2.2.1 Cytosine BEs

The cytosine BE (CBE) system converts codons CAA, CAG, CGA, and TGG into a stop codon by mediating the conversion of the target site C to T.¹¹⁰ BE3, BE4, and dCpf1-BE (d means catalytically dead) use Cas9 or Cpf1 variants to recruit cytidine deaminases to produce specific C to T or G to A changes via DNA mismatch repair (MMR) pathways (Figure 3B).^{95, 96} The BE3 system links rat cytidine deaminase apolipoprotein B mRNA editing enzyme catalytic subunit 1 (APOBEC1) with the Cas9 cleavage enzyme and uracil glycosylase inhibitor.¹¹¹ BE3 editing is confined to a five-nucleotide (nt) editing window. It is effective in correcting various point mutations associated with human disease but is less efficient in the GC environment.¹¹² Fourth-generation BEs BE4 (BE derived from SpCas9) and SaBE4 (BE4 derived from SaCas9) offer greater efficiency with higher product purity than BE3.⁹⁵ BE3-Gam, SaBE3-Gam, BE4-Gam, and SaBE4-Gam improve product purity by minimizing the formation of unneeded insertions and deletions in the base editing process.⁹⁵ However, the target range of BE is restricted by the sequence of the G/C-rich PAM it contains. The CRISPR–Cpf1-based BE recognizes PAM sequences rich in T and catalyzes the C-T transition with low insertion-deletion and off-target editing levels by combining APOBEC1 with the Cpf1 of Aureobasidium pullulans.⁹⁶ The base targeting horizon expands after fusing APOBEC1 with the catalytically inactive dLbCas12a, which has a six base pair (bp) activity window and can recognize T-rich PAMs.⁹⁶ The xCas9 variant facilitates various non-NGG PAMs by BE or DNA cleavage, requiring highly accurate matching between gRNA and target sequences.⁹⁷ CBE was constructed by replacing Cas9 with xCas9 in the BE3 vector, which can edit target PAM sequences of NGN, GAT, and GAA.^{26, 110, 113} The BE comprising the Cas9 incision enzyme and human apolipoprotein B mRNA editing enzyme catalytic subunit 3A (APOBEC3A; A3A-PBE) has a 17 nt editing window and can effectively achieve C-T conversion.¹¹⁴ Engineered A3A (eA3A) domains can reduce bystander mutations and off-target effects. Indeed, eA3A-BE3 can correct point mutations in some human diseases more accurately than BE3.⁹⁷ Recently, a study has developed highly active minibase editors miABE and miCBE by fusing adenosine deaminase and cytidine deaminase structural domains, opening a new chapter for minibases in the field of DNA single-base editing.¹¹⁵

2.2.2 Glycosylase BEs

Based on CBEs, C-to-G BEs (CGBEs) or glycosylase BEs (GBEs) can realize the transversion between C to G and C to A (Figure 3B).¹¹⁶ CGBE1 induces C to G editing in human cells rich in AT sequences. CGBEs can be engineered for efficient and accurate base editing.¹¹⁶ GBEs comprise cytidine deaminase, the Cas9 incision enzyme, and uracil-DNA glycosylase (UNG). The GBE APOBEC-nickase-Cas9 (nCas9)-UNG can cause C-to-A conversion in Escherichia coli and C-to-G conversion in mammalian cells, enabling the targeting of G-C pathogenic variants.⁹⁹ The development of CGBEs and GBEs has extended the research of DNA damage repair mechanisms and BE applications.^{99, 117} BEs with C:G to G:C activity, comprising nCas9 fused with cytidine deaminase and a base excision repair protein, have been developed to target cytidine in the context of WCW (W represents A/T), ACC, or GCT sequences and the precise trinucleotide window of the target protosepta, with therapeutic potential to correct human genetic diseases.⁹⁸ In other studies, the tRNA adenosine deaminase (TadA)-8e was repurposed for cytosine conversion. An N46L variant was introduced into TadA-8e to eliminate its adenine deaminase activity and construct the first new CGBE/CBE family of base codification Td-CGBE/Td-CBEs independent of the Activation-induced deaminase/APOBEC (AID/APOBEC) deaminase family.¹¹⁸ Td-CGBEs and Td-CBEs enable efficient and accurate C-G-to-G-C editing with a very low insertion-deletion effect and off-target editing, which can potentially be used for the clinical transformation of gene therapy. More recently, a novel DNA BE, glycosylase-based guanine BE, which enables efficient guanine single-base editing and does not depend on deamidation reactions, has been successfully developed to further enrich the base editing toolkit.¹¹⁹

2.2.3 Adenine BEs

About half of the known pathogenic SNPs are represented by C-G to T-A mutations. Correcting most disease-associated human SNPs by converting A-T bps to G-C bps using base-editing techniques is possible.⁹³ Adenine BEs (ABEs) were developed by fusing TadA with the SpCas9 cleavage enzyme to induce A-T to G-C conversion, significantly expanding the scope of base editing (Figure 3B).⁹³ About a quarter of the SNPs causing genetic diseases in humans require translocation of adenine bases such as A-to-T and A-to-C or the complementary chains of T-to-A and T-to-G. The development of adenine transversion BE AYBEv3 filled the gap that current BEs cannot efficiently perform A-to-C or A-to-T editing.¹⁰⁰

Studies have developed deaminase-embedded inlaid BEs (IBEs) based on ABE using protein engineering.¹⁰¹ IBEs induce the conversion of bases from A-to-G, positioning the deaminase domain in the interior of Cas9 to avoid the PAM restrictions.¹⁰¹ ABE8e was developed to pair with various Cas9 or Cas12 homologs to efficiently install natural mutations and greatly improve editing efficiency, improving the efficiency and adaptability of adenine editing.^{102, 103} ABEmax and mini-ABEmax (TadA* fused with Cas9n) variants have been constructed to develop engineered ABEs with minimal RNA off-target activity, retaining DNA-targeted editing activity while reducing RNA editing activity.¹⁰⁴

ABE7.10 has low efficiency in editing challenging sites in primary human cells. Several studies have developed ABE8s through the evolution of ABE7.10, which effectively recapitulate the natural alleles on the hemoglobin (Hb) subunit gamma 1 (HBG1) and 2 (HBG2) promoters of the γ-Hb genes of human CD34⁺ cells to induce HbF persistence.¹⁰⁵ Another study designed TadA7.10, a commonly used adenosine deaminase that has a D108Q mutation and can reduce the cytosine editing activity of ABEs, ABE8e, and ABE8s at target sites and is compatible with mutations that reduce off-target RNA editing. This base-specific editing tool for T-C to T-T or T-C to T-G conversions broadens the usefulness of BEs.¹²⁰ ABEs pioneered the A-G transition for base editing, but mutations and off-target editing effects among bystanders may cause safety concerns for gene editing. ABE8e with an N108Q mutation has been found to reduce bystander editing of adenine and cytosine, so ABE9 with an editing window reduced to 1–2 nt almost eliminates the off-target editing of cytosine and reduces DNA/RNA off-target events to background level. Theoretically, it provides a targeted editing tool for correcting nearly 50% of genetic pathogenicity point mutations.¹⁰⁶

