Gene Editing Therapy Restores Hearing in Mice with Genetic Deafness

To identify the most effective genome editor for targeting theallele, we designed sgRNAs for use withCas9 (SaCas9) sgRNA1 andCas9 (SpCas9) sgRNA2–sgRNA5 nucleases (). All Cas9/sgRNA combinations were incorporated into the same plasmid backbone containing 3 nuclear localization signals and optimized sgRNA scaffold () (). The Cas9/sgRNA plasmids and a puromycin resistance plasmid were cotransfected into the primary fibroblasts derived frommice, using nucleofection. The puromycin-resistant cells were collected for DNA extraction, amplicon amplification, and analysis by next-generation sequencing (NGS) for insertion-deletion (indel) identification (). P2rx2 V61L Staphylococcus aureus Streptococcus pyogenes P2rx2 V61L/+ Figure 1A ⁴³ Figure 1B Figure 1C

In primary fibroblasts, efficient editing, as indicated by indel reads, was detected in theallele across all gRNA designs for SpCas9 and SaCas9, with editing efficiencies ranging from 71.28% to 83.29% (). For SpCas9 sgRNAs (sgRNA-2, -3, -4 and -5), elevated levels of indels were detected in WTallele (8.73%–29.20%) (, and, A and B; supplemental material available online with this article;), indicating undesired off-target editing on the WT allele. The findings suggested that SpCas9/sgRNAs can tolerate 1 nucleotide mismatch between theallele and the WT allele, making it unsuitable for allele-specific genome editing. In contrast, NGS analysis of SaCas9–sgRNA-1 revealed a high level of indel formation in theallele (75.01% ± 4.55%) (), with negligible indels observed in the WT allele (0.45 ± 0.39%) (), indicatingallele-specific editing. For SaCas9–sgRNA-1–mediated editing, NGS revealed a variety of indels, with single-nucleotide deletions being the most common type (). Most of the indels (85.1%) caused frameshift mutations (), suggesting efficient disruption of the P2rx2protein by SaCas9–sgRNA-1. P2rx2 V61L P2rx2 + P2rx2 V61L P2rx2 V61L P2rx2 V61L V61L Figure 1D Figure 1D Supplemental Figure 1 Figure 1, D and E Figure 1, D and F Figure 1G Figure 1H https://doi.org/10.1172/JCI187872DS1↗

To assess the editing efficiency and specificity in the cochlear cells, we used the mouse organ of Corti cell line HEI-OC1 to generate HEI-OC1-P2rx2-V61L cells (OC1-V61L), which harbor theDNA fragment in the genome using the PiggyBac system (). NGS analysis of HEI-OC1-V61L cells after transfection with SaCas9–sgRNA-1 or SpCas9/sgRNA-2 plasmids revealed similarly robust indel formation. In the HEI-OC1 cells harboring the WT allele, SpCas9/sgRNA-2–mediated editing was evident from indel analysis, consistent with the primary fibroblasts results (and). In contrast, SaCas9–sgRNA-1 transfection produced negligible editing events in the HEI-OC1 cells with the WT allele (, D and G). The indel profile in SaCas9–sgRNA-1–edited HEI-OC1-V61L cells showed various types of indels (, E and F), with single-nucleotide deletions being the most common. P2rx2 V61L Supplemental Figure 1C Supplemental Figure 1D Figure 1D Supplemental Figure 1 Supplemental Figure 1

To comprehensively assess off-target editing, we performed circularization for in vitro reporting of cleavage effects by sequencing (CIRCLE-Seq) () on HEI-OC1-V61L cells edited by SaCas9–sgRNA-1 and identified no off-target sites beyond the on-target locus (). Furthermore, computational predictions were used to identify potential off-target loci (,). Analysis of the top 7 predicted off-target sites, identified using the cutting frequency determination score () in SaCas9–sgRNA-1–edited HEI-OC1-V61L cells, revealed no off-target editing events, whereas on-target editing efficiency exceeded 80% (and). Collectively, these results demonstrate that SaCas9–sgRNA-1 specifically targets theallele with minimal off-target effects in mouse primary fibroblasts or a cochlear cell line. ⁴⁴ Figure 1, I and J ^{45 46 45} Figure 1D Supplemental Figure 1H P2rx2 V61L

