Methods xxx (2015) xxx–xxx

Contents lists available at ScienceDirect

Methods

journal homepage: www.elsevier .com/locate /ymeth

Efficient genomic correction methods in human iPS cellsusing CRISPR–Cas9 system

http://dx.doi.org/10.1016/j.ymeth.2015.10.0151046-2023/� 2015 Published by Elsevier Inc.

⇑ Corresponding author at: 53 Shogoin Kawahara-cho, Sakyo-ku, Postal Code606-8507, Kyoto, Japan.

E-mail address: akitsu.hotta@cira.kyoto-u.ac.jp (A. Hotta).1 Current address: Division of Pediatric Hematology/Oncology, Boston Children’s

Hospital and Department of Biological Chemistry and Molecular Pharmacology,Harvard Medical School, Boston, MA 02115, USA.

Please cite this article in press as: H.L. Li et al., Methods (2015), http://dx.doi.org/10.1016/j.ymeth.2015.10.015

Hongmei Lisa Li a,1, Peter Gee a, Kentaro Ishida a,b, Akitsu Hotta a,b,⇑aCenter for iPS Cell Research and Application (CiRA), Kyoto University, Kyoto, Japanb Institute for Integrated Cell-Material Sciences (iCeMS), Kyoto University, Kyoto, Japan

a r t i c l e i n f o a b s t r a c t

Article history:Received 16 July 2015Received in revised form 16 October 2015Available online xxxx

Keywords:iPS cellsCRISPRGenome editingKnock-outKnock-inGene correction

Precise gene correction using the CRISPR–Cas9 system in human iPS cells holds great promise for variousapplications, such as the study of gene functions, disease modeling, and gene therapy. In this reviewarticle, we summarize methods for effective editing of genomic sequences of iPS cells based on our expe-riences correcting dystrophin gene mutations with the CRISPR–Cas9 system. Designing specific sgRNAs aswell as having efficient transfection methods and proper detection assays to assess genomic cleavageactivities are critical for successful genome editing in iPS cells. In addition, because iPS cells are fragileby nature when dissociated into single cells, a step-by-step confirmation during the cell recovery processis recommended to obtain an adequate number of genome-edited iPS cell clones. We hope that the tech-niques described here will be useful for researchers from diverse backgrounds who would like to performgenome editing in iPS cells.

� 2015 Published by Elsevier Inc.

1. Introduction

Programmable nucleases are artificial nucleases that bind andcut a certain DNA sequence of interest, thereby introducing adouble stranded DNA break at a desired genomic locus and activat-ing the host DNA repair response locally. During the DNA repairprocess, the host DNA sequence can be altered by the introductionof small deletion or insertion mutations via non-homologous endjoining (NHEJ) or by the insertion of extra sequences via homolo-gous recombination (HR) pathways. Taking advantage of theseDNA repair processes, scientists have established genome editingtechnology, with which they can alter a genomic sequence at adesired locus. This technology has been proven to be a powerfultool for disrupting, knocking-in and correcting an endogenous genein many organisms [1]. Among the several platforms ofprogrammable nucleases, the clustered regularly interspaced shortpalindromic repeat (CRISPR) and CRISPR associated 9 (Cas9)endonuclease system provides great flexibility and ease of use forgene modification [2–5].

1.1. The CRISPR–Cas9 system

The CRISPR–Cas9 system is a prokaryotic defense system thatconfers resistance against foreign genetic elements such as viralor plasmid DNA. CRISPR RNA (crRNA) and trans-activating crRNA(tracrRNA) form a complex with Cas9 protein to recognize andcut a sequence complementary to the crRNA. By conjugating crRNAand tracrRNA into a single-guide RNA (sgRNA), DNA cleavage canbe induced at the desired location coded by the sgRNA [6]. Soonafter the discovery of this mechanism, several groups demon-strated that CRISPR–Cas9 is a versatile genome editing tool inhuman cells [7–10]. Now, the CRISPR–Cas9 system has becomean indispensable tool for genome editing applications in a varietyof experimental models.

1.2. Genome editing in iPS cells

A fertilized egg can give rise to essentially all cell types thatmake up the human body. Embryonic stem (ES) cells can be iso-lated from the blastocyst stage of an early developing embryo,yet retain unlimited self-renewal capacity and pluripotency to giverise to various cell types. A more recent stem cell type is theinduced pluripotent stem (iPS) cell, which is indistinguishablefrom ES cells but can be directly converted from the somatic cellsof a donor by transiently transducing a cocktail of transcriptionalfactors [11,12]. The innovation of iPS cells has enabled the isolation

2 H.L. Li et al. /Methods xxx (2015) xxx–xxx

of patient-derived iPS cells, which serve as a valuable resource tostudy disease or test drugs in a dish. iPS cell technology can alsobe used for regenerative medicine, and was applied to replacedamaged retinal pigment epithelium cell sheets in a patient withwet-type aged-related macular degeneration [13] (UMIN ClinicalTrial Registry: UMIN000011929).

The potential of iPS cell technology is enormous, but untilrecently researchers have lacked effective genomic alternationmethods except for gene insertions by viral vectors or transposonvectors. The emergence of new genome editing technologies hasgreatly facilitated the ease of genomic alterations in human iPScells [14]. Now, genome editing in iPS cells is widely used forreporter knock-in, gene knockout and gene correction in humaniPS cells [15–18]. In this review, we describe effective genome edit-ing methods by using the CRISPR–Cas9 system in human iPS cells,shedding light on genome editing techniques.

2. Genome editing experiments in iPS cells

2.1. Strategies for gene disruption, knock-in, and correction

2.1.1. NHEJ mediated frame-shiftFor loss-of-function studies of protein-coding genes, inducing

small deletions or insertions (indels) is required to disrupt theopen reading frame. By targeting an exon by CRISPR–sgRNA, small(typically less than 20 bp) deletions are induced via the NHEJ ormicrohomology-mediated end joining (MMEJ) DNA repair path-ways [19,20]. If the beginning of the protein-coding region, forinstance right after the ‘‘ATG” start codon, is targeted, then it ispossible to completely disrupt the expression of the protein ofinterest. However, the presence of an alternative start codonshould be taken into consideration. In addition, a particular partof the coding region can be targeted to disrupt a specific function,such as an enzymatic domain. In this case, disruption of theprotein-coding frame relies on the size of the deletion inducedby CRISPR–Cas9, and theoretically only two-thirds of deletionevents disrupt the open reading frame.

