Crossing enhanced and high fidelity SpCas9 nucleases to optimize specificity and cleavage
© The Author(s). 2017
Received: 26 April 2017
Accepted: 11 September 2017
Published: 6 October 2017
The propensity for off-target activity of Streptococcus pyogenes Cas9 (SpCas9) has been considerably decreased by rationally engineered variants with increased fidelity (eSpCas9; SpCas9-HF1). However, a subset of targets still generate considerable off-target effects. To deal specifically with these targets, we generated new “Highly enhanced Fidelity” nuclease variants (HeFSpCas9s) containing mutations from both eSpCas9 and SpCas9-HF1 and examined these improved nuclease variants side by side to decipher the factors that affect their specificities and to determine the optimal nuclease for applications sensitive to off-target effects.
These three increased-fidelity nucleases can routinely be used only with perfectly matching 20-nucleotide-long spacers, a matching 5′ G extension being more detrimental to their activities than a mismatching one. HeFSpCas9 exhibit substantially improved specificity for those targets for which eSpCas9 and SpCas9-HF1 have higher off-target propensity. The targets can also be ranked by their cleavability and off-target effects manifested by the increased fidelity nucleases. Furthermore, we show that the mutations in these variants may diminish the cleavage, but not the DNA-binding, of SpCas9s.
No single nuclease variant shows generally superior fidelity; instead, for highest specificity cleavage, each target needs to be matched with an appropriate high-fidelity nuclease. We provide here a framework for generating new nuclease variants for targets that currently have no matching optimal nuclease, and offer a simple means for identifying the optimal nuclease for targets in the absence of accurate target-ranking prediction tools.
Cas9 proteins are RNA-guided endonucleases that can be directed to cleave a chosen DNA sequence [1–4]. This process requires complementarity between the Cas9-associated single guide RNA (sgRNA) and the target site in addition to the presence of a short protospacer-adjacent motif (PAM) at the 3′ end of the target [5–13]. Although it has been demonstrated that the Streptococcus pyogenes Cas9 (SpCas9) nuclease can be used for genome engineering, its widespread use has been limited by its off-target activity; i.e., the nuclease also cleaves targets that show limited, imperfect complementarities with the associated sgRNA [14–21]. The off-target sequences are difficult to predict and have been shown to contain mismatches in up to 5 or 6 positions [15, 22, 23], a property that may interfere with many research applications as well as therapeutic uses. Much effort has been devoted to circumvent these confounding effects of the nuclease, such as reducing the amount of active Cas9 in the cell [14, 24, 25], using truncated sgRNAs that bear shortened regions of target site complementarity , engineering SpCas9 mutants , using paired SpCas9 nickases [27, 28], or using pairs of catalytically inactive SpCas9 fused to a non-specific FokI nuclease domain [29–31]. Recently, attempts to use structure-guided engineering of SpCas9 to reduce its off-target activities have been reported: the enhanced SpCas9 [eSpCas9(1.1), K848A/K1003A/R1060A] , hereinafter referred to as eSpCas9, was developed to decrease the affinity of the protein for the non-target DNA strand, hence increasing the strand’s propensity for reinvading the RNA–DNA hybrid helix and, therefore, decreasing the stability of mismatch-containing helices. By contrast, mutations contained in the high fidelity Cas9 (SpCas9-HF1, N497A/R661A/Q695A/Q926A)  that weaken the interactions of the protein with the target DNA strand are aimed at decreasing the energetics of the SpCas9–sgRNA complex so that it retains a robust on-target activity but has a diminished ability to cleave mismatched off-target sites. Both mutants exhibited considerably reduced off-target effects when assessed by unbiased whole-genome off-target analysis: their cleavage activities toward off-targets with multiple mismatches were almost completely eliminated, although some off-targets, mainly with single-base mismatches, were found. However, a subset of targets, referred to as atypical, with repetitive or homopolymeric sequences were still cleaved with considerable off-target effects . While these results are very encouraging, it is difficult to decide which SpCas9 variant is superior for applications where the avoidance of off-target activity is of paramount interest because they were characterized in differing experimental setups that exploited different sets of targets in different (either U2OS or HEK) cells and employed different methods (GUIDE-seq  versus BLESS ) to assess their genome-wide specificity.
Here, we generate new variants (“Highly enhanced Fidelity” or HeFSpCas9s) of SpCas9, containing combinations of mutations from both eSpCas9 and SpCas9-HF1, that show higher fidelity specifically with respect to those targets for which eSpCas9 and SpCas9-HF1 exhibit a higher off-target propensity. Furthermore, we directly compare these highly improved nucleases in the same system to understand the factors that affect their specificity and to help select for which off-target-sensitive applications each would be most suitable.
To assess the activity of the Highly enhanced Fidelity SpCas9 nuclease (HeFSpCas9) we performed direct comparisons with eSpCas9, SpCas9-HF1, and the wild type (WT) SpCas9 on a large number of on-target sites. Testing 16 genomic targets in N2a cells, we measured either the on-target HR-mediated integration of a GFP cassette with 1000-bp-long homology arms  (Additional file 1: Figure S1a) or the indel frequencies by Tracking of Indels by DEcomposition (TIDE)  on five genomic loci (Pten, Prnp, Rbl2, Ttn, and Tp53) (Additional file 1: Figure S1b). To our surprise, the increased fidelity nuclease variants performed poorly in these assays in contrast to those reported earlier [32, 33]. A closer inspection of the results revealed that many of these targets had been targeted by 21-nucleotide-long spacers bearing a guanine (G) extension at the start. This is a commonly applied modification to comply with the preference for a G nucleotide as the transcription initiation site for the human U6 promoter  that is used for the sgRNAs here and which is the most commonly used promoter for mammalian sgRNA expression vectors .
5′ Extended sgRNAs diminish the activities of improved fidelity nucleases more with a matching than with a mismatching G nucleotide
Ranking by activity and target discrimination/selectivity of improved fidelity nuclease variants
Fidelity ranking among the mutant SpCas9 nuclease variants
These results show that there is a big difference in fidelity between WT SpCas9 and eSpCas9 or SpCas9-HF1. Whereas WT SpCas9 barely distinguishes perfect matches from mismatches for the majority of the targets examined, eSpCas9 and SpCas9-HF1 are capable of strong discrimination (Fig. 4; Additional file 1: Figure S3b). eSpCas9 does not exhibit any detectable cleavage (< 3%) with 20 out of 48 mismatch positions of sgRNAs, whereas SpCas9-HF1 shows no cleavage with 28 out of 47. These confirm earlier reports that eSpCas9 and SpCas9-HF1 have greatly increased target fidelity not only when presented with multiple- but also with single-base mismatched sequences. Our results also reveal that SpCas9-HF1 possesses higher fidelity, exhibiting less off-target cleavage than eSpCas9. The number of spacers with mismatching positions that result in higher than 80% of the corresponding disruption levels found with the matching spacers are: 42 for the WT protein, 18 for eSpCas9, and only three for SpCas9-HF1 (Additional file 1: Figure S3b). Interestingly, for some targets (targets 1, 3, 4, and 5 in Fig. 4a) even SpCas9-HF1 shows higher off-target effects, although the difference between the sequence characteristics of these targets with low and high off-target cleavage is not apparent. Importantly, HeFSpCas9 exhibits high-fidelity activity particularly on these targets (Fig. 4a).
Since several off-target positions resulted in disruption levels at around or under the detection limit of the assay, we performed deep sequencing on 67 samples to assess more precisely their respective fidelity on single-base mismatched sequences (for details of the read numbers see Additional file 2: Deepseq reads). SpCas9-HF1 also proved to be a higher fidelity nuclease than eSpCas9 by deep sequencing, showing considerably lower off-target cleavage (greater than twofold) with 7 out of 14 mismatch-position containing guide RNAs (Fig. 4c). When examining those targets by deep sequencing (targets 1, 3, 4, and 5) for which eSpCas9 and SpCas9-HF1 failed to achieve high fidelity, HeFSpCas9 showed detectable off-target activity only with 4 out of 12 mismatching positions, with the rest exhibiting off-target activity indistinguishable from the background. These results demonstrate a spectacularly increased specificity of HeFSpCas9 on these targets of about 50- to 400-fold over the two other increased fidelity nucleases (Fig. 4d). Thus, the rank of the nucleases is, in order of increasing fidelity, eSpCas9 < SpCas9-HF1 < HeFSpCas9.
