Reflective blog assignment

VIEWPOINTS

Collateral damage and CRISPR genome

editing

Mark ThomasID 1*, Gaetan BurgioID

2 , David J. AdamsID

1 , Vivek IyerID

1*

1 Wellcome Sanger Institute, Wellcome Genome Campus, Hinxton, Cambridge, United Kingdom,

2 Department of Immunology and Infectious Disease, The John Curtin School of Medical Research, The

Australian National University, Canberra, Australia

* [email protected] (MT); [email protected] (VI)

Abstract

The simplicity and the versatility of clustered regularly interspaced short palindromic

repeats/CRISPR-associated protein (CRISPR-Cas) systems have enabled the genetic

modification of virtually every organism and offer immense therapeutic potential for the treat-

ment of human disease. Although these systems may function efficiently within eukaryotic

cells, there remain concerns about the accuracy of Cas endonuclease effectors and their

use for precise gene editing. Recently, two independent reports investigating the editing

accuracy of the CRISPR-Cas9 system were published by separate groups at the Wellcome

Sanger Institute; our study—Iyer and colleagues [1]—defined the landscape of off-target

mutations, whereas the other by Kosicki and colleagues [2] detailed the existence of on-tar-

get, potentially deleterious deletions. Although both studies found evidence of large on-tar-

get CRISPR-induced deletions, they reached seemingly very different conclusions.

So, what do scientists using CRISPR gene-editing technology need

to know?

Off targets—the need for controls

Iyer and colleagues used whole-genome sequencing (WGS) to identify potential off-target

damage in mouse embryos and assessed the impact of genomic variation on de novo mutation

calls. This study was undertaken in response to Schaefer and colleagues [3]—subsequently

retracted—which reported a greater number of mutations at unexpected off-target locations in

mice following the delivery of CRISPR reagents in mouse zygotes. Iyer and colleagues empha-

sised proper experimental controls and use of best practice in the choice of a single-guide

RNA (sgRNA) and Cas9-mediated mutagenesis conditions. We demonstrated that if Cas9-me-

diated mutagenesis were causing off-target mutations, then the rate of these mutations was not

distinguishable from the background de novo mutation rate in mouse embryos. The study

concluded that efficient CRISPR gene editing was possible without a significant increase in de

novo mutation rates, supporting the development of CRISPR-Cas9 as a therapeutic tool.

PLOS Genetics | https://doi.org/10.1371/journal.pgen.1007994 March 14, 2019 1 / 8

a1111111111

a1111111111

a1111111111

a1111111111

a1111111111

OPEN ACCESS

Citation: Thomas M, Burgio G, Adams DJ, Iyer V

(2019) Collateral damage and CRISPR genome

editing. PLoS Genet 15(3): e1007994. https://doi.

org/10.1371/journal.pgen.1007994

Editor: Lin He, University of California Berkeley,

UNITED STATES

Published: March 14, 2019

Copyright: © 2019 Thomas et al. This is an open access article distributed under the terms of the

Creative Commons Attribution License, which

permits unrestricted use, distribution, and

reproduction in any medium, provided the original

author and source are credited.

Funding: This work was undertaken at the

Wellcome Sanger Institute, supported by core

funding from the Wellcome Trust. Dr Burgio is

supported by the National Collaborative Research

Infrastructure (NCRIS) via the Australian

Phenomics Network (APN). The funders had no

role in the preparation of the article.

Competing interests: The authors have declared

that no competing interests exist.

On target—damage closer to home

Weeks later, Kosicki and colleagues reported unexpected on-target alterations at target loca-

tions in mouse embryonic stem cells, hematopoietic progenitors, and a differentiated human

cell line. Although the vast majority of the CRISPR-induced double strand breaks (DSBs)

resulted in small indels (<50 nt), up to 20% of editing events resulted in significantly larger

deletions (>250 nt) and more complex genomic rearrangements than previously reported [4].

Using Pacific Biosciences (California) long-read sequencing and long-range PCR (over 5 kb),

some of these events were shown to extend up to several kilobases from the protospacer adja-

cent motif (PAM) at the target site. The authors posited that these events would likely be

missed using standard genotyping methods. Because this observation has significant implica-

tions for both research and therapeutic applications, the authors correctly concluded that com-

prehensive genomic analyses are warranted to fully characterise CRISPR-targeted cells.

