Abstract

Recent advances in the era of genetic engineering have significantly improved our ability to make precise changes in the genomes of human cells. Throughout the years, clinical trials based on gene therapies have led to the cure of diseases such as X-linked severe combined immunodeficiency (SCID-X1), adenosine deaminase deficiency (ADA-SCID) and Wiskott–Aldrich syndrome. Despite the success gene therapy has had, there is still the risk of genotoxicity due to the potential oncogenesis introduced by utilising viral vectors. Research has focused on alternative strategies like genome editing without viral vectors as a means to reduce genotoxicity introduced by the viral vectors. Although there is an extensive use of RNA-guided genome editing via the clustered regularly interspaced short palindromic repeats (CRISPR) and associated protein-9 (Cas9) technology for biomedical research, its genome-wide target specificity and its genotoxic side effects remain controversial. There have been reports of on- and off-target effects created by CRISPR–Cas9 that can include small and large indels and inversions, highlighting the potential risk of insertional mutagenesis. In the last few years, a plethora of in silico, in vitro and in vivo genome-wide assays have been introduced with the sole purpose of profiling these effects. Here, we are going to discuss the genotoxic obstacles in gene therapies and give an up-to-date overview of methodologies for quantifying CRISPR–Cas9 effects.

Introduction

Vector-mediated genotoxicity is a recognised safety concern of viral vector-based gene therapy [1]. Understanding the causative agents in viral vector genotoxicity will aid in improving pre-clinical safety assessment of these vectors. Starting with viral vector-mediated insertional mutagenesis (gain-of-function mutations, gene truncations, genomic translocations, etc.), the vector design, integration site and insertion profile become central matters in viral vector-related genotoxicity. Efforts have been made to develop non-viral vector delivery with less insertional mutagenesis risk [24]. There has been an increase in investigating new technologies, such as clustered regularly interspaced short palindromic repeats (CRISPR) and associated protein-9 (Cas9), as a more specific targeted genome editor. Several research groups have devised many viral and non-viral vectors as appropriate delivery systems, especially since CRISPR–Cas9 allows for broad compatibility due to the small size of its two components.

Since the development of CRISPR–Cas9 system over a decade ago [5,6] and its first use in genomic engineering research [710], this technology has revolutionised the world of gene editing. Nowadays, the CRISPR–Cas9 system has been applied to many different projects in order to correct specific genomic sites that are related to disease. The technology utilises short single guide RNA sequences (sgRNA) that bind to a specific target DNA sequence in the genome and to the Cas9 enzyme which scans the genome for protospacer adjacent motifs (PAMs). Upon recognition of complementary DNA sequence by the sgRNA and its associated PAM, Cas9 proceeds with the targeted cleavage of the DNA. The outcome of Cas9-mediated DNA cleavage is a double-stranded break (DSB) within the target DNA (∼3–4 nucleotides upstream of PAM). Although the DSB can be repaired through the error-prone non-homologous end joining pathway resulting in loss of function mutations, researchers have been able to trigger an alternative repair pathway, homologous-directed repair (HDR), which relies on the cell's own repair mechanism to add, delete or change the genetic material upon provision of exogenous customised DNA template. Although the targeting specificity of Cas9 is thought to be tightly controlled by the gRNA and its adjacent PAM, potential off-target cleavage activity could still occur [1113]. These could be around the cut site or in homologous regions around the genome which Cas9 could recognise. Furthermore, optimising the composition of donor DNA, which can be circular or linear double-stranded DNA, or single-stranded oligodeoxynucleotide, has shown increased HDR efficiency [14] which can significantly reduce off-target indel generation [1517]. Validation of all major off-target events is crucial for clinical use. In this review, we are going to summarise the main quantification methodologies used to detect CRISPR–Cas9-mediated off-target effects.

Tools for quantifying CRISPR–Cas9

The assays that have been introduced so far for quantifying off-targets in CRISPR–Cas9-edited cells fall into three subcategories: in silico, in vitro and in vivo assays. There are advantages and disadvantages in using any of these and deciding on one heavily relies on the sample material and the research question.

In silico assays

Computational algorithms that detect potential off-target sites are based massively on the sequence of the gRNA. The first web application described in the literature by Hsu et al. [18] can be used as guidance for the selection and validation of specific target sequences as well as off-target analyses (http://www.genome-engineering.org/). They generated a set of sgRNAs targeting multiple sites within two human genomic loci with different trans-activating CRISPR RNA (tracrRNA) 3′ truncations. Based on the criteria produced by their experimental evidence, the group formulated a computational tool for selecting and validating sgRNAs and predicting potential off-target loci. The authors argue in the end that further in vivo investigation into the thermodynamics and stability of sgRNA–DNA duplexes as well as exploration of spCas9 variants and orthologues will improve the prediction power and specificity of their tool for off-targets. E-CRISP [19] is the next chronologically available online software which provides gRNA sequence design tools and evaluates off-target effects using alignment methodologies. Many such tools followed including but not limited to Cas-OFFinder [20], COSMID [21] and Breaking-Cas [22] (Table 1). Initially, there were limitations with each of these tools in terms of number of mismatches allowed for finding off-targets, but since their creation, they have each updated their versions with more relaxed parameters allowing for greater amounts of off-target identification.

