How cytosine base editors work — and why the mechanism matters

Cytosine base editors convert cytosine (C) to thymine (T) — or, in newer transversion variants, to guanine (G) — by fusing a cytidine deaminase domain to a Cas9 nickase (nCas9) or catalytically dead Cas9 (dCas9), guided to a target locus by a single guide RNA (sgRNA). The deaminase converts C to uracil (U) within a defined activity window on the non-template strand of the displaced R-loop, while uracil DNA glycosylase inhibitor (UGI) domains suppress base-excision repair, biasing repair toward the T outcome. No donor DNA template is required and double-strand breaks are not generated, substantially reducing the oncogenic liability associated with classical CRISPR cutting.

The core sub-domains of CBE research identified in this dataset span five interconnected engineering layers. Deaminase engineering focuses on swapping or mutating the deaminase component — APOBEC1, APOBEC3B, APOBEC3G, PmCDA1, and novel variants — to reduce spurious off-target deamination of unintended cytosines in both DNA and RNA. Cas9 ortholog diversification replaces the standard SpCas9 with compact (NmeCas9, SaCas9) or PAM-flexible (Cas9-NG, SpRY) variants to broaden targetable genomic space and enable AAV packaging. UGI architecture optimization, demonstrated in Intellia Therapeutics filings, separates UGI-encoding mRNAs from the deaminase-nickase fusion to improve manufacturing modularity. Delivery system engineering encompasses split-intein AAV, lipid nanoparticle-delivered mRNA, and ribonucleoprotein (RNP) formats. Finally, transversion CBEs (CGBEs) incorporate uracil DNA glycosylase (UNG) into the CBE to redirect editing outcomes from C-to-T toward C-to-G.

This mechanistic architecture makes CBEs particularly attractive for treating point-mutation diseases: according to NIH databases, a significant proportion of catalogued pathogenic human variants are single-nucleotide substitutions amenable in principle to base editor correction. The BEable-GPS database (ShanghaiTech University, 2019) profiled 20 base editors against all catalogued pathogenic SNVs, providing a genomic accessibility map that directly informs which CBE variant can address which disease mutation.

From foundational patents to clinical-stage filings: the 2017–2026 innovation arc

The CBE intellectual foundation was established in 2017–2018, when Harvard University (Howard Hughes Medical Institute) published seminal work on improved base excision repair inhibition to yield higher-efficiency C:G-to-T:A base editors, and described high-fidelity protein engineering strategies such as HF-BE3 for RNP delivery. Harvard’s AAV delivery of split nucleobase editors — filed as a patent in 2019 in the IL jurisdiction — represents an important early infrastructure patent that anchors the field’s freedom-to-operate map.

The 2019–2021 period marks a proliferation and diversification phase. Beam Therapeutics published and filed extensively on next-generation CBEs featuring alternative deaminases (RrA3F, AmAPOBEC1, SsAPOBEC3B, PpAPOBEC1) with 12- to 69-fold reductions in off-target RNA deamination relative to the rAPOBEC1-containing BE4. Kobe University published the AID-2S/AID-3S architecture with truncated PmCDA1 (tCDA1EQ) in 2021, achieving a CBE compact enough for AAV vectors. The Institute of Zoology, Chinese Academy of Sciences reported HDAC6-inhibitor small molecules boosting CBE efficiency 2.45- to 9.95-fold in human cells in 2021.

The 2022–2026 period signals a transition toward therapeutic-grade product development. Intellia Therapeutics filed patents in 2023 covering mRNA-encoded cytidine deaminase and nickase constructs with separately encoded UGI. Consiglio Nazionale delle Ricerche received an active EP patent in 2024 covering cytidine deaminase variants with specific point mutations for Cas9 fusion. Cellectis S.A. filed a JP patent in 2025 on TALE base editors for gene and cell therapy of primary T cells and hematopoietic stem cells. The Institute of Genetics and Developmental Biology, Chinese Academy of Sciences filed a BR application in 2025 on newly identified cytosine deaminase enzymes for plant-applicable base editing systems.

Deaminase engineering and Cas9 diversification: the off-target reduction race

The dominant engineering challenge in CBE development is minimising unguided cytosine deamination in both genomic DNA and transcriptome-wide RNA — a challenge that multiple groups independently addressed by screening diverse non-rAPOBEC1 cytidine deaminases as drop-in replacements. Beam Therapeutics’ 2020 publications screened 153 CBE variants through a high-throughput cellular deamination assay, identifying eight next-generation editors using RrA3F, AmAPOBEC1, SsAPOBEC3B, and PpAPOBEC1 variants that achieve 12- to 69-fold reductions in C-to-U RNA edits and up to 45-fold reduction in unguided off-target DNA deamination relative to BE4-rAPOBEC1.