2.2.4 Dual ABEs and CBEs

Currently, available ABEs and CBEs can induce the modifications of only one type, limiting the scope of DNA alterations. Some studies have combined traditional CBE and ABE tools to develop two-base gene editing tools. Simultaneous programmable adenine and cytosine editing allows both A-to-G and C to T substitutions to be introduced and minimizes RNA off-target editing.¹²¹ Target-ACEmax comprises a Cas9 notch enzyme, cytosine deaminase Petromyzon marinus cytidine deaminase 1 (PmCDA1), and adenosine deaminase TadA and has high C-T and A-G editing activity. Dual ABE and CBE (A&C-BEmax) can achieve C-T and A-G conversion at the same target site and have high cytosine editing activity.¹⁰⁷ Compared with the single-BEs, A&C-BEmax shows significantly improved base transition rate, product purity, and simultaneous A/C conversion activity (Figure 3B).¹⁰⁸ Another study integrated a CGBE and ABE to develop the double deaminase-mediated base editing system A-to-G BE (AGBE), which can introduce four types of base conversion in mammalian cells: C-to-G, C-to-T, C-to-A, and A-to-G.¹⁰⁹ Compared with ABEs and CGBEs, AGBEs are efficient and safe, have a much broader editing window, have no apparent off-target effects at the DNA or RNA level, and are useful for building mutant libraries containing more gene mutation types.¹⁰⁹

2.3.1 PE1, PE2, PE3, and PE3b

The original PE1 is a wild-type RT of Moloney murine leukemia virus (M-MLV) ligated to the C-terminus of the SpCas9 H840A cut enzyme via a flexible junction to extend the pegRNA binding site (PBS) to 8–15 bases, PE1 can mediate cellular single-base transitions, insertions and deletions, but with low efficiency, with a maximum editing efficiency of only 0.7–5.5%.¹²² To improve the base editing efficiency of PE1, the study introduced five mutations, D200N, L603W, T306K, W313F, and T330p, into M-MLV RT and constructed PE2. PE2 increased the editing efficiency to 1.6–5.1-fold of that of PE1, and with a shortened sequence of the PBS, was able to perform more targeted insertions and deletions, and the study further showed that that the first base of pegRNA in the 3' extension is not a C also contributes to the editing efficiency of PE2. Since the newly synthesized DNA editing strand in PE2 needs to undergo an endogenous repair process before it can be copied to the complementary strand, the study added additional sgRNAs to the PE2 system for guiding the SpCas9 H840A incision enzyme to cleave the unedited strand, which further boosted the editing efficiency of PE3 to 1.5–4.2-fold that of PE2, but since there will be two DNA strands at the same time nicks, PE3 has the probability of generating an indel. Studies optimized sgRNAs with spacers matching the edited strand to avoid editing the original allele, generating the PE3b BE, and PE3b significantly reduced the formation of insertion deletions whilst providing editing efficiencies similar to those of PE3. The study confirmed that PE3 was able to achieve HBB E6V mutations in HEK293T cells with 44% efficiency and less than 5% insertional deletions, suggesting that PE has the potential for use in the treatment of human diseases.

2.3.2 PE4 and PE5

After the development of PE1, PE2, PE3, and PE3b, it was found that DNA MMR in the PE system is a major limiting factor for editing efficiency and increases the production of by-products. The study optimized PE2 and PE3 using a transiently expressed dominant-negative MMR protein (MLH1dn), which is less susceptible to MMR activity, and constructed PE4 (PE2+MLH1dn) and PE5 (PE3+MLH1dn), two editors that were 7.7-fold and 2.0-fold more efficient than the editing efficiencies of PE2 and PE3, respectively.¹²⁶ The study further designed and optimized Cas9 proteins and nucleation signals, and so on, and synergistically applied PEmax and epegRNA with PE4 or PE5, which showed greater enhancement of editing efficiency for PE4 and PE5, and constructed PE4max and PE5max, which showed higher editing efficiencies in all seven mammalian cell types.

2.3.3 Homology 3' extension-mediated PE

Based on the developed PE editors, a homology 3' extension-mediated PE (HOPE) has been developed. The two pegRNAs of HOPE target the positive and negative DNA strands, respectively, and perform precise gene editing at the same locus of the genome, which greatly improves the editing efficiency of PE, and achieves highly efficient editing in HEK293T and human colorectal cancer cells HCT116 editing. Compared with PE2 and PE3, the efficiency of HOPE base editing was higher than that of PE2, and the purity of the product was better than that of PE3, which extended the application prospect of PE editing. The study also provided an in-depth investigation of factors such as the spacing of DNA double-stranded PAMs and the length of PBS versus RT, which provided directions for the optimization of double pegRNAs.¹²⁷

2.3.4 Twin plasmid editor

Twin plasmid editor (twinPE) is a new version of PE technology that addresses the limitation of the initial PE to edit only a few dozen bps, enabling safe and precise insertion of complete genes into human cells.¹²⁸ twinPE uses a plasmid editing protein and two pegRNAs to provide programmable DNA sequence replacement or excision at human endogenous genomic loci, enabling targeted DNA plasmid (>5000 bp) incorporation and reversal of 40 kb of the target sequence in human cells, expanding the ability to perform precise gene editing with few by-products (12). In twinPE editing, each pegRNA directs the editing protein to create single-stranded gaps in the DNA at different target sites in the genome, avoiding the generation of meaningless byproducts, and subsequently synthesizing the desired sequence between the gaps, thereby inserting, replacing, or deleting sequences of up to 800 bps. twinPE may act synergistically with another tool to make corrections or additions to large or complex human disease-causing genes for correction or complementation.

2.3.5 PE6

For further optimization of the PE system, the study used protein evolution and engineering to generate PE editors with smaller sizes and higher efficiency. The study first developed RTs different from M-MLV RT and then used PE editing phage-assisted sequential evolution to screen and evolve three RTs with smaller sizes, Gs RT, Ec48, and Tf1, to generate PE editors that are smaller than the previous PE editors. It was found that different RTs specializing in different types of PE editors helped to improve PE editing efficiency, and improved structural domains of Cas9 also had a positive effect on editing efficiency. The study obtained a series of variants that are superior to previous PE editors in terms of size (PE6a and PE6b), RT activity (PE6b, PE6c, and PE6d), and Cas9-dependent editing efficiency (PE6e–PE6g) through the screening and modification of different RTs and the evolution of Cas9, and a series of variants that are more efficiently customized according to the actual needs of gene editing can be selected.¹²⁹

2.3.6 PE7

Considering that the PE system is currently a more recent gene editor, a study using a genome-wide scale CRISPR interference screen (CRISPRi screening) identified the small RNA-binding protein La (La protein) as a key regulator of PE efficiency. La binds polycytidylic acid at the 3′ end of RNA polymerase III transcriptions, and La was found to functionally interact with the 3′ end of polycytidylic acidic lead editing gRNA (pegRNAs). The study fused La proteins and LA(1–194) truncated proteins into PEmax and showed that fusion at the C-terminal end of La(1–194) fusion construct (PEmax-C) was the most efficient, and the study thus constructed the PE7 editor. The editing efficiency of PE7 in U2OS cells was significantly higher than that of PEmax, and the construction of PE7 provides new ideas for the optimization and development of PE editors.¹³⁰

4.1.1 Induction of HBB gene correction

SCD mutation correction is an effective strategy for preclinical studies of SCD using gene editing technology, which is divided into two methods: one is to directly repair the mutated SCD codon (valine) to the normal codon (glutamate), and the other is to convert the mutated SCD codon to the benign nonsickle G-Makassar variant (alanine; Figure 4B).²¹⁸

