Viewpoint pubs.acs.org/synthbio
Targeting Specificity of the CRISPR/Cas9 System Ipek Tasan† and Huimin Zhao*,†,‡,§,∥,⊥,# †
Department of Biochemistry, ‡Department of Chemical and Biomolecular Engineering, §Carl R. Woese Institute for Genomic Biology, ∥Center for Biophysics and Quantitative Biology, ⊥Department of Chemistry, and #Department of Bioengineering, University of Illinois at Urbana−Champaign, Urbana, Illinois 61801, United States ABSTRACT: CRISPR/Cas9 system has accelerated research across many fields since its demonstration for genome editing. CRISPR also offers vast therapeutic potential, but an important hurdle of this technology is the off-target mutations it can induce. In this viewpoint, we will discuss recent strategies for improving CRISPR specificity, emphasizing how a complete mechanistic understanding of CRISPR/ Cas9 can benefit such efforts. We also propose that agreeing upon a consensus protocol with the highest specificity could benefit researchers working on CRISPR-based therapies. In addition to improving CRISPR/Cas9 specificity, accurate detection of off-target events is also crucial, and we will discuss various unbiased off-target detection methods in terms of their advantages and disadvantages. We suggest that using a combination of cell-based and cell-free methods can prove more useful. In addition, we point out that improving predictive algorithms for off-target sites would require pooling of the available off-target analysis data and standardization of the protocols used for obtaining the data. Moreover, we highlight the risk of insertional mutagenesis for gene correction applications requiring the use of donor DNA. We conclude by discussing future prospects for the field, as well as steps that can be taken to overcome the aforementioned challenges.
D
specificity of CRISPR/Cas9 and off-target detection methods. We also suggest strategies and considerations for further optimizations of CRISPR/Cas9. In addition, we emphasize that donor DNAs used for gene correction purposes can be another source of unintended mutations.
emonstration of clustered regularly interspaced short palindromic repeats (CRISPR)/CRISPR-associated protein 9 (Cas9) as a genome editing tool was one of the most significant breakthroughs in biology. CRISPR system consists of a Cas9-single guide RNA (sgRNA) complex, which recognizes a 20 bp target sequence with a downstream protospacer adjacent motif (PAM) and induces a site-specific double strand break (DSB). Although gene editing through nuclease-induced DSBs was feasible using zinc finger nucleases (ZFNs) or transcription activator-like effector nucleases (TALENs), the cost-effectiveness and ease of using CRISPR/ Cas9 made it a very attractive tool. CRISPR is applicable to a wide variety of organisms, which can be useful for various purposes, from crop engineering to human gene therapy. However, despite all the promises CRISPR has to offer, there are still debates over the extent of its off-target activity. Another debate was recently started after a provocative publication by Schaefer et al., which claims that CRISPR/Cas9 can create more than a thousand unintended mutations.1 Even though multiple scientists suggested that the reported abnormally high off-target activity was due to the flaws in experimental design and data analysis,2−4 it is still true that off-target mutations are an important caveat of CRISPR/Cas9 that needs to be addressed. Especially for therapeutic applications of CRISPR, even a low frequency of unintended mutations might have deleterious effects and improving CRISPR specificity is essential. In this article, we highlight recent developments in the field regarding improvements in the © 2017 American Chemical Society
■
RECENT STRATEGIES FOR IMPROVING SPECIFICITY OF THE CRISPR/CAS9 SYSTEM There have been many efforts for increasing CRISPR’s specificity including rational Cas9 engineering,5,6 Cas9 directed evolution,7,8 discovery and use of more specific Cas9 variants,9−12 creating obligate heterodimers by using Cas9 nickase or dCas9-FokI fusion proteins,13,14 and modification of the sgRNA by truncation or changing the scaffold sequence15,16 (Figure 1). Another approach is limiting the time of CRISPR/ Cas9 activity by transfecting synthetic or in vitro-transcribed sgRNA together with Cas9 protein or mRNA, instead of their expression from a plasmid.17 Inducible Cas9 variants can also be used for controlling the exposure time of the cells to the Cas9 activity.18−21 Another promising recent finding was antiCRISPR proteins, which could decrease off-target effects without significantly affecting on-target efficiency.22 It is hard to claim complete lack of off-target effects for any of these methods and the goal should be continuous refinement of the CRISPR specificity. For more efficient rational engineering, Received: July 31, 2017 Published: September 15, 2017 1609
DOI: 10.1021/acssynbio.7b00270 ACS Synth. Biol. 2017, 6, 1609−1613
Viewpoint
ACS Synthetic Biology