Are these studies directly comparable?

Not really. They focus on fundamentally different aspects of CRISPR-Cas9 genome editing,

specifically on-target versus off-target damage, within different biological contexts. Iyer and

colleagues characterized on-target and putative off-target alleles in mouse embryos, whereas

Kosicki and colleagues only characterized on-target mutations in pooled cell assays.

On-target mutation rates are comparable

With only 10 embryos, Iyer and colleagues lacks the statistical power to conclude anything

more about large on-target deletions, other than that they occur with a frequency of at least

10%. Specifically, all 10 of the CRISPR-edited zygotes examined by Iyer and colleagues had an

average of 2 mutant on-target alleles per embryo; 21 mutant alleles in total. Of these 21 alleles,

20 were “small” (<50 nt) deletions, detected with the bcftools small-variant caller. In one

embryo, a large 338 nt deletion was also detected using Pindel, a structural variant caller. By

contrast, Kosicki and colleagues only analyzed on-target effects. At the target PigA locus, the inferred overall proportion of “large” (>250 nt) alleles observed in the pool of cells from

Kosicki and colleagues was approximately 20% across all sgRNAs, in which each distinct dele-

tion is represented in a small number of cells.

No data to compare for off-target mutation rates

Even though Kosicki and colleagues did not study off-target events, it could be inferred that

large deletions or rearrangements at potential off-target locations may be missed by the WGS

approaches used to analyze these events in Iyer and colleagues. The Pindel structural variant

caller used to analyze all WGS reads from treated embryos in Iyer and colleagues identified

one large de novo candidate off-target deletion of 260 nt. There was, however, no coincidence

between this large deletion and any potential off-target site, and the deletion was therefore dis-

counted as a potential off-target effect; we stand by this conclusion. Because the Cas9 ribonu-

cleoprotein (RNPs) complexes are unlikely to persist beyond the two-cell stage, due to its short

half-life, any CRISPR-induced mutations (including any large deletions) should be well repre-

sented in the mouse zygote. As such, the frequency of these mutations should have been clearly

detected in the filtered variant calls from Iyer and colleagues, although none were observed.

Zygote microinjection and pool transfection are different experiments

It is important to note that there are several technical and biological reasons why editing

outcomes might be different between these two experimental approaches. The most notable

PLOS Genetics | https://doi.org/10.1371/journal.pgen.1007994 March 14, 2019 2 / 8

difference is the synchronized and nonsynchronized cell cycle statuses of the targeted cells.

Cytoplasmic microinjection was used by Iyer and colleagues to mutate fertilized single-cell

mouse zygotes (confirmed by the presence of 2-pronuclei) with Cas9 RNP complexes.

Although the exact cell cycle stage will have varied between each zygote, they are synchro-

nized, because all microinjections were completed prior to coalescence of the pronuclei and

therefore the completion of S-phase [5]. Any CRISPR-induced DSBs that occur at these

early stages of the cell cycle are likely to be repaired by nonhomologous end joining

(NHEJ), which is consistent with the formation of the small indels that were predominantly

observed. By contrast, PiggyBac transposon-mediated delivery was predominantly used by Kosicki and colleagues to introduce constitutively active CRISPR reagents, although lipofec-

tion and electroporation methods were also used to deliver transient Cas9 RNPs into pools

of mitotically active cells. Unlike Iyer and colleagues, however, the cells within these pools

were not synchronized and DNA damage was likely to have occurred at different stages of

the cell cycle, potentially altering DNA-repair outcomes. Indeed, cell cycle synchronization

has previously been shown to improve RNP-mediated homology-directed repair (HDR)

rates, although the effect on nontemplate DSB repair outcomes is less clear [6]. The impor-

tance of considering cell cycle changes was further demonstrated by Gu and colleagues, who

exploited an extended G2-S phase at the two-cell stage in mouse zygotes to significantly

increase the knock-in efficiency of large DNA fragments [7]. Furthermore, fundamental dif-

