Animals were euthanized either by quick decapitation or deep anesthesia (Fatal Plus, 50 mg/kg). Temporal bones were isolated and fixed overnight with 4% paraformaldehyde (in PBS) at 4°C. After fixation, cochleae were isolated and washed 3x for 5 min each in PBS, pH 7.4 at room temperature. All specimens, except those collected at P5, were then incubated for ∼24 h in 10% EDTA in PBS, pH 7.4 at room temperature. Specimens were thoroughly washed 3x for 5 min each in PBS, pH 7.4 at room temperature and were then dissected into whole mount samples. Other samples were prepared as frozen mid-modiolar sections (30 µm thickness). Those cochleae were incubated in 10% EDTA in PBS, pH 7.4 at room temperature for 3–5 days and then washed 3x for 5 min each in PBS, pH 7.4 at room temperature. Cochlear whole mounts and mid-modiolar sections were immunolabel for GFP-expressing macrophages (rabbit anti-GFP antibody, catalog #A11122, Invitrogen, 1:500 or chicken anti-GFP antibody, catalog #1010, Aves, 1:500), neurons were labeled with mouse monoclonal anti-β-III tubulin (catalog #MMS-435P, Covance, 1:500), combined with mouse anti-Neurofilament (Catalog #2H3, Developmental Studies Hybridoma Bank, 1:100) antibodies, and apoptotic cells were labeled using rabbit anti-cleaved caspase-3 antibody (catalog # 9,661, Cell signaling, 1:100). To avoid non-specific binding of the antibodies, samples were preincubated in a blocking solution consisting of 5% normal horse serum/0.2% Triton X-100 in PBS for 1 h at room temperature. Samples were then incubated overnight with primary antibodies prepared in PBS with 2% normal horse serum and 0.2% Triton X-100 at room temperature. Samples were washed 3x for 5 min each in PBS, pH 7.4 at room temperature and then labeled with secondary antibodies, conjugated with Alexa-488, Alexa-568 or 555, and Alexa 647, (Life Technologies, 1:500). Secondaries were prepared in PBS with 2% normal horse serum and 0.2% Triton X-100. Samples were incubated in secondaries for 2–3 h at room temperature. The secondary solution also contained DAPI, in order to counterstain cell nuclei (catalog #D9542, Sigma- Aldrich, 1 μg/ml). All samples were washed 3x for 5 min each in PBS, pH 7.4 at room temperature, mounted on glass slides in glycerol: PBS (9:1), and coverslipped.

Cellular imaging and analyses: Fluorescence images were obtained using an LSM 700 confocal microscope (Zeiss). For cochlear whole mounts and mid modiolar sections, Z-series images were obtained using the following objective lenses: ×5 (∼40-micron z-step-size), 20x (1 or 2 -micron z-step-size), or 63x (0.5 or 1.0-micron z-step-size). Images were processed and analyzed using Volocity 3D image analysis software (version 6.3, PerkinElmer) and Fiji (ImageJ2.0) (National Institutes of Health) and Adobe illustrator CS5.1.

Auditory brainstem responses (ABR): The hearing of 1-month old CX3CR1-GFP mice (T3-treated or saline controls) was assessed by quantifying thresholds for auditory brainstem responses (ABR). Prior studies have shown that CX3CR1^GFP/GFP mice have normal ABR thresholds (Kaur et al., 2019). Mice were anesthetized using intraperitoneal injections of ketamine (100 mg/kg) and xylazine (20 mg/kg). Responses were recorded from subcutaneous electrodes placed at the vertex (active electrode), and behind the right ear (reference electrode). A ground electrode was placed on the back. ABR thresholds were measured in response to tone pips at 5.6, 8, 11.2, 16, 22.6, 32, 45.2, and 64 kHz, using 5 ms tone pips (including 0.5 ms cosine² rise/fall) at a repetition rate of 20 per second. The responses were amplified 100,00x, filtered (100 Hz–3 kHz), and averaged using BioSig software (System 3; Tucker-Davis Technologies). Stimuli were presented in 5 dB steps from 5 to 100 dB SPL, in decreasing order, to the right ear. At each level, up to 1,000 responses were averaged. The lowest sound level at which a recognizable and reproducible wave 1 and 2 complex was noted, and considered threshold. If hearing threshold was not present at 100 dB SPL (the typical maximum sound level in our system), the threshold value was assigned as 100 dB for graphing and statistical purposes.

Macrophage, pyknotic nuclei and hair cell counts: Quantification was performed from 20X images, using Fiji software (ImageJ2.0) (National Institutes of Health). For counts of macrophages, hair cells and pyknotic nuclei, 2-3 areas from the apical and basal GER regions were selected. The Cell Counter plug-in was used for all counts. Macrophages were identified by strong GFP and CD45 labeling. Pyknotic nuclei were identified by small size and intense DAPI nuclear labeling, and apoptosis was verified by immunolabeling for cleaved caspase-3. All macrophages, pyknotic nuclei and hair cells counts were obtained manually, within the designed area of complete z-stack using Fiji software.