Direct correction or modification of genetic variation is a strategy to cure inherited diseases, preserving elements of regulation that maintain normal gene function and avoiding the adverse consequences that might result from mutating other genes.²²⁰ Correction of GTG to GAG in SCD patients is a simpler gene therapy approach. Previous studies have used the Cas9 ribonucleoprotein (RNP)/single-strand oligonucleotide donor (ssODN) to edit the SCD SNP in K562 cells or induce the editing of HDR-mediated mutations in the SCD HBB gene in HSPCs, achieving up to 33% SCD mutation sequence replacement. The differentiation of corrected HSPCs into RBCs produces wild-type Hb, resulting in a decrease in sickle β-globin and an increase in adult wild-type β-globin and fetal β-like (γ) globin.¹⁴⁵ Targeting abbreviations in the HBB locus were reduced using eSpCas9-1.1 and HF1, and an off-target analysis performed using a web-based software tool showed minimal off-target gene activity.²²¹ CRISPR–Cas9 and recombinant adeno-associated virus serotype 6 (rAAV6) targeted editing of HSCs achieved homologous recombination of the HBB gene, effectively correcting SCD-associated amino acid mutations at the HBB gene codon 6 (E6V) in HSPCs from multiple patients with SCD, realizing the treatment of patients with SCD.¹⁴² HiFi Cas9 mediated the correction of high p.E6V mutation levels in SCD-causing HSPCs of patients with SCD.²⁴ The HiFi Cas9 variant achieved more specific gene correction than hSpCas9 in SCD CD34⁺ HSPCs. The off-target sites were identified using the web-based simulation mismatch search tool COSMID (CRISPR Off-target Sites with Mismatches, Insertions, and Deletions) and genome-wide unbiased mismatch search tool GUIDE-Seq (Genome-wide Unbiased Identification of Double-stranded breaks Enabled by Sequencing).^{222, 223} The off-target effect of HSPC implantation in vivo was found to be very low, proving that the gene editing of HSPCs in patients with SCD was an effective gene therapy approach.²²⁴ Reversing the sickle phenotype in patients using HSCT requires at least 20% donor myeloid chimerism (i.e., achieving at least 20% of the HSC correction level).²²⁵ The genomic editing technology potentially corrects SCD mutations in HSCs to simultaneously produce adult Hb tetramers (HbA) and eliminate sickle Hb (HbS). In vitro, HiFi Cas9-AAV6 has been used to correct HBB gene mutations in HSCs from patients with SCD. In vivo, it achieved 20% correction efficiency in human-derived globin cluster Townes-SCD mouse models, achieving stable Hb A (HgbA) production and restoring RBC function.¹⁴⁸ HSC-targeted genome editing has been simulated in xenografted mice and rhesus monkeys in β-βs-globin conversion models, resulting in therapeutic-level gene correction in mice.¹⁷¹ Studies further developed a therapeutic cell product (gcHBB-SCD) for HBB correction in HSPCs using ex vivo Cas9-RNP and rAAV6 to achieve up to 34% HBB allele correction in HSPCs from donor patients with SCD without abnormal hematopoiesis, genotoxicity, or tumorigenicity.²²⁶ It is safe, effective, and reproducible for initiating phase I/II clinical trials in patients with SCD. Another study found that a high-fidelity Cas9 (R691A) variant enabled efficient gene editing in HSPCs and better in RNP delivery.²⁴

HSPCs are regular cellular targets for gene therapy of hematological diseases, frequently used in most studies. Induced pluripotent stem cells (iPSCs) have also been used to study SCD. iPSCs could be generated from various somatic cells with significant potential in personalized cell therapy.²²⁷ Correction of pathogenic iPSC mutations can restore function and provide an abundant source of cells for transplantation. Therefore, genome editing of autologous HSPCs and iPSCs mediated by CRISPR–Cas9 is a therapeutic option ideal for treating inherited hematologic diseases.²²⁷ iPSC banks have been established for patients with SCD with different ethnicities, HBB haplotypes, and HbF levels. iPSC technology is an ideal choice for screening drugs against the genetic variants found in the patient population, supporting specific therapies for patients with SCD.²²⁸ The gene editing of iPSC lines derived from patients with SCD using Cas9 RNP combined with an AAV6 vector achieved a monoallelic editing frequency of up to 94% at the HBB locus, greatly improving the clinical applicability of gene editing.²²⁹ The PE system has realized efficient target editing with few insertions and deletions and no detectable off-target editing. HEK293T cells homozygous for the SCD-inducing E6V mutation were treated with PE3 and programmed pegRNA, achieving 26–52% effective correction of the HBB E6V mutation to wild-type HBB.¹²² Recently, studies have used PE to correct the HBB^S allele in HSPCs from patients with SCD into the wild-type HBB^A allele. After transplantation into immunodeficient mice for 17 weeks, the edited HSPCs maintained the HBB^A level. It also showed similar hematopoietic differentiation to unedited HSPCs derived from healthy donors, supporting the feasibility of one-time PE for SCD treatment.²³⁰

The HbS allele of β-globin represents the change from GAG (Glu) to GTG (Val) at amino acid six of HBB, and this SNP is converted into GCG (Ala) by A•T-G•C conversion on the opposite strand using ABE, producing the HBB E6A genotype, the Macassar allele (HbG). There was no pathogenicity in homozygous or heterozygous individuals.²⁸ The targeting range of SpCas9 and its engineered variants is mainly limited to G-base-containing PAM sequences. Novel SpCas9 variants have been developed to effectively edit the bases of more pathogenic SNPs that are not G-PAM. Three newly evolved SpCas9 variants have been developed that jointly identify NRNH PAMs (H represents A, C, or T) and convert SCD-associated SNPs into Makassar mutations by A-T to G-C base editing via a CACC PAM. ABE-NRCH has shown the greatest editing activity with 41 ± 3.8% base conversion.²⁸ Base editing strategies using novel PAM variants offer a hopeful treatment for patients with SCD. In SCD, the GAG (Glu) codon encoding β-globin amino acid 6 mutates to GTG (Val), and the single BE ABE converts the A-T bp to G-C in living cells without needing DSBs or donor DNA templates. That is, it converts the pathogenic SCD-associated codon into GCG (Ala) to produce the rare, occurring naturally, nonpathogenic HBB variant Hb-Makassar (HBB^G), where editing more than 20% of HBB^S–HBB^G is sufficient to rescue the SCD phenotype.⁸ Some studies have developed IBEs based on ABE, which edited the pathogenic sickle cell Hb alleles in HSPCs derived from patients with SCD and transformed them into HbG-Makassar variants, which can potentially treat SCD.¹⁰¹ In vitro, editing of human HSPCs has been studied using a customized ABE (ABE8e-NRCH) by electroporation of ABE8e-NRCH+sgRNA RNP or ABE8e-NRCH mRNA with sgRNA (base editing using mRNA or RNP). The SCD-associated HBB^S allele was converted to HBB^G with a conversion rate of up to 80%.²³¹ The edited HSPCs could be permanently edited when transplanted into immunodeficient mice, with a 68% HBB^G frequency at 16 weeks posttransplantation. ABE8e-NRCH has the smallest nonresting bypass editing in patient CD34⁺ HSPCs. A single autologous treatment eliminates pathogenic HBB^S, produces benign HBB^G, and minimizes the DSB adverse consequences, suggesting that in vitro autologous base editing and HSC transplantation are promising methods for treating SCD.⁸

