Targeted genome editing restores auditory function in adult mice with progressive hearing loss caused by a human microRNA mutation

Abstract

Mutations in microRNA-96 (MIR96) cause autosomal dominant deafness-50 (DFNA50), a form of delayed-onset hearing loss. Genome editing has shown efficacy in hearing recovery through intervention in neonatal mice, yet editing in the adult inner ear is necessary for clinical applications, which has not been done. Here, we developed a genome editing therapy for the MIR96 mutation 14C>A by screening different CRISPR systems and optimizing Cas9 expression and the sgRNA scaffold for efficient and specific mutation editing. AAV delivery of the KKH variant of Staphylococcus aureus Cas9 (SaCas9-KKH) and sgRNA to the cochleae of presymptomatic (3-week-old) and symptomatic (6-week-old) adult Mir96^14C>A/+ mutant mice improved hearing long term, with efficacy increased by injection at a younger age. Adult inner ear delivery resulted in transient Cas9 expression without evidence of AAV genomic integration, indicating the good safety profile of our in vivo genome editing strategy. We developed a dual-AAV system, including an AAV-sgmiR96-master carrying sgRNAs against all known human MIR96 mutations. Because mouse and human MIR96 sequences share 100% homology, our approach and sgRNA selection for efficient and specific hair cell editing for long-term hearing recovery lay the foundation for the development of treatment for patients with DFNA50 caused by MIR96 mutations.

Fig. 1. Mir96^14C>A mutant mice present early-onset progressive hearing loss.

(A) Mir96 sequence in human (has-), mouse (mmu-), macaque (mne-), zebrafish (dre-), and 14C>A mutation. The blue text region shows 100% conservation. The mutated nucleotide in Mir96^14C>A is displayed in red. (B) Representative Sanger sequencing results showed the Mir96 mutation locus in WT mice, Mir96^14C>A/+, Mir96^14C>A/14C>A. The red arrow indicates the mutated nucleotide. (C to E) ABR thresholds in Mir96^14C>A/+ (het) mice (red) compared with WT mice (blue) and Mir96^14C>A/14C>A (homo) mice (black) at 4 weeks (C), 8 weeks (D), and 12 weeks (E), respectively. (F to H) DPOAE thresholds in Mir96^14C>A/+ ears (red) compared with WT ears (blue) and Mir96^14C>A/14C>A ears (black) at 4 weeks (F), 8 weeks (G), and 12 weeks (H), respectively. Values and error bars reflect mean ± SEM. Statistical analyses were performed by two-way ANOVA with Bonferroni correction for multiple comparisons: *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001. WT refers to WT C57BL/6N mice.

Fig. 2. Mir96 14C>A allele-specific genome editing using different CRISPR systems in mouse and human cells.

(A) Illustration of the Mir96 + 14 C>A mutation locus and sgRNA sequences. The mutated nucleotide in the Mir96^14C>A allele is displayed in red. The protospacer adjacent motif (PAM) nucleotides are displayed in blue. (B) Schematic overview of plasmid constructions for different CRISPR systems. (C) Illustration of primary fibroblast isolation from Mir96^14C>A/+ mice and subsequent genome editing and InDel analysis. (D) Bar chart showing InDel frequencies in Mir96^14C>A/+ and WT primary fibroblasts after genome editing using different Cas9/sgRNA combinations. n = 3 per treatment condition. Values and error bars reflect mean ± SD. (E and F) Representative NGS results from SaCas9-KKH/sgRNA-4 edited Mir96^14C>A/+ and WT primary fibroblasts. The red arrow indicates the double-stranded DNA cutting site. Reference sequence is the mutant allele. (G) Overview of human and mouse MIR96 + 14 C>A cell line generation; DNA fragment containing the MIR96 + 14 C>A mutation was integrated into the genome of different cell lines using PiggyBac transposons technology. (H) Bar chart showing the editing efficiency of Mir96 mutation locus and WT locus using SpCas9/sgRNA-1 and SaCas9-KKH/sgRNA-4 (n = 3). Each dot represents a unique sequencing reaction. Values and error bars reflect mean ± SD. (I and J) InDel profiles from SpCas9/sgRNA-1 and SaCas9-KKH/sgRNA-4 edited Mir96 + 14C>A HEK-293T cells. Negative numbers represent deletions, and positive numbers represent insertions.