Statistical analysis: All data analysis and statistical tests were carried out using GraphPad Prism version 6.0d. Data are presented as mean ± SD. Student’s t-tests or analyses of variance (ANOVA), followed by post hoc tests, were applied as appropriate. Results were considered statistically significant when p < 0.05.

The mouse cochlea undergoes considerable structural changes during the first two postnatal weeks (Ruben 1967; Anniko 1983; Rueda et al., 1987; Kamilya et al., 2001; Dayaratne et al., 2014). One critical event is the death of columnar epithelial cells of the GER, leading to the opening of the inner sulcus. This process progresses along the tonotopic gradient of the cochlea, starting at the base (high frequency region) and then moving towards the apex (low frequency region) (Ruben 1967; Anniko 1983; Rueda et al., 1987; Kamilya et al., 2001; Dayaratne et al., 2014). Numerous macrophages are also present in the developing cochlea (Dong et al., 2018), but the role of macrophages in cellular remodeling is unknown. To resolve this issue, we first characterized the patterns of programmed cell death and the corresponding recruitment of macrophages in the GER of developing cochlea, using CX3CR1 ^GFP/+ mice, which express GFP in macrophages, microglia, monocytes and related cells (Diehl et al., 2013; Kaur et al., 2015a). Specimens were fixed at various developmental timepoints and prepared as whole mounts or mid-modiolar frozen sections. Immunolabeling was used to enhance GFP fluorescence in macrophages, and to identify apoptotic cells (cleaved caspase-3), and neurons (TUJ1+NF). DAPI staining facilitated the identification of pyknotic nuclei (a hallmark of apoptosis). Increased numbers of dying cells with pyknotic nuclei and/or immunolabeled for activated caspase-3 were observed in the basal GER region between P5-10, with the number peaking at P7 (Figures 1B–E′,G). Over the next 2–3 days, the appearance of pyknotic nuclei progressed from the base toward the apex, and the number of pyknotic nuclei in the apical GER peaked at P10 (Figures 1D,G). Macrophages were also observed in the GER between P7-13. Large numbers of macrophages initially appeared in the basal GER and then spread toward the apex. The numbers of macrophages in the basal and apical GER regions peaked at P10 and P13, respectively (Figures 1B–E′,H). The spatial distribution of macrophages followed a similar base-to-apex pattern that was observed for developmental cell death, with a temporal delay of ∼3 days (Figure 1B’). By P13, very few apoptotic cells were present within the GER region (Figures 1E,E’), but macrophage numbers in the GER remained elevated after cell death had terminated (Figure 1H). We also observed cleaved-caspase 3-labeled (i.e., apoptotic) cells being engulfed by macrophages (Figure 1D). Similar patterns of cell death and macrophage recruitment were observed in mid-modiolar sections of the developing cochlea (Figures 2A–E). Macrophages in the neonatal cochlea possessed both ameboid and ramified morphologies, while those in the mature cochlea were predominantly ramified (Figures 2F,G). The neonatal cochlea also contained larger numbers of macrophages than were present in mature cochleae. Taken together, these observations indicate that macrophages are recruited into the cochlea in response to developmental cell death and are actively involved in clearing dying cells from the GER region.

Apoptotic cells secrete a number of signaling molecules that are chemoattractants for phagocytes (Diehl et al., 2013; Sokolowski et al., 2014; Hirose et al., 2017; Park and Kim 2017). Examples of such ‘find-me’ signals include lysophosphatidylcholine (LPC), sphingosine-1-phosphate (S1P), nucleotides (ATP and UTP) and the CX₃C motif chemokine ligand 1 (CX3CL1, also called as fractalkine) (Park and Kim 2017). CX3CL1 is the sole ligand for the CX3CR1 receptor, which is expressed by macrophages, microglia, monocytes and related cells. One function of fractalkine signaling is to meditate chemoattraction between dying cells and macrophages (Sokolowski et al., 2014; Park and Kim 2017). To investigate the role of fractalkine signaling in macrophage recruitment and GER apoptosis, we compared macrophage numbers and cell death in the developing cochleae of CX3CR1^GFP/+ mice (which possess one allele of CX3CR1 and retain sensitivity to fractalkine) and CX3CR1^GFP/GFP mice (in which both copies of CX3CR1 had been replaced with the gene for GFP). Cochleae were collected from mice of both genotypes at P5, P7, P10 and P13 and processed as whole mounts. Samples were immunolabeled for GFP (to enhance the fluorescent signal in macrophages), cleaved caspase-3 (to identify apoptotic cells), and stained with DAPI. These specimens displayed similar patterns of cell death and macrophage recruitment to those described above (Figure 3). Apoptotic cells were first detected in the basal region of the GER at P5 (Figure 3A’), and progressed towards the apex between P7-P10. By P13, few apoptotic cells were detected in the apical region and no apoptotic cells were found in the basal GER (Figures 3A–D′,E). Macrophages were first apparent in the basal GER at P7 and their numbers moved apically at later ages (Figure 3F). Macrophages remained present at P13, after cell death in the basal GER region had ended. The spatial and temporal patterns of macrophage recruitment were similar in both CX3CR1 ^GFP/+ and CX3CR1 ^GFP/GFP (Figure 1 and Figure 3), and there was no statistically significant difference between the number of recruited macrophages and pyknotic nuclei in the GER region of CX3CR1 ^GFP/+ vs. CX3CR1 ^GFP/GFP mice (Figures 3E,F). These results indicate that fractalkine signaling is not essential for apoptosis or for recruitment of macrophages in the developing GER.