Ranking among the targets
Applying the four nucleases to a number of targets and exploiting a great number of sgRNAs with mismatching guides (Fig. 4a) revealed another important feature: a ranking among the targets due to differences in the efficiency with which targets are cleaved. Targets efficiently cleaved by a nuclease with both perfectly matching and mismatching guides were likely to be cleaved efficiently by another higher fidelity nuclease with perfectly matching sgRNAs but not (or much less) with mismatching ones. In this way targets could be ranked according to their requirements for a type of nuclease to result in optimal (efficient with minimal off-targets), high specificity cleavage. In this ranking, at one end are those targets that are efficiently cleaved by all three nuclease variants; however, only the highest fidelity HeFSpCas9 cleaves them with little to no off-target effect. Such targets are, e.g., 1, 3, 4, and 5 in Figs. 3 and 4. On this kind of target both eSpCas9 and SpCas9-HF1 showed significant off-target activities with single-base mismatched guides: in particular they showed, on average, 93 and 60% of the corresponding on-target activities, respectively, while for the rest of the targets (12/16) these off-target activities were only 26 and 6%, respectively (calculated from the data shown in Fig. 4). By contrast, at the other end of the ranks are those targets that are cleaved by eSpCas9 with > 80% efficiency of WT SpCas9 (such as targets 8 and 24 in Fig. 4a) and without much off-target activity (note that for such targets even the WT protein has decreased off-target activity). Although the difference in terms of fidelity and target selectivity is much larger between SpCas9-HF1 and HeFSpCas9, target ranking is also clearly discernible between eSpCas9 and SpCas9-HF1, targets 2, 9, 14, and 16 being efficiently and much more accurately targeted by SpCas9-HF1.
To confirm that these results are not specific only for the mouse cell line used, we selected nine targets of various ranks (covering optimal sequences for each of the variants, e-, -HF1, or HeFSpCas9 and used previously in the experiment in Fig. 4a) and repeated the same experiment in human cells (HEK-293) using a cell line containing a single-copy integrated EGFP (HEK-293.EGFP), similar to the N2a.EGFP cells used. To each target site nine single-base mismatched guides were applied, thus using altogether 81 mismatching guides. The patterns revealed by these experiments on the nucleases (Additional file 1: Figure S4) are almost identical to those found previously (shown on Fig. 4a), supporting the idea that the characteristics of the increased fidelity nucleases and that of the targets, which have become apparent in this study, are intrinsic and are not specific to the particular cell line used in the experiments. The new observation about target ranking reported here is a key aspect of selecting the optimal variant for a particular target.
Since off-target effects are also dependent on the amount of the active nuclease present, we have also compared the activities as a function of the expression levels of these nucleases while making considerable efforts to match the expression levels of the variants as closely as we could (see Additional file 1: Figures S5 and S6 and Supplementary results 2 for more details). These experiments revealed that the relative efficiencies and specificities of these nuclease variants seen in these studies, particularly the differences between eSpCas9 and SpCas9-HF1, are primarily determined by their intrinsic characteristics; however, it cannot be ruled out that there might also be a contribution to the lower on-target activities and higher fidelity of SpCas9-HF1 seen here from its somewhat lower expression levels from these vectors.
From the observed target ranking it should follow that the higher the cleavage activity of a nuclease with one mismatching position-bearing spacer on a target, the higher is its activity with any other type of modified/imperfectly matching (extended, truncated, or with another mismatching position) guide on the same target, if they act using the same mechanism. Such a correlation is discernible from experiments employing sgRNAs with mismatching and extended spacers on the same set of targets (Fig. 4a; Additional file 1: Figure S2a, c). Correlation matrix analysis of the activity data shows positive Pearson correlations (0.83 or 0.78) that are significant between the disruption activities of either eSpCas9 or SpCas9-HF1 programmed with sgRNAs bearing single mismatching nucleotides and those being extended with a 5′ end mismatching G targeting the same site (Additional file 1: Figure S7a, c). In addition, differing positions of mismatches for the same targets also showed significant positive correlation (between 0.73 and 0.93 and 0.73 and 0.97 for eSpCas9 and SpCas9-HF1, respectively; Additional file 1: Figure S7b, d). This result further supports the idea of target ranking but also implies that a similar mechanism determines how these imperfectly matched or modified guides affect the cleavage activities of SpCas9 nucleases.
To confirm the observed nuclease and target ranking, an additional 26 spacers were screened for on-target cleavage activity in the EGFP disruption assay. eSpCas9 and SpCas9-HF1 demonstrate decreased activity on some of these targets compared to the WT protein to an extent comparable to that found previously (Additional file 1: Figure S8a, b). Importantly, while HeFSpCas9 shows no detectable activity on 22 out of 26 targets, it shows about 80% of the WT activity on two targets (EGFP sites 26 and 43). To test the fidelity of SpCas9-HF1 and HeFSpCas9 when dealing with these two targets, nine one-base mismatching sgRNAs were applied in case of both. In line with expectations, SpCas9-HF1 cleaves these targets with non-perfectly matching sgRNAs (Additional file 1: Figure S8c). By contrast, HeFSpCas9 demonstrated considerably less off-target activity in these experiments.
Generation of a new variant with in-between activity and fidelity
Kleinstiver and coworkers showed that SpCas9-HF1 efficiently cleaves the off-target site 1 of FANCF site 2 that bears a mismatch close to the PAM region . We expected that such types of target are a better match for the higher fidelity HeF nuclease variants and tested the nucleases on this target by measuring on-target and off-target indel generation through TIDE . All HeF variants showed only background-level cleavage activity at the off-target site, whereas there were differences in on-target activity; among them the activity of HeFm2SpCas9 was the highest and comparable to that of the WT (Fig. 5c). This supports our conjecture that HeFm2SpCas9 might be a good candidate to fill the gap between SpCas9-HF1 and HeFSpCas9 and demonstrates the relevance of our approach.
Binding of these nuclease variants to inefficiently cleaved target DNAs is apparently not diminished compared to the WT nuclease
Employing a more direct in vitro approach, we performed polyacrylamide-gel electrophoretic mobility shift assay of the cleaved target DNAs by the nucleases, exploiting three targets that were shown to be cleaved efficiently by WT SpCas9 but not by HeFSpCas9. By these experiments, we confirmed that although HeF nucleases do not show nuclease activity, they retain most of their DNA-binding abilities to these targets (Additional file 1: Figure S10). Furthermore, to understand the effect of appending an extra G nucleotide to the 5′ end of the guides, we examined the in vitro binding of eSpCas9 and SpCas9-HF1 charged with 21-nucleotide-long sgRNAs to selected targets that were only cleaved when the 20-nucleotide-long spacers were applied in the disruption assay (Fig. 2). We found that although the matching G extension fully diminishes the cleavage activities of these variants on these targets, their binding seems to remain unaffected (Additional file 1: Figure S10).
In principle, there are two types of approaches that have been successfully used to decrease off-target cleavage activities of SpCas9. On the one hand, the length of the target sequence necessary to elicit double strand cleavage may be increased by employing dCas9-FokI [29, 30, 31] or nickases [27, 28]. On the other hand, lessening the promiscuity of SpCas9 can be achieved by lowering its activity on off-target sequences while maintaining reasonably effective on-target activities. This is attempted by minimizing the exposure of the DNA to SpCas9 activity by limiting the time of its expression [24, 25, 54–57] or decreasing the level of protein [58, 59], employing modified sgRNAs (truncated  or 5′ extended with a GG dinucleotide ) or weakening the protein–DNA interactions [32, 33]. Although a systematic comparison of these approaches is still missing, they have led to varying success and it has been observed that different targets are cleaved with different off-target propensity by SpCas9. Unfortunately, in the absence of a sufficiently accurate prediction tool, the off-target propensity of SpCas9 on a given target can be determined only by time-consuming and laborious genome-wide off-target analysis, apart from when homopolymeric and repetitive sequences are encountered, which have been already reported to be cleaved with high off-target effects and hence could be easily avoided. For the targeting of sequences in the bulk of the genome our results take a step forward in overcoming these difficulties.
