Rescue of hearing by adenine base editing in a humanized mouse model of Usher syndrome type 1F

Abstract

Usher syndrome type 1F (USH1F), characterized by congenital lack of hearing and balance and progressive loss of vision, is caused by mutations in the PCDH15 gene. In the Ashkenazi population, a recessive truncation mutation accounts for a large proportion of USH1F cases. The truncation is caused by a single C→T mutation, which converts an arginine codon to a stop (R245X). To test the potential for base editors to revert this mutation, we developed a humanized Pcdh15^R245X mouse model for USH1F. Mice homozygous for the R245X mutation were deaf and exhibited profound balance deficits, while heterozygous mice were unaffected. Here we show that an adenine base editor (ABE) is capable of reversing the R245X mutation to restore the PCDH15 sequence and function. We packaged a split-intein ABE into dual adeno-associated virus (AAV) vectors and delivered them into cochleas of neonatal USH1F mice. Hearing was not restored in a Pcdh15 constitutive null mouse despite base editing, perhaps because of early disorganization of cochlear hair cells. However, injection of vectors encoding the split ABE into a late-deletion conditional Pcdh15 knockout rescued hearing. This study demonstrates the ability of an ABE to correct the PCDH15 R245X mutation in the cochlea and restore hearing.

Keywords: AAV; PCDH15; Usher syndrome; base editor; blindness; cochlea; deafness; gene editing; gene therapy; hair cell.

Graphical abstract

Figure 1

Generation of a Pcdh15^R245X Usher 1F mouse model with a human target sequence (A) An oligonucleotide containing 60 bases of the human PCDH15 sequence bearing the USH1F R245X (c.733 C→T) mutation, flanked by homologous sequences of mouse Pcdh15 in exon 9, was co-injected into C57BL/6 mouse embryos with the SpCas9 enzyme and gRNAs targeting the mouse Pcdh15 locus (5′-CCGTGCACAAAATCTGAA-3′ and 5′-CCTCACAGTAGATGTTCTAGATG-3′). Aside from the mutation, there are four nucleotide differences in this segment between human and mouse, but they are silent, so the encoded protein is the same. (B) Auditory brain stem response (ABR) thresholds of Pcdh15^R245X/+ heterozygous and Pcdh15^R245X/R245X homozygous mice at P30 in response to an 8-kHz tone. Homozygous mice have no response at the highest intensity presented. (C) ABR of Pcdh15^+/+, Pcdh15^R245X/+, and Pcdh15^R245X/R245X mice for pure tone and click stimuli. Heterozygotes hear normally, but homozygotes have no response at any frequency. (D) Vestibular deficit in Pcdh15^R245X/R245X mice at age P30, observed as circling in an open field locomotion assay. Mice were placed in a featureless box and monitored for 5 min, and their track was analyzed for rotation, distance, and angular velocity using Ethovision software. Homozygous mutant mice circled at more than seven times the rate of heterozygotes. (E) Rotational velocity in the open field assay. Homozygous mutant mice circled twice as quickly as heterozygotes. Mice homozygous for the R245X mutation are deaf and display circling behavior. Error bars are standard error of the mean. ∗p < 0.05, ∗∗∗p < 0.001.

Figure 2

Severe functional and morphological deficits Pcdh15^R245X/R245X KO mice at P6 (A) Bundle morphology in phalloidin-stained Pcdh15^R245X/+ and Pcdh15^R245X/R245X cochlear hair cells. Severely disorganized hair cell bundles were observed in homozygous mutants (right) compared with heterozygote controls (left) of P6 mice. (B) Bundle morphology in vestibular hair cells from the utricle. Homozygous mutants had disorganized stereocilia. (C) Anti-PCDH15 labeling (magenta) of P6 OHCs in control mice along with phalloidin co-staining (cyan) demonstrated normal PCDH15 localization to the stereocilium tips in Pcdh15^R245X/+ mice (left), which is absent in Pcdh15^R245X/R245X mice (right). (D) Scanning electron micrographs of Pcdh15^R245X/+ mouse organ of Corti controls show largely normal OHC and IHC stereocilia morphology. Bundles have a regular appearance, with stereocilia aligned in rows. Tip links connect the tips of adjacent stereocilia along the hair bundles’ axis of sensitivity (arrows). (E) In Pcdh15^R245X/R245X mutants, the bundle structure was disrupted, with shortened stereocilia and gaps in the rows. While the lateral links were present, no tip links were detected. (F and G) At P6, FM1-43 dye uptake was abolished in Pcdh15^R245X/R245X mice, indicating no open transduction channels. Scale bars: 5 μm (A–C), 1 μm (D and E, top panels) and 200 nm (D and E, bottom panels), and 10 μm (F and G). Mice homozygous for the R245X mutation display disorganized hair bundles and lack MET channel functionality.