4.1.2 Induced increase in HbF

The γ-globin and α-globin subunits pair to form HbF.²³² HbF passivates the pathophysiology of SCD and moderates the clinical course. Increasing the HbF level is one of the effective treatments to improve the clinical manifestations of SCD.^{204, 233} The fetal γ-globin genes HBG1 and HBG2 are expressed in late gestation and induce erythrocyte HbF production. Around birth, HBG1 and HBG2 expression is suppressed, γ-globin levels decrease, and β-globin levels increase. While HbF is gradually converted to HbA under normal conditions, HbF is converted to HbS in patients with SCD. Therefore, inhibiting the conversion of γ-globin to β-globin is a useful strategy for treating SCD.²¹⁸ Gene editing technology can repair β-globin mutations in the HSCs of patients with SCD, add β-globin genes with an antisickle effect, activate target genes related to endogenous γ-globin expression, improve γ-globin levels, promote normal HbA formation, generate genetic changes to induce HbF expression, improve HbF levels, and obtain long-term phenotypic correction.²³⁴ KLF transcription factor 1 (KLF1) is a critical regulatory molecule engaged in the γ-β-globin gene conversion process, directly inducing β-globin gene expression and indirectly suppressing γ-globin expression. CRISPR–Cas9 can induce the insertion or deletion of the KLF1 gene in adult erythroid progenitor cells to promote the expression of the γ-globin gene and increase the HbF level.²³⁵ In HSPCs derived from patients with primary SCD, CRISPR–Cas9 targeted the deletion or inversion of a genomic region of the HbF silencer between a 13.6 kb δ- and γ-globin gene and a putative gamma-δ gene. It reactivated HbF synthesis in adult erythrocytes and induced high γ-globin expression in erythrocytes, which plays an antisickling role in SCD.²³⁶ CRISPR–Cas9 (Helper-dependent adenoviral vectors (HDAd)–HBG–CRISPR) was also used to transactivate HSPCs from human β-globin locus transgenic (β-YAC) mice in vitro and in vivo, disrupting the repressor binding region in the γ-globin promoter and converting human β-globin to γ-globin in adult mouse erythrocytes. The treatment of SCD was simplified, and no hematological abnormalities were observed, indicating that HBG1/2 promoter editing is a safe method for treating SCD.¹⁵³ Another study found that CRISPR–Cas9 genome editing of the repressor binding site within the γ-globin promoter could reactivate endogenous γ-globin. The gene-edited HSPCs were transduced into healthy mice and SCD mouse models, significantly increasing γ-globin levels in the erythrocytes of SCD mice, reaching 30% of the α and βS chains of adult human cells. Their HbF expression level was also increased to that of normal cells, completely correcting the sickle cell phenotype and providing new insights into treating SCD.¹⁵⁵

Oncogenes BCL11A and ZBTB7A (formerly called lymphoma-related factor [LRF]) are cis-regulatory elements (CREs) involved in inhibiting γ-globin expression, which directly bind separately to fetal β-globin promoter elements located at −115 and −200, shutting down HbF genes and inhibiting HbF expression early in life.^{156, 237-240} Reversing BCL11A inhibition of γ-globin expression and reactivating HbF via gene editing represents a promising strategy for SCD therapy (Figure 4B).^{241, 242} After editing the GATA binding protein 1 (GATA1) binding site in the BCL11A enhancer +58 using the Cas9 RNP complex to reduce BCL11A expression and increase fetal γ-globin, the HbF level in patients with SCD increased from 13.9 to 47.5%, indicating that the gene editing resulted in long-lasting HbF induction, and CIRCLE-seq (Circularization for In vitro Reporting of CLeavage Effects by Sequencing) did not detect off-target activity.²⁴³ In addition, a Cas9:sgRNA RNP complex was constructed to target the BCL11A binding motif in the γ-globin gene promoter in HSPCs, abolishing the BCL11A binding site and increasing the HbF level to a therapeutic level, which is a highly specific and safe treatment strategy for SCD.²⁴⁴ Gene editing of erythrocyte-specific BCL11A enhancers in HSPCs of patients with SCD can also induce HbF expression without detectable toxicity.^{244, 245} Due to the similarities between NHPs and humans in genomic mapping and hematopoietic and erythropoietic production, one study evaluated the autologous implantation and induction potential of HSPCs edited with erythrocyte-specific BCL11A enhancers in four NHPs.¹⁷⁰ The edited HSPCs lasted for at least 101 weeks after transplantation, produced strong γ-globin induction during erythropoiesis, provided therapeutic HbF levels in peripheral blood erythrocytes, and had no apparent hematological toxicity or off-target gene editing, suggesting that BCL11A erythrocyte enhancer genome editing is promising for further clinical transformation. There is frequent cytosine base editing at the BCL11A+58 RBC enhancer. Some studies have used BEs to target the +58 BCL11A RBC enhancer in HSPCs to transform C into T, G, or A, improve the globin chain imbalance, and further achieve efficient multiple editing by jointly abolishing the BCL11A RBC enhancer and correcting the HBB-28A>G promoter mutation to induce HbF expression and prevent sickle cell formation in RBC progeny.²⁴⁵ Base editing can also treat SCD by disrupting the CREs that negatively regulate HBG1/HBG2 expression to increase HbF levels. Previous studies have used ABEmax to insert an A•T to G•C change in the CREs of BCL11A, intergenic HBS1-like translational GTPase (HBS1L) to MYB proto-oncogene, transcription factor (TF) (MYB) region, KLF1, and β-like globin genes controlling erythrocyte HbF expression to quantify their impact on γ-globin gene regulated HbF expression, increases in HbF levels, and provide potential therapeutic insights for SCD.²⁴⁶ Studies have developed lentiviral vector (LV)-based gene addition and CRISP–Cas9 SCD treatments. HSPCs were genetically modified by combining antisickling globin (AS3) in Hb tetramers that disrupted the BCL11A binding site in the HBG promoter with CRISPR–Cas9-mediated disruption or downregulation of BCL11A-induced γ-globin reactivation.²⁴⁷ Editing of the HBG promoter enhanced the healing effect of the AS3 gene addition strategy, increased the total Hb content in β-globin treated transgenic cells, and induced HbF expression.

