Sex as a biological variable. The study did not consider the sex of the mice and NHPs. Initial experiments showed no sex-specific hearing or vision differences, so both male and female animals were used in the experiments.

Dual-PCDH15 study design. The dual-PCDH15 strategy was based on the hybrid strategy (30, 31, 33). To promote recombination between the 5′ and 3′ viruses, we inserted the highly recombinogenic AK sequence derived from the F1 phage genome (32). The 5′ vector harbors a splice donor site upstream of the AK sequence, while the 3′ vector harbors a splice acceptor site immediately downstream of the AK sequence. The coding sequence of the mouse CD2-1 isoform of Pcdh15 was used for the initial design (NM_001142742.1). Experiments in human retinal organoids and in the green monkey retina used the coding sequence for the human CD1-1 isoform (NM_001142763.2). The AAV expression cassettes tested in this study are shown in Figure 1A. We used an AAV transgene plasmid, flanked by AAV2 inverted terminal repeats. For all experiments, we used a 584 bp CMV promoter, which we had previously demonstrated to be effective in the cochlea (27). Additionally, 3′ viruses were engineered to include an IRES GFP element to facilitate the cotranslation of a fluorescent marker. These constructs also incorporated the woodchuck hepatitis virus post-transcriptional regulatory element (WPRE) and a bovine growth hormone (BGH) poly(A) sequence.

Mouse models. Animal handling and breeding were performed in the Harvard Medical School animal facility. All studies were performed on Pcdh15fl/fl, Myo15a-Cre+/– and Pcdh15R245X/R245X mice (here referred to as Pcdh15–/–), which were previously described (27, 28). The Pcdh15fl/fl mice lacking Cre recombinase displayed hearing sensitivity and bundle morphology comparable to those in wild-type mice (Pcdh15+/+). Therefore, in this study, Pcdh15fl/fl Cre– mice were used as normal-hearing controls. All experiments were performed on Pcdh15fl/fl, Myo15a-Cre mice with a mixed C57BL/6J–129/Sv genetic background. The Pcdh15fl/fl and Pcdh15R245X/R245X mice were generated at Harvard Medical School, the Myo15a-Cre mice were provided by Christine Petit (Institut Pasteur, Paris, France) with assistance from Ronna Hertzano (University of Maryland School of Medicine, Baltimore, Maryland, USA), and the C57BL/6J mice were obtained from The Jackson Laboratory. Genotyping for the Myo15a-Cre mouse lines was conducted as previously described (46). Pcdh15–/+ mice exhibited hearing sensitivity and bundle morphology similar to those in wild-type mice; thus, in this study Pcdh15–/+ mice were used as normal controls for rescue experiments in Pcdh15–/– mice. All Pcdh15–/– mice were on a C57BL/6J genetic background. Genotyping for the Myo15a-Cre mouse lines was conducted as previously described (28).

Viral vector production. Most AAVs produced in this study were manufactured by the Viral Vector Core at Boston Children’s Hospital (Boston, Massachusetts, USA). Specifically, AAV9-PHP.B vectors were generated in HEK293 cells by polyethylenimine-mediated cotransfection of pAAV transfer plasmid, pHelper plasmid, and RepCap plasmid pUCmini-iCAP-PHP.B. One hundred twenty hours after transfection, both the medium and cells were collected. Subsequently, AAV9-PHP.B viruses were extracted and subjected to ultracentrifugation using a discontinuous density iodixanol (OptiPrep, Axis Shield) gradient. After ultracentrifugation, AAV vector–containing iodixanol fractions were isolated and concentrated via diafiltration. The purified AAV vectors were quantified using quantitative PCR, divided into single-use aliquots, and stored at –80°C until required, with thawing immediately before in vivo injections. For NHP experiments, AAVs were produced by PackGene Biotech.

Transduction of HEK293 cells. HEK293 cells (ATCC, CRL-1573) were plated on glass coverslips in DMEM with 10% FBS (Gibco) and penicillin/streptomycin (Pen/Strep; Invitrogen). On the following day, cells were transfected with Lipofectamine 3000 (Thermo Fisher Scientific) according to the manufacturer’s protocol. Plates were incubated at 37°C for 24 hours and then at 30°C for an additional 48 hours to promote high levels of protein expression.