ferences between the nuclear organization of single-cell embryos and mitotically active cells

have also been noted by a recent study [8], which observed that there is a physical separation

of the maternal and paternal genomes in early stage mouse zygote development. Both

genomes exist on separate spindles, which are only aligned at the start of anaphase, prior to

cleavage at the two-cell stage. The persistence of distinct maternal and paternal genomes

until at least the two-cell stage may impair the formation of more complex rearrangements,

including the apparent interhomologue repair (IHR) observed by Kosicki and colleagues at

the Cd9 locus in F1 C57BL/6 × CAST/Ei embryonic stem (ES) cells. Although the nuclear organization within single-cell zygotes may not favor IHR events, this does not entirely

exclude the possibility of them occurring, as demonstrated by a recent study that used

RAD51-enhanced IHR to increase the efficiency of homozygous knock-in insertions in

mouse zygotes [9].

We have summarized the differences in on- and off-target data, along with experimental

methodologies, in Table 1.

Table 1. Comparison of experimental methodologies and results.

Criteria Iyer and colleagues Kosicki and colleagues

Transfection method Cytoplasmic microinjection Piggybac, lipofection, electroporation

Cells mutated Single-cell mouse zygotes, 2 pronuclei. Mixed-phase pools of mouse ES cells or human RPE1

cells.

Cas9 + sgRNA

delivery

Cas9 RNP Cas9 RNP

Observed large

on-target deletions?

Yes, at least 10% of zygotes (limited data). Yes, at 20% of the cell population.

Observed large

off-target deletions?

One large de novo deletion observed at non-target locus, unlikely to be true off-

target.

No data.

Abbreviation: ES, embryonic stem.

https://doi.org/10.1371/journal.pgen.1007994.t001

PLOS Genetics | https://doi.org/10.1371/journal.pgen.1007994 March 14, 2019 3 / 8

Supporting studies

Although the results of Iyer and colleagues are based on the targeting of a single gene, other

similar studies have subsequently been published that support these findings. Using trios con-

taining parents and embryos, these studies aimed to remove any variation within an animal

colony, to achieve a better estimate of the de novo mutation burden in treated animals. The

most recent study used a very similar approach to Iyer and colleagues—treating mouse zygotes

with individual sgRNAs [10]. Using WGS of trios, they were able to compare de novo muta-

tions in the CRISPR-treated mice to their untreated littermates. Like Iyer and colleagues, no

appreciable increase in the de novo mutation burden of treated animals was observed when

compared with untreated controls. In another study, WGS was used to analyse trios of Cas9-e-

dited monkeys [11]. In this case, two separate experiments were performed, with potential off-

target effects assessed in both by comparing the location of de novo mutations with predicted

off-target locations, and by comparing the de novo mutation burden as a function of Cas9 edit-

ing efficiency, which was used as a surrogate for untreated control animals. As described for

the above-mentioned mouse experiments, there was no association between Cas9 editing effi-

ciency and de novo mutation burden. Moreover, as the monkey editing experiments generated

both knock-out and knock-in alleles, the presence of an HDR template does not appear to

have an effect on off-target activity. This does not, however, exclude the fact that off-target

mutations can occur, as demonstrated by deep-sequencing analysis of 81 gene editing experi-

ments in mouse and rat [12]. Detailed analysis of 10 mouse embryos and their genetic parents

identified 43 true Cas9-generated off-target mutations consisting of small insertions or dele-

tions. Although a considerable number of off-target mutations were detected, the authors

acknowledge that this probably represents a worst-case scenario, because the specificity score

of the sgRNA used in this experiment was very low. Collectively, these and other studies [13]

highlight the importance of controlling for the effect of confounding genetic variation within a

colony of animals when seeking to identify possible off-target mutations.

An elegant alternative approach to WGS is circularization for in vitro reporting of cleavage

effects by sequencing (CIRCLE-seq) [14], an in vitro screening method that uses short-read

sequencing of circularized sheared genomic DNA to identify CRISPR-induced DNA damage

at all susceptible sites; both on- and off- target. As expected, the number of off-target muta-

tions identified by CIRCLE-seq increases with the number of sgRNA mismatches with a pro-

miscuous sgRNA generating multiple off-targets both in vitro and in vivo in somatically edited

mice [15].

