Determinants of Base Editing Outcomes from Target Library Analysis and Machine Learning

DNA base editing corrects common hemophilia A mutations and restores factor VIII expression in in vitro and ex vivo models

BACKGROUND

Replacement and nonreplacement therapies effectively control bleeding in hemophilia A (HA) but imply lifelong interventions. Authorized gene addition therapy could provide a cure but still poses questions on durability. FVIIIgene correction would definitively restore factor (F)VIII production, as shown in animal models through nuclease-mediated homologous recombination (HR). However, low efficiency and potential off-target double-strand break still limit HR translatability.

OBJECTIVES

To correct common model single point mutations leading to severe HA through the recently developed double-strand break/HR-independent base editing (BE) and prime editing (PE) approaches.

METHODS

Screening for efficacy of BE/PE systems in HEK293T cells transiently expressing FVIII variants and validation at DNA (sequencing) and protein (enzyme-linked immunosorbent assay; activated partial thromboplastin time) level in stable clones. Evaluation of rescue in engineered blood outgrowth endothelial cells by lentiviral-mediated delivery of BE.

RESULTS

Transient assays identified the best-performing BE/PE systems for each variant, with the highest rescue of FVIII expression (up to 25% of wild-type recombinant FVIII) for the p.R2166∗ and p.R2228Q mutations. In stable clones, we demonstrated that the mutation reversion on DNA (∼24%) was consistent with the rescue of FVIII secretion and activity of 20% to 30%. The lentiviral-mediated delivery of the selected BE systems was attempted in engineered blood outgrowth endothelial cells harboring the p.R2166∗ and p.R2228Q variants, which led to an appreciable and dose-dependent rescue of secreted functional FVIII.

CONCLUSION

Overall data provide the first proof-of-concept for effective BE/PE-mediated correction of HA-causing mutations, which encourage studies in mouse models to develop a personalized cure for large cohorts of patients through a single intervention.

CRISPR Base Editing to Create Potential Charcot-Marie-Tooth Disease Models with High Editing Efficiency: Human Induced Pluripotent Stem Cell Harboring SH3TC2 Variants

Human induced pluripotent stem cells (hiPSCs) represent a powerful tool to investigate neuropathological disorders in which the cells of interest are inaccessible, such as in the Charcot-Marie-Tooth disease (CMT), the most common inherited peripheral neuropathy. Developing appropriate cellular models becomes crucial in order to both study the disease's pathophysiology and test new therapeutic approaches. The generation of hiPS cellular models for disorders caused by a single nucleotide variation has been significantly improved following the development of CRISPR-based editing tools. In this study, we efficiently and quickly generated, by CRISPR editing, the two first hiPSCs cellular models carrying alterations involved in CMT4C, also called AR-CMTde-SH3TC2. This subtype of CMT is associated with alterations in the SH3TC2 gene and represents the most prevalent form of autosomal recessive demyelinating CMT. We aimed to develop models for two different SH3TC2 nonsense variants, c.211C>T, p.Gln71* and the most common AR-CMTde-SH3TC2 alteration, c.2860C>T, p.Arg954*. First, in order to determine the best CRISPR strategy to adopt on hiPSCs, we first tested a variety of sgRNAs combined with a selection of recent base editors using the conveniently cultivable and transfectable HEK-293T cell line. The chosen CRISPR base-editing strategy was then applied to hiPSCs derived from healthy individuals to generate isogenic CMT disease models with up to 93% editing efficiency. For point mutation generation, we first recommend to test your strategies on alternative cell line such as HEK-293T before hiPSCs to evaluate a variety of sgRNA-BE combinations, thus boosting the chance of achieving edited cellular clones with the hard-to-culture and to transfect hiPSCs.

Taming AID mutator activity in somatic hypermutation

Activation-induced cytidine deaminase (AID) initiates somatic hypermutation (SHM) by introducing base substitutions into antibody genes, a process enabling antibody affinity maturation in immune response. How a mutator is tamed to precisely and safely generate programmed DNA lesions in a physiological process remains unsettled, as its dysregulation drives lymphomagenesis. Recent research has revealed several hidden features of AID-initiated mutagenesis: preferential activity on flexible DNA substrates, restrained activity within chromatin loop domains, unique DNA repair factors to differentially decode AID-caused lesions, and diverse consequences of aberrant deamination. Here, we depict the multifaceted regulation of AID activity with a focus on emerging concepts/factors and discuss their implications for the design of base editors (BEs) that install somatic mutations to correct deleterious genomic variants.

Accelerated drug-resistant variant discovery with an enhanced, scalable mutagenic base editor platform

Personalized cancer therapeutics bring directed treatment options to patients based on their tumor's genetic signature. Unfortunately, tumor genomes are remarkably adaptable, and acquired resistance through gene mutation frequently occurs. Identifying mutations that promote resistance within drug-treated patient populations can be cost, resource, and time intensive. Accordingly, base editing, enabled by Cas9-deaminase domain fusions, has emerged as a promising approach for rapid, large-scale gene variant screening in situ. Here, we adapt and optimize a conditional activation-induced cytidine deaminase (AID)-dead Cas9 (dCas9) system, which demonstrates greater heterogeneity of edits with an expanded footprint compared to the most commonly utilized cytosine base editor, BE4. In combination with a custom single guide RNA (sgRNA) library, we identify individual and compound variants in epidermal growth factor receptor (EGFR) and v-raf murine sarcoma viral oncogene homolog B1 (BRAF) that confer resistance to established EGFR inhibitors. This system and analytical pipeline provide a simple, highly scalable platform for cis or trans drug-modifying variant discovery and for uncovering valuable insights into protein structure-function relationships.

Machine learning prediction of prime editing efficiency across diverse chromatin contexts

The success of prime editing depends on the prime editing guide RNA (pegRNA) design and target locus. Here, we developed machine learning models that reliably predict prime editing efficiency. PRIDICT2.0 assesses the performance of pegRNAs for all edit types up to 15 bp in length in mismatch repair-deficient and mismatch repair-proficient cell lines and in vivo in primary cells. With ePRIDICT, we further developed a model that quantifies how local chromatin environments impact prime editing rates.

nCas9 Engineering for Improved Target Interaction Presents an Effective Strategy to Enhance Base Editing

Base editors (BEs) are a recent generation of genome editing tools that couple a cytidine or adenosine deaminase activity to a catalytically impaired Cas9 moiety (nCas9) to enable specific base conversions at the targeted genomic loci. Given their strong application potential, BEs are under active developments toward greater levels of efficiency and safety. Here, a previously overlooked nCas9-centric strategy is explored for enhancement of BE. Based on a cytosine BE (CBE), 20 point mutations associated with nCas9-target interaction are tested. Subsequently, from the initial positive X-to-arginine hits, combinatorial modifications are applied to establish further enhanced CBE variants (1.1-1.3). Parallel nCas9 modifications in other versions of CBEs including A3A-Y130F-BE4max, YEE-BE4max, CGBE, and split-AncBE4max, as well as in the context of two adenine BEs (ABE), likewise enhance their respective activities. The same strategy also substantially improves the efficiencies of high-fidelity nCas9/BEs. Further evidence confirms that the stabilization of nCas9-substrate interactions underlies the enhanced BE activities. In support of their translational potential, the engineered CBE and ABE variants respectively enable 82% and 25% higher rates of editing than the controls in primary human T-cells. This study thus demonstrates a highly adaptable strategy for enhancing BE, and for optimizing other forms of Cas9-derived tools.