HEK293 cells were transduced with AAV9-PHP.B vectors when they reached a low confluence of approximately 40%, using a multiplicity of infection of approximately 7 × 105. The vectors were diluted in 200 μL of DMEM containing 1% FBS, penicillin, and streptomycin, and then applied directly onto coverslips to create a hydration bubble. These coverslips were then placed in a 37°C incubator. After 16–24 hours, the cells were washed with medium containing 1% FBS and 0.5× Pen/Strep and subsequently replenished with 2 mL of fresh medium. They were then cultured at 30°C for an additional 48 hours to promote high levels of protein expression. Cells were collected for mRNA or fixed for immunofluorescence.

RNA extraction, cDNA production, reverse transcription, PCR amplification, and sequencing. Total RNA was extracted from transduced HEK293 (ATCC, CRL-1573) cells using the Zymo Quick-RNA Microprep Kit (Zymo Research, R1050). Reverse transcription was performed with Invitrogen’s SuperScript IV VILO Master Mix with ezDNAse enzyme (Invitrogen, 11766050). Pcdh15-CD2 cDNA was amplified with a forward primer that anneals to a sequence in the 5′ vector (5′-AGGATGAAAACGATCACCCCC-3′) and a reverse primer that anneals to a sequence in the 3′ vector (5′-GGTATGATGAGCCGGTAGGC-3′). The expected product size for the properly spliced Pcdh15-CD2 cDNA splice junction site is 154 bp. The DNA Gel Extraction Kit (Monarch, T1020S) was used to gel-extract PCR products. Purified PCR products were subcloned, transformed into competent cells, mini-prepped, and Sanger-sequenced using the NEB PCR Cloning Kit (New England Biolabs, E1202S).

Expression in human retinal organoids. Wild-type human retina organoids were generated and maintained at the Institute of Molecular and Clinical Ophthalmology Basel as was described previously (62). These organoids were subsequently transduced with AAV vectors in a 96-well plate using the following dosages: dual AAV (AAV-HA.Pcdh15 5′ + AAV-Pcdh15 3′), 1.7 × 1012 VGC; AAV-HA.Pcdh15 5′, 8.5 × 1011 VGC; AAV-Pcdh15 3′, 8.5 × 1011 VGC; AAV-CMV-GFP, 8.5 × 1011 VGC. Transduced organoids were maintained in 50 μL of 3:1 N2 medium at 37°C in 5% CO2. The composition of 3:1 N2 medium included DMEM (Gibco, 10569-010) supplemented with 20% Ham’s F12 Nutrient mix (Gibco, 31765-027), 10% heat-inactivated FBS (Millipore, ES-009-b), 1% N2 Supplement (Gibco, 17502-048), 1% NEAA Solution (MilliporeSigma, M7145), 100 μM taurine (MilliporeSigma, T0625), and 1 μM retinoic acid (MilliporeSigma, R2625) (62). After 4 hours, 50 μL of medium was added to each well. One day later, 100 μL of medium was added to each well. After 24 hours and every 48 hours thereafter, the solution was completely exchanged with fresh medium. Five weeks later, samples were fixed with 4% formaldehyde.

Expression in NHP retina. For in vivo delivery in a species with greater relevance to humans, we chose an Old World primate, the green monkey (Chlorocebus sabaeus). Injections were carried out by Virscio Inc. Before injection, the animal underwent a thorough ophthalmological assessment conducted by a veterinary ophthalmologist, and baseline screening to assess AAV9 neutralizing antibody seronegativity. The animal received methylprednisolone (40 mg i.m.) weekly for 4 weeks, starting on the day prior to dosing. Anesthesia was achieved with intramuscular ketamine (8 mg/kg) and xylazine (1.6 mg/kg). Pupil dilation with accomplished using topical 10% phenylephrine, 1% tropicamide, and/or 1% cyclopentolate. After placement of the scleral ports, a contact vitrectomy lens was positioned on the cornea using 0.9% saline as a coupling agent. A 25-gauge light pipe was inserted through the left scleral port into the vitreous cavity for intraocular illumination. Simultaneously, a subretinal cannula was introduced through the second scleral port. The cannula was gently advanced to touch the retinal surface in a specific location. Once the retinal surface showed a slight blanching at the point of contact, the vector was administered through the cannula. Three blebs were introduced, in the superior, inferior, and temporal regions of the retina, each containing dual AAV vectors at a dosage of 2.5 × 1012 VGC per bleb (totaling 7.5 × 1012 per eye), with each bleb receiving 50 μL of virus. The second eye served as a control and received an injection of a formulation buffer consisting of PBS with 0.001% F-68, with the same distribution of 3 blebs in the superior, inferior, and temporal regions of the retina, each containing 50 μL.