To evaluate the therapeutic potential of SaCas9–sgRNA-1, we packaged it in AAV2 and administered it to the cochlea of 4-week-oldmice via RWM+CF at a dose of 4 × 10vg/ear. After 8 weeks, the cochleae were harvested, and DNA and RNA were extracted for NGS and indel analysis (). In uninjected mice, background indel frequencies ranged between 0% and 0.05%. In vivo NGS of the cochlea showed the presence of indels at thelocus in the injectedears after 8 weeks, with an indel rate of 2.62% ± 0.64% and no detectable editing of the WT allele (). We also injected AAV2–SaCas9–sgRNA-1 into the cochlea of WT adult mice; NGS analysis did not detect indels in the injected WT cochlea (). We determined the expression of SaCas9 after cochlear injection by qPCR. SaCas9 mRNA level peaked around 4 weeks after injection, followed by a decline over time (). By the 12th week after injection, SaCas9 expression was no longer detectable (), which potentially is attributable to either CMV promoter silencing or the immune response triggered by SaCas9 in adult mice, as reported by previous studies (,–). P2rx2 V61L/+ P2rx2 V61L P2rx2 V61L/+ 9 Figure 2A Figure 2, B and C Figure 2, D and E Figure 2F Figure 2F ^{43 47 49}

Given that on-target editing efficiency would directly impact hearing rescue, we measured editing efficiency by examining changes intranscripts in the injectedcochleae.is specifically expressed in the hair cells (HCs). We performed NGS analysis of cDNAs from injected and uninjected contralateral control cochleae and detected multiple indel-containingtranscripts in the injected samples, whereas none were found in the uninjected control cochleae (). Additionally, exon 2–skippingtranscripts were detected in the injected cochleae, accounting for 1.46% of total reads, suggesting that genome editing at thelocus disrupted normal splicing (). This is likely due to the SaCas9–sgRNA-1 cutting site targeting the exon 2 splicing acceptor, which was disrupted by indels. Rare, alternatively spliced transcripts were detected in both uninjected and injected samples, constituting less than 5% of all reads (). This suggests that genome editing did not alter the expression of these rare transcripts. P2rx2 P2rx2 V61L/+ P2rx2 P2rx2 P2rx2 P2rx2 V61L Figure 2, G and H Figure 2H Figure 2, G and H

Compared with uninjectedcochleae, a significant decrease in the ratio of uneditedtranscripts to WTtranscripts was observed in the injected ears (). After SaCas9–sgRNA-1–mediated editing of theallele, the proportion of uneditedtranscripts in the injected ears decreased by 28.2%, to 71.2% ± 3.0%, relative to the uninjected ears (). These findings suggest that approximately 28% of-expressing cells were successfully edited at theallele through genome editing. Given that cells incochleae carry only 1 copy of the V61L allele, these findings suggest that genome editing successfully disrupted the mutant allele in approximately 28% of-expressing HCs. P2rx2 V61L/+ P2rx2 V61L P2rx2 P2rx2 V61L P2rx2 V61L P2rx2 P2rx2 V61L P2rx2 V61L/+ P2rx2 Figure 2I Figure 2I

To precisely determine editing efficiency in HCs, we isolated HCs using the FM1-43 uptake assay (). Eight weeks after injection of SaCas9–sgRNA-1 into adultears, the cochleae were dissected and the cells dissociated using 0.25% trypsin-EDTA, then incubated with FM1-43FX to label HCs through open mechanotransduction channels (,). Subsequently, the labeled HCs were collected and lysed to obtain genomic DNA, which was then analyzed for indel frequency by NGS (). In contrast to uninjected cochleae, a wider range of indel types was observed in isolated HCs (), with an indel frequency of 26.96% ± 4.2% in theallele (). The editing frequency based on the HC DNA was consistent with that from thetranscript analysis. Furthermore, the combined reads from the edited and uneditedlocus were equal to thereads in the injected ear (). This finding indicates that no noticeable chromatin rearrangements (e.g., lesions, large deletions, insertions) were caused by genome editing, reinforcing the safety and accuracy of the approach in targeted HCs. Supplemental Figure 2A ^{50 51} Figure 2J Figure 2K Figure 2L Figure 2M P2rx2 V61L/+ P2rx2 V61L P2rx2 V61L P2rx2 V61L P2rx2 +

AAV integration into the genome as a consequence of induced DNA breaks (,) is a potential safety concern in editing therapy via AAV delivery (). To study AAV integration at the target locus in the inner ear, we designed specific primers for PCR and NGS to detect integration between thelocus and AAV inverted terminal repeats (ITRs) (). We first evaluated AAV integration in the HEK-293T-P2rx2-V61L cells after AAV2–SaCas9–sgRNA-1 infection. A low level of-ITR integration was observed, and the rate of integration was correlated with AAV2–SaCas9–sgRNA-1 dose (). At a dose of 10vg/cell, on-target editing approached peak efficiency with virtually no detectable integration (). At the AAV dose exceeding 10vg/cell, a-ITR integration rate of 5.0% ± 0.2% was detected (). These findings strongly support that optimizing AAV dose allows for maximum editing efficiency while minimizing AAV integration. ^{48 52} Supplemental Figure 2B Supplemental Figure 2C Supplemental Figure 2D Supplemental Figure 2D Supplemental Figure 2D P2rx2 P2rx2 V61L P2rx2 V61L 3 4