Table 1
A comparative summary for the in silico techniques described here
In silica assaysAdvantagesDisadvantagesPublication
Genome-engineering User friendly Restrictive criteria [18
E-CRISP User friendly User-dependent off-target threshold [19
Cas-OFFinder Written in OpenCL for heterogenous programming platforms
Uses a type of sequence pattern clustering for alignment instead of alignment tools 
Restrictive parameters [20
COSMID Provides application-specific primers
Identifies indels between target DNA and gRNA 
Problematic prediction accuracy [21
Breaking-Cas Possibility of using all eukaryotic genomes from ENSEMBL
Placing variable PAM sequences at 5′ or 3′
Setting the gRNA length and the scores per nucleotide 
A heuristic threshold of up to 4 base pair mismatches allowing for user override [22
CRISTA High prediction accuracy across different prediction techniques
Accounts for bulges in prediction process
Discovers patterns that underlie the mechanism of action in the CRISPR–Cas9 system 
As with all machine learning algorithms, CRISTA becomes more powerful with time; the more information is given in its learning dataset, the more accurate it becomes [23
In silica assaysAdvantagesDisadvantagesPublication
Genome-engineering User friendly Restrictive criteria [18
E-CRISP User friendly User-dependent off-target threshold [19
Cas-OFFinder Written in OpenCL for heterogenous programming platforms
Uses a type of sequence pattern clustering for alignment instead of alignment tools 
Restrictive parameters [20
COSMID Provides application-specific primers
Identifies indels between target DNA and gRNA 
Problematic prediction accuracy [21
Breaking-Cas Possibility of using all eukaryotic genomes from ENSEMBL
Placing variable PAM sequences at 5′ or 3′
Setting the gRNA length and the scores per nucleotide 
A heuristic threshold of up to 4 base pair mismatches allowing for user override [22
CRISTA High prediction accuracy across different prediction techniques
Accounts for bulges in prediction process
Discovers patterns that underlie the mechanism of action in the CRISPR–Cas9 system 
As with all machine learning algorithms, CRISTA becomes more powerful with time; the more information is given in its learning dataset, the more accurate it becomes [23

The most exciting, maybe, in silico methodology described in the literature today is an algorithm that goes beyond predicting the off-target cleavage loci [23]. CRISTA (CRISPR Target Assessment) not only predicts cleavage efficacies but also employs machine learning tools providing a learning process in the patterns that underlie the mechanism of action of the CRISPR–Cas9 system. Abadi et al. have described the algorithm as a state-of-the-art approach that considers many different features of the CRISPR–Cas9 system, including RNA thermodynamics, sequence similarity and DNA or RNA bulges that would have a pivotal role on CRISPR–Cas9 efficiency and selectivity [23]. Other such tools, utilising deep learning techniques, have also been described [24,25]. By the use of deep convolutional neural networks and deep feedforward neural networks, computational models were trained and tested on any released off-target datasets in CRISPR–Cas9 gene editing. These methodologies have been shown to outperform the current state-of-the-art prediction methods on two different datasets, CRISPOR and GUIDE-seq [25].

A comparative summary of these in silico techniques is shown in Table 1. However, based on the parameters used, in silico predictions can be very broad and often biased; therefore, a combinatorial approach, employing additional in vitro and/or in vivo techniques, would be needed for evaluation purposes.

In vitro assays

When Cas9 and similar nucleases cut the genome, they create DSBs. This is the principle used by most of the in vitro assays in order to investigate off-target events. They use Cas9 (or other nucleases) to cleave cell-free genomic DNA, sequence the material and then computationally detect DSBs in the sequencing data. By doing so, these assays become quite sensitive in identifying off-targets at a frequency as low as 0.1%. However, the use of cell-free DNA renders them unable to predict off-targets that occur within the cells. A summary of the in vitro techniques can be found in Table 2.

Table 2
A descriptive summary for all in vitro techniques discussed here
In vitro assaysAdvantagesDisadvantagesPCR biasBioinformaticsPublication
TC-seq Can detect chromosomal translocations
Can read through rearrangement breakpoints 
Underrepresentation of GC-rich genomic segments Yes No pipeline [26
DiGenome-seq Relies on DNA cleavage rather than binding. DNA/RNA bulges captured
Detects at a frequency as low as 0.1 
Requires in vivo cleavage verification No Pipeline available [29
DIG-seq Same as Digenome-seq using cell-free chromatin DNA and histone-free genomic DNA Requires in vivo cleavage verification No Pipeline available [32
SITE-seq Easy to follow NGS-based methodology Time-consuming Yes No pipeline [34,35
CIRCLE-seq High efficiency /low background
Detects at a frequency <0.1 
Possibility of false positives Yes Pipeline available [36
In vitro assaysAdvantagesDisadvantagesPCR biasBioinformaticsPublication
TC-seq Can detect chromosomal translocations
Can read through rearrangement breakpoints 
Underrepresentation of GC-rich genomic segments Yes No pipeline [26
DiGenome-seq Relies on DNA cleavage rather than binding. DNA/RNA bulges captured
Detects at a frequency as low as 0.1 
Requires in vivo cleavage verification No Pipeline available [29
DIG-seq Same as Digenome-seq using cell-free chromatin DNA and histone-free genomic DNA Requires in vivo cleavage verification No Pipeline available [32
SITE-seq Easy to follow NGS-based methodology Time-consuming Yes No pipeline [34,35
CIRCLE-seq High efficiency /low background
Detects at a frequency <0.1 
Possibility of false positives Yes Pipeline available [36