In addition to targeting HbF repressors such as BCL11A, another gene editing strategy to treat SCD uses gene editing technology to induce deletion of the β-globin gene cluster with rare naturally occurring β-globin mutations in patients with SCD, resulting in HPFH-related or similar mutations to improve SCD clinical severity.²³⁶ HPFH is a rare benign disease caused by mutations upstream of the transcriptional start site in HBG1 and HBG2. HPFH induces sustained high HbF expression by reducing the conversion of γ-globin to β-globin and disrupting the binding element of BCL11A, a γ-globin gene repressor (Figure 4B).^{248, 249} Some studies used CRISPR–Cas9 to modify the 13-nt sequence in the HBG1/2 gene promoter of HSPCs to simulate the HPFH-related mutation. After gene editing, the HbF level in RBCs produced by HSPCs increased, reaching a potentially therapeutic level and providing a theoretical basis for treating SCD with HPFH-related mutations induced by gene editing.²⁴⁹ Some studies have used CRISPR–Cas9 to introduce HPFH-related point mutations into erythrocytes, disrupt the γ-globin gene promoter, inhibit BCL11A and ZBTB7A/LRF binding to the γ-globin gene promoter, induce and enhance γ-globin gene expression, improve the HbF level, and effectively improve the clinical symptoms of SCD.²¹⁹ Other studies have used CRISPR–Cas9 to edit γ-globin repressor ZBTB7A/LRF binding sites to simulate HPFH mutations, restore HbF synthesis, and correct SCD phenotypes, suggesting that ZBTB7A/LRF binding sites are effective targets for SCD genome editing therapy.²⁵⁰ In addition to the above, TAL1 is an erythroid regulator that acts synergistically with GATA1, enabling T>C at the TAL1-GATA1 binding site at γ-bead protein-175 to mimic HPFH.²⁵¹ Studies have found that deleting the CTCF binding site (CBS) at the 3′ end of the β-globin locus using CRISPR–Cas9 technology induces HPFH, increases γ-globin gene expression, and induces HbF expression in HSPCs. Gene editing of the CBS 3′HS1 is a new therapeutic strategy for SCD.²⁵² ABE and CBE have been used to introduce −123T>C and −124T>C HPFH-like mutations within the homologous HBG1 and HBG2 promoters to drive γ-Hb enrichment by creating a KLF1 binding site that increases HbF-positive cell numbers and overall HbF production.²⁵³ HPFH is also associated with cascades of deletions within the β-globin locus downstream of the fetal HBG2 gene. Using CRISPR to induce mutations that reduce Hb subunit beta (ΗΒΒ) promoter activity and increase HbF expression has been studied as an effective treatment for SCD.²⁵⁴ There have been studies using ABE (ABE8e) for installing the −113 A>G HPFH mutation in the γ-globin promoter of healthy CD46/β-YAC mice with the human β-globin locus, inducing >60% effective −113 A>G conversion, activating γ-globin expression, and correcting the erythrocyte phenotype.¹⁵⁴ No off-target editing was detected in CIRCLE-Seq, suggesting that base edit-induced HPFH mutations have clinical potential in SCD. Some studies have used base editing ABE and CBE to edit the HBG promoter −200 region of HSPCs in patients to induce HPFH-like mutations, create KLF1 activator binding sites, and induce high levels of HbF expression.¹⁵⁰ Base editing of the HBG promoter avoids insertion, deletion, and large amounts of genome rearrangement and can induce higher γ-globin levels and save the SCD cell phenotype. GUIDE-seq found that a small amount of off-target production did not significantly affect protein expression, suggesting that base editing-induced HPFH mutation is a safe strategy for SCD treatment.

4.2.1 Targeting the HBB gene

Correction of HBB gene mutations in β-thalassemia using gene editing contributes to the restoration of functional HBB gene expression. One study used CRISPR/Cas9 to achieve targeted insertion of the HBB cDNA in exon 1 of the HBB gene to correct the HBB mutation in human iPSCs derived from β-thalassemia patients, which allowed the iPSCs to expand and differentiate to generate erythrocytes that could produce corrected HBB protein.²⁸⁰ A study applied CRISPR–Cas9 combined with piggyBac transposon to seamlessly correct the HBB mutation in iPSCs derived from β-thalassemia patients. The gene-corrected iPSCs did not show off-target effects and restored the expression of the HBB gene when differentiated into erythrocytes in vitro, suggesting that stem cell-based gene therapy could be applied to single-gene disorders.²⁸¹ It has been shown that seamless correction of β-41/42 deletion mutations in β-thalassemia patient-specific iPSCs has been achieved using CRISPR–Cas9 and ssODN with high efficiency and that the corrected cells exhibited minimal mutation loading, no off-target mutagenesis, and the genetically corrected cells expressed normal β-globin transcripts when the iPSCs were differentiated into erythrocytes.²⁸² Mutations in β-thalassemia produce aberrant splice sites, disrupting the splicing of the HBB gene and interrupting the alleles of the aberrant splice sites are effective methods to restore gene function. It was shown that CRISPR–Cas9 targeted correction of the IVS-1-110 (G>A) mutation site in the HBB locus obtained the wild-type HBB gene, confirming that gene editing is effective in HBB gene correction.²⁸³ β⁶⁵⁴-Thalassemia is the major subtype of β-thalassemia in China. The study used CRISPR/Cas9 to target the receptor splice sites of IVS2-654 (C>T) and IVS-2-579 to correct aberrant β-bead proteins for RNA splicing and to improve the clinical symptoms of β⁶⁵⁴ mice, and the gene-edited heterozygous β⁶⁵⁴ mice showed significantly higher survival rates.¹⁷³ Another study used Cas9 and Cas12a RNPs in β-thalassemia HSPCs to disrupt aberrant splice sites by acting on the IVS1-110G>A and IVS2-654C>T mutations, respectively, and erythrocyte progeny of edited patient HSPCs showed reversal of aberrant splicing and restoration of β-globin expression, effectively correcting a portion of the TDT genotype.²⁸⁴ Studies using multiple neighboring sgRNA-guided CRISPR–Cas9/RNP editing of HSPC were able to induce deletions of large segments, directly restoring the damage to HBB gene function caused by the IVS2-654 (C>T) mutation without off-target effects, making it a highly effective and safe treatment for β-thalassemia.²⁸⁵ In a study, PE3 was used to correct the IVS-II-654 mutation (C>T) in a mouse model of β-thalassemia, and components of PE3 were microinjected into the fertilized eggs of the mice, and gene-corrected mice with an editing efficiency of 14.29% were successfully obtained, and editing of PE3 restored the normal splicing of β-globin protein mRNA, which resulted in curative treatment of the mice, and off-target analyses confirmed the genomic safety of PE3 editing.²⁸⁶ One additional study used CRISPR–Cas9 to correct the β⁰³⁹ mutation, which resulted in a significant increase in the efficiency of production and accumulation of β-cadherin and HbA. β-Thalassemia is associated with the number of inherited α-globin genes, and when the β-globin chain is absent, free α-cadherin deposition disrupts the cellular membranes, generating ineffective erythrocytes and hemolysis. It has been shown that induction of HBA2 deletion in human HSPC using CRISPR/Cas9 can reconstruct α-thalassemia traits downregulate α-globin expression, and subsequently improve β0-thalassemia phenotypes by targeting the endogenous HBA promoter to combine the deletion of HBA2 with the gene replacement of HBB to increase β-globin gene expression.²⁸⁷ The HBB-28 (A>G) mutation is one of the most common mutations in β-thalassemia patients in China and Southeast Asia. A study has used BE to precisely correct the HBB-28 (A>G) mutation in primary cells of patients, with a gene correction efficiency of more than 23.0% in nuclear-transplanted embryos, and a BE variant with a narrow deamidation window could precisely facilitate HBB – 28 loci, confirming the promise of base editing systems for the curative treatment of human somatic and embryonic genetic diseases.²⁸⁸ There is a study using CRISPR/sgRNA-RNP complexes with rAAV6 donor-mediated HDR to edit patient-derived HSCs, to perform in situ gene correction for different HBB mutations in HSCs, to restore β-globin content in differentiated erythrocytes, and, the edited HSCs have the ability to be transplanted into the bone marrow of immunodeficient NOD–scid–IL2Rg−/− mouse bone marrow, providing an effective and safe method for the treatment of β-thalassemia patients by HBB locus correction.²⁸⁹ HbE β-thalassemia accounts for approximately 50% of all types of β-thalassemia, and severe β-thalassemia is caused when there is a point mutation in the codon 26 allele of the human HBB gene (GAG→AAG, E26K), and any mutation at the other locus of the allele. The study used the CRISPR/Cas9 system to correct the HBB gene in iPSCs from HbE/β-thalassemia patients, and the corrected iPSCs could be differentiated into HSCs for autologous transplantation in HbE/β-thalassemia patients.²⁹⁰ Another study used the ABE base editing strategy to correct HbE mutations to wild-type or Hb E26G with a normal phenotype of the Hb Aubenas variant, achieving editing efficiencies of up to 90% in primary human CD34+ cells and analyzing the off-target effects using CIRCLE-seq, confirming the base editing strategy as a therapeutic strategy for HbE β-thalassemia.²⁹¹

