Precision Genome Editing Using Cytosine

PROTOCOL
https://doi.org/10.1038/s41596-020-00450-9
Precision genome editing using cytosine and

adenine base editors in mammalian cells
1,2,3 ✉
Tony P. Huang1,2,3, Gregory A. Newby1,2,3 and David R. Liu
Genome editing has transformed the life sciences and has exciting prospects for use in treating genetic diseases. Our
laboratory developed base editing to enable precise and efficient genome editing while minimizing undesired byproducts
and toxicity associated with double-stranded DNA breaks. Adenine and cytosine base editors mediate targeted A•T-to-
G•C or C•G-to-T•A base pair changes, respectively, which can theoretically address most human disease-associated
single-nucleotide polymorphisms. Current base editors can achieve high editing efficiencies—for example, approaching
100% in cultured mammalian cells or 70% in adult mouse neurons in vivo. Since their initial description, a large set of
base editor variants have been developed with different on-target and off-target editing characteristics. Here, we describe
a protocol for using base editing in cultured mammalian cells. We provide guidelines for choosing target sites, appropriate
base editor variants and delivery strategies to best suit a desired application. We further describe standard base-editing
experiments in HEK293T cells, along with computational analysis of base-editing outcomes using CRISPResso2. Beginning
1234567890():,;
with target DNA site selection, base-editing experiments in mammalian cells can typically be completed within 1–3 weeks
1234567890():,;
and require only standard molecular biology techniques and readily available plasmid constructs.
Introduction
Genome-editing agents have had a transformative effect on the study of biological systems, the
engineering of cells and organisms with novel properties and even the treatment of genetic diseases in
human patients. The earliest class of widely used modern genome-editing agents are the program-
mable nucleases, including zinc-finger nucleases1–5, transcription activator-like effector nucleases6–11
and CRISPR-associated (Cas) nucleases12–15. These agents induce targeted double-stranded DNA
breaks (DSBs) to stimulate cellular repair processes that ultimately result in genome modification.
Cellular repair of DSBs, however, typically leads to various outcomes through multiple competing
pathways16–19, including potential cell cycle arrest from p53-mediated DNA damage responses20,21.
DSB-initiated non-homologous end joining and microhomology-mediated end joining, which, while
predictable22–24, are uncontrollable processes that can result in gene disruptions through insertions,
deletions, duplications or other DNA rearrangements16,17,25,26. The addition of a donor DNA tem-
plate can stimulate homology-directed repair (HDR) to precisely replace DNA alleles27,28, but this
process is largely limited to dividing cells. In addition, even in cells that support HDR, desired
products are typically disfavored relative to indels from end-joining processes, resulting in mixtures
of HDR-corrected products and small insertions or deletions28–31. Methods to install targeted single-
nucleotide mutations without inducing DSBs are therefore desirable, especially because single-
nucleotide polymorphisms (SNPs) comprise the largest class of known human pathogenic genetic
variations32–34.
To address this need, we developed base editing, a genome-editing method that directly converts
targeted base pairs without requiring DSBs, HDR or donor DNA templates35,36. Base editing typically
results in low indel formation and high product purities, defined as the percentage of all edited alleles
that contain the anticipated base conversion with no indels. Base editors contain two primary
components: a programmable DNA-binding protein, such as a catalytically impaired Cas nuclease
that cannot make DSBs, and a DNA-modifying enzyme35–37. Cas domain binding directed by a
programmable single guide RNA (sgRNA) to a target genomic locus exposes a small segment of
single-stranded DNA (ssDNA) in an R-loop12,38. This ssDNA R-loop then serves as the substrate for
the fused ssDNA-modifying enzyme during base editing. A high effective concentration of target
1
Merkin Institute of Transformative Technologies in Healthcare, The Broad Institute of Harvard and MIT, Cambridge, MA, USA. 2Department of
Chemistry and Chemical Biology, Harvard University, Cambridge, MA, USA. 3Howard Hughes Medical Institute, Harvard University, Cambridge, MA,
USA. ✉e-mail: drliu@fas.harvard.edu
NATURE PROTOCOLS | VOL 16 | FEBRUARY 2021 | 1089–1128 | www.nature.com/nprot 1089

PROTOCOL NATURE PROTOCOLS
a Cytosine base editing b Adenine base editing
ssDNA-specific Wild-type TadA monomer

cytidine deaminase Guide RNA (optional for ABE7.10, ABE8)
DNA target C DNA target A

recognition 5′ PAM 3′ recognition 5′ PAM 3′
3′ 5′ 3′ 5′
G T
5′ 3′ 5′ 3′
Cas effector
(nickase or dead)
Target C deamination Evolved TadA monomer
Cas effector Target A deamination
UGI (2× for in ssDNA bubble in ssDNA bubble
additional product purity) (nickase or dead)
DNA repair U DNA repair I

or replication or replication
5′ PAM 3′ 5′ PAM 3′ 5′ PAM 3′ 5′ PAM 3′
T 3′ 5′ G 3′ 5′
A C
3′ 5′ G 3′ 5′ T
5′ 3′ 5′ 3′
Fig. 1 | Overview of base editing. a, Cytosine base editing. A dead or nickase Cas protein is fused to a ssDNA-specific cytidine deaminase, which
converts cytosine to uracil. Cas binding by a programmed sgRNA generates an ssDNA R-loop, whereby the fused cytidine deaminase converts
exposed cytosines to uracil. Cas nicking of the unedited strand biases DNA repair toward repairing the unedited strand, while fused uracil glycosylase
inhibitor (UGI, 1× or 2×) impedes base excision repair of the uracil, promoting C•G-to-U•A conversions, which are converted to T•A base pairs after
replication or additional DNA repair35. b, Adenine base editing. A dead or nickase Cas protein is fused to an evolved TadA monomer. Cas binding by a
programmed sgRNA generates an ssDNA R-loop, whereby the fused TadA monomer converts exposed adenosines to inosine. Cas nicking of the
unedited strand biases DNA repair toward repairing the unedited strand, promoting A•T-to-I•C conversions, which are converted to G•C base pairs
after replication or additional DNA repair36. Optionally, a wild-type TadA monomer can also be fused, which can improve editing efficiencies, probably
by promoting the formation of dimeric TadA. PAM, protospacer adjacent motif.
DNA with respect to the ssDNA-modifying enzyme due to fusion of the latter with the Cas domain
promotes efficient and selective base modification. Editing is canonically localized to a narrow
window of bases (‘editing window’) exposed by Cas binding, although strategies have been introduced
to expand39–44 or further restrict45–47 this window. Recently, we developed a CRISPR-free base-
editing system that uses an enzyme that modifies double-stranded DNA fused to a programmable
DNA-binding protein37. These new RNA-free base editors enable precision editing of DNA in
mitochondria, which lack known pathways suitable for guide RNA import37.
After DNA modification, the resulting base pairing mismatch is subject to cellular DNA repair.
Most base editors bias the outcome of DNA repair by using a nickase form of the Cas protein (nCas)
to nick the non-deaminated strand35,36. Nicking stimulates resynthesis of the non-edited strand by
using the deaminated strand as a template48,49, resulting in conversion of both DNA strands at the
target position to the desired base pair35.
Two types of DNA base editors have been well characterized: cytosine base editors (CBEs), which
convert C•G base pairs to T•A base pairs (Fig. 1a), and adenine base editors (ABEs), which convert
A•T base pairs to G•C base pairs (Fig. 1b)35,36. More recently, C-to-G base editors (CGBEs) were
developed by exploiting the observation that CBEs occasionally generate C-to-non-T edits35,50–55.
However, current CGBEs have modest editing efficiencies, are highly site dependent and edit via an
as-yet undefined mechanism, limiting their widespread application50,51.
CBEs typically have three components: a cytidine deaminase, a Cas nickase and uracil glycosylase
inhibitor (UGI)35, which inhibits counterproductive DNA repair processes. ABEs are comprised of
two components: an evolved adenosine deaminase capable of recognizing ssDNA and a Cas ortholog
nickase. Although additional research is needed to develop robust transversion base editors, ABEs
and CBEs can generate all four transition mutations, which in theory could be used to install or
correct most known human pathogenic SNPs32,33,56.
Base-editing developments
Base editors exhibit various editing properties, with the importance of each feature dependent on the
desired application (Fig. 2). A diverse toolkit of base editor variants has been developed to enhance
the usage of base editing in different contexts. Modifications have been made to each base editor
component, including general improvements to base editor architecture and expression, usage of
1090 NATURE PROTOCOLS | VOL 16 | FEBRUARY 2021 | 1089–1128 | www.nature.com/nprot

NATURE PROTOCOLS PROTOCOL
Base editor class
CBEs enable conversion of C•G-to-T•A

ABEs enable conversion of A•T-to-G•C Editing window
Sequence context
APOBEC1 TC TU TC-preferred
Determines which bases can or cannot be
GC GU GC-disfavored edited within the protospacer
Editing efficiency can vary based on
deaminase sequence preferences
5′ 3′
3′ 5′
Off-target editing
Cas PAM requirement
5’ 3’
C
C NGG NCT
Deaminase-dependent Cas-dependent Sequence requirement to initiate cas
off targets off targets binding to target site
Fig. 2 | Key properties of base editors. The choice of base editor class (red box) determines the type of base conversion: C•G-to-T•A for CBEs or
A•T-to-G•C for ABEs. Some deaminases exhibit sequence context preferences (green box), which can alter the efficiency of target or non-target
nucleotide conversion. Base editing occurs within an editing window, canonically ~5 nt wide, ~15 ± 2 nucleotides upstream of the PAM (gray box). The
size and location of the editing window is influenced by the binding properties of the deaminase and the choice of Cas protein. The Cas protein’s PAM
compatibility (purple box) determines the location of potential target DNA engagement within the genome, and thus the position of the editing
window. The Cas protein and the deaminase also determine the potential location and extent of off-target editing (orange box).
protospacer adjacent motif (PAM) variants or high-fidelity Cas proteins and development of
nucleobase deaminases with different editing properties.
General improvements and modifications

Since the development of BE3 and ABE7.10, the canonical cytosine and adenine base editors,
researchers have improved the efficiency and applicability of base editing. To address potential base
editor expression and localization limitations, codon usage and nuclear localization signals (NLS)
were optimized57,58, resulting in substantially enhanced editing efficiencies. Additional UGI
domains55,59 or other fusion partners (Gam, ssDNA-binding domain from Rad51) have been added
to improve CBE product purity and reduce indel formation55, or further increase editing
efficiencies60.
Modified fusion strategies have also been used to alter or improve the targeting capabilities of base
editors. C-terminal fusion of deaminase partners can narrow and shift the editing window of base
editors relative to their N-terminal fusion counterparts54,61. Although only certain deaminases are
compatible, these C-terminal deaminase CBEs potentially afford access to target sites less accessible to
canonical N-terminal CBEs61. Alternatively, shortened, modified linkers have also shown success in
narrowing the editing window47. In contrast, Yao and coworkers found that deaminases could be
installed directly within the PAM-interacting domain of SpCas9, resulting in a base editor, BE-PIGS,
with an expanded editing window44. Finally, Gao, Yachie, Li, Joung and their respective coworkers
recently reported on dual fusion of cytidine and adenosine deaminases to an nCas variant, resulting in
base editors capable of generating simultaneous C•G-to-T•A and A•T-to-G•C edits within a target
genomic locus62–65.
Base editors have also been split into a deaminase component and a Cas component, with each
fused to one partner of a well-characterized, high-affinity protein–protein or protein–RNA interac-
tion40–43,66–70. Such split base editors allow the deaminase:Cas protein stoichiometry to be drastically
altered, leading to greatly expanded editing windows and increased mutagenesis40–43. Split base
editors have also enabled efficient packaging of base editors into therapeutically relevant viral vectors
such as adeno-associated virus (AAV), which would not be possible with full-length base editors, as
they are larger than the AAV packaging limit66–70. Split base editors have been used to target several
biological systems66–70, and in some cases have even been observed to yield greater editing efficiency
than full-length editors, potentially because smaller genes can express more robustly66. Efforts to
couple cytidine or evolved adenosine deaminases with Cas9 variants smaller than SpCas9 (1,368 aa)

could result in single-AAV base editors that overcome remaining delivery limitations. Altogether,
these architectural improvements and variations to base editors have resulted in versions that are
better suited to enable base editing in various biological contexts.
Base editors with Cas domain variants

The first reported CBE and ABE both use the D10A nickase form of SpCas9 to direct DNA binding
and mismatch repair35,36,54. Because SpCas9 requires recognition of an NGG PAM to engage a target
DNA sequence12, the targeting scope of these initial base editors is limited to cytosines and adenines
located in a 4–5-nt window ~15 ± 2 nt upstream of an NGG PAM35,36,54. Although all pathogenic
SNPs resulting from transition mutations could in principle be installed or corrected by using CBEs
and ABEs, only 27% and 31%, respectively, of those mutations are actually targetable if the base editor
recognizes only sequences with an NGG PAM32,33,56.
Several natural, engineered and evolved Cas protein variants have been described with altered
PAM preferences71–82. These Cas orthologs have been widely adopted as effective alternatives for
wild-type SpCas9 to access different genomic sites for applications including gene knockout, tran-
scriptional activation or repression, epigenetic modification and HDR, among others19,83,84. We and
others hypothesized that base editing could also be generalized by using Cas orthologs compatible
with different PAM sequences. Indeed, both CBE and ABE have now been shown to be compatible
with various PAM-variant39,45,55,58,66,67,74,75,77–82,85–111, circularly permuted39,53,112 and high-
specificity Cas orthologs73,75,94,102,113–120. The current set of base editors reported place ~95% of
human pathogenic SNPs within the editing window of one or more CBEs or ABEs78, and thus the
vast majority of genomic sites can now in principle be accessed.
Base editors with deaminase domain variants

While the Cas ortholog–sgRNA complex generally dictates which sites in the genome can be targeted,
the fused ssDNA-modifying enzyme determines the type and extent of DNA modification after Cas
protein binding. Although the types of base conversions that can be reliably achieved are currently
limited to the four transition substitutions (A-to-G, G-to-A, T-to-C and C-to-T) and sometimes C-
to-G transversions, modifications and additional discoveries have been made to generate deaminases
with altered targeting properties, including enhanced product purity53, modified editing
windows41–43,45–47,61,103,106,109,112,121–127, altered sequence preferences36,41–43,57,61,94,103,121–123,128–130
and reduced off-target DNA and RNA editing45,46,61,103,111,112,121–123,125,126,131–133.
Engineering or evolution efforts to modify the substrate affinity of deaminase domains have been
effective strategies for modulating the on-target and off-target editing properties of base editors.
Deaminases engineered to have reduced ssDNA binding affinity exhibit both narrowed editing
windows and reduced deaminase-dependent off-target editing45,112,122,125,131,132. Similarly, engineered
deaminases with reduced native RNA binding affinity have been generated to minimize undesired
RNA editing of DNA base editors36,46,125,126,131,133. Importantly, RNA and DNA editing can
apparently be decoupled in cells, as variants with decreased off-target RNA editing but unaffected off-
target DNA editing, and vice versa, have been found126,131. Conversely, directed evolution124 or
phage-assisted continuous evolution106,121,134 of cytidine and adenosine deaminases have generated
variants with improved solubility134 or enhanced catalytic activity106,121,124. These evolved deami-
nases exhibit increased editing efficiencies, improved Cas ortholog compatibility, reduced sequence
context preferences and broadened editing windows when used as base editors106,121,124.
One limitation of broadly active deaminases is the potential increase in bystander editing—editing
that occurs at non-target cytosines or adenines within the editing window—because of the increase in
editing window size. Although bystander editing is inconsequential in some cases because of
degeneracy in the genetic code, bystander editing in other cases can lead to undesired non-silent
mutations135. One strategy to minimize bystander editing is to use context-specific deaminases94,103
rather than a single, sequence-agnostic deaminase. Although this strategy requires the identification
or development of multiple deaminases to cover the same sequence space, it can enable more precise
editing by reducing bystander and off-target editing if the undesired nucleotides are in a different
sequence context than that of the on-target nucleotide.
We and other groups have described the use of many cytidine deaminases in a base-editing
context, each of which offers distinct base-editing properties, including many with complementary
sequence context preferences41–43,46,47,54,57,60,61,87,90,94,96,101–104,109,115,121–123,125,128–133,136–143. Cur-
rently, laboratory-evolved TadA variants remain the only class of adenosine deaminases capable of
modifying DNA. However, many generations of TadA mutants have been reported, each with

different adenine base-editing activities, sequence preferences, Cas compatibilities and other prop-
erties36,106,133. Collectively, the general improvements to base editor construction, the development of
many different deaminases for base editing and the use of Cas orthologs with expanded PAM
compatibilities have resulted in a large, diverse toolkit of base editors that satisfy the editing prop-
erties necessary for a wide range of applications.
CRISPR-free base editing of mitochondrial DNA

Previously reported cytidine deaminases operate on single-stranded nucleic acids, and thus their use
in base editors has required unwinding of target DNA by a CRISPR–Cas protein84,135. Although
transition point mutations in mtDNA cause many genetic diseases, base editing within mtDNA has
thus far been stymied by the difficulty of transporting guide RNA into mitochondria144. In colla-
boration with the laboratories of Joseph Mougous and Vamsi Mootha, we recently reported DddAtox,
a double-strand DNA deaminase derived from an interbacterial toxin37. By fusing split-DddA halves
that are non-toxic and inactive until brought together on target DNA with programmable DNA-
binding TALE repeat arrays and UGI, we developed RNA-free DddA-derived cytosine base editors
(DdCBEs) that mediate targeted C•G-to-T•A conversions in human mtDNA with high target spe-
cificity and product purity37. DdCBEs enable the installation or correction of mtDNA mutations in
human cells, rather than the elimination of mtDNA copies that results from mtDNA cleavage by
targeted nucleases. As DdCBEs are a new development reported while this manuscript was in pre-
paration, this article will not cover them further.
Applications of base editing

Base editing has been widely adopted as both a tool for basic science and as a potential
therapeutic modality. Diverse organisms ranging from prokaryotes to plants, mice, monkeys
and human embryos have been shown to support base editing. These applications include
the following: installing or removing SNPs in tissue culture, plants or living animals35,36,58,66,67,69,85,
87–92,96,97,99,101,117,136,137,142,145–170
both in the nucleus and, more recently, in mitochondria by using
CRISPR-free base editors37; generating precise gene-silenced populations171,172; recording cellular
events138,173; tracing cell lineages174; diversifying antibodies43; deleting splice sites93; and many others
(all reviewed extensively elsewhere84,135,145,175).
Comparison with other technologies (Cas9 nucleases, HDR and prime editing)
Base editing along with prime editing (PE)176 are currently the two technologies capable of precisely
modifying single nucleotides at targeted sites in the genomes of live mammalian cells without gen-
erating DSBs. Earlier nuclease-based technologies require DSB repair (non-homologous end joining,
microhomology-mediated end joining and HDR) and result primarily in undesired indel forma-
tion16,17. Although HDR can generate precise, programmable corrections when a donor template is
supplied27,28, editing efficiencies are typically much lower than base editing, with average levels
reported to be ≤10%28,29 up to a maximum of ~60%30. HDR typically results in product distributions
that favor indels over desired HDR products, with some notable exceptions29. In contrast, base editing
routinely achieves high ratios (typically >10:1) of desired products to indels or other byproducts35,36.
HDR is also largely limited to dividing cells and typically shows very low efficiencies (<5%) in
non-dividing cells177,178.
We recently reported PE, which takes advantage of the RNA:DNA hybrid nature of Cas9 inter-
actions to enable a wide range of local targeted changes within the genome, including nucleotide
conversions, small insertions and small deletions176. Base editing and PE have complementary
properties. PE is less dependent on PAM availability and offers a wider variety of potential changes
than can be made with base editing. In contexts in which the desired modification is a transition
mutation that is positioned within a base-editing window, ABE and CBE generally offer higher
editing efficiencies than PE176 while generating fewer indels. Furthermore, base editors have had the
benefit of being optimized through multiple generations of improvements and variations over the
past 4 years, and have been validated to be compatible with many delivery modalities in various
biological systems and cell types84,135, making them robust tools for most applications requiring the
installation or correction of transition point mutations.