The first in vitro assay ever described, for quantifying off-target events, is Translocation-Capture sequencing (TC-seq) [26] that has been shown to study chromosomal rearrangements and translocations by infecting cells with a retrovirus expressing specific I-Scel sites with or without activation-induced cytidine deaminase (AIDCA or AID) protein [27,28]. Genomic DNA from these cells is then isolated and library prepped for sequencing. All AID-dependent chromosomal rearrangements are identified, while AID-independent translocations are discarded. Although it is an efficient protocol for studying chromosomal translocations within any given model or environment, PCR amplification errors and PCR biases in GC-rich templates could still occur.

DiGenome-seq was first introduced in 2015 [29] where off-target mutations are identified in cells with Cas9-digested genomes (digenomes). Digenome-seq relies on DNA cleavage rather than binding and it is performed at the genomic level where DNA/RNA bulges are captured and can detect at once off-targets at a frequency as low as 0.01% in up to 10 gRNAs [30]. Its main advantage is that the DSBs introduced by Cas9 will not be processed by the DNA repair machinery, as opposed to Breaks Labelling, Enrichment on Streptavidin and next-generation Sequencing (BLESS) and GUIDE-seq (see in vivo assays), increasing the possibility of detecting off-targets. Nonetheless, this technique could also lead to a lot of false positives due to not being able to map properly sequence reads that are around naturally occurring indel sites. In vivo cleavage confirmation would still be required, as would a high skilled bioinformatic analysis [31]. Recently, the same research team that created DiGenome-seq investigated the possibility of CRISPR–Cas9 on- and off- targets being affected by chromatin in eukaryotic cells [32] with an optimised version of DiGenome-seq. Cas9 fusion with chromatin is found to improve its activity by up to several folds [33]. Their found that chromatin affects genome-wide CRISPR specificity and they hence developed a new tool, DIG-seq, that uses chromatin DNA rather than histone-free DNA in vitro.

Another biochemical method to identify off-target harbouring sites is SITE-seq [34,35]. Using this, extracted and purified genomic DNA is cleaved with Cas9 and Cas9 cleavage sites are biochemically probed and enriched for next-generation sequencing (NGS). Bioinformatics are then used to identify off-target cleavage sites by selecting for targets with the highest possible activity and specificity. The authors of SITE-seq state that the signature is similar to the one observed with Digenome-seq [34].

The latest addition to the in vitro screens for genome-wide CRISPR–Cas9 nuclease off-targets is CIRCLE-seq [36]. In contrast with previously published in vitro methods, CIRCLE-seq can be performed using widely accessible NGS technology and requires no reference genome sequence. The protocol involves shearing and circularisation of the purified genomic DNA and degradation of any residual linear DNA. Cas9 nuclease is then used to linearise the circular DNA containing a Cas9 cleavage site and the cleaved ends are amplified and sequenced to detect off-targets [37]. By enriching for Cas9 nuclease-cleaved genomic DNA before sequencing, CIRCLE-seq becomes more sensitive in locating off-target events. Moreover, it removes the need for larger sample sizes and read depths that introduce background noise rendering identification of low-frequency cleavage events even harder as in the case of Digenome-seq [36]. However, careful bioinformatic analysis needs to be carried over due to the amplification bias during Cas9 nuclease-cleaved genomic DNA enrichment.

In vivo assays

In the last 5 years, there has been a flurry of development of different in vivo assays with the earliest and most cited one being ChIP-seq [13,3840]. It uses chromatin immunoprecipitation coupled with high-throughput sequencing to detect Cas9 binding sites and chemical modifications of histone proteins genome-wide. The limiting requirements of ChIP-seq include: (1) high-complexity libraries (80% of 10 million or more reads to be mapped to distinct genomic locations), (2) replicates of two per experiment should be carried out, where either 80 or 40% of the identified targets in replicate one should be among the targets of the replicate two and (3) large numbers of cells (∼10 million) [41]. However, there have been advances in the technology since ChIP-seq was first developed. For instance, the Encyclopaedia of DNA Elements (ENCODE) Consortium suggests limiting the genomic regions to be investigated to a few candidate regions, if possible, and then validating them using biological experiments for ChIP-seq to be able to efficiently detect targets in low-complexity libraries [42]. Since ChIP-seq was first used for identifying off-target events, other in vivo assays have been established (Table 3).