The most important outcomes of the experiments reported here are (i) the observed fidelity or activity and target selectivity ranking among the increased fidelity nucleases and (ii) a ranking of the cleavability of targets without off-target effects by these nuclease variants. These results may be understandable now, but before this study it was not clear [32, 33] whether the mutations in eSpCas9 or SpCas9-HF1, although aiming at non-specific contacts, altered their sequence specificity, i.e., if eSpCas9 and SpCas9-HF1 exhibit lower activities on distinct sets of targets, a result which would not support the existence of the target-ranking discerned here. It was also not clear whether SpCas9-HF1 had higher fidelity than eSpCas9 and whether off-target propensities varied according to the target.
According to our findings, eSpCas9 possesses the lowest target selectivity among the increased fidelity nucleases examined here. It has activity comparable to that of WT SpCas9 when employed with 20-nucleotide-long, perfectly matching sgRNAs and because of its higher fidelity its routine use is preferable to using WT SpCas9 for practically all applications. We did not find a single target for which WT SpCas9 shows higher specificity (defined as on-target/off-target ratio) than eSpCas9. In contrast, SpCas9-HF1 showed strongly decreased activities for some targets (such as targets 6, 18, 19, and 21 in Fig. 3a and 27, 40, 45 and 50 in Additional file 1: Figure S8a) in these experiments. Even though it exhibits markedly higher specificity for several targets (e.g., targets 2, 9, 14, and 16 in Fig. 4) than eSpCas9, it may not be advisable to use it routinely for any targets without pretesting. Kleinstiver et al. suggested that SpCas9-HF1 cleaves only “atypical”, repetitive, or homopolymeric targets with substantial off-target propensity . According to our experiments, some typical targets that are not homopolymeric or repetitive sequences are also cleaved with higher off-target propensity by SpCas9-HF1. The proportion of such targets exceeds 10% occurrence in our experiments. Unfortunately, at present we are not able to predict which targets would have a higher off-target propensity, which limits the unrestricted use of SpCas9-HF1 for DNA modifications aimed at avoiding off-target effects. The HeF nucleases developed here exhibit greatly increased fidelity and an accompanying increased target selectivity/lower activity, a finding that may not be surprising. However, we found it striking that they exhibit this high fidelity specifically for those targets for which SpCas9-HF1 fails to show improved fidelity. This suggests that HeF variants will be very useful complements to the already existing increased fidelity nucleases in genome engineering applications, perfectly fulfilling the anticipated role for which we generated them.
Another interesting result is the finding that 5′ extending the sgRNAs with a single G nucleotide diminishes the activities of the increased fidelity SpCas9 variants. A 5′ GG dinucleotide extension of the sgRNA has been reported to decrease the off-target cleavage propensity of SpCas9 with certain targets . Since the 5′ end of the sgRNAs protrudes relatively freely from the known SpCas9–sgRNA structures  (Additional file 1: Figure S11a, b), the mechanism by which the 5′ dinucleotide extension decreases off-target propensity was not apparent and it was suggested that it may alter the guide RNA’s stability, concentration, or secondary structure . More recently a structure of SpCas9 has been published, which provides the most detailed picture of a cleavage-competent state of the SpCas9–sgRNA complex . According to this structure the 5′ end of the sgRNA is not only buried inside the protein, but if extended it could only exit the protein’s structure at a separate opening that is at some distance from the target DNA strand and separated from it by parts of the protein (Additional file 1: Figure S11c–e). Such a structural arrangement would explain how a 5′ G extension may disturb the cleavage but not the binding of SpCas9. It also makes it understandable how a matching G extension that would lengthen the RNA:DNA hybrid helix may diminish much more the cleavage activities of SpCas9 than a mismatching G may, by causing larger distortions to the cleavage-competent structure.
The results of these studies also extend our knowledge of the factors that influence on-target and off-target cleavages of SpCas9 and its increased fidelity variants in several respects. High fidelity nucleases exhibit strongly reduced activities with truncated sgRNAs, leading to the conclusion that they are likely to improve specificity by a similar mechanism to that of decreasing the interaction energy of the protein–sgRNA complex or R-loop stability [22, 23]. Our results may further complement this picture. First, increased fidelity nuclease variants show decreased activity with 5′ extended guides as well, suggesting a similar mechanism of action between extending (or truncating) the guide and variants engineered to have reduced non-sequence specific DNA contacts. Second, this contention of similar mechanisms is further strengthened by the correlations we found between the activity-reducing effect of appending a 5′ G to the sgRNAs and of using imperfectly matching sgRNAs for cleaving the same target sequences with an increased fidelity nuclease (Additional file 1: Figure S7). Third, these correlations also show that highly reduced off-target effects vary according to the target, an effect that we term “off-targetless cleavability ranking” and is also apparent in Fig. 4a. Fourth, we show here that increased fidelity nucleases using sgRNAs that are incapable of cleaving the DNA are able to elicit transcription activation and bind DNA in vitro similarly to the WT protein (Fig. 6; Additional file 1: Figure S10). Altogether these observations suggest that these modifications, truncating or extending the guide and reducing non-sequence specific protein–DNA contacts, improve SpCas9 fidelity by primarily affecting the catalytic activity of the cleavage complex without much altering the target binding of SpCas9.
The most simple interpretation of these data is one similar to that originally proposed to explain the off-target activity of engineered zinc finger proteins [23, 61, 62]. However, in the case of SpCas9, the specificity of its cleavage activity is determined by the interaction energy of its cleavage complex rather than by its binding. In this scenario, the actual sequence of a target also contributes to the energy of the cleavage complex. With high energy-contributing targets, the cleavage complex may possess enough excess energy to tolerate mismatches in the target DNA:RNA hybrid helix. With low energy-contributing targets the cleavage complex has no or has less excess energy; thus, it is less prone to off-target cleavage. Approaches such as employing various increased fidelity nuclease variants and truncating or extending the sgRNAs decrease the energy of the cleavage complex to different extents: in some cases retaining considerable off-target propensity, or in other cases abolishing even the on-target activity, whereas in optimal cases eliminating most or all of the off-target while retaining high on-target cleavage activity.
Consequently, our data point to the fruitlessness of attempting to generate a “generally superior” SpCas9 nuclease variant with overall highest specificity. Rather, an optimal nuclease variant needs to be identified for each target; this knowledge is critical for achieving further minimization of unwanted, off-target cleavage of the genome. Our data suggest that the fidelity of the improved nuclease variants might be further increased by combining them with a nickase approach. However, since the binding activity of these increased fidelity nucleases does not seem to decrease for targets not being cleaved by the given variant (Additional file 1: Figure S10), a combination with a dSpCas9-FokI approach is less likely to be rewarding.
Our results on the use of 5′ G extended sgRNAs (Fig. 2) also offer an alternative solution for increasing the fidelity of WT SpCas9 cleavage. Given the fact that extending the sgRNA with a G nucleotide only modestly decreases the activity of WT SpCas9 (10% on average; Fig. 2), on the basis of the interpretations provided here, we predict a considerably decreased off-target cleavage with such targets. Furthermore, it is plausible to speculate that in contrast to the so called tru-sgRNAs , a matching G extension may not generate new off-target sites that are not detected using the same unmodified sgRNA-WT SpCas9 complex. In addition, it is likely to increase the fidelity of WT SpCas9 more than non-matching 5′ G extensions. These two ideas need further confirmation.
One of the biggest challenges in the field remains to develop methods for the prediction of the ranking of the targets and their matching to the optimal off-target minimizing strategy. Both the approaches we developed here and unbiased genome-wide off-target analyses are excessively laborious and time-consuming, and thus it is not feasible to apply them to reveal targets’ cleavability ranking for routine use of SpCas9. However, the ranking of the increased fidelity nucleases observed here offers a straightforward solution to this problem: a simple pretesting of the candidate target with eSpCas9, SpCas9-HF1, and HeFm2SpCas9 would reveal which is the highest fidelity nuclease exhibiting sufficient activity for the given application.