We further confirmed that no-ITR integration reads were detected in isolated HCs from in vivo samples after AAV2–SaCas9–sgRNA-1 injection (, E and F). Furthermore, qPCR analysis of the-ITR transcript revealed no significant difference between uninjected and injected animals (), indicating the absence of-ITR chimeric transcripts caused by genome editing. Together, these results demonstrate that AAV2-mediated genome editing at thelocus exhibits negligible AAV integration in the injected mature cochlea, providing a crucial piece of evidence for the safety and specificity of the therapeutic approach. P2rx2 V61L P2rx2 V61L P2rx2 V61L P2rx2 V61L Supplemental Figure 2 Supplemental Figure 2G

AAV2 is a well-established AAV serotype known for its high transduction efficiency in inner hair cells (IHCs) and outer hair cells (OHCs) of the cochlea, with preferential distribution from the apical turn to the basal turn (). Given its effective transduction capabilities, we selected AAV2 as the vector for genome editing therapy inmice. ¹³ P2rx2 V61L/+

To assess its distribution and transduction efficiency, we injected 1 × 10vg of AAV2-GFP with a titer of 1 × 10vg/mL into the cochleae of 4-week-old mice via RWM+CF (). Cochleae were harvested 4 weeks after injection and analyzed for GFP expression across the apical (8 kHz), middle (16 kHz), and basal (32 kHz) turns (, B and C). AAV2-GFP efficiently transduced all IHCs across all the cochlear turns (). In OHCs, transduction efficiency varied, decreasing from the apex to the base (). Intensity quantification revealed uniformly high GFP intensity in IHCs, whereas intensity in OHCs was reduced to approximately 50% in the apex and approximately 25% in the base compared with IHCs (). 10 13 Supplemental Figure 3A Supplemental Figure 3 Supplemental Figure 3D Supplemental Figure 3D Supplemental Figure 3E

Because the titer of AAV2–SaCas9–sgRNA was 4.46 × 10vg/mL, we injected 4 × 10vg of AAV2-GFP into the cochleae of 4-week-old mice to better match the viral dose (). Four weeks after the injection, we observed that IHC transduction remained at 100%, whereas OHC transduction rates were approximately 80% in apical and middle turns and 65% in the basal turn (, G and H). The auditory brainstem response (ABR) and distortion product otoacoustic emissions (DPOAE) tests showed no hearing loss in injected ears compared with uninjected controls (, I and J). 12 9 Supplemental Figure 3F Supplemental Figure 3 Supplemental Figure 3

In summary, AAV2 effectively transduces 100% of IHCs with robust transgenes expression. However, it transduces OHCs less efficiently, both in transduction rate and expression level. AAV2 administration through RWM + CF does not impair normal hearing in adult mice.

To evaluate the impact of AAV2–SaCas9–sgRNA-1–mediated CRISPR/Cas9 editing on HC survival, we examined HC numbers at 9 months after injection at 4 weeks of age inmice. Compared with WT inner ears, uninjectedears had significant loss of HCs across all cochlear turns, as shown by immunolabeling and quantification, with complete loss in the basal turn and a partial loss in the middle and apical turns (). In the injectedears, a complete loss of basal HCs was observed (). However, significantly more HCs, especially OHCs, survived in the middle and apical turns (), compared with the uninjectedears (). We conclude that AAV2–SaCas9–sgRNA-1–mediated editing markedly promotes long-term HC survival, especially in the middle and apical turns. P2rx2 V61L/+ P2rx2 V61L/+ P2rx2 V61L/+ P2rx2 V61L/+ Figure 3, A, B, D–I, M, and N Figure 3, C, L, M, and N Figure 3, C, J, K, M, and N Figure 3, B, G, H, M, and N

Mature IHCs have an elongated shape with a length of approximately 27 μm at 1 month and 31 μm at 4 months of age (). In contrast, we found that IHC length was shortened to approximately 21 μm between 1 and 4 months of age in adultmice (). To assess the effect of AAV2–SaCas9–sgRNA-1 on IHC length, we performed 3D reconstruction and analysis (). Nine months after injection at 4 weeks of age, IHC lengths in injected ears were approximately 9 μm longer in the apical turn and 18 μm longer in the middle turn compared with uninjected ears (, P and Q). These results suggest AAV2–SaCas9–sgRNA-1 significantly ameliorates IHC length abnormalities in adultmice. ^{41 41} Figure 3O Figure 3 P2rx2 V61L/+ P2rx2 V61L/+

