CRISPR 101: Cytosine and Adenine Base Editors

By Mary Gearing

Originally published Aug 16, 2016 and last updated Aug 6, 2020 by Jennifer Tsang.

When we talk about CRISPR applications, one negative often comes up: the low editing efficiency of homology-directed repair (HDR). Compared to non-homologous end joining, HDR occurs at a relatively low frequency, and in nondividing cells, this pathway is further downregulated. Rather than try to improve HDR, scientists have developed two classes of base editors: cytosine base editors (CBEs) and adenine base editors (ABEs). 

(There are also RNA base editors, but we’ll just be covering DNA base editors here. To learn more about RNA base editors head over to this blog post: CRISPR 101: RNA Editing with Cas13)

What bases can base editors edit?

CBEs mediate a C to T change (or a G to A change on the opposite strand). ABEs make an A to G change (or a T to C change on the opposite strand). This only accounts for four of 12 possible changes. 

More recently, the development of prime editing, which uses a different mechanism than CBEs or ABEs, from David Liu’s lab has allowed scientists to make all 12 possible base-to-base changes.

How do base editors work?

Base editing requires three elements. Broadly:

  1. A Cas nickase or Cas fused to a deaminase that makes the edit
  2. A gRNA targeting Cas to a specific locus
  3. A target base for editing within the editing window specified by the Cas protein

These elements were the starting point towards the development of the first cytosine base editors from postdocs Alexis Komor and the first adenine base editors from Nicole Gaudelli in David Liu’s lab in 2016.

Cytosine base editing

The beginnings of cytosine base editing

Komor created the first cytosine base editor by coupling a cytidine deaminase with the inactive dCas9 (Komor et al., 2016). These fusions convert cytosine to uracil without cutting DNA. Uracil is then subsequently converted to thymine through DNA replication or repair. Fusing an inhibitor of uracil DNA glycosylase (UGI) to dCas9 prevents base excision repair which changes the U back to a C mutation. To increase base editing efficiency, you’d need a way to force the cell to use the deaminated DNA strand as a template. To do so, the lab used a Cas nickase, instead of dCas9. The resulting editor, BE3, nicks the unmodified DNA strand so that it appears “newly synthesized” to the cell. Thus, the cell repairs the DNA using the U-containing strand as a template, copying the base edit.

Schematic of cytosine base editing. gRNA and Cas9-cytidine deaminase fusion come together, and forms a complex and binds the target. Cytidine deamination turns a C to a U on the free strand. Mismatch repair preserves edit if the modified strand is used as the template.
Cytidine deamination takes place on the free strand of DNA and converts a C to U without producing a double strand break.

The BE3 system increased editing frequency to above 30% for a variety of targets in human cell lines, with an average indel frequency of only 1.1%. These numbers are a vast improvement over Cas9-mediated HDR for the loci tested where average HDR-mediated editing frequency was only 0.5%, and more indels were observed than point modifications. CRISPR base editing persists through multiple cell divisions, indicating that this method produces stable edits. However, this system is also subject to off-target editing based on Cas9 off-target activity. 

Find CRISPR base editing plasmids at Addgene

Improving cytosine base editing scope and efficiency

Since the development of BE3, many research groups have made improvements to base editors including:

  • expanding targeting scope
  • improving editing efficiency
  • decreasing off-target effects

In 2016, Akihiko Kondo’s lab created the Target-AID base editor using a cytidine deaminase from sea lamprey fused to Cas9 nickase (Nishida et al., 2016). Target-AID acts similarly but not identically to BE3, modifying a 3-5 base window 18 bases upstream of the PAM. 

David Liu’s lab generated BE3 variants with other deaminases: AID, CDA1, and APOBEC3G (Komor et al., 2017). CDA1-BE3 and AID-BE3 edited Cs following a G more efficiently than BE3, but APOBEC3G displayed less predictable sequence preferences. 

The Liu lab also used natural and engineered Cas9 variants to develop five new base editors with distinct PAM sequences, expanding the number of available target sites for base editing (Kim et al., 2017). For each base editor, they observed editing activity with a minimum efficiency of ~50% and confirmed that the fusion protein retained the PAM properties of the individual Cas9. They also mutagenized the cytidine deaminase portion of the base editor to create SpCas9 base editors with editing windows as small as 1-2 nucleotides.

To reduce off-target effects associated with base editing, the lab created HF-BE3, a base editor containing high fidelity Cas9 variant HF-Cas9 (Rees et al., 2017). HF-BE3 showed 37-fold less off-target editing than BE3, with only a slight reduction in on-target editing efficiency. To further improve specificity, they purified HF-BE3 protein for delivery in ribonucleoprotein particles (RNPs) to both zebrafish embryos and the mouse inner ear.