To enhance the rate of knock-out, two sgRNAs might be used totarget the same gene. This is an effective approach not only for cod-ing genes, but also non-coding genes, as targeting two distal sitesby CRISPR–sgRNAs led to the successful removal of a miRNA clus-ter [21].

2.1.2. HR mediated knock-inHR is a more precise DNA repair pathway than NHEJ, and can be

induced by co-introducing Cas9/sgRNA and a donor DNA template.The targeted site is repaired based on the homology sequences ofthe 50 and 30 arms of a donor template, which contains a selectioncassette with an antibiotic resistant gene, allowing for a desiredDNA element to be inserted at a targeted locus. Since the occur-rence of HR mediated knock-in is rare (typically less than a fewpercent) [22–24], antibiotic selection is desired to obtain success-fully targeted clones. The Cre–loxP system, a well establishedmethod in the mouse genomics field, can be used to remove theselection cassette after the selection. The use of piggyBac DNAtransposon machinery to remove the cassette is an emergingnew approach that does not leave a footprint after the excision[25,26]. Alternatively, repeatedly subdividing the transduced cellswithin a 96-well plate (i.e. 2–10 cells per well of 96-well plate)and quantifying the copy number of knock-in cassettes by dropletdigital PCR in each well can be done to enrich the population withrecombination [24].

The knock-in approach can insert a wide variety of DNAsequences, therefore it has relatively wide application, such asinsertion of an expression cassette into a defined safe-harbor locus

Please cite this article in press as: H.L. Li et al., Methods (2015), http://dx.doi.o

[27], replacement of an endogenous sequence with an exogenousone [28], or the addition of a tag sequence (HA tag, Flag tag, orEGFP, etc.) into the N0- or C0-terminal of a protein [29].

2.1.3. Correcting dystrophin gene frame-shift mutationGenome editing techniques can be utilized to correct disease

causing mutations [30]. Duchenne muscular dystrophy (DMD) isa severe muscle-degenerative disease caused by disruption of thereading frame by frame shifting or nonsense mutation in the dys-trophin gene [31]. The dystrophin gene is located on the X chromo-some and consists of 79 exons, and the loss of exon(s) at the centerregion (i.e. exons 40–55) is the most frequent cause of DMD. In ourprevious study, we derived iPS cell lines from a DMD patient wholacks exon 44. To restore the dystrophin reading frame using theCRISPR–Cas9 system, we tested three different genome editingapproaches, including NHEJ-mediated frame shifting (withoutantibiotic selection) and HR-mediated knock-in of the missingexon 44 (with antibiotic selection) [20]. In the following sections,we provide our experiences on how to achieve successful genomemodification in iPS cells.

2.2. Design and evaluation of sgRNAs

Once the target region is determined, one should search forpotential target sequences that have a PAM (protospacer adjacentmotif) sequence corresponding with CRISPR–sgRNA. The mostwidely used Cas9 is derived from Streptococcus pyogenes (SpCas9)and recognizes the ‘‘NGG” trinucleotide sequence as a PAM and a20-nucleotide target sequence (Fig. 1A). Other Cas9s derived fromorthogonal species have different PAM sequence requirements.For example, Cas9 from Streptococcus thermophilus requires‘‘NAGAA” as a PAM, and Cas9 from Neisseria meningitidis (Nm) rec-ognizes ‘‘NNNNGATT” as a PAM sequence [32,33]. Recently, FengZhang’s group reported another orthogonal Cas9 from Staphylococ-cus aureus (3.1 kb), the cDNA of which is 1 kb shorter than that ofSpCas9 (4.1 kb) [34]. Using the protein 3D structural information ofCas9, Keith Joung’s group engineered the PAM recognition domainof SpCas9 to recognize PAM sequences other than ‘‘NGG” [35].Caution is advised when using orthogonal Cas9s, as they may havevariable DNA cleavage activities [34].

More recently, a novel CRISPR system, called Cpf1, was identi-fied to have DNA cleavage activity similar to Cas9, but severaldistinct features, such as the absence of tracrRNA (single crRNAis sufficient), the presence of ‘‘T” rich PAM (such as ‘‘TTTN”) locatedon the 50 side, and staggered DNA cleavage at the 30 side of crRNA[36]. Further studies are needed to reveal the activities and appli-cations of different CRISPR systems for the modification of iPS cells.

Other than the PAM requirement for the target sequence,extreme bias of ‘‘GC” content (i.e. >80% or <20%) in the targetsequence might affect the DNA cleavage activity [37]. To expressa sgRNA in mammalian cells, most expression vectors use anRNA polymerase III promoter, such as H1 or U6 promoter. Sincethe transcriptional start site of H1 or U6 promoter is typically a‘‘G” (or ‘‘A”), the initial nucleotide of the sgRNA is recommendedto be converted to ‘‘G” for efficient transcription [38]. The alter-ation of the initial nucleotide to ‘‘G” has negligible impact on targetspecificity, as the distal 50 region of sgRNA is flexible. Additionally,a poly ‘‘T” stretch of more than 5 bp should be avoided from thesgRNA target sequence, as ‘‘TTTTT” sequence acts as a terminationsignal for polymerase III promoters. The effect of the secondarystructure of sgRNA has not been elucidated. Other than thesequence restriction of the expression cassette of sgRNA, the epige-netic status of the target DNA sequence may impact the bindingand cleavage activity of Cas9, although direct comparison has notbeen fully addressed. It is reported that CRISPR can cleave methy-lated DNA without significant loss of its cleavage activity [39].

rg/10.1016/j.ymeth.2015.10.015

Fig. 1. Construction of CRISPR–sgRNAs. (A) Targeting sequence for Streptococcus pyogenes (Sp) CRISPR–sgRNA. The initial TSS (transcriptional start site) is better as ‘‘G” or ‘‘A”for transcription from H1 polIII promoter, and the last two nucleotides (position 22 and 23) must be ‘‘GG” for PAM recognition. (B) Oligonucleotide primer sequences forcloning an Sp-sgRNA into pHL-H1-ccdB-EF1a-RiH vector. Copy the initial 20 bp of the position 1–20 sequence from (A) into the forward (fwd) primer (indicated by blue andred ‘‘Ns”). Reverse (rev) primer contains the scaffold from tracrRNA and a terminator signal and can be universally used for any target sequence. (C) Conduct PCR amplificationof the sgRNA part using the two oligo primers from (B) and insert into the BamHI–EcoRI site of the pHL-H1-ccdB vector by In-Fusion reaction or standard ligation. ClonedsgRNA will be driven by H1 promoter, and mRFP expression can be used to monitor transfection efficiency. Hygromycin resistance gene can be used to enrich transfectedcells.