Thyroid hormone plays an important role in cochlear development and in the onset of hearing function ( Kelley and Forrest 2001; Ng et al., 2013 ). Both hypothyroidism and mutation of the thyroid hormone receptor β (THRB) gene delay the onset of GER remodeling and result in subsequent hearing loss (Kelley and Forrest 2001; Christ et al., 2004; Johnson et al., 2007; Mustapha et al., 2009; Sundaresan et al., 2016). It has also been shown that treating early neonatal mice with T3 hormone induces premature GER remodeling and accompanying hearing loss (Peeters et al., 2015). We investigated whether treatment with thyroid hormone (T3) also leads to earlier macrophage recruitment. Early apoptosis was induced in CX3CR1 ^GFP/+ mice by injecting T3 at P0 and P1 (1.5 µg/pup, s. c.). We collected cochleae from both T3-treated and saline-injected (control) mice at various developmental time points. Whole mount specimens were labeled with anti-GFP (to enhance GFP signal in macrophages, cleaved caspase-3 (apoptotic marker), and DAPI. Both T3-and saline-treated cochleae showed a base-to-apex pattern of GER remodeling. However, treatment with T3 resulted in a significant increase in cell death at P3, which was followed by early recruitment of macrophages into the GER region at P5 (Figure 4A). In contrast, cell death in saline treated samples began at P5 and was followed by recruitment of macrophages at P7 (Figures 4A–E). Similarly, cell death in T3-treated cochleae terminated earlier than in control specimens (Figures 4A–E). We also quantified hearing function in both T3-and saline-treated animals, using ABRs. Consistent with previous studies (Peeters et al., 2015 ), we observed significantly elevated hearing thresholds in the T3-treated animals, when compared with saline-treated animals (Figure 4F). Together, these data suggest that systemic injection of T3 hormone at P0 and P1 promotes premature apoptosis in the GER, which is followed by early recruitment of macrophages. This outcome is consistent with a positive correlation between developmental cell death and macrophage recruitment.

We investigated whether deletion of macrophages delays or impairs GER remodeling during cochlear development. We treated CX3CR1^DTR/+ mice and wild type control mice with diphtheria toxin (Diehl et al., 2013 ) at P0, P2, P4, P6, P8, P10 and P12. Multiple doses of DT were delivered to prevent macrophage repopulation during the experimental period. Cochleae were fixed at P5, P7, P10 and P13 and processed as whole mounts (Figure 6A). Samples were immunolabeled for CD45 (to identify macrophages) and cleaved caspase-3 (to identify apoptotic cells). Cell nuclei were also stained with DAPI. As depicted in Figures 5B–D′, DT treatment led to nearly-complete deletion of macrophages from the developing cochlea of CX3CR1^DTR/+ mice. Macrophage infiltration in the GER region of control animals was normal, but no macrophages were observed in the GER region of CX3CR1^DTR/+ mice (Figures 5E–L′). However, we observed normal patterns of apoptosis and GER regression in the developing cochlea of the CX3CR1^DTR/+ and control mice (Figures 5E–L’). Except in the apical region of P13 cochleae, we observed no significant increase in number of apoptotic cells in CX3CR1^DTR/+ mice, in comparison with controls. These data suggest that depletion of macrophages may cause a slight delay in the clearance of cellular debris in the apex, but does not otherwise affect GER regression (Figures 5M,N). Interestingly, we also noted that the DT treatment leads to loss of OHCs (Figures 6A,B). We observed missing OHC nuclei and labeling for cleaved caspase-3 in both CX3CR1^DTR/+ and WT control mice that were treated with DT at P10. No missing nuclei or apoptotic OHCs were observed in mice that did not receive DT.

In addition to the CX3CR1^DTR/+ mouse model, we also used a pharmacological approach to eliminate macrophages from the developing cochlea. CX3CR1^GFP/+ mice were given s. c. injections of the CSF1R antagonist BLZ945 (Milinkeviciute et al., 2019) at P2, P4, P6, P8, P10 and P12. (Multiple doses of BLZ945 were delivered, in order to avoid macrophage repopulation.) Cochleae were fixed at P5, P7, P10 and P13, and processed as whole mounts (Figure 7A). Treatment with BLZ945 eliminated nearly all macrophages from the developing cochlea (Figures 7B–D′), including the GER region (Figures 7E–L’). Notably, we observed normal patterns of apoptosis and GER regression in the developing cochleae of both the BLZ945-treated and control animals (Figure 7M). Macrophage depletion led to no significant difference in the numbers of pyknotic nuclei in the GER, when compared to controls (Figures 7M,N). Collectively, these findings suggest that macrophage depletion has no substantial effect on GER remodeling during development.