Fourth-generation base editors

The fourth-generation base editors, BE4, reduce the undesired C->G or C->A conversions that can happen with BE3’s. T These byproducts likely resulted from excision by uracil N-glycosylase (UNG) during base excision repair. Adding a second copy of the UNG inhibitor, UGI, increases base editing product purity. The lab also extended the APOBEC1-Cas9n and Cas9n-UGI linkers to improve product purity, and these three improvements represent the fourth generation of base editors. Compared to BE3, BE4 offers a 2.3 fold decrease in C->G and C->A products as well as a 2.3 fold decrease in indel formation.

To further decrease indel formation 1.5-2 fold, the team fused bacteriophage protein Gam to the N-terminus of BE4 (Komor et al., 2017). Gam binds the free ends of DSBs, which may lead to cell death rather than NHEJ repair, thus removing these cells from the edited population. 

Another way to improve base editing efficiency for mammalian edits is to ensure the editors make it into the nucleus and that they are expressed well. The Liu lab modified the nuclear localization signals and codon usage BE4 to create BE4max and AncBE4max with a 4.2-6-fold improvement in editing efficiency (Koblan et al., 2018).

Sharing science just got easier... Subscribe to the blog

Adenine base editors

The beginning of adenine base editing

Nicole Gaudelli from David LIu’s lab created an adenine base editor that would convert adenine to inosine, resulting in an A to G change (Gaudelli et al., 2017). Creating an adenine base editor requires an additional step because there are no known DNA adenine deaminases.  They used directed evolution to create one from the RNA adenine deaminase TadA.  

After seven rounds of molecular evolution, they obtained four adenine base editors (ABEs). ABE7.10 is the most active editor, displaying an average editing efficiency of 53% with an editing window of target positions 4-7. ABEs 6.3, 7.8, and 7.9 display slightly wider editing windows of position 4-9, although editing efficiency may be lower at positions 8 and 9. While cytosine base editors often produce a mixed population of edits, ABEs do not display significant A to non-G conversion at target loci. The removal of inosine from DNA is likely infrequent, thus preventing the induction of base excision repair.

In terms of off-target effects, ABEs also compare favorably to other methods. In a head to head comparison with Cas9, Cas9 modified 9/12 known off-targets with a 14% indel rate, while ABE7.10 modified only 4/12 off-targets with a frequency of 1.2%. Although the lab did not conduct comprehensive genome-wide studies of ABE specificity, their other experiments suggest that ABEs are robust but specific editors.

Improving adenine base editing

When the Liu lab created BE4max and AncBE4max mentioned above, they also made an adenine base editor with improved nuclear localization and expression. This base editor was named ABEmax.

In 2020, two papers were published describing additional ABEs evolved from ABE7.10 that have improved base editor targeting flexibility and specificity. In this first, the Liu lab used phage-assisted evolution selection systems to generate ABE8e(TadA-8e V106W), which edits ~590-fold faster than the TadA from ABE7.10 without increasing off-target activity (Richter et al., 2020). This is important because ABEs have generally been slow meaning that the Cas9 domain often lets go of the DNA before the edit is made.

Using ABE7.10 as a starting point, Gaudelli evolved the base editor into 40 new ABE8 variants (Gaudelli et al., 2020). Compared to ABE7.10, ABE8s resulted in 1.5-fold more editing at protospacer positions A5-A7 and 3.2-fold more editing at positions A3-A4 and A8-A10 at NGG PAM and 4.2-fold higher editing efficiency at non-NGG PAM variants compared to ABE7.10. ABE8s have an improved base editing capacity, even at sites previously difficult to target. ABE8s can achieve 98-99% target modification in primary T cells making them a promising tool for cell therapy applications.

Dual base editors

What about combining the function of both editors into one? This has recently been done by fusing the adenine and cytosine editing components together.

Keith Joung’s lab created a dual-deaminase editor called SPACE (synchronous programmable adenine and cytosine editor) by fusing miniABEmax-V82G and Target-AID to Cas9 (Grünewald et al., 2020). Another lab took a similar approach. Dali Li’s lab’s A&C-BEmax consists of a fusion of both cytidine and adenosine deaminases with a Cas9 nickase (Zhang et al., 2020). It has increased CBE activity and reduced RNA off-targeting activity compared to ABEmax.

Base Editing Publication Highlights

In the table below, you'll find a list of publications we've highlighted in this article. 


