Protocols for housing and handling of mice were approved by Case Western Reserve University’s Institutional Animal Care and Use Committee. CD1 Pax2-Cre [Tg(Pax2-cre)1Akg] mice (Ohyama and Groves, 2004) were obtained from the Mutant Mouse Resource & Research Center (stock #10569). We used Macf1^fl/– (Wu et al., 2008), Macf1^fl/+, Macf1^fl/fl, and Macf1^+/–; Pax2-Cre mouse lines. Each strain was outcrossed with FVB/NJ mice for at least seven generations. Specifically, C57BL/6 Macf1^fl/fl mice were outcrossed to FVB/NJ mice (Wu et al., 2008). In addition, to improve the likelihood of Cre-mediated recombination in the cochlea, we generated a germline transmissible knockout allele by crossing Macf1^fl/+ with BALB/c CMV-Cre mice (Tg(CMV-Cre)1Cgn/J obtained from the Jackson Laboratory stock #003465), with resulting Macf1^+/−; CMV-Cre offspring outcrossed to FVB/NJ mice (Schwenk et al., 1995) for more than seven generations before use in generating experimental data. Note, early publications of Macf1^fl describe loxP sites flanking exons 6 and 7 (Wu et al., 2008); however, after modern genomic analyses, the loxP sites were shown to flank exons 11 and 13. Ai3-YFP [Gt(ROSA)26Sor^{tm 3(CAG–EYFP)Hze}] (Madisen et al., 2010) reporter mice were crossed to Pax2-Cre mice. One-month old or P5 mice of either sex were used for experiments.

Mouse genotypes were determined by a standard PCR amplification procedure (Truett et al., 2000). Floxed and wildtype Macf1 alleles were genotyped using forward primer (F1′: 5′-AAAGAAACGGAAATACTGGCC-3′; exon 10) and reverse primer (R1′: 5′-GCAGCTTAATTCTGCCAAATTC-3′; exon 11). T_{A (annealing)} = 56°C; T_{E (extension)} = 1 min. The Macf1 knock out (KO) allele was separately genotyped using the forward primer F1′ with reverse primer targeting exon 14 (R2′: 5′-AAAGAAACGGAAATACTGGCC-3′; T_A = 50°C; T_E = 2.5 min) (Fassett et al., 2013). The Pax2-Cre allele was genotyped with the primer pair F′ (5′-GCCTGCATTACCGGTCGATGCAACGA-3′) and R′ (5′-GTGGCAGATGGCGCGGCAACACCATT-3′). T_A = 67°C; T_E = 1 min (Truett et al., 2000). PCR was conducted in a thermocycler (Applied Biosystems^TM SimpliAmp^TM Thermal Cycler and on a Bio-Rad PTC-200 DNA Engine^® Cycler).

One-month old mice cochleae were dissected and fixed in 4% paraformaldehyde in 1× phosphate-buffered saline (PBS) overnight at 4°C. Apical, middle, and basal turns of cochleae were defined according to Supplementary Figure 1 (Gilels et al., 2017). The tissue was rinsed three times in 1× PBS for 10 min each and decalcified with 1 mM ethylenediaminetetraacetic acid (EDTA) at 4°C for 2–3 days. Cochleae were blocked and permeabilized simultaneously with 0.05% Triton-X100 (Sigma) and 10% goat serum solution in 1× PBS for 2 h at room temperature. To visualize cells, the tissue was washed and labeled with anti-myosin7a (Proteus Biosciences, Inc.) primary antibody at a 1:500 dilution with blocking solution at 4°C overnight. After three, 5-min washes in blocking solution, tissue was labeled with secondary antibodies at a 1:1,000 dilution for Alexa Fluor 488 IgG (Invitrogen) and a 1:70 dilution for Alexa Fluor 633 phalloidin (Invitrogen) at 4°C overnight, washed, mounted (ProLong Diamond Antifade Mountant), and imaged on a TCS SP8 confocal microscope (Leica). Ai3-YFP/Pax2-Cre mice were imaged at an excitation wavelength of 512 nm for YFP.