The structure of HC stereocilia is critical for function and hearing, with P2X2 receptors most abundant on the stereocilia and cuticular plates facing the endolymph (,). To evaluate the effect of AAV2–SaCas9–sgRNA-1 on the stereocilia structure, we performed scanning electron microscopy to examine the apex-middle turn HC stereocilia of WT, treated, and untreatedears at 9 months after injection at 4 weeks of age. Hearing was rescued at 3 frequencies in the injected ears (). In WT ears, the stereocilia of OHC and IHC were well organized (). In contrast, uninjected P2rx2ears exhibited absent, partially missing, or disorganized OHC stereocilia (). The stereocilia of IHCs were present but lacked the shorter stereocilia (). In injected ears, the overall V-shape pattern of OHC stereocilia, as well as the structure, was maintained (). The stereocilia of IHC were present in the injected ears but were generally longer (). These scanning electron microscopy data reveal that AAV2–SaCas9–sgRNA-1 ameliorates HC stereocilia defects, especially in OHCs ofmice, supporting its potential to preserve HC morphology and function. ^{53 54} Figure 4, A and B Figure 4, C–F Figure 4, G and H Figure 4, I and J Figure 4, K and L Figure 4, M and N P2rx2 V61L/+ P2rx2 V61L/+ V61L/+

Themouse model had delayed-onset hearing loss at P21 and progressed to profound deafness over time (). This pattern mirrors the onset of hearing loss in patients with DFNA41, which typically is detected in the second decade of life (), suggesting that P21 could be an effective intervention time point. P2rx2 V61L/+ ^{41 55}

To assess the therapeutic potential of AAV2–SaCas9–sgRNA-1, we aimed to mimic the clinical treatment scenarios by treatingmice at 2 time points: (a) 4 weeks old (young adult) and (b) 2 weeks old (earliest detectable hearing). We injected 4 × 10vg AAV2–SaCas9–sgRNA-1 per ear via RWM + CF (). Auditory functions were evaluated at 1, 3, 6, and 9 months after injection using ABR and DPOAE across 6 frequencies from 5.66 to 32 kHz (). P2rx2 V61L/+ 9 Figure 5A Figure 5A

In the 4-week-injection group, 1 month after injection, a significant 6 dB reduction in ABR threshold was detected at 8.00 kHz in injected ears compared with uninjectedears (). No significant difference in DPOAE thresholds was detected between injected and uninjected ears (). P2rx2 V61L/+ Figure 5C Supplemental Figure 4A

At 3 months after injection, treatedears showed substantial hearing preservation, whereas untreated ears had further deterioration across all frequencies. Treated ears had significantly reduced ABR thresholds at 5.66, 8.00, and 11.32 kHz, with an average reduction of 13 dB (). DPOAE threshold was reduced by 11 dB at 11.32 kHz (). P2rx2 V61L/+ Figure 5D Supplemental Figure 4B

Six months after injection, more significant hearing preservation was observed at 5.66, 8.00, and 11.32 kHz (). At this point, ABR thresholds in the untreatedears exceeded 85 dB across all frequencies, whereas injected ears had an average reduction of 21 dB at these frequencies (). DPOAE threshold at 11.32 kHz was reduced by 10 dB (). Figure 5E Figure 5E Supplemental Figure 4C P2rx2 V61L/+

At 9 months after injection, hearing preservation persisted in treated ears (). Average ABR thresholds in untreated ears exceeded 95 dB across all frequencies, whereas injected ears had significant preservation at 5.66, 8.00, and 11.32 kHz, with an average ABR threshold reduction of 17 dB (). DPOAE thresholds at 11.32 kHz were reduced by 7 dB (). Figure 5F Figure 5F Supplemental Figure 4D

These findings demonstrate that AAV2–SaCas9–sgRNA-1 editing therapy provides robust and long-term preservation of hearing at specific frequencies inmice. P2rx2 V61L/+

To determine if earlier intervention could result in more efficacious hearing rescue, we injected AAV2–SaCas9–sgRNA-1 inmice at P14 and subsequently evaluated auditory function. One month after injection, injected ears had a significant 8 dB reduction in ABR threshold at 22.6 kHz compared with uninjectedears (). No significant difference in DPOAE thresholds was observed between injected and uninjected ears (). P2rx2 V61L/+ P2rx2 V61L/+ Figure 5G Supplemental Figure 5A