Structure-guided discovery of highly efficient cytidine deaminases with sequence-context independence

The applicability of cytosine base editors is hindered by their dependence on sequence context and by off-target effects. Here, by using AlphaFold2 to predict the three-dimensional structure of 1,483 cytidine deaminases and by experimentally characterizing representative deaminases (selected from each structural cluster after categorizing them via partitional clustering), we report the discovery of a few deaminases with high editing efficiencies, diverse editing windows and increased ratios of on-target to off-target effects. Specifically, several deaminases induced C-to-T conversions with comparable efficiency at AC/TC/CC/GC sites, the deaminases could introduce stop codons in single-copy and multi-copy genes in mammalian cells without double-strand breaks, and some residue conversions at predicted DNA-interacting sites reduced off-target effects. Structure-based generative machine learning could be further leveraged to expand the applicability of base editors in gene therapies.

The Y49H cytochrome c variant enhances megakaryocytic maturation of K-562 cells

Five pathogenic variants in the gene encoding cytochrome c (CYCS) associated with mild autosomal dominant thrombocytopenia have been reported. Previous studies of peripheral blood CD34+ or CD45+ cells from subjects with the G42S CYCS variant showed an acceleration in megakaryopoiesis compared to wild-type (WT) cells. To determine whether this result reflects a common feature of the CYCS variants, the c.145T>C mutation (Y49H variant) was introduced into the endogenous CYCS locus in K-562 cells, which undergo megakaryocytic maturation in response to treatment with a phorbol ester. The c.145T>C (Y49H) variant enhanced the megakaryocyte maturation of the K-562 cells, and this effect was seen when the cells were cultured at both 18 % and 5 % oxygen. Thus, alteration of megakaryopoiesis is common to both the G42S and Y49H CYCS variants and may contribute to the low platelet phenotype. The Y49H CYCS variant has previously been reported to impair mitochondrial respiratory chain function in vitro, however using extracellular flux analysis the c.145T>C (Y49H) variant does not alter mitochondrial bioenergetics of the K-562 cells, consistent with the lack of a phenotype characteristic of mitochondrial diseases in CYCS variant families. The Y49H variant has also been reported to enhance the ability of cytochrome c to trigger caspase activation in the intrinsic apoptosis pathway. However, as seen in peripheral blood cells from G42S CYCS variant carriers, the presence of Y49H cytochrome c in K-562 cells did not significantly change their response to an apoptotic stimulus.

Development of multiplexed orthogonal base editor (MOBE) systems

Base editors (BEs) enable efficient, programmable installation of point mutations while avoiding the use of double-strand breaks. Simultaneous application of two or more different BEs, such as an adenine BE (which converts A·T base pairs to G·C) and a cytosine BE (which converts C·G base pairs to T·A), is not feasible because guide RNA crosstalk results in non-orthogonal editing, with all BEs modifying all target loci. Here we engineer both adenine BEs and cytosine BEs that can be orthogonally multiplexed by using RNA aptamer-coat protein systems to recruit the DNA-modifying enzymes directly to the guide RNAs. We generate four multiplexed orthogonal BE systems that enable rates of precise co-occurring edits of up to 7.1% in the same DNA strand without enrichment or selection strategies. The addition of a fluorescent enrichment strategy increases co-occurring edit rates up to 24.8% in human cells. These systems are compatible with expanded protospacer adjacent motif and high-fidelity Cas9 variants, function well in multiple cell types, have equivalent or reduced off-target propensities compared with their parental systems and can model disease-relevant point mutation combinations.

Susceptibility to innate immune activation in genetically mediated myocarditis

Myocarditis is clinically characterized by chest pain, arrhythmias, and heart failure, and treatment is often supportive. Mutations in DSP, a gene encoding the desmosomal protein desmoplakin, have been increasingly implicated in myocarditis. To model DSP-associated myocarditis and assess the role of innate immunity, we generated engineered heart tissues (EHTs) using human induced pluripotent stem cell-derived cardiomyocytes (hiPSC-CMs) from patients with heterozygous DSP truncating variants (DSPtvs) and a gene-edited homozygous deletion cell line (DSP-/-). At baseline, DSP-/- EHTs displayed a transcriptomic signature of innate immune activation, which was mirrored by cytokine release. Importantly, DSP-/- EHTs were hypersensitive to Toll-like receptor (TLR) stimulation, demonstrating more contractile dysfunction compared with isogenic controls. Relative to DSP-/- EHTs, heterozygous DSPtv EHTs had less functional impairment. DSPtv EHTs displayed heightened sensitivity to TLR stimulation, and when subjected to strain, DSPtv EHTs developed functional deficits, indicating reduced contractile reserve compared with healthy controls. Colchicine or NF-κB inhibitors improved strain-induced force deficits in DSPtv EHTs. Genomic correction of DSP p.R1951X using adenine base editing reduced inflammatory biomarker release from EHTs. Thus, EHTs replicate electrical and contractile phenotypes seen in human myocarditis, implicating cytokine release as a key part of the myogenic susceptibility to inflammation. The heightened innate immune activation and sensitivity are targets for clinical intervention.

Base editor-mediated large-scale screening of functional mutations in bacteria for industrial phenotypes

Base editing, the targeted introduction of point mutations into cellular DNA, holds promise for improving genome-scale functional genome screening to single-nucleotide resolution. Current efforts in prokaryotes, however, remain confined to loss-of-function screens using the premature stop codons-mediated gene inactivation library, which falls far short of fully releasing the potential of base editors. Here, we developed a base editor-mediated functional single nucleotide variant screening pipeline in Escherichia coli. We constructed a library with 31,123 sgRNAs targeting 462 stress response-related genes in E. coli, and screened for adaptive mutations under isobutanol and furfural selective conditions. Guided by the screening results, we successfully identified several known and novel functional mutations. Our pipeline might be expanded to the optimization of other phenotypes or the strain engineering in other microorganisms.