Whole-mount immunolabeling of cochleae was performed with anti-ACF7 (Karakesisoglou et al., 2000) and mouse anti-acetylated tubulin (T6793; Sigma) as described by Pataky et al. (2004), with modifications being a 1 h fixation, 2 h permeabilization, and 2 h blocking (1% bovine serum albumin) (Pataky et al., 2004). Anti-ACF7 rabbit polyclonal antiserum against amino acids 1,622–1,817 of a partial sequence of the primary ACF7 isoform was used (Bernier et al., 1996; Karakesisoglou et al., 2000). Anti-ACF7 and anti-acetylated tubulin were used at a 1:100 dilution overnight at 4°C. Treatment with goat anti-rabbit and anti-mouse Alexa Fluor 488 IgG (1:200) and Alexa Fluor 633 phalloidin (1:50) occurred for 2 h at room temperature. Mounting and imaging were conducted as above.

For hair cell enumeration, outer and inner hair cells were manually counted using LAS X software (Leica). A hair cell was considered present when the hair bundle and cuticular plate were observed and myosin7a signal detected (Raphael and Altschuler, 1991; Anttonen et al., 2014; Monzack et al., 2015). We used the hair cells from the entire cochlea for calculations. To calculate the percent of present hair cells we used: % hair cells present = present hair cells/(missing + present hair cells) × 100. For organelle length measurements, we measured kinocilia of all useable cells from the apical turn in a confocal series similar to methods described by May-Simera et al. (2016). Kinocilium length measurements of hair cells labeled with anti-acetylated tubulin in P5 mice were obtained with LAS X software (Leica).

Mice of one month of age were anesthetized with a ketamine hydrochloride and xylazine hydrochloride solution (100 and 20 mg/kg, respectively) via intraperitoneal injection, and kept on a homeothermic heating pad at 37°C for the duration of sedation. Platinum subdermal needle electrodes were placed below the pinna of the left and right ears to record the ABRs, and on the back of the mice for electrical grounding (Akil et al., 2016; Geng et al., 2017). Electrical responses of the cochlear ganglion neurons and the nuclei were recorded at different levels of pure tone frequencies played for 100 ms at 8, 16, and 32 kHz (Intelligent Hearing System SmartEP 4.20 system). The lowest sound pressure level (SPL) that could generate an electrical response at the ABR thresholds of these different frequencies was considered. To evaluate if the lack of ACF7 has an impact on ABR thresholds, Macf1^fl/–; Pax2-Cre mice were tested with control littermates under the same testing conditions. The intensities of SPL ranged from 100 dB SPL to 20 dB SPL, in 10 dB SPL intervals, and the responses were averaged over 1,024 sweeps. Tones were presented to mice via high-frequency transducers placed in the ear, with ABRs recorded in a soundproofed and closed, free-field system. ABR thresholds were determined post-procedure by identifying the lowest stimulus level that yielded a detectable ABR waveform at various time points.

All statistics were performed with GraphPad Prism version 8. Data are reported as mean ± SEM or SD. Comparisons between groups were tested with unpaired two-tailed Student’s t-tests. R-Studio was used to generate circular histograms (RStudio, Inc.).

To understand the requirement of ACF7 in hair cells of mammals, we used a genetic approach. Because mice that lack ACF7 die pre-implantation (Kodama et al., 2003; Chen et al., 2006), we conditionally targeted the associated gene by using the genetically modified mouse Macf1^fl/–, which has two loxP sites flanking a region encompassing exons 11 through 13 (Wu et al., 2008). This genetic strain has been successful in revealing the role of the ACF7 protein as a cytoskeletal integrator in other cell types as deletion of these exons are predicted to shift the coding sequence of downstream exons and has been shown to produce cells that are devoid of ACF7 (Fassett et al., 2013). In addition, we applied a strategy that used a Pax2-Cre allele as a Cre-driver line. Pax2 is expressed throughout the otic placode at E9.5 and thus should excise exons that are in-between loxP sites in all of the cells in the developing inner ear, including the hair cells in the cochlea (Ohyama and Groves, 2004; Figure 1).

Initially, to visualize Cre expression in the organ of Corti, Pax2-Cre mice were crossed with the Cre recombinase reporter Ai3-YFP (Madisen et al., 2010). In Ai3-YFP, transcription of the YFP reporter only occurs following Cre mediated excision of a premature stop codon. We observed YFP expression in most inner and outer hair cells (Figure 1B), indicating that the Pax2 driver has the spatial and temporal expression to knock out ACF7 in hair cells.