Box 1 | Designing sgRNAs for base editing

The first step in sgRNA design is to identify available PAMs that would place the target nucleotide within the editing window of a characterized
base-editing deaminase. Candidate sgRNAs should be designed to exactly match the 5′-to-3′ DNA sequence containing the target nucleotide and
directly 5′ of the PAM (3′ for Cas12a applications), as the genomic strand displaced by sgRNA binding is the substrate for deamination. For U6
promoter-driven transcripts, a 5′ guanosine is preferred and typically added for strong transcription193. For some target sites, adding this 5′
guanosine will result in a mismatch between the sgRNA and the target, which has been shown to affect the activity of some Cas variants,
particularly high-fidelity variants75,194.
Because all base-editing deaminases were first optimized or characterized with wild-type SpCas9, base editing tends to be most efficient when
using wild-type SpCas9, which recognizes an NGG PAM. If multiple NGG PAMs are available, priority should be given to sgRNAs that minimize the
number of undesired bystander edits within the editing window and the number of predicted Cas off-target sites with high sequence similarity to
the target protospacer. Webtools including CHOPCHOP v2195 (https://chopchop.cbu.uib.no/), CRISPOR196 (crispor.tefor.net) or Cas-OFFinder187
(http://www.rgenome.net/cas-offinder/) can be used for in silico prediction of Cas-dependent off-targets.
If an NGG PAM is not available or would place the editing window such that it includes an undesired bystander edit, determine if any of the
available PAMs that optimally position the target edit while excluding bystander edits can be recognized by an alternative Cas ortholog. An
example sgRNA selection process is shown in Fig. 3. Because certain sgRNAs may work unusually well or poorly with base editors, it is usually
worthwhile to examine multiple suitable sgRNAs.
Guide RNA design for broad mutagenesis applications is generally less restrictive. For these applications, a web-based sgRNA design tool, such as
CHOPCHOP v2 or BE-designer197 (http://www.rgenome.net/be-designer/), can be used to search the target locus for potential sgRNA sequences
with minimal predicted Cas off-targets and highest on-target binding efficiency. To achieve the highest editing efficiency, we recommend
prioritizing sgRNAs that place the target(s) in the center of the editing window (positions 5–6) when applicable.
Limitations of base editing

The two main requirements for using base editors are (i) targetability and (ii) delivery. Targeting of a
base editor requires the presence of a suitable PAM for Cas binding and an editing window that
includes the target nucleotide while excluding unacceptable bystander edits. Base editor variants with
enlarged editing windows39,40,44,112 or altered PAM specificity39,45,55,58,66,67,74,75,77–82,85–111 have been
generated to overcome targeting scope limitations. Editing at off-target genomic loci can occur if the
target protospacer is highly similar to other sites in the cellular genome (Cas-dependent off-target
DNA editing)35,36, or if the base editor uses a deaminase with high native nucleic acid affinity (Cas-
independent off-target DNA or RNA editing)125,132. Off-target editing can be minimized by careful
sgRNA design (Box 1 and Fig. 3), using high-fidelity Cas variants94,114–119, selecting deaminase
variants that have diminished DNA- or RNA-binding capacity45,46,61,103,112,121–123,125,126,131–133 and/
or decreasing the dose or duration of base editor exposure106,112,114,149,179.
Although base editors generally support high product purities, they can occasionally generate
unintended byproducts35,52–55,133. For CBEs, base editing can sometimes result in C-to-R (where
R = A or G) edits, probably due to cellular repair of abasic sites that result from uracil excision. CBE
architectures containing certain engineered deaminases (eA3A(T44D, S45A) or rAPOBEC1(H47E,
S48A))53 and/or multiple copies of UGI (eBE359, BE455, FNLS58, BE4max57, etc.) tend to have much
higher C-to-T product purity. For ABEs, base editing can sometimes occur at nearby cytosines within
the target protospacer, resulting in C-to-T (from off-target deamination) or C-to-G (hypothesized
from cellular repair of uracil) conversions52,53,133.
A method to deliver base editor proteins, or DNA or mRNA encoding them, must be considered
for each cell type or biological system of interest. Lipofection or electroporation of base editor protein
or base editor materials can lead to efficient editing in cell types that are amenable to these delivery
methods35,36,114,132,142,179,180. Viruses such as lentivirus, adenovirus, and AAV have proven to support
efficient base editing when used as a delivery vehicle in various tissues and cell types both in vitro and
in vivo58,66,67,69,70,88. However, some organisms and cell types remain recalcitrant to current delivery
methods66. The future development of improved delivery methods would significantly advance the
applicability of base editors.
Experimental design
A typical base-editing experiment comprises four phases (Fig. 4): (i) target sgRNA design and base
editor selection, which requires identifying target protospacers and associated Cas proteins, candidate
deaminases and suitable base editor architectures; (ii) reagent selection and preparation, which
requires selecting a base editor delivery method and preparing requisite reagents; (iii) base editor
delivery, which involves delivery of base editor components into a biological system of choice; and
(iv) base-editing analysis, which is done via Sanger or high-throughput sequencing (HTS).
Base editors are commonly used for two major classes of applications: (i) precise allele replacement
(nucleotide substitution) or (ii) broad mutagenesis within a target locus or region. Although many

Human chromosome 3 CTNNB1 locus
Nearby bystander Target base

bases
41,202,536 41,202,604
5′ 3′
3′ 5′
Guide A
Guide B
Guide C
HEK293T Guide D
Guide E
Desired outcome: efficient, precise correction of target adenine, minimize off-target editing
Predicted Cas OTs Guide

Guide ID Guide sequence (5’–3’) with editing window PAM (≤ 3 mismatches) rank* Rationale
Advantages:
Target adenine in optimal editing window
Wild-type SpCas9 NGG PAM
A A 1C 2A 3A 4A 5C 6C 7A 8G 9A 10A 11G 12C 13C 14G 15C 16T 17C 18C 19T 20 GGG 9 1 Neutral considerations:
Average predicted Cas OTs
Disadvantages:
Two bystander adenines in editing window
Advantages:
Wild-type SpCas9 NGG PAM
Neutral considerations:
B G 1A 2C 3A 4A 5A 6C 7C 8A 9G 10A 11A 12G 13C 14C 15G 16C 17T 18C 19C 20 TGG 7 3 Average predicted Cas OTs
Disadvantages:
Three bystander adenines in editing window
Target adenine out of optimal editing window
Advantages:
C C 1A 2A 3A 4C 5C 6A 7G 8A 9A 10G 11C 12C 13G 14C 15T 16C 17C 18T 19G 20 GGC 36 2 NG PAM requirement, possibly lower activity
One bystander adenine in editing window
Disadvantages:
High potential Cas off-targets if using SpCas9-NG
Advantages:
D A 1C 2C 3A 4G 5A 6A 7G 8C 9C 10G 11C 12T 13C 14C 15T 16G 17G 18G 19C 20 TAT 8 4 Average predicted Cas OTs
Disadvantages:
NAT PAM requirement (few available Cas variants)
Advantages:
NG PAM requirement, possibly lower activity
E C 1A 2G 3A 4A 5G 6C 7C 8G 9C 10T 11C 12C 13T 14G 15G 16G 17C 18T 19A 20 TGT 42 5 Disadvantages:
High potential Cas off-targets if using SpCas9-NG
Target adenine out of optimal editing window
*sgRNAs can be qualitatively filtered/ranked based on their potential desired vs. undesired properties. Properties can include: positioning of bystander edits relative
to the editing window, positioning of the desired edit in the center of the window and/or compatible Cas variant and protospacer combinations with the fewest predicted off-targets.
Fig. 3 | Example base-editing target selection. Example target selection for eliciting an A-to-G transition within the human CTNNB1 locus. Ranking
sgRNAs qualitatively by their potential for desired outcomes (e.g., positioning of target base) versus undesired outcomes (e.g., low activity due to a
poorly accessed PAM) can be a good way to filter which sgRNA sequences to test. The desired outcome (e.g., maximize on-target editing and
minimize detrimental off-target editing) dictates which editing properties to evaluate. In this example, efficient, precision correction is required, which
biases the decision toward sgRNA sequences that have an easily accessible PAM (Guide A = B > C = E > D), minimal Cas-dependent off-targets
(Guide B > D > A > C > E) and optimal positioning of the target adenine (Guide A = C = D > B = E) relative to bystander adenines (Guide C > A = E >
D > B). These considerations may be weighted differently depending on the application, and we generally recommend empirically validating at least
two or three of the most-attractive guides when possible. The gray boxes indicate the opimal editing window. OT, off-target.
base editors have been shown to work in various biological systems84,126, identifying an optimal
combination of sgRNA, deaminase, Cas protein, architecture and delivery method depends on the
specific needs of the target application. For precise allele replacements, there are six types of suitable
base editors that can be categorized into three common use cases: (i) maximized editing efficiency
(standard or high-activity base editors), (ii) minimized bystander editing (narrow-window or context-
specific base editors) and (iii) minimized off-target DNA/RNA editing (reduced DNA off-target and/
or reduced RNA off-target base editors). For broad mutagenesis, expanded-window base editors
enable installation of C-to-T or A-to-G SNPs throughout a target protospacer or locus, while non-
specific mutagenic base editors enable installation of non-specific SNPs throughout a target proto-
spacer or locus. The common classes of base editors and their suggested usages are summarized in
Fig. 5. The subsequent considerations are guidelines to help a user design a base editing strategy and
select base editor components that best suit the desired application.
Identifying candidate protospacers and Cas proteins

The first step in designing a base-editing strategy is to search for suitable protospacers and PAMs to
place the target nucleotide within the base-editing window. Considerations for sgRNA design are

5′ 3′
Target design and BE selection
Timing: 1 d 3′ 5′
Steps 1 & 2
Start with high-activity BE

(evoCDA1/evoAPOBEC1 or TadA-8e/ABE8)
sgRNA P2A-GFP
vector (optional) BE mRNA BE protein
Reagent selection and preparation
Timing: 3–20 d
Or
Steps 3–46 BE Or Or
vector
Chemically modified
sgRNA
Transfect Nucleofect
Or
BE delivery
Timing: 3–4 d
Steps 47–63
Enrich by FACS
(Optional)
Isolate gDNA
Base-editing analysis
Timing: 3 d Ref
Steps 64–87 50%
Sequence
analyze 25%
14%
5%
4%
1%
1%
(Optional)
Refine target and BE selection
for desired outcome
Fig. 4 | Typical workflow and timeline of a base-editing experiment. Suitable target sites and base editors (BEs) are first designed based on the
application. The most promising target sites and corresponding BEs are tested in the biological system of interest. Editing outcomes can be further
enriched by FACS or directly evaluated via high-throughput sequencing (HTS). Assessment of on- and off-target editing frequencies informs
refinement of the guide RNA (gRNA) choice, BE variant choice or delivery method. gDNA, genomic DNA; Ref, reference genomic sequence.
detailed in Box 1 and Fig. 3. Target sgRNAs can be prepared as DNA plasmids containing a human
U6 promoter-driven sgRNA expression construct, or as chemically modified synthetic sgRNAs.
Plasmid expression constructs can be generated using any of several one-step cloning strate-
gies35,36,181. Here, we describe a USER cloning strategy (Step 3A) and a KLD cloning strategy
(Step 3B); the former is recommended for sgRNAs for which a previously generated U6-sgRNA
cassette is not readily available, and the latter is recommended for sgRNAs in which a template
cassette containing the desired tracrRNA is available.
Cas variants that have been used for base editing and their associated PAM specificities can be
found in Fig. 6 and Supplementary Table 1. If the desired outcome is maximum on-target efficiency,
select a Cas variant that places the target edit as close to the center of the editing window as possible.
For precision allele substitutions in which minimized bystander or off-target editing is desired, select
Cas variants that position bystander edits near the edges, or outside the editing window, while
keeping the target edit within the editing window. Importantly, the editing window can be shifted
between Cas variants derived from different bacterial species39,45, which must be considered for

a Target locus
b Target locus
Target nucleotide
Standard base editors High activity base editors Expanded-window base editors
Maximize editing efficiency
(good on-target activity, standard editing windows, (high on-target activity, large editing windows, (editing windows that can span over half of the protospacer)
some sequence context) minimal sequence context, higher off-target editing)
Example uses: commonly used for screening editing sites or for sites without off-target or bystander concerns
Narrow-window base editors Context-specific base editors

Minimize bystander editing
Example uses: localized C-to-G or A-to-G mutations

within a gene for gene knockdown, gene disruption
Bystander base Bystander base Nonspecific mutagenic base editors

(altered base editing repair or dual base
editors resulting in nonspecific mutation)
Example uses: sites with bystander edits that Example uses: sites with bystander edits that cannot be
can be spatially excluded by altering window size spatially excluded but have different sequence contexts
Reduced DNA off-target base editors Reduced RNA off-target base editors
Minimize off-target editing
Off-target locus Off-target transcript
Example uses: biological systems where deamination Example uses: biological systems where transient Example uses: gene diversification within a domain,
at off-target loci is detrimental off-target deamination of RNA is detrimental antibody diversification
c
Key for Table 1 Suggested starting variant(s)*
Base editor type Deaminases Cas variants Architectures
and Figs. 6 and 7 (addgene plasmid ID)
Precise allele substitution
BE2, BE3, BE4, BE4-Gam,

rAPOBEC1, A3A, pmCDA1, etc. Target-AID/C-term, BE4max, BE4max (112093), BE3-FNLS (110836)
Standard base editors FNLS, hyBE4max S
TadA7 variants WT SpCas9 if possible; ABEmax (112095), miniABEmax (131311)
otherwise PAM or broad-window ABE, ABEmax, miniABE/ABE8
variant placing target in
evoAPOBEC1-BE4max (125616),
Activity evolved cytidine deaminases editing window center BE4max, FNLS, hyBE4max
High-activity base editors H evoCDA1-BE4max (125613)
TadA8e/ABE8 variants ABEmax, miniABE/ABE8
ABE8e (138489), ABE8 variants (136295-301)
Increased Km cytidine deaminases, BE4max, Target-AID/C-term, YE1-BE4max (138155), BE3-PAPAPAP (145036),

Narrow-window base editors truncated deaminases (CDA∆, A3A∆) WT SpCas9 if possible; otherwise, FNLS, BE3-PAPAPAP, hyBE4max N SECURE (123614-15)
TadA7.10 F148A PAM variant placing target ABEmax, miniABE/ABE8 ABE7.10-F148A (n/a)
in editing window center and
bystander edits outside or
eA3A, A3Gctd BE4max, FNLS, hyBE4max eA3A-BE5 (152998)
Context-specific base editors near edges of window C
TadA6.3 ABEmax, miniABE/ABE8 ABE6.3 (102916)
YE1-BE4max (138155), SECURE (123614-15),

Increased Km cytidine deaminases High-fidelity SpCas9 variants BE4max, FNLS, eBE3, hyBE4max
Reduced DNA off-target base editors D BE4-PpAPOBEC1/SsAPOBEC3B/etc. (see ref., 138349, etc.)
TadA V106W variants if possible ABEmax, miniABE/ABE8
ABEmaxAW (125647)
Increased Km cytidine deaminases WT SpCas9 if possible; otherwise, YE1-BE4max (138155), SECURE (123614-15),
BE4max, FNLS, hyBE4max
Reduced RNA off-target base editors TadA V106W variants, PAM variant placing target R BE4-PpAPOBEC1/SsAPOBEC3B/etc. (see ref., 138349, etc.)
ABEmax, miniABE/ABE8
SECURE variants in editing window center ABEmaxAW (125647), SECURE ABE (131312, 131313)
Broad mutagenesis of target region
Broad editing window BE-PLUS, BE4max, FNLS, BE-PLUS (113025, 113029),

cytidine deaminases WT SpCas9 (if BE-PLUS),
Expanded window base editors hyBE4max E CP1028/1041-BE4max (119802, 119803)
TadA8e/ABE8 variants broad- window Cas variant
ABEmax, miniABE/ABE8 CP1028/1041-ABE8e (138492, 138493)
WT SpCas9 or broad-window
AID/rAPOBEC1, TadA7.10/SECURE; Cas variant if possible; otherwise,
CRISPRx, TAM CRISPR-X (AID; 85413, 85415, and 85417), TAM (83260)
Nonspecific mutagenic base editors activity evolved deaminases M
PAM variant that accesses STEME, target-ACE Target-ACE (131124), SPACE (140242)
should also be compatible the target locus
*These can be good starting points, but are not necessarily optimal for the target application and all have SpCas9 (NGG PAM). Consult the rest of this table, along with Table 1 and Figs. 6 and 7 for a comprehensive list of viable base editors
and their properties. n/a = not available on Addgene, green = CBE specific, orange = ABE specific.
Fig. 5 | Base editor classes and usage. Base editors are most commonly used for two major types of applications: precise allele substitution (a) and
broad mutagenesis of a target region (b). For precise allele substitution (a), there are six major base editor types, not necessarily mutually exclusive,
that determine editing outcomes. For broad mutagenesis applications (b), two additional types are available. c, Suggested components for each type of
base editor are listed in a table.
target/bystander edit positioning. If Cas-dependent off-target sgRNA binding is a concern and the
available PAM is NGG, consider using high-fidelity SpCas9 variants (SpCas9-HF1/273,94,114,115,117,119,
eSpCas9113,115,119, HypaCas9115,119,182, Sniper-Cas9116,118 and evoCas9119,120) to minimize the chance
of base editing at Cas-dependent off-targets. Enhanced specificity variants of other Cas orthologs have
been described (eSaCas9113, HiFi-Sc++ 81, SpG-HF177 and SpRY-HF177) but have not yet been
validated with base editing. Because PAM availability is less restrictive when using base editors to
perform broad mutagenesis of a target locus, we recommend finding a nearby NGG PAM and using
WT SpCas9 or circularly permuted SpCas9 together with high-activity deaminases, such as evolved
TadA-8e/ABE8106,124 or evoCDA1121, if possible.

Editing window PAM scope
Cas variant (BE4max, (First three non-degenerate PAM bases)b Suggested base editor
2
(aa) ABEmax)a 1 3 AA AC AG AT CA CC CG CT GA GC GG GT TA TC TG TT PAM scope consensus
c
usage (Fig. 4) Notes and refs.
A
D All reported deaminases
S N E 4 8
SpCas9 C (see Table 2 and
G Supplementary Tables
1,368
H C R M 2 and 3)
T
4 8
A
C S N D 20 nt
SpCas9-NG 39, 53, 74, 78, 101–106,
1,368 G 109, 112, 124
H C R M
T
A
C S N D 3 12
SpCas9-NRRH 78
1,368 G
H C R M
T
A
7 11
C S N D
SpCas9-NRTH
78 22 nt
1,368 G
H C R M
T
A
C S N D
SpCas9-NRCH
78 8 13
1,368 G
H C R M
T
A
SpCas9 C S N D 7 14
VQR/VRQR 39, 45, 85, 90, 94, 97, 109
G 23 nt
1,368 H C R M
T
A
C S N D
SpCas9-VRER 39, 45, 97, 109
1,368 G
H C R M
T
A
C S N D
xCas9 39, 75, 94, 102, 109
1,368 G
H C R M
T
A
C S N D
SpG 77
1,368 G
H C R M
T
A
C S N D
SpRY 77
1,368 G
H C R M
T
A
C S E
SpCas9-CP1012 38
1,368 G
H M
T
A
C S E
SpCas9-CP1028 39, 53, 106, 112
1,368 G
H M
T
A
C S E
SpCas9-CP1041
39, 53, 106
1,368 G
H M
T
A
C S E
SpCas9-CP1249
39
1,368 G
H M
T
A
SpCas9
C
HF1/HF2 D 94, 114, 115, 117, 119
G
1,368
T
A
eSpCas9 C
D 115, 119
1,368 G
T
A
HypaCas9 C
D 94, 115, 119
1,368 G
T
A
Sniper-Cas9 C
D 116, 118
1,368 G
T
A CBE not validated
evoCas9 C
D
1,368 G
T 119
A
C S N D
ScCas9
80
1,375 G
H C R M
T
A ABE not validated, CBE
C S N D
ScCas9+ preprint only
1,375 G
H C R M
T 81
A
C S N D
Spymac 82, 100
1,360 G
H C R M
T
A ABE not validated
C S N D
iSpymac
1,360 G
H C R M 82
T
A
S N D E
SaCas9 C
39, 45, 97, 106, 124
1,053 G
H C R M
T
A
S N D E
SaCas9-KKH C
39, 45, 85, 96–97,
1,053 G
H C R M 106, 124
T
A
S N D E
St1Cas9-LMD9 C
108
1,121 G
H C R M
T
A
S N D E
SauriCas9 C
107
1,061 G
H C R M
T
A Only ABE verified, only
C S N D E
CjCas9 one PAM validated
984 G
H C R M
T 110
A 5′ PAM, only compatible

ABE8
C S N D E
LbCas12a with ABE8
1,228 G
H C R M
T 67, 98, 106, 111
A 5′ PAM, ABE not validated
C S N D E
LbCas12a-RVRR
1,228 G
H C R M
T 79

ABE8 S N D E
AsCas12a C with ABE8, not CBE
1,308 G compatible
H C R M
T 106
ABE8 S N D E
enAsCas12a C with ABE8
1,308 G
H C R M
T 86, 106

Fig. 6 | Cas proteins compatible with base editing. aThe editing window symbols refer to the key to the right of the
table. These windows are estimates based on available data from publications and can vary depending on target site
and experimental conditions. The height of the curve reflects expected relative editing efficiencies at each position of
the protospacer. The position of the PAM is highlighted in purple. Green = CBE specific, and orange = ABE specific.
b
Targetable PAMs are highlighted in purple. Partially covered PAMs or non-consensus PAMs that have been
reported but not thoroughly verified for base editing are highlighted in gray. The nucleotide on the left reflects the
first non-degenerate nucleotide, and the nucleotide codes on top indicate the following two. cRegions shaded gray
indicate the positions compared in the PAM scope column. Note that certain PAMs will shift the positioning of the
protospacer. WT, wild-type.
Identifying candidate deaminase domains

Deaminases compatible with base editing are found in Table 1 and Supplementary Tables 2 and 3.
Each deaminase offers a unique combination of editing window, editing sequence preference, off-
target editing and on-target editing properties. For maximal targeting scope or editing efficiency,
deaminases evolved for increased deamination activity and minimal sequence preference should be
used (evoCDA1121, evoFERNY121 and evoAPOBEC1121 for CBE; ABE8106,124 for ABE). If the target
site falls outside the editing window of these high-activity deaminases, alternative deaminases with
even wider editing windows can be considered (LpCDA1L1_361, SsAPOBEC3B122, etc. for CBEs;
currently not possible for ABE). In general, more active deaminases with wider editing windows are
accompanied by increased off-target DNA editing and bystander editing106,121, which are detrimental
in applications requiring highly precise editing. If some undesired edits can be tolerated or can be
avoided with some sequence requirements or slightly smaller editing windows, standard deaminases
(pmCDA154, FERNY121, rAPOBEC135, etc. for CBEs and ABE7.1036 for ABE) can be used to retain
good on-target efficiencies while avoiding some bystander or off-target edits relative to evolved,
sequence-agnostic deaminases. For instance, the strong GC anti-preference of wild-type rAPOBEC1
makes it a better potential candidate than evoAPOBEC1 for target sites that have nearby bystander
edits within GC motifs.
If bystander editing must be minimized, narrow-window deaminases (YE145, AALN112, etc. for
CBE and TadA7.10 F148A125 for ABE) can be used if the target edit is sufficiently far from bystander
edits (>2–3 nt away) such that a narrow editing window can spatially exclude bystander edits. If
bystander edits are difficult to eliminate via careful target/bystander positioning, context-specific
deaminases can be used to selectively modify target nucleotides within a preferred sequence context.
Examples include eA3A, which strongly favors cytosines that occur in a TCR (where R = G or A)
context53,94; A3Gctd, which prefers cytosines that occur in a CC context103; and eAID N51G, which
prefers cytosines that occur in GC contexts104. For adenine base editing, early evolved TadA variants
such as TadA6.3, TadA7.8, TadA7.9 and TadA7.10 exhibit a moderate preference for TadA’s native
YAC (where Y = T or C) context36,183.
To minimize off-target DNA or RNA editing, select deaminases engineered to have reduced
affinity for the undesired type of off-targets45,46,61,103,112,121–123,125,126,131–133. Examples include ABE
V106W variants (TadA-8e V106W, ABE8.17-m-V106W and TadA7.10 V106W)106,124,126 and some
natural (RrA3F, SsAPOBEC3B, AmAPOBEC1 and PpAPOBEC1)122 or engineered (RrA3F F130L122,
SsAPOBEC3B R54Q122, PpAPOBEC H122A122, PpAPOBEC1 R33A122, YE145,112 and SECURE
variants46) cytidine deaminases, all of which have both reduced off-target DNA and RNA editing.
Notably, many deaminases optimized for minimized bystander or off-target editing tend to have
more restrictive targeting requirements (e.g., narrow editing windows resulting in less PAM flex-
ibility, sequence context requirements). Checking that the target site meets these requirements
increases the likelihood of retaining high on-target editing efficiency.
Identifying candidate base editor architectures