At 3 months after injection, ABR thresholds in injected ears were significantly reduced at all frequencies except for 32 kHz, with reductions ranging from 16 dB at 5.66 kHz to 11 dB at 22.6 kHz, averaging 14 dB across the 5 frequencies (). DPOAE thresholds in injected ears were significantly reduced at 11.32 and 16 kHz by 17 and 14 dB, respectively (). Figure 5H Supplemental Figure 5B

Six months after injection, ABR thresholds had further deteriorated in uninjected ears, and injected ears had significant reductions at 5.66, 8.00, 11.32, and 16.00 kHz, with an average reduction of 21 dB (). DPOAE thresholds were significantly reduced by 16 dB at 11.32 kHz and by 6 dB at 16.00 kHz (). Figure 5I Supplemental Figure 5C

At 9 months after injection, uninjected ears exhibited profound hearing loss with ABR thresholds exceeding 95 dB across all frequencies. In contrast, injected ears had significant ABR threshold reductions at 5.66, 8.00, 11.32, and 16.00 kHz, averaging 17 dB (). DPOAE threshold was significantly reduced at 11.32 kHz by 7 dB in the injected ears (). Figure 5J Supplemental Figure 5D

Finally, at 12 months after injection, uninjected ears had ABR thresholds over 100 dB at all frequencies. Injected ears, however, had significant ABR threshold reduction at 8.00 and 11.32 kHz, averaging 16 dB (). DPOAE threshold did not differ between injected and uninjected ears (). Supplemental Figure 6A Supplemental Figure 6B

To evaluate how AAV2–SaCas9–sgRNA-1 injection affects normal hearing, we injected AAV2–SaCas9–sgRNA-1 into 2-week-old WT CBA mice and studied their hearing. The ABR and DPOAE thresholds in the injected WT mice were indistinguishable from WT mice without injection up to 12 months after injection (,, A–D, and, A and B), demonstrating that the editing complex does not affect normal hearing, consistent with mutation-specific editing. Figure 5, G–J Supplemental Figure 5 Supplemental Figure 6

Consistent with ABR and DPOAE threshold improvement, significantly greater wave I P1 amplitudes at 11.32 kHz at 90 and 100 dB sound pressure levels (SPLs) were observed in 4-week-old injectedears 9 months after injection compared with untreated ears, in which P1 amplitudes were negligible (, and). Additionally, in treated ears 9 months after injection at 4 weeks of age, wave I latency at 11.32 kHz across 20 to 100 dB SPL was reduced to a pattern more closely resembling that of WT ears compared with untreated ears (). P2rx2 V61L/+ Figure 5, K and L Supplemental Figure 7 Figure 5M

Based on the results, we conclude that AAV2–SaCas9–sgRNA-1 injection into the adultinner ear effectively preserves auditory function over the long term. Earlier intervention further enhances therapeutic efficacy by broadening the range of frequencies preserved. P2rx2 V61L/+

P2X2 receptors protect against NIHL by mitigating cation burden from high-intensity noise stimulation (,).-mutant mice are more susceptible to noise exposure (>95 dB), resulting in permanent threshold shift due to defects such as disruption of HC ribbon synapses (). Indeed,mice exhibited greater sensitivity to NIHL thanmice under identical noise exposure conditions (97 dB at the octave band of 1–20 kHz for 2 hours) (, and, A–D), demonstrating remarkable similarities between mice and humans in terms of hearing deficits. ^{56 57 58} Figure 6, B–E Supplemental Figure 8 P2rx2 P2rx2 V61L/+ P2rx2 +/+

Because AAV2–SaCas9–sgRNA-1–mediated editing disrupts the V61L mutation responsible for increased NIHL sensitivity inmice, we hypothesized that this editing would protect the treated ears against NIHL. To test the hypothesis, we injected AAV2–SaCas9–sgRNA-1 intomice at P14. Four weeks later, the mice were exposed to 97 dB SPL at the octave band of 1–20 kHz for 2 hours, followed by a hearing test 2 weeks after noise exposure (). Controls included contralateral uninjected ears exposed to noise andmice without noise exposure. P2rx2 V61L/+ P2rx2 V61L/+ P2rx2 V61L/+ Figure 6A

In untreatedears, noise exposure led to significant hearing loss, as evidenced by elevated ABR thresholds at 5.66, 8.00, and 11.32 kHz compared with age-matchedmice without noise exposure (). In contrast, AAV2–SaCas9–sgRNA-1 injected ears had complete attenuation of NIHL, with the ABR thresholds virtually identical to those of nonexposedmice (). No significant change in DPOAE thresholds was observed before or after noise exposure (), indicating that the noise did not affect the OHC function. P2rx2 V61L/+ P2rx2 V61L/+ P2rx2 V61L/+ Figure 6, B and D Figure 6, C and D Figure 6E