For the therapeutic usage of the SpCas9-based technology it is of paramount priority to reduce the possibilities for incidental off-targets to a minimum, even below the detection limit of the currently existing approaches (<0.1% imposed by the current NGS technology). These applications frequently involve the optimization of a procedure based on exploiting one or a few targets that are at appropriate positions for their later routine use. The rankings observed here among these nucleases and among targets and the extremely high fidelity of these increased fidelity nucleases on certain targets suggest that, by careful matching of the nuclease to the target being exploited, the off-target cleavage may be reduced even further to the level of incidentally occurring off-targets that are not detectable by the current methods. These, in turn, would greatly reduce the whole genome sequencing effort required to validate engineered cells for clinical applications without unwanted genomic alterations.
Increased fidelity nucleases can routinely only be used with perfectly matching 20-nucleotide-long spacers. A 5′ G extension of the sgRNAs is especially detrimental for increased fidelity nucleases if the appended G matches the target diminishing their cleavage activity. A fidelity and target selectivity ranking of the increased fidelity SpCas9 nucleases and a cleavability ranking of targets could be revealed here due to our special experimental design: application of SpCas9 variants with significantly different fidelities compared side by side on the same set of targets using a number of imperfectly matching sgRNAs. These experiments revealed that eSpCas9 has lower fidelity than SpCas9-HF1; however, owing to its low target selectivity it may be used instead of WT SpCas9 for higher fidelity gene editing in most of applications. Interestingly, contrary to expectations, the higher fidelity SpCas9-HF1 also exhibited considerable off-target propensity not only on “atypical” but also on “typical” targets that feature no homopolymeric or repetitive sequences. By combining the mutations of these two increased fidelity nucleases, we generated HeFSpCas9, a highly enhanced fidelity nuclease, that is a very useful complement to SpCas9-HF1 by cleaving exactly those targets with little to no off-target effects (with a considerable on-target activity) for which SpCas9-HF1 presented lower fidelity. The observed ranking of the nucleases offers straightforward means for finding the optimal nuclease variant for efficient on-target and minimal off-target modifications. Our approach also provides a framework for the generation of new high-fidelity nuclease variants with fidelity in between that of the existing ones, as it is demonstrated here by the development of HeFm2SpCas9. This approach allows “optimal” increased fidelity nuclease variants to tailored to individual targets, matching them to achieve accurate genome editing with minimal off-target effects.
Restriction enzymes, Klenow polymerase, T4 ligase, Dulbecco’s modified Eagle medium (DMEM; Gibco), fetal bovine serum (Gibco), Turbofect, TranscriptAid T7 High Yield Transcription Kit and penicillin/streptomycin were purchased from Thermo Fischer Scientific, protease inhibitor cocktail was purchased from Roche Diagnostics. DNA oligonucleotides and GenElute HP Plasmid Miniprep kit used in plasmid purifications were acquired from Sigma-Aldrich. Q5 High-Fidelity DNA Polymerase was from New England Biolabs Inc. NucleoSpin Gel and PCR Clean-up kit was purchased from Macherey-Nagel. All plasmid constructs and PCR products were sequenced by Microsynth AG.
Vectors were constructed using standard molecular biology techniques including one-pot cloning method , Escherichia coli DH5α-mediated DNA assembly method , and Body Double cloning method . For detailed cloning and sequence information see Additional file 3. sgRNA target sites and mismatching sgRNAs are available in Additional file 2. The sequences of all plasmid constructs were confirmed by Sanger sequencing.
Plasmids acquired from the non-profit plasmid distribution service Addgene (http://www.addgene.org/) were the following: pX330-U6-Chimeric_BB-CBh-hSpCas9 (Addgene #42230) , eSpCas9(1.1) (Addgene #71814) , VP12 (Addgene #72247) , pX335-U6-Chimeric_BB-CBh-hSpCas9n(D10A) (Addgene #42335) , pET-dCas9-VP64-6xHis (#62935) , sgRNA(MS2) cloning backbone (Plasmid #61424) , MS2-P65-HSF1_GFP (Plasmid #61423) .
Plasmids developed by us and that are deposited at Addgene are the following: pX330-Flag-wtSpCas9 (without sgRNA; Addgene #92353), pX330-Flag-eSpCas9 (without sgRNA; Addgene #92354), pX330-Flag-SpCas9-HF1 (without sgRNA; Addgene #92102), HeFSpCas9 (Addgene #92355), HeFm1SpCas9 (Addgene #92110), HeFm2SpCas9 (Addgene #92111), pX330-HA-dSpCas9 (Addgene #92112), pX330-Flag-dSpCas9 (Addgene #92113), pX330-Flag-deSpCas9 (Addgene #92114), pX330-Flag-dSpCas9-HF1 (Addgene #92115), pX330-Flag-dHeFSpCas9 (Addgene #92116), pET-deSpCas9-VP64-6xHis (Addgene #92117), pET-dSpCas9-HF1-VP64-6xHis (Addgene #92118), pET-dHeFSpCas9-VP64-6xHis (Addgene #92119), pmCherry_gRNA (Addgene: #80457), U6-sgRNA(MS2)_EF1a-MS2-P65-HSF1 (Addgene #92120).
Cell culture and transfection
N2a (neuro-2a mouse neuroblastoma cells, ATCC, CCL-131) cells, N2a.EGFP and HEK-293.EGFP cells (both cell lines containing a single integrated copy of an EGFP cassette driven by the Prnp promoter), and HeLa (ATCC, CCL-2) cells were grown at 37 °C in a humidified atmosphere of 5% CO2 in high glucose DMEM supplemented with 10% heat-inactivated fetal bovine serum, 4 mM L-glutamine (Gibco), 100 units/ml penicillin and 100 μg/ml streptomycin. N2a, N2a.EGFP, and HEK-293.EGFP cells were plated one day prior to transfection on 48-well plates at a density of approximately 30,000 cells (35,000 cells in the case of HEK-293.EGFP). HeLa cells were plated one day prior to transfection in 24-well plates at a density of approximately 60,000 cells. Transfections were performed with TurboFect transfection reagent according to the manufacturer’s recommended protocol and as detailed below for each corresponding assay performed. Transfections were performed in triplicate unless otherwise noted.
Flow cytometry analyses were carried out on Attune Acoustic Focusing Cytometer (Applied Biosystems) or on CytoFLEX Flow Cytometer (Beckman Coulter). For data analysis Attune Cytometric Software v.2.1.0 and CytExpert 2.0 were used. Viable single cells were gated based on side and forward light scatter parameters and a total of 5000–10,000 viable single cell events were acquired in all experiments. Attune Acoustic Focusing Cytometer parameters: the GFP fluorescence signal was detected using the 488-nm diode laser for excitation and the 530/30-nm filter for emission; the mCherry fluorescent signal was detected using the 488-nm diode laser for excitation and a 640LP filter for emission. CytoFLEX Flow Cytometer parameters: the GFP fluorescence signal was detected using the 488-nm diode laser for excitation and the 525/40-nm filter for emission; the mCherry fluorescent signal was detected using the 638-nm diode laser for excitation and a 660/20-nm filter for emission.
EGFP disruption assay
N2a.EGFP and HEK-293.EGFP cells were co-transfected with two types of plasmids: SpCas9 expression plasmid (137 ng) and sgRNA and mCherry coding plasmid (78–97 ng) using 1 μl TurboFect reagent per well in 48-well plates. Where indicated, less SpCas9 plasmid variants were used and their amounts are completed to 137 ng by a mock plasmid with identical size. Transfected cells were analysed ∼ 72 and ∼ 168 h post-transfection by flow cytometry. Transfection efficacy was calculated via mCherry-expressing cells measured ∼ 72 h post-transfection. EGFP-positive cells were counted both ∼ 72 and ∼ 168 h post-transfection. Background level of EGFP for each experiment was determined using non-transfected cells and also two types of control transfected population: (i) using co-transfection of a mock plasmid (pmCherry_sgRNA) and an active SpCas9 plasmid; or (ii) using co-transfection of a dead SpCas9 expression plasmid and a targeted sgRNA and mCherry coding plasmid.