Base editor architectures, the arrangement of individual protein components and linkers within a base
editor, can influence base-editing properties. Common architectures that have been described and
their key features are given in Fig. 7 and Supplementary Table 4. For most base-editing applications,
the architectures optimized for nuclear localization, expression and enhanced activity are suitable
(FNLS58, BE4max57, hyBE4max60 for CBE and ABEmax57 or miniABE/ABE8106,124,133 for ABE).
Modulating the stoichiometry or spatial positioning of the deaminase and Cas domains can result in
base editors with further modified editing windows useful for narrow-window or expanded-window
base editors (narrow-window: C-terminal deaminase CBEs54,61 and BE3-PAPAPAP47; expanded-
window: BE-PLUS40), or enable dual A-to-G and C-to-T editing when both cytidine and adenosine

Table 1 | Base-editing deaminases and their properties

Cytidine deaminases
Approximate editing window (SpCas9 editor)b Sequence preferencec

d,e
Editing window Off-target editing Suggested base editor Notes and
Deaminasea 1 20 PAM 5′- N C N -3′ usage (Fig. 5) refs.
5′ -3′
Optimal efficiency Observed A C G T A C G T DNA RNA

4 8 Standard Standard
rAPOBEC1 (rA1) S M 35, also see note f
OTs OTs
1 10 41-42, 53, 94, 112

AID S E M
122, 123, 133
36.1 4 8
S 134
(rA1 F113C/A123E/F205S)
43.1 4 8
S 134
(rA1 G45D/Y75H/F113C/A123E/H166N/F205S)
43.2-rev 4 8
(rA1 F113C/A123E/A165T/F205S) S 134
4 8
rAPOBEC1-H47E/S48A S 53
4 7
Anc689 S 57, 121
4 7
APOBEC3A-Y132D S 123
3 8
APOBEC3A-D131Y S 123
3 8 112, 122, 123

APOBEC3B (A3B) S 128
3 8
A3Bctd S 128-130
–1 9 47, 53-54, 94, 112

pmCDA1 S E
121, 123, 133
eAID 1 9
(AID K10E, T82I, E156G, P182X) S 104
1 9
eAID-N51G S 104
1 8
LpCDA1 S 61
evoAPOBEC1 3 8
(rA1 E4K/H109N/H122L/D124N/ H M 53, 121
R154H/A165S/P201S/F205S)
evoFERNY 3 8
(FERNY H102P/D104N) H M 121
evoCDA1 3 12
(pmCDA1 F23S/A123V/I195F) H E M 53, 121
5 8
eA3A (APOBEC3A N57G) C D 53, 94, 112, 133
5 8
eA3A-T44D/S45A C 53
4 8
APOBEC3A-N57A C 94
4 8
APOBEC3A QF (A3A N57Q/Y130F) C 94
5 7
A3Gctd-D316R/D317R C N 103
YFE 4 6
(rA1 W90Y/Y120F/R126E) N 127
5 7
A3Gctd N 103
47, 109
CDA1Δ190 3 4
see publication for
(pmCDA1 E190X) N
additional ∆ variants
109
APOBEC3AΔ182 4 5
see publication for
(APOBEC3A H182X) N
additional ∆ variants
YE2
(rA1 W90Y/R132E) N 45, 94, 112
EE
(rA1 R126E/R132E) D N 45, 94
YEE 6 45, 47, 94, 98

(rA1 W90Y/R126E/R132E) D N
112
AALN 4 6
D N 112
(rA1 R33A/K34A/H122L/D124N)
4 6
FERNY D N 112, 121
5 7 103, 112, 122-123

APOBEC3G (A3G) D N 128
YE1 5 7 45, 94, 98, 112, 125

(rA1 W90Y/R126E) D N R 131
SECURE 5 6 46, 112, 122, 131,

(rA1 R33A) D N R
also see note f
SECURE 56
(rA1 R33A/K34A) D N R 46, 112, 122, 131
4 7
rAPOBEC1-Y120F D 122, 127
2 10
SsAPOBEC3B-R54Q D 122
2 6
LjCDA1 D 61
7 12
LjCDA1L2_1 D 61
5 9
PpAPOBEC1 D R 122
5 8
PpAPOBEC1-H122A D R 122
6 9
PpAPOBEC1-R33A D R 122
6 12
AmAPOBEC1 D R 122
4 10
RrA3F D R 122
2 10
RrA3F-F130L D E 122
1 13
LpCDA1L1_3 D E 61
2 12
SsAPOBEC3B D E R 122
4 8
APOBEC3A-Y130F R 123, 125, 131
3 9 94, 112, 121-123, 125

APOBEC3A (A3A) R
129-131, 133, 143
8 12 g 111
BEACON1
R
(A3A W98Y/Y132D) Only Cas12a validated
BEACON2 8 12 g 111
R
(A3A W98Y/W104A/Y130F) Only Cas12a validated
3 13
PmCDA1L1_4 E 61
–1 13
LjCDA1L1_4 E 61
1 14
LpCDA1L1_1 E 61
1 15
LjCDA1L1_1 E 61
1 12
LpCDA1L1_4 E 61
Adenosine deaminases
4 8 Standard Standard
TadA7.10 S M 36, 53
OTs OTs
4 8
TadA7.8 S 36
4 8
TadA7.9 S 36
4 7
TadA7.10-D53E S 125
4 8
TadA-8e H E M 106
4 8
TadA-8e-V106W H 106
4 7 124
ABE8.x variants H E M see publication for
additional variants
4 7
ABE8.17-m-V106W H 124
4 9
TadA6.3 C 36
4 5
TadA7.10-F148A N R 125, 133
4 7
ABE-AW (TadA-E59A & TadA7.10-V106W) D R 126
4 7
SECURE ABE (TadA7.10-K20A/R21A) R 133
4 7
SECURE ABE (TadA7.10-V82G) R M 133
Δ, truncated. aMany of the publications reporting these deaminases examined other alternative deaminases. Deaminases that are not well characterized or were not recommended by the original
authors are not reported here or are described in Supplementary Table 3 for clarity. bThe extent to which editing windows have been characterized varies. The editing windows shown here are
estimates based on available data from the original publications and can vary with target site and experimental conditions. Dark regions indicate the editing window resulting in optimal editing
efficiencies, and lighter regions indicate regions where less editing has been observed. cStrong sequence preference or anti-preference is shown in dark green/orange or dark gray, respectively.
Moderate or weak sequence preference or anti-preference is shown in light green/orange or light gray, respectively. No preference is shown in white. Diagonal lines signify preference not
characterized. dOff-target DNA editing characterization indicates Cas-dependent and/or Cas-independent DNA off-targets observed relative to canonical CBE (BE3 with rAPOBEC1) or ABE
(ABE7.10). eLight gray/green arrows indicate a moderate increase or decrease, dark gray/green arrows indicate a strong increase or decrease and a dash indicates no significant change in
off-target editing relative to canonical CBE (BE3 with rAPOBEC1) or ABE (ABE7.10). Diagonal lines indicate that the sequence context and/or off-target editing with these deaminases are not well
characterized. fThese deaminases have recently been characterized to be able to elicit programmable C-to-G transversion edits in certain base-editor architectures (CGBEs)50,51,53. gThe deaminases
BEACON1 and BEACON2 have been validated only with Cas12a, which has a 5′ PAM and a 23-nt protospacer. The approximate editing window is shown relative to these Cas properties.

Suggested base editor usage
Namea Base editor architecture Class Key propertiesb (Fig. 5) Notes and refs.
16 aa 4 aa
BE3 Canonical CBE architecture S 35, also see 109
CBE
16 aa 4 aa Reduced indels, reduced editing efficiency relative to
BE2 S 35
BE3, improved product purity relative to dCas base editor
CBE
32 aa 10 aa 10 aa
BE4 IImproved product purity and editing efficiency relative to BE3 S 55
CBE
16 aa 32 aa 10 aa 10 aa S
BE4-Gam Gam Reduced indels relative to BE4 55
CBE Improved base editor expression and editing efficiency

BE4max 32 aa 10 aa 10 aa S H N C D R E 57
relative to BE4
32 aa 4 aa Improved base editor expression and editing efficiency 58

FNLS CBE S H N C D R E
relative to BE3
32 aa 10 aa 10 aa 57
BE-p2A-GFP p2A EGFP CBE Enabled flow sorting of CBE-expressing cells S H N C D R E
16 aa 4 aa 58
FNLS-p2A-Puro p2A PuroR CBE Enabled selection of base editor–expressing cells S H N C D R E
7 aa 4 aa 47
BE3-PAPAPAP CBE Greatly narrowed editing window relative to BE3 N
-PAPAPAP-
CBE CBE
100 aa
Target-AID/C-terminal CBE Enables editing by certain deaminases; exhibits altered S N 54, 61
base editor editing windows relative to N-terminal deaminase fusion
30 aa
Broad mutagenesis, increase in C-to-non-T 42, 43
CRISPR-X CBE M
MS2 edits relative to BE3
47 aa
TAM CBE Broad mutagenesis and C-to-non-T edits relative to BE3 M 41
9x 22 aa
27 aa
GCN4 GCN4 Greatly enlarged editing window size relative to BE3
BE-PLUS CBE E 40
16 aa 5 aa 18 aa
scFv GB1
3x
eBE3 16 aa 4 aa CBE Improved product purity, reduced indels relative to BE3 S D 59
p2A
32 aa 32 aa 10 aa 10 aa D R E 60
hyBE4max DBD Improved editing efficiency relative to BE4max S H N C
32 aa
N-term NpuN
Split CBE CFN CBE Enabled packaging of cytosine base editor constructs into AAV S 66, 67, 70
10 aa
NpuC C-term
32 aa 32 aa 36
ABE ABE Canonical ABE architecture S
*
32 aa 32 aa 57
ABEmax ABE Increased expression and editing efficiency relative to ABE S H N C D R E
*
32 aa 32 aa 57
ABE-p2A-GFP p2A EGFP ABE Enabled flow sorting of ABE expressing cells S H N C D R E
*
ABE
32 aa Reduced off-target RNA editing relative to canonical ABE, but 106, 124, 133
miniABE/ABE8 ABE S H N C D R E
* reduced editing efficiency except when using TadA-8 variants
32 aa 32 aa
N-term NpuN
Split ABE CFN
* Enabled packaging of adenine base editor constructs into AAV S 66, 69
NpuC C-term
32 aa 32 aa 93 aa SPACE only contains

Target-ACE/SPACE Both Dual C-to-T and A-to-G editing M TadA* monomer
* 63, 64
BOTH
48 aa 32 aa 32 aa
STEME/A&C-BE Both
* Dual C-to-T and A-to-G editing M 62, 65
16/32 aa Reduced indels but also reduced editing relative to

dCas base editor Deaminase Both S 35, 36
BE3/ABE7.10
nCas variant UGI NLS

* Cytidine deaminase dCas variant
Codon optimized
TadA TadA
(WT) (evolved)
Fig. 7 | Common base editor architectures and their key properties. aArchitecture names reported here reflect the nomenclature given when first
reported or as commonly used. Linker lengths may vary slightly between versions. bKey properties represent distinguishing reasons for selecting that
specific architecture, as described in the original publication or subsequent studies. CFN, 3-aa extein motif required by Npu-intein for splicing; DBD,
single-stranded DNA binding domain from Rad51; dCas, catalytically dead Cas variant; GB1, solubility tag, a binding domain of protein G from group G
Streptococcus; NpuC, C-intein from Nostoc punctiforme; NpuN, N-intein from N. punctiforme; PuroR, puromycin resistance cassette; scFv, single-chain
variable-fragment antibody recognizing GCN4.
deaminases are fused (Target-ACE63/SPACE64, STEME62/A&C-BE65). For CBEs, adding or removing

UGI domains results in unique architectures that either have greatly improved product purity (BE4/
BE4max35 and eBE359) or can be used for non-specific mutagenesis (TAM41 and CRISPR-X42). For
ABEs, using the monomeric form of evolved TadA106,124,133 results in smaller ABE constructs that
may be amenable to single-AAV packaging or have reduced off-target RNA editing. Finally, addi-
tional components can be added to allow for downstream selection or visualization of base
editor–treated cells or tissues (adding a P2A-GFP57 or P2A-selection marker tag58).
Selected Cas proteins, deaminases and base editor architectures can have synergistic properties.
For instance, a narrow-editing-window deaminase (e.g., YE1) can be coupled with an expanded-
editing-window Cas protein (e.g., SpCas9-CP1028) to achieve an editing window size similar to that
of a standard CBE112. The resulting editor (YE1-CP1028) has fewer targeting restrictions than the

corresponding YE1-SpCas9 base editor but retains the desirable minimized off-target DNA editing
exhibited by the deaminase112. Carefully considering the interactions between base editor compo-
nents will result in an increased likelihood of achieving optimal desired editing outcomes while
minimizing undesired editing effects. A website for this protocol can be found at https://be-keeper.
broadinstitute.org and assists in selecting and evaluating candidate base editor components. If the
selected base editor is one of the 10 variants (ABEmax, BE4max, ABEmax-CP1041, BE4max-CP1028,
EA-BE4max, eA3A-BE4max, CDA-BE4max, AID-BE4max, evoAPOBEC1-BE4max or eA3A-BE5)
that has been thoroughly characterized via high-throughput, machine-learning analysis, the BE-Hive
webtool (https://www.crisprbehive.design)53 can be used to predict editing outcomes.
Identifying candidate delivery strategies

Different cell types are amenable to different delivery methods for genome-editing agents. Plasmid
DNA delivery is the simplest and most affordable approach to delivering these agents when using
readily transfected cell types. Many base editor plasmids are readily available through Addgene
(Supplementary Table 5). If the chosen cell type is amenable to lipofection or electroporation of
plasmid DNA and can robustly express encoded proteins, this delivery method is recommended first
(Step 52A). However, plasmid delivery leads to a longer exposure time to the base editor, which can
cause increased levels of undesired off-target editing. More transient delivery methods, including base
editor mRNA (Step 52B) or protein (Steps 4–46 and 52C) delivery, can be used to reduce off-target
editing while retaining high on-target efficiencies78,112,114,132,142,149,179,184. In addition to the delivery
methods described in this protocol, lentivirus, adenovirus and AAV have also been used to deliver
base editors to cells that do not efficiently take up and express plasmid DNA58,66,67,69,70,88.
Selection or enrichment of base-edited cells

If an edit is inefficient or if the cell line used is not readily amenable to the chosen delivery strategy,
FACS may be used to enrich base editor–treated cells (Steps 53–63). Cells that more strongly express
the base editor tend to be the most efficiently edited (Fig. 8d,e)57,58. To FACS-enrich, editor DNA or
mRNA should be tagged with a P2A self-cleaving peptide followed by a fluorescent gene (usually
GFP). After delivery of the editor and 12–72 h of incubation, FACS may be conducted to collect
fluorescent cells, which can then be processed for downstream analysis. Notably, this strategy can also
be used to isolate clonal alleles176. A higher gating stringency can be used to increase the likelihood of
isolating homozygous edited genotypes. Isolated cells can then be expanded into new cell lines if
desired. A common alternative method, not described in this protocol, is to use a selectable antibiotic
resistance marker to enrich for base editor–expressing cells58. Cells are transduced with lentivirus
containing a base editor construct linked to an antibiotic resistance cassette (e.g., puroR for pur-
omycin resistance) by a P2A self-cleaving peptide58,63. Starting 2 d after transduction, treated cells are
subjected to antibiotic selection for several days.
Analysis of base-editing outcomes

We recommend using next-generation sequencing to assess genome editing efficiency (Steps 64–87)
because it is the most sensitive method to quantify precise allele outcomes and indels. Sample
preparation involves PCR to amplify a region around the intended target nucleotide or nucleotides
from edited gDNA. Primers to amplify this region are designed to minimize off-target or biased
amplification and can be designed using NCBI’s Primer-BLAST tool (https://www.ncbi.nlm.nih.gov/
tools/primer-blast/).
In general, we recommend an amplicon length between 150 and 300 bp centered around the target
base, as shorter amplicons or asymmetric amplicons can lead to false-negative detection of longer
indels or low-quality reads if the target site is too close to the ends of the amplicon (<20 bp away)185.
Longer amplicons can also be designed following the same principles but usually require more costly
sequencing kits.
Required sequencing coverage can vary depending on library preparation, variant allele frequency
and sequencing error rates186. For single amplicon sequencing, we find that 10,000-read coverage on
an Illumina MiSeq is suitable for most base-editing experiments. Fewer reads can also be used with
increasingly poorer sensitivity of the measurement. The amount of gDNA template used for PCR
amplification also limits the sensitivity of the measurement. Ideally, an excess of template genomes
should be amplified with respect to the final number of measured sequencing reads.
Sequenced base-editing outcomes can be analyzed using various software packages. User-friendly
online tools including BE-Analyzer187 (https://rgenome.net/be-analyzer/) or CRISPResso2188 (https://

a c
HEK293T untransfected control population
Sample allele frequency table: Site 1 BE4max-treated, all GFP+ sorted
Gate A: Gating on population Gate B: doublet exclusion Gate C: GFP+ cells
2.0 M 2.0 M 2.0 M PAM Target protospacer (RC)
5′ 3′ Ref. Sequence
SSC-A :: SSC-A
FSC-H :: FSC-H
SSC-A :: SSC-A
1.5 M 1.5 M 1.5 M
69.64% (20,267 reads)
6.36% (1,850 reads)
4.44% (1,292 reads)
1.0 M 1.0 M 1.0 M C: 0.0% 3.65% (1,063 reads)
3.38% (985 reads)
B: 86.3% 1.81% (527 reads)
500 K 500 K 500 K 0.67% (195 reads)
0.63% (184 reads)
A: 86.2% 0.39% (113 reads)
0.38% (112 reads)
0.37% (108 reads)
500 K 1.0 M 1.5 M 2.0 M 500 K 1.0 M 1.5 M 2.0 M 101 102 103 104 105 106 107 0.33% (97 reads)
FSC-A :: FSC-A FSC-A :: FSC-A FL2-A :: FITC-A 0.32% (92 reads)
0.29% (84 reads)
0.25% (72 reads)
0.24% (70 reads)
Gate D (optional): top GFP+ cells 0.21% (60 reads)
2.0 M
Expected nick/cut site Optimal editing Allele frequency
Population Count % of parent % of total Mean: FITC-A
window
SSC-A :: SSC-A
1.5 M All events 20,061 100 100 n/a

Population sorting 17,288 86.2 86.2 n/a
D: 0.0% n/a
1.0 M Singlets 14,927 86.3 74.4
GFP positive 0 0 0
d 100 BE4max unsorted Untreated
e 100 ABEmax unsorted Untreated
ABEmax, all GFP+ sorted
Top 10% GFP 0 0 0 BE4max, all GFP+ sorted
500 K
% of total sequencing reads with
% of total sequencing reads with

80 80
C•G converted toT•A
A•T converted to G•C

101 102 103 104 105 106 107
FL2-A :: FITC-A 60 60
40 40
b HEK293T ABEmax-p2A-GFP site 1 population
Gate A: gating on population Gate B: doublet exclusion Gate C: GFP+ cells
2.0 M 2.0 M 2.0 M 20 20
SSC-A :: SSC-A
FSC-H :: FSC-H
SSC-A :: SSC-A
1.5 M 1.5 M 1.5 M 0 0

C3 C7 C8 C6 C7 C4 C6 A4 A5 A6 A9 A5 A8 A9 A3 A5 A7 A8 A9
1.0 M 1.0 M 1.0 M C: 67.1
Site 1 Site 2 Site 3 Site 1 Site 2 Site 3
500 K 500 K B: 76.5 500 K

A: 83.3
f g 600
500 K 1.0 M 1.5 M 2.0 M 500 K 1.0 M 1.5 M 2.0 M 101 102 103 104 105 106 107
% of total sequencing reads
100
with C•G converted to T•A
FSC-A :: FSC-A FSC-A :: FSC-A FL2-A :: FITC-A
On:off-target R-loop
400
80
editing ratio
Gate D (optional): top GFP+ cells
2.0 M 60
200
Population Count % of parent % of total Mean: FITC-A 40
SSC-A :: SSC-A
1.5 M All event s 20,000 100 100 n/a

Population sorting 16,653 83.3 83.3 n/a 20 10
D: 10.4 5
1.0 M Singlets 12,741 76.5 63.7 n/a
0 0
GFP positive 8,543 67.1 42.7 2.19 × 105 DNA mRNA Protein DNA mRNA Protein
500 K
Top 10% GFP 888 10.4 4.44 8.85 × 105 Delivery method
Delivery method
101 102 103 104 105 106 107

FL2-A :: FITC-A
Fig. 8 | Anticipated results for a typical base-editing transfection experiment in HEK293T cells with FACS sorting. a and b, FACS gating on a
negative population (a) and a treated condition (b) to enable GFP-positive sorting of cells treated with base editor–P2A–GFP. Cells can be sorted for all
GFP-positive cells (blue gate) or for a population of the highest GFP-expressing cells (orange gate) to enrich for the likelihood of base-editing
outcomes. If no difference in GFP signal is observed between the untreated and treated samples, first ensure that the GFP coding sequence is correct
via sequencing and then check that the amount of base editor being delivered is not too cytotoxic. FACS plots shown in a and b reflect sorting on a
population of cells from a single biological replicate. c, Example allele frequency table resulting from CRISPResso2 analysis; each line represents a
unique genotype along with its frequency. Note that for this amplicon, the 5′-to-3′ sequence of the strand bound by the sgRNA (reverse complement
of the strand containing the target base) is shown. Bases that are edited relative to the reference amplicon are shown in bold text. In the target site
shown, there are two cytosines (C7 and C8, in which the number reflects the position within the protospacer, counting the PAM as positions 21–23)
that fall within the optimal editing window of BE4max (positions 4–8, gray box). This site is efficiently edited (>50% C-to-T conversion), but both C7
and C8 are edited, as well as C3 to an extent. For screening purposes, such an outcome would probably be a successful experiment; however, for
applications in which selective editing of one of the cytosines is required, additional refinement would be necessary. d, Bulk population C•G-to-T•A
conversion frequency at three genomic sites in HEK293T cells treated with BE4max, unsorted or sorted for all GFP-positive cells. Edited cytosines are
shown by their position within the protospacer (counting the PAM as positions 21–23). e, Bulk population A•T-to-G•C conversion frequency at three
genomic sites in HEK293T cells treated with ABEmax, unsorted or sorted for all GFP-positive cells. Edited adenines are shown by their position within
the protospacer (counting the PAM as positions 21–23). Values in d and e reflect a single biological replicate. At genomic sites amenable to base
editing, current base editors can achieve 50% efficiency at cytosines or adenines that fall within the optimal editing window by transfection, and can be
further enriched to nearly 100% base conversion when treated cells are sorted for cells that express the base editor. Although the data shown in d and
e (Supplementary Data 2) were uniquely collected for this protocol, both the genomic sites39 and the FACS sorting process57 have been validated and
previously reported. f and g, Editing comparison of cells treated with BE4max plasmid DNA and mRNA, and BE4max protein, at one genomic locus in
HEK293T cells and their on:off-target editing ratios as determined by an orthogonal R-loop assay112. Values and error bars reflect the means and s.e.m.
of three independent biological replicates performed on different days. Although the data shown in f and g (Supplementary Data 3) were uniquely
collected for this protocol, a similar experiment was reported previously112. FITC-A, FITC area; FL2-A, FL2 area; FSC-A, forward scatter area; K,
thousand; M, million; RC, reverse complement; SSC-A, side scatter area.
crispresso.pinellolab.partners.org) can be used to rapidly process single sequencing files. For higher-
throughput analysis, we have found that the command-line, batch functionality of CRISPResso2
(CRISPRessoBatch) provides rapid, efficient data processing for grouped samples, allowing direct

comparison of different sample treatments at the same amplicon. CRISPRessoBatch provides plots for
indel analysis, base editing efficiency and allelic frequencies that can be directly used in addition to
raw data files that can be reprocessed. Alternatively, Sanger sequencing of PCR-amplified genomic
targets can be a fast and affordable way to roughly estimate population-level editing efficiency. The
EditR189 webtool (http://baseeditr.com) may be used to calculate editing from Sanger sequence
chromatograms.
As discussed previously, one limitation of base editing is potential Cas-dependent or Cas-
independent editing at undesired genomic loci. Prudent selection of sgRNAs and base editor reagents
can reduce the frequency of off-target editing (Box 1 and Fig. 3). For some applications such as
therapeutic editing, however, off-target editing should be carefully assessed. Cas-dependent off-target
edits can be reasonably approximated with existing Cas off-target characterization methods, including
GUIDE-seq190, CIRCLE-seq191 and Digenome-Seq192, among others. In general, the process requires
identifying candidate off-target loci targeted by the corresponding Cas enzyme in its nuclease form
for the chosen sgRNA, and then sequencing these candidate off-target loci to detect potential
off-target base editing after treatment of the target site with base editors. Cas-independent off-target
editing can be assessed by whole-genome sequencing or, more efficiently and cost effectively, by using
recently developed tools such as an orthogonal R-loop assay that measures the frequency of
Cas-independent (unguided) DNA deamination in the ssDNA regions of R-loops created by
orthogonal dead Cas proteins and their guide RNAs112,122. As this protocol largely focuses on initial
design and execution of base editor experiments, we refer readers interested in downstream off-target
characterization to relevant publications cited above for more information on those methods.
Considerations for experimental controls