Regarding Kosicki and colleagues, it should be noted that large on-target deletions have

also been observed at a low frequency in mice in some circumstances following treatment of

single-cell mouse zygotes with an individual sgRNA [16, 17]. Unlike Iyer and colleagues, these

studies microinjected Cas9 mRNA into the cytoplasm, which may have extended the activity

of the Cas9 protein beyond the two- or four-cell stage. The potential outcomes of CRISPR-

induced DSBs at these early stages in the developing zygote are therefore highly context- and

method-dependent.

Best-practice to reduce the risk of unwanted CRISPR damage

sgRNA selection criteria for minimising unwanted on-target damage

As demonstrated in Kosicki and colleagues, the potential DNA-repair outcomes resulting

from individual CRISPR-induced DSBs may be significantly larger and more complex than

previously anticipated. Although the full extent of these outcomes remains to be determined,

these events are likely to be highly context-dependent. In a recent study that investigated the

PLOS Genetics | https://doi.org/10.1371/journal.pgen.1007994 March 14, 2019 4 / 8

targeting outcomes for more than 1,000 sgRNAs, nucleotides within the target sequence at the

−2 to −5 position relative to the PAM were found to be critical for defining the editing preci- sion [18]. The composition of the −4 nucleotide position adjacent to the cleavage site was par- ticularly significant, with either an “A” or “T” frequently associated with a nucleotide

insertion. Conversely, editing outcomes at target sites with a “G” nucleotide at the −4 position were found to be the most imprecise, inducing a variety of unpredictable deletions. Similar

results were observed in a larger study, involving more than 40,000 sgRNAs, in which cell-

line–specific differences were observed [19]. These studies have derived methods for predict-

ing Cas9 editing outcomes based on nucleotide composition context at the target site. With

the further development of such tools, it may be possible to bias DNA-repair outcomes in

favor of smaller NHEJ-associated indels, as opposed to larger deletions associated with either

microhomology-mediated end joining (MMEJ) or homologous recombination (HR).

sgRNA selection criteria for minimizing off-target damage

Reducing the risk of on-target deletions does not alter the fact that complex rearrangements or

mutations at DSBs might occur at off-target as well as on-target locations. Indeed, it is the

innate ability of the native CRISPR-Cas systems to recognize both the target sequence and

highly similar sequences [20] that raises the most concerns for CRISPR applications. As shown

by Iyer and colleagues, it is possible to mitigate potential off-target effects by selecting sgRNAs

with minimal potential off-target sites, as demonstrated by the specific targeting of the Tyr locus.

It may not always be possible to select sgRNAs with minimal off-targets, because a target

region can impose specific constraints. Regions that are either repetitive or duplicated—such

as with pseudogenes, paralogous gene expansions, or copy number variants—can result in

sgRNAs with a much higher predicted off-target burden. These sgRNAs are likely to recognize

highly similar sequences in other genomic regions that might contain as few as one or two

nucleotide differences. Although such sgRNAs are unlikely to be used for therapeutic applica-

tions, they have been identified within pooled CRISPR libraries. These sgRNAs may confound

the analysis of whole-genome screens, influencing results in a cell-line–specific manner that

can lead to false positives and biased essentiality scores [21].

Finally, the current tools used for predicting off-targets are primarily based on the analysis

of reference genome assemblies, which, although sufficient for most research purposes, will

need to include personal genome variation for therapeutic applications. Such applications can

be expected to employ highly annotated sgRNAs with well-defined off-target profiles that

could potentially exclude highly sensitive genomic regions, such as tumor suppressor loci.

Cas9 specificity

Although sgRNA selection is a critical component for minimizing off-target effects, it is also

dependent on the specificity of the Cas9 endonuclease. Because the native Cas9 endonucleases

are known to tolerate mismatches [22, 23], there have been significant efforts to engineer

improved versions of the Cas9 endonuclease with increased on-target–binding specificity and

reduced off-targets [24–27]. However, a consequence of this increased specificity is that the

overall activity of the higher fidelity Cas9 versions can be diminished for some sgRNA targets

[28]. This is sure to be an area in which further advances are made in the coming years.