To determine if this stratagem was effective at eliminating ACF7 from hair cells, we labeled the organ of Corti of Macf1^fl/–; Pax2-Cre and control mice with ACF7 antiserum and phalloidin to mark the actin of the cuticular plate (Karakesisoglou et al., 2000). In control inner hair cells, ACF7 localized to the cuticular plate and the circumferential band; however, the target protein was absent from inner hair cell cuticular plates and circumferential bands in the Macf1^fl/–; Pax2-Cre mice, demonstrating that this gene targeting strategy was successful (Figure 1C). We did not observe labeling of outer hair cells in control and experimental groups. This could be because of different mounting procedures. Here, we use whole-mount labeling; however, previously we labeled cultured cochleae, indicating that labeling of outer hair cells may be sensitive to the immunological procedure (Antonellis et al., 2014). Since labeling of the cuticular plate and circumferential band is absent in inner hair cells, the conditional targeting strategy was effective.

To determine if the presence of ACF7 is required for the survival of hair cells in any of the three turns of the cochlea, tissue of Macf1^fl/–; Pax2-Cre and control mice were labeled for myosin7a and for actin using phalloidin (Figure 2). Macf1^fl/–; Pax2-Cre mice are indistinguishable from their littermate controls, showing no apparent defects in the organ of Corti in either hair cells or supporting cells (Figures 2A–F). Quantification of hair cells between Macf1^fl/–; Pax2-Cre and control littermates in one-month old mice revealed similar numbers of hair cells (Figures 2G,H), indicating that ACF7 is not required for hair cell survival in any of the cochlear turns.

Since ACF7 has a compelling localization pattern in and around the cuticular plate, we tested the hypothesis that this protein may be involved in PCP. Macf1^fl/–; Pax2-Cre mice and littermate controls were examined for defects in PCP by labeling actin followed by a hair bundle orientational assay on the organ of Corti (Figures 3A,A′,B,B′). The orientation of individual stereociliary bundles at the approximate midpoints along the lengths of the cochleae (Figures 3D,E) were measured and plotted in circular histograms (Figure 3C). Average bundle deviation from the midlateral axis of hair bundles across different animals were measured for Macf1^fl/–; Pax2-Cre mice and control mice (Figure 3F). In each row, OHC1, OHC2, OHC3, or IHC, there was no significant difference in the average stereociliary bundle deviation between the Macf1^fl/–; Pax2-Cre mice and the controls.

In prenatal mice, ACF7 is required for normal kinocilia lengths (May-Simera and Kelley, 2012). Therefore, we tested if kinocilia lengths were altered in P5 Macf1^fl/–; Pax2-Cre mice, a timepoint during cochlear development that was chosen because it is before the kinocilia degenerate (Wang and Zhou, 2021) but should still also reflect defects in kinocilia formation if they do exist in these conditional knockout animals. Using anti-acetylated tubulin to label the kinocilia and confocal microscope software for measurements, we showed that the mean length of kinocilia from control mice is 2.49 ± 0.02 μm (mean ± SEM) (Figure 4). In contrast, the mean length of kinocilia from the Macf1^fl/–; Pax2-Cre mice is 2.20 ± 0.01 μm. The significant reduction of mean kinocilium length in the conditional knockout mice demonstrates that ACF7 is required for normal kinocilium length after birth.

To further investigate whether the lack of ACF7 has an impact on inner hair cell function, we compared ABR thresholds of Macf1^fl/–; Pax2-Cre mice to littermate control mice. We observed no significant difference (Figure 5) in ABR thresholds between the two groups at one-month of age. ABR thresholds (Mean ± SEM) for the control group were 23.9 ± 2.2 (8 kHz), 29.4 ± 1.7 (16 kHz), and 30.0 ± 5 (32 kHz); in contrast, ABR thresholds for the Macf1^fl/–; Pax2-Cre group were 25.0 ± 6.6 (8 kHz), 33.8 ± 5.2 (16 kHz), and 31.3 ± 3.5 (32 kHz).