The cytosine base editor (CBE), an evolution of the CRISPR/Cas9 system, known as the base editor (BE), consists of two essential components: a single-guide RNA (sgRNA) and a fusion protein comprising a Cas9 variant lacking double-strand cleavage activity (dCas9 or nCas9) and a cytosine deaminase enzyme (Komor et al., 2016). dCas9 (dead Cas9) and nCas9 (nickase Cas9) are generated through mutations in the RuvC and HNH domains of the Cas9 nuclease (Jinek et al., 2012). dCas9 preserves DNA binding without cleaving the backbone, while nCas9 exclusively cleaves one DNA strand, causing a single-strand break. By using Cas9 proteins that have lost cleavage activity or have only one strand of cleavage activity to achieve targeted base substitutions at the target site, the base editing system does not rely on double-strand breaks formation. The fusion protein, guided to genomic DNA by sgRNA, enables cytosine deaminase to transform cytosine (C) into uracil (U), which is then changed to thymine (T) during DNA repair or replication, effectuating a C to T base conversion (Figure 2B) (Komor et al., 2016). Cytidine deaminases in nature predominantly exhibit enzymatic activity on RNA. In the quest to identify a suitable DNA cytidine deaminase for base editing applications, researchers conducted evaluations on the single-stranded DNA (ssDNA) deamination activities of various naturally occurring cytidine deaminases (hAID, hAPOBEC3G, rAPOBEC1, and pmCDA1). Among these variants, rAPOBEC1, originating from rats, exhibited the highest deamination activity on ssDNA. The inaugural base editor, BE1 (rAPOBEC1–XTEN–dCas9), integrates rAPOBEC1 attached to the N-terminus of dCas9 via an XTEN linker, composed of 16 residues (Komor et al., 2016). In vitro, BE1 exhibits a base editing efficiency between 25% and 40%, with its activity covering roughly a 5-nucleotide (nt) window, initiating from the fourth to eighth position when counted from the most distal end of the protospacer adjacent motif (PAM) sequence. In light of the notably low base editing efficiency of BE1 in mammalian cells, quantified at a mere 0.8% to 7.7%, it is postulated by the scientists that this inefficiency may be attributed to the action of uracil DNA glycosylase (UDG) (Komor et al., 2016). UDG is theorized to identify and act upon the intermediate UG base pair, restoring it to its original CG configuration through the mechanism of the base excision repair pathway (Kunz et al., 2009; Komor et al., 2016). This hypothesized intervention by UDG could be the underlying cause for the significant reduction in base editing efficiency observed within cellular environments. David Liu and colleagues ingeniously integrated a uracil glycosylase inhibitor (UGI) to the C-terminus of BE1 to create a second-generation base editor, BE2 (rAPOBEC1-XTEN-dCas9-UGI), to inhibit the base excision repair pathway and improve the editing efficiency of the cytosine editor (Komor et al., 2016). BE2 demonstrated high editing efficiency in human cells, up to 20%, which was 3-fold higher than BE1. To overcome the theoretical limitation of BE2, which is constrained by its exclusive modification of the C base in the CG base pair, limiting the maximum theoretical C-to-T conversion efficiency to 50%, researchers developed the third-generation base editor BE3 (rAPOBEC-XTEN-nCas9-UGI), substituting dCas9 with nCas9 (Komor et al., 2016). This advancement led to a significantly enhanced editing efficiency in cutting the non-edited strand of BE3, 2–6 times greater than BE2, achieving an impressive editing efficiency of approximately 37%.

In human pathogenic point mutations, the majority involve the conversion of CG to TA base pairs, representing almost half of all mutations, while only about 14% are AT to GC mutations (Gaudelli et al., 2017). This predominance highlights the immense potential of developing base editors capable of converting AT to GC pairs. In 2017, David Liu’s laboratory made a significant advancement in this realm by developing adenine base editors (ABEs) using protein evolution and engineering techniques (Gaudelli et al., 2017). These ABEs can effectively convert AT base pairs to GC pairs within genomic DNA without necessitating DNA cleavage. They are particularly promising for correcting a wide range of single nucleotide polymorphisms related to human diseases. Notably, ABEs demonstrate superior editing efficacy and a diminished incidence of nonspecific genomic alterations, compared to the earlier base editor BE3, marking a significant leap in the performance and application of base editing technology.

The core components of ABE consist of nCas9 and a synthetically evolved adenosine deaminase, enabling the direct conversion of AT base pairs to GC base pairs (Figure 2B) (Gaudelli et al., 2017). ABE functions by integrating nCas9 with adenine deaminase, which is directed by sgRNA to target specific genomic DNA sites. During the enzymatic process, the adenine deaminase acts on ssDNA, converting A to inosine (I) within a specific range. This inosine is subsequently read or replicated as G by a polymerase, culminating in the direct transformation of AT base pairs into GC base pairs. A pivotal challenge in developing ABE was the innate limitation of natural adenine deaminases, which typically modify RNA rather than ssDNA (Harris et al., 2002). Then, Researchers selected TadA, an adenine deaminase from Escherichia coli, for extensive evolutionary engineering. This process led to the creation of a modified enzyme with the novel ability to act on ssDNA. The resultant ABE system, named ABE7.10, incorporates this engineered enzyme (ecTadA-ecTadA*-nCas9) and demonstrating higher efficiency and broader applicability within human cells (Gaudelli et al., 2017). The ABE7.10 system features an editing window spanning positions 4 to 9 of the sgRNA. In human cells, it achieves an impressive editing efficiency of approximately 58%, with product purity reaching 99%. A notable feature of ABE7.10 is its exceptionally low off-target activity, reported to be under 0.1%, indicating high precision and specificity in genomic editing. These characteristics of ABE, particularly its high efficiency and specificity, underscore its potential as a transformative tool in genetic research and therapeutic applications, expanding the horizons of genomic engineering.

The development of the ABE system, particularly ABE7.10 and its evolved versions, has significantly advanced the field of genomic engineering in both mammalian and non-mammalian organisms, including plants. ABE7.10 is recognized for its precise base replacement capability and minimal indels, thus optimizing this tool focuses on improving editing efficiency, expanding its activity window, and broadening its genomic editing scope. One significant advancement in this regard is ABEmax, which outperforms ABE7.10 in terms of editing efficiency (Koblan et al., 2018). ABEmax incorporates varying amounts of NLSs and optimized codon sequences, leading to a substantial boost in the editing capabilities of ABE7.10. Further developments led to CP1012-ABEmax, CP1028-ABEmax, CP1041-ABEmax, and CP1249-ABEmax, created by modifying CP-Cas9 endonuclease with bis-bpNLS and codon optimization (Huang et al., 2019; Oakes et al., 2019). These variants not only maintain the editing efficiency equivalent to ABEmax, but also expand the editing activity window from positions 4 to 9 to positions 4 to 12. While the CP-ABEmax variants expanded the editing window of ABE, many therapeutic targets could benefit from a more active ABE. ABE8, an evolution of ABE7.10 developed through an adenine deaminase variant library, displays significantly enhanced activity, achieving editing levels 1.94 times higher than those of ABE7.10 (Gaudelli et al., 2020). This increased activity makes ABE8 particularly useful for therapeutic targets that require more active base editing. Moreover, ABE8e represents a further enhancement, offering increased compatibility with Cas structural domains and heightened activity (Richter et al., 2020). Developed through phage-assisted non-continuous and continuous evolution, ABE8e features eight additional mutations that increase deamination kinetics by 590 times. It also improves expression in regions poorly edited by ABE7.10, like the BCL11A enhancer or HBG promoter, thus amplifying the effectiveness and scope of adenine base editing (Richter et al., 2020).

Despite these advancements, ABE8 variants such as ABE8e and ABE8s (Gaudelli et al., 2020; Richter et al., 2020), developed through molecular evolution, have shown a propensity for causing unintended alterations, leading to bystander mutations and off-target editing (Kim et al., 2019; Jeong et al., 2021). Addressing this, researchers developed ABE9, a precise and safe adenine base editor (Chen et al., 2023). ABE9 narrows the editing window to 1–2 nt and almost completely eliminates off-target editing on cytosine. Remarkably, ABE9 shows negligible off-target effects at both DNA and RNA levels, effectively addressing the off-target risks associated with traditional ABE systems and enhancing the safety and precision of adenine base editing (Chen et al., 2023).