H.L. Li et al. /Methods xxx (2015) xxx–xxx 3

In terms of sgRNA design, the most critical criterion is speci-ficity. This is particularly important when targeting the humangenome, as almost half of the human genome consists of repeatsequences. When designing a sgRNA, such repeat sequences shouldbe avoided. Furthermore, sequence recognition by the Cas9–sgRNAcomplex is imperfect, meaning that some mismatches can be toler-ated [39–41]. Therefore, it is preferable to identify a unique targetsequence that has minimal or no sequence similarity elsewhere inthe genome.

To aid the design of the sgRNA target sequence, there are sev-eral web-based design tools available. Each design tool has itsown algorithm to screen the sgRNA with off target sites acrossthe whole genome. For example, CRISPR Design Tool [39] providesa list of candidate sgRNA target sequences with the off-target scorebased on the sequence’s specificity in the genome of a given spe-cies. Other similar web tools to search sgRNA targeting sitesinclude ZiFiT Targeter [42], E-CRISP [43], GT-Scan [44], CRISPRdi-rect [45], COSMID [46], Cas-OFFinder [47], and so on. These webtools are highly useful to identify candidate sgRNAs in a givenregion of interest. However, to identify the region of interest, i.e.which exon to be targeted, cannot be easily assessed, the above

Please cite this article in press as: H.L. Li et al., Methods (2015), http://dx.doi.o

websites have certain limitations for the size of the querysequences due to computational limitations. To this end, we devel-oped a customized tool to pre-calculate and visualize the genome-wide distribution of unique small sequences (k-mers) in advance.This tool is available on the iGEATs website [20]. In addition, anumber of potential off-target sites can be roughly assessed byDNA sequence alignment or mapping software, such as BLAST[48], Bowtie [49], BWA [50], or GGGenome (https://gggenome.db-cls.jp/en/). It is worth noting that off-target risk or cleavage activitycannot be accurately predicted by any method currently available[51]. We advise analyzing multiple genome-edited clones to con-firm a robust phenotype of interest. So long as the sgRNA is target-ing a specific sequence, it is statistically unlikely that the sameoff-target mutagenesis will occur in multiple cells.

2.3. Construction and evaluation of sgRNA expression vector

2.3.1. Choice of Cas9-sgRNA expression vector and cloningOnce the target sequence for the sgRNA is determined (Fig. 1A),

expression vectors for Cas9 and sgRNA are required. SeveralCRISPR expression vectors are available from Addgene

rg/10.1016/j.ymeth.2015.10.015

4 H.L. Li et al. /Methods xxx (2015) xxx–xxx

(https://www.addgene.org/) or commercial companies. As of July2015, there are 81 entries of Cas9/sgRNA mammalian expressionvectors in the Addgene depository. For efficient expression ofCas9 in human pluripotent stem cells, the EF1a promoter is recom-mended, as it is stronger than the CAG promoter in iPS cells [52].On the other hand, neither the CMV nor SV40 promoter are suit-able for use in iPS cells due to strong transcriptional silencing[53]. In our experiments, we used the pHL-EF1a-SphcCas9-iP-A(Addgene ID: 60599) vector to express a human-codon optimizedSpCas9 under the human EF1a promoter and conjugated with IRES(internal ribosome entry site) and puromycin resistance gene. Toclone a sgRNA into our expression vector, we designed twooligonucleotide primers, as indicated in Fig. 1B: a forward primerwhich contains the sgRNA target sequence and a reverse primerwhich contains a scaffold of sgRNA. The two primers are PCRamplified and cloned into pHL-H1-ccdB-mEF1a-RiH (Addgene ID:60601) by In-Fusion reaction to express a desired sgRNA underH1 RNA polymerase III promoter (Fig. 1C). To enhance the cloningefficiency of sgRNA, the Escherichia coli negative selection markerccdB gene is incorporated in the sgRNA cloning (BamHI–EcoRI) siteto abolish self-ligation of the vector or contamination of non-cleaved vectors. In addition, transfection efficiency can be assessedby mRFP1 expression and hygromycin resistance gene.

2.3.2. SSA assay using a luciferase reporter assay in HEK293T cellsSince cleavage activity of a sgRNA may vary based on the

sequence composition or unknown properties, it is highly recom-mended to construct several sgRNAs for one target site and evalu-ate the cleavage activity of each in advance (Fig. 2A). There areseveral assay systems for evaluating the cleavage activity ofCRISPR–sgRNA, as described in Section 3.2 below, but we normallyevaluate the sgRNA activity by either a single strand annealing(SSA) assay or T7 endonuclease I (T7EI) assay in HEK293 or 293Tcells. HEK293 is a widely used human cell line which has hightransfection efficiency and proliferation ability and is thereforeideal to evaluate sgRNA activity quickly. For the SSA assay, thesgRNA target locus (which contains some sgRNA target sites) isincorporated into the middle of a luciferase gene flanked by homol-ogy regions. When the target site is successfully cleaved, the luci-ferase gene will be restored via the SSA DNA repair pathway(Fig. 2B). CRISPR–Cas9 and sgRNA expression vectors are co-transfected into 293T cells in a 96-well plate using Lipofectamine2000 (Life Technologies). The luciferase activity is assessed as theCRISPR–sgRNA cutting activity towards the targeted region after24 h using Dual-Glo Luciferase assay system (E2920, Promega).Instead of luciferase, GFP-based SSA reporter is also available[54]. In our dystrophin targeted site, all five sgRNAs we con-structed showed consistently high cleavage activity by the SSAassay in 293T cells (Fig. 2C). The SSA assay can also be performedin target iPS cells, however, the CMV promoter needs to bereplaced with CAG or EF1a to avoid transcriptional silencing inhuman iPS cells [53].

2.4. Construction of targeting vector

A donor template vector for knock-in experiments is con-structed by conjugating two homology arms (50 and 30 arm) anda hygromycin selection cassette using the In-Fusion HD cloningkit (Clontech). Our targeting vector, pENTR-DMD-Donor, is avail-able from Addgene (Addgene ID: 60605) to replace both homologyarms. The 50 arm with a loxP site can be replaced by XbaI digestion,and the 30 arm can be replaced by NdeI and BamHI digestion. Thelength of the homology arms should be around 700–1000 bp [25]and start adjacent to the sgRNA target site. The homology armsshould not include the sgRNA targeting site (i.e. introduction ofsilent mutation at PAM) or repetitive regions (i.e. LINE or SINE).