Joint genotypic and phenotypic outcome modeling improves base editing variant effect quantification

CRISPR base editing screens enable analysis of disease-associated variants at scale; however, variable efficiency and precision confounds the assessment of variant-induced phenotypes. Here, we provide an integrated experimental and computational pipeline that improves estimation of variant effects in base editing screens. We use a reporter construct to measure guide RNA (gRNA) editing outcomes alongside their phenotypic consequences and introduce base editor screen analysis with activity normalization (BEAN), a Bayesian network that uses per-guide editing outcomes provided by the reporter and target site chromatin accessibility to estimate variant impacts. BEAN outperforms existing tools in variant effect quantification. We use BEAN to pinpoint common regulatory variants that alter low-density lipoprotein (LDL) uptake, implicating previously unreported genes. Additionally, through saturation base editing of LDLR, we accurately quantify missense variant pathogenicity that is consistent with measurements in UK Biobank patients and identify underlying structural mechanisms. This work provides a widely applicable approach to improve the power of base editing screens for disease-associated variant characterization.

Revolutionising healing: Gene Editing's breakthrough against sickle cell disease

Recent advancements in gene editing illuminate new potential therapeutic approaches for Sickle Cell Disease (SCD), a debilitating monogenic disorder caused by a point mutation in the β-globin gene. Despite the availability of several FDA-approved medications for symptomatic relief, allogeneic hematopoietic stem cell transplantation (HSCT) remains the sole curative option, underscoring a persistent need for novel treatments. This review delves into the growing field of gene editing, particularly the extensive research focused on curing haemoglobinopathies like SCD. We examine the use of techniques such as CRISPR-Cas9 and homology-directed repair, base editing, and prime editing to either correct the pathogenic variant into a non-pathogenic or wild-type one or augment fetal haemoglobin (HbF) production. The article elucidates ways to optimize these tools for efficacious gene editing with minimal off-target effects and offers insights into their effective delivery into cells. Furthermore, we explore clinical trials involving alternative SCD treatment strategies, such as LentiGlobin therapy and autologous HSCT, distilling the current findings. This review consolidates vital information for the clinical translation of gene editing for SCD, providing strategic insights for investigators eager to further the development of gene editing for SCD.

Expanding the flexibility of base editing for high-throughput genetic screens in bacteria

Genome-wide screens have become powerful tools for elucidating genotype-to-phenotype relationships in bacteria. Of the varying techniques to achieve knockout and knockdown, CRISPR base editors are emerging as promising options. However, the limited number of available, efficient target sites hampers their use for high-throughput screening. Here, we make multiple advances to enable flexible base editing as part of high-throughput genetic screening in bacteria. We first co-opt the Streptococcus canis Cas9 that exhibits more flexible protospacer-adjacent motif recognition than the traditional Streptococcus pyogenes Cas9. We then expand beyond introducing premature stop codons by mutating start codons. Next, we derive guide design rules by applying machine learning to an essentiality screen conducted in Escherichia coli. Finally, we rescue poorly edited sites by combining base editing with Cas9-induced cleavage of unedited cells, thereby enriching for intended edits. The efficiency of this dual system was validated through a conditional essentiality screen based on growth in minimal media. Overall, expanding the scope of genome-wide knockout screens with base editors could further facilitate the investigation of new gene functions and interactions in bacteria.

Deconstructing cancer with precision genome editing

Recent advances in genome editing technologies are allowing investigators to engineer and study cancer-associated mutations in their endogenous genetic contexts with high precision and efficiency. Of these, base editing and prime editing are quickly becoming gold-standards in the field due to their versatility and scalability. Here, we review the merits and limitations of these precision genome editing technologies, their application to modern cancer research, and speculate how these could be integrated to address future directions in the field.

Adenine transversion editors enable precise, efficient A•T-to-C•G base editing in mammalian cells and embryos

Base editors have substantial promise in basic research and as therapeutic agents for the correction of pathogenic mutations. The development of adenine transversion editors has posed a particular challenge. Here we report a class of base editors that enable efficient adenine transversion, including precise A•T-to-C•G editing. We found that a fusion of mouse alkyladenine DNA glycosylase (mAAG) with nickase Cas9 and deaminase TadA-8e catalyzed adenosine transversion in specific sequence contexts. Laboratory evolution of mAAG significantly increased A-to-C/T conversion efficiency up to 73% and expanded the targeting scope. Further engineering yielded adenine-to-cytosine base editors (ACBEs), including a high-accuracy ACBE-Q variant, that precisely install A-to-C transversions with minimal Cas9-independent off-targeting effects. ACBEs mediated high-efficiency installation or correction of five pathogenic mutations in mouse embryos and human cell lines. Founder mice showed 44-56% average A-to-C edits and allelic frequencies of up to 100%. Adenosine transversion editors substantially expand the capabilities and possible applications of base editing technology.

A benchmarked, high-efficiency prime editing platform for multiplexed dropout screening

Prime editing installs precise edits into the genome with minimal unwanted byproducts, but low and variable editing efficiencies have complicated application of the approach to high-throughput functional genomics. Leveraging several recent advances, we assembled a prime editing platform capable of high-efficiency substitution editing across a set of engineered prime editing guide RNAs (epegRNAs) and corresponding target sequences (80% median intended editing). Then, using a custom library of 240,000 epegRNAs targeting >17,000 codons with 175 different substitution types, we benchmarked our platform for functional interrogation of small substitution variants (1-3 nucleotides) targeted to essential genes. Resulting data identified negative growth phenotypes for nonsense mutations targeted to ~8,000 codons, and comparing those phenotypes to results from controls demonstrated high specificity. We also observed phenotypes for synonymous mutations that disrupted splice site motifs at 3' exon boundaries. Altogether, we establish and benchmark a high-throughput prime editing approach for functional characterization of genetic variants with simple readouts from multiplexed experiments.

High-throughput evaluation of genetic variants with prime editing sensor libraries

Tumor genomes often harbor a complex spectrum of single nucleotide alterations and chromosomal rearrangements that can perturb protein function. Prime editing has been applied to install and evaluate genetic variants, but previous approaches have been limited by the variable efficiency of prime editing guide RNAs. Here we present a high-throughput prime editing sensor strategy that couples prime editing guide RNAs with synthetic versions of their cognate target sites to quantitatively assess the functional impact of endogenous genetic variants. We screen over 1,000 endogenous cancer-associated variants of TP53-the most frequently mutated gene in cancer-to identify alleles that impact p53 function in mechanistically diverse ways. We find that certain endogenous TP53 variants, particularly those in the p53 oligomerization domain, display opposite phenotypes in exogenous overexpression systems. Our results emphasize the physiological importance of gene dosage in shaping native protein stoichiometry and protein-protein interactions, and establish a framework for studying genetic variants in their endogenous sequence context at scale.