“Eight next-generation CBEs achieve 12–69-fold reductions in C-to-U RNA edits and up to 45-fold reduction in unguided off-target DNA deamination relative to BE4-rAPOBEC1 — identified from a screen of 153 CBE variants.”

Cas9 ortholog diversification runs in parallel. The University of Massachusetts demonstrated NmeCas9 fused to nucleotide deaminase enabling C-to-T editing at sites inaccessible to SpCas9 base editors, leveraging NmeCas9’s compact size and broader PAM compatibility (2021, SG). The Max Planck Institute for Molecular Plant Physiology engineered CDA1 truncations with Cas9 variants recognising alternative PAMs to create high-precision base editors targeting C−15, C−16, or C−18 relative to PAM NGG, substantially reducing off-target effects (2020). Kobe University’s truncation of PmCDA1’s DNA-binding domain produced the AID-2S (N-terminal fusion) and AID-3S (inlaid) architectures compatible with smaller Cas9 orthologs and within AAV packaging limits.

An orthogonal approach to off-target reduction comes from small-molecule augmentation. The Institute of Zoology, Chinese Academy of Sciences identified HDAC6 inhibitors Ricolinostat and Nexturastat A via high-throughput screening of 1,813 compounds, demonstrating 2.45- to 9.95-fold boosts in CBE editing efficiency in human cells (2021). This pharmacological augmentation strategy is transferable across CBE variants and may offer a clinical co-dosing lever. For novel deaminase discovery, the Italian CNR’s active EP patent (2024) on cytidine deaminase variants with specific point mutations, and the Chinese Academy of Sciences’ 2025 BR filing on newly identified cytosine deaminase enzymes, signal ongoing prospecting for deaminases with superior activity windows and distinct sequence preferences — an open competitive frontier for IP entrants, as described by WIPO patent landscape methodology.

Delivery architectures: from split-intein AAV to mRNA/LNP two-component systems

Safe and efficient delivery is described across this dataset as the primary translational bottleneck for CBE therapeutics. Three principal formats have been characterised: split-intein AAV, ribonucleoprotein (RNP), and mRNA/LNP — each with distinct advantages for in vivo versus ex vivo applications.

The split-intein AAV approach, patented by Harvard (IL, 2019), delivers each half of the split Cas9/nucleobase editor construct via recombinant AAV, with each half fused to intein-N or intein-C; upon co-expression, the full editor is reconstituted via protein splicing. RNP delivery of BE3 and HF-BE3, described by Boston Children’s Hospital and Harvard University (2017), achieves DNA-free base editing in mammalian cells with higher specificity than plasmid transfection — demonstrating that protein delivery format strongly influences the off-target profile. According to standards tracked by FDA guidance on gene therapy products, the transient expression profile of RNP and mRNA formats reduces integration risk compared to viral vector approaches.

The most significant recent development is Intellia Therapeutics’ 2023 mRNA-based two-component delivery architecture (IL, pending), which separates the cytidine deaminase–nickase fusion mRNA from a distinct UGI-encoding mRNA. This eliminates the need for UGI as part of a single fusion construct, improving manufacturing modularity and avoiding the size constraints of all-in-one fusion constructs. The architecture is compatible with lipid nanoparticle (LNP) delivery for in vivo systemic administration — the same platform validated for approved RNA therapeutics by organisations including EMA.

Therapeutic and agricultural application domains

Monogenic disease correction

The clearest therapeutic application across this dataset is the correction or modelling of pathogenic point mutations in monogenic diseases. The translational review from Sirius University of Science and Technology (2022) enumerates CBE and ABE applications to rare genetic diseases in vitro and in vivo, noting that nCas9- or dCas9-based systems can correct disease-causing mutations without double-strand break risk. The University of Pennsylvania’s review (2021) specifically demonstrates C-to-T correction of a prevalent mutation causing hereditary tyrosinemia type 1 using CBEs in human cells. The BEable-GPS database (ShanghaiTech University, 2019) profiled 20 base editors against all catalogued pathogenic SNVs, providing a genomic accessibility map that directly informs which CBE variant can address which disease mutation.

Blood disorders represent a particularly active sub-domain. The Imagine Institute / INSERM UMR 1163 review (2021) identifies CBEs as promising for sickle cell disease, beta-thalassemia, and other hemoglobinopathies driven by point mutations, noting the broad expansion of CBE tools enabling editing at previously inaccessible loci. The WHO estimates that haemoglobin disorders affect millions of births annually, underscoring the scale of unmet need that CBE platforms are being positioned to address.