Currently, gene therapy primarily utilizes two delivery forms: viral and nonviral (Figure 4). Among these, viral vectors are a popular choice for in vivo gene editing, with adeno-associated virus (AAV), lentivirus (LV), and adenovirus (Ad) being the most common. These carriers are engineered to transport editing tools and necessary components into various tissues or organs through injection or targeted delivery. Viral vectors offer advantages like high efficiency, specificity, and the ability to deliver across a myriad of cell types and tissues. AAV is particularly notable for its versatility and safety in gene therapy applications (Table 1). They consist of a protein capsid enclosing a 4.7 kb ssDNA genome and are non-pathogenic (Wang et al., 2019). Their benefits include high delivery efficiency, low immunogenicity, broad cell-transducing capabilities, and stable, long-term expression (Wang et al., 2019). AAV is versatile in its delivery to various tissues, such as the liver, central nervous system, muscles, and ocular tissues (Naso et al., 2017; Wang et al., 2019). Clinically, AAV has a solid track record and has been extensively used in gene therapy trials (Ail et al., 2023; Muramatsu and Muramatsu, 2023). However, delivering large base editors via AAV presents challenges due to size constraints. To overcome AAV packaging limitations, researchers have developed a dual AAV strategy. This involves two separate AAV vectors, each carrying parts of the editing tools and components, which reassemble into a functional editing enzyme or protein through intein-mediated trans-splicing within the target cells (Truong et al., 2015). It has been reported that multiple groups have utilized a dual AAV system to deliver base editing treatments for diseases such as amyotrophic lateral sclerosis (Lim et al., 2020), duchenne muscular dystrophy (DMD) (Ryu et al., 2018; Xu L. et al., 2021; Chemello et al., 2021), metabolic liver diseases (Villiger et al., 2018), hutchinson-gilford progeria syndrome (Koblan et al., 2021b), and hearing loss (HL) (Yeh et al., 2020) in mouse models. In addition, researchers have developed smaller Cas nucleases to enable single AAV vector delivery (Chen et al., 2022; Kweon et al., 2023) and strategies to reduce the long-term expression of edited genes via AAV delivery (Ibraheim et al., 2021; Zhang H. et al., 2023). Despite these advancements, the clinical application of AAV vectors still faces certain limitations, with pre-existing immunity against AAV being a major challenge in AAV gene therapy (Kruzik et al., 2019).

The majority of in vivo gene editing applications have harnessed AAV, whereas certain preclinical investigations have availed themselves of lentiviruses or adenoviruses. Lentiviruses, derived from the human immunodeficiency virus (HIV), differ from AAV in their ability to carry larger genetic payloads and, in certain cases, achieve higher delivery efficiency (Ferreira et al., 2021). Integrating the editing tools into the lentivirus genome enables its direct injection or targeted delivery into tissues or organs in the body. For example, lentivirus and integration-deficient lentiviral vectors (IDLVs) have been used to deliver ABEmax for correcting the RPE65 pathogenic gene in the retinal pigment epithelium of mice (Suh et al., 2021). Ortinski and colleagues demonstrated efficient genome editing using IDLVs in HEK293T cells and post-mitotic brain neurons (Ortinski et al., 2017). However, a significant drawback of in vivo lentiviral delivery is the potential for genome integration (Kymäläinen et al., 2014). Adenoviruses, on the other hand, are non-enveloped DNA viruses that infect a wide range of host cells with high efficiency (Lee C. S. et al., 2017). Research has shown that adenovirus-mediated base editors introduce precise nucleotide changes in cells and tissues of mammals (Chadwick et al., 2017; Carreras et al., 2019). While Ad facilitates potent gene editing, it bears the potential to induce the production of antibodies against Cas9 (Wang et al., 2015).

Viral vectors are essential delivery tools in gene therapy and genetic editing, with current challenges focusing on delivery efficiency, specificity, and the potential for triggering immune responses, especially in in vivo applications. Future research is aimed at improving these vectors to enhance delivery efficiency and specificity while minimizing immune responses, making them safer and more effective. Additionally, combining viral vectors with other delivery methods or employing genome editing to optimize viral vectors themselves are promising areas of development.

Nonviral delivery methods play a crucial role in the field of drug and gene therapy, offering a range of advantages over traditional viral vector-based methods (Table 2). One of the key benefits of nonviral delivery methods is their reduced immunogenicity. Unlike viral vectors, nonviral methods typically do not elicit strong immune responses in the body. This reduced immunogenicity is crucial for repeated administration and long-term treatments, as it minimizes the risk of the body developing resistance or adverse reactions to the therapy (Cheng H. et al., 2021; Raguram et al., 2022). Another advantage is the lower toxicity associated with nonviral delivery systems. These methods are generally considered safer and more biocompatible, reducing the risk of adverse reactions and side effects that can be associated with viral vector-based therapies.

Duchenne muscular dystrophy (DMD) is a progressive genetic muscle disorder caused by mutations in the gene encoding the dystrophin protein on the X chromosome (Long et al., 2018; Verhaart and Aartsma-Rus, 2019). The DMD gene is the largest known human gene, spanning 2.4 megabases, and its extensive size increases the likelihood of mutations, with the majority of DMD mutations being due to deletions or duplications (Bladen et al., 2015). Dystrophin, the protein encoded by this gene, is essential for stabilizing and protecting muscle cells. Mutations in the dystrophin gene lead to the production of non-functional dystrophin, compromising the structural integrity of muscle cell membranes and resulting in the progressive destruction of muscle cells (Verhaart and Aartsma-Rus, 2019). As one of the most common lethal genetic disorders, approximately one in every 3,500 to 5,000 newborn males are affected by DMD (Mendell et al., 2012). Despite extensive research, both basic and clinical, an effective and precise treatment for DMD remains elusive. Corticosteroids (such as prednisone or deflazacort) are one of the main methods of treatment for DMD, widely used to delay the loss of muscle function (Roberts et al., 2023). They primarily provide direct and indirect protection to muscle cells by inhibiting inflammatory responses and through immunomodulatory effects (Ricotti et al., 2013; McDonald et al., 2018). However, long-term use of corticosteroids may lead to a range of side effects, including weight gain, osteoporosis, and insulin resistance (Ricotti et al., 2013). Eteplirsen and Golodirsen are two exon-skipping agents approved by the United States Food and Drug Administration (FDA), which work by encouraging cells to skip over mutated exons during the mRNA splicing process, providing a treatment option for DMD patients with specific exon mutations (Nelson and Miceli, 2017; Heo, 2020; Shirley, 2021). Although exon-skipping therapy offers a new treatment approach, its applicability is limited, and the development and production costs are high. CRISRPR-mediated exon knockout restores expression and function of dystrophin at the cellular and animal levels (Xu et al., 2016; El Refaey et al., 2017), but Cas9-induced double-strand breaks may lead to large genomic deletions and even chromosomal rearrangements (Shin et al., 2017; Kosicki et al., 2018). As our comprehension of the underlying mechanisms of DMD advances and gene editing technology matures, the application of base editing in DMD treatment is emerging as a focal area of research, representing a promising frontier in the quest for an effective therapy.

The base editor has demonstrated efficient and precise repair of duchenne muscular dystrophy gene mutations in mouse models. This achievement led to the restoration of dystrophin protein expression and subsequent improvement of symptoms in the mice, showcasing the potential of this technology as a therapeutic approach for DMD. In a study by Ryu and colleagues, in vivo base editing was used to correct a DMD-causing nonsense mutation in exon 20 of the Dmd gene in mice (Ryu et al., 2018). They employed extended gRNAs to expand the editing window of ABE7.10. When ABE7.10 was delivered into the myoblasts of a DMD mouse model using a double trans-splicing adeno-associated virus (tsAAV) (Sun et al., 2000; Lai et al., 2005) vector system, it restored dystrophin protein expression in approximately 17% ± 1% of myofibers (Ryu et al., 2018). This level of restoration is particularly significant, as achieving more than 4% of normal dystrophin protein expression is considered sufficient to improve muscle function in DMD mice (van Putten et al., 2013). Xu et al. then tested the feasibility and long-term efficacy of systemic iABE-NGA-based therapy in the DMD mouse (mdx^4cv) model (Xu L. et al., 2021). They packaged iABE-NGA into a double AAV9 vector mediated by the intronic peptide Gp41-1 (referred to as AAV9-iNG) and administered it via tail vein injection. Remarkably, 10 months post-injection, over 95% of cardiac myocytes in mdx^4cv mice showed restored dystrophin expression, with about 15% dystrophin reconstitution also observed in skeletal muscles (Xu L. et al., 2021). These mice exhibited reduced myocardial fibrosis and enhanced muscle contractile function, significantly ameliorating the pathological characteristics of their muscle tissue. Additionally, the study assessed potential immune responses and off-target effects in the mdx^4cv mice treated with AAV9-iNG. The results indicated no significant toxic side effects or notable genomic off-target events, underscoring the safety and precision of this therapeutic approach.