Deep learning models to predict the editing efficiencies and outcomes of diverse base editors

Applications of base editing are frequently restricted by the requirement for a protospacer adjacent motif (PAM), and selecting the optimal base editor (BE) and single-guide RNA pair (sgRNA) for a given target can be difficult. To select for BEs and sgRNAs without extensive experimental work, we systematically compared the editing windows, outcomes and preferred motifs for seven BEs, including two cytosine BEs, two adenine BEs and three C•G to G•C BEs at thousands of target sequences. We also evaluated nine Cas9 variants that recognize different PAM sequences and developed a deep learning model, DeepCas9variants, for predicting which variants function most efficiently at sites with a given target sequence. We then develop a computational model, DeepBE, that predicts editing efficiencies and outcomes of 63 BEs that were generated by incorporating nine Cas9 variants as nickase domains into the seven BE variants. The predicted median efficiencies of BEs with DeepBE-based design were 2.9- to 20-fold higher than those of rationally designed SpCas9-containing BEs.

Generation of precision preclinical cancer models using regulated in vivo base editing

Although single-nucleotide variants (SNVs) make up the majority of cancer-associated genetic changes and have been comprehensively catalogued, little is known about their impact on tumor initiation and progression. To enable the functional interrogation of cancer-associated SNVs, we developed a mouse system for temporal and regulatable in vivo base editing. The inducible base editing (iBE) mouse carries a single expression-optimized cytosine base editor transgene under the control of a tetracycline response element and enables robust, doxycycline-dependent expression across a broad range of tissues in vivo. Combined with plasmid-based or synthetic guide RNAs, iBE drives efficient engineering of individual or multiple SNVs in intestinal, lung and pancreatic organoids. Temporal regulation of base editor activity allows controlled sequential genome editing ex vivo and in vivo, and delivery of sgRNAs directly to target tissues facilitates generation of in situ preclinical cancer models.

Accurate top protein variant discovery via low-N pick-and-validate machine learning

A strategy to obtain the greatest number of best-performing variants with least amount of experimental effort over the vast combinatorial mutational landscape would have enormous utility in boosting resource producibility for protein engineering. Toward this goal, we present a simple and effective machine learning-based strategy that outperforms other state-of-the-art methods. Our strategy integrates zero-shot prediction and multi-round sampling to direct active learning via experimenting with only a few predicted top variants. We find that four rounds of low-N pick-and-validate sampling of 12 variants for machine learning yielded the best accuracy of up to 92.6% in selecting the true top 1% variants in combinatorial mutant libraries, whereas two rounds of 24 variants can also be used. We demonstrate our strategy in successfully discovering high-performance protein variants from diverse families including the CRISPR-based genome editors, supporting its generalizable application for solving protein engineering tasks. A record of this paper's transparent peer review process is included in the supplemental information.

Deep learning models incorporating endogenous factors beyond DNA sequences improve the prediction accuracy of base editing outcomes

Adenine base editors (ABEs) and cytosine base editors (CBEs) enable the single nucleotide editing of targeted DNA sites avoiding generation of double strand breaks, however, the genomic features that influence the outcomes of base editing in vivo still remain to be characterized. High-throughput datasets from lentiviral integrated libraries were used to investigate the sequence features affecting base editing outcomes, but the effects of endogenous factors beyond the DNA sequences are still largely unknown. Here the base editing outcomes of ABE and CBE were evaluated in mammalian cells for 5012 endogenous genomic sites and 11,868 genome-integrated target sequences, with 4654 genomic sites sharing the same target sequences. The comparative analyses revealed that the editing outcomes of ABE and CBE at endogenous sites were substantially different from those obtained using genome-integrated sequences. We found that the base editing efficiency at endogenous target sites of both ABE and CBE was influenced by endogenous factors, including epigenetic modifications and transcriptional activity. A deep-learning algorithm referred as BE_Endo, was developed based on the endogenous factors and sequence information from our genomic datasets, and it yielded unprecedented accuracy in predicting the base editing outcomes. These findings along with the developed computational algorithms may facilitate future application of BEs for scientific research and clinical gene therapy.

Domain-inlaid Nme2Cas9 adenine base editors with improved activity and targeting scope

Nme2Cas9 has been established as a genome editing platform with compact size, high accuracy, and broad targeting range, including single-AAV-deliverable adenine base editors. Here, we engineer Nme2Cas9 to further increase the activity and targeting scope of compact Nme2Cas9 base editors. We first use domain insertion to position the deaminase domain nearer the displaced DNA strand in the target-bound complex. These domain-inlaid Nme2Cas9 variants exhibit shifted editing windows and increased activity in comparison to the N-terminally fused Nme2-ABE. We next expand the editing scope by swapping the Nme2Cas9 PAM-interacting domain with that of SmuCas9, which we had previously defined as recognizing a single-cytidine PAM. We then use these enhancements to introduce therapeutically relevant edits in a variety of cell types. Finally, we validate domain-inlaid Nme2-ABEs for single-AAV delivery in vivo.

Direct delivery of stabilized Cas-embedded base editors achieves efficient and accurate editing of clinically relevant targets

Over the last 5 years, cytosine base editors (CBEs) have emerged as a promising therapeutic tool for specific editing of single nucleotide variants and disrupting specific genes associated with disease. Despite this promise, the currently available CBE's have the significant liabilities of off-target and bystander editing activities, in part due to the mechanism by which they are delivered, causing limitations in their potential applications. In this study we engineeredhighly stabilized Cas-embedded CBEs (sCE_CBEs) that integrate several recent advances, andthat are highly expressible and soluble for direct delivery into cells as ribonucleoprotein (RNP) complexes. Our resulting sCE_CBE RNP complexes efficiently and specifically target TC dinucleotides with minimal off-target or bystander mutations. Additional uracil glycosylase inhibitor (UGI) protein in trans further increased C-to-T editing efficiency and target purity in a dose-dependent manner, minimizing indel formation to untreated levels. A single electroporation was sufficient to effectively edit the therapeutically relevant locus for sickle cell disease in hematopoietic stem and progenitor cells (HSPC) in a dose dependent manner without cellular toxicity. Significantly, these sCE_CBE RNPs permitted for the transplantation of edited HSPCs confirming highly efficient editing in engrafting hematopoietic stem cells in mice. The success of the designed sCBE editors, with improved solubility and enhanced on-target editing, demonstrates promising agents for cytosine base editing at other disease-related sites in HSPCs and other cell types.