Percentage of EGFP disruption in the case of N2a.EGFP cells was calculated as follows. The measured percentage of EGFP-positive cells for each sample was subtracted from the total average obtained on the controls and was weighted by its transfection efficiency factor. The transfection efficiency factor was obtained utilizing the mCherry present in the samples: by measuring the percentage of mCherry-positive cells in individual samples and calculating their deviation from the average percentage of mCherry-positive cells obtained on all the transfected samples/wells as (Average mCherry – Sample mCherry)/(Average mCherry). For each sample three parallels were processed and their values were averaged. The errors associated with the final values were estimated by taking into account the errors of each term, mCherry and EGFP controls as well, using Gaussian error propagation on the standard deviations. When data were further used for ratio plotting, the errors were also processed further by Gaussian propagation to yield the final represented error bars. In the case of HEK-293.EGFP cells the mCherry fluorescence was not used to normalize the data.
HR-mediated integration assay
Cells were co-transfected with three types of plasmids: an expression plasmid for EGFP flanked by 1000-bp-long homology arms to the Prnp gene (referred to as Prnp.HA-EGFP plasmid; Tálas et al. ; 166 ng), SpCas9 expressing plasmid (42 ng), and an sgRNA/mCherry coding plasmid (42 ng), giving 250 ng total plasmid DNA, using 1 μl TurboFect reagent per well on 48-well plates. Transfected cells were analyzed ∼ 72 h and 12 days post-transfection by flow cytometry. Transfection efficacy was calculated via mCherry-expressing cells measured ∼ 72 h post-transfection. EGFP-positive cells were counted 12 days post-transfection on three parallels. Background level of EGFP was determined by control co-transfection using dead SpCas9 expression plasmid, Prnp.HA-GFP plasmid and a mouse Prnp site 1B targeting-sgRNA/mCherry coding plasmid. Values obtained for each sample well were weighted to count for their transfection efficiency utilizing the sample’s mCherry fluorescence and the average mCherry fluorescence value obtained based on all transfected wells as described in an EGFP disruption assay. The weighted values of the samples were averaged for parallels and corrected for the control average values. The error was estimated by Gaussian error propagation of the errors (standard deviation) associated with each term experimentally determined that was used for calculation of the value.
The Tracking of Indels by DEcomposition (TIDE) method  was applied for analyzing mutations and determining their frequency in a cell population using different sgRNAs and SpCas9 proteins.
N2a cells were co-transfected with 137 ng of SpCas9 expressing plasmid and 97 ng of sgRNA and mCherry coding plasmid (250 ng total plasmid DNA) using 1 μl TurboFect reagent per well on 48-well plates.
HeLa cells were co-transfected with 400 ng of SpCas9-expressing plasmid and 600 ng of sgRNA/mCherry coding plasmid (1 μg total plasmid DNA) using 2 μl TurboFect reagent per well on 24-well plates.
Control samples were made for each different genomic target site by co-transfecting a dead SpCas9 expression plasmid and the targeted sgRNA and mCherry coding plasmid.
Transfected cells were divided ∼ 72 h post-transfection as follows: 20% of the cells were analyzed for transfection efficacy via mCherry fluorescence by flow cytometry and from the rest of the cells genomic DNA was extracted by following the Gentra DNA Purification protocol (Gentra Puregen Handbook, Qiagen) from the mix of the triplicates. From the isolated genomic DNA PCR was conducted with Q5 High-Fidelity DNA Polymerase (for PCR primer details, see Additional file 2). Genomic PCR products were gel excised (in the case of experiments shown in Additional file 1: Figure S1b) or directly purified (in the case of experiments shown in Fig. 5c) via NucleoSpin Gel and PCR Clean-up kit and were Sanger sequenced. Indel efficiencies were analyzed using the TIDE webtool (https://tide.nki.nl/) by comparing Cas9-treated and control samples.
Indel analysis by next-generation sequencing
Off-targets with low cleavage efficacy in the EGFP disruption assay together with their on-targets were examined via targeted resequencing. The genomic DNA was extracted at ∼ 7 days post-transfection from the mix of the triplicates by following the Gentra DNA Purification protocol. PCR fragments for NGS analysis were generated in two-step PCR reactions. Briefly, first step PCR primers (Additional file 2) contained both the PCR handles for the second round amplification and the target-specific sequence to amplify genomic regions of interest. The second ten-cycle PCR step, the purification (Ampure Bead clean up), the pooling, the gel excised purification, and the 150-bp single-end sequencing on an Illumina MiSeq instrument were performed by Microsynth AG.
N2a.EGFP cells were cultured on a 48-well plate and transfected as described above in the “EGFP disruption assay” section. Three days post-transfection, eight parallel samples corresponding to each type of Cas9 transfected were washed with PBS, then trypsinized and mixed, and were analyzed for transfection efficacy via mCherry fluorescence level by using flow cytometry. The cells from the mixtures were pelleted (at 200 rcf for 5 min at 4 °C). The pellet was resuspended in ice cold Harlow buffer (50 mM Hepes pH 7.5; 0.2 mM EDTA; 10 mM NaF; 0.5% NP40; 250 mM NaCl; Proteinase Inhibitor Cocktail 1:100; Calpain inhibitor 1:100; 1 mM DTT) and lysed for 20–30 min. The cell lysates were centrifuged at 19,000 rcf for 10 min. The supernatants were transferred into new tubes and total protein concentration was measured by Bradford protein assay. Before SDS gel loading, samples were boiled in Protein Loading Dye for 10 min at 95 °C. Proteins were separated by SDS-PAGE using 7.5% polyacrylamide gels and were transferred to PVDF membrane, using a wet blotting system (Bio-Rad). Membranes were blocked by 5% non-fat milk in Tris buffered saline with Tween20 (TBST; blocking buffer) for 2 h. Blots were incubated overnight at 4 °C with primary antibodies [anti-FLAG (F1804, Sigma) at 1:1000 dilution; anti-β-actin (A1978, Sigma) at 1:4000 dilution in blocking buffer]. The next day after washing steps in TBST the membranes were incubated for 1 h with HRP-conjugated secondary anti-mouse antibody 1:20,000 (715-035-151, Jackson ImmunoResearch) in blocking buffer. The signal from detected proteins was visualized by ECL (Pierce ECL Western Blotting Substrate, Thermo Scientific) by CCD camera (Bio-Rad ChemiDoc MP).
N2a.EGFP cells were co-transfected with three types of plasmids as follows: 91 ng of SpCas9-expressing plasmid, 83 ng of the mixture of 5 sgRNA coding plasmid, which expresses MS2-p65-HFS1 fusion protein as well, and 75 ng of mCherry coding plasmid (pcDNA3-mCherry) using 1 μl TurboFect reagent per well in 48-well plates. Transfected cells were analyzed ∼ 72 h post-transfection. The transfection efficacy was calculated via mCherry fluorescence level. The relative upregulation was calculated from the median of the EGFP intensity. Background level of EGFP was determined using a negative control for transfections: a mock plasmid ([MS2-p65-HFS1_sgRNA(MS2)] without spacer sequence) co-transfected with SpCas9 coding plasmid and with mCherry coding plasmid. The medians obtained for EGFP fluorescence were averaged between three parallel samples and the error was estimated by Gaussian error propagation of the component errors (standard devitation) associated with the measured variables.
In vitro transcription
sgRNAs were in vitro transcribed using TranscriptAid T7 High Yield Transcription Kit and PCR-generated double-stranded DNA templates carrying a T7 promoter sequence. Primers used for the preparation of the DNA templates are listed in Additional file 2. sgRNAs were quality checked using 10% denaturing polyacrylamide gels and ethidium bromide staining.
pET-dCas9-VP64-6xHis (#62935) was acquired from Addgene and the other Cas9 variants (pET-deSpCas9-VP64-6xHis, pET-dSpCas9-HF1-VP64-6xHis, pET-dHeFSpCas9-VP64-6xHis) were subcloned by us (for detailed cloning information and sequence information see the “Plasmid construction” section and Additional file 3). The resulting fusion constructs contained an N-terminal hexahistidine (His6) tag.
