice. The current study methods employed decalci cation for histological preparation, prohibiting assessments of cochlear neo-ossi cation after implantation. This aspect is important for future studies, as CSF1R inhibition is associated with alterations in osteoclast activity that could impact post-CI neo-ossi cation and differentially effect electrode impedance compared to the less dense, non-mineralized brotic tissue. (Brun et al., 2020) Degeneration We would like to mention a potential limitation of the method of SGN quanti cation that we used. We observed variation in Thy1-driven YFP expression among the SGN population. Therefore, use of Thy1-reporter expression as a marker for SGN might present issues with reliability. Moreover, the sensitivity of Thy1-reporter as a marker for SGN is not currently known.    Quanti cation of CX3CR1+ macrophage density in cochlea following cochlear implantation. Cochlear implantation was performed in CX3CR1 +/GFP Thy1 +/ YFP mice, fed on chow with 1200ppm of PLX-5622 (PLX) or control chow (No PLX) for 7 days. Following surgery, mice were continued with respective chow (PLX or No PLX). Starting on post-operative day 7, mice within stimulation cages, connected to the CI processor were stimulated for 5 h per day, 5 days a week. Mice were euthanized at 10-, 28-and 56-days post-CI and imaged as in Figure 3. Image analysis was performed in IMARIS image analysis software. In 30µm thick midmodiolar sections, CX3CR1+ macrophage cells were counted on maximum intensity zprojections of 3D confocal image stacks. The outline of Rosenthal's canal (RC) and lateral wall at base, middle and apex of cochlea, scala tympani of the base of the cochlea and modiolus were traced and volume of each area was measured. CX3CR1+ macrophages were counted using automated counting system in IMARIS image analysis software aided by custom made macros. The density of macrophages with visible, Hoechst+ nuclei was calculated per 10 5 µm 3 in each area mentioned. Values derived from every region of cochlea for an individual animal were averaged together from 3 sections; "n" is the total number of mice used in the study. Error bars indicate SEM.

Figure 4
Quanti cation of cellular density into scala tympani of the base of cochlea following cochlear implantation. Cochlear implantation was performed in CX3CR1 +/GFP Thy1 +/ YFP mice, fed on chow with 1200ppm of PLX-5622 (PLX) or control chow (No PLX) for 7 days. Treatment with respective diets and electrical stimulation was continued until the desired endpoints (10, 28 or 56 days). Nuclei were labeled with Hoechst 3342 in 30µm thick mid-modiolar sections. Image analysis was performed in IMARIS image analysis software. Hoechst+ cells were counted on maximum intensity z-projections of 3D confocal image stacks in scala tympani of the base of the cochlea. Volume of scala tympani and nuclear counts were made using automated counting system in IMARIS image analysis software aided by custom made macros. Nuclear density (Hoechst+ cells) was calculated per 10 5 µm 3 . An average of 3 sections per animal was taken with 'n' being the number of animals. Error bars indicate SEM.

Figure 5
Quanti cation of SMA+ tissue within scala tympani of the base of the cochlea following cochlear implantation. Following 7-day feeding on chow with 1200ppm of PLX-5622 (PLX) or control chow (No PLX), cochlear implantation was performed in CX3CR1 +/GFP Thy1 +/ YFP mice, Respective diets were resumed following recovery from surgery. Electrical stimulation was continued until 28-day post-CI. Mice were euthanized at the desired endpoints (10, 28 or 56 days). Following euthanasia, 30µm midmodiolar sections were labeled with anti-alpha SMA antibodies. The volume of the scala tympani and SMA+ tissue volumes were measured using IMARIS image analysis software. Fibrosis was measured by dividing the volume of SMA+ tissue by volume of scala tympani, expressed in % volume. Error bars indicate SEM.  Quanti cation of spiral ganglion neuron density following cochlear implantation. 7-day feeding on chow with 1200ppm of PLX-5622 (PLX) or control chow (No PLX) was followed by cochlear implantation in CX3CR1 +/GFP Thy1 +/ YFP mice. After recovery from surgical anesthesia, respective diets were resumed. Electrical stimulation was done as described before. Cochlea harvested at desired endpoints (10, 28 or 56 days) were sectioned at 30µm thickness. After measurement of spiral ganglia volume and quanti cation