Please cite this article in press as: H.L. Li et al., Methods (2015), http://dx.doi.o

3. Transfection of CRISPR–Cas9 into human iPS cells

For successful cleavage of DNA at a target site, the efficienttransfection of Cas9 and sgRNA into the cells of interest isabsolutely essential. There are several transfection approaches,such as lipofection or electroporation methods available [55], butwe recommend to use electroporation as the transfection effi-ciency is higher. Among several electroporators, such as Nucleofec-tor (Lonza), Neon (Life Technologies) and NEPA21 (Nepagene),Nucleofector is the most widely used for genome editing experi-ments in human iPS cells [22]. We use the NEPA21 electroporator,as it can achieve high transfection efficiency under controllableelectric pulse conditions specified by the user and has low celltoxicity and low running cost.

3.1. Cas9/sgRNA plasmid transfection using NEPA21 electroporator

To achieve the highest plasmid transfection efficiency in ourhuman iPS cells, we optimized the conditions of eletroporation,such as the pulse voltage and pulse wide, in advance. We trans-fected an EGFP expression vector into three different human iPScell lines, 201B7 [11], 404C2 [56], and DMD patient-derived iPS cellline CiRA00111 [20], and assessed the transfection efficiency usingflow cytometry (Fig. 3A). Among the conditions tested, weobserved the highest transfection efficiency at voltage 125 V andpulse width 5 ms of poring pulse.

Higher transfection efficiency in human iPS cell lines usingNEPA21 can be achieved by manipulating cell growth conditions,acquiring high purity of the plasmids, and avoiding cell damageduring dissociation. The iPS cells should be cultured under ahealthy undifferentiated state at a logarithmic growth stage andnever allowed to become confluent. Transfected plasmid DNAshould be prepared with high purity using an endotoxin-free col-umn purification system and dissolved in TE (Tris–EDTA) bufferat 4–5 lg ll�1 concentration. Contamination of ionic salt greatlyaffects the condition of the electric field during electroporation,hence the carry-over volume of the plasmid DNA solution shouldbe minimized. Lastly, human ES/iPS cells easily induce apoptosisby activation of the ROCK (Rho-associated protein kinase) pathwaywhen dissociated into single cells. Therefore, ROCK inhibitorY-27632 (final 10 lM) should be added at least 1 h in advance oftransfection to prevent cell death, and electroporated cells shouldbe transferred into media containing Y-27632 as soon as possible.After transfection, the cells can be seeded on feeder layer for bettercell viability or feeder-free culture, such as Matrigel-mTeSR mediaor Laminin-511 E8 fragment culture [57], to avoid contaminationof mouse genomic DNA during genotyping. Y-27632 should bekept in culture media for 2–3 days after seeding. The transfectediPS cells should recover from the transfection damage after oneto two passages. At the same time, unseeded cells should be col-lected to extract genomic DNA for checking the cleavage activityof CRISPR–sgRNA as a bulk population using the methodsdescribed below (Fig. 3B).

3.2. Evaluation of CRISPR-mediated cleavage activity in human iPScells

As shown in Table 1, there are several methods for evaluatingthe cleavage activities of CRISPR–sgRNA at the target sequence.

Among the described methods above, we recommend the T7EIassay and/or restriction fragment length polymorphism (RFLP)assay for initial testing in human iPS cells. The T7EI assay amplifiesthe target sequence by a pair of PCR primers and then treatment ofthe re-hybridized PCR product by T7 Endonuclease I specificallycleaves imperfectly matched DNA. This assay is available for

rg/10.1016/j.ymeth.2015.10.015

Fig. 2. Testing cleavage activity of CRISPR–sgRNAs in 293T cells by SSA assay. (A) Examples of designed Sp-CRISPR–sgRNAs targeting exon 45 of the dystrophin gene onhuman X chromosome. sgRNA1-4 target the sense strand, whereas sgRNA5 targets the antisense strand relative to the dystrophin gene. (B) Principle of the SSA assay fordetecting the DNA cleavage activity of CRISPR–sgRNAs. Luciferase cDNA is split by the target sequence of the five sgRNAs. Once DNA cleavage is induced, luciferase cDNA isrestored via the single strand annealing (SSA) DNA repair pathway. (C) The SSA activities of CRISPR–sgRNAs measured in HEK293T cells. All five sgRNAs showed similarcleavage activities in this case. The values represent mean ± SD among technical replicates (n = 3).

Fig. 3. Evaluation of CRISPR activity in human iPS cells by NEPA21 transfection. (A) Optimization of the transfection condition using NEPA electroporator in human iPS celllines. Voltage and pulse width of the poring pulses were investigated. EGFP expressing plasmid was transfected, and the percentage of positive cells was analyzed byLSRFortessa (BD). (B) Schematic overview of the transfection procedures. (C) After the electroporation of Cas9 and sgRNAs into iPS cells, the cleavage activity of theendogenous dystrophin gene was assessed by the T7EI assay. Band intensities of the cleaved bands (indicated by red arrowhead) were detected by TapStation (Agilent), andvalues are indicated below the gel-electrogram image. (D) The same genomic DNA samples from (C) were also analyzed by the RFLP assay using the restriction enzyme XcmI(50-CCANNNNN^NNNNTGG-30). Band intensities of the uncleaved band (red arrowhead) are indicated below. Since sgRNA5 cleaves far from the XcmI site, no cleavage activitywas detected by the RFLP assay.

H.L. Li et al. /Methods xxx (2015) xxx–xxx 5

Please cite this article in press as: H.L. Li et al., Methods (2015), http://dx.doi.org/10.1016/j.ymeth.2015.10.015

Table 1Various assay systems for quantifying CRISPR–Cas9/sgRNA cleavage activity.