4.2.2 Induced increase in HbF

In addition to the correction of gene mutations, inducing an increase in HbF expression is an effective strategy for the treatment of β-thalassemia.²⁷⁷ Restoration of HbF expression has a positive therapeutic effect in most TDT patients compared with repair of the HBB gene.²⁹² Consistent with SCD, HPFH in β-thalassemia likewise improves the clinical phenotype of patients and is a promising treatment for β-thalassemia.²⁹³ After disruption of the BCL11A binding site in the HBG1/HBG2 promoter, the introduction of the natural HPFH mutations -113A>G, -114C>T, -1175T>>C, -195C>G, and -198T>C into HSPC using CRISPR–Cas9/AAV6 significantly increased HbF expression.²⁹⁴ In addition to BCL11A, regulators of HbF such as KLF-1, ZBTB7A/LRF, SOX6, ZNF410, and RBM12 expanded the range of possible genome editing targets.^295-297 The study used a transforming BE (tBE) to directly disrupt the BCL11A binding motif in the HBG1/2 promoter, which could produce durable therapeutic editing in HSPC from β⁰/β⁰ thalassemia patients, significantly enhancing the level of γ-globin, increasing the level of HbF expression, and without detectable off-targeting.²⁹⁸ Transplantation of tBE-edited HSPC into immunodeficient mice was detected 4 months later to reconstitute the hematopoietic system in the mice. Both in vivo and in vitro experiments confirmed that tBE editing of BCL11A binding motif induced HbF levels superior to ABE8e and hA3A-BE3 editing of BCL11A enhancers, and was a highly efficient strategy to reactivate HbF in patients’ HSC. There is a study that used CRISPR–Cas9 to disrupt a 13.6 kb genomic region containing the δ- and β-globin genes as well as a silencing factor for HbF between the γ-δ genes, which increased γ-globin expression while inducing activation of HbF, with beneficial effects on the treatment of β-thalassemia.²³⁶ It has been shown that deletion of a 200 bp genomic region encompassing the GATAA motif within the BCL11A enhancer of the K562 cell line using CRISPR–Cas9 induced hereditary persistence of HPFH, which resulted in increased HbF expression.²⁹⁹ The study used CRISPR–Cas9 to correct gene mutations in erythroid precursor cells from homozygous β039 thalassemia patients, inducing accumulation of the β-globin gene and efficient production of HbA without genomic toxicity.³⁰⁰ It was further shown that HbA and HbF can be induced using rapamycin to induce ab initio expression of HbA and HbF at the same time as CRISPR–Cas9 editing of the β-globin gene, inducing an increase in HbF.³⁰¹ In order to investigate the difference in the effect of a single gene editing method versus multilocus editing, the study used CRISPR–Cas9 to simultaneously edit both cis (HBG-promoter binding site) and trans (BCL11A enhancer) target sites and maintained the same high editing efficiency in NBSGW mice as in CD34 cells.³⁰² The study optimized the vector capable of editing the HDAd5/35 CRISPR of HSPC to the adenoviral vector Ad-dualCRISPR that could be used for dual genome editing to further increase the level of HbF reactivation in β⁰/β⁰-thalassemia mice, expanding the therapeutic window of gene editing to provide a safe and effective therapeutic strategy for phenotypically severe patients. In addition to gene editing using CRISPR–Cas9, a study developed a near PAM-free ABE variant, ABE8e-SpRY, which uses ABE8e-SpRY RNP to correct the HbE (CD 26, G>A) and IVS II-654 (C>T) mutations in patient-derived CD34 HSPC, disrupting the BCL11A enhancer +58 site, upregulates the expression of γ-globin in the BCL11A enhancer or HBG promoter, mimics the HPFH mutation, induces HbF levels at or above the clinical threshold, produces durable therapeutic editing in human HSC and confirms that the editing has a safety profile using the Cas-OFFinder tool.³⁰³ There were also studies comparing five strategies using CRISPR–Cas9 or ABE in HSPC, which showed that base-edited -175A>G erythrocyte colonies expressed 81±7% of HbF, whereas unedited controls expressed 17±11% of HbF and CRISPR–Cas9 induced lower levels of HbF than base editing.³⁰⁴ When HSPC was transplanted into mice, base editing induced -175A>G to produce HbF equally better than CRISPR–Cas9, and the study demonstrated better therapeutic efficacy of base editing compared with CRISPR–Cas9. A study using CRISPR/Cas9 and BE to edit HSPC derived from β-thalassemia patients achieved efficient editing of the HBG promoter, which resulted in activation of γ-globin and a significant enhancement of HbF expression levels.³⁰⁵ Studies have utilized the HDAd5/35++ vector to deliver ABE, target editing of the BCL11A enhancer, or induce HPFH mutations in the HBG1/2 promoter, to efficiently achieve target base conversion and reactivate γ-globin production.³⁰⁶ It was shown that the ratio of γ-globin to erythrocytes in the peripheral blood of the ABE vector targeting the -113A>G HPFH mutation (HDAd–ABE–sgHBG-2) was greater than 40% with no detectable off-target editing, and the rate of HBG1/2 deletion after ABE editing was much lower than that induced by CRISPR/Cas9 in the β-YAC mouse model, suggesting that the HDAd-vector delivery of ABE is a promising strategy for the treatment of β-hemoglobinopathies.

4.3.1 Acute myeloid leukemia

AML is an aggressive disease with defects in normal hemopoiesis.³¹¹ Gene editing technology can be applied to genome-wide screening of AML, which can help to identify targets for differential therapy in AML.³¹² One study established a CRISPR screening method using an in situ xenograft model and confirmed that SLC5A3, which is associated with AML cell metabolism, and MARCH5, a key gene for apoptosis, are candidate targets for AML treatment.³¹³ Double-negative T cells (DNT) are able to kill most AML patients’ primitive cells, but about 30% of AML patients’ primitive cells are resistant to the cytotoxicity of DNT. Studies using CRISPR/Cas9 screening to identify genes in AML cells that are susceptible to DNT treatment have shown that inactivation of CD64 leads to resistance in AML and that the expression level of CD46 is associated with the AML cell susceptibility to DNT correlated.³¹⁴ In addition to CRISPR screening for AML, gene editing technology has also enabled the treatment of AML by editing CAR-T cells. CAR-T cell therapy has an important role in hematological tumor therapy, and one study used CRISPR/Cas9-RNP to knock down CD45 in CAR-T cells, which after long-term cultivation in vitro still had efficient therapeutic effects.³¹⁵ In T cells and CAR-T cells from AML patients, the expression of activated CD69 and HLA-DR as well as the markers of exhaustion PD1 and LAG3 was increased, leading to a decrease in antitumor efficacy.³¹⁶ Study has combined CRISPR/Cas9 with the viral-free gene transfer strategy of the Sleeping Beauty transposon to edit CAR-T cells from patients, generating HLA-I^KO/TCR^KO CAR-T cells with HLA-I and TCR complexes removed, with a high degree of specificity and safety of gene editing, which has enhanced the therapeutic efficacy of CAR-T cells in AML and is a promising option for the treatment of AML CAR-T cell therapy is limited by the lack of cancer-restrictive surface markers, and CAR-T targeting of CD33, a classical marker in AML, destroys normal myeloid cells, the study used CRISPR/Cas9 to knockdown CD33 in normal HSPC, and achieved long-term implantation and differentiation in immunodeficient mice and rhesus monkeys, and the xenografts of CD33 KO HSPC with CD33 CAR T constitutes a functional hematopoietic system that can specifically target AML and enable treatment of AML without myelotoxicity, and the findings confirm that on-target and off-tumor toxicity could be avoided by genetically engineering the host, providing an important reference for selective tumor targeting in cancer immunotherapy.³¹⁷ According to a study using a CD276-CAR construct to transduce natural killer (NK)-92 cells and CRISPR–Cas9 to knock down three different inhibitory checkpoints (CBLB, NKG2A, and TIGIT) in NK cells, triple-knockdown CD276-CAR-NK-92 cells showed a significant enhancement of toxicity against leukemia cells, and are a promising candidate for the treatment of AML and B cell acute lymphoblastic leukemia.³¹⁸

