A review of application of base editing for the treatment of inner ear disorders

Abstract

The development of applications for the CRISPR/Cas9 (clustered regularly interspaced short palindromic repeats/CRISPR-associated nuclease 9) system has increased greatly in recent years, especially in the area of gene therapy by efficient in vivo genome editing. Although great success has been achieved in repairing and rewriting genomes through homology-directed repair coupled with Cas9 nuclease cleavage, its in vivo efficiency is insufficient for gene therapy. Base editing is a next-generation genome-editing tool that does not involve double-stranded DNA breaks and uses components of the CRISPR system together with other enzymes to make point mutations directly in cellular DNA or RNA. Base editors, composed of an engineered deaminase and a catalytically impaired CRISPR/Cas9 variant, are powerful tools for targeted base editing in cells and organisms. In non-dividing cells, base editors can directly transform one base or base pair into another, efficiently installing a point mutation. Undesired by-products of editing are seldom generated during this procedure. Herein we review the different base-editing platforms, including their deaminase recruitment strategies and editing outcomes, and the in vivo delivery of base editors. Additionally, we summarize therapeutic applications of base editing in disorders of the inner ear.

Introduction

The CRISPR/Cas9 (clustered regularly interspaced short palindromic repeats/CRISPR-associated nuclease 9) system has revolutionized the life sciences since it was introduced into mammalian cell lines in 2013, and it opens up the possibility of gene therapy for human diseases. It was developed from an adaptive immune defense system that exists widely in bacteria which fights against invading viruses and exogenous DNA.^[1] The CRISPR/Cas9 system consists of a single-stranded guide RNA and a Cas protein with endonuclease activity that can localize and cleave the target nucleic acid sequence (pre-spacer sequence) via Watson-Crick base pairing, resulting in DNA double-strand breaks (DSBs). In response to DSBs, the integrity of the DNA can be restored by non-homologous end joining (NHEJ) or homology-directed repair (HDR). NHEJ leads the 2 strands of the break to rejoin directly without inserting templates, which is suitable for the research and treatment of dominant genetic diseases.^[2,3] However, NHEJ cannot be used in most genetic diseases because most genetic diseases require correction of a mutation at the target locus rather than disruption of the gene. In contrast, HDR can repair a defective gene by introducing an exogenous gene template into the chromosome of a recipient cell.^[4,5]

Cas9-mediated genome editing has greatly expanded many areas of research and has accelerated the generation of transgenic models.^[6,7] Numerous knock-out or knock-in animal models have been implemented in embryonic stem cells and zygotes.^[7–10] Originally, CRISPR/Cas9 was constructed to induce small insertions and deletions and later developed to induce point mutations and even chromosomal translocations.^[11] The rapid generation of animal models of diseases has provided scientists a convenient method to explore the mysteries of life and seek approaches to eradicate diseases that afflict humans. For example, researchers have used an adenovirus vector to deliver a Cas9 system targeting Pten into wild-type mouse liver, achieving abnormal liver function consistent with genetic Pten-deficient mice at 4 months.^[12] Microinjection of the CRISPR/Cas9 system into NOD/SCID IL2rgc null embryos (an immunodeficient mouse strain) enabled creation of an immunodeficient Col7a1^−/− mouse model of recessive dystrophic epidermolysis bullosa.^[13] Thus, the CRISPR/Cas9 tool facilitates the creation of cost-effective and large-scale rodent models for in vivo mutagenesis studies.

Currently, clinical genetic diseases can generally not be cured. Gene therapy on animal models with human genetic disorders provides a potential therapeutic experimental platform and, combined with in vivo CRISPR/Cas9–mediated genome editing of pathogenic mutations, is a promising approach for the treatment of genetic disorders.^[14] The first Cas9-based gene therapy in an animal model with a human genetic disease was performed using a CRYGC gene–related cataract mouse model. By injecting Cas9 mRNA and a single-guide RNA (sgRNA) into the zygote of Crygc heterozygous mutated mice, the cataract was cured successfully.^[14] Since then, numerous studies have demonstrated that Cas9-based genome editing is an efficient method to rescue disease phenotypes in models with human monogenetic diseases.^[15–18] For example, CRISPR/Cas9 has been applied to an mdx mouse model with Duchenne muscular dystrophy and a mouse model with hereditary tyrosinemia type I through NHEJ- or HDR-mediated gene editing, successfully correcting both diseases in genotype and phenotype.^[16,18] Dystrophic epidermolysis bullosa, a blistering skin disorder caused by mutations in the COL7A1 gene, represents another disorder edited using the CRISPR/Cas9 system and NHEJ-mediated gene disruption.^[15] Moreover, the CRISPR/Cas9 system has been used to disrupt the genes that encode T-cell signaling molecules or inhibit receptors to improve chimeric antigen receptor T-cell function.^[17]