Fig. 3. Optimization of CRISPR constructs for inner ear delivery.

(A) The design of the optimized AAV structure of SaCas9-KKH sgRNA vector with multiple NLS sites. (B and C) Sequence of unmodified and optimized SaCas9-KKH sgRNA, with sequence changes in bold. (D) Schematic view of the tdTomato reporter structure in the primary fibroblasts. (E) Representative fluorescence images of primary fibroblasts after editing by unmodified and optimized SaCas9-KKH/sgRNA systems. tdTomato⁺ cells are edited. Three technical replicates. (F) Bar chart showing the editing efficiency by the quantification of tdTomato⁺ cells after editing. Values and error bars reflect mean ± SD. Each dot represents one independent experiment.

Fig. 4. AAV2-CRISPR mediated targeted genome editing at the Mir96^14C>A locus in hair cells of Mir96^14C>A/+ mice.

(A) Schematic view of the AAV2-CMV-SaCas9-KKH-sgRNA4 design. (B) Timeline of in vivo studies. (C) Representative NGS results of cochlea samples from AAV2-SaCas9-KKH-sgRNA-4–injected and the contralateral uninjected ears of Mir96^14C>A/+ mice. Reference sequence is the mutant allele. (D) Quantification of InDel frequency from AAV2-SaCas9-KKH-sgRNA-4–injected, AAV2-SaCas9-KKH-sgCtrl–injected, and uninjected ears from Mir96^14C>A/+ mice, as well as AAV2-SaCas9-KKH-sgRNA-4–injected ears from WT mice (n = 9). Cochleae were collected 4 and 8 weeks after AAV injection. Each dot represents a unique sequencing reaction from a combination of three cochleae. Values and error bars reflect mean ± SD. (E) Schematic overview of the experimental protocol of hair cell isolation, cell lysis, and NGS. (F) Representative NGS result of isolated hair cells from AAV2-SaCas9-KKH-sgRNA-4–injected cochlea. Reference sequence is the mutant allele. (G) Quantification of Mir96^14C>A allele-specific InDel frequency from NGS of hair cell and cochlea samples after AAV2-SaCas9-KKH-sgRNA-4 injection (n = 3). Each dot represents a unique sequencing reaction from three cochleae combination. Values and error bars reflect mean ± SD. (H) Percentage of Mir96 WT allele reads, 14C>A reads, and InDel-containing reads in the NGS results from AAV2-SaCas9-KKH-sgRNA-4–injected hair cells from three independent experiments.

Fig. 5. AAV2-SaCas9-KKH-sgRNA-4 delivery into cochleae of 6-week-old Mir96^14C>A/+ mice promotes sustained rescue of hearing in adulthood.

(A) Timeline of AAV2-SaCas9-KKH-sgRNA-4 delivery into adult Mir96^14C>A/+ mouse cochleae and subsequent auditory function assays. (B to G) Frequency-dependent ABR and DPOAE thresholds in injected (blue) versus uninjected contralateral ears (red) at 16 weeks of age (B and C), 20 weeks of age (D and E), and 26 weeks of age (F and G). n = 10. (H to K) Representative ABR waveforms recorded from an injected (left) and an uninjected ear (right) of a mouse of 20 weeks of age in response to 11.3-kHz (H and I) and 16-kHz auditory stimulation (J and K). Single traces represent responses to different stimulation intensities [20 to 100 decibels (dB)]. The thresholds were determined by the detection of peak 1 (green color traces). Values and error bars reflect mean ± SEM. Statistical analyses were performed by two-way ANOVA with Bonferroni correction for multiple comparisons: *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001.

Fig. 6. AAV2-SaCas9-KKH-sgRNA-4 delivery into cochleae of adult Mir96^14C>A/+ mice promotes hair cell survival and maintenance of stereocilia integrity.