RNA extraction, cDNA production, and quantitative reverse transcription PCR in NHPs. Three cynomolgus monkey (Macaca fascicularis) cadavers were acquired post-euthanasia from an unrelated study at Massachusetts General Hospital (Boston, Massachusetts, USA) and transferred to Harvard Medical School as approved by IRB Ex Vivo Animal Tissue Importation Amendment ID 14-111-A02. Cochlear and vestibular system tissue from 5 inner ears and retinal tissue from 6 eyes were extracted and transferred into Trizol solution (Invitrogen) surrounded by a liquid nitrogen bath. Once cochlear and retinal tissues were digested, RNA and DNA isolation was carried out per the Trizol manufacturer’s instructions. Approximately 1–5 μg of RNA was used to create cDNA using the SuperScript III First-Strand Synthesis System for reverse transcription PCR (RT-PCR) (Thermo Fisher Scientific, 18080051). Quantitative RT-PCR (RT-qPCR) was carried out with SYBR Select Master Mix (Thermo Fisher Scientific, 4472908) on an ABI StepOnePlus qPCR machine using the following primers: PCDH15_CD1 forward primer 5′-CTCTATGAAGAACTTGGAGACAGCT-3′, reverse primer 5′-GGAAGAAAAGGGCATCACAACTTG-3′; PCDH15_CD2 forward primer 5′-CTCTATGAAGAACTTGGAGACAGCT-3′, reverse primer 5′-CCTCACTAGGCTCTCTAATTTCAACTT-3′; PCDH15_CD3 forward primer 5′-CTCTATGAAGAACTTGGAGACAGCT-3′, reverse primer 5′-CTCGATCTACAACTAACTTGATCATTCT-3′. Expression of PCDH15-CD1 isoforms was measured via ΔΔCt of the most prominent isoform in comparison with GAPDH or RPS19 calibrator genes.

AAV round window membrane injection in neonatal mice. P1 pups were anesthetized using cryoanesthesia and kept on an ice pack during the procedure. Injections were done through the round window membrane as previously described (43). Briefly, a small incision was made beneath the external ear. The round window niche was identified visually, and the viral vector solution was delivered via a micropipette needle at a controlled rate of 150 nL/min. The surgical incision was closed using two 7-0 Vicryl surgical sutures (Ethicon). After the injection, standard postoperative care protocols were implemented.

ABR and DPOAE testing. ABRs and DPOAEs were recorded following established procedures (27, 67), using a custom acoustic system developed by Massachusetts Eye and Ear. Adult mice aged 5 weeks were given anesthesia using a ketamine/xylazine mixture and were placed on a 37°C heating pad throughout the recording session. For ABR recordings, 3 subdermal needle electrodes were used. Tone-pip stimuli with a duration of 5 milliseconds and a rise-fall time of 0.5 milliseconds were delivered at frequencies ranging from 4.0 kHz to 32 kHz. Sound levels were increased in 5-dB increments, starting from approximately 20 dB sound pressure level (SPL) and increasing up to 80 dB. ABR Peak Analysis software (version 1.1.1.9, Massachusetts Eye and Ear) was used to determine ABR thresholds and measure peak amplitudes. DPOAEs were recorded for primary tones with a frequency ratio of f2/f1 = 1.2, where L1 was set as L2 + 10 dB. The f2 frequency ranged from 5.6 kHz to 32 kHz in half-octave increments. Primary tone levels were adjusted in 5-dB increments, spanning from 10 dB SPL to 70 dB SPL for f2.