Genotyping

Of course, none of these practices will entirely exclude the possibility of unwanted damage

occurring at either on-target or off-target locations, hence the requirement for in-vivo

PLOS Genetics | https://doi.org/10.1371/journal.pgen.1007994 March 14, 2019 5 / 8

characterization of sgRNAs in either mouse zygotes or defined cell lines. This should include

comprehensive genotyping with correct controls, either with WGS as described by Iyer and

colleagues or CIRCLE-seq [14]. Although the larger structural variation highlighted by Kosicki

and colleagues may be less common, multiple DSBs within the same chromosome associated

with either on-target or combined on- and off-target activity can result in complex rearrange-

ments [29]. In these instances, further validation using long-range genotyping approaches (for

review, see [30]) will be required, as suggested by Kosicki and colleagues.

Avoiding DSBs

Although all of these approaches are aimed at limiting the impact of unwanted effects associ-

ated with CRISPR-induced DSBs, it will ultimately be preferable to avoid, where possible, the

in vivo cleavage of DNA in future therapeutic applications. Depending on the required out-

come, it may be possible to alter the genome without creating DSBs, using either CRISPRa/i

[31], epigenome-editing [32], base-editing [33], or RNA-editing [34] approaches. If DNA

cleavage is required, an ex vivo approach may be used to edit a patient’s cells under laboratory

conditions. Under these circumstances, comprehensive genotyping should be used to confirm

the absence of unintended mutations and off-target effects before these cells are returned to

the patient.

Take home message

Genome editing can be used with precision to engineer the genome, by following best prac-

tices. Quoting Professor Rodolphe Barrangou, “Keep calm and CRISPR on” [35].

Author Contributions

Conceptualization: Mark Thomas, David J. Adams, Vivek Iyer.

Supervision: David J. Adams.

Writing – original draft: Mark Thomas, Vivek Iyer.

Writing – review & editing: Mark Thomas, Gaetan Burgio, David J. Adams, Vivek Iyer.

References

1. Iyer V, Boroviak K, Thomas M, Doe B, Riva L, Ryder E, et al. No unexpected CRISPR-Cas9 off-target

activity revealed by trio sequencing of gene-edited mice. PLoS Genet. 2018; 14(7):e1007503. https://

doi.org/10.1371/journal.pgen.1007503 PMID: 29985941

2. Kosicki M, Tomberg K, Bradley A. Repair of double-strand breaks induced by CRISPR–Cas9 leads to

large deletions and complex rearrangements. Nature biotechnology. 2018; 36:765. https://doi.org/10.

1038/nbt.4192 PMID: 30010673

3. Schaefer KA, Wu WH, Colgan DF, Tsang SH, Bassuk AG, Mahajan VB. Unexpected mutations after

CRISPR-Cas9 editing in vivo. Nature methods. 2017; 14(6):547–8. https://doi.org/10.1038/nmeth.4293

PMID: 28557981

4. Hendel A, Kildebeck Eric J, Fine Eli J, Clark Joseph T, Punjya N, Sebastiano V, et al. Quantifying

Genome-Editing Outcomes at Endogenous Loci with SMRT Sequencing. Cell Reports. 2014; 7(1):293–

305. https://doi.org/10.1016/j.celrep.2014.02.040 PMID: 24685129

5. Howlett SK, Bolton VN. Sequence and regulation of morphological and molecular events during the first

cell cycle of mouse embryogenesis. Journal of Embryology and Experimental Morphology. 1985; 87

(1):175–206.

6. Lin S, Staahl BT, Alla RK, Doudna JA. Enhanced homology-directed human genome engineering by

controlled timing of CRISPR/Cas9 delivery. eLife. 2014; 3:e04766. https://doi.org/10.7554/eLife.04766

PMID: 25497837

PLOS Genetics | https://doi.org/10.1371/journal.pgen.1007994 March 14, 2019 6 / 8

7. Gu B, Posfai E, Rossant J. Efficient generation of targeted large insertions by microinjection into two-

cell-stage mouse embryos. Nature biotechnology. 2018; 36(7):632–7. https://doi.org/10.1038/nbt.4166

PMID: 29889212

8. Reichmann J, Nijmeijer B, Hossain MJ, Eguren M, Schneider I, Politi AZ, et al. Dual-spindle formation in

zygotes keeps parental genomes apart in early mammalian embryos. Science (New York, NY). 2018;

361(6398):189–93.