Assay Principle Required reagents Required equipment

SSA (single strandannealing) assay [54,58]

A target sequence is inserted into the middle of a reporter gene(i.e. luciferase) flanking two homology regions. DSB induces HRor SSA repair pathways to restore the reporter gene

Reporter plasmid, reporter detection kit(i.e. luciferase assay kit for luciferasereporter)

Microplate reader (forluciferase) or flowcytometer (for GFP)

Surrogate reporter (frame-shifted EGFP) [59]

A target sequence is inserted in front of the reporter gene(s) (i.e.GFP or lacZ). DSB induces NHEJ mediated indels to restore thereading frame of the reporter gene(s)

Reporter plasmid Flow cytometer (forGFP) or microplatereader (for luciferase)

Heteroduplex dsDNAspecific nuclease (suchas CEL-I or T7EI) assay[60]

Original and mutated sequences form a heteroduplex after re-annealing. The heteroduplex region is cleaved by a nuclease

Heteroduplex dsDNA specific nuclease,such as CEL-I, Surveyor nuclease (CEL-II), orT7EI (T7 endonuclease I)

PCR, Gelelectrophoresis and gelimager

RFLP (Restriction FragmentLength Polymorphism)assay [61]

A restriction enzyme site overlapping the nuclease targeting site. PCR primers and reagents to amplify thetarget sequence and appropriate restrictionenzyme

PCR, Gelelectrophoresis and gelimager

E. coli subcloning andsanger sequencing

PCR amplification of the target region and subcloning by E. coli.Sanger sequence dozens of clones to identify the indel rate

PCR primers and reagents to amplify thetarget sequence, ligase, competent E. coli,and Sanger sequencing reagents

PCR, E. coli incubator,Sanger sequencer

Deep sequencing [62] PCR amplify the target region and massive parallel sequencer isused to identify indel rate

PCR primers and reagents to amplify thetarget sequence and deep sequencer librarypreparation kit

PCR, Deep sequencersuch as MiSeq (Illumia)

High-resolution meltanalysis [63,64]

PCR amplify the target region and measure the meltingtemperature. Original and mutated PCR products form aheteroduplex, which shows lower melting temperature

PCR primers and reagents to amplify thetarget sequence, and HRMA specific DNAdye (i.e. LCGreen Plus)

HRMA compatible Realtime PCR machine (i.e.LigtScanner)

6 H.L. Li et al. /Methods xxx (2015) xxx–xxx

basically any endogenous target locus, but PCR primers mustamplify the target locus specifically and should avoid heterozygousmutation or SNP sites and contamination of mouse genomic DNAfrom mouse feeder layer. Assay conditions of T7EI enzyme incuba-tion or the re-hybridization process might need optimization inadvance to obtain cleaner results.

Another assay method is the RFLP assay, which can be used toquantify the amount of non-cleaved PCR product and becomesresistant to restriction enzyme treatment. To perform this assay,a restriction enzyme site must exist in or around the CRISPR–Cas9 cleavage site. Although the incubation time with the restric-tion enzyme requires longer than the T7EI assay and resultsdepend on the type of restriction enzyme used, the RFLP assaytends to give more sensitive results than the T7EI assay, sincethe sequence recognition ability (1 bp difference) of restrictionenzymes is typically better than the recognition of heteroduplexes(or single-strand DNA) by the T7EI enzyme.

3.2.1. T7EI assayTo assess cleavage activity, Cas9- and sgRNA-transfected human

iPS cells are better to be cultured without feeders to avoid contam-ination of mouse genomic DNA and to be seeded in a small well (i.e.12-well plate) for better and quicker recovery of the cells. Then,design a pair of PCR primers to amplify the target region (500–1200 bp in size) while keeping the CRISPR–sgRNA target site off-set from the center of the amplicon so that the cleaved PCR productscan be separated by electrophoresis. A high-fidelity polymerase (i.e.PrimeSTAR HS DNA Polymerase, TaKaRa) should be used for ampli-fication and PCR product should be purified by Wizard SV Gel andthe PCR Clean-Up System (Promega). If non-specific amplificationis observed, we recommend optimizing the PCR condition (i.e. rais-ing Tm) or re-designing the primers. In our experiments, purifiedPCR products (400 ng) were denatured (95 �C for 5 min) and re-annealed (gradually cooled from 95 �C to 85 �C at �2 �C sec�1 and85 �C to 25 �C at �0.1 �C sec�1) in NEBuffer 2 (NEB) using a thermo-cycler. Then, digest the re-annealed PCR product with 10 units ofT7EI (NEB) for 15 min at 37 �C. Stop the reaction by adding0.25 M EDTA solution, and place the sample on ice. Finally, analyzethe treated PCR products by agarose gel or High Sensitivity D1000ScreenTape (5067–5584, Agilent Technologies) using Agilent 2200TapeStation system (Agilent Technologies) (Fig. 3C). Make sure totreat the non-transfected control sample as a negative control, as

Please cite this article in press as: H.L. Li et al., Methods (2015), http://dx.doi.o

sometimes minor background bands may be present even in thenegative control sample due to the background cleavage of T7EInuclease or in the rare case due to the presence of heterozygousSNP or genomic mutations in the cell line.

3.2.2. RFLP assayFor the RFLP assay of NHEJ products, the sgRNA cleavage site

must contain a restriction enzyme site. The RFLP assay can alsobe used to assess HR mediated knock-in efficiency by insertingan additional restriction enzyme site in the donor template. Thesame PCR primers designed for the T7EI assay can be utilized forthe RFLP assay. Digest purified PCR products (generated the sameway as in the T7EI assay) by a restriction enzyme with the appro-priate buffer. Analyze the amount of digested and undigested PCRproducts by 2% agarose gel electrophoresis. We tested the activityof our sgRNAs by the RFLP assay with the XcmI restriction enzymelocated in the DMD-CRISPR cleavage site (Fig. 3D). Importantly, theRFLP assay is highly dependent on the distance between the cleav-age site and the restriction enzyme site used. In our case, sgRNA5cleaved 20 bp away from the XcmI site and its cleavage activitycould not be detected by RFLP assay.

Once the cleavage or mutation efficiency is determined as abulk, you can estimate how many clones should be analyzed toobtain the desired mutant clones. For example, if your indel rateis 10%, you need to screen at least 10 clones to obtain one clonewith some kind of indel. If the indel rate is 1%, you must screenat least 100 clones. Since the clonal isolation of human iPS cellsand the culture of multiple iPS cell clones is laborious and demand-ing, we highly recommend optimizing the efficiency of the trans-fection conditions and CRISPR–sgRNA cleavage activity before thesubcloning experiments.