In addition to correcting point mutations in the Dmd gene, base editors have the ability to restore dystrophin expression in DMD by inducing exon skipping. Gapinske and colleagues developed a base editing approach named CRISPR-SKIP, which achieves permanent exon skipping by introducing C to T or A to G mutations at splice acceptor sites in genomic DNA (Gapinske et al., 2018). Based on this strategy, Gapinske and colleagues implemented a split-intein mediated dual AAV system to deliver the ABE and CBE base editors, successfully achieving the editing of exon 45 in the DMD gene both in human cells and in vivo in mice (Gapinske et al., 2023). In another study, the team led by Chemello utilized a dual AAV9 delivery system to administer ABEmax-SpCas9-NG to a DMD mouse model with exon 51 deletion, targeting the Dmd gene for exon skipping (exon 49 spliced to exon 52), resulting in the restoration of dystrophin expression in 96.5% ± 0.7% of muscle fibers (Figure 6A) (Chemello et al., 2021). These innovative therapeutic strategies expand the range of genome editing techniques available for treating DMD.

Historically, respiratory failure was the leading cause of death in DMD patients, but with advancements in respiratory support, cardiac failure has now emerged as the primary cause of morbidity and mortality (Eagle et al., 2002; McNally et al., 2015). Currently, cardiac symptoms in DMD patients still lack effective therapeutic interventions. Yuan et al. utilized targeted AID-mediated mutagenesis (TAM) to achieve exon 50 skipping in induced pluripotent stem cells (iPSCs) derived from DMD patients (Yuan et al., 2018). This approach restored dystrophin protein expression and function in iPSC-derived cardiomyocytes. They also identified a Dmd^E4* mouse model, which mimics the cardiac pathology of DMD, exhibiting progressive cardiac dysfunction similar to that in human patients (Li J. et al., 2021). Intraperitoneal injection of the AAV9-packaged base editor eTAM into postnatal day 2 or day 3 Dmd^E4* mice achieved skipping of Dmd exon 4 (exon skipping efficiency of 59.98% ± 4.74%), which modestly restored dystrophin protein expression. Dystrophin levels in the hearts of Dmd^E4* mice remained close to wild-type levels after 12 months, alleviating the dystrophic condition and prolonging survival (Li J. et al., 2021). Wang et al. created a DMD hiPSC cell line with exon 48 to 54 deletions (ΔE48-54) using CRISPR-Cas9 (Wang P. et al., 2023). They restored dystrophin expression in DMD hiPSC-derived cardiomyocytes to 42.5% ± 11% of wild-type levels by inducing exon 55 skipping with the ABE8eV106W-SpCas9 base editor (Wang P. et al., 2023). Furthermore, Gapinske and colleagues accomplished systemic delivery of ABE8e-UGI using the MyoAAV system with RGD motifs, achieving DNA editing in cardiac tissue (Gapinske et al., 2023).

RNA editing also shows great promise in DMD treatment. In a study conducted by Li and colleagues, a previously uncharacterized c.4174C>T nonsense point mutation in the DMD gene was identified and its pathogenicity was confirmed in the humanized DMD^E30mut mouse model (Li et al., 2023). Utilizing a mini-dCas13X-mediated RNA adenine base editing system (mxABE), they achieved up to 84% A-to-G editing efficiency in the DMD^E30mut mice, leading to differential restoration of dystrophin protein expression in various muscle tissues including the diaphragm, tibialis anterior, and heart (Li et al., 2023). Continuous treatment with mxABE can restore the expression of dystrophin protein to 50% of wild-type levels in DMD^E30mut mice, significantly improving muscle growth and function in the mice. This study strongly suggests that mxABE-based strategies can be used to effectively treat genetic diseases caused by nonsense point mutations. In a similar vein, another small compact RNA base editor (ceRBE) successfully repaired the Q1392X mutation in the DMD gene in DMD^Q1392X humanized mice (68.3% ± 10.1%), in which dystrophy were rescued to a high level of 68.1% ± 9.4% in right tibialis anterior tissue (Wang et al., 2023c). These studies underscore the potential of RNA editing, particularly mxABE and ceRBE systems, as powerful tools for treating DMD and possibly other genetic diseases.

Inherited retinal diseases (IRDs) are a genetically and clinically heterogeneous group of disorders that affect the function and structure of the retina, leading to visual impairment and even blindness (Sahel et al., 2014). Owing to the heterogeneity of IRDs, their clinical manifestations exhibit considerable variability, encompassing diverse symptoms, severity levels, and modes of inheritance (Pulman et al., 2022; Jo et al., 2023a). Examples of these disorders include Leber congenital amaurosis (LCA), retinitis pigmentosa (RP), color blindness, and choroideremia, among others. The identification of hundreds of genes associated with IRDs underscores the genetic complexity of these conditions, complicating both diagnosis and treatment. Traditional approaches to managing IRDs, such as visual aids, gene therapy, and cell transplantation, are constrained in their effectiveness and accompanied by numerous challenges (Yue et al., 2016; Gasparini et al., 2019; Singh et al., 2021).

LCA is a rare, severe genetic retinal disorder, often considered the earliest and most severe form within IRDs. It primarily exhibits autosomal recessive inheritance and is characterized by progressive retinal degeneration and vision loss (Cremers et al., 2002). LCA is caused by critical gene mutations responsible for the development and function of the retina or retinal pigment epithelium (RPE), with most LCA patients experiencing severe visual impairments in infancy or early childhood, leading to complete blindness usually by the age of 30–40 (den Hollander et al., 2008). RPE-specific 65-kDa protein (RPE65) is an essential isomerase in the classical visual cycle, playing a pivotal role in converting 11-cis-retinol to 11-cis-retinal (Redmond et al., 1998). Loss-of-function mutations in the RPE65 gene disrupt this crucial process, impairing the visual cycle and leading to a range of retinal diseases (Redmond et al., 1998). Among these, mutations in RPE65 are a significant cause of leber congenital amaurosis type 2 (LCA2), accounting for about 16% of all LCA cases (Choi et al., 2022). Luxturna (voretigene neparvovec-rzyl) is an FDA-approved treatment for LCA and other retinal diseases caused by mutations in the RPE65 gene (FDA, 2017b). It aims to deliver a normal RPE65 gene directly into the eye in a one-time procedure, compensating for the gene that has lost function due to mutation. It is worth noting that clinical trials for Luxturna included incidents of mild ocular discomfort, and the FDA’s prescribing information for Luxturna contains specific warnings about serious ocular adverse events (FDA, 2017c; Russell et al., 2017). The efficiency of Rpe65 nonsense mutation correction mediated by CRISPR-Cas9 is very low and is associated with a high incidence of indels (Jo et al., 2019). Therefore, there is a broader therapeutic demand for the treatment of LCA. Research has demonstrated that base editing can effectively repair mutations in the Rpe65 gene, thereby improving visual function in mouse models of LCA (Suh et al., 2021; Choi et al., 2022; Jo et al., 2023b). Suh and colleagues used rd12 mice, an LCA model with a nonsense mutation in Rpe65 gene (c.130C>T; p. R44X), to demonstrate the efficacy of base editing (Figure 6B) (Suh et al., 2021). Utilizing a LV vector, they delivered the base editor ABEmax subretinally, achieving a 29% correction of the mutant gene in this model with no observable off-target effects. Subsequent assessment revealed the reinstatement of the visual cycle, preservation of the visual pathway from the retina to the primary visual cortex, and recovery in cortical responses in treated rd12 mice. Jo et al. employed a dual AAV system-mediated base editor, NG-ABE, in rd12 mice (Jo et al., 2023b). This approach effectively corrected Rpe65 mRNA transcripts (80%) and restored visual function, with retinal electroretinogram (ERG) wave amplitudes reaching 60% of those wild-type mice. While both ABEmax and NG-ABE strategies successfully improved visual function in rd12 mice, it remains unclear whether they can prevent further degeneration of retinal photoreceptors. In another study, an optimized NG-ABE base editor was delivered via LV vector in rd12 mice, successfully correcting the Rpe65 mutation and leading to the re-expression of the truncated Rpe65 protein, thereby restoring cone cell function (Choi et al., 2022). These findings suggest that base editing can correct pathogenic mutations and potentially offer long-term protection to the retina, preventing photoreceptor degeneration. The exploration of nonviral vectors for delivering base editors in the treatment of LCA marks a significant step towards enhancing safety and specificity. An illustrative example is the use of the purified NG-ABEmax RNP complex, which has been applied to the rd12 mouse retina for gene correction (Jang et al., 2021). This approach offers a notable advantage by avoiding the accidental, random integration of DNA and reducing cellular immune responses caused by excess sgRNA, thereby improving the safety profile of base editing. Further advancing this nonviral delivery approach, Kabra and colleagues utilized silicon nanocapsules (SNCs) to deliver the ABE8e base editor (Kabra et al., 2023). They focused on editing the W53X mutation in the Kcnj13 gene of LCA16 mice (Kcnj13 ^W53X/+ΔR ), which led to the restoration of Kir7.1 potassium ion channel function in the RPE. This intervention not only improved the functionality of the RPE in diseased mice but also underscored the potential of nonviral delivery systems in preserving vision in LCA16. This study and others like it represent important advancements in the field of gene therapy, offering safer and more specific alternatives to traditional viral vector-based methods.