Optimization of base editors for the functional correction of SMN2 as a treatment for spinal muscular atrophy

Spinal muscular atrophy (SMA) is caused by mutations in SMN1. SMN2 is a paralogous gene with a C•G-to-T•A transition in exon 7, which causes this exon to be skipped in most SMN2 transcripts, and results in low levels of the protein survival motor neuron (SMN). Here we show, in fibroblasts derived from patients with SMA and in a mouse model of SMA that, irrespective of the mutations in SMN1, adenosine base editors can be optimized to target the SMN2 exon-7 mutation or nearby regulatory elements to restore the normal expression of SMN. After optimizing and testing more than 100 guide RNAs and base editors, and leveraging Cas9 variants with high editing fidelity that are tolerant of different protospacer-adjacent motifs, we achieved the reversion of the exon-7 mutation via an A•T-to-G•C edit in up to 99% of fibroblasts, with concomitant increases in the levels of the SMN2 exon-7 transcript and of SMN. Targeting the SMN2 exon-7 mutation via base editing or other CRISPR-based methods may provide long-lasting outcomes to patients with SMA.

C-to-G editing generates double-strand breaks causing deletion, transversion and translocation

Base editors (BEs) introduce base substitutions without double-strand DNA cleavage. Besides precise substitutions, BEs generate low-frequency 'stochastic' byproducts through unclear mechanisms. Here, we performed in-depth outcome profiling and genetic dissection, revealing that C-to-G BEs (CGBEs) generate substantial amounts of intermediate double-strand breaks (DSBs), which are at the centre of several byproducts. Imperfect DSB end-joining leads to small deletions via end-resection, templated insertions or aberrant transversions during end fill-in. Chromosomal translocations were detected between the editing target and off-targets of Cas9/deaminase origin. Genetic screenings of DNA repair factors disclosed a central role of abasic site processing in DSB formation. Shielding of abasic sites by the suicide enzyme HMCES reduced CGBE-initiated DSBs, providing an effective way to minimize DSB-triggered events without affecting substitutions. This work demonstrates that CGBEs can initiate deleterious intermediate DSBs and therefore require careful consideration for therapeutic applications, and that HMCES-aided CGBEs hold promise as safer tools.

Advancing genome editing with artificial intelligence: opportunities, challenges, and future directions

Clustered regularly interspaced short palindromic repeat (CRISPR)-based genome editing (GED) technologies have unlocked exciting possibilities for understanding genes and improving medical treatments. On the other hand, Artificial intelligence (AI) helps genome editing achieve more precision, efficiency, and affordability in tackling various diseases, like Sickle cell anemia or Thalassemia. AI models have been in use for designing guide RNAs (gRNAs) for CRISPR-Cas systems. Tools like DeepCRISPR, CRISTA, and DeepHF have the capability to predict optimal guide RNAs (gRNAs) for a specified target sequence. These predictions take into account multiple factors, including genomic context, Cas protein type, desired mutation type, on-target/off-target scores, potential off-target sites, and the potential impacts of genome editing on gene function and cell phenotype. These models aid in optimizing different genome editing technologies, such as base, prime, and epigenome editing, which are advanced techniques to introduce precise and programmable changes to DNA sequences without relying on the homology-directed repair pathway or donor DNA templates. Furthermore, AI, in collaboration with genome editing and precision medicine, enables personalized treatments based on genetic profiles. AI analyzes patients' genomic data to identify mutations, variations, and biomarkers associated with different diseases like Cancer, Diabetes, Alzheimer's, etc. However, several challenges persist, including high costs, off-target editing, suitable delivery methods for CRISPR cargoes, improving editing efficiency, and ensuring safety in clinical applications. This review explores AI's contribution to improving CRISPR-based genome editing technologies and addresses existing challenges. It also discusses potential areas for future research in AI-driven CRISPR-based genome editing technologies. The integration of AI and genome editing opens up new possibilities for genetics, biomedicine, and healthcare, with significant implications for human health.

An adenine base editor variant expands context compatibility

Adenine base editors (ABEs) are precise gene-editing agents that convert A:T pairs into G:C through a deoxyinosine intermediate. Existing ABEs function most effectively when the target A is in a TA context. Here we evolve the Escherichia coli transfer RNA-specific adenosine deaminase (TadA) to generate TadA8r, which extends potent deoxyadenosine deamination to RA (R = A or G) and is faster in processing GA than TadA8.20 and TadA8e, the two most active TadA variants reported so far. ABE8r, comprising TadA8r and a Streptococcus pyogenes Cas9 nickase, expands the editing window at the protospacer adjacent motif-distal end and outperforms ABE7.10, ABE8.20 and ABE8e in correcting disease-associated G:C-to-A:T transitions in the human genome, with a controlled off-target profile. We show ABE8r-mediated editing of clinically relevant sites that are poorly accessed by existing editors, including sites in PCSK9, whose disruption reduces low-density lipoprotein cholesterol, and ABCA4-p.Gly1961Glu, the most frequent mutation in Stargardt disease.

CRISPR/Cas9-gene editing approaches in plant breeding

CRISPR/Cas9 gene editing system is recently developed robust genome editing technology for accelerating plant breeding. Various modifications of this editing system have been established for adaptability in plant varieties as well as for its improved efficiency and portability. This review provides an in-depth look at the various strategies for synthesizing gRNAs for efficient delivery in plant cells, including chemical synthesis and in vitro transcription. It also covers traditional analytical tools and emerging developments in detection methods to analyze CRISPR/Cas9 mediated mutation in plant breeding. Additionally, the review outlines the various analytical tools which are used to detect and analyze CRISPR/Cas9 mediated mutations, such as next-generation sequencing, restriction enzyme analysis, and southern blotting. Finally, the review discusses emerging detection methods, including digital PCR and qPCR. Hence, CRISPR/Cas9 has great potential for transforming agriculture and opening avenues for new advancements in the system for gene editing in plants.

Deep learning and CRISPR-Cas13d ortholog discovery for optimized RNA targeting

Effective and precise mammalian transcriptome engineering technologies are needed to accelerate biological discovery and RNA therapeutics. Despite the promise of programmable CRISPR-Cas13 ribonucleases, their utility has been hampered by an incomplete understanding of guide RNA design rules and cellular toxicity resulting from off-target or collateral RNA cleavage. Here, we quantified the performance of over 127,000 RfxCas13d (CasRx) guide RNAs and systematically evaluated seven machine learning models to build a guide efficiency prediction algorithm orthogonally validated across multiple human cell types. Deep learning model interpretation revealed preferred sequence motifs and secondary features for highly efficient guides. We next identified and screened 46 novel Cas13d orthologs, finding that DjCas13d achieves low cellular toxicity and high specificity-even when targeting abundant transcripts in sensitive cell types, including stem cells and neurons. Our Cas13d guide efficiency model was successfully generalized to DjCas13d, illustrating the power of combining machine learning with ortholog discovery to advance RNA targeting in human cells.

