Genome editing with programmable base editors in human cells

Abstract

Genome editing has garnered significant attention over the last decade, resulting in a massive expansion of the genome engineering toolbox. Base editors encompass a class of tools that enable installing single-nucleotide changes in genomic DNA without the use of double-strand breaks. With the ever-increasing development of new and/or improved base editor systems, it is easy to be overwhelmed by the abundance of options. Here, we provide clear guidance to facilitate the selection of a base editor and to design guide RNAs (gRNAs) to suit various needs. Additionally, we describe in detail how to generate gRNA plasmids, transfect various mammalian cell types, and evaluate editing efficiencies. Finally, we give alternative methods and troubleshooting tips for some common pitfalls encountered during base editing.

1. Key resources table

Resources will vary by protocol. This table provides a summary of key resources needed to conduct every protocol included within this document.

Reagent or Resource	Source	Identifier
Bacterial and Virus Strains
Mix&Go E. Coli DH5α	Zymo	T3009
Biological Samples
KOLF 2.1J iPSC cell line (Pantazis et al., 2022)	Pantazis et al., 2022	N/A
HEK293T cell line	ATCC	CRL–3216
K562-s cell line	ATCC	CRL–3343
Chemicals, Peptides, and Recombinant Proteins
2xYT bacterial medium and agar plates
Tryptone	Fisher Chemical	BP1421–500
Yeast extract	VWR	97063–370
Sodium chloride	Macron Fine Chemicals	7532–06
Carbenicillin	GoldBio	C–103–100
Agar	VWR	J637–1KG
Polybrene	MilliporeSigma™	TR-1003G
Sodium Dodecyl Sulfate (SDS), 10%	VWR	0227–500G
Agarose gel
Agarose	VWR	97062–250
TAE Buffer
Tris	Thermo Scientific™	17926
Acetic acid	Fisher Chemical	A38SI–212
EDTA	Thermo Scientific™	17892
Gibco™ Penicillin-Streptomycin (10,000 U/mL)	Life Technologies	15140122
Critical Commercial Assays/Reagents
T4 ligase buffer	New England Biolabs	B0202S
T4 PNK	New England Biolabs	M0201S or M0201L
T4 ligase	New England Biolabs	M0202S or M0202L
BsmBI-v2 Restriction Enzyme	New England Biolabs	R0739S or R0739L
ZymoPURE™ Plasmid Miniprep	Zymo	11–550B
Lipofectamine™ 2000	Invitrogen™	11668019
Gibco™ Opti-MEM™	Thermo Scientific™	31985062
Gibco™ DMEM	Thermo Scientific™	10566024
Gibco™ Fetal Bovine Serum (FBS)	Thermo Scientific™	A5256701
Gibco™ Phosphate Buffered Saline (PBS)	Thermo Scientific™	10010049
Neon™ Transfection System, 10 μL Kit	Thermo Scientific™	MPK1096
Accumax	Innovative Cell Technologies	AM105
BioCoat® Matrigel® 6-well Plate	Corning	354671
ACCUTASE™	STEMCELL™ Technologies	07920
mTeSR™ Plus	STEMCELL™ Technologies	100–0276
Lipofectamine™ Stem	Invitrogen™	STEM00003
Proteinase K	New England Biolabs	P8107S
Phusion® High-Fidelity DNA Polymerase (Kit)	New England Biolabs	M0530S or M0530L
Phusion® High-Fidelity DNA Polymerase
DMSO
Phusion® HF Buffer Pack
Phusion® GC Buffer Pack
Deoxynucleotide (dNTP) Solution Mix	New England Biolabs	N0447S or N0447L
Monarch® Spin DNA Gel Extraction Kit	New England Biolabs	T1120S or T1120L
Oligonucleotides
Forward and reverse oligonucleotides with corresponding spacer sequence (see Table 2)	IDT	https://www.idtdna.com/pages/products/custom-dna-rna/dna-oligos/custom-dna-oligos
Primers with NGS adapters (see Table 4)	IDT
Illumina-specific barcode primers (see Supplementary Table 2)	IDT
Plasmids
See Supplementary Table 1 for BE and gRNA plasmids	Addgene	Multiple – see Supplementary Table 1
Super piggyBac™	System Biosciences	PB210PA–1
Lentivirus packaging
pMD	Addgene	12259
psPAX2	Addgene	12260
Other
HyPure™ Cell Culture Grade Water	Cytiva	SH30529.02
Neon™ Electroporation System	Thermo Scientific™	MPK5000
NanoDrop™ 2000c UV-Vis Spectrophotometer	Thermo Scientific™	E112352
S3e Cell Sorter	BioRad	145–1006
Namocell® Pala™ Cell Sorter and Single Cell Dispenser	Bio-Techne	NI006
Falcon™ Round-Bottom Polystyrene Test Tubes with 35 μm Cell Strainer Snap Cap	Corning	352235

Table 2.

pU6-GG-gRNA oligonucleotide sequences. The adapters for the overhangs of the Golden Gate cloning are given and must be included when ordering the oligonucleotides for the spacer insert.

	Oligonucleotide	Sequence
If spacer sequence starts with G	Forward	CACC[spacer sequence]
	Reverse	AAAC[reverse complement of spacer sequence]
If spacer sequence starts with A, C, or T	Forward	CACCG[spacer sequence]
	Reverse	AAAC[reverse complement of spacer sequence]C

Table 4.

Round 1 primer adapter sequences. These sequences must be added to the 5′ end of the annealing region of the primers designed for amplification. Illumina-specific round 2 primers can be found in Supplementary Table 2.

Primer	Sequence
NGS fwd primer	ACACTCTTTCCCTACACGACGCTCTTCCGATCTNNNN [fwd annealing region]
NGS rev primer	TGGAGTTCAGACGTGTGCTCTTCCGATCT [rev annealing region]

2. Introduction

Since the mechanistic characterization of CRISPR-Cas9 systems as programmable nucleases in 2012, (Jinek et al., 2012) the field of genome editing has quickly expanded, with an abundance of new tools developed over the past decade. One such new class of tools is base editors (BEs), which take advantage of the programmability of CRISPR-Cas systems, and use deaminase enzymes to install single nucleotide changes in genomic DNA. Applications of BEs include, but are not limited to, disease modeling, evaluation of the impact of genetic variants, and the correction of disease-causing mutations.

Programmable DNA binding by Cas9 is facilitated by a short piece of RNA called a single guide RNA (sgRNA) or, more simply, guide RNA (gRNA). In the natural bacterial or archaeal CRISPR systems, the gRNA consists of two separate RNA molecules, the CRISPR RNA (crRNA) and the trans-activating CRISPR RNA (tracrRNA, pronounced “tracer” RNA). To facilitate a more streamlined use of the system, the 3′ end of the crRNA was joined to the 5′ end of the tracrRNA by a four-nucleotide linker sequence to produce the sgRNA construct that is used today (Jinek et al., 2012). The Cas9 protein first searches the genome for protospacer adjacent motifs (PAMs), which are short sequence motifs specific to the Cas9 homolog being used (Sternberg, Redding, Jinek, Greene, & Doudna, 2014). The most commonly used Cas9 homolog is the Streptococcus pyogenes Cas9 (Sp-Cas9), which natively recognizes an NGG (N = A, C, G, or T) PAM. Once the PAM sequence is located, the Cas9 protein unwinds the DNA directly 5′ to the PAM (called the “protospacer”, Fig. 1) and checks for sequence complementarity with the 20-nucleotide sequence at the 5′ end of the gRNA (called the “spacer”, Fig. 1). While wild-type Cas9 is an endonuclease that cleaves the DNA backbone on both strands, BE systems are fusions between nucleic acid editing enzymes and a partially catalytically inactivated Cas9 (called nickase Cas9, or nCas9), which cleaves only the DNA strand that is base-paired with the gRNA (Fig. 1), which prompts the cell to replace this strand and use the opposite as a template.

Fig. 1.

Overview of base editing. (A) Binding of the BE to genomic DNA is facilitated by sequence complementarity between the gRNA spacer and the genomic DNA (the protospacer). A protospacer adjacent motif (PAM) must also be present next to the protospacer (in this case, the PAM is NGG). Once the BE is bound to the DNA, a stretch of nucleotides on the PAM-distal end of the protospacer (the “base editing window”, shown in gray) is exposed to the ssDNA deaminase enzyme (a cytidine deaminase for CBEs, an adenosine deaminase for ABEs). The base editing window depends on the nCas9 homolog and the deaminase, see Fig. 2B for editing windows of specific BEs. The bases within the protospacer are numbered 1 (PAM-distal end) to 20 (PAM-proximal end) according to the illustration. A U·G intermediate (in the case of ABEs this is an I·T intermediate) is introduced into the genomic DNA following deamination of target nucleotides within the base editing window and nCas9-induced nicking of the unedited strand. This intermediate is processed by the cell to result in an overall C‧G to T‧A conversion (in the case of ABEs this is an A‧T to G‧C conversion). A uracil glycosylase inhibitor (UGI) component is also included with CBEs to increase editing efficiencies. (B) Overview of mutations facilitated by ABEs and CBEs, with intermediates shown.

There are two main classes of BEs: cytosine base editors (CBEs) and adenine base editors (ABEs) (Gaudelli et al., 2017; Komor, Kim, Packer, Zuris, & Liu, 2016). CBEs facilitate the conversion of C·G base pairs to T·A (Fig. 1) and generally consist of nCas9, a cytidine deaminase enzyme, and a uracil DNA glycosylase inhibitor (UGI, Fig. 1). After nCas9 binds to the genomic DNA, the cytidine deaminase enzyme converts target cytidines to uracils (U), which are read as thymine by DNA replication and repair machinery (Fig. 1). The cytidine deaminase enzyme can only deaminate cytidines within accessible single-stranded DNA, thus the “base editing window” is located on the PAM-distal end of the DNA strand that lacks a base-pairing partner (Fig. 1A). It is important to note that the base editing window is generally a stretch of four to eight nucleotides (See Fig. 1A, bottom left panel). However, the enzyme may edit outside of this window in low frequencies. If multiple Cs are present within this window (the target C as well as neighboring “bystander” Cs), all may become edited concurrently in a process called “bystander editing”. This will be important later on for judicious gRNA design. The UGI component of CBEs increases the editing efficiency by preventing the endogenous DNA repair protein uracil N-glycosylase (UNG) from excising the uracil intermediate. ABEs work analogously to CBEs but use an adenosine deaminase enzyme, and therefore convert A·T base pairs to G·C through an inosine (I) intermediate (Fig. 1). Furthermore, the use of ABEs does not require an equivalent UGI component due to inefficient inosine excision by the cell’s DNA repair machinery. The use of nCas9 to nick the gRNA-paired, unmodified DNA strand promotes the cell to preferentially replace this strand and, in the process, use the U or I as a template, increasing overall base editing yields.

Whether using base editing to model a genetic disease, to study the consequences of genetic variants, or to develop a therapeutic, it is valuable to navigate through the suite of tools available and best practices when employing them. In this protocol, we discuss the process for selecting the appropriate BE and designing the most suitable gRNAs for a given target of interest. Additionally, we describe approaches for reliable generation and delivery of gRNAs (together with BEs) into various cell types. Finally, we provide guidelines for increasing base editing efficiencies and analyzing base editing outcomes.