Table 3
In vivo off-target detection assays: a summary table
In vivo assaysAdvantagesDisadvantagesBioinformaticsPublication
CHIP-seq Identifies binding site locations for individual proteins and histones High number of cells required
Low-resolution analysis 
Pipeline available [13,3840
BLESS Detects DSBs at nucleotide resolution
Independent to proteins binding to DSBs ssDNA-independent 
High background
Cell fixation-dependent 
Pipeline available [4344
GUIDE-seq Generates global specificity landscapes for RGNs in living human cells
Targeted sequencing reduces costs 
Low sensitivity Pipeline available [45
HTGTS Higher efficiency than WGS Underestimates the frequency of DSBs
Limited chromatin accessibility 
Pipeline available [46
LAM-HTGTS Sensitive detection of DSBs within chromosomes
Limited PCR bias. Detects translocations 
Bait-prey DSBs-dependent
Not available to limited amounts of material 
Pipeline available [47
IDLVs Detects efficiently single nucleotide skipping from sgRNA or its genomic target Low sensitivity Pipeline available [4849
In vivo assaysAdvantagesDisadvantagesBioinformaticsPublication
CHIP-seq Identifies binding site locations for individual proteins and histones High number of cells required
Low-resolution analysis 
Pipeline available [13,3840
BLESS Detects DSBs at nucleotide resolution
Independent to proteins binding to DSBs ssDNA-independent 
High background
Cell fixation-dependent 
Pipeline available [4344
GUIDE-seq Generates global specificity landscapes for RGNs in living human cells
Targeted sequencing reduces costs 
Low sensitivity Pipeline available [45
HTGTS Higher efficiency than WGS Underestimates the frequency of DSBs
Limited chromatin accessibility 
Pipeline available [46
LAM-HTGTS Sensitive detection of DSBs within chromosomes
Limited PCR bias. Detects translocations 
Bait-prey DSBs-dependent
Not available to limited amounts of material 
Pipeline available [47
IDLVs Detects efficiently single nucleotide skipping from sgRNA or its genomic target Low sensitivity Pipeline available [4849

BLESS is a genome-wide in vivo approach to map DSBs at nucleotide resolution by direct in situ Breaks Labelling, Enrichment on Streptavidin and NGS [43,44]. The advantages of this methodology are that it can detect DSBs at a nucleotide resolution and does not depend on proteins that bind to DSBs nor single-stranded DNA, minimising bias (Table 3). The disadvantages are that the data have a high background, they only map unjoined ends and they are susceptible to artefacts associated with cell fixation [31,45].

The same group that developed the in vitro assay CIRCLE-seq [36] had originally introduced an in vivo assay, GUIDE-seq, which uses the integration of blunt-ended double-stranded oligodeoxynucleotides (DSOs) into DSBs, genome-wide [45]. The DSO integration sites are then mapped to the exact positions in the genome at nucleotide level using amplification steps and sequencing [45]. Using this method, specific landscapes for the RNA-guided nucleases can be generated globally in living human cells. Moreover, targeted sequencing decreases the costs of sequencing tremendously when compared with other known protocols for the investigation of off-targets. However, as shown in the CIRCLE-seq comparison experiments [36], certain sites are undetectable by GUIDE-seq due to lower read counts. This detection limit could compromise validation of off-targets sites in cells analysed by targeted sequencing, as the lower limit of detection by NGS remains at 0.1% (Table 3). Instead, high-depth targeted amplicon sequencing using genomic DNA from cell-based GUIDE-seq experiments could address this concern [36].

As early as 2011, a research group developed an in vivo assay to study genome-wide translocations that occur in haematopoietic malignancies [46]. The assay involved high-throughput, genome-wide translocation sequencing (HTGTS) in mammalian cells, particularly identifying translocation events induced by AID-dependent IgH class switching [46] and CRISPR–Cas9 [30]. It is proved to have a higher efficiency when compared with whole-genome sequencing (WGS) but tends to underestimate the frequency of DSBs [31] and is limited by chromatin accessibility [47]. Moving forward, the research team that first performed HTGTS using Cas9:sgRNA (Cas9:single guide RNA) successfully, introduced a modified version of HTGTS that included a linear amplification-mediated PCR step (LAM-PCR) upgrading the technique [47]. The now called LAM-HTGTS uses ‘prey’ DSBs that bind to ‘bait’ DSBs (Table 3), and their junctions from the isolated genomic DNA are detected in a robust and unbiased manner [31]. It is the only assay, thus far, that can efficiently detect all recurrent DSBs that occur during a period of time in a population of cells [31]. Although this is an exceptionally sensitive method for detecting large genomic rearrangements, it relies on the presence of both a ‘prey’ and a known ‘bait’ DSB, excluding its use on previously isolated genomic DNA without priori knowledge of a recurrent DSB that can serve as a ‘bait’. Moreover, LAM-HTGTS only produces information about the ‘prey’ DSB that binds to a ‘bait’ DSB and misses any data on the ‘prey’ DSB that could persist as a DSB. Finally, it may have a constrained outcome if input material is limited.

The latest development in the in vivo predictions for off-target sequences is the use of integrase-defective lentiviral vectors (IDLVs) which was first introduced for the detection of off-target cleavage sites of zinc finger nucleases [48] but was further optimised for the CRISPR–Cas9 system too [49]. IDLVs have been shown to detect off-targets to a frequency as low as 1%, which is relatively insensitive in comparison with other techniques (DiGenome-seq, CIRCLE-seq), but is also able to recognise a single nucleotide skipping mutation from sgRNA or its genomic target efficiently [49] (Table 3).