RP is a common genetically heterogeneous retinal disorder, characterized by the progressive loss of rod and cone photoreceptor cells (Hartong et al., 2006). Initially manifesting as night blindness, RP gradually progresses to a more severe loss of vision, ultimately resulting in total blindness (Hartong et al., 2006). To date, over 100 genes have been identified in various subtypes of RP with different genetic patterns, including the rhodopsin gene (RHO), the pre-mRNA processing factor 31 gene (PRPF31), and the peripherin 2 gene (PRPH2), among others (Bhardwaj et al., 2022; Qin et al., 2023b). Although Luxturna offers a treatment option for a minority of RP patients carrying RPE65 mutations, the genetic diversity of RP means that a universal cure for all patients is currently lacking (Cross et al., 2022). Against this backdrop, base editing technology provides new hope for correcting pathogenic genes, demonstrating potential for the treatment of RP. Mutations in the PDE6B gene impact the protein that is the β-subunit of the rod cell cyclic guanosine monophosphate (cGMP)-phosphodiesterase (PDEβ), with such mutations being one of the common causes of autosomal recessive inherited RP (Danciger et al., 1995). The rd10 mouse is a commonly used model for RP, characterized by a missense mutation (c.1678C>T, p. R560C) in its Pde6b gene, exhibiting a phenotype similar to that of typical human RP patients (Chang et al., 2007; Gargini et al., 2007). Su and colleagues employed the retinal tropism vector AAV8 for subretinal delivery of NG-ABE8e in rd10 mice, achieving an effective correction of the pathogenic mutation in the Pde6b with a correction efficiency of up to 54.97% at the cDNA level (Figure 6B) (Su et al., 2023). The application of NG-ABE8e not only corrects the mutation in the Pde6b gene in rd10 mice but also preserves both rod and cone cells, significantly improving the retinal structure and visual behavior in the treated mice. In another related study, Wu and colleagues utilized a dual AAV5 vector system to precisely correct the Pde6b mutation in rd10 mice through the application of SpRY-ABE8e (Wu et al., 2023). They achieved an average cDNA editing efficiency of 34.07% ± 7.12%, resulting in the restoration of functional protein expression, extended survival of photoreceptor cells, and enhanced visual function.

Aniridia is a rare congenital eye disorder predominantly associated with dominant loss-of-function mutations in the transcription factor paired box 6 (PAX6) (Hingorani et al., 2012). It is characterized by the absence or underdevelopment of the iris and may also exhibit other ocular abnormalities, including cataracts, retinal pigmentary degeneration, and other eye-related anomalies (Hingorani et al., 2012). Previous studies have indicated that the use of nonsense suppression and mitogen-activated protein kinase kinase (MEK or MAP2K) inhibitors can rescue the retinal and visual function in a mouse model of aniridia (Wang X. et al., 2017; Cole et al., 2022). More recently, Adair and colleagues developed a humanized aniridia mouse model (CHuMMMS) and a corresponding cell line (Adair et al., 2023). In the study conducted, researchers successfully achieved a high genome correction rate, averaging 76.8% ± 0.48%, by employing the adenine base editor ABE8e (Adair et al., 2023). Additionally, the study involved the use of LNP-mediated ABE8e to restore Pax6 protein expression in primary neurons, attaining a correction rate of 24.8%.

Genetic hearing loss (HL) is an auditory impairment resulting from genetic mutations or abnormalities, primarily classified into syndromic hearing loss (30%) and non-syndromic hearing loss (70%) based on inheritance patterns and clinical manifestations (Yang et al., 2019). Syndromic hearing loss involves auditory impairment along with pathological changes in other systems or organs, whereas non-syndromic hearing loss manifests solely as auditory impairment (Tropitzsch et al., 2023). Traditional treatments for genetic HL, such as sound amplification devices or cochlear implants that stimulate the auditory nerve, provide limited improvement. They improve hearing but do not restore the natural functionality of the ear. In terms of genetic interventions, CRISPR/Cas9-mediated HDR and NHEJ have shown potential in silencing or knocking out dominant pathogenic mutations, offering benefits in genetic HL treatment (Mianné et al., 2016; Noh et al., 2022; Xue et al., 2022). However, these methods generally do not address mutations that cause recessive functional loss. Recessive genetic HL requires correction rather than disruption or silencing of pathogenic alleles to prevent hearing loss, posing a series of challenges in the treatment of genetic HL. Mutations in the transmembrane channel-like 1 (TMC1) gene can result in dominant or recessive deafness, as the encoded protein is closely linked to the formation of mechanosensitive ion channels involved in auditory sensation (Corey et al., 2019). Recessive mutations in the TMC1 gene lead to rapid degeneration of hair cells, resulting in swift and complete deafness, with the TMC1 mutations accounting for 4% to 8% of hereditary deafness in certain populations (Kitajiri et al., 2007; Sirmaci et al., 2009). Treating mice with recessive Tmc1 mutations via AAV gene therapy partially restores hearing, but the effect is limited in duration and does not genuinely edit or repair recessive gene mutations (Askew et al., 2015; Landegger et al., 2017). David Liu and colleagues achieved a significant breakthrough by successfully repairing Tmc1 gene mutations in the inner ear of Baringo mice through the application of base editing technology (Figure 6C) (Yeh et al., 2020). This achievement represents the first successful application of a base editor in treating recessive hearing loss. The Baringo mice (Tmc1 ^Y182C/Y182C ; Tmc2 ^+/+ ) harbor a TA to CG mutation in the Tmc1 gene, resulting in the onset of severe deafness by 4 weeks of age (Manji et al., 2012). The team utilized a dual AAV strategy to deliver the optimized adenine base editor, AID-BE4max, featuring enhanced AID deaminase, into the inner ear of Baringo mice, successfully achieving a notable 50% editing efficiency in the Tmc1 sequence (Yeh et al., 2020). The AID-BE4max treatment restored inner hair cell sensory transduction and cell morphology in the mice, temporarily rescuing low-frequency hearing. However, the therapeutic effect lasted only for about 4 weeks, underscoring the need for further enhancements in efficiency and durability for this strategy to advance as a viable treatment for hereditary deafness.

The MYO6 gene encodes an atypical myosin, Myosin IV, serving as a molecular motor expressed in inner ear hair cells and playing a crucial role in auditory and vestibular function (Ahmed et al., 2003). Pathogenic variations in the MYO6 gene have been found to be associated with autosomal dominant or recessive hereditary hearing loss (Melchionda et al., 2001; Ahmed et al., 2003). In a significant study, Xiao et al. utilized an AAV-mediated mxABE to precisely correct the Myo6 gene mutation in the inner ear of Myo6^C442Y/+mice, a model for autosomal dominant hereditary deafness (Xiao et al., 2022). The mxABE editing technique enhanced the survival rate of inner ear hair cells and ameliorated the auditory function in mice, with the therapeutic effects persisting for a duration of up to 3 months. This therapeutic approach underscores the substantial potential of RNA base editing as a treatment for dominant hereditary hearing loss, further supporting the development and application of RNA correction therapies in genetic medicine.