(A) Timeline of AAV2-SaCas9-KKH-sgRNA-4 delivery, confocal analysis, and SEM study of ears harvested at 10 and 14 weeks after injection, respectively. (B to E) Representative confocal z-stack images of whole-mount cochleae from uninjected (B and C) and AAV2-SaCas9-KKH-sgRNA-4–injected (D and E) Mir96^14C>A/+ mice. Hair cells were stained for MYO7A (green). Asterisks point to missing hair cells. Scale bar, 20 μm. Experiments were repeated independently in three cochleae. (F and G) Quantification and comparison of the number of OHCs (F) and IHCs (G) across the cochlear turns from injected and uninjected Mir96^14C>A/+ mice 10 weeks after injection. Uninjected WT mice were used as a control. Error bar represents SD. *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001. (H and I) SEM images of injected (H) uninjected (I) Mir96^14C>A/+ OHC bundles at the apical turn. The asterisks indicate the missing stereocilia in an OHC from an uninjected ear. Scale bar, 2 μm. (J and K) Images of SEM of injected (J) and uninjected (K) Mir96^14C>A/+ IHC bundles at the apex turn. Scale bar, 2 μm.

Fig. 7. Safety assessment of AAV2-SaCas9-KKH-sgRNA-4 delivery into cochleae of 3- and 6-week-old mice.

(A) qPCR of SaCas9-KKH expression in the injected cochleae and the contralateral uninjected cochleae in 6-week-old mice. Ctrl refers to the uninjected cochlea. Each dot represents data from a combination of two cochleae (n = 6). (B) Western blotting of SaCas9-KKH protein in both 3-week-old and 6-week-old injected cochleae and the contralateral uninjected cochleae. (C) Primer design for the AAV vector integration assay; the red arrows indicate the location of the primers; P1-F and P2-R were used for amplifying Mir96 loci in mouse genome; P3-F and P4-R were used for detecting AAV vector; P1-F and ITR-R were used for detecting AAV vector integration at Mir96 loci. (D) Quantification of InDel frequency and AAV vector ITR integration ratio from edited HEK-293T-Mir96^14C>A cells. The vertical dashed lines represent the optimal dosage range. (E and F) Gel image of the PCR products, showing the bands of the Mir96 locus, AAV vectors, and miR96-ITR integration in edited cochlea samples (E) and isolated hair cells from injected ears (F). Asterisk indicates the putative integration fragment position. (G) MiR96-ITR integration reads from NGS of isolated hair cells from injected ears. (H) qPCR analysis of Mir96 in injected and uninjected cochleae. Each dot represents an independent result from two cochleae combined (n = 6). (I) qPCR analysis of miR96-ITR RNA in injected and uninjected cochleae (n = 3). (J) CIRCLEseq analysis of SaCas9-KKH-sgRNA-4 in Mir96^14C>A/+ primary fibroblasts genomic DNA. (K and L) Quantification of InDel frequency of potential off-target sites in vitro (K) and in vivo (L). Values and error bars reflect mean ± SD.

Fig. 8. A dual-AAV system carrying multiple gRNAs targeting all known Mir96 seed region mutations in human cells specifically and efficiently.

(A) Sequence information of the human MIR96 locus and the sgRNA design to target three known seed mutations. Red: mutation site. Blue: the PAM nucleotides. (B) Schematic view of dual-AAV constructions. One contains the “U1a-SpCas9-polyA” cassette, and the other contains three U6-sgRNA cassettes. (C) Sequence of the optimized SpCas9 sgRNA; the bold letters indicate the changes compared with the unmodified sequence. (D and E) Representative NGS results from SpCas9/sgmiR96-master edited HEK-miR96 (15A to T) cells and SpCas9/sgmiR96-master edited HEK293T WT cells. (F and G) InDel profiles from SpCas9/sgmiR96-master edited HEK-miR96 (15A to T) cells. Negative numbers represent deletions, and positive numbers represent insertions. Experiments were repeated three times. (H) The InDel frequency in the HEK-miR96 mutation cell lines and WT HEK293T cells after genome editing using SpCas9/sgmiR96-master. Each dot represents an independent experiment. Values and error bars reflect mean ± SD. (I) Off-target analysis in SpCas9/sgmiR96-master edited HEK-miR96 (15A to T) cells.