3. Selecting a base editor

Selecting the appropriate BE for a given target is the first step in designing a base editing experiment. However, the choice may seem overwhelming given the repertoire of BEs; there are currently over 700 BE constructs on Addgene for use in mammalian cells. Here we will simplify the process of selecting a BE to enable even non-specialists to design their own experiments (Method 1). Construct maps of the basic architectures of most ABEs and CBEs are shown in Fig. 2A, and consist of two main components – the Cas protein and the deaminase. Many variations exist for each of these two components, resulting in a large number of combinations. We will discuss each component separately and distill all the apparent options down to 8 recommended BEs (4 CBEs and 4 ABEs). Note the two basic architectures that we recommend (BE4 for CBEs and ABE8 for ABEs) are comprised of an N-terminal deaminase fused to the Cas enzyme, with both N- and C-terminal nuclear localization signals (NLS). These basic architectures are among the most well-characterized, have optimal nuclear import, and have well-defined editing windows, described below. The BE4 architecture also employs two copies of UGI, which helps to increase editing efficiencies and minimize C·G to non-T·A editing outcomes when using CBEs, which can occur if intracellular UNG enzymes excise the U intermediate efficiently enough (Koblan et al., 2018). We also include a “P2A-GFP” motif in our constructs, in which GFP is co-expressed with the BE via the self-cleaving P2A sequence. This allows for evaluation of transfection efficiencies and selection of transfected cells via cell sorting methods.

Fig. 2.

Recommended Base Editors. (A) General base editor construct maps. ″BE4″ refers to the most optimized CBE architecture, while “ABE8″ refers to the most efficient ABE constructs. The Cas component can be replaced with any nCas9 or dCas12 variant, and the deaminase can be replaced with a variety of deaminases. The P2A-GFP enables visualization of BE expression. (B) Recommended deaminase options with sequence motif preferences and editing windows when used in conjunction with a Sp-Cas9 variant (Arbab et al., 2020; Cowan et al., 2024; Gaudelli et al., 2020; Koblan et al., 2018; Richter et al., 2020; Thuronyi et al., 2019; Xiao, Wu, & Tang, 2024). The target nucleotide is bolded and underlined. B = C, T, or G. Highest editing occurs in the middle of the editing window. Position 1 is furthest from the PAM and position 20 is closest to the PAM (illustrated in Fig. 1). (C) Recommended nCas9 variants and their protospacer adjacent motifs (PAMs) (Walton, Christie, Whittaker, & Kleinstiver, 2020). PAMs are in white.

The choice of Cas protein for a BE will have a major impact on the size of the base editing window and the number of potential protospacers that can be designed (via the presence of appropriate PAMs). Cas proteins evolved from bacteria and archaea as natural defense systems, and a multitude of Cas9 and Cas12 homologs have been identified and characterized for genome editing applications. Cas12 enzymes function similarly to Cas9 enzymes. While certain Cas12 enzymes are slightly smaller than some Cas9 enzymes, they cannot be mutated to cleave the appropriate DNA strand in the context of base editing. All Cas12-derived BEs must utilize completely catalytically dead Cas12 (dCas12) which does not nick the DNA strand that is base-paired with the gRNA. We thus do not recommend the use of Cas12-derived BEs, as their efficiencies are much lower than nCas9-derived systems. Of the many nCas9 enzymes, the Sp-nCas9 homolog is the most commonly used nCas9 protein for base editing due to its PAM flexibility, high efficiency in mammalian cells, and precise base editing window. Furthermore, researchers have engineered the “PAM-relaxed” Sp-nCas9 variants Sp-nCas9-NG (recognizes NG PAMs) and SpRY-nCas9 (recognizes NR [R = G or A], and to a lesser extent, NY [Y = C or T] PAMs), (Nishimasu et al., 2018; Walton et al., 2020) which offer the greatest flexibility for gRNA design (Fig. 2C).

For the purposes of this protocol piece, we will exclusively recommend BEs derived from Sp-nCas9-NG and SpRY-nCas9, with the choice between the two depending on the protospacer sequence (detailed in the next section). The use of BEs derived from wild-type Sp-nCas9 is also recommended, but identifying an appropriate BE gRNA that is compatible with an NGG PAM is less likely, and thus we have not included plasmid constructs for these BEs with this work. Sp-nCas9-derived BEs will generally have a base editing window of positions 3–9 within the protospacer, in which the PAM-distal end is 1 and the base directly next to the PAM is 20 (Fig. 1), with slight variations depending on the choice of deaminase. This editing window offers, in our opinion, an optimal balance between the likelihood of an optimally-positioned PAM and minimizing potential bystander editing. However, nCas9 homologs from other species have been repurposed as BEs and do offer certain benefits. For example, BEs derived from the KKH variant of Staphylococcus aureus (Sa, NNNRRT PAM) nCas9 homolog (Kleinstiver et al., 2015) have a smaller size overall (which provides cloning and delivery benefits), but have a more restrictive PAM and much wider editing windows (typically, positions 3 to 14), increasing the likelihood of bystander editing. While this may be a drawback for certain applications (for instance, in cases where editing of only a single A or C is desired), in some scenarios a wider editing window is either necessary or preferable.

In addition to the continued discovery of various Cas9 enzymes, there has been extensive work to discover, engineer, or evolve deaminases for use in base editing. The choice of deaminase for a BE will have a major impact on editing efficiency for a given target sequence (as most deaminase enzymes have certain motif preferences and aversions) and the size of the base editing window. The adenosine deaminases used in ABEs were evolved from an E. coli tRNA modifying enzyme, as no naturally occurring ssDNA adenosine deaminase enzymes exist. As such, the deaminase options for ABEs are limited and we recommend either the TadA8e deaminase (which will have an editing window of positions 2 through 11 when used with Sp-nCas9) or the TadA8.20 deaminase (which will have a narrower editing window of positions 2 through 9 when used with Sp-nCas9) (Cowan et al., 2024; Gaudelli et al., 2020; Richter et al., 2020). While neither of these two evolved TadA enzymes has a particular sequence motif aversions, both prefer As that are present within TA sequence motifs. Further, TadA8e is slightly more efficient at editing As within CA motifs when compared to TadA8.20 (Fig. 2B) (Xiao et al., 2024). Cytidine deaminase enzymes are a naturally occurring family of proteins, and as such, there are many more potential options for deaminases in CBEs than ABEs. We have found the most ubiquitous cytidine deaminases for use in base editing to be the “ancestral APOBEC” variant (ancAPOBEC, which will have an editing window of positions 3 through 9 when used with Sp-nCas9) and the “evolved APOBEC” (evoAPOBEC), which will have an editing window of positions 3 through 8 when used with Sp-nCas9 (Koblan et al., 2018; Thuronyi et al., 2019). evoAPOBEC has no dominant motif preference, but has a slight aversion to Cs within AC motifs (Arbab et al., 2020), while ancAPOBEC prefers Cs within TC motifs and has an aversion to editing Cs within GC contexts (Thuronyi et al., 2019). A brief summary of editing windows and sequence preferences is shown in Fig. 2B, and the eight BEs that we recommend are summarized in Table 1.

Table 1.

Plasmid information for recommended BEs. All plasmids have been deposited on Addgene.

Type of BE	Construct name	Deaminase	nCas9
CBE	evoBE4-NG	evoAPOBEC	Sp-nCas9-NG
	evoBE4-SpRY	evoAPPOBEC	SpRY-nCas9
	ancBE4-NG	ancAPOBEC	Sp-nCas9-NG
	ancBE4-SpRY	ancAPOBEC	SpRY-nCas9
ABE	ABE8e-NG	TadA8e	Sp-nCas9-NG
	ABE8e-SpRY	TadA8e	SpRY-nCas9
	ABE8.20-NG	TadA8.20	Sp-nCas9-NG
	ABE8.20-SpRY	TadA8.20	SpRY-nCas9

As mentioned previously, when designing a base editing experiment, potential bystanders must always be identified and considered. Bystanders are A or C bases next to (generally, within 5 base pairs of) the target A or C base, which may get edited concurrently (or instead of) the desired base. This is because certain deaminases (such as TadA8e) are particularly processive and will deaminate multiple bases at a time. It is important to identify all potential bystander bases near the target editing site, recognize if such bystander editing is problematic or not (e.g., certain silent mutations may be acceptable), and note their sequence contexts as compared to the intended target base.

3.1. Method 1 – selecting a base editor

1. Identify which nucleotide conversion is desired: C•G to T•A or A•T to G•C

   1. C•G to T•A: use a CBE
      Note: If the mutation is listed as a G to A mutation on the coding strand, this is equivalent to a C to T mutation on the template strand, and a CBE can be used.
   2. A•T to G•C: use an ABE
      Note: If the mutation is listed as a T to C mutation on the coding strand, this is equivalent to a A to G mutation on the template strand, and an ABE can be used.

2. Note the sequence motif of the nucleotide to be edited. If this motif is one of the “disliked” motifs of a given deaminase, eliminate that deaminase as an option.

3. Identify potential bystander edits, their sequence motifs, and whether or not it is acceptable if they are edited.

4. If possible, identify a deaminase that prefers the motif of the target C or A and has an aversion to the motif of potential bystanders. If this is not possible, simply choose a deaminase that prefers or does not have an aversion to the motif of the target C or A.

5. Move to Method 2 (designing gRNAs) with both BEs (NG and SpRY) derived from the deaminase(s) of choice. The selection between NG and SpRY will depend on the protospacer(s).

Fig. 2 covers a minimal number of BEs, but includes those that are the most universal and thus will simplify the process of designing an initial base editing experiment, while maximizing the chances of successful genome editing. These eight BEs have been deposited to Addgene (Table 1, Supplementary Table 1). For a more comprehensive list of BEs, we recommend the 2020 review by Anzalone, Koblan, and Liu (Anzalone, Koblan, & Liu, 2020). New BEs are regularly in development and additional literature may be beneficial to determine whether another editor has been developed for specific needs not addressed by the more commonly used BEs.

4. Designing guide RNAs (gRNAs)

Designing a suitable gRNA for a given target is the next step. The gRNA consists of a constant 80-nt 3′ end, with a variable 20-nt 5′ end termed the spacer. The spacer must be designed based on the desired edit of interest, and is identical to the protospacer sequence within the DNA. While some tools exist to aid in spacer design, most were not developed specifically for BE gRNA design and thus we describe here processes for identifying and designing BE gRNA spacers manually, using either a “PAM-first” approach (Method 2a, Fig. 3) or an “edit-first” approach (Method 2b, Fig. 4). In the “PAM-first” approach, the identification of efficient PAMs (such as NG and NA) is prioritized, while in the “edit-first” approach, optimal positioning of the target C or A within the protospacer (positions 5, 6, and 7) is prioritized. These two methods may therefore identify slightly different gRNAs (as demonstrated below). While manually designing gRNAs for a manageable number of targets is practical, if a large library of gRNAs is desired, we recommend the “Base editor design tool” code written by Mudra Hegde and Ruth Hanna (https://github.com/mhegde/base-editor-design-tool) (Hanna et al., 2021).

Fig. 3.

PAM-first approach for gRNA design (Method 2a).

Fig. 4.

Edit-first approach for gRNA design (Method 2b).

4.1. Method 2a – designing a gRNA: PAM-first approach

1. Identify the strand on which the target C or A occurs (Fig. 3A and E).

2. Using a molecular biology visualization tool (e.g., Benchling), annotate the target C or A. Be sure to annotate the appropriate DNA strand (Fig. 3A and E).

3. Scan the bases that are 9 to 18 bases away from the target C or A on the 3′ side and locate and annotate all PAMs (NGG, NG, or NA motifs are preferred, but NC and NT can be used with SpRY-BEs, Fig. 3B and F).

4. For each identified PAM, annotate the 20 bases on the 5′ side of the PAM as a potential protospacer (Fig. 3C and G).

5. For each potential protospacer, determine at which position the target C or A is, and whether this is within the base editing window for the BE(s) that was chosen. Discard the protospacers in which the target C or A is outside of the editing window (Fig. 3D and H).