Hearing impairment is a common public health problem, affecting 360 million people worldwide—about 5% of the world's population.^[19] Approximately 90% of hearing loss cases are related to sensorineural hearing loss, in which the root cause is in the cochlea or the auditory nerve.^[20] The auditory pathway and pathogenic factors of sensorineural hearing loss are complex and closely related to genetics and environment. Although hearing aids and cochlear implants are currently effective treatments for persons with deafness, there is no effective biological treatment.^[21,22] Many challenges exist in treating sensorineural hearing loss using traditional methods, and gene therapy may provide a solution to restore hearing with more natural sound perception. Gene editing has been shown to have potential for inner ear therapy. Injection of Cas9-guide RNA-lipid complexes into the cochlea of neonatal Tmc1^Bth/+ (Beethoven) mice to target the Tmc1^Bth allele of the transmembrane channel-like gene 1 (Tmc1) can significantly reduce progressive hearing loss.^[23] Locus-specific RNA-Cas9 ribonucleoproteins have been used on fibroblasts of a patient with Usher syndrome who carries a homozygous c.2299delG mutation, which highlights a new possibility for treatment of Usher syndrome.^[24] The ATOH1-2A-eGFP cell line was generated by CRISPR/Cas9 and exhibits electrophysiological properties similar to those of native sensory hair cells.^[25] The CRISPR/Cas9 gene editing tool has been used to correct the MYO15A mutation in induced pluripotent stem cells, repairing the morphology and function of the derived hair cell-like cells.^[26]

However, there are some obstacles to the use of CRISPR/Cas9, the major one being the low efficiency of HDR.^[27] Therefore, most applications of CRISPR/Cas9 are based on NHEJ induced by DSBs.^[27] Although HDR works precisely, it is not known how to improve its efficiency in vivo. Most hereditary diseases—especially those that cause deafness—are recessive, which theoretically cannot be treated by NHEJ. Even though many attempts have been made to improve the efficiency of HDR and decrease the use of the less-accurate NHEJ, it remains inefficient to modify point mutations using HDR, especially in nondividing cells. Alternative approaches that do not require DSBs are urgently needed to correct point mutations in genomic DNA.

To improve gene editing efficiency and reduce random insertions caused by DNA breaks, Liu and colleagues developed a DNA base editor (BE) that could convert one base directly to another at a programmable target locus without DSBs, which increases the efficiency of gene editing and prevents the insertion of random indels.^[28] Since then, researchers have improved and optimized this system and developed many variants. Chang Xing and colleagues developed the dead Cas9 (dCas9) activation-induced cytidine deaminase-P128X (AIDx) single base editing system, which works as a human cytidine deaminase that induces high-frequency mutations. Combined with inactivated Cas9 and uracil DNA glycosylase (UDG) inhibitors, activation-induced cytidine deaminase can induce specific C-to-T transitions.^[29] Base editing has been widely used in the life sciences because of its precision and high efficiency.

Database search strategy

We performed an electronic search of the PubMed database for literature describing the principle and development of base editing from 1989 to 2018 using the following search terms: base editing AND principle OR development, animal experimentation. The results were further screened by title and abstract to select only applications in the inner ear. Non-base-editing experiments and review articles were excluded.

In addition, we searched PubMed for applications of base editing. This included publications with the following search criteria: base editing, applications in vivo or in vitro, animal or plant experiments. Subsequent searches were completed that were specifically relevant to the inner ear. Articles that did not correspond to applications of base editing were excluded.