Figure 3

Efficiency of editing in HEK293 cells transfected with plasmids encoding full-length editors (A) Comparison of base editing by five different editors at three genomic sites in normal HEK293T cells. (B) Base editing window efficiency of editors on the genomic sites used in editor selection. (C) Coding and complement sequences for the human PCDH15 knockin segment with the C>T nonsense mutation (G>A on the complement strand) underlying R245X. Three gRNAs are shown that target an ABE to the complement strand, where the mutated base is adenosine. Numbering is shown for gRNA1 with the 5'-TGG-3' PAM. (D) Comparison of base editing of the PCDH15 R245X mutant nucleotide in HEK293T cells harboring the R245X target sequence. ∗∗∗∗∗∗∗p < 10⁻⁷. (E) Editing efficiency at different times post transfection (normalized to efficiency at 48 h). (F) Effect of gRNA:editor molar ratio on editing efficiency. There were no significant differences. (G) Effects of lengthening (+) or shortening (−) the spacer length on editing efficacy of ABEmax and ABE8e for the R245X target. Numbers indicate the lengthening or shortening of the g1 20bp gRNA. The ABE8e editor performed significantly better on the R245X target sequence despite all editors working similarly on the target sequences used to isolate and develop the original editors. One-way ANOVA detected no significant difference between ABEmax and ABE8e guides (compared within same base editors); the g2 and ALT are significantly different. (H) Editing efficiency at the R245X mutation by editors with different Cas9 variants, using the most efficient gRNA for the targeted PAM. Although the editors had similar editing efficiencies at the genomic sites on which they were first selected, they showed a different editing ability on the R245X site. All comparisons were significant at p < 10⁻⁴. (I) Editing window efficiencies of ABE8e-SpRY using guides that place the A17 targeted base at positions 12–16+ PAM. gRNA1 vs. other gRNAs was significant at p < 0.01. All error bars indicate standard error of the mean.

Figure 4

Base editing in HEK293T cells transduced with PCDH15 R245X reporter constructs (A) Diagrams of the PCDH15-FLAGMYC and FlashLight constructs. (B) Western blot of PCDH15-FLAGMYC after transfection with full-length base editors and gRNAs targeting the R245X mutation. (C) Fluorescence of FlashLight HEK293T cells after editing. Many cells express mCherry, which does not require editing. Some cells—especially with ABE8e—express GFP as well, which does require editing. (D) Western blot of PCDH15-FLAGMYC after transfection with split-intein base editors and gRNAs targeting the R245X mutation. (E) Fluorescence of FlashLight HEK293T cells after editing. (F) Flow cytometry of FlashLight R245X cells after editing. Unedited cells express just mCherry (red); edited cells express mCherry and GFP (green). The proportion of cells edited was highest when the editor was transfected as either full-length (ABE8e) or intein-linked (Int-ABE8e) ABE8e. Significance: ∗p < 0.05, ∗∗p < 0.01, others N.S. (G) Quantification of editing in FlashLight HEK293 cells by sequencing. Editing was highest at the targeted adenine at position A17, although A10 was also edited. For the PCDH15 coding sequence, editing at A10 causes a silent mutation. ABE editor effects were quickly observable using the reporter cell lines, which paralleled the results of NGS. Error bars indicate standard error of the mean.

Figure 5

R245X DNA editing in HEK293 FlashLight cells and in the Pcdh15^R245X mouse cochlea by dual AAV delivery of intein-linked base editors (A and B) Editing in HEK293 cells carrying the FlashLight reporter. Dual AAVs encoding ABEmax or ABE8e were added to cultures at a multiplicity of infection of 10⁶. The number of GFP-fluorescent cells, indicating successful editing, was much larger with ABE8e. (C) DNA editing in the Pcdh15^R245X mouse cochlea after injection of intein-linked ABEmax or ABE8e. DNA from whole cochlea was sequenced. Editors performed equally well, with no significant difference. (D) cDNA sequencing to detect editing in hair-cell Pcdh15 mRNA. ABE8e performed about twice as well as ABEmax. Significance: ∗p < 0.05. (E) Failure to rescue ABR thresholds with dual-AAV editor delivery in the Pcdh15^R245X/R245X mouse model. Mice receiving dual AAVs encoding intein-linked ABEmax or ABE8e showed auditory sensitivity, as assessed by ABR at P30, no better than the deaf untreated mice (n = 3 ABE injected, n = 4 uninjected). All error bars indicate standard error of the mean.

Figure 6

Rescue of hearing by ABEmax and ABE8e in the late-deletion Pcdh15^R245X/fl, Myo15-Cre mouse model (A) Normal ABR responses at P30 to 8-kHz tones in an untreated Cre-control mouse (Pcdh15^R245X/fl, Myo15-Cre-). The threshold is indicated with a green arrowhead. (B) ABR in an untreated KO mouse (Pcdh15^R245X/fl, Myo15-Cre+). (C) ABR in a Cre-control mouse (Pcdh15^R245X/fl, Myo15-Cre-), treated with dual AAV delivery of ABEmax. (D) ABR in a Cre-control mouse, treated with dual AAV delivery of ABE8e. (E) Rescue of a Cre+ KO mouse (Pcdh15^R245X/fl, Myo15-Cre+) with dual AAV delivery of ABEmax. The threshold, at about 80 dB, is about 40 dB improved from the untreated mouse. (F) Rescue of a Cre+ KO mouse with dual AAV delivery of ABE8e. The threshold, at about 100 dB, is slightly improved from the untreated KO. (G) ABR threshold plots for all conditions. Overall, ABEmax delivered the best rescue. In Cre- and WT control mice, it showed some elevation of the threshold at high frequencies. (Cre+, uninjected n = 4; Cre+, ABE8e n = 15; Cre+, ABEmax n = 15; WT, ABEmax n = 6; Cre−, ABEmax n = 15; Cre−, ABE8e n = 7; Cre−, uninjected n = 4). In Pcdh15^R245X/fl, Myo15-Cre+ mice, ABRs from ABEmax- and ABE8e-injected Cre+ mice were significantly different at each frequency measured, except for 22.6, 32, and 45 kHz (p < 0.05). (H) Durability of rescue in older mice. Mice for the durability study were selected as the best-rescued mice of (G). Rescue of hearing by ABEmax and ABE8e persisted at ages P60–P80, although to a lesser extent than at P30. n = 4 mice for each condition. ABRs were measured in the injected ears. All error bars indicate standard error of the mean.