There are several controls that should be included when executing a base-editing experiment. When
delivering base editors to a target mammalian cell culture, it is important to include untreated
samples (e.g., cells grown without delivered base editor), because many cell lines contain hetero-
geneous SNPs that can confound base-editing analysis if not identified. At a minimum, untreated cells
should be plated alongside cells that will be treated. Note that for typical (n = 3) biological replicates,
there should be three untreated samples corresponding to the cells used in each replicate of the
editing experiment.
For a more rigorous control, untreated samples should be treated with the delivery vehicle only, or
the delivery vehicle containing base editor and a non-targeting sgRNA or no sgRNA. In the former
condition, effects resulting from the delivery vehicle, such as lipofection or AAV toxicity, can be
evaluated. In the latter condition, guide-independent editing can be assessed.
If flow cytometry sorting is used to enrich for base-editing outcomes, an untreated or vehicle-
treated sample should be included to determine the sorting gates necessary for the population of
interest. When treated cells are ready to be analyzed for base-editing outcomes, ensure that the
untreated samples are sequenced at the same loci that were treated with base editor. Base editing in
treated samples should then be normalized against SNPs or apparent editing that appear in the
untreated condition.
Materials
Reagents
sgRNA and base editor preparation
● Plasmids: SpCas9 sgRNA (pFYF1320, Addgene ID: 47511); canonical BE4max (Addgene ID: 112093)
or canonical ABEmax (Addgene ID: 112010); for other base editor constructs available on Addgene,
see Supplementary Table 5
● PCR primers for sgRNA cloning, sequencing and amplifying a template for mRNA transcription can
be designed and ordered as shown in Table 2 from Integrated DNA Technologies (IDT) or another
preferred vendor. Alternatively, chemically modified, custom sgRNAs can be ordered from IDT,
Trilink Biotechnologies, Synthego, Agilent, BioSpring or other vendors CRITICAL Note that the 3′
c
ends of the primers for amplifying a template for mRNA transcription are variable and should anneal
to desired 5′ and 3′ untranslated regions (UTRs) cloned into the base editor plasmid.
●
Chemically modified, custom base editor mRNA can be ordered from Trilink Biotechnologies. We
recommend Cap1 (CleanCap), N1-methyl pseudouridine or 5-methoxyuridine substituted in place of
uridine and a 120-nt polyA tail. For applications in mammalian cells, Trilink proprietary mammalian
UTR sequences can be used. The product concentration can be normalized to 2 µg/µl.

Table 2 | Example primer sequences for sgRNA cloning and HTS sequencing
Step Primer name Sequence (5′–3′) Purpose Primer usage
3A(iii) USER-Fw AGAGCTAGAAA/deoxyU/ USER cloning a user- Anneals to the tracrRNA of SpCas9 in pFYF1320.
AGCAAGTTAAAATAAGG defined sgRNA sequence In cases where a Cas ortholog is used, the 3′ end
into pFYF1320 can be extended to include the entire desired
tracrRNA sequence and the annealing region
shifted into the backbone of pFYF1320. Note that
doing so requires the USER junction to be
changed for both USER-Fw and USER-Rv primers.
The USER junction is shown in italics.
USER-Rv ATTTCTAGCTC/deoxyU/ USER cloning a user- Replace N20–22 with the reverse complement of
AAAAC(N20–22) defined sgRNA sequence the target sgRNA sequence. The USER junction is
GGTGTTTCGTCCTTTCCAC into pFYF1320 shown in italics and the annealing region is
underlined.
3B(i) KLD-Fw (N20–22) GTTTTAGAGCTAG KLD cloning a user-defined Replace N20–22 with the target sgRNA sequence.
AAATAGCA sgRNA sequence into The underlined sequence anneals to the
AGTTAAAATAAG pFYF1320 tracrRNA of SpCas9 and can be replaced
accordingly depending on the Cas variant and
template plasmid used.
KLD-Rv GGTGTTTCGT KLD cloning a user-defined Anneals to the U6 promoter of pFYF1320 and
CCTTTCCACAAG sgRNA sequence into can be used for cloning an sgRNA plasmid for
pFYF1320 any Cas ortholog
4 BE-protein-Fw AGCGAAT/deoxyU/ Amplifying a desired base Anneals to the N terminus of the base editor
TGAAAGCCCG editor fragment for USER (see Supplementary Data 1). The underlined
AAAAAAAAGCGT cloning into a protein- annealing region can be replaced with a
AAAGTTAGCGAAGTTGA expression vector sequence that binds to the N-terminal sequence
ATTTAGCCAC of the desired base editor to be inserted. The
USER junction is shown in italics.
BE-protein-Rv ATTCAAATTC/deoxyU/ Amplifying a desired base Anneals to the C terminus of the base editor (see
GATCCATCGG editor fragment for USER Supplementary Data 1). The underlined annealing
CTGTGCGTTTC cloning into a protein- region can be replaced with a sequence that
GAACCACCGCTGTCAC expression vector binds to the C-terminal sequence of the desired
base editor to be inserted. The USER junction is
shown in italics.
BacteriaExpressBB-Fw AGAATTTGAA/deoxyU/ Amplifying the plasmid Anneals to the C-terminal NLS 3′ of the base
CTCCGAAAAA backbone for USER cloning editor insert (see Supplementary Data 1). The
GAAACGCAAG a base editor protein- USER junction (shared with BE-protein-Rv) is
expression vector shown in italics.
BacteriaExpressBB-Rv AATTCGC/deoxyU/ Amplifying the plasmid Anneals to N-terminal NLS 5′ of the base editor
ACCATCTGCGGTACGTTTG backbone for USER cloning insert (see Supplementary Data 1). The USER
a base editor protein- junction (shared with BE-protein-Fw) is shown in
expression vector italics.
52B(i) mRNA-F TAATACGACTCACTATAGGG PCR amplification of a Includes a T7 promoter (italics). (NAnneal) should
(NAnneal) template for in vitro be replaced with a sequence that anneals to the
transcription of desired 5′ UTR sequence cloned into the plasmid
editor mRNA of interest.
52B(i) mRNA-R TTTTTTTTTTTTTTTTTT PCR amplification of a Includes a 120-nt poly A tail. (NAnneal) should be
TTTTTTTTTTTT template for in vitro replaced with a sequence that anneals to the
TTTTTTTTTTTTTTTT transcription of desired 3′ UTR sequence cloned into the plasmid
TTTTTTTTTTTTTT (NAnneal) editor mRNA of interest.
69 PCR1-Fw ACACTCTTTCCCTACACGAC Fw primer for amplifying Replace NAnneal with a sequence matching the 5′
GCTCTTCCGATCTNNNN gDNA in preparation end of the targeted genomic site (for a 200-bp
(NAnneal) for HTS amplicon, this primer should typically anneal to
the genomic region ~100 nt 5′ of the target base
edit). The Illumina PCR1 Fw adapter is shown in
italics.
PCR1-Rv TGGAGTTCA Rv primer for amplifying Replace NAnneal with a sequence matching the
GACGTGTGCTCTTCCGAT gDNA in preparation reverse complement of the 3′ end of the targeted
(NAnneal) for HTS genomic site (for a 200-bp amplicon, this primer
should typically anneal to the genomic region
~100 nt 3′ of the target base edit). The Illumina
PCR1 Rv adapter is shown in italics.
Fw, forward; Rv, reverse.

CRITICAL HPLC purification of mRNAs did not help base editor activity in our hands.
c
● Alternately, mRNA can be transcribed in the laboratory using a kit such as HiScribe T7 ARCA (NEB,
cat. no. E2065S).
● Nuclease-free water (Qiagen, cat. no. 129115)
● PhusionU Green Multiplex PCR Master Mix, 2× (Thermo Fisher Scientific, cat. no. F564S/F564L)
CRITICAL For USER cloning with uracil-containing primers, a uracil-tolerant polymerase such as
c
this must be used. However, for KLD cloning purposes, any alternative high-fidelity polymerase can
be used.
● SeaKem LE Agarose (Lonza, cat. no. 50004)
● Ethidium bromide solution, 10 mg/ml (Millipore Sigma, cat. no. E1510-10ML)
● UltraPure TAE Buffer, 10× (Thermo Fisher Scientific, cat. no. 15558026)
●
Quick-Load Purple 1 kb Plus DNA Ladder (NEB, cat. no. N0550S/N0550L)
● QIAquick PCR purification kit (Qiagen, cat. no. 28104)
● One Shot Mach1 T1 Phage-Resistant Chemically Competent Escherichia coli (Thermo Fisher, cat. no.
C862003) or preferred cloning strain

● S.O.C. medium (Thermo Fisher Scientific, cat. no. 15544034)
● LB medium (United States Biological, cat. no. L1505)
● LB agar medium (Millipore Sigma, cat. no. L2897)
● Carbenicillin, 50 mg/ml, sterile filtered (Gold Biotechnology, cat. no. C-103)
● Illustra TempliPhi 100 Amplification Kit (Cytiva, cat. no. 25640010)
● Qiagen Plasmid Plus Midi Kit (Qiagen, cat. no. 12945)
● TE Buffer, 1× (Thermo Fisher Scientific, cat. no. 12090015)
USER cloning
● USER enzyme (NEB, cat. no. M5505/M5505L). Cutsmart Buffer, 10× is provided with the product, but
can also be ordered separately (NEB, cat. no. B7204S).

● Dpn1 enzyme (NEB, cat. no. R0176S/R0176L)
KLD cloning
● KLD enzyme mix (NEB, cat. no. M0554S)
Protein purification
● Vector for protein purification: pD881-SR (Atum, cat. no. FPB-27E-269)
● One Shot BL21 Star (DE3) Chemically Competent E. coli (Thermo Fisher Scientific, cat. no. C601003)
or preferred protein-expression strain

● Luer-lock syringe for sterile filtration (VWR, cat. no. 89215-218)
● Syringe filter unit, 0.22 µm, PVDF (Millipore Sigma, cat. no. SLGVM33RS)
● Kanamycin, 25 mg/ml, sterile filtered (Gold Biotechnology, cat. no. K-120)
● Terrific broth medium, dissolved in distilled water and autoclaved per the manufacturer’s instructions
(Thermo Fisher Scientific, cat. no. 22711022)

● L-Rhamnose, 30% (wt/vol) stock, sterile filtered (Gold Biotechnology, cat. no. R-105-250)
● Liquid nitrogen
●
Sodium chloride, 1 M solution (Millipore Sigma, cat. no. S9888)
● Tris-HCl, pH 8.0, 1 M solution (Thermo Fisher Scientific, cat. no. 15568025)
● Tris-(2-carboxyethyl)phosphine (TCEP), powdered (Millipore Sigma, cat. no. C4706)
● Glycerol, 20% (vol/vol) solution (Millipore Sigma, cat. no. G5516)
● cOmplete, EDTA-free Protease Inhibitor Cocktail (Millipore Sigma, cat. no. 04693132001)
● DNase I (Thermo Fisher Scientific, cat. no. 90083)
● Nickel-nitrilotriacetic acid (Ni-NTA) Resin (G-Biosciences, cat. no. 786-940)
● Imidazole, 3 M solution, sterile filtered (Millipore Sigma, cat. no. I5513)
● NuPAGE 4–12%, Bis-Tris, 1.0 mm, Mini Protein Gel, 10-well (Thermo Fisher Scientific,
cat. no. NP0321BOX)
● NuPAGE LDS Sample Buffer (4×) (Thermo Fisher Scientific, cat. no. NP0007)
●
Precision Plus Protein Dual Color Standards (Bio-rad, cat. no. 1610374)
● NuPAGE MOPS SDS Running Buffer (20×) (Thermo Fisher Scientific, cat. no. NP0001)
● InstantBlue protein stain (Millipore Sigma, cat. no. ISB1L)
● Pierce bicinchoninic acid (BCA) Protein Assay Kit (Thermo Fisher Scientific, cat. no. 23225)

Mammalian cell culture
● HEK293T cell line (ATCC, cat. no. CRL-3216; RRID: CVCL_0063) ! CAUTION Cell lines used for
base-editing experiments should be regularly tested for mycoplasma to ensure that they are authentic.
● DMEM, high glucose, GlutaMAX Supplement (Thermo Fisher Scientific, cat. no. 10566016; phenol-red
free: 21063029)
CRITICAL FBS should be
c
● Fetal bovine serum (FBS) (Thermo Fisher Scientific, cat. no. 16000044)
divided into aliquots and frozen at −20 °C if not in use for culture medium.
● PBS, pH 7.4 (1×) (Thermo Fisher Scientific, cat. no. 10010023)
● TrypLE Express Enzyme (1×), phenol red (Thermo Fisher Scientific, cat. no. 12605036; phenol-red
free: 12604021)
● Lipofectamine 2000 Transfection Reagent (Thermo Fisher Scientific, cat. no. 11668019) or
Lipofectamine 3000 Transfection Reagent (Thermo Fisher Scientific, cat. no. L3000015)
● Opti-MEM Reduced Serum Medium (Thermo Fisher Scientific, cat. no. 31985070)
● GFP transfection marker: pmaxGFP (provided in Lonza Nucleofector kits such as SF Cell Line
4D-Nucleofector X Kit S; Lonza, cat. no. V4XC-2032)

● Trypan blue stain (0.4% (wt/vol)) (Thermo Fisher Scientific, cat. no. T10282)
● Proteinase K (NEB, cat. no. P8107S)
● Tris-HCl, pH 7.5, 1 M solution (Thermo Fisher Scientific, cat. no. 15567027)
● SDS, 10% (wt/vol) solution (Thermo Fisher Scientific, cat. no. 15553027)
● SF Cell Line 4D-Nucleofector X Kit S, for electroporation of editors as mRNA or protein (Lonza, cat.
no. V4XC-2032)
Next-generation sequencing analysis

● Q5 Hot Start High-Fidelity Mastermix, 2× (NEB, cat. no. M0494S)
● QIAquick Gel Extraction Kit (Qiagen, cat. no. 28704)
● Qubit double-stranded DNA High-Sensitivity Assay Kit (Thermo Fisher Scientific, cat. no. Q32854) or
KAPA Library Quantification Kit (Roche, cat. no. 07960140001)

● MiSeq Reagent Kit v2 (300-cycles) (Illumina, cat. no. MS-103-1002—Micro kit, ~4 million reads; MS-
102-2002—standard kit, ~15 million reads)
Equipment
●
Filtered sterile pipette tips (e.g., Biotix, Rainin, VWR)
● Serological pipettes, assorted (Corning)
● Standard microcentrifuge tubes, 1.5 ml (Neptune Scientific, cat. no. 4445.X)
● Standard PCR tube strips, 8 tubes/strip, 0.2 ml (Corning, cat. no. PCR-0208-C)
● Standard PCR 1 × 8 strip caps, for 0.2-ml PCR tubes (Corning, cat. no. PCR-2CP-RT-C)
● Falcon centrifuge tubes, polypropylene, 15 ml (VWR, cat. no. 62406-200)
● Falcon centrifuge tubes, polypropylene, 50 ml (VWR, cat. no. 21008-940)
● Corning 50-ml Mini Bioreactor (Corning, cat. no. 431720)
● Corning vacuum filter/storage bottle system, 0.22-µm pore, 33.2 cm2 polyethersulfone (PES)
membrane (Corning, cat. no. 431097)
● 8-tube PCR strips, white for qPCR (Bio-rad, cat. no. TLS0851)
●
Flat PCR tube 8-cap strips, optical, ultraclear (Bio-rad, cat. no. TCS0803)
● VWR 96-well PCR plate (VWR, cat. no.: 89218-296)
● Hard-shell 96-well PCR plates, for qPCR (Bio-rad, cat. no. HSP9655)
● Microseal ‘F’ PCR plate seal, foil (Bio-rad, cat. no. MSF1001)
● PCR plate heat seal, clear, optical, for qPCR reactions (Bio-rad, cat. no. 1814030)
● Plastic inoculating loops, 10 µl (Copan, cat. no. COP-S10)
● Non-tissue culture–treated bacteriological Petri dish, 100 × 15 mm (VWR, cat. no. 470210-568)
● Poly-D-lysine or collagen 96- or 48-well clear flat-bottom TC-treated microplates with lids (Corning,
cat. no. 354461 for 96-well, poly-D-lysine; 354407 for 96-well, collagen)
● Falcon TC-treated cell culture flask with vented cap, 75 cm
2
(Corning, cat. no. 353136)
● Countess cell-counting chamber slides (Thermo Fisher Scientific, cat. no. C10314)
●
Light microscope with filters for fluorescence (Zeiss Axio Observer or comparable system)
● Gel casting system (Bio-rad, cat. no. 1704412—caster; and Bio-rad, cat. no. 1704416—gel tray)
● Gel electrophoresis system (Bio-rad, cat. no. 1704401)
● Mini gel tank for protein gels (Thermo Fisher Scientific, cat. no. A25977)

●
Power supply for gel electrophoresis (Bio-rad, cat. no. 1645070)
● PCR thermocycler with 48- and/or 96-well heating blocks (Bio-rad, cat. no. 1851197)
● Real-time PCR detection system (e.g., Bio-rad CFX96 or comparable system)
● Benchtop microcentrifuge (Eppendorf, cat. no. 5405000441)
● Tabletop centrifuge with rotor fitting 50-ml conical tubes (Eppendorf, cat. no. 022623508 or
comparable system)
● Floor centrifuge with rotors for 1-liter and 50-ml volumes (Sorvall LYNX 4000 or comparable system
specified to reach ≥20,000g)

● Centrifuge bottles with sealing closure, 1 liter (Fisher Scientific, cat. no. 05-579-13)
● Shaker Erlenmeyer flasks with baffles, 4 liters (VWR, cat. no. 32645-064)
● Cell homogenizer (Avestin EmulsiFlex-C3 or comparable system)
●
Disposable gravity chromatography columns (Bio-Rad, cat. no. 7321010)
● Fast protein liquid chromatography (FPLC) instrument (Cytiva Akta Pure or comparable system)
● Injection loop for FPLC (Cytiva, cat. no. 18111382)
● Cation exchange column, 5 ml SP HP (Cytiva, cat. no. 17115201)
● Centrifugal filters for protein concentration, 15 ml, 100-kDa molecular-weight cutoff (MWCO)
(Millipore Sigma, cat. no. UFC9100)

● Centrifugal filters for protein concentration, 0.5 ml, 100-kDa MWCO (Millipore Sigma, cat. no.
Z677906)
● Qubit 4 fluorometer (Thermo Fisher Scientific, cat. no. Q33238) (optional if using qPCR library
quantitation)
● Countess II cell counter (Thermo Fisher Scientific, cat. no. AMQAZ1000) or other preferred cell
quantification method
●
Lonza 4D Nucleofector with X unit for electroporation of base editors as mRNA or protein (Lonza, cat.
no. AAF-1002X and AAF-1002B)
● 37 °C, humidity- and CO -regulated incubator (Thermo Fisher Scientific, cat. no. 51030284 or
2
comparable system)
● Benchtop vortexer (Fisher Scientific, cat. no. 02-215-414 or comparable system)
● Blue-light transilluminator for gel cutting (VWR, cat. no. 76151-834 or comparable system)
● Gel-imaging system (Bio-rad ChemiDoc or comparable system)
● Miseq system (Illumina, cat. no. SY-410-1003)
Software
● CRISPResso2 (https://github.com/pinellolab/CRISPResso2)
●
Docker (https://www.docker.com/products/docker-desktop)
● Geneious or preferred comparable software (https://www.geneious.com/)
Reagent setup
Lysis buffer for extraction of gDNA from HEK293T cells
To prepare 1 liter of lysis buffer, combine 10 ml of 1 M Tris-HCl, pH 7.5 with 5 ml of 10% (wt/vol)
SDS solution. Add nuclease-free water to a final volume of 1 liter. This solution can be stored at room
temperature (25 °C) for a few months. Before usage, prepare an aliquot of the necessary volume of
lysis buffer and supplement freshly with 1:1,000 (vol/vol) proteinase K.
DMEM culture medium with 10% (vol/vol) FBS for culturing HEK293T cells
To prepare 500 ml of medium, sterile-filter 450 ml of DMEM supplemented with 50 ml of thawed
FBS. DMEM prepared with FBS should be stored at 4 °C and can be used for 2–3 weeks before it
should be remade.
HTS PCR primer solutions, 10 mM

Dissolve lyophilized primers in TE buffer to a stock concentration of 100 µM. Dilute stocks 10-fold in
nuclease-free water to a concentration of 10 µM before use. Diluted primer stocks can be stored at
room temperature for several days or for several months at −20 °C.
Low-salt buffer for protein purification

To prepare 1 liter of low-salt buffer, mix 700 ml of nuclease-free water with 100 ml of 1 M Tris-HCl,
pH 8.0 and 200 ml of glycerol. Add 1.4 g of TCEP powder and stir until dissolved. Sterile-filter the

solution and store at 4 °C for ≤2 months. Final composition: 100 mM Tris-HCl, pH 8.0; 5 mM TCEP;
20% (vol/vol) glycerol.
High-salt buffer for protein purification

To prepare 1 liter of high-salt buffer, mix 700 ml of nuclease-free water with 100 ml of 1 M Tris-HCl,
pH 8.0 and 200 ml of glycerol. Add 1.4 g of TCEP powder and 58.4 g of NaCl and then stir until
dissolved. Sterile-filter the solution and store at 4 °C for ≤2 months. Final composition: 1 M NaCl;
100 mM Tris-HCl, pH 8.0; 5 mM TCEP; 20% (vol/vol) glycerol.
Bacterial lysis buffer

Prepare freshly before use. For each liter of bacterial culture, add two tablets of EDTA-free cOmplete
protease inhibitor to 30 ml of high-salt buffer. Supplement with 30 µl of 2.5 U/µl DNase I solution.
Keep on ice until ready for use.
Nickel resin elution buffer

Prepare freshly before use. To prepare 10 ml of nickel resin elution buffer, mix 1 ml of 3 M imidazole
stock with 4.5 ml of high-salt buffer and 4.5 ml of low-salt buffer. Keep on ice until ready for use.
Procedure
Design of target sgRNAs and selection of base editor variants ● Timing 1 d
1 Follow the process outlined in Box 1 to design target sgRNAs.
2 Select the appropriate base editor variant(s) (see ‘Experimental design’). In general, we recommend
starting with a high-activity base editor (evoCDA/evoAPOBEC1 for CBE and ABE8/ABE8e for
ABE) to validate that the site is amenable to base editing.
Preparation of the sgRNA construct

3 We suggest conducting USER cloning (option A) or KLD cloning (option B) to generate sgRNA
plasmids, as used when delivering base editors as DNA. When delivering editors as protein or
mRNA, chemically modified sgRNAs can be purchased as RNA for co-delivery (option C).
(A) Generation of an sgRNA plasmid by USER cloning ● Timing 3 d
CRITICAL This method is recommended if the cloning procedure involves more than
c
simply swapping the protospacer sequence of a previously available plasmid.