Hematological hereditary diseases, originating from or impacting the hematopoietic system and accompanied by hematological abnormalities, are characterized by anemia, hemorrhage, fever, and coagulation disorders. Among these, SCD and β-thalassemia are notably common. SCD is caused by a specific mutation in the β-globin gene (HBB), leading to the production of abnormal hemoglobin (HbS), with symptoms including, but not limited to, pain crises, anemia, increased risk of infections, and organ damage (Bauer and Orkin, 2015). β-thalassemia, similarly resulting from mutations in the HBB gene, impairs the normal synthesis of β-globin, leading to its deficiency and causing symptoms such as chronic anemia, fatigue, growth delay, and facial bone deformities (Saraf et al., 2014). The persistent presence of fetal hemoglobin (HbF), comprised of γ-globin and α-globin, is closely correlated with the severity of SCD and β-thalassemia (Xu et al., 2011; Bauer et al., 2013; Canver et al., 2015). Elevating the levels of HbF within patients is one of the key strategies for the treatment or mitigation of these types of anemia. The γ-globin gene, encoded by the HBG gene and typically silenced in adulthood, is functional in fetal stages and can compensate for β-globin deficiency (Fontana et al., 2023). The BCL11A gene plays a crucial role in repressing γ-globin gene expression in erythrocytes (Bauer et al., 2013). Consequently, therapeutic strategies that involve reactivating γ-globin gene expression in patients to supplement insufficient β-globin and employing targeted editing of the BCL11A gene to reactivate HbF present viable methods for addressing these disorders. Hydroxyurea, one of the first-line treatment options approved by the FDA for SCD, effectively reduces the incidence of pain crises and the need for transfusions by increasing fetal hemoglobin levels (Charache et al., 1995; Steinberg et al., 2003; Lee and Ogu, 2022). Although hydroxyurea demonstrates significant benefits in treating SCD, its use is also accompanied by some drawbacks and potential side effects, such as dose dependency, bone marrow suppression, and reproductive toxicity (Brawley et al., 2008). L-glutamine, another medication approved by the FDA for the treatment of SCD following hydroxyurea, offers patients a new path to relief by reducing oxidative stress (FDA, 2017a; Niihara et al., 2018). However, its side effects include conditions such as nausea, diarrhea, and abdominal pain (Quinn, 2018; Ogu et al., 2021). In recent years, the FDA and the European Medicines Agency (EMA) have approved two innovative treatments for SCD: the monoclonal antibody Crizanlizumab and the small molecule compound Voxelotor (FDA, 2019; EMA, 2020a). The former reduces the incidence of pain crises by inhibiting the adhesion of red blood cells to the vascular wall, while the latter alleviates the sickling of red blood cells by increasing their affinity for oxygen (Oksenberg et al., 2016; Ataga et al., 2017; Kutlar et al., 2019; Vichinsky et al., 2019). However, both Crizanlizumab and Voxelotor may cause side effects such as headaches, fever, and chest pain (Ataga et al., 2017; Vichinsky et al., 2019). For β-thalassemia, traditional treatment strategies primarily rely on periodic blood transfusions, which necessitate chelation therapy to manage the consequent iron load accumulation and prevent organ damage from iron overload (Cappellini et al., 2018). While these treatments are effective in the short term, they do not cure the disease and require lifelong management. Luspatercept, an innovative recombinant fusion protein that promotes the maturation of late-stage red blood cells and increases hemoglobin levels, offers a new treatment avenue for transfusion-dependent β-thalassemia (Markham, 2020; Sheth et al., 2023). Although it opens new paths for the treatment of transfusion-dependent β-thalassemia, it may lead to adverse events such as bone pain, joint pain, and hyperuricemia (Cappellini et al., 2020). Furthermore, hematopoietic stem cell transplantation offers a potential curative treatment for SCD and β-thalassemia under certain conditions, but its high risk and limited applicability restrict its widespread use (Chou, 2013; Cappellini et al., 2018).

Base editing technology facilitates precise modification of the BCL11A erythroid enhancer, attenuating its regulatory function and elevating HbF expression levels. The transcription factor GATA1 is involved in regulating red blood cell development and hemoglobin synthesis (Crispino and Horwitz, 2017). Mutations in the GATA1 binding sites on the erythroid-specific enhancer of BCL11A result in diminished regulatory activity on the BCL11A gene (Canver et al., 2015). David Liu and colleagues achieved successful introduction of two point mutations in a GATA1 binding site at the +58 BCL11A erythroid enhancer by employing the adenine base editor ABE8e (Richter et al., 2020). This resulted in simultaneous editing of two target adenines in 54.4% ± 12.5% of the alleles. In a related study, Zeng et al. employed electroporation to introduce the A3A(N57Q)-BE3 base editor into CD34⁺ hematopoietic stem/progenitor cells (CD34⁺ HSPCs) from patients with SCD and β-thalassemia, targeting and editing the +58 BCL11A erythroid enhancer (Zeng et al., 2020). This study led to the downregulation of BCL11A expression in red blood cells, effectively inducing the expression of γ-globin and HbF. Furthermore, the researchers conducted dual-site editing at −28 HBB and +58 BCL11A in β-thalassemia HSPCs, leading to the in vivo differentiation of cells producing functionally normal hemoglobin (Zeng et al., 2020).

Base editing has shown promise in introducing hereditary persistence of fetal hemoglobin (HPFH) mutations into the HBG1/2 promoter region, effectively increasing the expression of HbF. HPFH is typically caused by deletions in the β-globin gene cluster and point mutations in the γ-globin gene promoters, characterized by the persistent presence of excess HbF in adult red blood cells (Forget, 1998). Patients with HPFH generally do not exhibit severe anemia symptoms, and in some cases, the persistence of HbF can even provide a protective effect (Lu et al., 2022). Wang et al. delivered the base editor hA3A-BE3 to healthy or patient-derived CD34⁺ HSPCs via electroporation, targeting the HBG1/2 promoters to induce HPFH (Wang et al., 2020). This editing led to a substantial elevation in γ-globin levels in patient-derived cells, increasing from 6.8% to 44.2%, reaching a level that has potential clinical benefits for patients with β-hemoglobinopathies. Li and colleagues advanced the application of base editing for in vivo induction of HPFH mutations, promoting β-globin production (Li C. et al., 2021). They initially used the HDAd5/35++ vector to deliver base editors AncBE4max and ABEmax, targeting the +58 BCL11A erythroid enhancer or reconstructing HPFH mutations in the HBG1/2 promoters. This resulted in reactivated γ-globin in HUDEP-2 erythroid progenitor cells. Following this, β-YAC mice (Peterson et al., 1993) were treated using HDAd-ABE-sgHBG-2, achieving an average editing efficiency of 20% in HSPCs and leading to normal fetal hemoglobin expression with no detectable off-target effects (Li C. et al., 2021). In another study conducted by Li and colleagues, CD34 cells derived from β-thalassemia and SCD patients were transduced with an HDAd5/35++ vector expressing ABE8e, successfully introducing the −113 A>G HPFH mutation into these cells (Li C. et al., 2022). In β-YAC/CD46 mice, a single intravenous injection of the HDAd-EF1α.ABE8e vector achieved an in vivo hematopoietic stem cell (HSC) editing efficiency of up to 60%, with 30% γ-globin of β-globin expressed in 70% of erythrocytes (Li C. et al., 2022). Furthermore, utilizing base editors to create new transcription factor binding sites in the γ-globin promoter represents another viable strategy for inducing HbF expression (Cheng L. et al., 2021; Ravi et al., 2022).