6. Pair each protospacer with the appropriate BE based on its PAM sequence (i.e., use either an NG- or SpRY-BE for any with NG PAMs, and use a SpRY-BE for any with NA PAMs) and the deaminase(s) that was identified previously (Fig. 3D and H).

4.2. Examples (Fig. 3)

See Fig. 3 to follow along with these two examples. In Fig. 3A, the target nucleotide for a C to T edit on the coding strand is shown in red. Note that the motif of the target C (GC) is a disliked motif of the ancAPOBEC deaminase. Thus, only evoAPOBEC-derived CBEs will be considered for this target. In Step 3 (Fig. 3B), PAMs are identified on the appropriate strand, 3′ of the target nucleotide. Preferred PAMs are underlined here (NGG, NG, and NA), with PAMs corresponding to Sp-nCas9 (NGG) and Sp-nCas9-NG (NG) underlined pink, and PAMs corresponding to SpRY-nCas9 (NA) underlined blue. Next, per Step 4, 20-nt protospacers for each PAM are shown in green (Fig. 3C). Finally, the position of the target nucleotide within each protospacer is identified (Step 5). As can be seen in Fig. 3D, the target nucleotide lies in position 8, 7, or 6 of the identified protospacers, which are all compatible with evoAPOBEC-derived CBEs (Fig. 2B). The C8 protospacer would be paired with evoBE4 (which uses wild-type nCas9) or evoBE4-NG, the C7 protospacer would be paired with evoBE4-NG, and the C6 protospacer would be paired with evoBE4-SpRY.

In Fig. 3E, the target nucleotide for a G to A edit on the coding strand is shown in blue. The C on the template strand (shown in red) that is base paired with the G is the actual target of the CBE that will yield the desired G•C to A•T conversion. Note that the motif of the target C (CC) is NOT a disliked motif of either CBE deaminases. Thus, both ancAPOBEC and evoAPOBEC-derived CBEs will be considered for this target. In Step 3 (Fig. 3F), PAMs are identified on the appropriate strand, 3′ of the target nucleotide. Preferred PAMs are underlined here (NGG, NG, and NA), with PAMs corresponding to Sp-nCas9 (NGG) and Sp-nCas9-NG (NG) underlined pink, and PAMs corresponding to SpRY-nCas9 (NA) underlined blue. Next, per Step 4, 20-nt protospacers for each PAM are shown in green (Fig. 3G). Finally, the position of the target nucleotide within each protospacer is identified (Step 5). As can be seen in Fig. 3H, the target nucleotide is in position 3, 5, or 6 of the identified protospacers, which are all compatible with both ancAPOBEC and evoAPOBEC-derived CBEs (Fig. 2B). In this example, there is a likely bystander edit, shown in orange, immediately 5′ of the target cytosine. However, this will result in a silent mutation, as both CAA and CAG code for Gln, which is the desired amino acid change in this example (the codon is indicated by a purple dotted outline). Furthermore, the motif of this bystander C (AC) is a disliked motif of the evoAPOBEC deaminase, and it is outside of the editing window in the C3 protospacer. It is therefore recommended to use evoAPOBEC-NG with the C6 protospacer, evoAPOBEC-SpRY with the C5 protospacer, and either evoAPOBEC-NG or ancAPOBEC-NG with the C3 protospacer.

4.3. Method 2b – designing a gRNA: edit-first approach

1. Identify the strand on which the target C or A occurs (Fig. 4A and D).

2. Using a molecular biology visualization tool (e.g., Benchling), annotate the target C or A. Be sure to annotate the appropriate DNA strand (Fig. 4A and D).

3. Design an initial protospacer in which the target C or A is at position 6: annotate the five bases 5′ of the target C or A, the target C or A, and the 14 bases 3′ of the target C or A (Fig. 4B and E).

4. Identify the PAM for this protospacer. This will determine which BE must be used. Be aware of less favorable PAMs, such as NC, which may work with SpRY-nCas9 but are not optimal (Fig. 4C and F).

5. Repeat steps 3 and 4, annotating additional protospacer sequences that place the target C or A within the editing window of potential BEs by shifting the initial protospacer one nucleotide to the 5′ or 3′ side at a time.

6. Pair each protospacer with the appropriate BE based on its PAM sequence (i.e. use either an NG- or SpRY-BE for any with NG PAMs, and use a SpRY-BE for any with NA PAMs) and the deaminase(s) that was identified previously. If no NG or NA PAMs exist that place the edit in a favorable position, SpRY BE may still be successful.

4.4. Examples (Fig. 4)

See Fig. 4 to follow along with these two examples. In Fig. 4A, the target nucleotide for a C to T edit on the coding strand is shown in red. Note that the motif of the target C (GC) is a disliked motif of the ancAPOBEC deaminase. Thus, only evoAPOBEC-derived CBEs will be considered for this target. In Step 3 (Fig. 4B), protospacers that place the target nucleotide in positions 5, 6, or 7 of the editing window are identified. These are underlined in green. This step could be repeated to find protospacers with the target nucleotide in additional positions within the editing window, determined by the deaminase being used (Fig. 2B), if needed. Next, per Step 4, PAMs for the spacers are identified (Fig. 4C). PAMs compatible with SpRY-nCas9 are underlined in blue, while those compatible with Sp-nCas9-NG are in pink. The PAM for the C5 protospacer is NC, which is a less favorable option as the SpRY-nCas9 variant is less efficient at binding to these targets. The C7 protospacer would be paired with evoBE4-NG, and the C6 and C5 protospacers would be paired with evoBE4-SpRY.

In Fig. 4D, the target nucleotide for a G to A edit on the coding strand is shown in blue. The C on the template strand (shown in red) that is base paired with the G is the actual target of the CBE that will yield the desired G•C to A•T conversion. Note that the motif of the target C (CC) is NOT a disliked motif of either CBE deaminases. Thus, both ancAPOBEC and evoAPOBEC-derived CBEs will be considered for this target. In Step 3 (Fig. 4B), protospacers that place the target nucleotide in positions 5, 6, or 7 of the editing window are identified. These are underlined in green. This step could be repeated to find protospacers with the target nucleotide in additional positions within the editing window, determined by the deaminase being used (Fig. 2E), if needed. Next, per Step 4, PAMs for the spacers are identified (Fig. 4F). PAMs compatible with SpRY-nCas9 are underlined in blue, while those compatible with S-npCas9-NG are in pink. The PAM for the C7 protospacer is NC, which is a less favorable option as the SpRY-nCas9 variant is less efficient at binding to these targets. As can be seen in Fig. 4F, there is a likely bystander edit, shown in orange, that will occur to the cytosine immediately 5′ of the target cytosine. However, this will result in a silent mutation, as both CAA and CAG code for Gln, which is the desired amino acid change in this example (this codon is indicated by a purple dotted outline). Furthermore, the motif of this bystander C (AC) is a disliked motif of the evoAPOBEC deaminase. It is therefore recommended to use evoAPOBEC-NG with the C6 protospacer, and evoAPOBEC-SpRY with the C5 and C7 protospacers.

After a potential protospacer is selected, it is important to make several assessments. First, confirm that the target nucleotide is on the appropriate DNA strand (the target C or A should be found within the protospacer sequence, and not its reverse complement), and is within the editing window of the BE that it is being paired with (Fig. 2B). Next, identify if there are nearby bystanders within the editing window that, if edited, would be unacceptable (e.g., non-silent mutations, splice site mutations). Lastly, if there are unacceptable bystanders, consider using an alternative protospacer that positions the bystander bases in a less accessible position. For each target base, there may be multiple BE:gRNA options. We recommend trying all options. If there are none or only a single potential protospacer, repeat the process and look for NC (preferred) or NT PAMs, to use with a SpRY-BE variant.

Once the protospacers are selected, the corresponding spacer sequence should be cloned into a plasmid backbone with the rest of the gRNA backbone. A variety of gRNA-expressing plasmids are available from Addgene, including pU6-GG-gRNA (detailed in Fig. 5A), which we recommend. pU6-GG-gRNA contains a dual Type IIS restriction enzyme (BsmBI) site between the U6 promoter and the gRNA backbone, also called a Golden Gate (GG) site. This enables custom spacer sequences to be easily ligated at this site (Fig. 5B). Type IIS restriction enzymes cleave outside of their recognition sequence, allowing for scarless restriction enzyme cloning known as Golden Gate cloning (Engler, Kandzia, & Marillonnet, 2008). To facilitate assembly into the digested backbone, custom overhangs must be appended on either side of the spacer sequence. Note that the U6 promoter requires a 5′G to initiate transcription. If your spacer sequence naturally starts with a G, this is sufficient for intracellular gRNA transcription. If your spacer does not, an extra 5′G must be added. For pU6-GG-gRNA, the forward and reverse oligonucleotides in Table 2 must be ordered for each custom spacer.

Fig. 5.

Construct maps of plasmids recommended for gRNA cloning (top) and for enriching for cells with high BE activity (bottom four) are shown in (A). pU6-GG-gRNA is recommended for gRNA cloning. It contains a dual BsmBI site for restriction enzyme digestion and ligation of custom spacer sequences, as shown in (B). It also contains a “UA flip” in the gRNA backbone which reduces polymerase stalling during transcription. pCBE-dGFP and pABE-dGFP are plasmids with non-functional GFP (dGFP) which are converted to functional GFP when a critical mutation (Y93H or A11V) is successfully base edited back to wild type. Matching gRNA expressing plasmids that target the appropriate BE to the deactivating mutation (pCBE-dGFP-gRNA and pABE-dGFP-gRNA, respectively) are also shown.

The U6 promoter utilizes RNA polymerase III (RNAPIII) for expression of high levels of the gRNA. RNAPIII recognizes poly-T stretches of 4–6 thymines as termination signals (Artsimovitch & Belogurov, 2015), which may cause issues with certain spacers. If one or more of the designed spacer sequences contains a poly-T stretch, it may cause issues with expression in cells. To overcome this, we recommend the use of synthetic gRNAs, which can be purchased commercially (e.g., from IDT, Synthego, or Thermo Fisher Scientific). These can be transfected into cells alongside BE-encoding mRNA (optimally) or a BE-encoding plasmid. Notably, the natural Sp-gRNA backbone sequence has a poly-T stretch immediately 3′ of the spacer sequence. Therefore, to improve intracellular gRNA transcription, the gRNA backbone sequence in pU6-GG-gRNA incorporates two single-nucleotide changes known as the “UA flip” (Chen et al., 2016).

4.5. Method 3 – anneal and phosphorylate oligonucleotides

4.5.1. Materials and equipment

• Thermocycler

• Micropipettes

• Ice bucket

• 0.2 mL PCR tubes

• Forward and reverse oligonucleotides (25 nmol scale, standard desalting purification specifications from IDT are recommended)

• Sterile water

• T4 ligase buffer

• T4 PNK

4.5.2. Protocol

1. Resuspend any new, lyophilized oligonucleotides to a working concentration of 100 μM in sterile water.

2. On ice, mix the following, per reaction (i.e., per protospacer):

   Component/Reagent	Volume (μL)
   Sterile water	16.5
   10X T4 ligase buffer*	2
   T4 PNK	0.5
   Forward oligo (100 μM)	0.5
   Reverse oligo (100 μM)	0.5
   Total Volume	20

   ^*

   T4 ligase buffer is used instead of T4 PNK buffer as it includes ATP, which is necessary for the reaction to proceed.

3. Using a thermocycler, incubate the reaction for 30 min at 37 °C, followed by 95 °C for 1 min. Ramp the temperature down to either 12 °C or 4 °C by 0.1 °C per second.