4.3.2 Multiple myeloma

As a malignant tumor of the hematological system, multiple myeloma (MM) is diagnosed in approximately 588,161 people worldwide each year. In MM, there is an abnormal growth of clonal plasma cells in the bone marrow, which may cause symptoms such as kidney damage, osteolytic bone disease, and anemia.³¹⁹ Moreover, immune dysfunction in patients with MM increases the risk of infectious diseases, and currently, there is still no curative treatment option for MM, only a prolongation of the patient's survival.³²⁰ Consistent with its role in AML, CRISPR gene editing technology is used for genome-wide screening of MM therapeutic targets and editing of immune cells in vitro. As existing therapeutic options do not target specific oncogenic mutations in MM, one study used CRISPR for genomic analysis of 19 MM cell lines and hundreds of normal cell lines to systematically characterize the molecular-dependent preferences of MM and identify 116 important genes affecting MM, which provide valuable targets for targeted therapy in MM.³²¹ The clinical use of new drugs has improved the treatment of MM, but most patients end up relapsing due to drug resistance, and the relationship between patient-specific mutations and sensitivity to drugs has not been explored. The study performed whole exome sequencing of patient samples and CRISPR–Cas9 resistance screening for relapse-specific mutations against commonly used drugs such as dexamethasone and identified 15 functional mutations that were resistant to different drugs, and the study of resistance due to specific genetic alterations could help to personalize the treatment of MM patients.³²² Unlike CAR-T cell therapy, T-cell receptors (TCRs) recognize antigens of intra- and extracellular origin, expanding the scope of targeted therapy. Transfer of tumor antigen-specific TCR genes into T cells is the more popular therapeutic option, and T-cell therapies generating high-frequency transgenic TCRs (tgTCRs) have improved TCR gene therapy. Studies using CRISPR/Cas9 to simultaneously edit the TRAC or TRBC motifs have achieved KO of endogenous TCRαβ, elevated tgTCR expression, improved recognition of MM tumor targets, and effectively prolonged treatment duration, making it an enhanced and highly effective T cell therapy.³²³ NK cells are cytotoxic, and an increase in the number of NK cells improves patient survival in MM therapy, but NK cell function is impaired due to up-regulation of inhibitory receptors. The study used CRISPR–Cas9 to knock down the killer cell lectin-like receptor C1 (KLRC1) motif in the immune checkpoint NKG2A in primary NK cells. The gene-edited NK cells showed significantly increased toxicity to some MM cells and could still produce NK cell-specific cytokines, the disruption of immune checkpoints in primary NK cells has the potential to be overcome in clinical treatment immune checkpoint suppression, overcome tumor immune evasion and improve therapeutic effects on MM.³²⁴

4.3.3 Lymphomas

Lymphomas are divided into three main groups: B-cell lymphomas, NK/T-cell lymphomas (NKTL), and Hodgkin's lymphomas.³²⁵ Diffuse large B-cell lymphoma (LBCL) accounts for approximately 30% of non-Hodgkin's lymphomas, and patients have symptoms such as progressive lymph node enlargement; more than half of the patients can be cured by R-CHOP immunochemotherapeutic agents such as rituximab and adriamycin, but the prognosis of patients who are not cured is poor, and new therapeutic strategies need to be developed.³²⁶ NKTL is an Epstein–Barr virus (EBV)-associated non-Hodgkin's lymphoma, lymphoma cells are transformed from NK cells or T cells, EBV infection and cytogenetic alterations are the main factors in the development of NKTL, and most patients are treated with radiation therapy, a few patients with extensive disease are treated with chemotherapy, and immunotherapy and hematopoietic cell transplantation treatments are still in the early stages of exploration.³²⁷ Classical Hodgkin's lymphoma is a highly curable cancer and patients are usually treated with a combination of first-line chemotherapy and radiotherapy, but some patients are not cured and die due to the toxic effects of the treatment.³²⁸ CRISPR/Cas9-based genome editing offers clinical therapeutic potential for immunotherapies such as TCR, CAR T cells, and over-the-counter cell transfer (ACT).³²⁹ CAR T-cell therapies are primarily designed to target diseases individually, and one study used ABE base editing to install a function-preserving mutation that protects CAR T cells and healthy hematopoietic cells from the toxic effects of CD45 targeting, to develop universal CAR T-cell therapies targeting panleukopoietic CD45 markers for AML, B-cell lymphoma and acute T-cell leukemia.³³⁰ Another study used CRISPR–Cas9 to generate gene-specific targeted nonviral CAR-T cells and inserted an anti-CD19 CAR cassette into the AAVS1 safety motif to develop innovative anti-CD19 CAR-T cells integrated with PD1. This two-in-one approach was therapeutic for patients with relapsed/refractory aggressive B-cell non-Hodgkin's lymphoma, with a remission rate of 87.5% in eight patients, confirming the safety and efficacy of nonviral, gene-specific integration of CAR-T cells in oncology therapy.³³¹ Super enhancers (SEs) could drive key oncogenes in malignant tumors. The study analyzed primary tumor samples of SEs NKTL and found that TOX2 was aberrantly overexpressed in NKTL. Subsequently, the regulation of SE oncogenes by TFs was explored using techniques such as CRISPR–dCas9, assessed the effect of TOX2 on the degree of malignancy of NKTL in vitro and in vivo, and demonstrated that targeting TOX2 may be an important therapeutic intervention for NKTL patients.³³² Brentuximab vedotin (BV), a drug-coupled anti-CD30 antibody, is an effective agent in refractory/recurrent Hodgkin's lymphoma, but many patients end up developing resistance to BV, a study established a CRISPR library screening platform for BV-sensitive gene identification in Hodgkin lymphoma for this purpose, and the results showed that ubiquitin editing enzyme A20 was negatively correlated with the activity of NF-κB, and that NF-κB could affect the sensitivity to BV treatment by regulating ABCB1 expression; therefore, targeting and regulating the activity of NF-κB could synergistically modulate the sensitivity of patients to BV.³³³ The above studies suggest that CRISPR gene editing technology plays an important role in both gene therapy of diseases and the exploration of drug resistance mechanisms.