Base editing offers another promising strategy for treating hemoglobinopathies, like SCD, by converting pathogenic hemoglobin variants into benign forms. This approach involves transforming the aberrant SCD β-globin gene (HBB ^S ) into a naturally occurring, non-pathogenic variant, such as the Makassar β-globin gene (HBB ^G ) (Viprakasit et al., 2002; Chu et al., 2021; Newby et al., 2021). In a study by Chu and colleagues, the research team employed deaminase-inlaid base editors (IBEs) on fibroblasts derived from a patient with homozygous SCD (Chu et al., 2021). This process achieved an impressive Makassar editing efficiency of up to 50%. Furthering this approach, Newby and his team employed ABE8e-NRCH mRNA and sgRNA through electroporation into CD34⁺ HSPCs obtained from SCD patients (Figure 6D) (Newby et al., 2021). This intervention successfully converted 80% of HBB ^S to HBB ^G . Following the transplantation of these base-edited hematopoietic stem cells into SCD mouse models, the frequency of HBB^G editing was maintained at 68% even after 16 weeks, effectively preserving the hematopoietic stem cells (Newby et al., 2021). The study also performed secondary transplantation of these edited stem cells, demonstrating that the modified cells retained functionality comparable to healthy hematopoietic stem cells. This finding not only underscores the efficacy of base editing in treating SCD but also highlights the durability and stability of the edited cells, marking a significant advancement in gene therapy for blood disorders.

To optimize genome editing for HbF, David Liu and colleagues compared five different gene editing approaches that were mediated either by Cas9 nucleases or adenine base editors (Mayuranathan et al., 2023). Their findings revealed that the adenine base editor of γ-globin-175A > G was the most effective in inducing HbF. This was further validated by the successful application of this editing approach in regenerated human hematopoietic stem cells within transplanted mice. In these models, the −175A > G editing notably reduced the formation of hypoxia-induced sickle-shaped cells, which are characteristic of SCD. This outcome demonstrates the potential of base editing in directly addressing the underlying cause of SCD. An important aspect of this study was the direct comparison between Cas9 editing and base editing techniques. The team observed that, compared to Cas9 editing, base editing resulted in precise nucleotide changes, exhibiting uniform HbF induction and being independent of TP53-mediated DNA damage response (Mayuranathan et al., 2023). This independence is a significant advantage, as it could reduce potential complications related to DNA repair pathways, which are a concern with traditional Cas9-mediated editing. Therefore, the findings from Mayuranathan and colleagues suggest that base editing could present a superior alternative to Cas9 for therapeutically inducing HbF. This holds great promise for developing more effective and safer treatments for diseases like SCD. As research in this area progresses, base editing technologies may offer new avenues for treating a range of genetic disorders.

Phenylketonuria (PKU) is a genetic disorder caused by a deficiency or reduced activity of phenylalanine hydroxylase (PAH) in the liver, typically inherited in an autosomal recessive manner (van Spronsen et al., 2021). PKU usually exhibits mild or imperceptible symptoms in infancy, but the accumulation of phenylalanine can lead to neurological damage and delayed intellectual development, with severe untreated cases potentially resulting in intellectual disability, behavioral issues, and epilepsy (van Vliet et al., 2019; Klaassen et al., 2021). Currently, the primary treatments for PKU include strict dietary control, pharmacotherapy, enzyme replacement therapy, and gene therapy, which is still under research (Regier et al., 2022). Dietary management is the cornerstone of PKU treatment, aiming to reduce phenylalanine (Phe) levels in the blood by limiting the intake of foods high in phenylalanine. This approach requires patients to adhere to a low-protein diet for life, which, although effective, can significantly impact the quality of life, especially for the growth and development of children (Ashe et al., 2019). Sapropterin dihydrochloride (Kuvan^®) is a medication approved in the United States and the European Union for the treatment of PKU, working by enhancing the activity of the PAH enzyme to reduce Phe levels in the blood (Burton et al., 2007; FDA, 2014; EMA, 2020b). It is important to note that Kuvan^® is only suitable for PKU patients who have a response to tetrahydrobiopterin (BH4) (Levy et al., 2007; Muntau et al., 2019). Palynziq (pegvaliase-pqpz) constitutes an enzyme replacement therapy approved by both the FDA and the EMA (FDA, 2018; Levy et al., 2018; EMA, 2019). The therapeutic efficacy of this treatment is attributed to its principal active ingredient, pegvaliase, a phenylalanine ammonia-lyase (PAL) that has been modified through PEGylation technology. This modification enables the direct conversion of phenylalanine in the bloodstream into safe, metabolizable entities such as trans-cinnamic acid and additional amino acids (Levy et al., 2018). A distinctive feature of this process is its independence from the conventional metabolic pathway of PAH, thus providing an efficacious means of reducing phenylalanine levels in the blood for patients suffering from PAH deficiency. While Palynziq marks a considerable advancement in the therapeutic landscape of PKU, its application is not devoid of risks. The medication is accompanied by a black box warning on its label, explicitly cautioning against the potential for severe allergic reactions, including anaphylactic shock (FDA, 2018; Harding et al., 2018; Thomas et al., 2018). Accordingly, the deployment of this therapeutic strategy necessitates rigorous medical oversight to ensure the safety of patients undergoing treatment and the immediate enactment of appropriate interventions upon the manifestation of any allergic response signs. Gene therapy represents the future direction of PKU treatment, offering a potential cure by repairing or replacing the defective PAH gene. Base editors have demonstrated promising capabilities in correcting pathogenic mutations within the livers of PKU model mice (Figure 6E) (Shedlovsky et al., 1993; Villiger et al., 2018; Villiger et al., 2021). Villiger and his research team achieved successful correction of the Pah ^enu2 c.835T>C mutation in the PKU mouse model by employing the dual AAV system to deliver the base editor nSaKKH-BE3 (Villiger et al., 2018). Significantly, this intervention effectively maintained blood phenylalanine levels within the physiological range. The study also observed a high mRNA correction rate of up to 63% in mouse liver extracts, reversing disease-associated phenotypes (Villiger et al., 2018). However, concerns about the potential risks associated with cytosine base editing have been raised. Some studies suggest that cytosine base editors might induce RNA mutations in cell lines (Grünewald et al., 2019a; Grünewald et al., 2019b), and the in vivo overexpression of the deaminase rAPOBEC1 could have detrimental effects on the mouse liver (Yamanaka et al., 1995; Yamanaka et al., 1996). Addressing these concerns, subsequent investigations focused on evaluating the off-target effects and overall safety of the cytosine base editor SaKKH-CBE3 in the liver (Villiger et al., 2021). Comprehensive RNA sequencing (RNA-Seq) and whole-genome sequencing (WGS) analyses indicated that SaKKH-CBE3 did not cause significant RNA and DNA off-target mutations. Importantly, no evidence of malignant liver transformation was observed. Further expanding the potential therapeutic approaches, researchers explored the delivery of SaKKH-CBE3 via LNPs in Pah^enu2 mice (Villiger et al., 2021). This method successfully corrected the pathogenic gene, mitigating the pathological phenotype without noticeable off-target effects across the transcriptome and genome. These studies broaden the application scope for cytosine base editors in the treatment of hereditary liver diseases.

Familial hypercholesterolemia (FH) is a hereditary disease caused by abnormalities in lipoprotein metabolism, primarily characterized by abnormally elevated levels of LDL cholesterol in the blood (Sniderman et al., 2011; Kersten, 2020). LDL is a critical lipoprotein that transports cholesterol and lipids. Excessively elevated LDL levels can cause cholesterol deposition on arterial walls, significantly increasing the risk of atherosclerosis and related cardiovascular and cerebrovascular diseases (Meng et al., 2023). A key factor in this process is proprotein convertase subtilisin/kexin type 9 (PCSK9), predominantly expressed in the liver. When PCSK9 undergoes loss-of-function mutations, it results in lowered levels of LDL cholesterol, thereby reducing the risk of atherosclerotic cardiovascular diseases (Cohen et al., 2005; Meng et al., 2023). In the management of FH, the prevailing therapeutic strategies encompass pharmacotherapy, lifestyle interventions, and, in certain cases, innovative therapeutic approaches (Lan et al., 2023). Pharmacotherapy forms the cornerstone of FH management, with statins serving as the preferred option to lower low-density lipoprotein cholesterol (LDL-C) levels in the blood, including Atorvastatin, Simvastatin, and Rosuvastatin. These medications work by reducing cholesterol production in the liver, effectively lowering LDL-C levels, but they may cause musculoskeletal adverse reactions such as muscle pain and gastrointestinal discomfort (de Sauvage Nolting et al., 2002; Braamskamp et al., 2015; Thompson and Taylor, 2017). As a complement, additional medications like bile acid sequestrants, intestinal cholesterol absorption inhibitors, and PCSK9 inhibitors are available, but these too necessitate long-term use and carry the risk of potential side effects (Huijgen et al., 2010; Nissen et al., 2016a; Nissen et al., 2016b; Watts et al., 2016). On the other hand, dietary improvements, regular exercise, and maintaining a healthy weight can serve as adjunctive therapy methods for FH treatment. However, lifestyle changes alone rarely achieve target LDL-C levels in most FH patients. On the horizon, emergent modalities like gene editing and gene therapy present potential curative prospects by directly addressing the genetic underpinnings of FH through the correction or replacement of the causative gene defects.