4. This product can be stored at 4 °C for up to two weeks or at –20 °C for up to a year.

4.6. Method 4 - golden gate cloning (Fig. 5B)

4.6.1. Materials and equipment

• Thermocycler or heat blocks set for 42 °C and 60 °C

• Micropipettes

• Ice bucket

• 0.2 mL PCR tubes or 1.5 mL microcentrifuge tubes

• Annealed and phosphorylated oligonucleotides from Method 3

• Sterile water

• T4 ligase buffer

• T4 ligase

• Backbone plasmid with Golden Gate site (pU6-GG-gRNA recommended)

• Compatible Type IIS restriction enzyme (BsmBI if using pU6-GG-gRNA)

4.6.2. Protocol

1. On ice, mix the following, per reaction (i.e., per protospacer):

   Component/Reagent	Volume (μL)
   Sterile water	to a total volume of 10
   10X T4 ligase buffer	1
   T4 ligase	0.5
   BsmBI-v2 (or other Type IIS restriction enzyme)	0.5
   Backbone	0.01 pmol
   Insert (from Method 3)	0.05 pmol
   Total Volume	10

2. Incubate the reaction for 30 min at 42 °C.

   Note: If there are issues with efficiency, the reaction can be cycled on a thermocycler as follows: 42 °C for 5 min, 16 °C for 5 min, 4–6 cycles.

3. Transfer the reaction to 60 °C for 10 min.

4. This product can be stored between 4 °C and −20 °C overnight, depending on the restriction enzyme used. For best results proceed to the transformation step immediately.

4.7. Method 5 – transformation, inoculation, and sequencing

4.7.1. Materials and equipment

• Micropipettes

• Ice bucket

• 0.2 mL PCR tubes or 1.5 mL microcentrifuge tubes

• Heat block

• Bacterial loops or spreaders

• Bacterial incubator set to 37 °C

• Bacterial incubator/shaker set to 37 °C

• Competent bacterial cells (E. Coli DH5α are recommended for most plasmids)

• Bacterial medium (LB or 2XYT) with appropriate antibiotics (ampicillin or carbenicillin for pU6-GG-gRNA)

• Agar plate(s) with appropriate growth medium (LB or 2XYT) and antibiotics (ampicillin or carbenicillin for pU6-GG-gRNA)

• Miniprep kit (e.g., ZymoPURE^™ Plasmid Miniprep)

4.7.2. Protocol

This protocol uses Zymo’s Mix&Go! competent E. Coli DH5α cells. A heat shock step might be necessary if other types of competent cells are used.

1. Thaw cells on ice.

2. Add 5 μL of Method 4 product to 50 μL cells and gently flick the tube to mix.

   Notes:

   1. If the transformation efficiency is too high, less Golden Gate product can be added. However, it is not recommended to add less than 1 μL of product.
   2. Instead of reducing the volume added, the product can be diluted prior to transformation.

3. Incubate the cells on ice for 5 min if using Zymo Mix&Go! cells.

   Note: If using competent cells other than Zymo Mix&Go!, increase incubation time to 10–20 min.

4. If necessary, heat shock cells for 75 s at 55 °C, and incubate on ice for another 2 min after.

5. Spread the cells on a pre-warmed bacterial plate with 100 μg/mL carbenicillin or ampicillin.

   Notes:

   1. The type of antibiotic used may vary depending on the backbone. pU6-GG-gRNA contains ampicillin resistance.
   2. Recovery may be necessary if using an antibiotic other than carbenicillin or ampicillin.
   3. If your transformation efficiency is too high, dilute the transformed bacteria prior to plating or plate a smaller volume.

6. Place in 37 °C incubator for 12–20 h.

7. Upon successful cloning and transformation, select two or more colonies and inoculate each in 5 mL media with appropriate antibiotics.

8. Place cultures in a shaker at 37 °C for 12–16 h.

9. Isolate the plasmids from the cells using a miniprep kit or equivalent.

10. Send an aliquot of the isolated plasmid for sequencing to ensure the cloning was successful.

    Notes:

    1. If using a backbone for the first time or that has been in storage for a while, whole plasmid sequencing is recommended to ensure there are no unexpected mutations.
    2. If using a backbone that has been recently used and sequence-verified, only the protospacer region needs to be sequence-verified. Design a primer that is more than 100 bp away from the beginning of the protospacer to use for Sanger sequencing. The suggested Sanger sequencing primer for pU6-GG-gRNA is 5′-TAC GTG ACG TAG AAA GTA AT-3′.
    3. If needing to sequence more than 700 bp via Sanger sequencing, generate tiled primers or a reverse primer to ensure robust sequence data of the area that you desire.

It is important to ensure that any plasmid that will be used with a mammalian cell line is endotoxin-free. Most plasmid isolation kits come with some method of removing endotoxins from the isolation. It is imperative that this step is performed. If endotoxins are present, transfection efficiencies will be severely reduced and decreased cell viability is highly likely. Additionally, any results from assays performed with plasmids containing endotoxins are unreliable.

While this method is quite efficient for generating gRNA plasmids, there are some common pitfalls to be aware of. First, if the backbone:insert ratio is not optimal, both the transformation efficiency and likelihood of obtaining colonies with the desired plasmid sequence will suffer. The insert should be more plentiful than the backbone. In this case, the insert is the annealed and phosphorylated oligonucleotides from Method 3, and a backbone:insert ratio of 1:5 (molar ratio) has been reliably successful. Additional decreases in transformation efficiency may be seen if the competent bacterial cells are mishandled. Thaw competent cells on ice and do not introduce bubbles or vortex the cells. Another variable that may impact transformation efficiency is the amount of DNA introduced to the competent cells. Sometimes, adding a smaller amount of ligation product may yield greater transformation efficiencies. If efficiency is poor, vary the amount of ligation mixture that is transformed.

5. Transfecting mammalian cells and harvesting DNA

To effectively perform genome editing, the gRNA and BE selected and generated in the previous sections must be delivered into the cells of interest. The most common method for this process is using a cationic lipid-based reagent, such as Lipofectamine^™, to perform a transfection. These cationic lipids can be used to perform both forward and reverse transfections (described below), and can effectively deliver plasmid DNA as well as other nucleic acids such as siRNA, mRNA, synthetic gRNA, and even purified Cas9 protein if it is first complexed with a gRNA.

We describe in detail an optimal method for the transfection of plasmids encoding BE and gRNA into HEK293T cells in a 48-well plate format using Lipofectamine^™ 2000 (L2000). However, if working with other cell types, both the amount of DNA used and the ratio of L2000 to DNA should be optimized. Additionally, even if using HEK293T cells, the amount of DNA used per well does not scale linearly based on surface area and should be optimized when scaling up or down. Finally, different cationic lipid reagents should be evaluated if working with other cell types, or if delivering nucleic acids other than plasmid DNA (such as synthetic gRNA and mRNA).

An additional consideration for transfections is cell density. Ideally, cells should be experiencing logarithmic growth when transfected. This typically is when cells are approximately 80 % confluency. However, some cell types may be much more or much less dense than this to achieve optimal transfection efficiency. Optimization of the cell density upon transfection is highly recommended for new cell lines. When optimizing, we recommend that densities both well above and well below 80 % confluency are tested, as some cell types may appear more confluent than they truly are due to different morphologies and growth patterns.

Here, we present two methods for transfecting cells: both forward (Method 6a) and reverse (Method 6b) transfection. A forward transfection is when cells are plated first, and a reverse transfection is when the reagents are plated first and a cell suspension is added on top. Forward transfections are gentler on sensitive cell types, as the process of attaching to the plate while taking up plasmid (as in reverse transfections) can cause toxicity. This can typically be compensated for by plating a larger number of cells. Reverse transfections generally afford higher transfection efficiencies with adherent cells, as there is theoretically a greater amount of cell surface area with which the transfection reagents come into contact with. Additionally, they require one less day of laboratory work (as the cells are plated on the same day they are transfected), and often require less DNA input. As with all deviations from a standard protocol, we recommend optimization of the conditions for a reverse transfection. However, plating twice the density of cells required for an optimized forward transfection is a decent starting point for a reverse transfection.

Cells should not be transfected freshly from cryopreservation. Cells should be passaged a minimum of two times prior to transfection to ensure they have successfully recovered from cryopreservation. While cells may be transfected sooner, transfection efficiency and cell viability will decrease significantly.

Some cell types may be challenging to transfect utilizing cationic lipids. This may be due to a particularly delicate cell type, or extreme resistance to transfection via cationic lipids. In this case, alternate methods exist to deliver nucleic acid payloads into cells. Electroporation is the next most common method, which uses a transient electric pulse to briefly open the cell membrane, allowing cargo to pass through. Here, we present a method for electroporating K562 suspension cells (Method 6c). Finally, certain sensitive cell types such as stem cells exhibit extreme toxicity when exposed to plasmid DNA. Further, long-term expression of BEs can also be toxic in these cell types. We therefore offer a method for genomically encoding a small molecule-inducible BE construct into induced pluripotent stem cells (iPSCs) via piggyBac^™ transposition and delivering the gRNA via lenti-viral transduction, to facilitate base editing in stem cells (Method 6d).

5.1. Method 6a – forward transfection of adherent mammalian cells in 48-well plate

5.1.1. Materials and equipment

• 1.5 mL microcentrifuge tubes

• 0.2 mL PCR tubes

• Tissue culture treated plate(s)

• 15 mL conical tubes

• 50 mL conical tubes

• Micropipettes

• Centrifuge

• Tissue culture incubator set to optimal conditions for cell type

• Microscope with fluorescence imaging capabilities

• Lipofectamine^™ 2000

• Opti-MEM^™

• Antibiotic-free medium

• PBS

• Plasmid DNA encoding gRNA

• Plasmid DNA encoding BE

5.1.2. Protocol

5.1.2.1. Day one

1. Plate cells in antibiotic-free medium such that they will be experiencing logarithmic growth upon transfection (around 80 % confluency for many cell types).

2. Ensure wells are included for a no-color control, and color control(s). A color control should be present for each potential fluorescent marker used. For example, if only using a plasmid with GFP, include one well for a no-color control and one well for a GFP+ control. If using GFP alongside another marker such as mCherry, three control wells are necessary: one no-color, one GFP+ , and one mCherry+.

   Notes:

   1. Some cell types and their optimal plating densities can be found in Table 3.
   2. It is highly recommended to use antibiotic-free medium at this point, as antibiotics in the medium will cause excessive cell death upon transfection.

Table 3.

Recommended seeding densities for a 48-well plate. Additional optimization may be necessary if cells appear to be growing at atypical rates. Cells must be experiencing logarithmic growth when transfected. These values provide starting densities for potential further optimization by the user.

Cell type	Seeding density (cells per well of 48-well plate)
	Forward transfection	Reverse transfection
HEK293T	50,000	100,000
HEK293A	35,000	70,000
RPE1	55,000	110,000
AC16	40,000	80,000
HeLa	30,000	60,000

5.1.2.2. Day two

1. Bring Opti-MEM^™ or similar serum-reduced transfection medium to room temperature.

2. For each transfection condition, there will be two tubes - one will contain DNA and the other will contain the L2000.

   1. Tube 1 – combine the appropriate volume of gRNA and BE plasmids. Dilute with Opti-MEM^™ (or equivalent) to 12.5 μL total volume. For transfecting HEK293T cells in a 48-well plate, we recommend using 750 ng of BE plasmid with 250 ng of gRNA plasmid.
   2. Tube 2 – combine 1.5 μL L2000 and 11 μL Opti-MEM^™ (or equivalent).

3. Mix the contents of Tube 1 and Tube 2 for a total volume of 25 μL per well. Incubate for 15 min at room temperature.

   Notes:

   1. These dilutions and mixes can be done in either 0.2 mL PCR tubes or 1.5 mL microcentrifuge tubes, depending on the final volume needed.
   2. If transfecting multiple wells, make a master mix with extra volume to account for error.