Summary

  • No tool can be used in isolation for identifying off-targets in CRISPR–Cas9 and hence a combinatorial approach of two or more techniques is advised based on the research hypothesis of each laboratory.

  • Technical limitations will always arise as a result of small-sized amplicons that NGS could overcome by maintaining high coverage and large read depth to allow for the detection of very rare large indels, an apparent issue in many of the assays. Therefore, choosing the most appropriate sequencing platform and careful analysis of the sequencing data is essential.

  • Special care must be taken when discarding false-positive sequence reads that result from amplification artefacts during PCR. In such cases, including a negative control (with no nuclease expression) at each target site could prove to be beneficial [50].

  • Optimising the delivery efficiency could also further reduce off-targets. Studies have shown that a shorter Cas9 expression duration [51], the use of Cas9 nickases [52], Cas9-Fokl chimeric proteins [53], proper modifications to residues of the Cas9 protein [54,55] could greatly increase the specificity for genome editing.

  • Most importantly, there have been reports on significant, unexpected on-target mutations, such as large deletions and complex mutations, in mouse embryonic stem cells, mouse haematopoietic progenitors and a human differentiated cell line [56]. This on-target mutagenesis could potentially lead to pathogenicity if used in a clinical setting. Thorough examination of the genome and a comprehensive genomic analysis is pivotal for the identification of normal genomes prior to patient treatment.

Abbreviations

     
  • AID

    activation-induced cytidine deaminase

  •  
  • BLESS

    breaks labelling, enrichment on streptavidin and next-generation sequencing

  •  
  • Cas9

    CRISPR-associated protein 9

  •  
  • CRISPR

    clustered regularly interspaced short palindromic repeats

  •  
  • CRISTA

    CRISPR Target Assessment

  •  
  • DSB

    double-stranded break

  •  
  • DSOs

    double-stranded oligodeoxynucleotides

  •  
  • HDR

    homologous-directed repair

  •  
  • HTGTS

    high-throughput, genome-wide translocation sequencing

  •  
  • IDLVs

    integrase-defective lentiviral vectors

  •  
  • NGS

    next-generation sequencing

  •  
  • PAMs

    protospacer adjacent motifs

  •  
  • sgRNA

    single guide RNA

  •  
  • TC-seq

    translocation-capture sequencing

  •  
  • WGS

    whole-genome sequencing

Competing Interests

The Author declares that there are no competing interests associated with this manuscript.