To further assess the optimal timing for genome editing intervention against noise sensitivity, we injected AAV2–SaCas9–sgRNA-1 into 4-week-oldmice, followed by noise exposure 4 weeks later at 97 dB SPL at the octave band of 1–20 kHz for 2 hours, followed by hearing tests 2 weeks later (). Again, injection in 4-week-oldmice significantly attenuated ABR threshold shifts at 8.00 and 11.32 kHz compared with uninjected ears (, A and B). No significant changes in DPOAE thresholds or hair cell loss were observed after noise exposure (, C–E). Thus, editing in oldermice partially protects against NIHL. P2rx2 V61L/+ P2rx2 V61L/+ P2rx2 V61L/+ Supplemental Figure 8A Supplemental Figure 9 Supplemental Figure 9

Ribbon synapses, essential for converting mechanical signals from HCs to electrical signals for auditory neurons, are highly vulnerable to noise exposure. CtBP2, a presynaptic protein, serves as a marker of synaptic integrity (,). In noise-exposed WT mice, loss of ribbon synapses was observed despite no threshold shift, compared with nonexposed WT controls (, E–G). In contrast, injectedears exposed to noise exhibited significantly higher CtBP2 counts per IHC in the apex turn compared with uninjected noise-exposed ears (). This finding demonstrates that editing inprotected against synapse loss. ^{59 60} Supplemental Figure 8 Figure 6, F–H P2rx2 V61L/+ P2rx2 V61L/+

These results demonstrate that AAV2–SaCas9–sgRNA-1 treatment effectively attenuates heightened sensitivity to NIHL inmice, with improved outcomes after early intervention. This strongly supports a promising therapeutic strategy to mitigate NIHL in patients withmutations. P2rx2 V61L/+ P2RX2

Our previous study demonstrated thatmice exhibited vestibular dysfunction (). To assess whether AAV2–SaCas9–sgRNA-1–mediated CRISPR/Cas9 editing improves vestibular HC survival and subsequent vestibular functional recovery, we first evaluated AAV2 transduction efficiency in vestibular organs. We injected an AAV2-GFP vector into the cochleae of adult WT mice via the RWM+CF approach and examined GFP expression in vestibular HCs. Moderate GFP transduction was observed in vestibular HCs (), suggesting that AAV2–SaCas9–sgRNA-1 inner ear delivery could potentially preserve some vestibular function inmice. P2rx2 V61L/+ P2rx2 V61L/+ ⁴¹ Supplemental Figure 10

To further investigate feasibility, we evaluated the impact of AAV2–SaCas9–sgRNA-1–mediated CRISPR/Cas9 editing on utricle HC survival. We quantified utricle HC numbers at 9 months after injection in 4-week-oldmice (). In uninjectedears, approximately 70% of utricle HCs were lost compared with the WT utricle, as shown by immunolabeling and quantification (). In contrast, in injectedears, significantly more HCs were preserved, with approximately 30% more HCs compared with the uninjected utricle (). P2rx2 V61L/+ P2rx2 V61L/+ P2rx2 V61L/+ Figure 7, A–E Figure 7, B, D, and E Figure 7, C and E

To assess vestibular function, we performed open field and rotarod tests 9 months after injection. In the open field test, untreatedmice displayed hyperactive behavior, including increased activity across the entire field and more frequent full-body rotations compared with untreated WT mice (). In contrast, injectedmice showed normalized behavior, preferring to explore along the border of the chamber with minimal full-body rotations, similar to WT mice (). Treatedmice also had decreased velocity and travel distance compared with untreatedmice, aligning more closely with WT levels (). P2rx2 V61L/+ P2rx2 V61L/+ P2rx2 V61L/+ P2rx2 V61L/+ Figure 7, F and H Figure 7G Figure 7, I and J

In the rotarod test, uninjectedmice displayed poor rotarod performance in maintaining their balance within the allocated time (). In contrast, injectedmice showed partially improved rotarod performance, comparable to that of WT mice (). P2rx2 V61L/+ P2rx2 V61L/+ Figure 7K Figure 7K

Collectively, these results from the open field and rotarod tests indicate AAV2–SaCas9–sgRNA-1 treatment effectively rescues vestibular function inmice over the long term. P2rx2 V61L/+