4. Clonal isolation of human iPS cells and genotyping

Once a bulk cell population is successfully obtained with a rea-sonable indel rate or HR-mediated knock-in, proceed to subcloningand genotyping to obtain the clones of interest. For NHEJ mediatedinduction of indels or ssODN mediated knock-in experiments, drugselection is not applicable, so simply perform subcloning withoutselection. For ssODNmediated knock-in experiments, a small pool-ing approach or sub-selection method might be preferred, as theknock-in efficiency is typically very low (less than 1%) [24].

rg/10.1016/j.ymeth.2015.10.015

Fig. 4. Flowchart for isolating genome-edited iPS cell clones. (A) Isolation of genome-edited iPS cell clones without drug selection. Once successful genome editing isconfirmed in the bulk cell population by the T7EI or RFLP assay, spread the cells into two dishes; one dish for clump passaging and making frozen stocks and the other dish fordissociating into single cells and seeding onto feeder dishes at a density of 200–400 cells per 100 mm dish for clonal colony formation. After the formation of iPS cell colonies,scratch a part of the colony as a template for the PCR reaction and subsequent Sanger sequencing. The leftover cells in the same colony can be transferred into a well of a 24-well plate for further expansion. (B) Isolation of genome-edited iPS cell clones with drug selection. We recommend starting the drug selection after passaging the transfectedcells rather than right after the transfection to avoid extra stress on the damaged cells. It is important to apply the same drug selection to non-transfected cells as a control forthe drug selection. Ideally, the concentration of antibiotic drug should be determined in advance. Once drug-resistant cells reach a sub-confluent state, split the cells into twodishes and one microcentrifuge tube. Then, extract genomic DNA for bulk PCR analysis (by using primers flanking the knock-in cassette) to roughly estimate the rate of knock-in events. Colony isolation is the same as described in (A).

H.L. Li et al. /Methods xxx (2015) xxx–xxx 7

4.1. Single cell isolation without drug selection

To subclone iPS cell clones without drug selection, we performthe procedures listed in the flowchart in Fig. 4A. For genotyping,we recommend performing Sanger sequencing to sequence the tar-get site. Assessing a single copy gene, such as the dystrophin geneon X chromosome in male iPS cells, is straightforward, but autoso-mal genes are more challenging, as there are two copies andCRISPR mediated mutation is sometimes heterozygous. Still, San-ger sequencing results can be a mixture of two electrograms fromthe point of the cleavage site. It is advised to decide and optimizethe assay methods in advance to screen the isolated clones.

Please cite this article in press as: H.L. Li et al., Methods (2015), http://dx.doi.o

(a) Dissociate the bulk population of edited iPS cells into singlecell suspension by Accutase or TrypLE under Y-27632 treat-ment. Count the cells and seed 200–400 cells into a 100-mmdish with feeder layer. Cultivate for around 10–12 days toform colonies (2–4 mm in diameter).

(b) Aim for well-isolated and round shape iPS cell colonies.Scratch a center portion of the colony by a clean 10 ll tipand put the scratched cells into PCR buffer (5 ll per well)containing Proteinase K (final concentration, 0.2 lg ll–1) ina 96-well PCR plate. Only a small portion of the cells(�100 cells) is normally sufficient to serve as a templatefor PCR analysis (Fig. 4A).

rg/10.1016/j.ymeth.2015.10.015

8 H.L. Li et al. /Methods xxx (2015) xxx–xxx

(c) Scratch off the leftover portion of the colony by anotherclean crystal tip and transfer all of the cells into a well of a24-well culture plate with iPS cell medium (0.5 ml) contain-ing Y-27632 (final concentration, 10 lM). Repeat the colonypicking procedures to obtain as many clones as possible.

(d) Seal the 96-well plate from (b) by sticking a seal and incu-bate at 55 �C for 3–6 h to lyse the cells. After adding 10 llof milli Q water into each well, use 0.5–1.0 ll (50–200 ngof genomic DNA) for the template of the PCR reaction. Per-form PCR amplification and run on gel electrophoresis toconfirm specific and sufficient amplification. Treat 5 ll ofPCR product by exonuclease I (0.1 unit, TaKaRa) and shrimpalkaline phosphatase (0.25 unit, TaKaRa) at 37 �C for 30 minto remove free primers and dNTPs. Then, use 1–2 ll of thetreated PCR product as a template and perform Sangersequencing to detect the mutation at the target site.

(e) Based on the sequencing results from above, expand thepositive clones of interest further to a larger culture plate(i.e. 6-well plate). It is always a good idea to keep one ortwo non-modified clones as a negative control.

4.2. Single cell isolation with drug selection

For knock-in experiments, the basic procedures are the same asabove except for co-transfection of a donor DNA template (as cir-cular plasmid DNA) and drug selection (Fig. 4B). Another markeddifference is the genotyping process. Because a donor vectorcontains a foreign reporter element, the PCR assay for detectingsuccessful knock-in is effective at distinguishing heterozygous orhomozygous targeted clones when using a pair of primers flankingthe outside of both homology arms (c.f. short PCR product for wildtype allele and long PCR product for the targeted allele). Afterscreening by the PCR assay, Southern blot analysis is recommendedto assess the copy number of the selection cassette, as randomintegration or concatemerized knock-in is also sometimesobserved.

5. Off-target analysis

Off-target analysis is recommended for genome-edited clonesbefore phenotypic experiments, as undesired genomic alterationsmay confuse experimental results (i.e. disruption of anotherimportant gene). In particular, we encourage confirming normalkaryotyping (i.e. G-banding) of the parental and isolated iPS cellclones, as abnormal karyotyping is often observed in prolongediPS cell culture and the affected gene number could be generallylarge. For basic research purposes, it is important to check thereproducibility of the results in multiple clones or multiplesgRNAs, as it is less likely that multiple clones generated by multi-ple sgRNAs result in the same off-target mutagenesis. If desired,candidate off-target sites can be PCR amplified to check potentialmutations by T7EI assay or deep sequencing. Several other non-biased genome wide assays for addressing off-target mutagenesisare discussed elsewhere [51,65].

6. Conclusions

In this review, we summarized fundamental yet essential tech-niques for achieving efficient genome editing in iPS cells by theCRISPR–Cas9 system, such as the determination of a target region,construction of CRISPR–sgRNAs, optimization of transfection, andevaluation of sgRNA cleavage activities. Moreover, to obtain theedited clones of interest with confidence, we provide a step-by-step flowchart to confirm each procedure to deal with the fragilenature of human iPS cells. Efficient genomic editing in human iPS

Please cite this article in press as: H.L. Li et al., Methods (2015), http://dx.doi.o

cells has tremendous opportunity for disease modeling, functionalgenomics, gene therapy and beyond. We hope that our methodswill be helpful for the editing genes of interest and move forwardresearch fields using iPS cells.

Acknowledgments

The authors apologize to all whose work could not be cited dueto space limitations. We gratefully acknowledge Dr. PeterKaragiannis for critically reading the manuscript. A. H. is supportedin part by JSPS KAKENHI grant (15H05581) and AMED ResearchCenter Network for Realization of Regenerative Medicine grants.

References

[1] T. Yamamoto, Targeted Genome Editing Using Site-Specific Nucleases,Springer, 2015.