References

References
1
David
,
R.M.
and
Doherty
,
A.T.
(
2017
)
Viral vectors: the road to reducing genotoxicity
.
Toxicol. Sci.
155
,
315
325
2
Wang
,
M.
,
Zuris
,
J.A.
,
Meng
,
F.
,
Rees
,
H.
,
Sun
,
S.
,
Deng
,
P.
et al (
2016
)
Efficient delivery of genome-editing proteins using bioreducible lipid nanoparticles
.
Proc. Natl Acad. Sci. U.S.A.
113
,
2868
2873
3
Yin
,
H.
,
Song
,
C.-Q.
,
Dorkin
,
J.R.
,
Zhu
,
L.J.
,
Li
,
Y.
,
Wu
,
Q.
et al (
2016
)
Therapeutic genome editing by combined viral and non-viral delivery of CRISPR system components in vivo
.
Nat. Biotechnol.
34
,
328
333
4
Zuris
,
J.A.
,
Thompson
,
D.B.
,
Shu
,
Y.
,
Guilinger
,
J.P.
,
Bessen
,
J.L.
,
Hu
,
J.H.
et al (
2015
)
Cationic lipid-mediated delivery of proteins enables efficient protein-based genome editing in vitro and in vivo
.
Nat. Biotechnol.
33
,
73
80
5
Jore
,
M.M.
,
Lundgren
,
M.
,
van Duijn
,
E.
,
Bultema
,
J.B.
,
Westra
,
E.R.
,
Waghmare
,
S.P.
et al (
2011
)
Structural basis for CRISPR RNA-guided DNA recognition by Cascade
.
Nat. Struct. Mol. Biol.
18
,
529
536
6
Wiedenheft
,
B.
,
Zhou
,
K.
,
Jinek
,
M.
,
Coyle
,
S.M.
,
Ma
,
W.
and
Doudna
,
J.A.
(
2009
)
Structural basis for DNase activity of a conserved protein implicated in CRISPR-mediated genome defense
.
Structure
17
,
904
912
7
Cho
,
S.W.
,
Kim
,
S.
,
Kim
,
J.M.
and
Kim
,
J.-S.
(
2013
)
Targeted genome engineering in human cells with the Cas9 RNA-guided endonuclease
.
Nat. Biotechnol.
31
,
230
232
8
Cong
,
L.
,
Ran
,
F.A.
,
Cox
,
D.
,
Lin
,
S.
,
Barretto
,
R.
,
Habib
,
N.
et al (
2013
)
Multiplex genome engineering using CRISPR/Cas systems
.
Science
339
,
819
823
9
Jinek
,
M.
,
East
,
A.
,
Cheng
,
A.
,
Lin
,
S.
,
Ma
,
E.
and
Doudna
,
J.
(
2013
)
RNA-programmed genome editing in human cells
.
eLife
2
,
e00471
10
Mali
,
P.
,
Yang
,
L.
,
Esvelt
,
K.M.
,
Aach
,
J.
,
Guell
,
M.
,
DiCarlo
,
J.E.
et al (
2013
)
RNA-guided human genome engineering via Cas9
.
Science
339
,
823
826
11
Cradick
,
T.J.
,
Fine
,
E.J.
,
Antico
,
C.J.
and
Bao
,
G.
(
2013
)
CRISPR/Cas9 systems targeting beta-globin and CCR5 genes have substantial off-target activity
.
Nucleic Acids Res.
41
,
9584
9592
12
Fu
,
Y.F.
,
Foden
,
J.A.
,
Khayter
,
C.
,
Maeder
,
M.L.
,
Reyon
,
D.
,
Joung
,
J.K.
et al (
2013
)
High-frequency off-target mutagenesis induced by CRISPR-Cas nucleases in human cells
.
Nat. Biotechnol.
31
,
822
13
Kuscu
,
C.
,
Arslan
,
S.
,
Singh
,
R.
,
Thorpe
,
J.
and
Adli
,
M.
(
2014
)
Genome-wide analysis reveals characteristics of off-target sites bound by the Cas9 endonuclease
.
Nat. Biotechnol.
32
,
677
683
14
Song
,
F.
and
Stieger
,
K.
(
2017
)
Optimizing the DNA donor template for homology-directed repair of double-strand breaks
.
Mol. Ther.-Nucleic Acids
7
,
53
60
15
Chen
,
X.
,
Janssen
,
J.M.
,
Liu
,
J.
,
Maggio
,
I.
,
‘t Jong
,
A.E.J.
,
Mikkers
,
H.M.M.
et al (
2017
)
In trans paired nicking triggers seamless genome editing without double-stranded DNA cutting
.
Nat. Commun.
8
,
657
16
Nakajima
,
K.
,
Zhou
,
Y.
,
Tomita
,
A.
,
Hirade
,
Y.
,
Gurumurthy
,
C.B.
and
Nakada
,
S.
(
2018
)
Precise and efficient nucleotide substitution near genomic nick via noncanonical homology-directed repair
.
Genome. Res.
28
,
223
230
17
Richardson
,
C.D.
,
Ray
,
G.J.
,
DeWitt
,
M.A.
,
Curie
,
G.L.
and
Corn
,
J.E.
(
2016
)
Enhancing homology-directed genome editing by catalytically active and inactive CRISPR-Cas9 using asymmetric donor DNA
.
Nat. Biotechnol.
34
,
339
344
18
Hsu
,
P.D.
,
Scott
,
D.A.
,
Weinstein
,
J.A.
,
Ran
,
F.A.
,
Konermann
,
S.
,
Agarwala
,
V.
et al (
2013
)
DNA targeting specificity of RNA-guided Cas9 nucleases
.
Nat. Biotechnol.
31
,
827
832
19
Heigwer
,
F.
,
Kerr
,
G.
and
Boutros
,
M.
(
2014
)
E-CRISP: fast CRISPR target site identification
.
Nat. Methods
11
,
122
123
20
Bae