6.1.1 Three types of commonly used off-target prediction tools

Three primary techniques exist for genome-wide detection of potential off-target sites induced by nucleases: sequence-homology-based bioinformatics prediction, cell-capture-based off-target editing events, and using genomic DNA templates for cleavage site characterization in vitro.⁴¹³ These approaches are used for predicting or identifying off-target effects in gene editing, detecting gene editing outcomes, and assessing the safety of gene editing (Table 5).⁴¹¹ The commonly used computer prediction techniques mainly include E-CRISP,⁴¹⁴ COSMID,⁴¹⁵ Cas-OFFinder,⁴¹⁶ CRISPRseek,⁴¹⁷ CCTOP (Consensus Constrained TOPology prediction),⁴¹⁸ CROP-IT (a web-based CRISPR/Cas9 Off-target Prediction and Identification Tool),⁴¹⁹ CHOPCHOP,⁴²⁰ CHOPCHOP v2,⁴²¹ CHOPCHOP v3,⁴²² Crisflash,⁴²³ CRISPOR,⁴²⁴ and CRISTA.⁴²⁵ Computational prediction algorithms are powerful and convenient but often miss most confirmed off-target sites. They are also ineffective in ranking active off-target sites within cells.⁴¹³

Cell-based and in vitro methods for identifying potential off-target sites at a genome-wide scale are used to circumvent the limitations of computational prediction tools.⁴¹¹ Cell-based methods directly observe genome editing activity within cells. One of the primary cell-based methods for off-target detection is WGS (whole-genome sequencing),⁴²⁶ HTGTS (high-throughput whole-genome translocation sequencing),⁴²⁷ BLISS (breaks labeling in situ and sequencing),⁴²⁸ GUIDE-seq,⁴²⁹ iGUIDE-seq (improved GUIDE-seq),⁴³⁰ DISCOVER-seq (discovery of in situ cas off-targets and verification by sequencing),^{431, 432} DISCOVER-Seq+,⁴³³ CAST-Seq (chromosomal aberration analysis by single targeted linker-mediated PCR sequencing),⁴³⁴ and GOTI (genome-wide off-target analysis by two-cell embryo injection).⁴³⁵

While cell-based approaches are vulnerable to limitations in the efficiency and safety of nuclease or DNA delivery, in vitro off-target detection techniques directly purify the genome and perform CRISPR in vitro cutting experiments to capture the sites where off-target cutting occurs.⁴³⁶ In vitro, off-target detection methods include Digenome-seq (digested genome sequencing),^437-439 Digenome-seq using cell-free chromatin DNA (DIG-seq),⁴⁴⁰ SITE-seq (Selective enrichment and Identification of Tagged genomic DNA Ends by Sequencing),⁴⁴¹ CIRCLE-seq,^{436, 442} CHANGE-seq (Circularization for High-throughput Analysis of Nuclease Genome-wide Effects by sequencing),^{443, 444} NucleaSeq (nuclease digestion and deep sequencing),⁴⁴⁵ SURRO-seq,⁴⁴⁶ and Extru-seq.⁴⁴⁷ Extru-seq is a novel genome-wide off-target prediction method that combines the advantages of both cell-based and in vitro approaches, showing high validation rates and retention of information about the intracellular environment, low leakage rates, which would be easily performed in clinically relevant cell types.⁴⁴⁷

6.1.2 Off-target assessment tool for BEs

In addition to CRISPR–Cas9-based gene editing technology that requires off-target detection, the application of BE-based gene editing technology also requires the assessment and determination of genome-wide off-target effects. BEs can precisely edit or modulate the genome without triggering random mutations, and the off-target effects generated mainly originate from base-edited deaminases or lead-edited RTs, and the off-target sites are usually different from those of Cas9 alone, which requires an independent assessment of their genome-wide specificity.⁴⁴⁸ EndoV-seq, the first assay to assess the genome-wide off-target effect of ABE, used endonuclease V to cleave in vitro the inosine-containing DNA strands of genomic DNA deaminated by ABE, and WGS of the processed DNA to identify off-target-sites, demonstrating that the off-target effect of ABE is much less than that of the classical Cas9 nuclease.⁴⁴⁹ A Modification of Digenome-seq for assessment of APOBEC1-nCas9 BE (BE3) genome-wide off-target activity has been investigated.⁴⁴⁸ It has been shown that a modified version of Digenome-seq has been optimized to detect ABE off-target sites and to assess genome-wide target specificity for ABE.⁴⁵⁰ Another study developed nickase-based Digenome-seq (nDigenome-seq) to probe the genomic specificity of the PE, which can be further improved in human cells by integrating mutations from engineered Cas9 variants (eSpCas9 and Sniper Cas9) into PEs.¹²⁵ Prime editor assisted off-target characterization by sequencing (PEAC-seq) is a strategy for comprehensively identifying CRISPR off-target sites and DNA translocation events in vitro and in vivo. PEAC-seq uses PE to insert sequence-optimized tags into editing sites and enriches the tagged regions with site-specific primers for high-throughput sequencing. Translocations have greater genotoxicity but are often overlooked by other off-target detection methods, the development of PEAC-seq fills this gap.⁴⁵¹ TAgmentation of Prime Editor sequencing (TAPE-seq) is a cell-based method that accurately and sensitively predicts genome-wide off-target effects of PE.⁴⁵² TAPE-seq has a lower leakage rate than other methods and identifies valid off-target sites that are missed by other methods.⁴⁵²

Despite advances in off-target assays, the development of credible, high-sensitive, unbiased, and accurate ways to detect the effects of gene editing off-target continues to pose a challenge for gene editing clinical therapies, and optimization of sgRNAs and nuclease proteins is still needed to enhance the safety of gene editing tools in clinical treatment applications.

6.2.1 Virus vector delivery systems

Viral vectors efficiently deliver CRISPR–Cas9 to host cells, transferring nucleic acids based on plasmids into mammalian cells for long-term expression in vitro and in vivo. AdVs have extensive tissue tropism and do not usually integrate into the host genome, minimizing the off-target mutagenesis risk of the CRISPR–Cas9 system.⁴⁶⁷ AAVs are nonimmunogenic and specific to multiple serotypes and are one of the most widely used in vivo CRISPR–Cas delivery vectors.^{468, 476} Small Cas9 homologs such as SaCas9 can improve the delivery efficiency of AAV vectors.^{32, 469, 477} LVs have mild immunogenicity and deliver genes that can be expressed for a long time.^{470, 478} The integrase-defective LV (IDLV) overcomes the risk of insertion mutagenesis and genotoxicity problems and is a simple and feasible approach for in vivo gene therapy.⁴⁷⁹ Virus-like particles are an emerging targeted delivery vehicle expressed instantaneously in vivo with little risk of genome integration.⁴⁸⁰

6.2.2 Nonvirus delivery system

Nonviral chemical delivery systems are a promising approach to effectively avoid the potential immunogenic risks of viruses.^{471, 481} LNPs are a crucial tool in gene editing, pharmaceuticals, and vaccination. LNPs have been used in vivo in clinical studies for COVID-19 mRNA vaccination and have great potential to realize gene therapy.^{482, 483} AuNPs and CRISPR-Gold are delivery vectors for Cas9 RNP and donor DNA in vivo.^{472, 484} Microinjection and electroporation are the two main physical delivery methods. Microinjection is highly specific and reproducible.⁴⁷⁴ It is usually applied in vitro, where gene editing components of various molecular weights, such as RNPs, are directly delivered to the target position in living cells under a microscope.⁴⁶⁰ It is not restricted by various cell components and can control the number of delivery goods, effectively improving the occurrence of off-target effects. However, it is unsuitable for large-scale gene editing therapy.⁴⁶⁰ Electroporation is a widely used method for delivering proteins and nucleic acids into mammalian cells.⁴⁷⁵ It is mainly used for in vitro gene editing and is suitable for various cell types. It can effectively deliver CRISPR cargo such as DNAs, RNAs, or RNPs into target cells but is limited by cytotoxicity induced by high voltage shock.⁴⁵⁸

The challenges for successful in vivo gene editing therapies involve achieving maximal delivery to target cells, minimizing transfer to nontarget cells, reducing immunogenicity, and achieving stable expression. The development of new methods for delivering CRISPR–Cas9 systems safely and efficiently remains an active research field in gene editing, and further optimization of in vivo delivery methods for gene editing systems is necessary to improve the targeting and safety of delivery systems.