The expression constructs of the dead Cas9 variants were transformed into BL21 Rosetta 2 (DE3) cells and were processed similarly, as follows. Cells were grown in LB medium at 37 °C for 5 h and 10 ml of this culture was inoculated into 1 l of Terrific Broth growth media and cells were grown at 37 °C to a final cell density of 0.6 OD600 and then chilled to 18 °C. The protein was expressed at 18 °C for 16 h following induction with 0.2 mM IPTG. The protein was purified by a combination of chromatographic steps using an NGC Scout Medium-Pressure Chromatography System (Bio-Rad). Briefly, the cells were harvested and resuspended in 30 ml of lysis buffer (40 mM Tris pH 8.0, 500 mM NaCl, 20 mM imidazole, 1 mM TCEP) supplemented with protease inhibitor cocktail (complete, EDTA-free) and were sonicated on ice. The lysate was cleared by centrifugation at 18,000 rcf for 40 min at 4 °C. The supernatant was bound in batch to a 5 ml Mini Nuvia IMAC Ni-Charged column (Bio-Rad). The resin was washed extensively with 40 mM Tris pH 8.0, 500 mM NaCl, 20 mM imidazole, and the bound protein was eluted by 40 mM Tris pH 8.0, 250 mM imidazole, 150 mM NaCl elution buffer. The protein was dialyzed 2 × 1 h against 20 mM HEPES pH 7.5, 150 mM KCl, 1 mM DTT, 1% glycerol. The dialyzed protein was purified on a 3 × 1 ml Bio-Scale Mini Macro-Prep High S column (Bio-Rad), eluting with 1 M KCl, 20 mM HEPES pH 7.5, 1 mM DTT. The protein was further purified by size exclusion chromatography on a Superdex 200 16/60 column in 20 mM HEPES pH 7.5, 150 mM KCl, and 1 mM DTT. The eluted protein was tested on SDS-PAGE and Coomassie brilliant blue r-250 staining and was stored at −20 °C until use.
Electrophoretic mobility shift assay
Binding assays were performed in buffer containing 20 mM HEPES pH 7.5, 100 mM KCl, 5 mM MgCl2, 0.1 mg/ml heparin, and 1 mM TCEP in a total volume of 20 μl. The sgRNA was supplied at two times the molar amount of protein. The target DNA (40 nM) was incubated with protein–sgRNA complex (160 nM; or in the case of the serial dilution experiments with 160, 320, 640, and 1280 nM protein–sgRNA complex concentrations, respectively) for 30 min at 37 °C. Samples were resolved at 4 °C on an 8% native polyacrylamide gel containing 0.5× TBE and 10 mM MgCl2. The gel was stained with ethidium bromide. (For detailed information about sgRNAs and DNA targets see Additional file 2).
Differences between samples were tested using Welch’s one-way Anova with Games–Howell post hoc tests for samples with unequal variances and/or sample size (Figs. 2a, b, 3b, 5a; Additional file 1: Figures S2b, S3b, S8b) and by one-way Anova with Tukey’s post-hoc test for homoscedastic samples (Fig. 3b; Additional file 1: Figures S1f and S2a, 2e). Homogeneity of variances was tested by Levene’s test. Statistical tests were performed using R version 3.4.1 (C: 2017 The R Foundation for Statistical Computing) and packages FSA, Car, multcomp, userfriendlyscience, dplyr, ggplot2. Test results are in Additional file 4.
The investigators were not blinded to group assignment and outcome assessment.
We thank Bernadett Czene, Ildikó Szűcsné Pulinka, Luca Pájer-Turgyán, Anita Paulovitsné Csatári, and Dávid Fetter for their excellent laboratory assistance, Lőrinc Pongor, Adrienn Éva Borsy, Márk Saskőy, Éva Varga, Dániel Bátora, István Vida, György Várady, Tamás Hegedűs, and Antal Nyeste for their valuable help.
This work was supported by the National Research, Development and Innovation Office (KFI_16-1-2016-0240).
Availability of data and materials
Plasmids used in this study are available at Addgene: details in the “Plasmid construction” section. Sequences of the constructs are listed in Additional file 3. The deep sequencing data from this study have been submitted to the NCBI Sequence Read Archive (SRA; http://www.ncbi.nlm.nih.gov/sra/) under accession number SRR5005491.
PIK and EW conceived of and designed experiments, PIK, AT, ET, KH, ZL, and EF performed all experiments. PIK, AT, ET, KH, ZL, and EF analyzed the data. PIK, EF, and EW wrote the manuscript with input from all the authors. All authors read and approved the final manuscript.
Ethics approval and consent to participate
Ethics approval was not needed for the study.
The authors declare that they have no significant competing financial, professional, or personal interests that might have influenced the performance or presentation of the work described in this manuscript.
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.
- Sander JD, Joung JK. CRISPR-Cas systems for editing, regulating and targeting genomes. Nat Biotechnol. 2014;32:347–55.View ArticlePubMedPubMed CentralGoogle Scholar
- Hsu PD, Lander ES, Zhang F. Development and applications of CRISPR-Cas9 for genome engineering. Cell. 2014;157:1262–78.View ArticlePubMedPubMed CentralGoogle Scholar
- Doudna JA, Charpentier E. Genome editing. The new frontier of genome engineering with CRISPR-Cas9. Science. 2014;346:1258096.View ArticlePubMedGoogle Scholar
- Jinek M, East A, Cheng A, Lin S, Ma E, Doudna J. RNA-programmed genome editing in human cells. Elife. 2013;2:e00471.View ArticlePubMedPubMed CentralGoogle Scholar
- Garneau JE, Dupuis ME, Villion M, Romero DA, Barrangou R, Boyaval P, Fremaux C, Horvath P, Magadan AH, Moineau S. The CRISPR/Cas bacterial immune system cleaves bacteriophage and plasmid DNA. Nature. 2010;468:67–71.View ArticlePubMedGoogle Scholar
- Jinek M, Chylinski K, Fonfara I, Hauer M, Doudna JA, Charpentier E. A programmable dual-RNA-guided DNA endonuclease in adaptive bacterial immunity. Science. 2012;337:816–21.View ArticlePubMedGoogle Scholar
- Mali P, Yang L, Esvelt KM, Aach J, Guell M, DiCarlo JE, Norville JE, Church GM. RNA-guided human genome engineering via Cas9. Science. 2013;339:823–6.View ArticlePubMedPubMed CentralGoogle Scholar
- Cong L, Ran FA, Cox D, Lin S, Barretto R, Habib N, Hsu PD, Wu X, Jiang W, Marraffini LA, Zhang F. Multiplex genome engineering using CRISPR/Cas systems. Science. 2013;339:819–23.View ArticlePubMedPubMed CentralGoogle Scholar
- Anders C, Niewoehner O, Duerst A, Jinek M. Structural basis of PAM-dependent target DNA recognition by the Cas9 endonuclease. Nature. 2014;513:569–73.View ArticlePubMedPubMed CentralGoogle Scholar
- Anders C, Bargsten K, Jinek M. Structural Plasticity of PAM Recognition by Engineered Variants of the RNA-Guided Endonuclease Cas9. Mol Cell. 2016;61:895–902.View ArticlePubMedPubMed CentralGoogle Scholar
- Jiang F, Zhou K, Ma L, Gressel S, Doudna JA. Structural biology: a Cas9-guide RNA complex preorganized for target DNA recognition. Science. 2015;348:1477–81.View ArticlePubMedGoogle Scholar