Efficient gene editing in induced pluripotent stem cells enabled by an inducible adenine base editor with tunable expression

The preferred method for disease modeling using induced pluripotent stem cells (iPSCs) is to generate isogenic cell lines by correcting or introducing pathogenic mutations. Base editing enables the precise installation of point mutations at specific genomic locations without the need for deleterious double-strand breaks used in the CRISPR-Cas9 gene editing methods. We created a bulk population of iPSCs that homogeneously express ABE8e adenine base editor enzyme under a doxycycline-inducible expression system at the AAVS1 safe harbor locus. These cells enabled fast, efficient and inducible gene editing at targeted genomic regions, eliminating the need for single-cell cloning and screening to identify those with homozygous mutations. We could achieve multiplex genomic editing by creating homozygous mutations in very high efficiencies at four independent genomic loci simultaneously in AAVS1-iABE8e iPSCs, which is highly challenging with previously described methods. The inducible ABE8e expression system allows editing of the genes of interest within a specific time window, enabling temporal control of gene editing to study the cell or lineage-specific functions of genes and their molecular pathways. In summary, the inducible ABE8e system provides a fast, efficient and versatile gene-editing tool for disease modeling and functional genomic studies.

Systematic Exploration of Optimized Base Editing gRNA Design and Pleiotropic Effects with BExplorer

Base editing technology is being increasingly applied in genome engineering, but the current strategy for designing guide RNAs (gRNAs) relies substantially on empirical experience rather than a dependable and efficient in silico design. Furthermore, the pleiotropic effect of base editing on disease treatment remains unexplored, which prevents its further clinical usage. Here, we presented BExplorer, an integrated and comprehensive computational pipeline to optimize the design of gRNAs for 26 existing types of base editors in silico. Using BExplorer, we described its results for two types of mainstream base editors, BE3 and ABE7.10, and evaluated the pleiotropic effects of the corresponding base editing loci. BExplorer revealed 524 and 900 editable pathogenic single nucleotide polymorphism (SNP) loci in the human genome together with the selected optimized gRNAs for BE3 and ABE7.10, respectively. In addition, the impact of 707 edited pathogenic SNP loci following base editing on 131 diseases was systematically explored by revealing their pleiotropic effects, indicating that base editing should be carefully utilized given the potential pleiotropic effects. Collectively, the systematic exploration of optimized base editing gRNA design and the corresponding pleiotropic effects with BExplorer provides a computational basis for applying base editing in disease treatment.

Engineering of cytosine base editors with DNA damage minimization and editing scope diversification

Cytosine base editors (CBEs), which enable precise C-to-T substitutions, have been restricted by potential safety risks, including DNA off-target edits, RNA off-target edits and additional genotoxicity such as DNA damages induced by double-strand breaks (DSBs). Though DNA and RNA off-target edits have been ameliorated via various strategies, evaluation and minimization of DSB-associated DNA damage risks for most CBEs remain to be resolved. Here we demonstrate that YE1, an engineered CBE variant with minimized DNA and RNA off-target edits, could induce prominent DSB-associated DNA damage risks, manifested as γH2AX accumulation in human cells. We then perform deaminase engineering for two deaminases lamprey LjCDA1 and human APOBEC3A, and generate divergent CBE variants with eliminated DSB-associated DNA damage risks, in addition to minimized DNA/RNA off-target edits. Furthermore, the editing scopes and sequence preferences of APOBEC3A-derived CBEs could be further diversified by internal fusion strategy. Taken together, this study provides updated evaluation platform for DSB-associated DNA damage risks of CBEs and further generates a series of safer toolkits with diversified editing signatures to expand their applications.

Engineered lentivirus-derived nanoparticles (LVNPs) for delivery of CRISPR/Cas ribonucleoprotein complexes supporting base editing, prime editing and in vivo gene modification

Implementation of therapeutic in vivo gene editing using CRISPR/Cas relies on potent delivery of gene editing tools. Administration of ribonucleoprotein (RNP) complexes consisting of Cas protein and single guide RNA (sgRNA) offers short-lived editing activity and safety advantages over conventional viral and non-viral gene and RNA delivery approaches. By engineering lentivirus-derived nanoparticles (LVNPs) to facilitate RNP delivery, we demonstrate effective administration of SpCas9 as well as SpCas9-derived base and prime editors (BE/PE) leading to gene editing in recipient cells. Unique Gag/GagPol protein fusion strategies facilitate RNP packaging in LVNPs, and refinement of LVNP stoichiometry supports optimized LVNP yield and incorporation of therapeutic payload. We demonstrate near instantaneous target DNA cleavage and complete RNP turnover within 4 days. As a result, LVNPs provide high on-target DNA cleavage and lower levels of off-target cleavage activity compared to standard RNP nucleofection in cultured cells. LVNPs accommodate BE/sgRNA and PE/epegRNA RNPs leading to base editing with reduced bystander editing and prime editing without detectable indel formation. Notably, in the mouse eye, we provide the first proof-of-concept for LVNP-directed in vivo gene disruption. Our findings establish LVNPs as promising vehicles for delivery of RNPs facilitating donor-free base and prime editing without formation of double-stranded DNA breaks.

APOBEC Reporter Systems for Evaluating diNucleotide Editing Levels

Precision genome editing has become a reality with the discovery of base editors. Cytosine base editor (CBE) technologies are improving rapidly but are mostly optimized for TC dinucleotide targets. Here, we report the development and implementation of APOBEC Reporter Systems for Evaluating diNucleotide Editing Levels (ARSENEL) in living cells. The ARSENEL panel is comprised of four constructs that quantitatively report editing of each of the four dinucleotide motifs (AC/CC/GC/TC) through real-time accumulation of eGFP fluorescence. Editing rates of APOBEC3Bctd and AIDΔC CBEs reflect established mechanistic preferences with intrinsic biases to TC and GC, respectively. Twelve different (new and established) base editors are tested here using this system with a full-length APOBEC3B CBE showing the greatest on-target TC specificity and an APOBEC3A construct showing the highest editing efficiency. In addition, ARSENEL enables real-time assessment of natural and synthetic APOBEC inhibitors with the most potent to-date being the large subunit of the Epstein-Barr virus ribonucleotide reductase. These reporters have the potential to play important roles in research and development as precision genome engineering technologies progress toward achieving maximal specificity and efficiency.

Prediction of prime editing insertion efficiencies using sequence features and DNA repair determinants

Most short sequences can be precisely written into a selected genomic target using prime editing; however, it remains unclear what factors govern insertion. We design a library of 3,604 sequences of various lengths and measure the frequency of their insertion into four genomic sites in three human cell lines, using different prime editor systems in varying DNA repair contexts. We find that length, nucleotide composition and secondary structure of the insertion sequence all affect insertion rates. We also discover that the 3' flap nucleases TREX1 and TREX2 suppress the insertion of longer sequences. Combining the sequence and repair features into a machine learning model, we can predict relative frequency of insertions into a site with R = 0.70. Finally, we demonstrate how our accurate prediction and user-friendly software help choose codon variants of common fusion tags that insert at high efficiency, and provide a catalog of empirically determined insertion rates for over a hundred useful sequences.