Publication Plasmids Highlights
Komor et al., 2016 BE1, BE2, BE3 BE3 displays highest editing efficiency but higher indel formation than BE2
Nishida et al., 2016 Target-AID Edits 3-5 base window surrounding -18 position upstream of the PAM
Kim et al., 2017 SaBE3, BE3 PAM variants, BE3 editing window variants Greatly expands the number of target loci for base editing
Rees et al., 2017 HF-BE3 HF-BE3 and ribonucleoprotein delivery decrease BE3 off-target activity
Komor et al., 2017 BE4 and BE4-Gam; AID, CDA1 and APOBEC3G BE3 variants A second copy of UGI improves product purity. Gam decreases indel frequency.
Koblan et al., 2018 BE4max, ArcBe4max, ABEmax Improved nuclear localization and expression
Gaudelli et al., 2017 Adenine base editors (ABE7.10) A->I (A->G) editing with high product purity and low off-target editing
Richter et al., 2020 ABE8 590x faster adenine base editing than ABE7.10
Gaudelli et al., 2020 ABE8e Efficient editing in primary cells
Zhang et al., 2020 A&C-BEmax Has both cytosine and adenine base editing
Grünewald et al., 2020 SPACE Has both cytosine and adenine base editing

Download Addgene's CRISPR 101 eBook!

Jennifer Tsang contributed to updating this article.


Gaudelli NM, Komor AC, Rees HA, Packer MS, Badran AH, Bryson DI, Liu DR (2017) Programmable base editing of A•T to G•C in genomic DNA without DNA cleavage. Nature 551:464–471 .

Gaudelli NM, Lam DK, Rees HA, Solá-Esteves NM, Barrera LA, Born DA, Edwards A, Gehrke JM, Lee S-J, Liquori AJ, Murray R, Packer MS, Rinaldi C, Slaymaker IM, Yen J, Young LE, Ciaramella G (2020) Directed evolution of adenine base editors with increased activity and therapeutic application. Nat Biotechnol 38:892–900 .

Grünewald J, Zhou R, Lareau CA, Garcia SP, Iyer S, Miller BR, Langner LM, Hsu JY, Aryee MJ, Joung JK (2020) A dual-deaminase CRISPR base editor enables concurrent adenine and cytosine editing. Nat Biotechnol 38:861–864 .

Kim YB, Komor AC, Levy JM, Packer MS, Zhao KT, Liu DR (2017) Increasing the genome-targeting scope and precision of base editing with engineered Cas9-cytidine deaminase fusions. Nat Biotechnol 35:371–376 .

Koblan LW, Doman JL, Wilson C, Levy JM, Tay T, Newby GA, Maianti JP, Raguram A, Liu DR (2018) Improving cytidine and adenine base editors by expression optimization and ancestral reconstruction. Nat Biotechnol 36:843–846 .

Komor AC, Kim YB, Packer MS, Zuris JA, Liu DR (2016) Programmable editing of a target base in genomic DNA without double-stranded DNA cleavage. Nature 533:420–424 .

Komor AC, Zhao KT, Packer MS, Gaudelli NM, Waterbury AL, Koblan LW, Kim YB, Badran AH, Liu DR (2017) Improved base excision repair inhibition and bacteriophage Mu Gam protein yields C:G-to-T:A base editors with higher efficiency and product purity. Sci Adv 3:eaao4774 .

Nishida K, Arazoe T, Yachie N, Banno S, Kakimoto M, Tabata M, Mochizuki M, Miyabe A, Araki M, Hara KY, Shimatani Z, Kondo A (2016) Targeted nucleotide editing using hybrid prokaryotic and vertebrate adaptive immune systems. Science 353:aaf8729–aaf8729 .

Rees HA, Komor AC, Yeh W-H, Caetano-Lopes J, Warman M, Edge ASB, Liu DR (2017) Improving the DNA specificity and applicability of base editing through protein engineering and protein delivery. Nat Commun 8: .

  • Find plasmids from this publication at Addgene.

Richter MF, Zhao KT, Eton E, Lapinaite A, Newby GA, Thuronyi BW, Wilson C, Koblan LW, Zeng J, Bauer DE, Doudna JA, Liu DR (2020) Phage-assisted evolution of an adenine base editor with improved Cas domain compatibility and activity. Nat Biotechnol 38:883–891 .

Tsai SQ, Zheng Z, Nguyen NT, Liebers M, Topkar VV, Thapar V, Wyvekens N, Khayter C, Iafrate AJ, Le LP, Aryee MJ, Joung JK (2014) GUIDE-seq enables genome-wide profiling of off-target cleavage by CRISPR-Cas nucleases. Nat Biotechnol 33:187–197 .

Zhang X, Zhu B, Chen L, Xie L, Yu W, Wang Y, Li L, Yin S, Yang L, Hu H, Han H, Li Y, Wang L, Chen G, Ma X, Geng H, Huang W, Pang X, Yang Z, Wu Y, Siwko S, Kurita R, Nakamura Y, Yang L, Liu M, Li D (2020) Dual base editor catalyzes both cytosine and adenine base conversions in human cells. Nat Biotechnol 38:856–860 .

Additional Resources on the Addgene Blog

Additional Resources on

Topics: CRISPR, Base Editing

Leave a Comment

Sharing science just got easier... Subscribe to our blog