,
S.
,
Park
,
J.
and
Kim
,
J.S.
(
2014
)
Cas-OFFinder: a fast and versatile algorithm that searches for potential off-target sites of Cas9 RNA-guided endonucleases
.
Bioinformatics
30
,
1473
1475
21
Cradick
,
T.J.
,
Qiu
,
P.
,
Lee
,
C.M.
,
Fine
,
E.J.
and
Bao
,
G.
(
2014
)
COSMID: a web-based tool for identifying and validating CRISPR/Cas off-target sites
.
Mol. Ther.-Nucleic Acids
3
.
22
Oliveros
,
J.C.
,
Franch
,
M.
,
Tabas-Madrid
,
D.
,
San-León
,
D.
,
Montoliu
,
L.
,
Cubas
,
P.
et al (
2016
)
Breaking-Cas-interactive design of guide RNAs for CRISPR-Cas experiments for ENSEMBL genomes
.
Nucleic Acids Res.
44
,
W267
W271
23
Abadi
,
S.
,
Yan
,
W.X.
,
Amar
,
D.
,
Mayrose
,
I.
and
Xu
,
H.
(
2017
)
A machine learning approach for predicting CRISPR-Cas9 cleavage efficiencies and patterns underlying its mechanism of action
.
PLoS Comput. Biol.
13
,
e1005807
24
Chuai
,
G.H.
,
Ma
,
H.
,
Yan
,
J.
,
Chen
,
M.
,
Hong
,
N.
,
Xue
,
D.
et al (
2018
)
DeepCRISPR: optimized CRISPR guide RNA design by deep learning
.
Genome Biol.
19
,
80
25
Lin
,
J.
and
Wong
,
K.C.
(
2018
)
Off-target predictions in CRISPR-Cas9 gene editing using deep learning
.
Bioinformatics
34
,
656
663
26
Klein
,
I.A.
,
Resch
,
W.
,
Jankovic
,
M.
,
Oliveira
,
T.
,
Yamane
,
A.
,
Nakahashi
,
H.
et al (
2011
)
Translocation-capture sequencing reveals the extent and nature of chromosomal rearrangements in B lymphocytes
.
Cell
147
,
95
106
27
Jankovic
,
M.
,
Feldhahn
,
N.
,
Oliveira
,
T.Y.
,
Silva
,
I.T.
,
Kieffer-Kwon
,
K.-R.
,
Yamane
,
A.
et al (
2013
)
53BP1 alters the landscape of DNA rearrangements and suppresses AID-induced B cell lymphoma
.
Mol. Cell
49
,
623
631
28
Pefanis
,
E.
,
Wang
,
J.
,
Rothschild
,
G.
,
Lim
,
J.
,
Chao
,
J.
,
Rabadan
,
R.
et al (
2014
)
Noncoding RNA transcription targets AID to divergently transcribed loci in B cells
.
Nature
514
,
389
393
29
Kim
,
D.
,
Bae
,
S.
,
Park
,
J.
,
Kim
,
E.
,
Kim
,
S.
,
Yu
,
H.R.
et al (
2015
)
Digenome-seq: genome-wide profiling of CRISPR-Cas9 off-target effects in human cells
.
Nat. Methods
12
,
237
243
30
Mei
,
Y.
,
Wang
,
Y.
,
Chen
,
H.
,
Sun
,
Z.S.
,
Ju
,
X.-D.
et al (
2016
)
Recent progress in CRISPR/Cas9 technology
.
J. Genet. Genomics
43
,
63
75
31
Hu
,
J.Z.
,
Meyers
,
R.M.
,
Dong
,
J.
,
Panchakshari
,
R.A.
,
Alt
,
F.W.
and
Frock
,
R.L.
(
2016
)
Detecting DNA double-stranded breaks in mammalian genomes by linear amplification-mediated high-throughput genome-wide translocation sequencing
.
Nat. Protoc.
11
,
853
871
32
Kim
,
D.
and
Kim
,
J.S.
(
2018
)
DIG-seq: a genome-wide CRISPR off-target profiling method using chromatin DNA
.
Genome Res.
28
,
1894
1900
33
Ding
,
X.
,
Seebeck
,
T.
,
Feng
,
Y.
,
Jiang
,
Y.
,
Davis
,
G.D.
and
Chen
,
F.
(
2019
)
Improving CRISPR-Cas9 genome editing efficiency by fusion with chromatin-modulating peptides
.
CRISPR J.
2
,
51
63
34
Cameron
,
P.
,
Fuller
,
C.K.
,
Donohoue
,
P.D.
,
Jones
,
B.N.
,
Thompson
,
M.S.
,
Carter
,
M.M.
et al (
2017
)
Mapping the genomic landscape of CRISPR-Cas9 cleavage
.
Nat. Methods
14
,
600
606
35
Cameron
,
P.S.
,
Settle
,
A.H.
,
Fuller
,
C.K.
,
Thompson
,
M.S.
,
Cigan
,
A.M.
,
Young
,
J.K.
et al (
2017
)
SITE-Seq: A Genome-Wide Method to Measure Cas9 Cleavage
,
Nature Publishing Group
36
Tsai
,
S.Q.
,
Nguyen
,
N.T.
,
Malagon-Lopez
,
J.
,
Topkar
,
V.V.
,
Aryee
,
M.J.
and
Joung
,
J.K.
(
2018
)
Corrigendum: CIRCLE-seq: a highly sensitive in vitro screen for genome-wide CRISPR-Cas9 nuclease off-targets
.
Nat. Methods
15
,
394
394
37
Lazzarotto
,
C.R.
,
Nguyen
,
N.T.
,
Tang
,
X.
,
Malagon-Lopez
,
J.
,
Guo
,
J.A.
,
Aryee
,
M.J.
et al (
2018
)
Defining CRISPR-Cas9 genome-wide nuclease activities with CIRCLE-seq
.
Nat. Protoc.
13
,
2615
2642
38
Barski
,
A.
,
Cuddapah
,
S.
,
Cui
,
K.
,
Roh
,
T.-Y.
,
Schones
,
D.E.
,
Wang
,