Open field test. We used a square 37 × 37 cm2 arena with uniform, low-level illumination for our experiments. The testing took place when the animals were age P35. Each animal was positioned at the side of the arena, and position was recorded with video for a duration of 4 minutes. To prevent any potential olfactory distractions, the arena was thoroughly cleaned between test sessions. Video footage was subsequently analyzed using ImageJ software (NIH), and open field path outlines were generated. During the 4-minute observation period, we quantified the number of full-circle rotations, including both clockwise and counterclockwise turns.

Rotarod and swimming tests. The rotarod test was conducted over 2 days. On the first day, mice were positioned within an enclosed housing on a rotating rod, initially spinning at a constant rate of 4 revolutions per minute (rpm) for 5 minutes. Mice that fell during this training session were promptly placed back on the rotating rod. On the second day, the trained mice were once again placed on the spinning rod, but this time with a start speed of 4 rpm and acceleration rate of 20 rpm/min. The time each animal managed to remain on the device before falling to the floor of the housing was monitored by a timer and recorded after each trial. A 5-minute resting interval was enforced between trials, and a total of 5 trials were conducted for each mouse. The latency to fall off the rotarod was recorded. In the swimming test, the mice were placed in a tank filled with water, which forced them to swim. The time mice could swim before needing rescue was recorded.

FM1-43 loading in adult cochlea. Adult mice were subjected to anesthesia using isoflurane through an open drop method and were subsequently euthanized by cervical dislocation followed by decapitation. The otic capsules from the mice were carefully extracted and then placed in Leibovitz’s L-15 medium (Gibco). Under a stereomicroscope, the apical and mid-apical regions of the cochlea were microdissected. The tectorial membrane was gently pulled away to expose the sensory epithelium. A solution of FM1-43 (2 μM in L-15) was directly applied to the exposed epithelium at room temperature and left on for 1 minute. Subsequently, a solution of SCAS (0.2 mM) was applied. Imaging of the organs of Corti was carried out using an Olympus upright FV1000 confocal microscope equipped with a ×60 1.1-NA water-dipping objective lens.

Immunofluorescence labeling of mouse cochleae and HEK293 cells. Adult mice were anesthetized, then humanely euthanized by cervical dislocation followed by decapitation. The cochleae were dissected and fixed with 4% formaldehyde in HBSS for 1 hour at room temperature. Afterward, they were decalcified in 10% EDTA for 2 days. Once decalcified, the organs of Corti were microdissected and blocked with 10% donkey serum. Subsequently, the samples were stained with an anti-HA antibody (Abcam ChIP Grade, ab9110) diluted to 1:500 in 10% donkey serum and incubated overnight, followed by multiple rinses in HBSS. The samples were then incubated in a blocking solution (10% donkey serum) for 30 minutes at room temperature. After that, they were incubated overnight at room temperature with a donkey anti-rabbit IgG secondary antibody conjugated to Alexa Fluor 594 (Invitrogen, R37119), diluted to 1:500 in the blocking solution, which also included Alexa Fluor 405 phalloidin (Invitrogen, A30104) diluted to 1:20 to label actin. After the secondary antibody steps, the samples underwent several rinses and were mounted on Colorfrost glass slides (Fisher Scientific) using ProLong Gold Antifade mounting medium (Thermo Fisher Scientific).

The imaging was conducted using a Nikon Ti2 inverted spinning disk confocal microscope with Nikon Elements Acquisition Software AR 5.02, using the following objectives: a Plan Apo λ ×100/1.45 oil, a Plan Fluor λ ×40/1.3 oil, and a Plan Apo λ ×60/1.4 oil.