4. After incubation, add the DNA/L2000 mix dropwise to the well of cells.

5. Return the plate to the incubator.

5.1.2.3. Day three

1. Observe the cells using a fluorescence microscope. If the BE used has a fluorescent marker, fluorescence will be visible if the transfection was successful (Fig. 6A–C).

2. Some cationic lipid transfection reagents can be toxic to some cell types. If cells are more sensitive or delicate, medium should be changed at this time.

   1. Carefully aspirate medium and gently add 150 μL PBS.
   2. Carefully aspirate PBS and add 250 μL fresh, prewarmed medium.
      Note: Medium can contain antibiotics at this point.

3. Return the plate to the incubator.

Fig. 6.

Fluorescence imaging of cells after a successful transfection. (A–B) HEK293T cells were transfected using the forward (A) and reverse (B) methods with a gRNA and a base editor with a GFP marker, and images were taken after 48 h of incubation. (C) AC16 cells were transfected using the forward method with a gRNA, base editor, and dGFP turn-on plasmid. Images were taken after 48 h of incubation. (D) hiPSCs with a genomically integrated base editor (using the piggyBac transposition method, in which the BE is co-expressed with mCherry) and transduced gRNA were treated with doxycycline for 48 h to induced base editor expression.

5.1.2.4. Day five

1. Harvest the cells via either bulk lysis (see Method 7) or flow cytometry (see Enriching Cells section).

5.2. Method 6b – reverse transfection of adherent mammalian cells in 48-well plate

5.2.1. Materials and equipment

• 1.5 mL microcentrifuge tubes

• 0.2 mL PCR tubes

• Tissue culture treated plate(s)

• 15 mL conical tubes

• 50 mL conical tubes

• Micropipettes

• Centrifuge

• Tissue culture incubator set to optimal conditions for cell type

• Microscope with fluorescence imaging capabilities

• Lipofectamine^™ 2000

• Opti-MEM™

• Antibiotic-free medium

• PBS

• Plasmid DNA encoding gRNA

• Plasmid DNA encoding BE

5.2.2. Protocol

5.2.2.1. Day one

1. Bring Opti-MEM^™ or similar serum-reduced transfection medium to room temperature.

2. Detach cells from the culture vessel. Spin the cells down and aspirate the medium. Resuspend in a minimal amount of prewarmed, antibiotic-free medium.

   Note: For a fully confluent T25 flask, 1 mL of medium is sufficient.

3. While detaching cells, generate the transfection mixtures. Ensure wells are included for a no-color control, and color control(s). See Method 6a, Day One, Step 2 for additional details.

4. For each transfection condition, there will be two tubes – one will contain DNA and the other will contain the L2000.

   1. Tube 1 – combine the appropriate volume of gRNA and BE plasmids. Dilute with Opti-MEM^™ (or equivalent) to 12.5 μL total volume. For transfecting HEK293T cells in a 48-well plate, we recommend using 750 ng of BE plasmid with 250 ng of gRNA plasmid.
   2. Tube 2 – combine 1.5 μL L2000 and 11 μL Opti-MEM^™ (or equivalent).

5. Mix the contents of Tube 1 and Tube 2 for a total volume of 25 μL per well. Incubate for 15 min at room temperature.

   Notes:

   1. These dilutions and mixes can be done in either 0.2 mL PCR tubes or 1.5 mL microcentrifuge tubes, depending on the final volume needed.
   2. If transfecting multiple wells, make a master mix with extra volume to account for error.

6. While the DNA/L2000 mix incubates, dilute the cells such that the number of cells per well is double what they would be for a forward transfection. Note that 250 μL of cell mixture will be added to each well.

   Note: See Table 3 for some cell types and their optimal plating densities.

7. After the DNA/L2000 mix has finished incubating, add it dropwise to the empty plate well.

8. Gently add 250 μL of cells to each well.

9. Return the plate to the incubator.

5.2.2.2. Day two

1. Proceed per Method 6a, Day Three.

5.2.2.3. Day four

1. Proceed per Method 6a, Day Five.

5.3. Method 6c – electroporation of suspension mammalian cells in a 12-well plate

5.3.1. Materials and equipment

• Electroporator (e.g., Neon^™)

• 1.5 mL microcentrifuge tubes

• 0.2 mL PCR tubes

• Tissue culture treated plate(s)

• 15 mL conical tubes

• 50 mL conical tubes

• Micropipettes

• Centrifuge

• Tissue culture incubator set to optimal conditions for cell type

• Microscope with fluorescence imaging capabilities

• Antibiotic-free medium

• Plasmid DNA encoding gRNA

• Plasmid DNA encoding BE

• PBS

• Neon^™ Transfection System 10 μL Kit

5.3.2. Protocol

This protocol uses the Neon^™ Electroporation System.

5.3.2.1. Day one

1. Warm appropriate antibiotic-free medium and PBS to 37 °C.

2. Aliquot 1 mL of media into each well of a 12-well plate. Place the plate in the incubator.

3. Prepare the Neon^™ tube by inserting it into the pipette station. Add 3 mL of Buffer E to the Neon^™ tube.

4. Add the DNA mixture (250 ng gRNA plasmid and 750 ng BE plasmid) to a 1.5 mL centrifuge tube.

5. Count cells and transfer the desired number of cells per transfection into a 1.5 mL centrifuge tube. For K562-s cells, 200,000 cells per well for a 12-well plate format is sufficient.

6. Centrifuge the cells to generate a cell pellet. For K562-s cells, centrifuge at 200 xg for 8 min. Aspirate the media and add 1 mL of prewarmed PBS to wash the cells.

7. Centrifuge the cells as before and aspirate the PBS.

8. Resuspend the cells in Buffer R so that there are 10,000 cells per microliter. That is, for every 200,000 cells, add 20 μL.

9. Transfer the 20 μL of cells into the previously prepared 1.5 mL centrifuge tube with the DNA mixture and gently pipette to mix.

10. Load the 10 μL Neon^™ tip into the Neon^™ pipette.

11. Slowly aspirate the cell/DNA mixture, being careful to avoid generating bubbles, as bubbles will prohibit adequate electroporation.

12. Submerge the pipette tip in the Buffer E within the Neon^™ tube and dock the pipette. A click sound indicates appropriate docking.

13. Run the protocol on the instrument. For K562-s cells, use 1700 V for 20 ms and 1 pulse.

14. Transfer the cells to the pre-warmed plate. Repeat steps 11–13 using the same tip for the remaining cell-payload mixture.

15. Repeat for as many wells as needed, then return the plate to the incubator.

5.3.2.2. Day two

1. Observe the cells using a fluorescence microscope. If the BE used has a fluorescent marker, fluorescence will be visible if the transfection was successful.

5.3.2.3. Day four

1. Harvest the cells via either bulk lysis or flow cytometry (see Enriching Cells section).

In some cases, it may be necessary to split or passage cells after transfection. This is advised when cells are too confluent or if the incubation period will be longer than 48 h. This may also be useful if two identical populations of transfected cells are desired for different assays.

5.4. Method 6d – piggyBac^™ BE transposition and gRNA transduction of human iPSCs in a 6-well plate

5.4.1. Materials and equipment

• 1.5 mL microcentrifuge tubes

• 0.2 mL PCR tubes

• Tissue culture treated plate(s), coated with Matrigel^®

• 15 mL conical tubes

• 50 mL conical tubes

• Micropipettes

• Centrifuge

• Tissue culture incubator set to optimal conditions for cell type

• Human induced pluripotent stem cell (hiPSC) line (this protocol is optimized for the KOLF 2.1 J line)

• piggyBac^™ cargo plasmid with BE and fluorescent marker (see Supplementary Table 1; the constructs that we provide use mCherry as a marker for BE expression)

• piggyBac^™ transposase plasmid (this protocol uses Super piggyBac^™: PB210PA-1, System Biosciences)

• Lentivirus containing gRNA with fluorescent marker (see Supplementary Table 1 and the protocol available at https://primer3.ut.ee/20)

• PBS

• ACCUTASE^™

• mTeSR^™ Plus

• ROCK inhibitor

• Lipofectamine^™ Stem

• Opti-MEM^™

• Polybrene

5.4.2. Protocol – transposition

5.4.2.1. Day one

1. Aspirate medium from iPSC culture. Wash once with 1 mL PBS.

2. Add 1 mL of ACCUTASE^™ and incubate at 37 °C for 10–12 min, until cells appear in single or 2-cell clumps.

3. Add 1 mL medium and mix well to break up any remaining clumps.

4. Measure the cell concentration using a hemocytometer or equivalent.

5. Centrifuge cells for 5 min at 100 × g to pellet them.

6. Aspirate media and resuspend cells in mTeSR^™ Plus with 5–10 μM ROCK inhibitor to a final concentration of 250,000 cells/mL.

7. Add 2 mL of the cell mixture to one well of a 6-well plate, to transfect 500,000 cells.

   Note: the number of cells transfected may be varied and can be optimized by the user as necessary.

8. In 1.5 mL microcentrifuge tubes, prepare the DNA mixture to add to the cells. Include the following conditions:

   1. Negative control: transposase only (600 ng)
   2. Experimental: cargo plasmid (1800 ng) + transposase (600 ng)
      Note: a 3:1 ratio of cargo to piggyBac^™ transposase is recommended. However, some optimization may be necessary and desired.

9. Mix the DNA mixture with OptiMEM^™ to total volume of 100 μL.

10. Make a Lipofectamine^™ Stem (LipoStem) master mix by adding 5 μL to 95 μL OptiMEM^™ for a total volume of 100 μL per sample.

11. For each sample, add 100 μL of the LipoStem mixture to the 100 μL DNA mixture and vortex. Incubate the LipoStem/DNA mixture for 10 min at room temperature.

12. Add the 200 μL of the LipoStem/DNA mixture to the cells in the 6-well plate. Swirl or tap to ensure even distribution within the medium.

13. Incubate at 37 °C overnight.

5.4.2.2. Day two

1. Replace with fresh mTeSR^™ Plus medium.

5.4.2.3. Day three

1. Once cells are fully confluent, replace medium with mTeSR^™ Plus medium containing 50 μg/mL hygromycin. Continue replacing the medium with fresh medium containing hygromycin every other day until all negative control cells are dead.

2. If cells are not yet fully confluent, continue replacing the medium every other day with fresh medium without antibiotics until cells are confluent. Then, switch to medium with hygromycin and proceed as described previously.

5.4.2.4. Day ten to fourteen

1. Once all negative control cells are dead, bank cells by expanding cell culture up to two 10 cm dishes and cryopreserving 20–25 vials of cells.

   Note: Cell banking is required prior to doxycycline induction of the BE. After induction commences, we do not recommend ceasing induction and re-using the cells, as silencing of the BE locus often occurs.

5.4.2.5. After cell banking

1. Perform a doxycycline titration to determine the optimal concentration of doxycycline for BE induction. We recommend testing concentrations ranging from 100 ng/mL to 2000 ng/mL. These concentrations are the final concentrations in the well, so a single well of a 6-well plate at the lowest concentration would have 200 ng doxycycline. See Fig. 6D for an example of successful BE expression induction.

5.4.3. Protocol – transduction

5.4.3.1. Prior to commencing transduction

1. Generate lentivirus containing the desired gRNA construct using preferred lentivirus generation protocol.

2. Freeze lentivirus in single-use aliquots, as freeze-thaw cycles reduce lentivirus efficacy.

3. Perform a function titration experiment to determine virus titer required for desired multiplicity of infection (MOI).