- Jinek M, Jiang F, Taylor DW, Sternberg SH, Kaya E, Ma E, Anders C, Hauer M, Zhou K, Lin S, et al. Structures of Cas9 endonucleases reveal RNA-mediated conformational activation. Science. 2014;343:1247997.View ArticlePubMedPubMed CentralGoogle Scholar
- Nishimasu H, Ran FA, Hsu PD, Konermann S, Shehata SI, Dohmae N, Ishitani R, Zhang F, Nureki O. Crystal structure of Cas9 in complex with guide RNA and target DNA. Cell. 2014;156:935–49.View ArticlePubMedPubMed CentralGoogle Scholar
- Hsu PD, Scott DA, Weinstein JA, Ran FA, Konermann S, Agarwala V, Li Y, Fine EJ, Wu X, Shalem O, et al. DNA targeting specificity of RNA-guided Cas9 nucleases. Nat Biotechnol. 2013;31:827–32.View ArticlePubMedPubMed CentralGoogle Scholar
- Fu Y, Foden JA, Khayter C, Maeder ML, Reyon D, Joung JK, Sander JD. High-frequency off-target mutagenesis induced by CRISPR-Cas nucleases in human cells. Nat Biotechnol. 2013;31:822–6.View ArticlePubMedPubMed CentralGoogle Scholar
- Pattanayak V, Lin S, Guilinger JP, Ma E, Doudna JA, Liu DR. High-throughput profiling of off-target DNA cleavage reveals RNA-programmed Cas9 nuclease specificity. Nat Biotechnol. 2013;31:839–43.View ArticlePubMedPubMed CentralGoogle Scholar
- Cradick TJ, Fine EJ, Antico CJ, Bao G. CRISPR/Cas9 systems targeting beta-globin and CCR5 genes have substantial off-target activity. Nucleic Acids Res. 2013;41:9584–92.View ArticlePubMedPubMed CentralGoogle Scholar
- Tsai SQ, Zheng Z, Nguyen NT, Liebers M, Topkar VV, Thapar V, Wyvekens N, Khayter C, Iafrate AJ, Le LP, et al. GUIDE-seq enables genome-wide profiling of off-target cleavage by CRISPR-Cas nucleases. Nat Biotechnol. 2015;33:187–97.View ArticlePubMedGoogle Scholar
- Frock RL, Hu J, Meyers RM, Ho YJ, Kii E, Alt FW. Genome-wide detection of DNA double-stranded breaks induced by engineered nucleases. Nat Biotechnol. 2015;33:179–86.View ArticlePubMedGoogle Scholar
- Wang X, Wang Y, Wu X, Wang J, Wang Y, Qiu Z, Chang T, Huang H, Lin RJ, Yee JK. Unbiased detection of off-target cleavage by CRISPR-Cas9 and TALENs using integrase-defective lentiviral vectors. Nat Biotechnol. 2015;33:175–8.View ArticlePubMedGoogle Scholar
- Cho SW, Kim S, Kim Y, Kweon J, Kim HS, Bae S, Kim JS. Analysis of off-target effects of CRISPR/Cas-derived RNA-guided endonucleases and nickases. Genome Res. 2014;24:132–41.View ArticlePubMedPubMed CentralGoogle Scholar
- Tsai SQ, Joung JK. Defining and improving the genome-wide specificities of CRISPR-Cas9 nucleases. Nat Rev Genet. 2016;17:300–12.View ArticlePubMedGoogle Scholar
- Fu Y, Sander JD, Reyon D, Cascio VM, Joung JK. Improving CRISPR-Cas nuclease specificity using truncated guide RNAs. Nat Biotechnol. 2014;32:279–84.View ArticlePubMedPubMed CentralGoogle Scholar
- Zetsche B, Volz SE, Zhang F. A split-Cas9 architecture for inducible genome editing and transcription modulation. Nat Biotechnol. 2015;33:139–42.View ArticlePubMedGoogle Scholar
- Davis KM, Pattanayak V, Thompson DB, Zuris JA, Liu DR. Small molecule-triggered Cas9 protein with improved genome-editing specificity. Nat Chem Biol. 2015;11:316–8.View ArticlePubMedPubMed CentralGoogle Scholar
- Kleinstiver BP, Prew MS, Tsai SQ, Topkar VV, Nguyen NT, Zheng Z, Gonzales AP, Li Z, Peterson RT, Yeh JR, et al. Engineered CRISPR-Cas9 nucleases with altered PAM specificities. Nature. 2015;523:481–5.View ArticlePubMedPubMed CentralGoogle Scholar
- Ran FA, Hsu PD, Lin CY, Gootenberg JS, Konermann S, Trevino AE, Scott DA, Inoue A, Matoba S, Zhang Y, Zhang F. Double nicking by RNA-guided CRISPR Cas9 for enhanced genome editing specificity. Cell. 2013;154:1380–9.View ArticlePubMedPubMed CentralGoogle Scholar
- Mali P, Aach J, Stranges PB, Esvelt KM, Moosburner M, Kosuri S, Yang L, Church GM. CAS9 transcriptional activators for target specificity screening and paired nickases for cooperative genome engineering. Nat Biotechnol. 2013;31:833–8.View ArticlePubMedPubMed CentralGoogle Scholar
- Tsai SQ, Wyvekens N, Khayter C, Foden JA, Thapar V, Reyon D, Goodwin MJ, Aryee MJ, Joung JK. Dimeric CRISPR RNA-guided FokI nucleases for highly specific genome editing. Nat Biotechnol. 2014;32:569–76.View ArticlePubMedPubMed CentralGoogle Scholar
- Guilinger JP, Thompson DB, Liu DR. Fusion of catalytically inactive Cas9 to FokI nuclease improves the specificity of genome modification. Nat Biotechnol. 2014;32:577–82.View ArticlePubMedPubMed CentralGoogle Scholar
- Wyvekens N, Topkar VV, Khayter C, Joung JK, Tsai SQ. Dimeric CRISPR RNA-guided FokI-dCas9 nucleases directed by truncated gRNAs for highly specific genome editing. Hum Gene Ther. 2015;26:425–31.View ArticlePubMedPubMed CentralGoogle Scholar
- Slaymaker IM, Gao L, Zetsche B, Scott DA, Yan WX, Zhang F. Rationally engineered Cas9 nucleases with improved specificity. Science. 2016;351:84–8.View ArticlePubMedGoogle Scholar
- Kleinstiver BP, Pattanayak V, Prew MS, Tsai SQ, Nguyen NT, Zheng Z, Joung JK. High-fidelity CRISPR-Cas9 nucleases with no detectable genome-wide off-target effects. Nature. 2016;529:490–5.View ArticlePubMedPubMed CentralGoogle Scholar
- Crosetto N, Mitra A, Silva MJ, Bienko M, Dojer N, Wang Q, Karaca E, Chiarle R, Skrzypczak M, Ginalski K, et al. Nucleotide-resolution DNA double-strand break mapping by next-generation sequencing. Nat Methods. 2013;10:361–5.View ArticlePubMedPubMed CentralGoogle Scholar
- Tóth E, Weinhardt N, Bencsura P, Huszár K, Kulcsár PI, Tálas A, Fodor E, Welker E. Cpf1 nucleases demonstrate robust activity to induce DNA modification by exploiting homology directed repair pathways in mammalian cells. Biol Direct. 2016;11:46.View ArticlePubMedPubMed CentralGoogle Scholar
- Brinkman EK, Chen T, Amendola M, van Steensel B. Easy quantitative assessment of genome editing by sequence trace decomposition. Nucleic Acids Res. 2014;42:e168.View ArticlePubMedPubMed CentralGoogle Scholar
- Goomer RS, Kunkel GR. The transcriptional start site for a human U6 small nuclear RNA gene is dictated by a compound promoter element consisting of the PSE and the TATA box. Nucleic Acids Res. 1992;20:4903–12.View ArticlePubMedPubMed CentralGoogle Scholar
- Reyon D, Tsai SQ, Khayter C, Foden JA, Sander JD, Joung JK. FLASH assembly of TALENs for high-throughput genome editing. Nat Biotechnol. 2012;30:460–5.View ArticlePubMedPubMed CentralGoogle Scholar
- Haurwitz RE, Jinek M, Wiedenheft B, Zhou K, Doudna JA. Sequence- and structure-specific RNA processing by a CRISPR endonuclease. Science. 2010;329:1355–8.View ArticlePubMedPubMed CentralGoogle Scholar