For immunocytochemistry in HEK293 cells, transfected cells were fixed with 4% formaldehyde for 1 hour, then washed 3 times with HBSS, and subsequently blocked with 10% donkey serum. We used either a sheep polyclonal anti-PCDH15 antibody (R&D Systems, AF6729) or a rabbit anti-HA (C29F4) antibody (Cell Signaling Technology, 3724), both diluted to 1:200 in 10% donkey serum, and incubated samples for 24 hours at room temperature, followed by several rinses in HBSS. Next, the samples were incubated in a blocking solution for 30 minutes and then incubated overnight at room temperature with a donkey anti-sheep IgG secondary antibody conjugated to Alexa Fluor 594 (Invitrogen, A-11016) or donkey anti-rabbit immunoglobulin IgG secondary antibody conjugated to Alexa Fluor 594 (Invitrogen, R37119) diluted to 1:500 in the blocking solution. After the secondary antibody incubation, the samples underwent several rinses in HBSS and were mounted on Colorfrost glass slides using ProLong Gold Antifade mounting medium. Imaging was carried out using an Olympus FluoView 1000 confocal microscope equipped with a ×60/1.42-NA oil-immersion objective.

Immunofluorescence labeling of NHP retina and human retina organoids. NHPs were euthanized and perfused with heparinized saline followed by 4% formaldehyde. Collected eye globes were postfixed for another 24 hours in 4% formaldehyde before the retinas were dissected. Human retinal organoids were fixed with 4% formaldehyde followed by a triple wash with PBS. Next, retina samples and organoids were cryoprotected by incubation in gradient concentrations of sucrose and then embedded in OCT compound and stored at −80°C before sectioning. Cryosections were generated using a Leica CM 3050 S cryostat at 30 μm step size.

For immunofluorescence labeling, the following primary antibodies and secondary antibodies were used: mouse anti-HA antibody (1:200) (BioLegend, 16B12), rabbit anti-HA antibody (1:200) (Abcam, ab9110), rabbit anti-ARR3 antibody (1:200) (Sigma Aldrich, HPA063129), sheep polyclonal anti-PCDH15 (1:200) (R&D Systems, AF6729), mouse monoclonal anti-rhodopsin (1:500) (MilliporeSigma, MAB5316), donkey anti-rabbit IgG secondary antibody conjugated to Alexa Fluor 594 (1:200) (Invitrogen, R37119), donkey anti-sheep IgG conjugated to Alexa Fluor 488 (1:200) (Invitrogen, A-11015), donkey anti-mouse IgG conjugated to Alexa Fluor 488 (1:200) (Invitrogen, A32766), donkey anti-mouse IgG conjugated to Alexa Fluor 405 (1:200) (Invitrogen, A48257), and donkey anti-rabbit IgG conjugated to Alexa Fluor 488 (1:200) (Invitrogen, A-21206). Samples were blocked with 10% donkey serum for 1 hour at room temperature. Antibodies were diluted in 10% donkey serum and incubated overnight at room temperature, followed by several rinses in HBSS. Next, samples were incubated in a blocking solution for 30 minutes and incubated overnight at room temperature with a secondary antibody in the blocking solution. We used DAPI (Invitrogen, D1306) or Hoechst 34580 dye (Invitrogen, H21486) to label cell nuclei (1:500) and BODIPY (4,4-difluoro-4-bora-3a,4a-diaza-s-indacene; Invitrogen, D3835) to label membranes (1:1500). Imaging was performed with a Nikon Ti2 inverted spinning disk confocal microscope.

Quantification of confocal microscopy data. Microscopy data analysis and quantification were done in the Fiji distribution of ImageJ v1.53. Transduction efficiency in cells was evaluated as previously described (27). Hair cells were identified with phalloidin staining of bundles and photoreceptors identified with BODIPY, and transduced cells identified by positive HA-tag labeling. Control samples without AAV were used to correct for autofluorescence. Segments with dissection-related damage were removed from the analysis.

GraphPad Prism 7 software was used to generate the graphs and perform the statistical analysis. The results are shown as mean ± SEM or mean ± SD as indicated in figure legends. Randomization was used whenever possible.