9. Wilde JJ, Aida T, Wienisch M, Zhang Q, Qi P, Feng G. Efficient Zygotic Genome Editing via RAD51-

Enhanced Interhomolog Repair. bioRxiv. 2018.

10. Willi M, Smith HE, Wang C, Liu C, Hennighausen L. Mutation frequency is not increased in CRISPR-

Cas9-edited mice. Nature methods. 2018; 15(10):756–8. https://doi.org/10.1038/s41592-018-0148-2

PMID: 30275594

11. Luo X, He Y, Zhang C, He X, Yan L, Li M, et al. Trio deep-sequencing does not reveal unexpected muta-

tions in Cas9-edited monkeys. bioRxiv. 2018.

12. Anderson KR, Haeussler M, Watanabe C, Janakiraman V, Lund J, Modrusan Z, et al. CRISPR off-target

analysis in genetically engineered rats and mice. Nature methods. 2018; 15(7):512–4. https://doi.org/

10.1038/s41592-018-0011-5 PMID: 29786090

13. Wang G, Du M, Wang J, Zhu TF. Genetic variation may confound analysis of CRISPR-Cas9 off-target

mutations. Cell Discovery. 2018; 4(1):18.

14. Tsai SQ, Nguyen NT, Malagon-Lopez J, Topkar VV, Aryee MJ, Joung JK. CIRCLE-seq: a highly sensi-

tive in vitro screen for genome-wide CRISPR–Cas9 nuclease off-targets. Nature methods. 2017;

14:607. https://doi.org/10.1038/nmeth.4278 PMID: 28459458

15. Akcakaya P, Bobbin ML, Guo JA, Malagon-Lopez J, Clement K, Garcia SP, et al. In vivo CRISPR edit-

ing with no detectable genome-wide off-target mutations. Nature. 2018; 561(7723):416–9. https://doi.

org/10.1038/s41586-018-0500-9 PMID: 30209390

16. Shin HY, Wang C, Lee HK, Yoo KH, Zeng X, Kuhns T, et al. CRISPR/Cas9 targeting events cause com-

plex deletions and insertions at 17 sites in the mouse genome. Nature communications. 2017; 8:15464.

https://doi.org/10.1038/ncomms15464 PMID: 28561021

17. Adikusuma F, Piltz S, Corbett MA, Turvey M, McColl SR, Helbig KJ, et al. Large deletions induced by

Cas9 cleavage. Nature. 2018; 560(7717):E8–e9. https://doi.org/10.1038/s41586-018-0380-z PMID:

30089922

18. Chakrabarti AM, Henser-Brownhill T, Monserrat J, Poetsch AR, Luscombe NM, Scaffidi P. Target-Spe-

cific Precision of CRISPR-Mediated Genome Editing. Molecular cell. 2018.

19. Allen F, Crepaldi L, Alsinet C, Strong AJ, Kleshchevnikov V, De Angeli P, et al. Predicting the mutations

generated by repair of Cas9-induced double-strand breaks. Nature biotechnology. 2018.

20. Han P, Niestemski LR, Barrick JE, Deem MW. Physical Model of the Immune Response of Bacteria

Against Bacteriophage Through the Adaptive CRISPR-Cas Immune System. Physical biology. 2013;

10(2): https://doi.org/10.1088/478-3975/10/2/025004

21. Fortin JP, Tan J, Gascoigne KE, Haverty PM, Forrest WF, Costa MR, et al. Multiple-gene targeting and

mismatch tolerance can confound analysis of genome-wide pooled CRISPR screens. Genome biology.