Z.
et al (
2007
)
High-resolution profiling of histone methylations in the human genome
.
Cell
129
,
823
837
39
Johnson
,
D.S.
,
Mortazavi
,
A.
,
Myers
,
R.M.
and
Wold
,
B.
(
2007
)
Genome-wide mapping of in vivo protein-DNA interactions
.
Science
316
,
1497
1502
40
Mikkelsen
,
T.S.
,
Ku
,
M.
,
Jaffe
,
D.B.
,
Issac
,
B.
,
Lieberman
,
E.
,
Giannoukos
,
G.
et al (
2007
)
Genome-wide maps of chromatin state in pluripotent and lineage-committed cells
.
Nature
448
,
553
560
41
Furey
,
T.S.
(
2012
)
ChIP-seq and beyond: new and improved methodologies to detect and characterize protein-DNA interactions
.
Nat. Rev. Genet.
13
,
840
852
42
Landt
,
S.G.
,
Marinov
,
G.K.
,
Kundaje
,
A.
,
Kheradpour
,
P.
,
Pauli
,
F.
,
Batzoglou
,
S.
et al (
2012
)
ChIP-seq guidelines and practices of the ENCODE and modENCODE consortia
.
Genome Res.
22
,
1813
1831
43
Crosetto
,
N.
,
Mitra
,
A.
,
Silva
,
M.J.
,
Bienko
,
M.
,
Dojer
,
N.
,
Wang
,
Q.
et al (
2013
)
Nucleotide-resolution DNA double-strand break mapping by next-generation sequencing
.
Nat. Methods
10
,
361
365
44
Ran
,
F.A.
,
Cong
,
L.
,
Yan
,
W.X.
,
Scott
,
D.A.
,
Gootenberg
,
J.S.
,
Kriz
,
A.J.
et al (
2015
)
In vivo genome editing using Staphylococcus aureus Cas9
.
Nature
520
,
186
191
45
Tsai
,
S.Q.
,
Zheng
,
Z.
,
Nguyen
,
N.T.
,
Liebers
,
M.
,
Topkar
,
V.V.
,
Thapar
,
V.
et al (
2015
)
GUIDE-seq enables genome-wide profiling of off-target cleavage by CRISPR-Cas nucleases
.
Nat. Biotechnol.
33
,
187
197
46
Chiarle
,
R.
,
Zhang
,
Y.
,
Frock
,
R.L.
,
Lewis
,
S.M.
,
Molinie
,
B.
,
Ho
,
Y.-J.
et al (
2011
)
Genome-wide translocation sequencing reveals mechanisms of chromosome breaks and rearrangements in B cells
.
Cell
147
,
1640
1640
47
Frock
,
R.L.
,
Hu
,
J.
,
Meyers
,
R.M.
,
Ho
,
Y.-J.
,
Kii
,
E.
and
Alt
,
F.W.
(
2015
)
Genome-wide detection of DNA double-stranded breaks induced by engineered nucleases
.
Nat. Biotechnol.
33
,
179
186
48
Gabriel
,
R.
,
Lombardo
,
A.
,
Arens
,
A.
,
Miller
,
J.C.
,
Genovese
,
P.
,
Kaeppel
,
C.
et al (
2011
)
An unbiased genome-wide analysis of zinc-finger nuclease specificity
.
Nat. Biotechnol.
29
,
816
823
49
Wang
,
X.L.
,
Wang
,
Y.
,
Wu
,
X.
,
Wang
,
J.
,
Wang
,
Y.
,
Qiu
,
Z.
et al (
2015
)
Unbiased detection of off-target cleavage by CRISPR-Cas9 and TALENs using integrase-defective lentiviral vectors
.
Nat. Biotechnol.
33
,
175
178
50
Cho
,
S.W.
,
Kim
,
S.
,
Kim
,
Y.
,
Kweon
,
J.
,
Kim
,
H.S.
,
Bae
,
S.
et al (
2014
)
Analysis of off-target effects of CRISPR/Cas-derived RNA-guided endonucleases and nickases
.
Genome Res.
24
,
132
141
51
Chen
,
Y.
,
Liu
,
X.
,
Zhang
,
Y.
,
Wang
,
H.
,
Ying
,
H.
,
Liu
,
M.
et al (
2016
)
A self-restricted CRISPR system to reduce off-target effects
.
Mol. Ther.
24
,
2132
2132
52
Ran
,
F.A.
,
Hsu
,
P.D.
,
Lin
,
C.-Y.
,
Gootenberg
,
J.S.
,
Konermann
,
S.
,
Trevino
,
A.E.
et al (
2013
)
Double nicking by RNA-guided CRISPR Cas9 for enhanced genome editing specificity
.
Cell
155
,
479
480
53
Tsai
,
S.Q.
,
Wyvekens
,
N.
,
Khayter
,
C.
,
Foden
,
J.A.
,
Thapar
,
V.
,
Reyon
,
D.
et al (
2014
)
Dimeric CRISPR/Cas-based nucleases for high-precision genome editing in human cells
.
Human Gene Therapy
25
,
A120
A121
54
Kleinstiver
,
B.P.
,
Pattanayak
,
V.
,
Prew
,
M.S.
,
Tsai
,
S.Q.
,
Nguyen
,
N.T.
and
Joung
,
J.K.
(
2016
)
High-fidelity CRISPR-Cas9 nucleases with No detectable genome-wide off-target effects
.
Mol. Ther.
24
,
S288
S288
55
Slaymaker
,
I.M.
,
Gao
,
L.
,
Zetsche
,
B.
,
Scott
,
D.A.
,
Yan
,
W.X.
and
Zhang
,
F.
(
2016
)
Rationally engineered Cas9 nucleases with improved specificity
.
Science
351
,
84
88
56
Kosicki
,
M.
,
Tomberg
,
K.
and
Bradley
,
A.
(
2018
)
Repair of double-strand breaks induced by CRISPR–Cas9 leads to large deletions and complex rearrangements
.
Nat. Biotechnol.
36
,
899
899