   Notes:

   1. To do this, transduce variable amounts of virus, keeping cell count the same. This experiment will yield a range of MOIs and inform you of how much virus to add to a certain number of cells to achieve a specific MOI. These values scale linearly.
   2. For example, if adding 2 μL virus to 500,000 cells results in 50 % of the cells expressing GFP, this is an MOI of 0.5. To achieve the same MOI in 2000,000 cells, add 8 μL of virus.
   3. If a single integrant per cell is desired, it is recommended to strive for a MOI of 0.2–0.4.

5.4.3.2. Day one

1. Aspirate medium from iPSC culture. Wash once with 1 mL PBS.

2. Add 1 mL of ACCUTASE^™ and incubate at 37 °C for 10–12 min, until cells appear in single or 2-cell clumps.

3. Add 1 mL medium and mix well to break up any remaining clumps.

4. Measure the cell concentration using a hemocytometer or equivalent.

5. Centrifuge cells for 5 min at 100 g to pellet them.

6. Aspirate media and resuspend cells in mTeSR^™ Plus with 5–10 μM ROCK inhibitor and 8–10 μg/mL polybrene to a final concentration of 250,000 cells/mL.

7. Add 2 mL of the cell mixture to one well of a 6-well plate, to transduce 500,000 cells.

   Note: the number of cells transduced may be varied and can be optimized by the user as necessary.

8. Add the amount of virus necessary to reach the intended multiplicity of infection (MOI).

   Notes:

   1. First, measure the titer of your virus through a functional titration experiment, as described previously.
   2. Every time you freeze and thaw lentivirus, it loses efficacy. It is recommended to freeze lentivirus in single-use aliquots.

9. Incubate at 37 °C overnight.

5.4.3.3. Day two

1. Replace with fresh mTeSR^™ Plus medium.

5.4.3.4. Day three

1. Induce base editor expression with doxycycline at the concentration determined by the previous doxycycline titration experiment (see Fig. 6D for an example).

2. Maintain doxycycline on the cells for at least five days.

   Note: fluorescent marker expression peaks on day six or seven.

5.4.3.5. Day four or more

1. Once cells are confluent, expand or split cells as desired.

   Note: ensure you maintain doxycycline during the expansion or splitting process if prior to day eight.

2. Bank cells for potential future use.

5.4.3.6. Day six or seven

1. Confirm successful induction and integration using a fluorescent microscope (Fig. 6) flow cytometry.

   Note: this assumes both the BE and gRNA are tagged with fluorescent proteins. It is highly recommended that fluorescent proteins are used as the selection markers. If lentiviral selection markers instead confer antibiotic resistance (e.g., puromycin), modify protocol accordingly.

2. Confirm successful base editing using Methods 7–9.

5.5. Method 7 – lysing adherent mammalian cells following transfection in a 48-well plate

5.5.1. Materials and equipment

• Thermocycler

• 1.5 mL microcentrifuge tubes

• 0.2 mL PCR tubes

• Micropipettes

• Microscope with fluorescence imaging capabilities

• PBS

• Sterile water

• 1 M Tris

• 10 % SDS

• Proteinase K

5.5.2. Protocol

1. 48–72 h post-transfection, observe the cells using a fluorescence microscope. If the BE used has a fluorescent marker, fluorescence will be visible in successful transfections. Note the confluency of each well.

2. Prepare an appropriate amount of fresh lysis buffer:

   1. Scale the amount of lysis buffer to the confluency of each well. For a 100 % confluent well, use 100 μL of lysis buffer, scale linearly down (i.e. 50 % confluent well requires 50 μL lysis buffer).
      Note: If between two percentages, round down. It is better to add less and be able to add additional later, rather than add too much lysis buffer to begin with.
   2. Prepare fresh lysis buffer the day of lysis and keep at 4 °C until use. For 200 μL of buffer, use the below numbers and scale linearly as needed.

      Component/Reagent	Volume (μL)
      Sterile water	196.75
      1 M Tris	2
      10% SDS	1
      Proteinase K	0.25

      Note: it is not feasible nor recommended to pipette the volume of Proteinase K needed for a final total volume less than 200 μL.

3. Carefully aspirate growth medium and gently add 150 μL PBS.

4. Carefully aspirate PBS, ensuring all liquid has been removed from the well.

5. Add the appropriate amount of lysis buffer to each well.

6. Let cells incubate with lysis buffer for 1–5 min at room temperature, then resuspend thoroughly. Be sure to scrape the tip of the pipette across the bottom of the well to remove all cells. Lysis material should be viscous.

   Note: If lysis is not viscous, either too much lysis buffer was added, or the PBS wash was too aggressive and removed a significant portion of the cells.

7. Transfer resuspension into 0.2 mL PCR tubes.

8. Using a thermocycler, incubate the resuspension at 37 °C for 1 h and 80 °C for 30 min. Ramp the temperature down to either 12 °C or 4 °C by 0.1 °C per second.

9. Store lysis at 4 °C for up to 1 year, but ideally no more than 6 months.

6. Next-generation sequencing to quantify base editing efficiency

All genome editing experiments require a method for quantifying editing efficiency. Next Generation Sequencing (NGS), also known as High Throughput Sequencing (HTS), provides information on different sequence outcomes (alleles), along with quantitative editing efficiencies for each allele (the lower limit of detection is roughly 0.1 %). This is useful for sites with low editing percentages, protospacers with multiple editable bases (i.e. bystanders, in which some alleles may have only one base edited, while others may have more than one edited, see Fig. 7C), or for isogenic cell line characterization at sites that have multiple copies of the gene of interest. Here, we provide guidance for the preparation of samples for a 300-cycle paired-end NGS sequencing run (a 2 ×150 cycle run specifically) with amplicons of 200- to 250-bp in length (Fig. 7A). However, this protocol can be adjusted for different sequencing configurations. When quantifying editing efficiencies from “bulk samples” (i.e. those that are not clonally expanded and thus a heterogeneous mixture of genotypes), we do not recommend using Sanger sequencing as the relative peak heights of Sanger sequencing data at mixed-base sites (such as SNVs) are not necessarily a quantitative representative of the proportion of each base (the incorporation rate of ddNTP terminators is highly dependent on template concentrations and the sequence context surrounding the SNV location) (Carr et al., 2009). For isogenic cell lines (which are clonal and thus a homogeneous genotype), Sanger sequencing may be appropriate for genotyping. However, if a particular clone is heterozygous, and the cell line is known to have an unusual or unstable karyotype, NGS is recommended for genotyping.

Fig. 7.

Sequencing preparation and analysis examples. (A) Example schematic for designing amplicons and primers for Sanger and NGS analysis (top), and work flow of the two rounds of PCR for NGS preparation (bottom). (B) Agarose gel showing the round 1 and 2 PCR products with an increase in size between the rounds due to the Illumina adapters added in round 2 amplification. (C) CRISPResso2 analysis of allelespecific editing data. The target edited sequence is shown to have the highest prevalence, while the wildtype is the second most commonly identified allele. Other alleles, such as large deletions and bystander edits are undesired outcomes.

To quantify base editing efficiencies with NGS, the genomic locus surrounding the protospacer is amplified from the lysis obtained from Method 7, Method 9a, or Method 9b. First, design primers to amplify the protospacer plus additional sequence on either side to produce an amplicon of 200- to 250-bp with Illumina adapter sequences on either side (round 1 PCR, Fig. 7A and B). 200- to 250-bp provides optimal clustering, but a maximum amplicon size of 285-bp is possible. Specifically, we recommend designing the amplicon such that the forward and reverse reads have at least 10-bp of overlap between them in the middle of the amplicon, as this overlap is required for some commonly used software that merge paired-end reads. The annealing regions of these primers should be designed using a primer design software (e.g., Benchling’s Primer3), (Untergasser et al., 2012) as PCR of genomic DNA lysate can be quite challenging. Round 1 NGS primer adapters can be found in Table 4. After this initial amplification, Illumina-specific barcodes (index 1 and 2 sequences) and P5/P7 sequences are added (round 2 PCR) to enable pooling of multiple different samples in the same NGS run (Fig. 7A). These allow for de-multiplexing of the samples after the NGS run, separating the different samples into separate FASTQ files. Note that if performing an NGS run that differs from a 2 × 150 cycle run, the amplicon size will need to correlate with the sequencing run. For example, if running a 2 × 75 cycle run, your amplicon should be under 135-bp (to ensure at least 10-bp of overlap between the paired-end reads). After the NGS run is completed, the data can be processed using CRISPResso2 (Clement et al., 2019) software or custom NGS analysis codes.

6.1. Method 8 – next generation sequencing sample preparation

6.1.1. Materials and equipment

• Thermocycler

• UV-Vis spectrophotometer (e.g., NanoDrop™)

• Micropipettes

• Ice bucket

• Razor blade

• 0.2 mL PCR tubes

• Sterile water

• HF buffer

• GC buffer

• dNTPs

• Phusion^®

• DMSO

• Cell lysis

• Primers with NGS adapters

• Illumina-specific barcode primers

• 2 % agarose gel

• Gel extraction kit (e.g., Monarch® DNA Gel Extraction Kit)

6.1.2. Protocol

6.1.2.1. Round 1 preparation

1. Make a 20 μM mix of the found 1 forward and reverse NGS primers.

2. Mix per reaction in a 0.2 mL PCR tube:

   Component/Reagent	Volume (μL)
   Sterile water	8.17
   GC green buffer	2.50
   dNTPs	0.25
   DMSO (3%)	0.375
   Primer mix (20uM)	0.08
   Phusion®	0.125
   Cell lysate (DNA)	1.00
   Total Volume	12.5

   Notes:

   1. Be sure to add the Phusion^® or equivalent polymerase last and keep the reaction mixture on ice after the addition of the polymerase.
   2. This can be scaled up to make a master mix where all components except the cell lysate are added, and the individual cell lysates are added after the master mix is distributed.

   1. Add the sample(s) to a thermocycler and cycle per the protocol below:

      Cycles	Time	Temperature	Purpose
      1 cycle	1 min	98 °C	initial denaturation
      22–28 cycles	10 s	98 °C	denaturation
      	20 s	65 °C	annealing
      	10 s	72 °C	extension
      1 cycle	5 min	72 °C	final extension
      1 cycle	forever	4–12 °C	ramp to hold

      Note: it is important to run the fewest possible number of cycles to avoid PCR bias. This may require optimization when working with a new set of fwd/rev NGS primers.

   1. Run 1–3 μL of each sample on a 2 % agarose gel using a 100-bp ladder. It is recommended to run at 115–125 V for 15–25 min to resolve any potential primer dimer bands.
      Note: A clear, single band that is 66 bp longer than the amplicon should be observed. This added length accounts for the NGS adapter sequences. Refer to the troubleshooting sub-section if multiple bands are observed, no bands are observed, or the primer dimer band density outweighs the on-target band.

6.1.2.2. Round 2 preparation

1. Make a 1 μM mix of the round 2 forward and reverse Illumina adapter primers. Each sample should have its own unique fwd/rev combination.

2. Mix per reaction in a 0.2 mL PCR tube:

   Component/Reagent	Volume (μL)
   Sterile water	6.34
   HF green buffer	2.40
   dNTPs	0.24
   Primer mix (1uM)	2.40
   Phusion®	0.12
   Round 1 PCR product	0.50
   Total Volume	12

   Notes:

   1. Primers can be added individually at 1.2 μL per sample. It is recommended to pool either the forward or reverse primer (if one is held constant across all samples) in the master mix to make pipetting simpler and reduce error.
   2. Be sure to add the Phusion^® or equivalent polymerase last and keep the reaction mixture on ice after the addition of the polymerase.
   3. This can be scaled up to make a master mix where all components with the exception of the Round 1 PCR product and one of the primers (if one is held constant across all samples) are added, with these components being added after the master mix is distributed.