In the field of genetic editing, base editing has been successfully applied to introduce nonsense mutations in the Pcsk9 gene in the liver of mice (Chadwick et al., 2017; Carreras et al., 2019). This intervention led to marked decreases in plasma PCSK9 protein and cholesterol levels, underlining its potential effectiveness in managing cholesterol-related disorders. A notable application of this technology was demonstrated in a non-human primate model, the cynomolgus monkey (Figure 6F) (Musunuru et al., 2021). Researchers achieved pathogenic gene editing in cynomolgus monkeys by intravenously injecting LNPs to deliver the adenine base editor (ABE8.8). This approach led to over 60% editing of the PCSK9 gene in the liver, resulting in an approximately 90% reduction in blood PCSK9 protein levels. Remarkably, a single dose of this treatment maintained a long-term, stable reduction of LDL cholesterol by about 60% (Musunuru et al., 2021). Furthermore, the delivery of ABEmax using LNP has successfully achieved precise editing of the PCSK9 gene in liver tissues of mice and cynomolgus monkeys (Rothgangl et al., 2021). This editing resulted in significant reductions in PCSK9 and LDL protein levels in the blood, with decreases of 95% and 58% observed in mice, and 32% and 14% in cynomolgus monkeys, respectively. These results highlight the potential of base editing as a powerful tool in the prevention and treatment of cardiovascular diseases, especially in conditions like Familial Hypercholesterolemia where traditional treatment methods may be limited.

Dilated cardiomyopathy (DCM) is a severe cardiac disease characterized by an enlarged heart accompanied by impaired contractile function and a risk of sudden cardiac death, and it is also one of the most common causes of heart failure in humans (Hershberger et al., 2013; McNally et al., 2013). The management of DCM necessitates a multifaceted approach, encompassing pharmacotherapy to enhance cardiac function and prolong survival, lifestyle modifications to maintain cardiac health, the implantation of devices such as pacemakers and implantable cardioverter defibrillators (ICDs) for arrhythmia management, and, in severe cases, heart transplantation (Kummeling et al., 2015; Halliday et al., 2017; Fatkin et al., 2019). While each of these therapeutic modalities offers distinct advantages in managing DCM, they are also associated with limitations and challenges, including medication side effects, surgical risks, and post-transplant immunosuppression issues. Consequently, ongoing research and therapeutic innovation are critical for improving treatment outcomes and quality of life for DCM patients. A notable genetic factor contributing to DCM is mutations in the RNA binding motif protein 20 (RBM20) gene. Individuals with mutations in this gene typically experience an earlier onset of DCM, a higher likelihood of progressing to end-stage heart failure, and an elevated mortality rate (Jordan et al., 2021; Kornienko et al., 2023). ABEs were employed to successfully correct RBM20 gene mutations (RBM20^R634Q, RBM20^R636S) in iPSCs, achieving a high editing efficiency of up to 92% (Figure 6G) (Nishiyama et al., 2022). Building on this achievement, researchers developed a Rbm20^R636Q mutant mouse model. Subsequently, by intraperitoneally injecting AAV9 carrying ABEmax-VRQR-SpCas9, the researchers successfully restored cardiac function in the afflicted mice and prolonged their lifespan (Nishiyama et al., 2022). This offers a promising therapeutic approach for treating DCM and other diseases caused by mutant genes.

CD3δ severe combined immune deficiency (SCID) is a rare and severe hereditary disorder of the immune system, characterized by a profound impairment of immune system function (Picard et al., 2015). This disorder is related to developmental defects in the immune system and primarily involves genetic variations in the CD3δ gene, which is crucial for the assembly of the T-cell receptor (Garcillán et al., 2021). The absence of this gene precipitates a marked immunodeficiency. Bone marrow transplantation (or hematopoietic stem cell transplantation) serves as the principal therapeutic intervention for various forms of SCID, including CD3δ SCID (Grunebaum et al., 2006; Marcus et al., 2011; Fabio et al., 2020). This approach is distinguished by its capacity to afford enduring immune system reconstitution. Nevertheless, it is beset by several limitations, notably the challenge of identifying an appropriate donor, the elevated risks inherent in the transplantation procedure, and the possibility of graft rejection. In contrast, gene therapy represents a groundbreaking therapeutic avenue, fundamentally targeting the genetic aberrations responsible for disease by reinstating normal immune function. The application of ABE within HSPCs has been demonstrated to rectify the pathogenic mutations in the CD3δ gene associated with CD3δ SCID (Figure 6H) (McAuley et al., 2023). This correction reestablishes the T-cell developmental capabilities in HSPCs. Moreover, the sustained presence of amended hematopoietic stem cells post-transplantation presents a viable, one-off therapeutic avenue for individuals afflicted with CD3δ-SCID (McAuley et al., 2023). In the pathogenesis of SCID, dysfunctional RAG1 and IL2RG genes play a critical role (Suzuki et al., 2012; Gan et al., 2021). Zheng et al. have developed a technique employing the CBE4max system to deactivate the IL2RG and RAG1 genes, thereby successfully generating an immunodeficient monkey model (Zheng et al., 2023). These monkeys exhibit severely compromised immune systems, characterized by lymphocytopenia, atrophy of lymphoid organs, and an absence of mature T cells. The deployment of such immunodeficient monkeys can significantly augment the efficacy of preclinical trials, laying a robust groundwork for the advancement of biomedical science and its subsequent clinical translation.

Spinal muscular atrophy (SMA) is identified as a rare and severe hereditary neurological disorder, predominantly presenting in infancy or early childhood (Sugarman et al., 2012; Finkel et al., 2017). SMA results from homozygous deletion or mutation in the survival motor neuron 1 (SMN1) gene, leading to progressive muscle atrophy and weakness, ultimately causing impaired motor function (Prior, 2010; Finkel et al., 2017). SMN2, a human homologous gene to SMN1, produces a limited amount of functional SMN protein, but its stability is compromised by an exon deletion (Lorson et al., 1999). This variant protein partially compensates for SMN1 loss, alleviating spinal muscular atrophy symptoms to some extent. Recent years have witnessed significant advancements in the therapeutic landscape for SMA, with key treatments comprising antisense oligonucleotides such as nusinersen, AAV-mediated gene therapy onasemnogene abeparvovec, and the oral small molecule drug risdiplam (Yeo et al., 2024). Nusinersen modulates the splicing of the SMN2 gene, thereby augmenting the production of full-length SMN protein, demonstrating potential in improving muscle functionality and enhancing survival prospects in patients (Finkel et al., 2017; Mercuri et al., 2018). Nonetheless, this therapeutic approach necessitates periodic intrathecal administration and the significant financial burden of the therapy. Onasemnogene abeparvovec, distinguished as the inaugural gene therapy sanctioned for SMA, facilitates persistent SMN protein expression via a singular intravenous administration, designed to curtail the advancement of the disease (Mendell et al., 2017; Mendell et al., 2021). However, its disadvantages encompass the substantial one-time treatment cost and the need for further observation to ascertain long-term effects and safety. Risdiplam, the inaugural oral therapy for SMA, acts as an SMN2 mRNA splicing modifier, facilitating the production of more full-length SMN protein (Baranello et al., 2021; Masson et al., 2022). Its benefits include ease of use and continuous drug delivery, yet it also faces challenges regarding high treatment costs and the evaluation of long-term efficacy and safety. As innovative therapeutic approaches for SMA are developed, the challenges associated with these treatments, particularly in terms of their invasive nature and the need for sustained intervention, become apparent. Against this backdrop, the application of base editing for the effective correction of aberrant splicing in the SMN2 gene, thereby amplifying the expression of SMN protein in SMA patients, is emerging as a promising therapeutic strategy (Lin et al., 2020; Alves et al., 2023). An optimized D10-ABE base editing strategy (ABE8e-SpyMac and P8 sgRNA) targeting the mutation site in the SMN2 gene attained an average T6>C conversion rate of 87% in Δ7SMA mice (Figure 6I) (Arbab et al., 2023). This approach enhanced motor function and prolonged the average lifespan in mice afflicted with the disease. Moreover, the synergistic application of this strategy with nusinersen has shown compatibility, augmenting the therapeutic outcomes (Arbab et al., 2023). This integrative approach holds significant potential for future clinical applications in treating spinal muscular atrophy.