2019; 20(1):21. https://doi.org/10.1186/s13059-019-1621-7 PMID: 30683138

22. Hsu PD, Scott DA, Weinstein JA, Ran FA, Konermann S, Agarwala V, et al. DNA targeting specificity of

RNA-guided Cas9 nucleases. Nature biotechnology. 2013; 31:827. https://doi.org/10.1038/nbt.2647

PMID: 23873081

23. Lin Y, Cradick TJ, Brown MT, Deshmukh H, Ranjan P, Sarode N, et al. CRISPR/Cas9 systems have

off-target activity with insertions or deletions between target DNA and guide RNA sequences. Nucleic

Acids Research. 2014; 42(11):7473–85. https://doi.org/10.1093/nar/gku402 PMID: 24838573

24. Hu JH, Miller SM, Geurts MH, Tang W, Chen L, Sun N, et al. Evolved Cas9 variants with broad PAM

compatibility and high DNA specificity. Nature. 2018; 556(7699):57–63. https://doi.org/10.1038/

nature26155 PMID: 29512652

25. Kleinstiver BP, Pattanayak V, Prew MS, Tsai SQ, Nguyen NT, Zheng Z, et al. High-fidelity CRISPR-

Cas9 nucleases with no detectable genome-wide off-target effects. Nature. 2016; 529(7587):490–5.

https://doi.org/10.1038/nature16526 PMID: 26735016

26. Lee JK, Jeong E, Lee J, Jung M, Shin E, Kim YH, et al. Directed evolution of CRISPR-Cas9 to increase

its specificity. Nature communications. 2018; 9(1):3048. https://doi.org/10.1038/s41467-018-05477-x

PMID: 30082838

27. Slaymaker IM, Gao L, Zetsche B, Scott DA, Yan WX, Zhang F. Rationally engineered Cas9 nucleases

with improved specificity. Science (New York, NY). 2016; 351(6268):84–8.

PLOS Genetics | https://doi.org/10.1371/journal.pgen.1007994 March 14, 2019 7 / 8

28. Kulcsar PI, Talas A, Huszar K, Ligeti Z, Toth E, Weinhardt N, et al. Crossing enhanced and high fidelity

SpCas9 nucleases to optimize specificity and cleavage. Genome biology. 2017; 18(1):190. https://doi.

org/10.1186/s13059-017-1318-8 PMID: 28985763

29. Boroviak K, Fu B, Yang F, Doe B, Bradley A. Revealing hidden complexities of genomic rearrange-

ments generated with Cas9. Scientific reports. 2017; 7(1):12867. https://doi.org/10.1038/s41598-017-

12740-6 PMID: 28993641

30. Pollard MO, Gurdasani D, Mentzer AJ, Porter T, Sandhu MS. Long reads: their purpose and place.

Human molecular genetics. 2018; 27(R2):R234–r41. https://doi.org/10.1093/hmg/ddy177 PMID:

29767702

31. Gilbert Luke A, Larson Matthew H, Morsut L, Liu Z, Brar Gloria A, Torres Sandra E, et al. CRISPR-Medi-

ated Modular RNA-Guided Regulation of Transcription in Eukaryotes. Cell. 2013; 154(2):442–51.

https://doi.org/10.1016/j.cell.2013.06.044 PMID: 23849981

32. Amabile A, Migliara A, Capasso P, Biffi M, Cittaro D, Naldini L, et al. Inheritable Silencing of Endoge-

nous Genes by Hit-and-Run Targeted Epigenetic Editing. Cell. 2016; 167(1):219–32.e14. https://doi.

org/10.1016/j.cell.2016.09.006 PMID: 27662090

33. Komor AC, Kim YB, Packer MS, Zuris JA, Liu DR. Programmable editing of a target base in genomic

DNA without double-stranded DNA cleavage. Nature. 2016; 533:420. https://doi.org/10.1038/

nature17946 PMID: 27096365

34. Abudayyeh OO, Gootenberg JS, Essletzbichler P, Han S, Joung J, Belanto JJ, et al. RNA targeting with

CRISPR–Cas13. Nature. 2017; 550:280. https://doi.org/10.1038/nature24049 PMID: 28976959

35. Barrangou R. Keep Calm and CRISPR On. The CRISPR Journal. 2018; 1(1):1–3.

PLOS Genetics | https://doi.org/10.1371/journal.pgen.1007994 March 14, 2019 8 / 8