1. Add the sample(s) to a thermocycler and cycler per the protocol below:

   Cycles	Time	Temperature	Purpose
   1 cycle	1 min	98 °C	Initial denaturation
   8–16 cycles	10 s	98 °C	Denaturation
   	20 s	65 °C	Annealing
   	10 s	72 °C	Extension
   1 cycle	5 min	72 °C	Final extension
   1 cycle	forever	4–12 °C	Ramp to hold

1. Run 1–3 μL of each sample on a 2 % agarose gel using a 100-bp ladder. It is recommended to run at 115–125 V for 8–20 min to resolve any potential primer dimer bands.

   Note: A clear, single band that is 140 bp longer than the amplicon should be observed. This addition accounts for both the NGS adapter sequences from Round 1 and the Illumina-specific barcode additions from Round 2. Refer to the troubleshooting sub-section if multiple bands are observed, no bands are observed, or the primer dimer band density outweighs the on-target band.

6.1.2.3. Pooling and gel extraction

1. Pool all Round 2 PCR products that are a similar size together depending on the intensity of their band on the gel in step 8. Product sizes should vary no more than 50 base pairs, i.e. a 300 bp product and 350 bp product can be pooled.

   Notes:

   1. For bands brighter than the average intensity, pool less (i.e. half) than the normalized amount. For bands dimmer than the average intensity, pool more (i.e. double) than the normalized amount.
   2. Ensure enough of each of the Round 2 samples remain to feasibly repeat the pooling should a mistake be made past this step. If your library is small (< 10 samples) consider doubling the scale of your round 2 reaction to allow for multiple attempts with a larger pooling volume.

2. Run the entire pooled library on a 2 % agarose gel using a wide comb. Run at 100–110 V for 35–45 min for the best resolution of potential primer dimer bands.

   Note: Run as much volume as feasible in a well of the gel. Well size will vary depending on gel or gel caster.

3. Perform a gel extraction of the on-target band. Use a NanoDrop^™ spectrophotometer or equivalent to measure the concentration and assess DNA quality before proceeding.

   Notes:

   1. The minimum mass of DNA required to run during NGS is around 100 ng total DNA. The A260/A280 should be around 1.80 and A260/A230 should be around 2.20. However, if the peak is clean, minor deviations from these parameters are acceptable. If there is a shoulder on the peak, then an additional PCR cleanup of the product is recommended.
   2. Expect to lose half of the total yield of DNA with each additional cleanup.

4. Quantify the library using a Qubit dsDNA HS assay kit. This will quantify the actual amount of dsDNA in the sample, ignoring any other contaminants. This value should be used for dilution calculations when preparing to run on NGS and is considered more reliable than a measurement from a NanoDrop^™ spectrophotometer or equivalent.

5. Set up the NGS according to the Illumina protocols for the instrument, or submit this library to a sequencing core. To obtain robust, reliable, and quantitative sequencing data, we recommend obtaining at least 10,000 sequencing reads per sample.

6. The instrument will demultiplex the samples according to the forward and reverse barcode sequence combinations, with all of the reads from a given sample in separate fastq files. Those who are familiar with NGS datasets and analysis may prefer to analyze their fastq files for editing efficiencies with custom code. For those not familiar with NGS datasets, we recommend using CRISPResso2, which has a simple user interface and can analyze targeted amplicon sequencing data for genome editing efficiencies. To use the web version:

   1. Under “Editing tools”, select “Base editors”.
   2. Under “Sequencing design”, select “Paired end reads” unless a single end read run was completed.
   3. Under “Fastq file R1″, upload your read 1 (R1) fastq file, and under “Fastq file R2″, upload your read 2 (R2) fastq file.
   4. Under “Amplicon”, input the entire amplicon sequence (this will include the annealing regions of the forward and reverse round 1 primers, plus the intervening sequence). Note that CRISPResso2 requires 10-bp of overlap between the R1 and R2 reads for paired end reads.
   5. Under “sgRNA”, enter the 20-bp protospacer sequence (do not include the PAM). This will result in the program quantifying editing efficiencies within this region and providing an allele plot focused on this region as well.
   6. Under “Optional parameters”, select the appropriate base editor “target base” and “result base”. For CBE, this will be a target base of C and a result base of T. For ABE, this will be a target base of A and a result base of G.
   7. The remainder of the default parameters should be sufficient.
   8. Click on “Submit!” to analyze. See Fig. 7C for an example output.
   9. Alternatively, if there are many fastq files to analyze, it may be desired to download the source program from github (https://github.com/pinellolab/CRISPResso2) and run the analysis in “batch mode”, which allows for analysis of multiple fastq files at once (the web version can only handle up to 100 MB of data at a time).

Editing efficiency can vary depending on multiple factors. Some factors to consider include the protospacer selected (e.g., position of the edit and of bystanders), the base editor used (e.g., editing window, processivity, sequence motifs), the delivery method for the gRNA and base editor (e.g., electroporation, cationic lipid, transduction), the cell line and cell type (e.g., some cell lines are more challenging to transfect and edit in than others), and the collection and harvesting method (e.g., bulk lysis, FACS, enriching for transfection vs. enriching for BE expression). Therefore, we have provided detailed information from the transfections covered in this work in Table 5. The necessary level of editing for an experiment to be deemed a success may differ based on desired outcomes. However, if the goal is to generate an isogenic cell line, we recommend that editing efficiency be ≥ 10 %.

Table 5.

Editing data for transfections performed in this work. Additional editing data for experiments not performed in this work, but by these authors, is also provided. The protospacer sequence and corresponding gene are given in column 1.

Protospacer sequence (5′–3′) and gene target	Editor used	Delivery method	Cell type	Harvest method (bulk, FACS for BE expression, etc.)	Expected editing efficiency
GGCCCAGACTGAGCACGTGA (HEK site 3)	ancBE4 - NG	Method 6d	iPSC (KOLF 2.1J)	FACS for GFP (gRNA)	~50 %
CAAACCGCTGTGGGCAGAAG (ERCC2)	ancBE4-NG	Method 6a	HEK293T	Method 7: Bulk	~10 %
GGCTCGGTGCTCTCCTCCGT (SCN5A)	ancBE4-NG	Method 6a	HEK293T	Method 7: Bulk	~15 %
CCACTGTGCAGCCAGTGCCC (MUTYH)	ancBE4-NG	Method 6a	HEK293T	Method 7: Bulk	~15 %
CAAACCGCTGTGGGCAGAAG (ERCC2)	ancBE4-NG	Method 6a	HEK293T	Method 9a: enriched FACS	~25 %
GGCTCGGTGCTCTCCTCCGT (SCN5A)	ancBE4-NG	Method 6a	HEK293T	Method 9a: enriched FACS	~35 %*
CCACTGTGCAGCCAGTGCCC (MUTYH)	ancBE4-NG	Method 6a	HEK293T	Method 9a: enriched FACS	~35 %*
CAAACCGCTGTGGGCAGAAG (ERCC2)	ancBE4-NG	Method 6a	HEK293T	Method 9a: dGFP turn-on FACS for BE expression	~30 %
GGCTCGGTGCTCTCCTCCGT (SCN5A)	ancBE4-NG	Method 6a	HEK293T	Method 9a: dGFP turn-on FACS for BE expression	~45 %*
CCACTGTGCAGCCAGTGCCC (MUTYH)	ancBE4-NG	Method 6a	HEK293T	Method 9a: dGFP turn-on FACS for BE expression	~45 %*
CAAACCGCTGTGGGCAGAAG (ERCC2)	ancBE4-NG	Method 6b	HEK293T	Method 7: Bulk	~15 %
GGCTCGGTGCTCTCCTCCGT (SCN5A)	ancBE4-NG	Method 6b	HEK293T	Method 7: Bulk	~20 %
CCACTGTGCAGCCAGTGCCC (MUTYH)	ancBE4-NG	Method 6b	HEK293T	Method 7: Bulk	~20 %
CAAACCGCTGTGGGCAGAAG (ERCC2)	ancBE4-NG	Method 6b	HEK293T	Method 9a: enriched FACS	~25 %
GGCTCGGTGCTCTCCTCCGT (SCN5A)	ancBE4-NG	Method 6b	HEK293T	Method 9a: enriched FACS	~35 %
CCACTGTGCAGCCAGTGCCC (MUTYH)	ancBE4-NG	Method 6b	HEK293T	Method 9a: enriched FACS	~40 %
CAAACCGCTGTGGGCAGAAG (ERCC2)	ancBE4-NG	Method 6b	HEK293T	Method 9a: dGFP turn-on FACS for BE expression	~30 %
GGCTCGGTGCTCTCCTCCGT (SCN5A)	ancBE4-NG	Method 6b	HEK293T	Method 9a: dGFP turn-on FACS for BE expression	~45 %
CCACTGTGCAGCCAGTGCCC (MUTYH)	ancBE4-NG	Method 6b	HEK293T	Method 9a: dGFP turn-on FACS for BE expression	~50 %
CCACTGTGCAGCCAGTGCCC (MUTYH)	ancBE4-NG	Method 6c	K562-s	Method 7: Bulk	~10 %

Expected editing efficiency is determined by the allele-specific target base conversion percentage provided by NGS analysis, in technical triplicate, unless indicated (*) which is in technical duplicate.

6.1.3. Troubleshooting

6.1.3.1. Round 1 PCR

• No bands:

  • Lysis may be too concentrated or too dilute. Try a gradient of lysis input from 1 μL of a 1:10 dilution of the lysis, up to 2 μL of the undiluted lysis.

    • A smaller gradient of 0.5-2 μL of lysis may also be used.
  • Primers may not be binding. If the annealing region is too long, the annealing temperature may be too high. Either redesign the primers or try a temperature gradient on the annealing step from 54 °C to 72 °C.
• Note: Running a PCR with the annealing temperature equal to or higher than the extension temperature is not recommended as it can lead to PCR bias, but in cases of particularly tricky sites, and an inability to redesign the primers, this could help with getting any bands to move forward with in Round 2.

  • Primers may be forming self-dimers or heterodimers and not binding to the target. Use a primer analyzer tool such as IDT’s OligoAnalyzerTM tool (www.idtdna.com/calc/oligoanalyzer) to see if the primers are forming significant primer dimers or secondary structures. Include the adapter sequences in your analysis.
• Unbalanced primer dimer and on-target bands:

  • Ideally, the on-target band should be stronger than the primer dimer band. If not, try a temperature gradient on the annealing step from 54 °C to 72 °C. Also, try using half the amount of primer, or double the amount of lysis. Finally, try adding 3–5 % DMSO to the PCR reaction mixture.
• More than one band:

  • A primer dimer band is normal, but multiple bands in addition to a primer dimer band could be due to the primer binding at other regions in the genome. Be sure to check if the primer (including the NGS adapters) has any homology to other regions of the genome.
  • Adding 3–5 % DMSO can also help to reduce off-target binding.

6.1.3.2. Round 2 PCR

• Unbalanced primer dimer and on-target bands:

  • Try increasing the number of cycles up to 10–16.
  • Try increasing the amount of Round 1 PCR input (1–2 μL).

• Having no bands or multiple Round 2 bands is rare and likely due to contamination of reagents.

6.1.3.3. Gel extraction

• The goal is to have a clean and concentrated gel extraction product. An additional PCR cleanup on the gel extraction product may be necessary. Ensuring an A260/A280 peak around 1.80 and A260/A230 peak around 2.20 on a NanoDrop^™ spectrophotometer or equivalent is advised.

• Save some of the Round 2 samples so that re-pooling is possible should the gel extraction fail.