Figure 1. Strategies for improving CRISPR/Cas9 specificity. (A) Summary of strategies for improving specificity via use of novel Cas9 variants. In one example, directed evolution is utilized to obtain Cas9 variants recognizing longer and more unique alternative PAM sequences. In another example, Cas9 is engineered by introducing rational mutations based on the Cas9 crystal structure and its known mechanism of action. SpCas9-HF1 and eSpCas9 are two such variants obtained by rational protein engineering (corresponding mutations are shown in orange and teal, respectively, on the Cas9 protein map). Cas9-FokI fusion proteins and Cas9 nickase are other strategies that can improve specificity since they require target recognition by two sgRNAs simultaneously. Inducible Cas9 variants can be used for increasing specificity by limiting Cas9 activity. Some Cas9 orthologs (SaCas9, StCas9 and NmCas9) and CRISPR proteins were also discovered that could have high target specificity. (B) Modifications to the sgRNA could be another approach to increase specificity. Truncated spacers with 17−18 nt sequence can be more specific since they may have less tolerance to mismatches. Although not explored well, modifications to functional sequences within the scaffold can also improve specificity.
a complete understanding of the CRISPR/Cas9 system would be helpful. While a great deal of information has been gained from recent studies of the structural, biochemical and mechanistic features of CRISPR/Cas9, questions still remain about the conformational changes associated with Cas9 transitioning from the apo state to the holo form. Specifically, details regarding the structural transitions of the catalytic HNH domain remain unclear. Furthermore, while higher fidelity variants of Cas9 have been engineered, their mechanism of ontarget/off-target discrimination was incompletely understood, limiting the potential for further advances. Hence, more efforts for the full characterization of the CRISPR/Cas9 system are necessary and can help designing better strategies to decrease off-target effects without compromising on-target activity.
editing practices. Factors that might affect the reported specificity results, such as Cas9 amount, delivery method, the controls used, and the cell line and locus tested, should be standardized and any deviations from the standard protocols should be reported for an objective and strict comparison of various tools to improve CRISPR specificity. A careful analysis of the strategies to improve CRISPR specificity can also be helpful to define best practices for preclinical studies using CRISPR. It is also possible that the effect of these methods on specificity might be additive, which should be examined. As more strategies to improve the specificity are developed, they can be incorporated into the agreed-upon protocols for further refinement. Researchers working on the CRISPR-based therapeutic strategies should use the best practices available to report specificity data since their results have a direct impact on the clinical translation of the CRISPR technology. Using suboptimal conditions would subsequently lead to lower specificity, which would underestimate the current advances in the field and be misleading to the community.
■
COMPARING THE METHODS TO IMPROVE SPECIFICITY REQUIRES STANDARDIZATION In order to determine the best available method for improving specificity, the next step should be a side-by-side comparison of the above-mentioned strategies. A direct comparison of the data across laboratories could be more practical; however, a true comparison would necessitate standardization in the genome 1610
DOI: 10.1021/acssynbio.7b00270 ACS Synth. Biol. 2017, 6, 1609−1613
Viewpoint
ACS Synthetic Biology
Figure 2. Schematic of the novel strategies for unbiased genome-wide off-target detection. In IDLV capture method, IDLV double-stranded DNAs (dsDNAs) are integrated into CRISPR/Cas9-induced double strand breaks (DSBs) inside the cells. IDLV integration sites are determined by linear amplification-mediated PCR (LAM-PCR), followed by high-throughput sequencing. In GUIDE-Seq, double-stranded oligodeoxynucleotides (dsODNs) are captured at DSB sites. Integration sites are determined by tag-specific PCR, followed by high-throughput sequencing. HTGST requires an additional DSB at a “bait” site. Translocations between the “prey” and “bait” are detected by LAM-PCR with a biotinylated primer that anneals to the bait, streptavidin enrichment and high-throughput sequencing. In BLESS, nuclei are isolated from cells transfected with CRISPR/ Cas9, followed by permeabilization and adaptor ligation. After streptavidin enrichment, high-throughput sequencing is performed. In Digenome-Seq, purified genomic DNA (gDNA) is treated with Cas9 in vitro and after adaptor ligation, whole genome sequencing is performed. In CIRCLE-Seq, gDNA is sheared and circularized, followed by degrading any remaining linear DNA. After in vitro CRISPR/Cas9 treatment, circular DNAs with a CRISPR/Cas9 recognition site will be cleaved, linearizing the DNA. Cleaved DNA ends can be ligated with an adaptor, followed by paired-end highthroughput sequencing.