- Xie K, Minkenberg B, Yang Y. Boosting CRISPR/Cas9 multiplex editing capability with the endogenous tRNA-processing system. Proc Natl Acad Sci U S A. 2015;112:3570–5.View ArticlePubMedPubMed CentralGoogle Scholar
- Palermo G, Miao Y, Walker RC, Jinek M, McCammon JA. Striking plasticity of CRISPR-Cas9 and key role of non-target DNA, as revealed by molecular simulations. ACS Cent Sci. 2016;2:756–63.View ArticlePubMedPubMed CentralGoogle Scholar
- Gao YB, Zhao YD. Self-processing of ribozyme-flanked RNAs into guide RNAs in vitro and in vivo for CRISPR-mediated genome editing. J Integr Plant Biol. 2014;56:343–9.View ArticlePubMedGoogle Scholar
- Lee RTH, Ng ASM, Ingham PW. Ribozyme mediated gRNA generation for in vitro and in vivo CRISPR/Cas9 mutagenesis. Plos One. 2016;11:e0166020.View ArticlePubMedPubMed CentralGoogle Scholar
- Chavez A, Scheiman J, Vora S, Pruitt BW, Tuttle M, PRI E, Lin S, Kiani S, Guzman CD, Wiegand DJ, et al. Highly efficient Cas9-mediated transcriptional programming. Nat Methods. 2015;12:326–8.View ArticlePubMedPubMed CentralGoogle Scholar
- Gilbert LA, Larson MH, Morsut L, Liu Z, Brar GA, Torres SE, Stern-Ginossar N, Brandman O, Whitehead EH, Doudna JA, et al. CRISPR-mediated modular RNA-guided regulation of transcription in eukaryotes. Cell. 2013;154:442–51.View ArticlePubMedPubMed CentralGoogle Scholar
- Perez-Pinera P, Kocak DD, Vockley CM, Adler AF, Kabadi AM, Polstein LR, Thakore PI, Glass KA, Ousterout DG, Leong KW, et al. RNA-guided gene activation by CRISPR-Cas9-based transcription factors. Nat Methods. 2013;10:973–6.View ArticlePubMedPubMed CentralGoogle Scholar
- Maeder ML, Linder SJ, Cascio VM, Fu Y, Ho QH, Joung JK. CRISPR RNA-guided activation of endogenous human genes. Nat Methods. 2013;10:977–9.View ArticlePubMedPubMed CentralGoogle Scholar
- Chavez A, Tuttle M, Pruitt BW, Ewen-Campen B, Chari R, Ter-Ovanesyan D, Haque SJ, Cecchi RJ, Kowal EJ, Buchthal J, et al. Comparison of Cas9 activators in multiple species. Nat Methods. 2016;13:563–7.View ArticlePubMedPubMed CentralGoogle Scholar
- Gilbert LA, Horlbeck MA, Adamson B, Villalta JE, Chen Y, Whitehead EH, Guimaraes C, Panning B, Ploegh HL, Bassik MC, et al. Genome-scale CRISPR-mediated control of gene repression and activation. Cell. 2014;159:647–61.View ArticlePubMedPubMed CentralGoogle Scholar
- Qi LS, Larson MH, Gilbert LA, Doudna JA, Weissman JS, Arkin AP, Lim WA. Repurposing CRISPR as an RNA-guided platform for sequence-specific control of gene expression. Cell. 2013;152:1173–83.View ArticlePubMedPubMed CentralGoogle Scholar
- Konermann S, Brigham MD, Trevino AE, Joung J, Abudayyeh OO, Barcena C, Hsu PD, Habib N, Gootenberg JS, Nishimasu H, et al. Genome-scale transcriptional activation by an engineered CRISPR-Cas9 complex. Nature. 2015;517:583–8.View ArticlePubMedGoogle Scholar
- Kiani S, Chavez A, Tuttle M, Hall RN, Chari R, Ter-Ovanesyan D, Qian J, Pruitt BW, Beal J, Vora S, et al. Cas9 gRNA engineering for genome editing, activation and repression. Nat Methods. 2015;12:1051–4.View ArticlePubMedPubMed CentralGoogle Scholar
- Dahlman JE, Abudayyeh OO, Joung J, Gootenberg JS, Zhang F, Konermann S. Orthogonal gene knockout and activation with a catalytically active Cas9 nuclease. Nat Biotechnol. 2015;33:1159–61.View ArticlePubMedPubMed CentralGoogle Scholar
- Dow LE, Fisher J, O'Rourke KP, Muley A, Kastenhuber ER, Livshits G, Tschaharganeh DF, Socci ND, Lowe SW. Inducible in vivo genome editing with CRISPR-Cas9. Nat Biotechnol. 2015;33:390–U398.View ArticlePubMedPubMed CentralGoogle Scholar
- Nihongaki Y, Kawano F, Nakajima T, Sato M. Photoactivatable CRISPR-Cas9 for optogenetic genome editing. Nat Biotechnol. 2015;33:755–60.View ArticlePubMedGoogle Scholar
- Liu KI, Ramli MNB, Woo CWA, Wang YM, Zhao TY, Zhang XJ, Yim GRD, Chong BY, Gowher A, Chua MZH, et al. A chemical-inducible CRISPR-Cas9 system for rapid control of genome editing. Nat Chem Biol. 2016;12:980.View ArticlePubMedGoogle Scholar
- Wright AV, Sternberg SH, Taylor DW, Staahl BT, Bardales JA, Kornfeld JE, Doudna JA. Rational design of a split-Cas9 enzyme complex. Proc Natl Acad Sci U S A. 2015;112:2984–9.View ArticlePubMedPubMed CentralGoogle Scholar
- Zuris JA, Thompson DB, Shu Y, Guilinger JP, Bessen JL, Hu JH, Maeder ML, Joung JK, Chen ZY, Liu DR. Cationic lipid-mediated delivery of proteins enables efficient protein-based genome editing in vitro and in vivo. Nat Biotechnol. 2015;33:73–80.View ArticlePubMedGoogle Scholar
- Kim S, Kim D, Cho SW, Kim J, Kim JS. Highly efficient RNA-guided genome editing in human cells via delivery of purified Cas9 ribonucleoproteins. Genome Res. 2014;24:1012–9.View ArticlePubMedPubMed CentralGoogle Scholar
- Jiang FG, Taylor DW, Chen JS, Kornfeld JE, Zhou KH, Thompson AJ, Nogales E, Doudna JA. Structures of a CRISPR-Cas9 R-loop complex primed for DNA cleavage. Science. 2016;351:867–71.View ArticlePubMedPubMed CentralGoogle Scholar
- Beerli RR, Dreier B, Barbas 3rd CF. Positive and negative regulation of endogenous genes by designed transcription factors. Proc Natl Acad Sci U S A. 2000;97:1495–500.View ArticlePubMedPubMed CentralGoogle Scholar
- Pattanayak V, Ramirez CL, Joung JK, Liu DR. Revealing off-target cleavage specificities of zinc-finger nucleases by in vitro selection. Nat Methods. 2011;8:765–70.View ArticlePubMedPubMed CentralGoogle Scholar
- Engler C, Kandzia R, Marillonnet S. A one pot, one step, precision cloning method with high throughput capability. PLoS One. 2008;3:e3647.View ArticlePubMedPubMed CentralGoogle Scholar
- Kostylev M, Otwell AE, Richardson RE, Suzuki Y. Cloning should be simple: Escherichia coli DH5 alpha-mediated assembly of multiple DNA fragments with short end homologies. PLoS One. 2015;10:e0137466.View ArticlePubMedPubMed CentralGoogle Scholar
- Tóth E, Huszár K, Bencsura P, Kulcsár PI, Vodicska B, Nyeste A, Welker Z, Tóth S, Welker E. Restriction enzyme body doubles and PCR cloning: on the general use of type IIs restriction enzymes for cloning. PLoS One. 2014;9:e90896.View ArticlePubMedPubMed CentralGoogle Scholar
- Talas A, Kulcsar PI, Weinhardt N, Borsy A, Toth E, Szebenyi K, Krausz SL, Huszar K, Vida I, Sturm A, et al. A convenient method to pre-screen candidate guide RNAs for CRISPR/Cas9 gene editing by NHEJ-mediated integration of a 'self-cleaving' GFP-expression plasmid. DNA Res. 2017. doi: 10.1093/dnares/dsx029.
- Spitzer M, Wildenhain J, Rappsilber J, Tyers M. BoxPlotR: a web tool for generation of box plots. Nat Methods. 2014;11:121–2.View ArticlePubMedPubMed CentralGoogle Scholar