Cancer immunotherapy and allogeneic cell engineering

CBE technology is being applied to engineer allogeneic CAR-T cells by knocking out TCR components, HLA markers, and checkpoint genes without generating translocatable DSBs. Cellectis S.A.’s TALE base editor patent (JP, 2025) specifically demonstrates C-to-T editing of TRAC and CD52 in primary T cells, with flow cytometry confirming loss of TCR and CD52 surface expression. The Revvity aptamer-mediated Pin-point™ platform (2023) achieves simultaneous multiple gene knockouts and site-specific transgene integration in primary human T cells with significantly reduced chromosomal translocation frequency compared to DSB-based approaches.

Agricultural genomics and plant base editing

Multiple results in this dataset describe CBE deployment in rice, Arabidopsis, and other crops. Anhui Academy of Agricultural Sciences (2023) constructed Cas12a-based plant CBEs using a truncated APOBEC3B deaminase with RAD51 single-stranded DNA-binding domain, achieving editing efficiencies of up to 68.75% in T0 rice. New transversion editors (CGBEs) for rice were demonstrated with C-to-G efficiencies up to 27.3% at NG-PAM sites (Guangdong Laboratory for Lingnan Modern Agriculture, 2022). The Institute of Genetics and Developmental Biology, Chinese Academy of Sciences filed a 2025 BR application on newly identified cytosine deaminase enzymes applicable to plant base editing systems.

Antibiotic discovery and microbial engineering

The Korea Research Institute of Bioscience and Biotechnology (KRIBB, 2021) demonstrated CBE-mediated multiplex genome editing in Bacillus subtilis and Paenibacillus polymyxa, achieving simultaneous quintuple gene knockouts at 75% efficiency to uncover cryptic antibiotic biosynthetic gene clusters — demonstrating that CBE platforms extend well beyond therapeutic applications into industrial biotechnology.

Assignee landscape, IP concentration, and strategic implications

Innovation in the CBE space is concentrated among a small number of US academic and commercial players — Harvard, Beam Therapeutics, Intellia Therapeutics, University of Massachusetts, and Rutgers — with meaningful parallel activity in China (Chinese Academy of Sciences groups, ShanghaiTech University) and emerging European assignees (Cellectis, Italian CNR). The Israel (IL) jurisdiction appears repeatedly as a PCT national phase entry point rather than as a locus of Israeli origination.

“Freedom-to-operate is tightly concentrated around Harvard and Beam Therapeutics: the foundational CBE patents and next-generation off-target reduction IP are held or licensed through Harvard and commercialised by Beam Therapeutics, which holds active or pending US, AU, and WO filings.”

Beam Therapeutics Inc. (Cambridge, MA) is the most prolific CBE-specific patent assignee in this dataset, with filings in WO (2020), AU (2021, pending), and US (2022, pending) directed to nucleobase editors with reduced off-target deamination. President and Fellows of Harvard College holds the core foundational patent on AAV delivery of split nucleobase editors (IL, 2019, pending), reflecting the academic origination of the CBE platform under the David Liu laboratory. Intellia Therapeutics holds two parallel 2023 IL filings on mRNA-delivered cytidine deaminase systems, signalling a clinical-stage company’s CBE pipeline. Cellectis S.A. (France) holds the 2025 JP TALE base editor patent targeting primary T cells and HSCs, representing European therapeutic cell engineering activity.

For entrants, the strategic picture is clear: filing on newly characterised deaminase enzymes — especially those with narrow editing windows or distinct sequence contexts — represents a tractable IP entry point orthogonal to core Cas9 fusion architecture claims. TALE-based (Cellectis), aptamer-mediated (Rutgers, Revvity), and novel deaminase-based systems (Chinese Academy of Sciences, Italian CNR) all represent architecturally differentiated approaches that can be pursued without directly infringing the Harvard/Beam Therapeutics IP stack. Regulatory and safety differentiation will additionally require whole-genome and whole-transcriptome off-target deamination assays as IND-enabling study components — an area where standardised reporting benchmarks, as tracked by PatSnap’s innovation intelligence resources, remain actively evolving.

The field’s regulatory trajectory is being shaped by the need for quantitative off-target profiling. Across this dataset, the convergence on whole-genome and whole-transcriptome deamination assays as acceptance criteria for IND-enabling studies is clear. R&D investment in high-sensitivity detection platforms and standardised benchmarks — aligned with guidance from bodies such as EMA and PatSnap Insights — will be essential for clinical advancement.