The principle of base editing

DNA BEs consist of a catalytic dCas9 nuclease and a deaminase enzyme that converts one base to another on single-stranded DNA (ssDNA). dCas9 contains Asp10Ala and His840Ala mutations that impair its ability, as a nuclease, to cleave the DNA backbone but retain the ability to bind DNA in a guide RNA-programmed manner.^[30] A small segment of ssDNA is displaced when the guide RNA binds the target DNA strand. Then, DNA bases located in this ssDNA region are modified by the deaminase enzyme^[31] (Fig. 1).

Figure 1: Mechanism of base editing. Dead Cas9 (dCas9) binds DNA in a guide RNA-programmed manner, without cleaving the DNA backbone. A transfer RNA (tRNA)-specific adenosine deaminase (TadA) domain convert the DNA base to thymine to cytosine. gRNA = guide RNA.

There are 2 types of DNA BE: cytosine BEs (CBEs) convert a C to a T^[28,32] and adenine BEs (ABEs) convert an A to a G.^[33] CBEs and ABEs jointly mediate all 4 transition mutations: C(G) to T(A) and A(T) to G(C).

Development of CBEs

CBEs were the first demonstrated BE and aimed to deaminate the target cytosine to generate uracil to convert a C•G base pair to a T•A base pair. CBE is mainly composed of cytidine deaminase, sgRNA, and mutant dCas9 protein.^[28] Cytidine deaminase binds to the target sequence by forming a fusion protein with dCas9.^[28] Cytidine deaminase converts base C to base U in the editing window.^[28] During subsequent replication, base U can be recognized by DNA polymerase and paired with base A, and then ultimately completes the conversion of base C to base T.^[28]

However, base editing is less efficient in vivo because of the mechanism of base-excision repair in cells: UDG catalyzes removal of base U from DNA.^[28] Subsequently, the U:G pair is reversed to a C•G pair, making editing inefficient.^[28] To improve the editing efficiency, UDG inhibitor, which inhibits the conversion of uracil to cytosine, is bound to the deaminase-Cas9 complex.^[28]

To further enhance editing efficiency, researchers replaced dCas9 with Cas9n (D10A), which can cleave a non-edited DNA strand containing a G base and inhibit the repair of a U:G pair to a C:G pair to improve editing efficiency and significantly reduce off-target mutations .^[28]

When UDG catalyzes the removal of U from the DNA strand, it results in an apurinic or apyrimidinic site (AP).^[28] AP lyase can convert AP into a ssDNA gap to break the DNA.^[28] In addition, Cas9n (D10A) can cleave a single DNA strand containing a G base, which may result in DSB and then produce NHEJ.^[28] To further reduce contact of UDG and base U, researchers have increased UDG inhibitor to 2 copies, resulting in improved editing efficiency and a reduction in the indel mutation rate, which substantially improves the purity and efficiency of the edited product.^[28]

Development of ABEs

In all possible base-pair exchanges, uneven distributions dominate in pathogenic point mutations in the living system. The common type of pathogenic single nucleotide polymorphism (SNP) is the mutation of a G•C base pair to an A•T base pair in the ClinVar database, existing in ∼47% of diseases.^[34] A deaminase capable of converting an A•T base pair into a G•C base pair is of great significance. However, it is still unknown whether adenosine deaminase enzymes can act on ssDNA. To solve this problem, scientists have developed a deoxyadenosine deaminase enzyme that accepts ssDNA starting from an Escherichia coli transfer (t)RNA adenosine deaminase enzyme.^[33] Just as CBEs can convert a C:G pair to T:A pair, the TadA enzyme combined with sgRNA can alter its base pairing preferences.^[12] It can deaminate adenine at 12 to 17 bp upstream of protospacer adjacent motif (PAM) to inosine, and then inosine can be paired with cytosine and replaced by guanine according to the Watson-Crick principle to achieve the mutation from A to G. Transfer RNA (tRNA)-specific adenosine deaminase 35 (TadA35). E. coli cells contain defective antibiotic resistance genes.^[33] To improve editing efficiency, researchers generated several ABE1.2 variants to resist higher concentrations of chloramphenicol.^[33] Two new point mutations were introduced to speculate location of a helix adjacent to the TadA tRNA substrate, D147Y and E155 V to ABE2.1 (ABE1.2 + D147Y + E155 V).^[33] In mammalian cells, ABE2.1 had higher editing efficiency than ABE1.2, resulting in an average of 11 ± 2.9% A•T to G•C base editing.^[33]