[2] J.D. Sander, J.K. Joung, Nat. Biotechnol. 32 (2014) 347–355.[3] H. Kim, J.S. Kim, Nat. Rev. Genet. 15 (2014) 321–334.[4] P.D. Hsu, E.S. Lander, F. Zhang, Cell 157 (2014) 1262–1278.[5] J.A. Doudna, E. Charpentier, Science 346 (2014) 1258096.[6] M. Jinek, K. Chylinski, I. Fonfara, M. Hauer, J.A. Doudna, E. Charpentier, Science

337 (2012) 816–821.[7] L. Cong, F.A. Ran, D. Cox, S. Lin, R. Barretto, N. Habib, P.D. Hsu, X. Wu, W. Jiang,

L.A. Marraffini, F. Zhang, Science 339 (2013) 819–823.[8] P. Mali, L. Yang, K.M. Esvelt, J. Aach, M. Guell, J.E. DiCarlo, J.E. Norville, G.M.

Church, Science 339 (2013) 823–826.[9] S.-W. Cho, S. Kim, J.M. Kim, J.-S. Kim, Nat. Biotechnol. 31 (2013) 230–232.[10] M. Jinek, A. East, A. Cheng, S. Lin, E. Ma, J. Doudna, Elife 2 (2013) e00471.[11] K. Takahashi, K. Tanabe, M. Ohnuki, M. Narita, T. Ichisaka, K. Tomoda, S.

Yamanaka, Cell 131 (2007) 861–872.[12] K. Takahashi, S. Yamanaka, Cell 126 (2006) 663–676.[13] H. Kamao, M. Mandai, S. Okamoto, N. Sakai, A. Suga, S. Sugita, J. Kiryu, M.

Takahashi, Stem Cell Rep. 2 (2014) 205–218.[14] A. Hotta, S. Yamanaka, Annu. Rev. Genet. 49 (2015).[15] D. Hockemeyer, F. Soldner, C. Beard, Q. Gao, M. Mitalipova, R.C. DeKelver, G.E.

Katibah, R. Amora, E.A. Boydston, B. Zeitler, X. Meng, J.C. Miller, L. Zhang, E.J.Rebar, P.D. Gregory, F.D. Urnov, R. Jaenisch, Nat. Biotechnol. 27 (2009) 851–857.

[16] D. Hockemeyer, H. Wang, S. Kiani, C.S. Lai, Q. Gao, J.P. Cassady, G.J. Cost, L.Zhang, Y. Santiago, J.C. Miller, B. Zeitler, J.M. Cherone, X. Meng, S.J. Hinkley, E.J.Rebar, P.D. Gregory, F.D. Urnov, R. Jaenisch, Nat. Biotechnol. 29 (2011) 731–734.

[17] Q. Ding, Y.-K. Lee, E.A.K. Schaefer, D.T. Peters, A. Veres, K. Kim, N. Kuperwasser,D.L. Motola, T.B. Meissner, W.T. Hendriks, M. Trevisan, R.M. Gupta, A. Moisan,E. Banks, M. Friesen, R.T. Schinzel, F. Xia, A. Tang, Y. Xia, E. Figueroa, A. Wann, T.Ahfeldt, L. Daheron, F. Zhang, L.L. Rubin, L.F. Peng, R.T. Chung, K. Musunuru, C.A. Cowan, Cell Stem Cell 12 (2013) 238–251.

[18] Q. Ding, S.N. Regan, Y. Xia, L.A. Oostrom, C.A. Cowan, K. Musunuru, Cell StemCell 12 (2013) 393–394.

[19] S. Bae, J. Kweon, H.S. Kim, J.-S. Kim, Nat. Methods 11 (2014) 705–706.[20] H.L. Li, N. Fujimoto, N. Sasakawa, S. Shirai, T. Ohkame, T. Sakuma, M. Tanaka, N.

Amano, A. Watanabe, H. Sakurai, T. Yamamoto, S. Yamanaka, A. Hotta, StemCell Rep. 4 (2015) 143–154.

[21] T.T. Ho, N. Zhou, J. Huang, P. Koirala, M. Xu, R. Fung, F. Wu, Y.Y. Mo, NucleicAcids Res. 43 (2015) e17.

[22] F.A. Ran, P.D. Hsu, J. Wright, V. Agarwala, D.A. Scott, F. Zhang, Nat. Protoc. 8(2013) 2281–2308.

[23] L. Yang, M. Guell, S. Byrne, J.L. Yang, A. De Los Angeles, P. Mali, J. Aach, C. Kim-Kiselak, A.W. Briggs, X. Rios, P.-Y. Huang, G. Daley, G. Church, Nucleic AcidsRes. 41 (2013) 9049–9061.

[24] Y. Miyaoka, A.H. Chan, L.M. Judge, J. Yoo, M. Huang, T.D. Nguyen, P.P. Lizarraga,P.-L. So, B.R. Conklin, Nat. Methods 11 (2014) 291–293.

[25] K. Yusa, Nat. Protoc. 8 (2013) 2061–2078.[26] K. Yusa, S.T. Rashid, H. Strick-Marchand, I. Varela, P.-Q. Liu, D.E. Paschon, E.

Miranda, A. Ordóñez, N.R.F. Hannan, F.J. Rouhani, S. Darche, G. Alexander, S.J.Marciniak, N. Fusaki, M. Hasegawa, M.C. Holmes, J.P. Di Santo, D.A. Lomas, A.Bradley, L. Vallier, Nature 478 (2011) 391–394.

[27] F. Gonzalez, Z. Zhu, Z.D. Shi, K. Lelli, N. Verma, Q.V. Li, D. Huangfu, Cell StemCell 15 (2014) 215–226.

[28] S.M. Byrne, L. Ortiz, P. Mali, J. Aach, G.M. Church, Nucleic Acids Res. 43 (2015)e21.

[29] H. Yang, H. Wang, C.S. Shivalila, A.W. Cheng, L. Shi, R. Jaenisch, Cell 154 (2013)1370–1379.

[30] H.L. Li, T. Nakano, A. Hotta, Dev. Growth Differ. 56 (2014) 63–77.[31] C. Pichavant, A. Aartsma-Rus, P.R. Clemens, K.E. Davies, G. Dickson, S.a.i.

Takeda, S.D. Wilton, J.A. Wolff, C.I. Wooddell, X. Xiao, J.P. Tremblay, Mol. Ther.19 (2011) 830–840.