Joint genotypic and phenotypic outcome modeling improves base editing variant effect quantification

CRISPR base editing screens are powerful tools for studying disease-associated variants at scale. However, the efficiency and precision of base editing perturbations vary, confounding the assessment of variant-induced phenotypic effects. Here, we provide an integrated pipeline that improves the estimation of variant impact in base editing screens. We perform high-throughput ABE8e-SpRY base editing screens with an integrated reporter construct to measure the editing efficiency and outcomes of each gRNA alongside their phenotypic consequences. We introduce BEAN, a Bayesian network that accounts for per-guide editing outcomes and target site chromatin accessibility to estimate variant impacts. We show this pipeline attains superior performance compared to existing tools in variant classification and effect size quantification. We use BEAN to pinpoint common variants that alter LDL uptake, implicating novel genes. Additionally, through saturation base editing of LDLR, we enable accurate quantitative prediction of the effects of missense variants on LDL-C levels, which aligns with measurements in UK Biobank individuals, and identify structural mechanisms underlying variant pathogenicity. This work provides a widely applicable approach to improve the power of base editor screens for disease-associated variant characterization.

Prediction of base editor off-targets by deep learning

Due to the tolerance of mismatches between gRNA and targeting sequence, base editors frequently induce unwanted Cas9-dependent off-target mutations. Here, to develop models to predict such off-targets, we design gRNA-off- target pairs for adenine base editors (ABEs) and cytosine base editors (CBEs) and stably integrate them into the human cells. After five days of editing, we obtain valid efficiency datasets of 54,663 and 55,727 off-targets for ABEs and CBEs, respectively. We use the datasets to train deep learning models, resulting in ABEdeepoff and CBEdeepoff, which can predict off-target sites. We use these tools to predict off-targets for a panel of endogenous loci and achieve Spearman correlation values varying from 0.710 to 0.859. Finally, we develop an integrated tool that is freely accessible via an online web server http://www.deephf.com/#/bedeep/bedeepoff . These tools could facilitate minimizing the off-target effects of base editing.

A temperature-tolerant CRISPR base editor mediates highly efficient and precise gene editing in Drosophila

CRISPR nucleases generate a broad spectrum of mutations that includes undesired editing outcomes. Here, we develop optimized C-to-T base editing systems for the generation of precise loss- or gain-of-function alleles in Drosophila and identify temperature as a crucial parameter for efficiency. We find that a variant of the widely used APOBEC1 deaminase has attenuated activity at 18° to 29°C and shows considerable dose-dependent toxicity. In contrast, the temperature-tolerant evoCDA1 domain mediates editing of typically more than 90% of alleles and is substantially better tolerated. Furthermore, formation of undesired mutations is exceptionally rare in Drosophila compared to other species. The predictable editing outcome, high efficiency, and product purity enables near homogeneous induction of STOP codons or alleles encoding protein variants in vivo. Last, we demonstrate how optimized expression enables conditional base editing in marked cell populations. This work substantially facilitates creation of precise alleles in Drosophila and provides key design parameters for developing efficient base editing systems in other ectothermic species.

CRISPR single base-editing: in silico predictions to variant clonal cell lines

Engineering single base edits using CRISPR technology including specific deaminases and single-guide RNA (sgRNA) is a rapidly evolving field. Different types of base edits can be constructed, with cytidine base editors (CBEs) facilitating transition of C-to-T variants, adenine base editors (ABEs) enabling transition of A-to-G variants, C-to-G transversion base editors (CGBEs) and recently adenine transversion editors (AYBE) that create A-to-C and A-to-T variants. The base-editing machine learning algorithm BE-Hive predicts which sgRNA and base editor combinations have the strongest likelihood of achieving desired base edits. We have used BE-Hive and TP53 mutation data from The Cancer Genome Atlas (TCGA) ovarian cancer cohort to predict which mutations can be engineered, or reverted to wild-type (WT) sequence, using CBEs, ABEs or CGBEs. We have developed and automated a ranking system to assist in selecting optimally designed sgRNA that considers the presence of a suitable protospacer adjacent motif (PAM), the frequency of predicted bystander edits, editing efficiency and target base change. We have generated single constructs containing ABE or CBE editing machinery, an sgRNA cloning backbone and an enhanced green fluorescent protein tag (EGFP), removing the need for co-transfection of multiple plasmids. We have tested our ranking system and new plasmid constructs to engineer the p53 mutants Y220C, R282W and R248Q into WT p53 cells and shown that these mutants cannot activate four p53 target genes, mimicking the behaviour of endogenous p53 mutations. This field will continue to rapidly progress, requiring new strategies such as we propose to ensure desired base-editing outcomes.

Predicting Mutations Generated by Cas9, Base Editing, and Prime Editing in Mammalian Cells

The first fruits of the CRISPR-Cas revolution are starting to enter the clinic, with gene editing therapies offering solutions to previously incurable genetic diseases. The success of such applications hinges on control over the mutations that are generated, which are known to vary depending on the targeted locus. In this review, we present the current state of understanding and predicting CRISPR-Cas cutting, base editing, and prime editing outcomes in mammalian cells. We first provide an introduction to the basics of DNA repair and machine learning that the models rely on. We then overview the datasets and methods created for characterizing edits at scale, as well as the insights that have been derived from them. The predictions generated from these models serve as a foundation for designing efficient experiments across the broad contexts where these tools are applied.

Base editor screens for in situ mutational scanning at scale

A fundamental challenge in biology is understanding the molecular details of protein function. How mutations alter protein activity, regulation, and response to drugs is of critical importance to human health. Recent years have seen the emergence of pooled base editor screens for in situ mutational scanning: the interrogation of protein sequence-function relationships by directly perturbing endogenous proteins in live cells. These studies have revealed the effects of disease-associated mutations, discovered novel drug resistance mechanisms, and generated biochemical insights into protein function. Here, we discuss how this "base editor scanning" approach has been applied to diverse biological questions, compare it with alternative techniques, and describe the emerging challenges that must be addressed to maximize its utility. Given its broad applicability toward profiling mutations across the proteome, base editor scanning promises to revolutionize the investigation of proteins in their native contexts.

Revolutionizing DNA repair research and cancer therapy with CRISPR-Cas screens

All organisms possess molecular mechanisms that govern DNA repair and associated DNA damage response (DDR) processes. Owing to their relevance to human disease, most notably cancer, these mechanisms have been studied extensively, yet new DNA repair and/or DDR factors and functional interactions between them are still being uncovered. The emergence of CRISPR technologies and CRISPR-based genetic screens has enabled genome-scale analyses of gene-gene and gene-drug interactions, thereby providing new insights into cellular processes in distinct DDR-deficiency genetic backgrounds and conditions. In this Review, we discuss the mechanistic basis of CRISPR-Cas genetic screening approaches and describe how they have contributed to our understanding of DNA repair and DDR pathways. We discuss how DNA repair pathways are regulated, and identify and characterize crosstalk between them. We also highlight the impacts of CRISPR-based studies in identifying novel strategies for cancer therapy, and in understanding, overcoming and even exploiting cancer-drug resistance, for example in the contexts of PARP inhibition, homologous recombination deficiencies and/or replication stress. Lastly, we present the DDR CRISPR screen (DDRcs) portal , in which we have collected and reanalysed data from CRISPR screen studies and provide a tool for systematically exploring them.