ABE could efficiently convert A:T pairs to G:C pairs after 7 rounds of optimization through directional evolution of the TadA enzyme and engineering process.^[33] The editing efficiency of ABE7.10 is up to 58 ± 4.0% within an editing window of protospacer positions ∼4 to 7 bp, whereas ABE7.9 offers higher editing efficiency at positions closer to the PAM.^[33] Thus, ABEs are a powerful new tool to precisely convert a target A•T base pair to G•C in genomic DNA.^[33]

Delivery of base editors

The goal of a base editing system is the safe and effective delivery of all its components into the nuclei of the target tissue for clinical implementation of gene editing. To date, in vivo delivery of BEs has used viruses, lipid nanoparticles, and direct nucleic acid injections. These delivery strategies are applied at various stages and each has advantages and risks.

Viruses are the most common vector to deliver BEs for both in vivo applications and therapeutic research. Adenoviruses may induce inflammatory responses,^[35] whereas adeno-associated viruses (AAV) have a negligible immune response. An AAV-carried BE plasmid is thought to be a promising delivery vector because of its diverse serotypes and ability to infect dividing cells. Package limit is a major barrier for AAV, but this problem could be solved by using two trans-RNA splicing AAVs.^[36] Scientists have delivered the dual trans-RNA splicing AAV-mediated ABE7.10 to muscle cells in a mouse model of Duchenne muscular dystrophy to correct a nonsense mutation in the Dmd gene, demonstrating the feasibility of AAV-delivered base editing in adult mice.^[37]

However, because of sustained overexpression of AAV-carried BEs, BEs might mediate off-target editing after successful editing to erode DNA integrity. Lipid nanoparticles have been proved a suitable alternative to overcome this barrier.^[38] Cationic lipid-mediated Cas9 complexes have accessed cells both in vivo and in vitro and showed higher DNA specificity than virus-mediated lipofection.^[23,39] The third-generation BE (BE3) protein has also been purified and packaged with sgRNA and liposomes into cultured cells and zebrafish embryos.^[40] BE:RNA:lipid was injected into mouse inner ear to install a mutation that induced mitosis of cochlear supporting cells and cellular reprogramming in Wnt activation.^[41] These results have established a precise approach to gene therapy in post-mitotic tissues and the inner ear of postnatal mice.^[41]

BE3 has also been delivered through direct nucleic acid injections into Xenopus laevis embryos.^[42] Site-specific conversion rates of up to 20.5% were achieved by BE3 ribonucleoprotein particle targeting the tyrosinase (tyr) gene in early embryos without off-target mutations.^[42] Ribonucleoprotein particle delivery and electroporation can deliver BE tools and edit genes with low off-targets mutations.

Application of base editing

With the new technologies described above, applications for base editing in research are increasing in medicine and in biotechnology. BEs enable the creation or removal of a single nucleotide variation of interest in animals, plants, and cell lines to reveal its phenotypic consequences in an isogenic background. As mutations are largely responsible for genetic diversity within a species, BEs are the perfect tool with applications in genetics, therapeutics, and agriculture.^[37,41,43]

As a tool for engineering agricultural traits

Compared with conventional breeding, base editing to generate novel plant mutants is highly efficient.^[43] In 2 crop plants, point mutagenesis at genomic regions specified by sgRNAs was applied by fusion of CRISPR/Cas9 and activation-induced cytidine deaminase. In rice, multiple herbicide-resistance point mutations were induced by multiplexed editing using herbicide selection, and a marker-free tomato was generated using homozygous heritable DNA substitutions.^[44] Thus, applying base editing to crop improvement is feasible.