[32] Z. Hou, Y. Zhang, N.E. Propson, S.E. Howden, L.-F. Chu, E.J. Sontheimer, J.A.Thomson, Proc. Natl. Acad. Sci. 110 (2013) 15644–15649.

rg/10.1016/j.ymeth.2015.10.015

H.L. Li et al. /Methods xxx (2015) xxx–xxx 9

[33] K.M. Esvelt, P. Mali, J.L. Braff, M. Moosburner, S.J. Yaung, G.M. Church, Nat.Methods 10 (2013) 1116–1121.

[34] F.A. Ran, L. Cong, W.X. Yan, D.A. Scott, J.S. Gootenberg, A.J. Kriz, B. Zetsche, O.Shalem, X.B. Wu, K.S. Makarova, E.V. Koonin, P.A. Sharp, F. Zhang, Nature 520(2015) 186–191.

[35] B.P. Kleinstiver, M.S. Prew, S.Q. Tsai, V.V. Topkar, N.T. Nguyen, Z. Zheng, A.P.W.Gonzales, Z. Li, R.T. Peterson, J.-R.J. Yeh, M.J. Aryee, J.K. Joung, Nature 523(2015) 481–485.

[36] B. Zetsche, J.S. Gootenberg, O.O. Abudayyeh, I.M. Slaymaker, K.S. Makarova, P.Essletzbichler, S.E. Volz, J. Joung, J. van der Oost, A. Regev, E.V. Koonin, F. Zhang,Cell 163 (2015) 759–771.

[37] T. Wang, J.J. Wei, D.M. Sabatini, E.S. Lander, Science 343 (2014) 80–84.[38] H. Koike-Yusa, Y. Li, E.-P. Tan, M.D.C. Velasco-Herrera, K. Yusa, Nat. Biotechnol.

32 (2014) 267–273.[39] P.D. Hsu, D.A. Scott, J.A. Weinstein, F.A. Ran, S. Konermann, V. Agarwala, Y. Li, E.

J. Fine, X. Wu, O. Shalem, T.J. Cradick, L.A. Marraffini, G. Bao, F. Zhang, Nat.Biotechnol. 31 (2013) 827–832.

[40] Y. Fu, J.A. Foden, C. Khayter, M.L. Maeder, D. Reyon, J.K. Joung, J.D. Sander, Nat.Biotechnol. 31 (2013) 822–826.

[41] C. Kuscu, S. Arslan, R. Singh, J. Thorpe, M. Adli, Nat. Biotechnol. 32 (2014) 677–683.

[42] J.D. Sander, P. Zaback, J.K. Joung, D.F. Voytas, D. Dobbs, Nucleic Acids Res. 35(2007) W599–605.

[43] F. Heigwer, G. Kerr, M. Boutros, Nat. Methods 11 (2014) 122–123.[44] A. O’brien, T.L. Bailey, Bioinformatics 30 (2014) 2673–2675.[45] Y. Naito, K. Hino, H. Bono, K. Ui-Tei, Bioinformatics 31 (2015) 1120–1123.[46] T.J. Cradick, P. Qiu, C.M. Lee, E.J. Fine, G. Bao, Mol. Ther. Nucleic Acids 3 (2014)

e214.[47] S. Bae, J. Park, J.-S. Kim, Bioinformatics 30 (2014) 1473–1475.[48] S.F. Altschul, W. Gish, W. Miller, E.W. Myers, D.J. Lipman, J. Mol. Biol. 215

(1990) 403–410.

Please cite this article in press as: H.L. Li et al., Methods (2015), http://dx.doi.o

[49] B. Langmead, C. Trapnell, M. Pop, S.L. Salzberg, Genome Biol. 10 (2009) R25.[50] H. Li, R. Durbin, Bioinformatics 26 (2010) 589–595.[51] K. Ishida, P. Gee, A. Hotta, Int. J. Mol. Sci. 16 (2015).[52] H. Matsui, N. Fujimoto, N. Sasakawa, Y. Ohinata, M. Shima, S. Yamanaka, M.

Sugimoto, A. Hotta, PLoS ONE 9 (2014) e104957.[53] A. Hotta, J. Ellis, J. Cell. Biochem. 105 (2008) 940–948.[54] D. Mashiko, Y. Fujihara, Y. Satouh, H. Miyata, A. Isotani, M. Ikawa, Sci. Rep. 3

(2013) 3355.[55] M. Costa, M. Dottori, K. Sourris, P. Jamshidi, T. Hatzistavrou, R. Davis, L. Azzola,

S. Jackson, S.M. Lim, M. Pera, A.G. Elefanty, E.G. Stanley, Nat. Protoc. 2 (2007)792–796.

[56] K. Okita, Y. Matsumura, Y. Sato, A. Okada, A. Morizane, S. Okamoto, H. Hong, M.Nakagawa, K. Tanabe, K.-I. Tezuka, T. Shibata, T. Kunisada, M. Takahashi, J.Takahashi, H. Saji, S. Yamanaka, Nat. Methods 8 (2011) 409–412.

[57] M. Nakagawa, Y. Taniguchi, S. Senda, N. Takizawa, T. Ichisaka, K. Asano, A.Morizane, D. Doi, J. Takahashi, M. Nishizawa, Y. Yoshida, T. Toyoda, K. Osafune,K. Sekiguchi, S. Yamanaka, Sci. Rep. 4 (2014) 3594.

[58] H. Ochiai, K. Fujita, K.-I. Suzuki, M. Nishikawa, T. Shibata, N. Sakamoto, T.Yamamoto, Genes Cells 15 (2010) 875–885.

[59] S. Ramakrishna, S.-W. Cho, S. Kim, M. Song, R. Gopalappa, J.-S. Kim, H. Kim, Nat.Commun. 5 (2014) 3378.

[60] L. Vouillot, A. Thelie, N. Pollet, G3 5 (2015) 407–415.[61] F.D. Urnov, E.J. Rebar, M.C. Holmes, H.S. Zhang, P.D. Gregory, Nat. Rev. Genet.

11 (2010) 636–646.[62] C.C. Bell, G.W. Magor, K.R. Gillinder, A.C. Perkins, BMC Genomics 15 (2014)

1002.[63] T.J. Dahlem, K. Hoshijima, M.J. Jurynec, D. Gunther, C.G. Starker, A.S. Locke, A.

M. Weis, D.F. Voytas, D.J. Grunwald, PLoS Genet. 8 (2012) e1002861.[64] H.R. Thomas, S.M. Percival, B.K. Yoder, J.M. Parant, PLoS ONE 9 (2014) e114632.[65] T. Koo, J. Lee, J.S. Kim, Mol. Cells 38 (2015) 475–481.

rg/10.1016/j.ymeth.2015.10.015