(i) Preparation of a diluted sgRNA template. We recommend using pFYF1320 as a template, although
any mammalian expression vector containing a human U6 promoter can be used. Dilute the
template with water or TE buffer as purified plasmid to a final concentration of 50 ng/µl.
(ii) Design USER cloning primers containing the target sgRNA sequence. Examples can be
found in Table 2.
CRITICAL STEP For optimal cloning efficiency, the annealing temperature of the
c
designed primers should be between 60 and 65 °C and have no predicted hairpin

formations with an annealing temperature over 45 °C.
(iii) Prepare USER cloning PCR. The preferred polymerase mix is PhusionU Green Multiplex,
which is premixed to include both dNTPs and a dye for downstream gel-based analysis,
although any high-fidelity, uracil-tolerant polymerase master mix can be used. Prepare the
reaction as follows:
Component Volume Final concentration
PhusionU Green Multiplex PCR Master Mix, 2× 25 µl 1×

USER-Fw primer, 100 µM 0.25 µl 0.5 µM
USER-Rv primer, 100 µM 0.25 µl 0.5 µM
U6 template (pFYF1320 from Step 3A(i)) 1 µl 1 ng/µl
Distilled water 23.5 µl
Total 50 µl
(iv) Perform PCR under the following cycling conditions:

Cycle number Denature Anneal Extend
1 95 °C, 3 m
2–30 95 °C, 15 s 60 °C, 20 s 72 °C, 1 m
31 – – 72°C, 2 m
CRITICAL STEP Vary the annealing temperature depending on the predicted primer
c
properties.
(v) After the PCR reaction, verify successful amplification via agarose gel (1%, supplemented
with 1:10,000 (vol/vol) ethidium bromide or preferred nucleic acid staining reagent)
electrophoresis. Pour a 1% Load 2–5 µl of the PCR reaction directly into the gel along with
a DNA ladder and run in 0.5× TAE buffer at 140 V/cm for 20 min. Successfully amplified
products should run at ~2.2 kb in length.
? TROUBLESHOOTING
(vi) Purify the PCR products using the QIAquick PCR purification kit (Qiagen) according to
the manufacturer’s protocols.
(vii) Elute in EB buffer (provided in the kit) or water and normalize the concentration to 50 ng/µl.
PAUSE POINT Purified PCR products can be stored at 4 °C for several days or are stable
j
at −20 °C for several months.

(viii) USER reaction setup. Setup the USER reaction as a 0.1-pmol reaction as follows:
Component Volume (μl) Final concentration
Cutsmart buffer, 10× 1 1×

USER enzyme 1 1U
Dpn1 enzyme 1 1U
PCR product from Step 3A(vii) 3 150 ng (~0.1 pmol of 2.2-kb fragment)
Distilled water 4 –
Total 10 –
(ix) Perform the USER reaction under the following conditions in a thermocycler:
Step Temperature Duration (min)
Template digestion and uracil excision 37 °C 30

Overhang denaturation and enzyme 80 °C 5
inactivation
Annealing 12 °C (slow annealing at a rate of –
0.1 °C/s)
CRITICAL STEP Dpn1 digestion removes template plasmid that remains from the PCR
c
reaction. Because the selection cassette is unchanged, it is essential that this digestion runs
long enough to minimize re-transformation of the PCR template.
(x) Following completion of the USER assembly, place the reactions on ice.
(xi) Transformation of the USER reaction into chemi-competent E. coli. To each reaction, add a
10-µl aliquot of chemi-competent E. coli Mach1 cells, or a preferred transformation strain.
(xii) Transform the cells per the manufacturer’s instructions. Briefly, the assembly-cell mixture is
incubated on ice for 10 min, then heat-shocked at 42 °C for 30 s and placed back on ice for 1 min.
(xiii) Add 100 µl of S.O.C. medium and plate directly on a plate containing LB-agar and 50 µg/
ml carbenicillin. Incubate overnight at 37 °C.
CRITICAL STEP Cells do not need to be incubated for the outgrowth period after heat
c
shock when transformed with ampicillin/carbenicillin-resistant plasmids.

PAUSE POINT Transformed E. coli can be stored on plates for ≤2–3 weeks at 4 °C.
j
? TROUBLESHOOTING

(xiv) Pick single colonies into TempliPhi sample buffer and prepare rolling circle amplification
(RCA) reactions (Cytiva). Incubate for >6 h at 30 °C.
CRITICAL STEP Track the identity of colonies that are sequenced and store in LB
c
medium. We recommend RCA, but another method of sequence verification can be used if
preferred.
(xv) Sequence validation of the sgRNA plasmid. Submit RCA reactions for Sanger sequencing to
verify correct installation of the protospacer sequence.
(xvi) Transfer sequence-verified colonies into a 50-ml bioreactor tube (Corning) with 35 ml of
LB medium and 50 µg/ml carbenicillin. Incubate at 37 °C for 8–20 h.
(xvii) Isolate plasmid DNA using a QIAgen MidiPrep kit or an equivalent kit with an endotoxin
removal step, following the manufacturer’s instructions.
(B) Generation of a DNA sgRNA expression cassette with KLD cloning ● Timing 3 d
c CRITICAL STEP This method is recommended for generating sgRNA expression vectors for
which an existing cassette containing the desired sgRNA scaffold is available and only the
protospacer sequence needs to be altered.
(i) PCR amplify using a U6-sgRNA plasmid (e.g., pFYF1320 from Step 3A(i)) as template and
purify the resulting product for KLD following the same process as Step 3A(iii–vii).
Examples of how to design PCR primers for this reaction can be found in Table 2.
(ii) KLD reaction setup. For standard cloning, set up the KLD reaction (NEB) according to the
manufacturer’s protocol as follows:
Component Volume (µl) Final concentration
KLD buffer, 2× 5 1×
KLD enzyme mix (10×) 1 1×
PCR template from Step 3B(i) 0.5–1 –
Distilled water 3–3.5 –
Total 10 –
(iii) Incubate the KLD reaction at room temperature for 5 min and then place it on ice for
immediate transformation.
(iv) Transform the entire KLD reaction into 10 µl of chemi-competent E. coli Mach1 cells. Plate
the transformation and prepare and sequence-verify plasmid DNA following the same
protocol as Step 3A(xiii–xvii).
? TROUBLESHOOTING
(C) Acquiring purified, chemically modified, synthetic sgRNAs ● Timing 5–20 d
CRITICAL Instead of delivering sgRNA expression constructs, the sgRNA molecule can be
c
directly delivered into cells. We recommend this method if the editor will be delivered as
mRNA or protein, as opposed to a plasmid expression construct. Companies including
Agilent, BioSpring, IDT, Synthego and Trilink Biotechnologies produce custom sgRNAs
containing stabilizing modifications at the 5′ and 3′ ends, although turnaround times can be up
to 20 d. The most common type of synthetic sgRNA includes 2′O-methyl groups on the first
and last three nucleotides, as well as phosphorothioate bonds in the first and last three
dinucleotide bonds. These modifications are thought to increase the lifetime of the sgRNA in
the cell. Other modification patterns have also been described that may perform as well or
better. 10–50 pmol of sgRNA is delivered per sample in this protocol.
(i) Chemically modified sgRNAs are typically delivered in a lyophilized form. Upon receipt, dissolve
the sgRNA in TE buffer at a concentration of 100–300 µM and store at −20 oC for ≤1 year.
Preparation of E. coli for base editor protein expression (optional) ● Timing 3–4 d
CRITICAL This process is necessary only when choosing to deliver base editors as ribonucleoprotein
c
complexes. Alternatively, base editor protein can be purchased from a vendor such as Aldevron.
4 Clone the codon-optimized base editor gene for bacterial expression into the plasmid pD881-SR
(Atum) with an 8xHis tag at the N terminus. We recommend conducting USER cloning as
described in Step 3A(iii–xvii), with the following modifications. First, PCR-amplify the plasmid
backbone and the editor of interest as two separate reactions. Example primers are provided in
Table 2 that anneal to the editor and backbone, respectively, of the plasmid shown in

Supplementary Data 1. For the USER reaction described in Step 3A(viii), combine 1.5 µl
of each PCR product rather than 3 µl of a single product to assemble both components in the final
plasmid.
5 Transform the base editor expression plasmid into BL21 Star DE3–competent cells (Thermo Fisher
Scientific) or another Star (low RNase) strain according to the manufacturer’s protocol.
6 Add 100 µl of S.O.C. medium and incubate at 37 °C for ≥30 min of outgrowth.
7 Plate on 25 µg/ml kanamycin to select for transformants and place overnight at 37 °C.
8 Pick colonies for overnight growth in Terrific Broth supplemented with 25 µg/ml kanamycin at 37 °C.
9 Express the base editor protein. Dilute saturated overnight cultures to an OD600 of 0.05 in 1 liter of
Terrific Broth supplemented with 25 µg/ml kanamycin. Each liter of culture will generally yield
~0.5–3 mg of purified base editor protein.
10 Shake in baffled Erlenmeyer flasks at 37 °C until OD600 reaches ~1.5 (~2.5 h).
CRITICAL STEP Do not allow cells to overgrow beyond an OD600 of 1.8 before cold shock. If the
c
growth is too dense before induction, expression of the editor will be greatly reduced.
11 Cold-shock the flasks in an ice water slurry for 1 h and then add 26.6 ml of 30% (wt/vol)
L-rhamnose per liter of culture (final concentration of 0.8% (wt/vol)) to induce protein expression.
12 Incubate the flasks at 18 °C with shaking for 24 h.
13 Centrifuge the cells for 15 min at 6,000g and remove most of the supernatant, leaving
≤20 ml behind.
14 Scrape the cells into 50-ml centrifuge tubes, resuspending in residual medium if necessary.
15 Centrifuge again and discard the supernatant. The pellet should contain 15–25 g of bacteria per liter
of culture. Flash-freeze cells in liquid nitrogen and store at −80 °C overnight.
PAUSE POINT Frozen bacteria can be stored for several months.
j
16 Cell lysis. Prepare bacterial lysis buffer and resuspend cells from 1 liter of culture in 30 ml of cold
bacterial lysis buffer.
CRITICAL STEP Base editor proteins are prone to degradation during purification. Add protease
c
inhibitors to at least 2× the concentration recommended by the manufacturer and ensure that
samples are kept at 4 °C throughout this process.
17 Passage cells three times through a homogenizer (Avestin Emulsiflex-C3 or similar) at ~18,000 psi
to lyse. Alternatively, a sonicator may be used to lyse bacteria.
18 Pellet cell debris by centrifuging for 20 min at 20,000g at 4 °C.
19 Collect the supernatant, taking care to keep it cold. The pellet may be discarded.
Purification of base editor protein (optional) ● Timing 1 d

Immobilized metal affinity chromatography
20 Centrifuge 1.5 ml of Ni-NTA resin slurry per liter of bacterial culture for 5 min at 500g at room
temperature.
21 Remove the supernatant and resuspend in 0.75 ml of lysis buffer per liter of bacterial culture.
22 Repeat centrifugation, discard the supernatant and resuspend the resin in 0.75 ml of high-salt buffer
per liter of bacterial culture.
23 Add the washed Ni-NTA slurry from Step 22 to the cell lysate (supernatant from Step 19).
24 (Optional) Add 30 mM imidazole during Ni-NTA resin binding to greatly increase the purity of
eluted protein, at the expense of losing some yield.
25 Slowly rotate/nutate the tube for 1 h at 4 °C to bind protein to the resin.
26 Centrifuge the lysate for 5 min at 500g at 4 °C to pellet the Ni-NTA resin.
CRITICAL STEP Save 20 µl of supernatant for subsequent analysis of the protein preparation (see
c
Step 46). Store at −20 °C for ≤3 d.

27 Resuspend the resin in 5 ml of high-salt buffer. More high-salt buffer may be needed if the resin bed
volume is >5 ml.
28 Flow the solution through a disposable chromatography column at 4 °C so that the resin slowly
settles into a bed. The ‘column volume’ is approximately half the volume of Ni-NTA resin slurry
used. Split the resin so that the column volume is <3 ml per column used.
29 Wash the resin twice with 10 column volumes of cold, high-salt buffer.
30 Elute protein in one column volume of nickel resin elution buffer. Repeat elution in one additional
column volume of nickel resin elution buffer and combine the eluates.
CRITICAL STEP Store 10 µl of elution for later analysis (see Step 46). Store at −20 °C
c
for ≤3 d.

Cation exchange chromatography
31 Prepare a 5-ml cation exchange column (GE Healthcare HiTrap SP HP on an Äkta Pure25 FPLC or
comparable instrument). Wash the column with five column volumes of water followed by five
column volumes of low-salt buffer. Place a reservoir of >500 ml of low-salt buffer in flow line A.
Place a reservoir of >300 ml of high-salt buffer in flow line B.
32 Dilute 1–4 ml of immobilized metal affinity chromatography–purified protein with 40 ml of low-
salt buffer immediately before loading into the 50-ml injection loop.
33 Flow the diluted protein through the ion exchange column to bind.
34 Wash the column with three column volumes of low-salt buffer.
35 Initiate an increasing gradient starting from 100% low-salt buffer to a maximum of 80% high-salt
buffer over the course of 50 ml. Use a flow rate of 5 ml/min.
36 Collect 1-ml fractions during this ramp to high-salt buffer.
37 Wash the column with three column volumes of high-salt buffer, five column volumes of water and
then five column volumes of 20% ethanol. Discard all washes and then remove the column and seal
it with the screws included with its purchase. Moisten both sides using a squirt bottle of distilled
water before sealing the column to ensure no air bubbles enter it. The column may be stored at 4 °C
for ≤3 months and used again for a subsequent protein preparation.
38 Assess fractions of purified protein. Combine 5 μl of each fraction from Step 36 with 1.7 μl of
NuPAGE LDS Sample Buffer. Incubate 3–5 minutes at 98 °C, let cool on the bench and load onto a
4–12% gradient SDS-PAGE gel alongside a protein molecular weight ladder such as the Precision
Plus Protein Dual Color Standards. Cover the gel with 1× MOPS buffer and run the gel at 180 V
for 45 min.
39 Stain with InstantBlue protein stain for 15 min.
40 Remove the stain and nutate the gel with water to resolve bands. Base editors are generally among
the largest proteins that will be visible, between 180 and 250 kDa.
41 Combine fractions with the desired band. If multiple bands are present in some fractions, consider
discarding them for maximum purity. Sterile-filter using a luer lock syringe and filter unit.
? TROUBLESHOOTING
42 Concentrate purified protein. Centrifuge combined fractions in a 15-ml, 100-kDa-cutoff centrifugal
filter tube at 4 °C and 3,000g for 30–60 min.
43 If desired, further concentrate in a 0.5-ml, 100-kDa-cutoff centrifugal filter tube at 4 °C and 10,000g
for another 15–30 min.
? TROUBLESHOOTING
44 Quantify protein concentration using a BCA assay following the manufacturer’s instructions and
divide aliquots as desired into PCR strip tubes. Base editor proteins can generally be concentrated to
between 20 and 100 µM and yield a total of ~0.5–3 mg per liter of bacterial culture.
45 Flash-freeze protein aliquots in liquid nitrogen. Store at −80 °C.
PAUSE POINT Purified base editor protein can be safely stored at −80 °C for several
j
months.
46 Assess the quality of the purification procedure by running the stored aliquots from Steps 26, 30
and 45 on an SDS-PAGE gel, as described in Steps 38–40. Typically, 0.2 µl of the bacterial lysate
(from Step 26), 1 µl of the nickel elution (from Step 30) and 0.05 µl of the concentrated protein
stock (from Step 45) can be run side by side. This gel will visualize the additional purity achieved by
each purification step, and if the purification was not successful, it will help determine at which step
the protein was lost (Supplementary Fig. 1).
Verification of sgRNA activity in HEK293T cells ● Timing 4–5 d

CRITICAL Base editors have been verified in numerous cell types. Here, we describe base editing in
c
HEK293T cells, which are a good starting point for screening optimal base-editing target sites before
transitioning to more difficult or valuable cell lines or animal models.
47 HEK293T cell culture. Maintain HEK293T cells (ATCC) according to the vendor’s protocols.
Briefly, grow HEK293T cells in T75 flasks in DMEM supplemented with 10% (vol/vol) FBS at 37 °C
and 5% CO2.
48 Passage HEK293T cells at 70–80% confluency. To passage, remove the medium and wash the cells
with 1× PBS buffer, taking care not to directly pipette onto the cells as HEK293T cells readily
detach from the flask surface.
49 After washing and removal of PBS, add 2 ml of TrypLE and incubate at 37 °C for 5 min.

50 Add 10 ml of freshly warmed DMEM, making sure to pipette up and down to ensure thorough
dissociation of the cells.
51 Reseed dissociated cells into a fresh flask for later use, or prepare or plate cells depending on the
chosen delivery method (see Step 52).
CRITICAL STEP Antibiotics like penicillin and streptomycin can be used during maintenance;
c
however, they sometimes affect cell physiology and should particularly be avoided during
transfection of cells. In general, cells should not be grown to confluency higher than 80% or used
past passage 15. Passage cells at a 1:5 ratio or 1:10 ratio every 2 or 3 d, depending on when the cells
will be needed.
52 Delivery of the sgRNA and base editor. Base editors can be delivered into cells as expression
constructs on plasmids (option A), as purified mRNAs (option B) or as proteins (option C). For
option A, sgRNAs are also delivered as plasmids. For options B and C, we recommend using
chemically modified synthetic RNA as the guide RNA, though plasmids or in vitro transcribed
sgRNA can also be used.
(A) Plasmid delivery of base editors ● Timing 3–4 d
Order and prepare transfection-grade base editor plasmid(s) (Supplementary Table 5). If the
desired base editor plasmid(s) are not available from Addgene, they can be prepared with
USER cloning similarly to sgRNA preparation (Step 3A).
(i) Cell plating for transfection. Starting from a flask of dissociated cells (see Step 50), plate at
the manufacturer’s recommended density for the desired plate format, using the Countess
II cell counter or preferred cell-counting equipment to determine the starting cell densities
per the manufacturer’s instructions. In general, base-editing experiments in HEK293T cells
can be carried out in 48- or 96-well formats, which are seeded at densities of 5.5 × 104 and
1.75 × 104 cells in 250 or 100 µl of DMEM per well, respectively.
CRITICAL STEP Cells should not be plated over- or under-confluent, as the former can
c
result in low transfection efficiency, and the latter can result in excessive cell death from
lipid toxicity.
CRITICAL STEP Cells are transfected ~18–24 h after plating (Step 52A(viii)) and should
c
be ~70–85% confluent for maximal transfection efficiency.

(ii) Premix each desired sgRNA vector (250 ng) into 6.25 µl of serum-free Opti-MEM
per sample.
CRITICAL STEP Each well receives one guide and one base editor, but mastermixes can
c
be made for the total number of reactions needed for each sgRNA expression vector. For
instance, if guide A is to be tested with base editors A, B and C, then prepare at least a
3× mastermix of guide A.
(iii) Separately, premix each desired base editor vector (750 ng) into 6.25 µl of Opti-MEM
per sample.
CRITICAL STEP Each well receives one guide and one base editor, but mastermixes can
c
be made for the total number of reactions needed for each base editor expression vector.
(iv) Mix desired sgRNA/base editor combinations to a total volume of 12.5 µl per sample.
Optionally, co-transfect with a fluorescent marker (e.g., GFP-expressing plasmid) to enable
transfection efficiency to be tracked. Add 10 ng of this vector to each sample, ensuring that
the total volume is unchanged by scaling down the amount of Opti-MEM. Alternatively,
use a base editor vector containing a 2A-GFP cassette (e.g., pCMV_BE4max_P2A_GFP,
Addgene: 112099). The total DNA amounts and volumes should be as follows:
Component Amount (ng) Volume (µl)
sgRNA vector (Step 3A(xvii), 3B(iv) or 3C(i)) 250 6.25

Base editor vector (Step 52A(i)) 750 6.25
GFP vector (optional) 10 –
Total DNA 1,000–1,010 12.5
(v) This represents the vector amounts needed per well when using 48-well plates. When using
96-well plates, scale the mass of DNA and all volumes down by a factor of 3.
(vi) Prepare a lipid mastermix as 1.5 µl of Lipofectamine 2000 diluted to a total volume of
12.5 µl with Opti-MEM scaled by the number of desired wells.