Although base editors can achieve correction and modification of individual genes or segments, off-target effects remain an uncertain factor that they face. These effects are categorized into two types: expected off-targets, occurring in genomic regions with high sequence similarity to the target site, and unintended off-targets in unrelated areas (Zhang et al., 2015; Pacesa et al., 2022). Such effects could lead to genomic instability and disrupted gene functions, making the resolution of these off-target effects a crucial aspect of base editing. Off-target detection plays a crucial role in unraveling the off-target mechanisms of gene editing systems. Several off-target detection methods have been validated in assessing the specificity of the CRISPR/Cas9 system. These include sgRNA off-target prediction tools like Cas-OFFinder (Bae et al., 2014) and CFD (Doench et al., 2016), techniques such as GUIDE-seq (Tsai et al., 2015) for capturing DNA double-strand break sites, and in vitro strategies like Digenome-seq (Kim et al., 2016) and CIRCLE-seq (Tsai et al., 2017). In a comprehensive analysis using WGS, researchers evaluated the BE3, HF1-BE3, and ABE base editing systems in plants (Jin et al., 2019). They discovered that BE3 and HF1-BE3 induced a substantial number of SNVs in the rice genome, predominantly C-to-T mutations (Jin et al., 2019). Indeed, the study revealed that most of these additional SNVs did not coincide with off-target sites predicted by existing software, such as Cas-OFFinder, highlighting a disparity between the observed mutations and the predictive abilities of the software. The development of innovative off-target detection methods and the refinement of base editing tools are effective strategies that have propelled the advancement of gene editing technologies. A key breakthrough in this domain is the GOTI (genome-wide off-target analysis by two-cell embryo injection) technology, which is instrumental in detecting random off-target sites that were previously undetectable by conventional methods (Zuo et al., 2019). This enhancement in sensitivity significantly bolsters the precision of gene editing. In a related development, Lei et al. introduced Detect-seq, an unbiased off-target detection technique noted for its high sensitivity, specificity, and non-preferential nature (Lei et al., 2021). Detect-seq facilitates the detection of off-target sites induced by CBE across the whole genome, offering a more comprehensive understanding of gene editing impacts. Furthermore, the development of DeepABE and DeepCBE by Park and Kim represents a significant contribution (Park and Kim, 2023). These deep learning-based computational models are adept at accurately predicting the editing efficiency and outcomes of ABE and CBE, thereby streamlining the application of these tools in genome editing. In addition to these detection technologies, Xiong and colleagues have demonstrated remarkable efficacy with their SAFE (simple and fast off-target elimination) strategy for CBE/ABE (Xiong et al., 2023). SAFE enables efficient and low off-target base editing not only in plant models like rice and arabidopsis but also in human and yeast cells, underscoring its versatility. In conclusion, advancements in off-target detection and the enhancement of base editing tools are pivotal for propelling the field of gene editing forward. By intensifying research on off-target factors, mitigation strategies, and off-target detection technologies, we are optimistic that significant progress in overcoming off-target effects will be achieved in the near future. This focus not only promises to refine the precision of gene editing techniques but also enhances their safety and efficacy for clinical applications.

The advancement and practical application of base editing technology heavily depend on enhancing both the efficiency and specificity of base editing systems. High editing efficiency and specificity are essential because they ensure that the system can precisely and effectively alter the target sequence to bring about the intended genetic modifications. However, this efficiency is subject to influence by a myriad of factors, including the intrinsic activity of the enzyme, the modalities of its delivery, and the nature of the target site chosen for modification. To address these issues, researchers continually improve the design and performance of editing systems, including developing more specific enzyme variants, refining delivery methods, and optimizing the selection of modified target (Komor et al., 2017; Koblan et al., 2018; Huang et al., 2019; Gaudelli et al., 2020; Richter et al., 2020; Segel et al., 2021; Wang et al., 2023b; Kreitz et al., 2023). Additionally, the advent of high-throughput sequencing technologies and sophisticated machine learning models has led to the creation of online prediction tools (Zhang et al., 2023a; Kim et al., 2023). These tools serve as invaluable resources in guiding the selection of optimal base editing tools. They provide insights into the potential editing efficiency and off-target effects of various systems, thereby aiding in the optimization of their performance. Through innovative research, the development of advanced tools, and the utilization of predictive technologies, the goal is to tailor base editing systems that are both highly efficient and specific, paving the way for their safe and effective application in clinical settings.

The potential of base editing tools to induce cytotoxic effects or trigger immune responses is a significant factor limiting their feasibility in clinical applications. Although the in vitro editing efficiency of current base editing systems has significantly improved, broader potential applications will require in vivo editing. The use of this technology in vivo comes with challenges, one of which is the immune response to Cas9. As an exogenous microbial-derived nuclease, Cas9 may elicit human adaptive immune responses, posing potential obstacles to the safety and effectiveness of base editors when used for therapeutic purposes in patients. This concern was highlighted by the work of Charlesworth and colleagues, who found antibodies against SaCas9 and SpCas9 in human serum donors (Charlesworth et al., 2019). This finding indicates the potential for both humoral and cell-mediated adaptive immune responses to Cas9 in humans, underscoring the need to consider the human immune system’s impact on CRISPR/Cas9-based therapies during clinical trials. Additionally, the interaction of Cas9 with cellular components can influence DNA repair mechanisms. Research by Xu and colleagues discovered that Cas9 interacts with the KU86 subunit of the DNA-dependent protein kinase complex, affecting the repair of DNA double-strand breaks through the non-homologous end joining pathway (Xu et al., 2020). Such interactions could have implications for the safety and efficacy of CRISPR/Cas9-based treatments. Enache and his team found that overexpression of the Cas9 protein can activate the p53 pathway in various cell lines, leading to the selective enrichment of p53-inactivating mutations (Enache et al., 2020). The study also indicates that the Cas9-induced activation of p53 in cell lines is likely to interfere with the results of genetic and chemical screens. Given these findings, it is clear that our understanding of the safety risks and mechanisms underlying CRISPR/Cas9 system usage is still evolving. Consequently, precise validation through laboratory studies and preclinical assessments is essential when advancing the clinical application of CRISPR/Cas9-derived base editing technologies. Employing such an integrative approach enables the maximization of the potential of base editing technologies within the healthcare domain, while concurrently safeguarding patient safety and wellbeing.

The maturation and standardization of base editing technologies have mitigated associated risks and ambiguities, thereby catalyzing their extensive deployment in clinical settings (Table 4). Clinical trials currently underway encompass a spectrum of genetic disorders, including but not limited to homozygous familial hypercholesterolemia, sickle cell disease, β-thalassemia, and T-cell acute lymphoblastic leukemia (Amit et al., 2022; Chiesa et al., 2023; Hughes-Castleberry, 2023; Naddaf, 2023). As base editing technology continues to evolve, its application in the clinical landscape is expected to expand, encompassing a wider range of genetic disorders. As the scope of base editing technology broadens, encompassing an increasing variety of genetic disorders, its potential to profoundly affect human health becomes more pronounced. This expansion highlights the need for stringent ethical scrutiny concerning the application of the technology. Key ethical considerations encompass the long-term effects of gene editing on individuals and society, demanding comprehensive risk evaluations and ethical deliberations before implementation (Coller, 2019; Rothschild, 2020). Furthermore, the equity and accessibility of gene editing technologies have come under scrutiny, raising concerns about the potential for creating disparities in health, social, and economic advantages among individuals, thereby influencing public acceptance (Coller, 2019). Moreover, the rapid development and clinical application of base editing technology challenge existing legal and ethical frameworks, requiring updates and enhancements to policies to protect patient rights while promoting scientific innovation. In summary, the development of base editing technology presents unprecedented opportunities for the treatment of genetic diseases, but the accompanying ethical considerations and challenges cannot be overlooked. Through ongoing scientific research, ethical scrutiny, and policy support, we hope to fully realize the immense potential of base editing technology in the treatment of genetic diseases, ensuring patient safety and rights.