Conventional scanning electron microscopy. Scanning electron microscopy in adult mice, NHP retina, or retina organoids was performed as previously described (27, 68). Immediately after extraction, cochleae underwent prefixation by immersion in a solution containing 1% glutaraldehyde and 4% formaldehyde, both in 0.1 M cacodylate buffer (pH 7.2) supplemented with 2 mM CaCl2, for 1 hour at room temperature. After the prefixation step, the samples were postfixed using 2.5% glutaraldehyde in 0.1 M cacodylate buffer (pH 7.2), supplemented with 2 mM CaCl2, for an additional hour at room temperature. They were next rinsed in 0.1 M cacodylate buffer (pH 7.2) and then with distilled water. The cochlear bone was removed using a 27-gauge needle, followed by microdissection of the organ of Corti.

Next, the samples were immersed in a saturated aqueous solution containing 1% osmium tetroxide for an hour in a light-controlled environment, and then were postfixed using a 1% tannic acid aqueous solution for an hour in the dark. Finally, the samples underwent a series of rinsing, dehydration, and critical-point drying. The prepared samples were mounted on aluminum stubs equipped with carbon-conductive tabs. They were then sputter-coated with a platinum layer to a thickness of 5 nm using an EM ACE600 sputter coater (Leica) and imaged with a Hitachi S-4700 field-emission scanning electron microscope equipped with a backscattered electron detector.

Immunogold scanning electron microscopy in mouse cochlea and human retina organoids. Immunogold scanning electron microscopy was conducted in accordance with previously described protocols (27, 68). Cochleae and human retina organoids were fixed using 4% formaldehyde. After fixation and washing, samples were subjected to a 2-hour blocking step at room temperature using 10% normal goat serum. Next, the samples were incubated with primary antibodies for 24 hours at room temperature and were subsequently rinsed in HBSS. An anti–HA tag antibody (Abcam, ab9110) at a 1:200 dilution in 10% donkey serum was used to label the HA tag. After rinsing, the samples were blocked for 30 minutes at room temperature using 10% normal goat serum. They were then incubated overnight at room temperature with a secondary antibody solution consisting of 12 nm Colloidal Gold AffiniPure Goat Anti-Rabbit IgG (Jackson ImmunoResearch, 111-205-144) at a 1:30 dilution in the blocking solution. After the application of the secondary antibodies, the samples underwent rinses in HBSS.

Finally, the samples were prepared for observation. This involved a dehydration step, followed by critical-point drying, mounting, and sputter-coating with palladium in the range of 3–5 nm. Samples were imaged using a Hitachi S-4700 scanning electron microscope.

Statistics. All experiments, except for the ones noted below, were replicated in at least 3 independent experiments using separate samples, such as mice, organoids, or plates with HEK cells.

In NHP experiments, an initial detailed examination of PCDH15 expression and photoreceptor structure was conducted using immunofluorescence and scanning electron microscopy on 1 cynomolgus monkey, with 4 samples taken from different areas of the retina. Evaluation of dual-AAV delivery in retina was conducted in 1 green monkey, in both eyes. Three blebs were introduced in each eye, in the superior, inferior, and temporal regions of the retina. Quantification was performed on at least 5 regions from each bleb. For experiments involving RT-qPCR, 5 whole inner ears or the retinas from 6 eyes were used.

Only samples displaying notable dissection-related damage were excluded from the analysis. No other data were omitted. All figures were created using Adobe Illustrator 2024 (v28.5). The ABR data were organized using Microsoft Excel 2016 (v16.0.5378.1000, Microsoft). Graphs were generated and statistical analysis was conducted using Prism 10 software (v10.02.3, GraphPad Software Inc.). Statistical significance was assessed using 1-way ANOVA followed by Dunnett’s test for the behavioral vestibular tests, and 1-way ANOVA followed by post hoc Tukey’s test to reveal tissue-specific PCDH15 isoform expression. A P value less than 0.05 was considered statistically significant. The results are presented as mean ± SEM as specified in the figure legends. Randomization was employed whenever possible.

Study approval. All mouse experiments were carried out in accordance with ethical guidelines under protocol IS00001452, approved by the Institutional Animal Care and Use Committee at Harvard Medical School (Boston, Massachusetts, USA). These studies adhered to NIH guidelines.

Data availability. All data generated or analyzed during this study are included in this article and its supplemental information files. Values for all data points in graphs are reported in the Supporting Data Values file.