CRITICAL STEP Cells can be transfected with Lipofectamine 2000 or Lipofectamine
c
3000.
(vii) Add the lipid mixture to the sgRNA/base editor mixture to a total volume of 25 µl.
(viii) Incubate the final lipid/plasmid mixture at room temperature for 10 min before
adding it to plated cells from Step 52A(i). Gently tap or rock the plate to ensure
even distribution of the lipid/plasmid mixture and then place it back in the incubator
at 37 °C.
CRITICAL STEP The plasmid and lipid concentrations listed here are high and are
c
generally recommended for base-editing screening conditions or for achieving maximal
editing efficiency. The amounts can be titrated as needed depending on the desired balance
between cell death (from lipid toxicity) and base-editing efficiency.
(ix) (Optional) The day after transfection, transfection efficiency can be monitored by
observing GFP expression under a fluorescence microscope. The medium can be aspirated
and replaced with fresh DMEM to reduce lipid toxicity if needed.
? TROUBLESHOOTING
(B) mRNA delivery of base editors ● Timing 3–4 d
(i) Purchase base editor mRNA (see ‘Reagents’). Alternatively, base editor mRNA can be
transcribed from a PCR product using a kit such as the NEB HiScribe T7 ARCA mRNA
kit. Briefly, clone the editor into a plasmid with desired UTR sequences. Using this plasmid
as a PCR template, PCR-amplify the transcription template using primers that bind to the
editor and contain UTRs and a 120-nt poly-A tail (example primers are listed in Table 2).
Use agarose gel electrophoresis followed by a Qiagen gel extraction kit to purify the band
corresponding to the desired transcription template. Using this transcription template,
transcribe, cap and purify base editor mRNA via lithium chloride precipitation per the
manufacturer’s instructions.
(ii) Determine the cell concentration (see Step 52A(i)) and transfer 200,000 cells from Step 50
per sample to a centrifuge tube—preferably at ≥600,000 cells per tube to prevent the loss of
cells due to a small and loose pellet.
(iii) Centrifuge at room temperature with a force of 150g for 8 min.
(iv) Remove the supernatant and wash the pelleted cells in 1 ml of sterile PBS.
(v) Repeat centrifugation.
CRITICAL STEP This wash removes RNases that are present in most growth media,
c
either from the serum or secreted by the cultured cells. mRNA can be rapidly degraded if
this wash is not thorough. One wash is usually sufficient, but if poor activity is observed,
then consider increasing the number of washes.
(vi) During the centrifugations, formulate Lonza SF buffer. Mix 16.4 µl per sample of
nucleofector solution with 3.6 µl per sample of supplement solution provided with the
Lonza SF kit.
(vii) Prepare the editor reagent mixture. Add sgRNA and base editor mRNA stock to the
formulated Lonza SF buffer. If replicate cell lines will be treated, then multiply these
amounts by 1.1× the number of samples.
Component Amount (µl)
sgRNA (100 µM stock) (Step 3C(i)) 0.5

Base editor mRNA (2 µg/µl stock) (Step 52B(i)) 1.5
Lonza SF buffer 15
Total volume 17
(viii) Resuspend cells from Step 52B(v) in 5 µl of Lonza SF buffer per 200,000 cells.
(ix) Transfer 5 µl of cell suspension into 20-µl nucleocuvette wells included in the
Lonza kit.
(x) Add 17 µl of editor reagent mix from Step 52B(vii) to cells and pipette up and down three
times to mix.
(xi) Nucleofect using program CM-130 on a Lonza 4D nucleofector device.
(xii) After nucleofection, add 80 µl of warmed DMEM + 10% (vol/vol) FBS per well and
incubate for 5 min at 37 °C to allow cells to recover.

(xiii) After recovery, pipette cells up and down to mix and transfer 12.5 µl of nucleofected cells
into 100 µl of warmed DMEM + 10% (vol/vol) FBS in a 96-well plate for further growth
as desired.
(xiv) (Optional) If a P2A-containing base editor transcript is used, GFP fluorescence can be
monitored the day after nucleofection. Alternatively, cell viability can be approximated
under a light microscope (Zeiss Axio Observer or comparable system).
? TROUBLESHOOTING
(C) Protein delivery of base editors ● Timing 3–4 d
(i) Base editor proteins can be purified in the laboratory (see Steps 4–46) or purchased from a
protein-purification company such as Aldevron.
(ii) Prepare HEK293T cells from Step 50 for protein delivery as described in Step 52B(ii–v).
(iii) Prepare the editor reagent mixture. Add 1 µl of 90 µM base editor protein stock from Step
52C(i) to 15 µl of Lonza SF buffer per sample. Add 1 µl of 100 µM sgRNA from Step 3C(i)
per sample and pipette up and down to mix.
CRITICAL STEP If base editor protein is less concentrated, use a total of 2 µl of reagent
c
with a 1.1:1 molar ratio of sgRNA:protein.

(iv) Incubate at room temperature for 5 min to allow the ribonucleoprotein complex to form.
CRITICAL STEP Do not add concentrated protein to concentrated sgRNA solutions, as
c
this tends to cause precipitation of the editor and a loss of editing efficiency. First dilute the
protein in buffer, followed by addition of the sgRNA.
? TROUBLESHOOTING
(v) Nucleofect cells with the same process as Step 52B(xi–xiv).
(vi) (Optional) Cell viability can be monitored the day after nucleofection.
? TROUBLESHOOTING
Enrichment of treated cells with flow cytometry (optional) ● Timing 2–6 h

CRITICAL Use FACS to enrich for cells that express more base editor protein and thus are more likely
c
to contain the desired edit. Cells may be cultured further after FACS if desired.
53 Three days after plasmid delivery or 1 d after delivery of editor mRNA, prepare the cells for FACS.
Aspirate the transfection medium and wash cells once with one volume of 1× PBS (250 µl for a
48-well plate and 100 µl for a 96-well plate).
CRITICAL STEP FACS enrichment is not recommended after protein delivery because base editor
c
proteins have not been directly linked to fluorescent molecules, and the lifetime of delivered protein
in the cell may be too short to visualize.
54 Add TrypLE directly to each well of the plate (100 µl per well of a 48-well plate and 30 µl per well of
a 96-well plate) and incubate at 37 °C for 5 min.
55 (Optional) Monitor trypsinization with a light microscope. The incubation time can be increased or
decreased depending on how quickly the cells dissociate from the plate. When all cells have
dissociated from the plate surface, the incubation can be stopped.
56 Inactivate TrypLE by addition of prewarmed culture medium (1 ml for a 48-well plate and 200 µl
for a 96-well plate).
CRITICAL STEP For fluorescence-gated sorting, the culture medium should be phenol-red free.
c
57 Gently but thoroughly mix by pipetting to minimize cell clumping.

58 Pipette cells through a cell strainer snap cap (Corning) into a tube for FACS.
59 Place the samples on ice until FACS begins.
60 Consult literature regarding sorting of the cell type of interest. Gating for the desired population
should generally include removal of cell debris and events containing non-singlet cells.
? TROUBLESHOOTING
61 (Optional) Include a viability dye such as DAPI to facilitate collection of only live cells.
CRITICAL STEP Flow cytometry gates should be set using control populations of cells. In general,
c
an untreated population should be used to set the gates for removal of debris and isolation of singlet
cells of interest, and to determine the stringency of the fluorescence gate. Note that higher
stringencies of the fluorescence gate will be correlated to populations of cells with higher levels of
base editor expression, and therefore greater potential for desired base-edited outcomes.
62 After flow cytometry, pellet the collected cells by centrifugation for 10 min at 200g and 4 °C.
63 Aspirate the medium and then wash once with 1× PBS. The collected cells are now ready for lysis
and/or re-plating for further growth.

Preparation of mammalian cells for HTS ● Timing 1 d
CRITICAL For most SNP amplicon sequencing applications, gDNA can be harvested through a simple
c
cell lysis without subsequent purification.
64 Add proteinase K (NEB) at a 1:1,000-fold (vol/vol) dilution to cell lysis buffer.
65 Cell lysis. Add lysis buffer directly to plates from Step 52A(viii), 52B(xiii) or 52C(v) or pelleted cells from
Step 63, typically 3 days after transfection or nucleofection. In general, 30 µl of lysis buffer is sufficient
for each well of a 96-well plate, and 100 µl of lysis buffer is sufficient for each well of a 48-well plate.
CRITICAL STEP These volumes were determined empirically for HEK293T cells grown to
c
confluency in 48- and 96-well plates (30,000 and 100,000 cells, respectively). The amount of lysis
buffer can be adjusted depending on the cell type and cell density used.
66 Upon addition of lysis buffer, incubate plates at 37 °C for 1 h.
67 After incubation at 37 °C, transfer the lysis mixture to PCR strips and inactivate the proteinase K at
80 °C for 20 min. The resulting gDNA can now be used for downstream analysis.
CRITICAL STEP The lysis can immediately be transferred from plates to PCR strips and
c
incubated at 37 °C on a thermocycler instead. However, we recommend plate incubation, as the

lysis mixture is viscous and difficult to pipette early in the lysis process.
PAUSE POINT Harvested gDNA can be temporarily stored at 4 °C for several days or safely
j
frozen at −80 °C for months.

? TROUBLESHOOTING
HTS preparation for base-editing analysis ● Timing 1 d

68 Design suitable PCR1 primers using NCBI’s Primer-BLAST tool, or if the target site was designed
using CHOPCHOP, the program provides sets of potential PCR1 primers. In general, we
recommend designing the primers to have a melting temperature around 60 °C.
CRITICAL STEP Ensure that the primers amplify a region such that the target site is ≥20 bp away from
c
the edges of the amplicon. The 5′ ends of PCR1 primers are appended with barcode sequences (Table 2)
that are recognized by the primers used in the second, sample barcoding PCR (PCR2; see Step 73).
69 Prepare the PCR1 reaction as follows:
Component Volume (µL) Final concentration
Q5 Hot Start High-Fidelity Mastermix, 2× 12.5 1×

PCR1-Fw primer, 100 µM 0.125 0.5 µM
PCR1-Rv primer, 100 µM 0.125 0.5 µM
gDNA (Step 67) 0.5–1 Varies
Distilled water 11.25–11.75
Total 25
CRITICAL STEP gDNA yields can vary depending on the number of cells that were lysed and the
c
volume of the lysis buffer. In general, we recommend starting with 0.5–1 µl, but if these do not work,
testing a panel of gDNA template amounts can increase the chance of successful PCR amplification.
CRITICAL STEP When conducting CBE experiments, a uracil-tolerant polymerase such as
c
PhusionU, which copies uracil as thymine, should be used if the user wishes to include uracil-
containing genotypes. When the user does not wish to include unresolved uracils as edited
genotypes, then Q5 or a comparable DNA polymerase that does not copy amplicons containing
uracil should be used for PCR1 instead.
70 Perform PCR1 under the following cycling conditions:
1 95 °C, 3 min – –
2–30 95 °C, 15 s 60 °C, 20 s 72 °C, 20 s
31 – – 72 °C, 1 min
CRITICAL STEP The number of PCR cycles should be optimized using qPCR to the top of the
c
linear range to minimize biased amplification. We find that, using these recommended conditions,
24–29 cycles is usually a good starting point.

71 After PCR, evaluate the amplified product on an agarose gel (1% (wt/vol)). Successfully amplified
products should be a band at the size of the designed amplicon + 70 bp (for a typical 200- to
300-bp amplicon designed for a 300-cycle Illumina sequencing kit, this band will appear between
270 and 370 bp).
CRITICAL STEP Non-optimal PCR1 primers can sometimes result in multiple amplification
c
bands. Occasionally, these additional bands can be tolerated if they can reasonably be separated
during the gel extraction step after PCR2 (see Step 76).
? TROUBLESHOOTING
72 (Optional) Prepare 10 µM stocks of Illumina forward and reverse index primers. Standard PCR2
adapter sequences can be found on Illumina’s website (https://support.illumina.com/downloads/
illumina-adapter-sequences-document-1000000002694.html).
73 Reamplify PCR1 products in a second PCR step (PCR2) to add Illumina indices for identifying
unique samples. Prepare the PCR2 reaction as follows:
Component Volume (µL) Final concentration
Q5 Hot Start High-Fidelity Mastermix, 2× 12.5 1×

PCR2-Fw index primer, 10 µM 1.25 0.5 µM
PCR2-Rv index primer, 10 µM 1.25 0.5 µM
PCR1 product (Step 70) 1–2 Varies
Distilled water 8–9 –
Total 25 –
CRITICAL STEP Each sample to be sequenced should have a unique combination of Fw and Rv
c
indices.
74 Perform PCR2 under the following cycling conditions:
1 95 °C, 3 min – –
2–11 95 °C, 15 s 60 °C, 20 s 72 °C, 20 s
12 – – 72 °C, 1 min
CRITICAL STEP The number of PCR cycles should be optimized similarly to Step 70. Using these
c
recommended conditions, 10–15 cycles is generally a good starting point.

75 After PCR2, evaluate amplified products on an agarose gel (1% (wt/vol)). Products should be visible
as a band at the size of the designed amplicon + 140 bp (for a 200–300-nt amplicon, this band will
appear between 340 and 440 bp).
76 Pool ~2 µl of each PCR2 product together into one mastermix. Alternatively, if the amplicon sizes
vary substantially (>100-nt difference), pool amplicons separately based on size. Increase the
volume to 5–10 µl for fewer samples (<10 samples) to increase the efficiency of the gel extraction.
CRITICAL STEP The molar ratio added of each PCR2 product will be equivalent to the ratio of
c
total sequencing reads that will correspond to that amplicon after HTS. Depending on the quality of
the PCR2 reaction (evaluated via gel band density) and desired sequencing coverage of each
amplicon, the volume pooled of each PCR2 reaction should vary accordingly.
77 Load the pooled PCR2 reactions onto an agarose gel (1% (wt/vol)) for gel extraction. We
recommend running the gel for slightly longer (~25 min at 140 V) to enable better separation
between the desired PCR2 product and undesired unreacted primers or PCR1 template.
78 Gel-extract using the QIAquick Gel Extraction Kit (Qiagen) or preferred kit of choice, following the
manufacturer’s instructions.
CRITICAL STEP Library quantification can be affected by the presence of residual salts from gel
c
extraction. To ensure high-quality gel extraction, the gel should be fully dissolved before column
purification. In addition, we recommend washing three times with 750 µl of PE buffer (or the
corresponding ethanol-based wash buffer) before elution.

Box 2 | Guidelines for batch parameter files
The batch parameter file should be tab delimited and can be readily modified in Microsoft Excel as a .csv file (see Table 3 for an example).
● Row 1 comprises the parameter identifiers; at minimum, there are four required user inputs:
r1 – the first fastq file (for paired end reads, a second r2 parameter is necessary)
a – the nucleotide sequence of the amplicon
g – the nucleotide sequence of the sgRNA sequence not including the PAM
n – desired name of the output file
● If no other parameters are defined, CRISPResso2 will use the default settings (see documentation).
● Common parameters that are also useful for base-editing analysis include:
q – defines the minimum average phred quality score threshold that must be met for a read to be included in the analysis (no filtering is done in
the default settings; recommended value: 30)
w – defines the size of the window to be considered for classifying reads as modified or unmodified (recommended value: 20)
wc – defines the center of the window to be considered for classifying reads (recommended value: −10)
base_editor_output – parameter can be turned on to generate additional base-editing plots that can be used directly.
79 Elute the gel-extracted DNA into water or TE buffer and normalize the concentration such that the molarity
is between 8 and 12 nM. Concentrations can be calculated using the tool in Supplementary Table 6.
CRITICAL STEP It is critical that the library is initially normalized such that the true molar
c
concentration is >4 nM. We have found that the recommended library concentration (8–12 nM)
will generally fall within the sensitivity range of library quantification methods. Libraries that are
too concentrated can be subsequently corrected by additional dilution. In contrast, libraries that are
too dilute (<4 nM) are much harder to adjust.
80 Library quantification. Quantify the library concentration using one of three common methods for
library quantification, following the manufacturer’s instructions, ordered from lowest to highest
precision: by absorbance (NanoDrop), by fluorescence (Qubit) or by qPCR (KAPA Biosciences).
! CAUTION Absorbance provides the lowest precision and is not recommended except in cases
where the library preparation has been previously verified with an alternative method. If using
absorbance, directly normalize the library to 4 nM from the concentrated library in Step 79.
? TROUBLESHOOTING
81 Use the molar concentration determined from Step 80 to normalize the library concentration
to 4 nM.
82 Illumina MiSeq DNA sequencing. Complete the remaining library-preparation steps and load the
sequencer according to the Illumina user manual.
CRITICAL STEP We recommend loading a final library concentration between 10 and 12 pM
c
onto the sequencing kit.
CRISPResso analysis ● Timing 2–4 h

CRITICAL Detailed CRISPResso2 usage can be found in the original publication and code access
c
pages (https://github.com/pinellolab/CRISPResso2). Here, we describe an example workflow of base-

editing analysis after HTS commonly used in our laboratory.
Batch analysis of sequenced amplicons

83 After HTS, place demultiplexed files into folders based on amplicon identity. As CRISPResso
is designed primarily for single amplicon analysis, it is recommended to sort fastq files in
this fashion.
CRITICAL STEP Please see the documentation for alternative analyses possible with
c
CRISPResso2. Note that CRISPResso2 analysis works directly on zipped fastq files (fastq.gz).
84 Batch analysis on a single amplicon can readily be carried out using a batch parameter file
and the CRISPRessoBatch functionality. In each amplicon subfolder from Step 83, generate a .csv
text file named as desired (e.g., AMPLICON.csv), which serves as the batch parameter file. Populate
the file using the guidelines provided in Box 2 (Table 3). In rows 2 through n, input the sequences
and desired parameters for each fastq file to be analyzed.
CRITICAL STEP Batch analysis can also be carried out on fastq files associated with multiple
c
amplicons in one parameter file; however, certain summary tables and comparison plots may not be
available if analyzed this way.
85 In the command line, change to one of the directories containing sorted fastq files and the
associated batch parameter file and then implement the batch analysis (via Bioconda or Docker).

Table 3 | Example batch parameter entries for CRISPResso2 base-editing analysis
fastq_r1 a g n q w wc
Sample_1_FILENAME Amplicon sgRNA Sample_1 (desired 30 (desired phred 20 (width of −10 (center of the
sequence here sequence here name of output filter score) analysis window, window to be
folder) usually the size of considered for
the protospacer) classifying reads)
Sample_2_FILENAME Amplicon sgRNA Sample_2 39 20 −10
sequence here sequence here
Sample_3_FILENAME Amplicon sgRNA Sample_3 30 20 −10
sequence here sequence here
When completed, the output files should be present in a new directory (CRISPRessoBatch_-
on_AMPLICON). Within this directory, summary batch analysis files and separate directories
containing analysis for each fastq file can be found.
CRITICAL STEP Before continuing the analysis, it is important to check the Read_barplot.pdf file
c
in each individual analysis folder. Too few total reads or poorly aligned reads (low ratio of aligned
reads to total reads) can result in a biased analysis.
? TROUBLESHOOTING
86 Quantifying base-editing outcomes. CRISPResso directly provides plots of the nucleotide
distribution for aligned reads of each fastq file; use these directly for quantifying base-editing
outcomes. The raw nucleotide percentages can also be found in the Nucleotide_percentage_sum-
mary.txt file and reprocessed for alternative plotting.
87 Change to another directory containing fastq files for an amplicon to be analyzed and repeat Steps
83–86 for each sequenced amplicon.
Troubleshooting
Troubleshooting advice can be found in Table 4.
Table 4 | Troubleshooting table
Step Problem Possible reason Solution
3A(v) No amplification of Incorrect template, unsuitable Rerun the PCR as a gradient PCR with eight
sgRNA PCR annealing temperature or different annealing temperatures ranging from 58
unsuitable primer to 72 °C. If the PCR still does not work, or multiple
annealing region bands appear, a longer or alternative primer
binding region should be designed
3A(xiii) No colonies on plate or Too much template in the PCR Run the PCR reaction with a minimal amount of
colonies consist of starting reaction or incomplete Dpn1 template DNA. Incubate the USER reaction for a
template (USER assembly) digestion; USER junction longer period. Check to make sure that the USER
too short junction is unique and ≥8 bp in length. If the USER
junction has sequence homology anywhere else in
the plasmid, the assembly can fail
3B(iv) No colonies on plate or Too much template in the PCR The assembly can be run instead at 37 °C for
colonies consist of starting reaction or incomplete Dpn1 >5 min to increase the digestion time. Titrate the
template (KLD cloning) digestion amount of PCR template added to the KLD
reaction
41 The purified editor protein is Purified base editors are prone Increase the concentration of protease inhibitors
not the expected size or has to proteolysis, particularly in used during bacterial cell lysis, even beyond
multiple bands observed via the linker regions between manufacturer recommendations. Keep the protein
SDS-PAGE deaminase and Cas9 at 4 °C throughout the entire purification process
to minimize protease activity
43,52C(iv) The base editor protein Base editors are not soluble in Increase the glycerol concentration during
precipitates when all conditions/concentrations purification to enhance protein solubility; better
concentrating it or after mixing and become less soluble after solubility is observed at 20% up to 50%. When
with sgRNA complexing with sgRNA complexing with sgRNA, first mix sgRNA and
buffer to the most dilute concentration the
Table continued

Table 4 (continued)
Step Problem Possible reason Solution
experiment allows and then add protein and

pipette to mix. Let this complex at room
temperature for 5–10 min
52A(ix) Apparent cytotoxicity or cell Cells not confluent enough or Lipofectamine is cytotoxic, particularly at high
death (lipofection) poor DNA preparation concentrations. Using the recommended
conditions, some cytotoxicity is expected, but
substantial cell death suggests that the
concentration may be too high. If necessary, the
amounts can be scaled down. Ensure that plasmid
DNA is both endotoxin and RNA free
52A(ix), No or few GFP-positive cells Low transfection efficiency or Ensure that cells are no greater than 85%
52B(xiv) cells plated too densely confluent before transfection. Cells should not be
plated for >24 h to minimize clumped cell growth.
After preparation of the lipid mixture, it should be
allowed to sit for ≥10 min to ensure proper lipid
assembly
52B(xiv), Apparent cytotoxicity or cell Cells lost during washing and Cell pellets may be readily lost during the PBS
52C(vi) death (electroporation) pipetting or toxic materials wash, and cells may be killed by excessive and
electroporated harsh pipetting. To avoid this, consider washing
and resuspending fivefold additional cells to form
a more solid pellet and pipetting slowly and
gently, allowing extra recovery time after
electroporation. Ensure that the materials
nucleofected are of high purity and evaluate their
activity across a range of concentrations.
Consider using uridine analogs like N1-methyl
pseudouridine or 5-methoxyuridine to decrease
innate immunity of mRNA
60 Flow sorting shows a low Transfection was too cytotoxic FACS-analyze an untreated control sample. If
GFP signal there is a substantial difference in the forward
scatter/side scatter profiles of the untreated vs.
treated cells, the transfection may have been too
toxic to the cells. In these cases, the amount of
transfected base editor can be scaled down. Note
that higher editing efficiencies can be enriched by
setting a more stringent fluorescence gate
67 gDNA lysis too viscous Incomplete cell lysis The duration of the cell lysis depends on the
number of cells present. In general, we find that a
confluent plate after a base-editing experiment
requires ~1 h at 37 °C before the lysis is complete
71 PCR1 does not work Too much/little template Primer BLAST to check for undesired annealing
gDNA or primers not regions. Rerun the PCR by titrating the amount of
specific enough template gDNA
80 Library quantification Poor gel extraction or library Ensure removal of contaminants from isolated
inconsistent concentration too high DNA and normalize concentration to ~10 nM
85 Low sequencing read Nonspecific primers Reamplify the target amplicon with a different set
alignment of primers
Unexpected alleles Cell line polymorphisms Ensure low passage of cells and sequence an
untreated cell sample
Timing
Steps 1–2, design target sgRNAs and select base editor variants: 1 d
Step 3A, generate DNA sgRNA expression vector by USER cloning: 3 d
Step 3B, generate DNA sgRNA expression vector by KLD cloning: 3 d
Step 3C, acquire purified, chemically modified synthetic sgRNAs: 5–20 d
Steps 4–19, prepare expression strain for base editor purification: 3–4 d
Steps 20–46, purify base editor protein: 1 d
Steps 47–52, verify sgRNA base editing activity in HEK293T cells: 4–5 d
Steps 53–63, enrich base-edited cells by FACS: 2–6 h
Steps 64–67, isolate gDNA for HTS: 1 d
Steps 68–82, prepare and high-throughput sequence isolated gDNA: 1 d
Steps 83–87, computationally analyze base-editing outcomes: 2–4 h