Enrichment strategies to enhance genome editing

Genome editing technologies hold great promise for numerous applications including the understanding of cellular and disease mechanisms and the development of gene and cellular therapies. Achieving high editing frequencies is critical to these research areas and to achieve the overall goal of being able to manipulate any target with any desired genetic outcome. However, gene editing technologies sometimes suffer from low editing efficiencies due to several challenges. This is often the case for emerging gene editing technologies, which require assistance for translation into broader applications. Enrichment strategies can support this goal by selecting gene edited cells from non-edited cells. In this review, we elucidate the different enrichment strategies, their many applications in non-clinical and clinical settings, and the remaining need for novel strategies to further improve genome research and gene and cellular therapy studies.

Base editing-mediated one-step inactivation of the Dnmt gene family reveals critical roles of DNA methylation during mouse gastrulation

During embryo development, DNA methylation is established by DNMT3A/3B and subsequently maintained by DNMT1. While much research has been done in this field, the functional significance of DNA methylation in embryogenesis remains unknown. Here, we establish a system of simultaneous inactivation of multiple endogenous genes in zygotes through screening for base editors that can efficiently introduce a stop codon. Embryos with mutations in Dnmts and/or Tets can be generated in one step with IMGZ. Dnmt-null embryos display gastrulation failure at E7.5. Interestingly, although DNA methylation is absent, gastrulation-related pathways are down-regulated in Dnmt-null embryos. Moreover, DNMT1, DNMT3A, and DNMT3B are critical for gastrulation, and their functions are independent of TET proteins. Hypermethylation can be sustained by either DNMT1 or DNMT3A/3B at some promoters, which are related to the suppression of miRNAs. The introduction of a single mutant allele of six miRNAs and paternal IG-DMR partially restores primitive streak elongation in Dnmt-null embryos. Thus, our results unveil an epigenetic correlation between promoter methylation and suppression of miRNA expression for gastrulation and demonstrate that IMGZ can accelerate deciphering the functions of multiple genes in vivo.

Massively parallel evaluation and computational prediction of the activities and specificities of 17 small Cas9s

Recently, various small Cas9 orthologs and variants have been reported for use in in vivo delivery applications. Although small Cas9s are particularly suited for this purpose, selecting the most optimal small Cas9 for use at a specific target sequence continues to be challenging. Here, to this end, we have systematically compared the activities of 17 small Cas9s for thousands of target sequences. For each small Cas9, we have characterized the protospacer adjacent motif and determined optimal single guide RNA expression formats and scaffold sequence. High-throughput comparative analyses revealed distinct high- and low-activity groups of small Cas9s. We also developed DeepSmallCas9, a set of computational models predicting the activities of the small Cas9s at matched and mismatched target sequences. Together, this analysis and these computational models provide a useful guide for researchers to select the most suitable small Cas9 for specific applications.

Parallel engineering and activity profiling of a base editor system

Selecting the most suitable existing base editors and engineering new variants for installing specific base conversions with maximal efficiency and minimal undesired edits are pivotal for precise genome editing applications. Here, we present a platform for creating and analyzing a library of engineered base editor variants to enable head-to-head evaluation of their editing performance at scale. Our comprehensive comparison provides quantitative measures on each variant's editing efficiency, purity, motif preference, and bias in generating single and multiple base conversions, while uncovering undesired higher indel generation rate and noncanonical base conversion for some of the existing base editors. In addition to engineering the base editor protein, we further applied this platform to investigate a hitherto underexplored engineering route and created guide RNA scaffold variants that augment the editor's base-editing activity. With the unknown performance and compatibility of the growing number of engineered parts including deaminase, CRISPR-Cas enzyme, and guide RNA scaffold variants for assembling the expanding collection of base editor systems, our platform addresses the unmet need for an unbiased, scalable method to benchmark their editing outcomes and accelerate the engineering of next-generation precise genome editors.

Evolution of an adenine base editor into a small, efficient cytosine base editor with low off-target activity

Cytosine base editors (CBEs) are larger and can suffer from higher off-target activity or lower on-target editing efficiency than current adenine base editors (ABEs). To develop a CBE that retains the small size, low off-target activity and high on-target activity of current ABEs, we evolved the highly active deoxyadenosine deaminase TadA-8e to perform cytidine deamination using phage-assisted continuous evolution. Evolved TadA cytidine deaminases contain mutations at DNA-binding residues that alter enzyme selectivity to strongly favor deoxycytidine over deoxyadenosine deamination. Compared to commonly used CBEs, TadA-derived cytosine base editors (TadCBEs) offer similar or higher on-target activity, smaller size and substantially lower Cas-independent DNA and RNA off-target editing activity. We also identified a TadA dual base editor (TadDE) that performs equally efficient cytosine and adenine base editing. TadCBEs support single or multiplexed base editing at therapeutically relevant genomic loci in primary human T cells and primary human hematopoietic stem and progenitor cells. TadCBEs expand the utility of CBEs for precision gene editing.

Re-engineering the adenine deaminase TadA-8e for efficient and specific CRISPR-based cytosine base editing

Cytosine base editors (CBEs) efficiently generate precise C·G-to-T·A base conversions, but the activation-induced cytidine deaminase/apolipoprotein B mRNA-editing enzyme catalytic polypeptide-like (AID/APOBEC) protein family deaminase component induces considerable off-target effects and indels. To explore unnatural cytosine deaminases, we repurpose the adenine deaminase TadA-8e for cytosine conversion. The introduction of an N46L variant in TadA-8e eliminates its adenine deaminase activity and results in a TadA-8e-derived C-to-G base editor (Td-CGBE) capable of highly efficient and precise C·G-to-G·C editing. Through fusion with uracil glycosylase inhibitors and further introduction of additional variants, a series of Td-CBEs was obtained either with a high activity similar to that of BE4max or with higher precision compared to other reported accurate CBEs. Td-CGBE/Td-CBEs show very low indel effects and a background level of Cas9-dependent or Cas9-independent DNA/RNA off-target editing. Moreover, Td-CGBE/Td-CBEs are more efficient in generating accurate edits in homopolymeric cytosine sites in cells or mouse embryos, suggesting their accuracy and safety for gene therapy and other applications.