Frequencies of up to 43.48% have been achieved in the targeted conversion of cytosine to thymine from position 3 to 9 within the protospacer by using a CRISPR/Cas9 nickase-cytidine deaminase fusion in both protoplasts and regenerated rice, wheat, and maize plants.^[45]

ABE editing has high efficiency in rice, which was confirmed by 2 independent reports.^[46,47] New ABEs and CBEs were developed with engineered Streptococcus pyogenes Cas9 and Staphylococcus aureus Cas9 variants that substantially expanded the targetable sites in the rice genome. Adenine and cytosine base editing can be executed simultaneously in rice and will be useful in rice functional genomics research, advancing precision molecular breeding in crops.^[46]

Based on an evolved tRNA adenosine deaminase fused to the nickase CRISPR/Cas9, a new plant ABE enables A•T to G•C conversion at frequencies up to 7.5% in protoplasts and 59.1% in regenerated rice and wheat plants.^[48] In addition to increasing the candidates for precise editing in plants, all base pairs are rendered editable by this new ABE system.

Screening gene knockouts with CRISPR-STOP and induction of STOP codons

Precision base editing can also be used to screen the effects of gene knockouts. Early stop codons at arginine, glutamine, and tryptophan residues generated by the CRISPR-STOP and induction of STOP codon (iSTOP) systems can effectively inactivate genes.^[49,50]

CRISPR-STOP was introduced recently and is a more efficient and less deleterious alternative to wild-type Cas9 for gene knockout studies. Early stop codons can be introduced in ∼17,000 human genes. The application for genome-wide functional screenings of CRISPR-STOP-mediated targeted screening was confirmed by its comparable efficiency to wild-type Cas9.^[49]

Gene inactivation by iSTOP was facilitated by Billon and co-workers,^[50] and access to a database of over 3.4 million sgRNAs for iSTOP is available, allowing 97% to 99% of genes in 8 eukaryotic species to be targeted. A restriction fragment length polymorphism assay that allows the rapid detection of iSTOP-mediated editing in cell populations and clones was also described by the authors.

As a therapeutic approach

The most common form of pathogenic genetic variant is the point mutation. Using the BE system, the precise conversion of cytidine to thymidine in the BE system indicates a potential therapeutic approach for the targeted insertion of beneficial mutations.

Base editing was first used to demonstrate the correction of Alzheimer's disease–associated mutations in from Apoe4 to Apoe3r in mouse astrocytes and Tp53 mutations in a mouse breast cancer line.^[28] Since then, numerous examples have been reported of BE-induced gene correction in cultured cells.

By using an engineered human APOBEC3A (eA3A) domain to reduce bystander mutations, the new strategy demonstrated preferential deamination of cytidines in specific motif hierarchies. When compared with the widely used BE3 fusion in human cells, the eA3A-BE3 fusion greatly reduced editing on cytidines in substance of other sequence, with the same activities on cytidine motifs. eA3A-BE3 corrects a human β-thalassemia promoter mutation with much higher (>40-fold) precision than BE3.^[51]

A recent study found that BE variants, with a narrower deamination window, could promote G-to-A conversion at one particular gene in human embryos, suggesting that curing genetic disease in human somatic cells and embryos will be achieved by using the BE system.^[52]

Homozygous mutants can be generated at a rate of up to 77% of embryos by direct injection of mRNA encoding BE3, YE1-BE3, or YEE-BE3 together with a guide RNA, which survive to the blastomere stage.^[52–54]

Another prevailing method for in vivo correction of pathogenic mutations in mouse disease models is viral delivery of BEs. In a mouse model of muscular dystrophy, scientists used AAV to deliver ABE7.10 with a guide RNA programmed to correct a premature stop codon in the dystrophin gene (Dmd). Dystrophin expression was restored in 17% of muscle fibers, although only 3.3% of sequenced cells were corrected. The conclusion is that low levels of editing do not decrease the efficiency of therapeutically relevant phenotypic changes.^[37]

Scientists have screened BEs for activity in cultured cells, including human-induced pluripotent stem cells. Then, they delivered the most efficient BE into the livers of adult mice to assess whether site-specific nonsense mutations could be introduced into the Pcsk9 (proprotein convertase subtilisin/kexin type 9) gene.^[55] In adult mice, reduced plasma Pcsk9 protein levels (>50%) as well as reduced plasma cholesterol levels (∼30%) were shown, with no evidence of off-target mutagenesis, cytosine-to-thymine edits, or indels.^[55]