Anticipated results
Base editing can be successfully applied to achieve high-precision genome editing in mammalian cells.
Because base editing is highly robust, the technology tolerates multiple modifications and delivery
strategies, increasing the likelihood of achieving the editing properties necessary for a given appli-
cation. Here, we have demonstrated the editing efficiencies that can be typically expected for a base-
editing experiment in an amenable cell line (HEK293T). We show that base editing is robust across
multiple target sites, with efficiencies approaching 100% if further enriched by FACS (Fig. 8a–e).
Computational analysis with CRISPResso2 provides allelic information (Fig. 8c) on common off-
target genotypes and can inform subsequent target site refinement. Both pre- and post-processed files
for the data used to generate Fig. 8d,e can be found in the Supplementary Data 1 file and can be used
as an example dataset for CRISPRessoBatch analysis (Steps 83–87). Base-editing efficiencies are
consistent across multiple delivery methods (plasmid, mRNA or protein; Fig. 8f,g), providing flex-
ibility in selecting a strategy that exposes target cells to base editors for a time frame that optimizes
on- to off-target editing ratios.
Reporting Summary
Further information on research design is available in the Nature Research Reporting Summary
linked to this article.
Data availability
The data that support the test findings in this study are available from the corresponding author upon
reasonable request. The protein expression vector shown in Supplementary Data 1 has been deposited
to Addgene. A subset of the data used to generate Fig. 8 can be found in Supplementary Data 2 and 3.
Raw HTS files have been deposited to the NCBI Sequence Read Archive (PRJNA655949).
References
1. Kim, Y. G., Cha, J. & Chandrasegaran, S. Hybrid restriction enzymes: zinc finger fusions to Fok I cleavage
domain. Proc. Natl Acad. Sci. USA 93, 1156–1160 (1996).
2. Wolfe, S. A., Nekludova, L. & Pabo, C. O. DNA recognition by Cys2His2 zinc finger proteins. Annu. Rev.
Biophys. Biomol. Struct. 29, 183–212 (2000).
3. Bibikova, M., Beumer, K., Trautman, J. K. & Carroll, D. Enhancing gene targeting with designed zinc finger
nucleases. Science 300, 764 (2003).
4. Miller, J. C. et al. An improved zinc-finger nuclease architecture for highly specific genome editing.
Nat. Biotechnol. 25, 778–785 (2007).
5. Porteus, M. H. & Baltimore, D. Chimeric nucleases stimulate gene targeting in human cells. Science 300,
763 (2003).
6. Boch, J. et al. Breaking the code of DNA binding specificity of TAL-type III effectors. Science 326,
1509–1512 (2009).
7. Moscou, M. J. & Bogdanove, A. J. A simple cipher governs DNA recognition by TAL effectors. Science 326,
1501 (2009).
8. Christian, M. et al. Targeting DNA double-strand breaks with TAL effector nucleases. Genetics
186, 757–761 (2010).
9. Miller, J. C. et al. A TALE nuclease architecture for efficient genome editing. Nat. Biotechnol. 29,
143–148 (2011).
10. Mahfouz, M. M. et al. De novo-engineered transcription activator-like effector (TALE) hybrid nuclease
with novel DNA binding specificity creates double-strand breaks. Proc. Natl Acad. Sci. USA 108,
2623–2628 (2011).
11. Li, T. et al. TAL nucleases (TALNs): hybrid proteins composed of TAL effectors and FokI DNA-cleavage
domain. Nucleic Acids Res. 39, 359–372 (2011).
12. Jinek, M. et al. A programmable dual-RNA-guided DNA endonuclease in adaptive bacterial immunity.
Science 337, 816–821 (2012).
13. Jansen, R., Embden, J. D., Gaastra, W. & Schouls, L. M. Identification of genes that are associated with DNA
repeats in prokaryotes. Mol. Microbiol. 43, 1565–1575 (2002).
14. Gasiunas, G., Barrangou, R., Horvath, P. & Siksnys, V. Cas9-crRNA ribonucleoprotein complex
mediates specific DNA cleavage for adaptive immunity in bacteria. Proc. Natl Acad. Sci. USA 109,
E2579–E2586 (2012).
15. Cong, L. et al. Multiplex genome engineering using CRISPR/Cas systems. Science 339, 819–823 (2013).
16. Jeggo, P. A. DNA breakage and repair. in Advances in Genetics Vol. 38 (eds Hall, J. et al.) 185–218
(Academic Press, 1998).
17. Rouet, P., Smih, F. & Jasin, M. Introduction of double-strand breaks into the genome of mouse cells by
expression of a rare-cutting endonuclease. Mol. Cell. Biol. 14, 8096–8106 (1994).

18. Lukacsovich, T., Yang, D. & Waldman, A. S. Repair of a specific double-strand break generated within a
mammalian chromosome by yeast endonuclease I-Scel. Nucleic Acids Res. 22, 5649–5657 (1994).
19. Yeh, C. D., Richardson, C. D. & Corn, J. E. Advances in genome editing through control of DNA repair
pathways. Nat. Cell Biol. 21, 1468–1478 (2019).
20. Haapaniemi, E., Botla, S., Persson, J., Schmierer, B. & Taipale, J. CRISPR–Cas9 genome editing induces a
p53-mediated DNA damage response. Nat. Med. 24, 927–930 (2018).
21. Ihry, R. J. et al. p53 inhibits CRISPR–Cas9 engineering in human pluripotent stem cells. Nat. Med. 24,
939–946 (2018).
22. Shen, M. W. et al. Predictable and precise template-free CRISPR editing of pathogenic variants. Nature 563,
646–651 (2018).
23. van Overbeek, M. et al. DNA repair profiling reveals nonrandom outcomes at Cas9-mediated breaks. Mol.
Cell 63, 633–646 (2016).
24. Koike-Yusa, H., Li, Y., Tan, E. P., Velasco-Herrera, M. D. C. & Yusa, K. Genome-wide recessive genetic
screening in mammalian cells with a lentiviral CRISPR-guide RNA library. Nat. Biotechnol. 32,
267–273 (2014).
25. Kosicki, M., Tomberg, K. & Bradley, A. Repair of double-strand breaks induced by CRISPR–Cas9 leads to
large deletions and complex rearrangements. Nat. Biotechnol. 36, 765–771 (2018).
26. Adikusuma, F. et al. Large deletions induced by Cas9 cleavage. Nature 560, E8–E9 (2018).
27. Jasin, M. & Rothstein, R. Repair of strand breaks by homologous recombination. Cold Spring Harb. Perspect.
Biol. 5, a012740 (2013).
28. Paquet, D. et al. Efficient introduction of specific homozygous and heterozygous mutations using CRISPR/
Cas9. Nature 533, 125–129 (2016).
29. Rees, H. A., Yeh, W.-H. & Liu, D. R. Development of hRad51–Cas9 nickase fusions that mediate HDR
without double-stranded breaks. Nat. Commun. 10, 2212 (2019).
30. Richardson, C. D., Ray, G. J., Dewitt, M. A., Curie, G. L. & Corn, J. E. Enhancing homology-directed
genome editing by catalytically active and inactive CRISPR-Cas9 using asymmetric donor DNA. Nat.
Biotechnol. 34, 339–344 (2016).
31. Lin, S., Staahl, B. T., Alla, R. K. & Doudna, J. A. Enhanced homology-directed human genome engineering
by controlled timing of CRISPR/Cas9 delivery. eLife 3, e04766 (2014).
32. Landrum, M. J. et al. ClinVar: public archive of relationships among sequence variation and human
phenotype. Nucleic Acids Res. 42, D980–D985 (2013).
33. Landrum, M. J. et al. ClinVar: public archive of interpretations of clinically relevant variants. Nucleic Acids
Res. 44, D862–D868 (2015).
34. Stenson, P. D. et al. The Human Gene Mutation Database: towards a comprehensive repository of inherited
mutation data for medical research, genetic diagnosis and next-generation sequencing studies. Hum. Genet.
136, 665–677 (2017).
35. Komor, A. C., Kim, Y. B., Packer, M. S., Zuris, J. A. & Liu, D. R. Programmable editing of a target base in
genomic DNA without double-stranded DNA cleavage. Nature 533, 420–424 (2016).
36. Gaudelli, N. M. et al. Programmable base editing of A•T to G•C in genomic DNA without DNA cleavage.
Nature 551, 464–471 (2017).
37. Mok, B. Y. et al. A bacterial cytidine deaminase toxin enables CRISPR-free mitochondrial base editing.
Nature 583, 631–637 (2020).
38. Nishimasu, H. et al. Crystal structure of Cas9 in complex with guide RNA and target DNA. Cell 156,
935–949 (2014).
39. Huang, T. P. et al. Circularly permuted and PAM-modified Cas9 variants broaden the targeting scope of
base editors. Nat. Biotechnol. 37, 626–631 (2019).
40. Jiang, W. et al. BE-PLUS: a new base editing tool with broadened editing window and enhanced fidelity. Cell
Res. 28, 855–861 (2018).
41. Ma, Y. et al. Targeted AID-mediated mutagenesis (TAM) enables efficient genomic diversification in
mammalian cells. Nat. Methods 13, 1029–1035 (2016).
42. Hess, G. T. et al. Directed evolution using dCas9-targeted somatic hypermutation in mammalian cells. Nat.
Methods 13, 1036–1042 (2016).
43. Liu, L. D. et al. Intrinsic nucleotide preference of diversifying base editors guides antibody ex vivo affinity
maturation. Cell Rep. 25, 884–892.3 (2018).
44. Wang, Y., Zhou, L., Liu, N. & Yao, S. BE-PIGS: a base-editing tool with deaminases inlaid into Cas9 PI
domain significantly expanded the editing scope. Signal Transduct. Target. Ther. 4, 36 (2019).
45. Kim, Y. B. et al. Increasing the genome-targeting scope and precision of base editing with engineered Cas9-
cytidine deaminase fusions. Nat. Biotechnol. 35, 371–376 (2017).
46. Grünewald, J. et al. Transcriptome-wide off-target RNA editing induced by CRISPR-guided DNA base
editors. Nature 569, 433–437 (2019).
47. Tan, J., Zhang, F., Karcher, D. & Bock, R. Engineering of high-precision base editors for site-specific single
nucleotide replacement. Nat. Commun. 10, 439 (2019).
48. Pluciennik, A. et al. PCNA function in the activation and strand direction of MutLα endonuclease in
mismatch repair. Proc. Natl Acad. Sci. USA 107, 16066–16071 (2010).
49. Heller, R. C. & Marians, K. J. Replisome assembly and the direct restart of stalled replication forks. Nat. Rev.
Mol. Cell Biol. 7, 932–943 (2006).

50. Kurt, I. C. et al. CRISPR C-to-G base editors for inducing targeted DNA transversions in human cells. Nat.
Biotechnol. https://doi.org/10.1038/s41587-020-0609-x (2020).
51. Zhao, D. et al. Glycosylase base editors enable C-to-A and C-to-G base changes. Nat. Biotechnol. https://doi.
org/10.1038/s41587-020-0592-2 (2020).
52. Kim, H. S., Jeong, Y. K., Hur, J. K., Kim, J.-S. & Bae, S. Adenine base editors catalyze cytosine conversions in
human cells. Nat. Biotechnol. 37, 1145–1148 (2019).
53. Arbab, M. et al. Determinants of base editing outcomes from target library analysis and machine learning.
Cell 182, 463–480.e30 (2020).
54. Nishida, K. et al. Targeted nucleotide editing using hybrid prokaryotic and vertebrate adaptive immune
systems. Science 353, aaf8729 (2016).
55. Komor, A. C. et al. 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 (2017).
56. Landrum, M. J. et al. ClinVar: improvements to accessing data. Nucleic Acids Res. 48(D1),
D835–D844 (2020).
57. Koblan, L. W. et al. Improving cytidine and adenine base editors by expression optimization and ancestral
reconstruction. Nat. Biotechnol. 36, 843–846 (2018).
58. Zafra, M. P. et al. Optimized base editors enable efficient editing in cells, organoids and mice. Nat.
Biotechnol. 36, 888–893 (2018).
59. Wang, L. et al. Enhanced base editing by co-expression of free uracil DNA glycosylase inhibitor. Cell Res. 27,
1289–1292 (2017).
60. Zhang, X. et al. Increasing the efficiency and targeting range of cytidine base editors through fusion of a
single-stranded DNA-binding protein domain. Nat. Cell Biol. 22, 740–750 (2020).
61. Cheng, T.-L. et al. Expanding C–T base editing toolkit with diversified cytidine deaminases. Nat. Commun.
10, 3612 (2019).
62. Li, C. et al. Targeted, random mutagenesis of plant genes with dual cytosine and adenine base editors.
Nat. Biotechnol. 38, 875–882 (2020).
63. Sakata, R. C. et al. Base editors for simultaneous introduction of C-to-T and A-to-G mutations. Nat.
Biotechnol. 38, 865–869 (2020).
64. Grünewald, J. et al. A dual-deaminase CRISPR base editor enables concurrent adenine and cytosine editing.
Nat. Biotechnol. 38, 861–864 (2020).
65. Zhang, X. et al. Dual base editor catalyzes both cytosine and adenine base conversions in human cells. Nat.
Biotechnol. 38, 856–860 (2020).
66. Levy, J. M. et al. Cytosine and adenine base editing of the brain, liver, retina, heart and skeletal muscle of
mice via adeno-associated viruses. Nat. Biomed. Eng. 4, 97–110 (2020).
67. Villiger, L. et al. Treatment of a metabolic liver disease by in vivo genome base editing in adult mice. Nat.
Med. 24, 1519–1525 (2018).
68. Yeh, W.-H. et al. In vivo base editing restores sensory transduction and transiently improves auditory
function in a mouse model of recessive deafness. Sci. Transl. Med. 12, eaay9101 (2020).
69. Ryu, S.-M. et al. Adenine base editing in mouse embryos and an adult mouse model of Duchenne muscular
dystrophy. Nat. Biotechnol. 36, 536–539 (2018).
70. Yang, L. et al. Amelioration of an inherited metabolic liver disease through creation of a de novo start codon
by cytidine base editing. Mol. Ther. 28, 1673–1683 (2020).
71. Kleinstiver, B. P. et al. Engineered CRISPR-Cas9 nucleases with altered PAM specificities. Nature 523,
481–485 (2015).
72. Kleinstiver, B. P. et al. Broadening the targeting range of Staphylococcus aureus CRISPR-Cas9 by modifying
PAM recognition. Nat. Biotechnol. 33, 1293–1298 (2015).
73. Kleinstiver, B. P. et al. High-fidelity CRISPR–Cas9 nucleases with no detectable genome-wide off-target
effects. Nature 529, 490–495 (2016).
74. Nishimasu, H. et al. Engineered CRISPR-Cas9 nuclease with expanded targeting space. Science 361,
1259–1262 (2018).
75. Hu, J. H. et al. Evolved Cas9 variants with broad PAM compatibility and high DNA specificity. Nature 556,
57–63 (2018).
76. Ran, F. A. et al. In vivo genome editing using Staphylococcus aureus Cas9. Nature 520, 186–191 (2015).
77. Walton, R. T., Christie, K. A., Whittaker, M. N. & Kleinstiver, B. P. Unconstrained genome targeting with
near-PAMless engineered CRISPR-Cas9 variants. Science 368, 290 (2020).
78. Miller, S. M. et al. Continuous evolution of SpCas9 variants compatible with non-G PAMs. Nat. Biotechnol.
38, 471–481 (2020).
79. Toth, E. et al. Improved LbCas12a variants with altered PAM specificities further broaden the genome
targeting range of Cas12a nucleases. Nucleic Acids Res. 48, 3722–3733 (2020).
80. Chatterjee, P., Jakimo, N. & Jacobson, J. M. Minimal PAM specificity of a highly similar SpCas9 ortholog.
Sci. Adv. 4, eaau0766 (2018).
81. Chatterjee, P. et al. An engineered ScCas9 with broad PAM range and high specificity and activity. Nat.
Biotechnol. 38, 1154–1158 (2020).
82. Chatterjee, P. et al. A Cas9 with PAM recognition for adenine dinucleotides. Nat. Commun. 11, 2474 (2020).
83. Cox, D. B. T., Platt, R. J. & Zhang, F. Therapeutic genome editing: prospects and challenges. Nat. Med. 21,
121–131 (2015).

84. Anzalone, A. V., Koblan, L. W. & Liu, D. R. Genome editing with CRISPR–Cas nucleases, base editors,
transposases and prime editors. Nat. Biotechnol. 38, 824–844 (2020).
85. Yang, L. et al. Increasing targeting scope of adenosine base editors in mouse and rat embryos through fusion
of TadA deaminase with Cas9 variants. Protein Cell 9, 814–819 (2019).
86. Kleinstiver, B. P. et al. Engineered CRISPR–Cas12a variants with increased activities and improved targeting
ranges for gene, epigenetic and base editing. Nat. Biotechnol. 37, 276–282 (2019).
87. Zhang, Y. et al. Programmable base editing of zebrafish genome using a modified CRISPR-Cas9 system. Nat.
Commun. 8, 118 (2017).
88. Chadwick, A. C., Wang, X. & Musunuru, K. In vivo base editing of PCSK9 (proprotein convertase
subtilisin/kexin type 9) as a therapeutic alternative to genome editing. Arterioscler. Thromb. Vasc. Biol. 37,
1741–1747 (2017).
89. Liu, Z. et al. Efficient generation of mouse models of human diseases via ABE- and BE-mediated base
editing. Nat. Commun. 9, 2338 (2018).
90. Wu, Y. et al. Increasing cytosine base editing scope and efficiency with engineered Cas9-PmCDA1 fusions
and the modified sgRNA in rice. Front. Genet. 10, 379 (2019).
91. Ren, B. et al. A CRISPR/Cas9 toolkit for efficient targeted base editing to induce genetic variations in rice.
Sci. China Life Sci. 60, 516–519 (2017).
92. Lee, H. K. et al. Simultaneous targeting of linked loci in mouse embryos using base editing. Sci. Rep. 9,
1662 (2019).
93. Gapinske, M. et al. CRISPR-SKIP: programmable gene splicing with single base editors. Genome Biol. 19,
107 (2018).
94. Gehrke, J. M. et al. An APOBEC3A-Cas9 base editor with minimized bystander and off-target activities.
Nat. Biotechnol. 36, 977–982 (2018).
95. Zhou, C. et al. Highly efficient base editing in human tripronuclear zygotes. Protein Cell 8, 772–775 (2017).
96. Yuan, J. et al. Genetic modulation of RNA splicing with a CRISPR-guided cytidine deaminase. Mol. Cell 72,
380–394.e7 (2018).
97. Hua, K., Tao, X. & Zhu, J.-K. Expanding the base editing scope in rice by using Cas9 variants. Plant
Biotechnol. J. 17, 499–504 (2019).
98. Li, X. et al. Base editing with a Cpf1–cytidine deaminase fusion. Nat. Biotechnol. 36, 324–327 (2018).
99. Lee, C. et al. CRISPR-pass: gene rescue of nonsense mutations using adenine base editors. Mol. Ther. 27,
1364–1371 (2019).
100. Liu, Z. et al. Efficient base editing with expanded targeting scope using an engineered Spy-mac Cas9 variant.
Cell Discov. 5, 58 (2019).
101. Ren, B. et al. Cas9-NG greatly expands the targeting scope of the genome-editing toolkit by recognizing NG
and other atypical PAMs in rice. Mol. Plant 12, 1015–1026 (2019).
102. Zhong, Z. et al. Improving plant genome editing with high-fidelity xCas9 and non-canonical PAM-targeting
Cas9-NG. Mol. Plant 12, 1027–1036 (2019).
103. Liu, Z. et al. Precise base editing with CC context-specificity using engineered human APOBEC3G-nCas9
fusions. BMC Biol. 18, 111 (2019).
104. Liu, Z. et al. Improved base editor for efficient editing in GC contexts in rabbits with an optimized AID-
Cas9 fusion. FASEB J. 33, 9210–9219 (2019).
105. Endo, M. et al. Genome editing in plants by engineered CRISPR–Cas9 recognizing NG PAM. Nat. Plants 5,
14–17 (2019).
106. Richter, M. F. et al. Phage-assisted evolution of an adenine base editor with improved Cas domain com-
patibility and activity. Nat. Biotechnol. 38, 883–891 (2020).
107. Hu, Z. et al. A compact Cas9 ortholog from Staphylococcus Auricularis (SauriCas9) expands the DNA
targeting scope. PLOS Biol. 18, e3000686 (2020).
108. Agudelo, D. et al. Versatile and robust genome editing with Streptococcus thermophilus CRISPR1-Cas9.
Genome Res. 30, 107–117 (2020).
109. Tan, J., Zhang, F., Karcher, D. & Bock, R. Expanding the genome-targeting scope and the site selectivity of
high-precision base editors. Nat. Commun. 11, 629 (2020).
110. Li, X. et al. Programmable base editing of mutated TERT promoter inhibits brain tumour growth. Nat. Cell
Biol. 22, 282–288 (2020).
111. Wang, X. et al. Cas12a base editors induce efficient and specific editing with low DNA damage response.
Cell Rep. 31, 107723 (2020).
112. Doman, J. L., Raguram, A., Newby, G. A. & Liu, D. R. Evaluation and minimization of Cas9-independent
off-target DNA editing by cytosine base editors. Nat. Biotechnol. 38, 620–628 (2020).
113. Slaymaker, I. M. et al. Rationally engineered Cas9 nucleases with improved specificity. Science 351,
84–88 (2016).
114. Rees, H. A. et al. Improving the DNA specificity and applicability of base editing through protein engi-
neering and protein delivery. Nat. Commun. 8, 15790 (2017).
115. Xu, W. et al. Multiplex nucleotide editing by high-fidelity Cas9 variants with improved efficiency in rice.
BMC Plant Biol. 19, 511 (2019).
116. Lee, J. K. et al. Directed evolution of CRISPR-Cas9 to increase its specificity. Nat. Commun. 9, 3048 (2018).
117. Liang, P. et al. Effective gene editing by high-fidelity base editor 2 in mouse zygotes. Protein Cell 8,
601–611 (2017).