7. Enriching for cells with high BE expression or activity

If the initial base editing efficiency, as assessed from bulk samples, is low (<10 %), editing can be improved using Fluorescence-Activated Cell Sorting (FACS) (Fig. 8). The BE plasmids that we recommend here have fluorescent markers incorporated (via the self-cleaving P2A sequence), enabling users to enrich for cells with high BE expression using FACS. To improve editing efficiency further, fluorescent turn-on plasmids such as pCBE-dGFP and pABE-dGFP (Fig. 5A) can be used. Both plasmids contain a functional mCherry gene (which acts a fluorescent marker for transfection) and a GFP gene with an inactivating mutation (which we call “dGFP”). In pCBE-dGFP, a C·G to T·A mutation is required to restore fluorescence, while in pABE-dGFP, an A·T to G·C mutation is required. pCBE-dGFP-gRNA and pABE-dGFP-gRNA are the respective gRNA-expressing plasmids to facilitate GFP turn-on of these two dGFP plasmids. When employing this strategy, the dGFP and dGFP-gRNA plasmids are added to the transfection mixture in Methods 6a, 6b, or 6c in a ratio of 1:5 and 1:10 to the endogenous gRNA plasmid, respectively. The BE performs editing at both the endogenous target and the episomal dGFP target, and GFP+ cells are sorted (Fig. 9). It is important to note that pCBE-dGFP and pABE-dGFP cannot be used with BE plasmids that contain GFP fluorescent markers. Both of these strategies can greatly improve base editing efficiencies to levels that make isogenic cell line generation much more feasible (we recommend obtaining editing efficiencies of >10 % for this) (Fig. 8). If editing efficiency for a given target remains low (<5 %) even after incorporating these strategies, it is recommended to use the same gRNA with the appropriate wild-type Cas9 variant (Cas9, Cas9-NG, or SpRY-Cas9) and quantify indel introduction efficiencies at the target site. This will provide information on how efficiently the Cas9:gRNA complex can bind to the genomic locus (which is a prerequisite for base editing). If indel introduction efficiency is low, gRNA expression may be an issue, in which case we recommend trying with synthetic gRNA.

Fig. 8.

Target conversion efficiencies across various conditions in HEK293T cells. (A) Under forward transfection conditions, target conversion efficiencies are shown for three genomic targets within the ERCC2, SCN5A, and MUTYH genes. Bulk lysis, where cells are not sorted using FACS, is shown in gray. Enriched samples, shown in pink, were transfected with a BE construct that constitutively expresses GFP. Therefore, only cells that were successfully transfected were collected, lysed, and sequenced. In teal, dGFP turn-on samples are shown. Cells were transfected with a construct containing dGFP that can be revived if base editing is active (according to the constructs shown in Fig. 5A). Thus, only cells with active base editor are collected, lysed, and sequenced. (B) Reverse transfection target conversion efficiencies are shown in the same manner as (A). Notably, some genomic loci can be more challenging to edit for a variety of reasons. (C) Cells transfected using forward and reverse transfections are compared after bulk lysis. Under bulk lysis conditions, reverse transfections yield a higher rate of target conversions at all three loci. (D) Spacer sequences for the loci. The target nucleotide is bolded and underlined. The corresponding gRNAs were combined with the ancBE4-NG editor.

Fig. 9.

Fluorescence-Activated Cell Sorting (FACS) scatter plots. (A–B) HEK293T cells were transfected with base editor and GFP as a transfection marker. Forward transfected samples (A), and reverse transfected samples (B) are shown comparing the forward scatter area (x-axis) to GFP intensity (y-axis). The forward scatter area provides information on the size of the cells. When only a single color is measured, this can be plotted against forward scatter (to assess the homogeneity of the size of the cells) or as a histogram. (C) HEK293A cells with the dGFP turn-on enrichment plasmid are shown. These cells express mCherry as a transfection control, and GFP fluorescence is restored in cells with active base editor. This graph compares mCherry (y-axis) to GFP (x-axis) fluorescence. Cells in the green boxes were gated and sorted for enrichment, then later lysed and sequenced.

7.1. Method 9a – FACS of adherent mammalian cells for base editing enrichment

7.1.1. Materials and equipment

• Flow cytometry instrument (BioRad S3e or equivalent)

• Micropipettes

• Filtered cap FACS tubes (35 μm mesh size)

• 1.5 mL microcentrifuge tubes

• Ice

• Accumax

• PBS

• Fluorescent cells

7.1.2. Protocol

1. This protocol is to be performed after Method 6, instead of Method 7.

2. 48–72 h post-transfection, observe the cells using a fluorescence microscope to confirm that fluorescence is present.

3. Carefully aspirate growth medium and gently add 150 μL PBS.

4. Carefully aspirate PBS, ensuring all liquid has been removed from the well.

5. Add 60 μL of Accumax to each well.

   Note: Accumax is used over Trypsin detachment mediums as it is more efficient, particularly at separating clumps of mammalian cells. This is ideal for FACS applications as even small clumps of cells result in clogs on the instrument.

6. Incubate at room temperature for 3–5 min. The cells can be moved into a 37 °C incubator to shorten the incubation time to 1–2 min.

7. Resuspend each well in 140 μL of cold PBS. This decreases the probability of the cells reattaching to each other before they are run on the FACS instrument.

   Note: Keep the plate on ice for the duration of the sorting process.

8. Immediately before loading the cells onto the instrument, resuspend each well again, and push the resuspension through the filtered cap of a FACS tube.

9. Generate gates using a no-color control. Load the instrument and run the sample, accounting for cell death and singlets versus multiples.

10. Refine gates using color control(s). A single-color control is required for each color used. For example, if only using GFP, only a green color control is required. If using both mCherry and GFP, two color controls are required – one red, one green.

11. Place a 1.5 mL microcentrifuge tube containing 500 μL of cold PBS into the cell dispensing area of the FACS instrument.

12. Load the instrument with a sample well, using previously established gates from Steps 9 and 10 for either GFP+ or dual GFP+/mCherry + cells, depending on the enrichment strategy used.

    1. Note the number of cells collected. This is important for calculating the amount of lysis buffer to use, later.
    2. Collect a minimum of 2000 cells.
    3. Ideally, collect 10,000 or more cells.
    4. See Fig. 9 for an example of gating.

13. Collect the microcentrifuge tube and invert it gently to resuspend any cells on the wall of the tube. Keep on ice until the rest of the sorting is complete.

14. Spin the microcentrifuge tubes at 300 g for 10 min, ensuring the back tab of the tube is facing the same direction on all tubes so the cell pellet location is known. Ideally, face the tab outwards.

15. Gently pipette off the PBS, being very careful not to disturb the pellet.

16. Add the lysis buffer to the tube and resuspend.

    Notes:

    1. Refer to Method 7 step 4b for lysis buffer recipe.
    2. The amount of lysis buffer to use is dependent on the amount of cells collected. Use 0.5 μL lysis buffer per 1000 cells collected.

17. Add the resuspension to a 0.2 mL PCR tube.

18. Using a thermocycler, incubate the resuspension at 37 °C for 1 h and 80 °C for 30 min. Ramp the temperature down to either 12 °C or 4 °C by 0.1 °C per second.

19. Store lysis at 4 °C for up to 1 year, but ideally no more than 6 months.

20. Sequence the genomic locus of interest using Method 8.

If the goal of an experiment is to generate an isogenic cell line harboring the target mutation, individual cells must be isolated after the editing process has occurred and allowed to proliferate into a clonal population. To sort individual cells, we recommend using a FACS instrument called a Namocell^®. This instrument uses single-use sterile cartridges that are loaded with cells still suspended in growth medium, and sorts cells at a far lower psi (<2 psi) compared to traditional FACS instruments (20–70 psi). Traditional FACS instruments are known to frequently contaminate cells as they pass through shared internal tubing inside the machine, and the sterile cartridges of the Namocell^® eliminate this source of contamination. Furthermore, the higher psi at which cells are expelled in traditional FACS instruments is not only toxic to the cells, but also increases the possibility of multiple cells landing in a single well when sorting, which results in a non-clonal cell line comprised of a mixture of different genotypes. We have found that with the Namocell^®, approximately one third of cells survive the single-cell isolation and clonal expansion process, compared to approximately one in ten with traditional FACS instruments.

7.2. Method 9b - Namocell^® sorting for cell line generation with adherent mammalian cells using base editing

7.2.1. Materials and equipment

• Namocell^®

• Micropipettes

• Ice bucket

• 96-well plate(s)

• Filtered cap FACS tubes (35 μm mesh size)

• Accumax

• PBS

• Cell culture media

• Optional: Penicillin Streptomyocin

• Fluorescent cells

7.2.2. Protocol

1. This protocol is to be performed after Method 6, instead of Method 7. 48–72 h post-transfection, observe the cells using a fluorescence microscope to confirm that fluorescence is present.

2. Prepare a 96-well plate with 100 μL of growth medium in each well. Put the plate(s) into the incubator to warm before sorting.

   Notes:

   1. Add an additional 10–20 μL in edge wells to account for evaporation.
   2. For HEK293T or HEK293A cells, DMEM + 25 % FBS + 1 % Penicillin Streptomycin is recommended.
   3. Prepare as many plates as needed to sort. Assume a 10–40 % survival rate and be prepared to care for that many individual wells when deciding how many wells to sort into.

3. Carefully aspirate growth medium and gently add 150 μL PBS.

4. Carefully aspirate PBS, ensuring all liquid has been removed from the well.

5. Add 60 μL of Accumax to each well of the 48-well plate.

   Note: Accumax is used over Trypsin detachment mediums as it is more efficient, particularly at separating clumps of mammalian cells. This is ideal for FACS applications as even small clumps of cells result in clogs on the instrument.

6. Incubate at room temperature for 3–5 min. The cells can be moved into a 37 °C incubator to shorten the incubation time to 1–2 min.

7. Resuspend each well in 190 μL of cold growth medium. This decreases the probability of the cells reattaching to each other before they are run on the Namocell^®. Either keep the resuspension in the plate, or move into a 1.5 mL microcentrifuge tube, and place on ice.

   Note: Keep the plate or tube on ice for the duration of the sorting process.

8. Immediately before loading the instrument, resuspend each well again, and push the resuspension through a filtered cap of a FACS tube.

9. The Namocell^® requires an incredibly dilute sample volume (~5000–20,000 cells/mL) depending on the iteration of the Namocell^® in use (Hana or Pala). Dilute the sample accordingly with the appropriate volume of growth medium.

   Note: A 1:10 dilution is a good place to start.

10. After starting up and washing the machine, load ~500 μL of the sample into a new sterile sorting cartridge.

11. Set gates using your sample, gating for the appropriate fluorophores for the experiment. See Method 6a Day One Step 2 for necessary controls.

12. Add the prepared and warmed plate onto the stage of the Namocell^® and sort samples. Repeat as needed for additional plates.

13. Immediately place cells in the incubator after sorting. Leave them for at least 48 h before attempting to move them to a different incubator.

14. Single colonies can usually be observed after 1 week, be sure to note if multiple colonies are observed in one well.

15. After 2 weeks cells will begin to require passaging. Split all viable wells into a new “master plate” and a “sequencing plate.” These plates should be identical and will serve as the numbering for each colony going forward.

16. Once wells in the “sequencing plate” are confluent, lyse them according to Method 7. For a 96-well plate be sure to scale the lysis buffer down to 50 μL for a fully confluent well.

17. Keep the “master plate” alive and split wells as needed until all of the “sequencing plate” wells have been sequenced.

18. Keep 3 homozygous clones, 3 heterozygous clones, and 3 wild type clones as controls. Expand the wells up to a 6-well or T25 and cryopreserve these lines.