Base editing in the inner ear

It has been shown that the small molecule compound CHIR99021, which is used to inhibit glycogen synthase kinase 3 activity, induces a significant increase in the fraction of proliferating sphere-forming cells labeled by transgenic mice reporting cell cycle progression (fluorescent ubiquitination-based cell cycle indicator) markers and in the percentage of leucine-rich repeat-containing G protein-coupled receptor 5-GFP (green fluorescent protein)-positive cells.^[56] Also, a significant increment in the fraction of proliferating Sox2 supporting cells was detected in whole-mount cultures of the organ of Corti after CHIR99021 treatment.^[56]

Subsequently, 1 study showed that ABE upregulates Wnt signaling, which induces mitosis of cochlear supporting cells and cellular reprogramming through nuclease-free base editing. The researchers installed an S33F mutation in β-catenin that blocks phosphorylation of β-catenin and impedes its degradation.^[41] BE3 treatment leads to cellular reprogramming of other cells into cochlear hair cell-like cells, which is identified by the dissection and staining of treated hair cells. This differs from the effect of treatment with the Cas9 nuclease and an HDR template. These results establish the ability of base editing to work in post-mitotic cells that are resistant to DSB-stimulated HDR.^[57,58]

Further study showed that hearing loss in a mouse model of human genetic deafness can be ameliorated by cationic lipid-mediated in vivo delivery of Cas9-guide RNA complexes. Tao and co-workers designed and validated genome editing agents that preferentially disrupt the dominant deafness-associated allele in the Tmc1 Beethoven (Bth) mouse model both in vitro and in primary fibroblasts. The mutant Tmc1^Bth allele differs from the wild-type allele at only a single base pair.^[23] Compared with uninjected ears or ears injected with control complexes targeting an unrelated gene, higher hair cell survival rates and lower auditory brainstem response thresholds were achieved in injected ears.^[23]

Three genes needed for normal hearing function—vGlut3 (Slc17a8), otoferlin (Otof), and prestin (Slc26a5)—were individually mutated by using CRISPR-stop. Later, they successfully suppressed the expression of Slc17a8, Otof, and Slc26a5. This study demonstrated that CRISPR-stop can efficiently generate single or triple homozygous F₀ mouse mutants, eliminating the need to breed mice, a labor-intensive process.^[59]

Compared with traditional HDR-based CRISPR/Cas9 genome editing methods, base editing systems possess numerous advantages. Insertion and deletion rates, which may disrupt the function of the target gene through NHEJ, are significantly lower during base editing.^[28] Furthermore, base editing constructs (eg, YEE-BE3) have been devised to decrease the number of undesirable substitutions by narrowing the editing window to 1 to 2 bases.^[60] Greater specificity of base editing is demonstrated by the much lower frequency of off-target mutations.^[60] The continued development of base editing tools that maximize efficiency and targeting scope, while minimizing off-target base editing, will continue to yield opportunities for sophisticated applications including in inner ear therapy.^[22]

Acknowledgements

None.

Author contributions

XZ and ZS wrote the manuscript. WH and YT designed and supervised the review, and wrote the manuscript. All authors edited the manuscript.

Financial support

HW was supported by the Key Project of National Natural Science Foundation of China (No. 81330023) and the National Key Technology Research and Development Program of the Ministry of Science and Technology of China (No. SQ2017YFSF080012), Shanghai Key Laboratory of Translational Medicine on Ear and Nose Diseases, China (No. 14DZ2260300), Innovative Research Team of High-level Local Universities in Shanghai, China. YT was supported by National Natural Science Foundation of China (Nos. NSFC81800900), Shanghai Science and Technology Committee, China (Nos. 18411953600, 18ZR1422100, 18PJ1406900), Shanghai Municipal Health Commission, China (Nos. 2018YQ59), Shanghai Ninth People's Hospital, China (No. QC201804).

Conflicts of interest

The authors declare that they have no conflicts of interest.