118. Kim, D., Kim, D.-E., Lee, G., Cho, S.-I. & Kim, J.-S. Genome-wide target specificity of CRISPR RNA-guided
adenine base editors. Nat. Biotechnol. 37, 430–435 (2019).
119. Hong, R., Ma, S. & Wang, F. Improving the specificity of adenine base editor using high-fidelity Cas9.
Preprint at bioRxiv https://doi.org/10.1101/712109 (2019).
120. Casini, A. et al. A highly specific SpCas9 variant is identified by in vivo screening in yeast. Nat. Biotechnol.
36, 265–271 (2018).
121. Thuronyi, B. W. et al. Continuous evolution of base editors with expanded target compatibility and
improved activity. Nat. Biotechnol. 37, 1070–1079 (2019).
122. Yu, Y. et al. Cytosine base editors with minimized unguided DNA and RNA off-target events and high on-
target activity. Nat. Commun. 11, 2052 (2020).
123. Wang, X. et al. Efficient base editing in methylated regions with a human APOBEC3A-Cas9 fusion. Nat.
Biotechnol. 36, 946–949 (2018).
124. Gaudelli, N. M. et al. Directed evolution of adenine base editors with increased activity and therapeutic
application. Nat. Biotechnol. 38, 892–900 (2020).
125. Zhou, C. et al. Off-target RNA mutation induced by DNA base editing and its elimination by mutagenesis.
Nature 571, 275–278 (2019).
126. Rees, H. A., Wilson, C., Doman, J. L. & Liu, D. R. Analysis and minimization of cellular RNA editing by
DNA adenine base editors. Sci. Adv. 5, eaax5717 (2019).
127. Liu, Z. et al. Efficient base editing with high precision in rabbits using YFE-BE4max. Cell Death Dis.
11, 36 (2020).
128. Martin, A. S. et al. A panel of eGFP reporters for single base editing by APOBEC-Cas9 editosome com-
plexes. Sci. Rep. 9, 497 (2019).
129. St. Martin, A. et al. A fluorescent reporter for quantification and enrichment of DNA editing by
APOBEC–Cas9 or cleavage by Cas9 in living cells. Nucleic Acids Res. 46, e84 (2018).
130. Coelho, M. A. et al. BE-FLARE: a fluorescent reporter of base editing activity reveals editing characteristics
of APOBEC3A and APOBEC3B. BMC Biol. 16, 150 (2018).
131. Zuo, E. et al. A rationally engineered cytosine base editor retains high on-target activity while reducing both
DNA and RNA off-target effects. Nat. Methods 17, 600–604 (2020).
132. Zuo, E. et al. Cytosine base editor generates substantial off-target single-nucleotide variants in mouse
embryos. Science 364, 289–292 (2019).
133. Grünewald, J. et al. CRISPR DNA base editors with reduced RNA off-target and self-editing activities. Nat.
Biotechnol. 37, 1041–1048 (2019).
134. Wang, T., Badran, A. H., Huang, T. P. & Liu, D. R. Continuous directed evolution of proteins with improved
soluble expression. Nat. Chem. Biol. 14, 972–980 (2018).
135. Rees, H. A. & Liu, D. R. Base editing: precision chemistry on the genome and transcriptome of living cells.
Nat. Rev. Genet. 19, 770–788 (2018).
136. Shimatani, Z. et al. Targeted base editing in rice and tomato using a CRISPR-Cas9 cytidine deaminase
fusion. Nat. Biotechnol. 35, 441–443 (2017).
137. Sasaguri, H. et al. Introduction of pathogenic mutations into the mouse Psen1 gene by Base Editor and
Target-AID. Nat. Commun. 9, 2892 (2018).
138. Farzadfard, F. et al. Single-nucleotide-resolution computing and memory in living cells. Mol. Cell 75,
769–780.e4 (2019).
139. Shevidi, S., Uchida, A., Schudrowitz, N., Wessel, G. M. & Yajima, M. Single nucleotide editing without DNA
cleavage using CRISPR/Cas9-deaminase in the sea urchin embryo. Dev. Dyn. 246, 1036–1046 (2017).
140. Banno, S., Nishida, K., Arazoe, T., Mitsunobu, H. & Kondo, A. Deaminase-mediated multiplex genome
editing in Escherichia coli. Nat. Microbiol. 3, 423–429 (2018).
141. Wang, Y. et al. MACBETH: multiplex automated Corynebacterium glutamicum base editing method. Metab.
Eng. 47, 200–210 (2018).
142. Xie, J. et al. Efficient base editing for multiple genes and loci in pigs using base editors. Nat. Commun. 10,
2852 (2019).
143. Zong, Y. et al. Efficient C-to-T base editing in plants using a fusion of nCas9 and human APOBEC3A. Nat.
Biotechnol. 36, 950–953 (2018).
144. Gammage, P. A., Moraes, C. T. & Minczuk, M. Mitochondrial genome engineering: the revolution may not
be CRISPR-ized. Trends Genet. 34, 101–110 (2018).
145. Molla, K. A. & Yang, Y. CRISPR/Cas-mediated base editing: technical considerations and practical appli-
cations. Trends Biotechnol. 37, 1121–1142 (2019).
146. Hess, G. T., Tycko, J., Yao, D. & Bassik, M. C. Methods and applications of CRISPR-mediated base editing
in eukaryotic genomes. Mol. Cell 68, 26–43 (2017).
147. Liu, Z. et al. Highly efficient RNA-guided base editing in rabbit. Nat. Commun. 9, 2717 (2018).
148. Li, Q. et al. CRISPR-Cas9-mediated base-editing screening in mice identifies DND1 amino acids that are
critical for primordial germ cell development. Nat. Cell Biol. 20, 1315–1325 (2018).
149. Yeh, W.-H., Chiang, H., Rees, H. A., Edge, A. S. B. & Liu, D. R. In vivo base editing of post-mitotic sensory
cells. Nat. Commun. 9, 2184 (2018).
150. Zeng, Y. et al. Correction of the Marfan syndrome pathogenic FBN1 mutation by base editing in human
cells and heterozygous embryos. Mol. Ther. 26, 2631–2637 (2018).

151. Li, G. et al. Highly efficient and precise base editing in discarded human tripronuclear embryos. Protein Cell
8, 776–779 (2017).
152. Liang, P. et al. Correction of β-thalassemia mutant by base editor in human embryos. Protein Cell 8,
811–822 (2017).
153. Zong, Y. et al. Precise base editing in rice, wheat and maize with a Cas9-cytidine deaminase fusion. Nat.
Biotechnol. 35, 438–440 (2017).
154. Lu, Y. & Zhu, J. K. Precise editing of a target base in the rice genome using a modified CRISPR/Cas9 system.
Mol. Plant 10, 523–525 (2017).
155. Park, D.-S. et al. Targeted base editing via RNA-guided cytidine deaminases in Xenopus laevis embryos. Mol.
Cells 40, 823–827 (2017).
156. Rossidis, A. C. et al. In utero CRISPR-mediated therapeutic editing of metabolic genes. Nat. Med. 24,
1513–1518 (2018).
157. Kim, K. et al. Highly efficient RNA-guided base editing in mouse embryos. Nat. Biotechnol. 35,
435–437 (2017).
158. Tanaka, S. et al. In vivo targeted single-nucleotide editing in zebrafish. Sci. Rep. 8, 1–11 (2018).
159. Qin, L. et al. High-efficient and precise base editing of C•G to T•A in the allotetraploid cotton (Gossypium
hirsutum) genome using a modified CRISPR/Cas9 system. Plant Biotechnol. J. 18, 45–56 (2020).
160. Chen, Y. et al. CRISPR/Cas9-mediated base-editing system efficiently generates gain-of-function mutations
in Arabidopsis. Sci. China Life Sci. 60, 520–523 (2017).
161. Li, J., Sun, Y., Du, J., Zhao, Y. & Xia, L. Generation of targeted point mutations in rice by a modified
CRISPR/Cas9 system. Mol. Plant 10, 526–529 (2017).
162. Li, Y. et al. Programmable single and multiplex base-editing in Bombyx mori using RNA-guided cytidine
deaminases. G3 (Bethesda) 8, 1701–1709 (2018).
163. Qin, W. et al. Precise A•T to G•C base editing in the zebrafish genome. BMC Biol. 16, 1–8 (2018).
164. Li, C. et al. Expanded base editing in rice and wheat using a Cas9-adenosine deaminase fusion. Genome Biol.
19, 59 (2018).
165. Liang, P. et al. Effective and precise adenine base editing in mouse zygotes. Protein Cell 9, 808–813 (2018).
166. Ma, Y. et al. Highly efficient and precise base editing by engineered dCas9-guide tRNA adenosine deaminase
in rats. Cell Discov. 4, 1–4 (2018).
167. Kang, B.-C. et al. Precision genome engineering through adenine base editing in plants. Nat. Plants 4,
427–431 (2018).
168. Hua, K., Tao, X., Yuan, F., Wang, D. & Zhu, J.-K. Precise A· T to G· C base editing in the rice genome. Mol.
Plant 11, 627–630 (2018).
169. Song, C.-Q. et al. Adenine base editing in an adult mouse model of tyrosinaemia. Nat. Biomed. Eng. 4,
125–130 (2020).
170. Yan, F. et al. Highly efficient A· T to G· C base editing by Cas9n-guided tRNA adenosine deaminase in rice.
Mol. Plant 11, 631–634 (2018).
171. Kuscu, C. et al. CRISPR-STOP: gene silencing through base-editing-induced nonsense mutations. Nat.
Methods 14, 710–712 (2017).
172. Billon, P. et al. CRISPR-mediated base editing enables efficient disruption of eukaryotic genes through
induction of STOP codons. Mol. Cell 67, 1068–1079.e4 (2017).
173. Tang, W. & Liu, D. R. Rewritable multi-event analog recording in bacterial and mammalian cells. Science
360, eaap8992 (2018).
174. Hwang, B. et al. Lineage tracing using a Cas9-deaminase barcoding system targeting endogenous L1 ele-
ments. Nat. Commun. 10, 1234 (2019).
175. Yin, K., Gao, C. & Qiu, J.-L. Progress and prospects in plant genome editing. Nat. Plants 3, 17107 (2017).
176. Anzalone, A. V. et al. Search-and-replace genome editing without double-strand breaks or donor DNA.
Nature 576, 149–157 (2019).
177. Heyer, W.-D., Ehmsen, K. T. & Liu, J. Regulation of homologous recombination in eukaryotes. Annu. Rev.
Genet. 44, 113–139 (2010).
178. Moynahan, M. E. & Jasin, M. Mitotic homologous recombination maintains genomic stability and sup-
presses tumorigenesis. Nat. Rev. Mol. Cell Biol. 11, 196–207 (2010).
179. Jiang, T. et al. Chemical modifications of adenine base editor mRNA and guide RNA expand its application
scope. Nat. Commun. 11, 1979 (2020).
180. Zeng, J. et al. Therapeutic base editing of human hematopoietic stem cells. Nat. Med. 26, 535–541 (2020).
181. Ran, F. A. et al. Genome engineering using the CRISPR-Cas9 system. Nat. Protoc. 8, 2281–2308 (2013).
182. Chen, J. S. et al. Enhanced proofreading governs CRISPR–Cas9 targeting accuracy. Nature 550,
407–410 (2017).
183. Kim, J. et al. Structural and kinetic characterization of Escherichia coli TadA, the wobble-specific tRNA
deaminase. Biochemistry 45, 6407–6416 (2006).
184. Gao, X. et al. Treatment of autosomal dominant hearing loss by in vivo delivery of genome editing agents.
Nature 553, 217–221 (2018).
185. Ma, X. et al. Analysis of error profiles in deep next-generation sequencing data. Genome Biol. 20, 50 (2019).
186. Petrackova, A. et al. Standardization of sequencing coverage depth in NGS: recommendation for detection
of clonal and subclonal mutations in cancer diagnostics. Front. Oncol. 9, 851 (2019).

187. Hwang, G.-H. et al. Web-based design and analysis tools for CRISPR base editing. BMC Bioinformatics 19,
542 (2018).
188. Clement, K. et al. CRISPResso2 provides accurate and rapid genome editing sequence analysis.
Nat. Biotechnol. 37, 224–226 (2019).
189. Kluesner, M. G. et al. EditR: a method to quantify base editing from Sanger sequencing. CRISPR J. 1,
239–250 (2018).
190. Tsai, S. Q. et al. GUIDE-seq enables genome-wide profiling of off-target cleavage by CRISPR-Cas nucleases.
Nat. Biotechnol. 33, 187–197 (2015).
191. Tsai, S. Q. et al. CIRCLE-seq: a highly sensitive in vitro screen for genome-wide CRISPR–Cas9 nuclease
off-targets. Nat. Methods 14, 607–614 (2017).
192. Kim, D. et al. Digenome-seq: genome-wide profiling of CRISPR-Cas9 off-target effects in human cells.
Nat. Methods 12, 237–243 (2015).
193. Gao, Z., Harwig, A., Berkhout, B. & Herrera-Carrillo, E. Mutation of nucleotides around the +1 position of
type 3 polymerase III promoters: the effect on transcriptional activity and start site usage. Transcription 8,
275–287 (2017).
194. Kim, S., Bae, T., Hwang, J. & Kim, J.-S. Rescue of high-specificity Cas9 variants using sgRNAs with matched
5′ nucleotides. Genome Biol. 18, 218 (2017).
195. Labun, K., Montague, T. G., Gagnon, J. A., Thyme, S. B. & Valen, E. CHOPCHOP v2: a web tool for the next
generation of CRISPR genome engineering. Nucleic Acids Res. 44, W272–W276 (2016).
196. Haeussler, M. et al. Evaluation of off-target and on-target scoring algorithms and integration into the guide
RNA selection tool CRISPOR. Genome Biol. 17, 148 (2016).
197. Bae, S., Park, J. & Kim, J. S. Cas-OFFinder: a fast and versatile algorithm that searches for potential off-target
sites of Cas9 RNA-guided endonucleases. Bioinformatics 30, 1473–1475 (2014).
Acknowledgements
We thank the Pattern team at the Broad Institute for data visualization assistance and preparation of the associated website; K. Zhao,
S.Miller, B. Mok and T. Blum for helpful discussions; A. Vieira for assistance editing the manuscript; and all Liu laboratory members and
alumni who contributed to the development of these methods. This work was supported by US NIH U01 AI142756, RM1 HG009490,
R35 GM118062 and HHMI. G.A.N. was supported by a Helen Hay Whiteney post-doctoral fellowship.
Author contributions
T.P.H., G.A.N. and D.R.L. developed the protocol. T.P.H. and G.A.N. designed and performed the test experiments and computational
analyses. D.R.L. designed and supervised the test experiments. T.P.H. and D.R.L. drafted the manuscript, and all authors contributed to
editing the manuscript.
Competing interests
The authors declare competing financial interests. D.R.L. is a consultant and co-founder of Prime Medicine, Beam Therapeutics, Pairwise
Plants and Editas Medicine, companies that use genome editing. The authors are co-inventors on patent applications on base editing.
Additional information
Supplementary information is available for this paper at https://doi.org/10.1038/s41596-020-00450-9.
Correspondence and requests for materials should be addressed to D.R.L.
Peer review information Nature Protocols thanks Sangsu Bae, Zhanjun Li and the other, anonymous, reviewer(s) for their contribution to
the peer review of this work.
Reprints and permissions information is available at www.nature.com/reprints.
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Received: 29 March 2017; Accepted: 20 October 2020;

Published online: 18 January 2021
Related links
Key references using this protocol
Komor, A. C. et al. Nature 533, 420–424 (2016): https://doi.org/10.1038/nature17946
Gaudelli, N. M. et al. Nature 551, 464–471 (2017): https://doi.org/10.1038/nature24644
Arbab, M. et al. Cell 182, 463–480.e430 (2020): https://doi.org/10.1016/j.cell.2020.05.037
Anzalone, A. V. et al. Nat. Biotechnol. 38, 824–844 (2020): https://doi.org/10.1038/s41587-020-0561-9

nature research | reporting summary
Field-specific reporting
Please select the one below that is the best fit for your research. If you are not sure, read the appropriate sections before making your selection.
Life sciences Behavioural & social sciences Ecological, evolutionary & environmental sciences
For a reference copy of the document with all sections, see nature.com/documents/nr-reporting-summary-flat.pdf
Life sciences study design

All studies must disclose on these points even when the disclosure is negative.
Sample size Example data in Figure 6a-e are n = 1, while the example data in Figure 6f-g are n = 3
Data exclusions n/a
Replication n/a
Randomization n/a
Blinding n/a
Reporting for specific materials, systems and methods

We require information from authors about some types of materials, experimental systems and methods used in many studies. Here, indicate whether each material,
system or method listed is relevant to your study. If you are not sure if a list item applies to your research, read the appropriate section before selecting a response.
Materials & experimental systems Methods

n/a Involved in the study n/a Involved in the study
Antibodies ChIP-seq
Eukaryotic cell lines Flow cytometry
Palaeontology and archaeology MRI-based neuroimaging
Animals and other organisms
Human research participants
Clinical data
Dual use research of concern
Flow Cytometry
Plots
Confirm that:
The axis labels state the marker and fluorochrome used (e.g. CD4-FITC).
The axis scales are clearly visible. Include numbers along axes only for bottom left plot of group (a 'group' is an analysis of identical markers).
All plots are contour plots with outliers or pseudocolor plots.
A numerical value for number of cells or percentage (with statistics) is provided.
Methodology
Sample preparation HEK293T cells, trypsinized and sorted in DMEM media
April 2020
Instrument Sony SH800 Cell Sorter
Software Collection: manufacturer's software (Sony), Analysis: FlowJo
Cell population abundance Sorting for GFP+ cells as a proxy for base editor expression. Population can vary depending on the minimal level of GFP
expression desired. All GFP+ cells constituted ~50% of the cells sorted (n=20,000).
2
Gating strategy SSC:FSC to gate for populations of live/dead cells, then FSC-H:FSC-A to exclude multiplets, then by GFP (SSC-A:FITC-A, with
nature research | reporting summary

the gate determined from a negative control population).
Tick this box to confirm that a figure exemplifying the gating strategy is provided in the Supplementary Information.
April 2020

Precision Genome Editing Using Cytosine

Uploaded by

Document Informationclick to expand document information

Copyright:

Available Formats

Precision Genome Editing Using Cytosine

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Precision Genome Editing Using Cytosine

Uploaded by

Copyright:

Available Formats

PROTOCOL

Precision genome editing using cytosine and

NATURE PROTOCOLS | VOL 16 | FEBRUARY 2021 | 1089–1128 | www.nature.com/nprot 1089

a Cytosine base editing b Adenine base editing

ssDNA-specific Wild-type TadA monomer

DNA target C DNA target A

DNA repair U DNA repair I

1090 NATURE PROTOCOLS | VOL 16 | FEBRUARY 2021 | 1089–1128 | www.nature.com/nprot

CBEs enable conversion of C•G-to-T•A

General improvements and modiﬁcations

NATURE PROTOCOLS | VOL 16 | FEBRUARY 2021 | 1089–1128 | www.nature.com/nprot 1091

Base editors with Cas domain variants

Base editors with deaminase domain variants

1092 NATURE PROTOCOLS | VOL 16 | FEBRUARY 2021 | 1089–1128 | www.nature.com/nprot

CRISPR-free base editing of mitochondrial DNA

Applications of base editing

NATURE PROTOCOLS | VOL 16 | FEBRUARY 2021 | 1089–1128 | www.nature.com/nprot 1093

Box 1 | Designing sgRNAs for base editing

Limitations of base editing

1094 NATURE PROTOCOLS | VOL 16 | FEBRUARY 2021 | 1089–1128 | www.nature.com/nprot

Nearby bystander Target base

Predicted Cas OTs Guide

Identifying candidate protospacers and Cas proteins

NATURE PROTOCOLS | VOL 16 | FEBRUARY 2021 | 1089–1128 | www.nature.com/nprot 1095

Start with high-activity BE

1096 NATURE PROTOCOLS | VOL 16 | FEBRUARY 2021 | 1089–1128 | www.nature.com/nprot

Narrow-window base editors Context-specific base editors

Example uses: localized C-to-G or A-to-G mutations

Bystander base Bystander base Nonspecific mutagenic base editors

Off-target locus Off-target transcript

Precise allele substitution

BE2, BE3, BE4, BE4-Gam,

Increased Km cytidine deaminases, BE4max, Target-AID/C-term, YE1-BE4max (138155), BE3-PAPAPAP (145036),

YE1-BE4max (138155), SECURE (123614-15),

Broad mutagenesis of target region

Broad editing window BE-PLUS, BE4max, FNLS, BE-PLUS (113025, 113029),

NATURE PROTOCOLS | VOL 16 | FEBRUARY 2021 | 1089–1128 | www.nature.com/nprot 1097

A 5′ PAM, only compatible

A 5′ PAM, only compatible

1098 NATURE PROTOCOLS | VOL 16 | FEBRUARY 2021 | 1089–1128 | www.nature.com/nprot

Identifying candidate deaminase domains

Identifying candidate base editor architectures

NATURE PROTOCOLS | VOL 16 | FEBRUARY 2021 | 1089–1128 | www.nature.com/nprot 1099

Table 1 | Base-editing deaminases and their properties

Approximate editing window (SpCas9 editor)b Sequence preferencec

Optimal efficiency Observed A C G T A C G T DNA RNA

1 10 41-42, 53, 94, 112

3 8 112, 122, 123

–1 9 47, 53-54, 94, 112

YEE 6 45, 47, 94, 98

5 7 103, 112, 122-123

YE1 5 7 45, 94, 98, 112, 125

SECURE 5 6 46, 112, 122, 131,

3 9 94, 112, 121-123, 125

1100 NATURE PROTOCOLS | VOL 16 | FEBRUARY 2021 | 1089–1128 | www.nature.com/nprot

CBE Improved base editor expression and editing efficiency

32 aa 4 aa Improved base editor expression and editing efficiency 58

32 aa 32 aa 93 aa SPACE only contains