Our ultimate goal is to develop an editing strategy to rescue hearing in patients with DFNA41. To design gRNAs to abolish theV60L mutation in the human genome, we tested gRNA with different editors in patient-derived human induced pluripotent stem cells (hiPSCs) harboring themutation (). We screened various compact CRISPR systems that can be accommodated in a single AAV vector, including SaCas9-KKH (), enosCas12f1 (), and Cas12j-8 () (). Plasmids encoding CRISPR nucleases and sgRNAs were introduced into hiPSCs using nucleofection. NGS analysis revealed robust indel generation in cells edited with SaCas9-KKH/sgRNA-1, with an indel frequency of 37.9% ± 2.3% (). Indel profiling further revealed that in reads edited with SaCas9-KKH/sgRNA-1, a majority of indels were frameshift mutations (). P2RX2 P2RX2 c.178G > T ^{55 61 62 63} Figure 8A Figure 8B Figure 8, C and E

We performed an editing study in hiPSCs derived from a healthy control individual. We did not detect indels in the control hiPSCs (), demonstrating that our gRNA design is mutation specific and does not edit the normal allele. In comparison, the other compact CRISPR systems exhibited substantially lower on-target genome editing efficiency, with enosCas12f1/sgRNA-2 at 4.6% ± 2.2% and Cas12j-8 at less than 3% (). Neither enosCas12f1/sgRNA-2 nor Cas12j-8 edited the WT allele (). We further tested the top 10 potential off-target loci in the human genome, and NGS revealed no indels (). Collectively, these data demonstrate that we have identified a gRNA for SaCas9-KKH that enables efficient and specific editing of humanc.178G > T (V60L) dominant mutation, with minimal off-target effects. Figure 8, B and D Figure 8B Figure 8B Figure 8, F and G P2RX2

Our study examined male and female animals. Similar findings are reported for both sexes.

mice were maintained on a CBA/J background. A female mouse carrying ac179 G>C (p.V61L) was used for colony expansion by mating with a WTmale mouse, as described previously (). The mice were housed in groups of 2–4 per cage on a 12:12 hour–light-dark cycle and allowed free access to food and water. All the animals were maintained under standard conditions with a room temperature range of 22°C ± 2°C and humidity range 55% ± 10%. P2rx2 V61L/+ P2rx2 P2rx2 +/+ ⁴¹

U6-sgRNA sequence and SaCas9 (Addgene, 61591) () cDNAs were acquired from Addgene. Vectors for in vitro screening were constructed via Gibson assembly (New England Biolabs, E2611S) based on pMax-SpCas9 (,). AAV vectors are based on pX601 (Addgene, 61591) (). Plasmids encoding recombinant AAV genomes were cloned by Gibson assembly. All plasmids were purified using Plasmid Plus Miniprep or Maxiprep kits (Qiagen). ^{84 85 86 84}

AAV plasmids containing CMV-SaCas9-PA cassette, U6-sgRNA-1, or U6-sgCtrl cassette, were sequenced before packaging (Massachusetts General Hospital DNA Core) into AAV2/2 (). AAV vectors were produced by the Mass Eye and Ear vector core. The vector titer was 4.46 × 10vg/mL for AAV2–SaCas9–sgRNA-1, as determined by qPCR specific for the ITR of the virus. Supplemental Figure 1 12

Cochleae were harvested with the sensory epithelia dissociated using needles under a microscope (Axiovert 200M, Carl Zeiss). Inner ear tissue was immersed in 1 μM FM 1-43FX (Thermo Fisher Scientific,) for 15 seconds at room temperature in the dark, then washed by Dulbecco’s PBS. The sensory epithelia were treated with 100 μL of 0.05% trypsin-EDTA (Thermo Fisher Scientific, 25300054) for 20 minutes. During incubation, the tissue was carefully dispersed into small or single-cell clusters using a 200 μL Eppendorf pipette tip. Cells were transferred into a 10 mm culture dish and placed under a fluorescent microscope (Zeiss) equipped with a camera. FM 1-43FX–labeled HCs were collected by a 10 indel Eppendorf pipette tip, with approximately 300 cells collected from single cochlea. HCs were transferred into a PCR tube and lysed by 5 μL QuickExtract DNA Extraction Solution (Lucigen) for genomic DNA extraction. The cell lysis was used for genomic PCR amplification using NEBNext Ultra II Q5 Master Mix (New England Biolabs, M0544S) with the following program: 1 cycle: 98°C for 5 minutes; 42 cycles: 98°C for 15 seconds, 60°C 20 for seconds, 72°C for 10 seconds; and 1 cycle: 72°C for 4 minutes; 4°C. From 400 to 800 ng of purified PCR product was used for Next-Generation Sequencing (NGS). Samples were sequenced by the Massachusetts General Hospital Center for Computational and Integrative Biology DNA Core to visualize CRISPR variants. F35355

LONZA 4D-Nucleofector was used for nucleofection of fibroblasts. Cells were digested by trypsin-EDTA (0.05%) (Thermo Fisher Scientific) and dispersed into single cells. A total of 100,000 cells were resuspended in 20 μL of P3 reagent of the P3 Primary Cell 4D-Nucleofector X Kit S (Lonza Bioscience, V4XP-3032) and nucleofected by program EH-100. Cells were then treated with puromycin 24 hours later for 2 days. Four days after nucleofection, genomic DNA was collected using QuickExtract DNA Extraction Solution (Lucigen) for PCR using NEBNext Ultra II Q5 Master Mix (New England Biolabs, M0544S), with purified PCR product used for NGS. Individual SEQ files were analyzed by CRISPResso2 () as instructed (). The sgRNA protospacer sequences can be found in. http://crispresso.pinellolab.org/submission↗ ⁸⁷ Supplemental Table 3

The RWM+CF was performed in mice at P14 or P28. The mice were anesthetized using intraperitoneal injections of ketamine (100 mg/kg) and xylazine (10 mg/kg). The postauricular area hair was shaved from 150% of the surgical area and disinfected with 10% povidone-iodine and 70% ethanol for 3 cycles. The AAV-GFP (Vector Laboratories, 7004) or AAV2–SaCas9–sgRNA-1 was injected into the inner ears oformice. A volume of 1–1.2 μL was injected into each cochlea. P2rx2 V61L/+ P2rx2 +/+

Cochleae were extracted, perfused with 4% paraformaldehyde, postfixed for 12 hours at 4°C, and decalcified in 0.12 M EDTA for 48 hours. Samples were dissected into 2–3 pieces spanning apex to base and stained with the following antibodies: (a) mouse anti-CtBP2 (1:200; BD Biosciences, 612044) and (b) rabbit anti-MYO7A (1:100; Proteus Biosciences, 25-6790), followed by Alexa Fluor–conjugated secondary antibodies (1:1,000). Confocal Z-stacks were acquired using high-numerical-aperture objectives and ×2 digital zoom with 0.30 μm z-steps.

ABRs and DPOAEs were recorded to test auditory ability. ABRs were elicited by tone pips at 5.66, 8.00, 11.32, 16.00, 22.60, and 32.00 kHz (5 ms duration; 0.5 ms cosrise–fall; alternating polarity;= 512 sweeps each). ABR threshold was defined by visual inspection of stacked waveforms at 20–100 dB in 5–10 dB steps. Wave 1 amplitudes were extracted using a semiautomated algorithm. Subsets of cases were analyzed by a second investigator blinded to the experimental treatment. DPOAEs were evoked using f1 and f2 tones (f2 matching ABR frequencies; f2/f1 = 1.2; L1 = L2 + 10 dB, 5 dB steps). The 2f1–f2 response was extracted from ear canal sound pressure. Threshold was computed by interpolation as the f2 producing a DPOAE of 5 dB SPL. 2 n

Awake mice were placed in a small wire mesh cage without restraints directly below the acoustic horn in a reverberant exposure chamber with a sound-delivery loudspeaker. A 1–20 kHz noise band was delivered at 97 dB SPL for 2 hours. The calibration was processed before each exposure session. The variation of sound level was within 1–1.5 dB.

For the open-field test, awake mice were placed in a 38 cm × 38 cm arena positioned inside a sound chamber with overhead LED lighting, all inside a dimmed room. The mice were tested 1 at a time inside the square open field and allowed to explore for 3 minutes. Behavior was recorded by a Canon EOS R5 camera and tracking videos were analyzed by EthoVision XT, measuring the traveled distance and velocity. Raw measurements for the traveled distance were recorded in centimeters. We recorded videos We recorded videos during daytime hours from 8:00 a.m. to 5:00 p.m. to ensure mice adapted to their new environment. Open-field assessments were conducted by tester blinded to the experiments, and the same tester performed the experiments throughout the day.

The rotarod test was conducted to assess the mice’s performance on a rod that began rotating at 5 rpm for 5 minutes to familiarize mice with the equipment. The following day, the animals were placed on the rods for a total of 5 trials, with a 3 minute resting period between each trial. The same tester conducted all the experiments throughout the day. The duration of time the animals could remain on the rod before falling off was recorded for each trial.

Statistical analysis was performed with Prism 10 software (GraphPad). Data are presented as mean ± SEM or SD.values less than 0.05 were considered statistically significant. All group sizes, statistical tests, andvalues are reported in the figures or the figure legends. P P

All procedures were performed in accordance with NIH guidelines for use and care of laboratory animals and were approved by the Massachusetts Eye and Ear IACUC (protocol 2022N000065).

The NGS data are available with the Sequence Read Archive accession number: SUB15099459. Values for all data points in graphs are reported in thefile. Supporting Data Values