■
ACCURACY AND SENSITIVITY OF THE METHODS FOR DETERMINING OFF-TARGET MUTATIONS Development of accurate and comprehensive off-target detection methods has been another focus in the CRISPR/ Cas9 field. The earliest examples were based on evaluation of top predicted off-target sites, which is not comprehensive. Recently various methods to identify genome-wide off-target events in an unbiased manner were developed (Figure 2). In the integrase-defective lentiviral vector (IDLV) capture method, DSB sites were labeled by integration of the doublestranded IDLV DNA and these sites of integration were determined by high-throughput sequencing.23 However, this method suffers from low integration efficiency of IDLVs and potential false-positives due to random integration. GUIDE-Seq (genome-wide unbiased identification of DSBs enabled by sequencing) was a similar but a more sensitive approach that was based on capturing a small double-stranded oligodeoxynucleotide (dsODN) barcode instead of IDLV.24 Although GUIDE-Seq and IDLVs can also capture CRISPR-independent DSBs in the genome, use of proper controls, such as IDLV transduction or dsODN transfection only, can help identify such background DSBs. A more significant issue with GUIDESeq is that it is technically difficult to adapt it for in vivo experiments since it requires codelivery of dsODN. HTGTS
(high-throughput, genome-wide translocation sequencing) is a method that could be compatible with in vivo studies since it relies on detecting translocations induced by CRISPR/Cas9 and does not require additional components.25 However, translocations as a result of gene editing are very rare events which could be difficult to detect. BLESS (breaks labeling, enrichment on streptavidin and next-generation sequencing) is another method that was used for detecting CRISPR-induced DSBs in tissues in vivo.9 However, BLESS only gives a snapshot of DSBs created at the moment of analysis and it cannot detect DSBs already repaired. For in vivo analyses, whole-genome sequencing (WGS) has been a preferred method. Techniques directly capturing DSBs, such as GUIDE-Seq or BLESS, can be more powerful than WGS since WGS can also capture naturally occurring variations between controls and edited cells/animal models, which can complicate the analysis and be misleading. For WGS analysis, choosing proper controls can be especially crucial for a correct interpretation. Minimizing false-positive and negative rates for WGS would require analyzing many clones or animals, which is tedious and costly. There are also methods performed on purified DNA such as Digenome-seq (in vitro Cas9-digested whole-genome sequencing)26 or the recently developed CIRCLE-Seq (circularization 1611
DOI: 10.1021/acssynbio.7b00270 ACS Synth. Biol. 2017, 6, 1609−1613
ACS Synthetic Biology
Viewpoint
■
for in vitro reporting of cleavage effects by sequencing).27 Cellbased assays might have advantages over such cell-free assays since local chromatin structure might affect cutting efficiency and an off-target site detected in a cell-free setup might not be accessible within the natural chromatin environment, hence it might not be an actual threat to the cell. However, it is necessary to remember that cell-based assays might not represent true frequencies of off-target mutations either, due to cell fitness effects of some mutations. A combination of in vitro and cell-based strategies might be more informative.
CONCLUSIONS Genome editing tools hold great promise for the future, enabling rapid development of novel technologies that used to be far from reality. As we are already starting to discuss its potential use in humans, it is important to point out the downsides of CRISPR/Cas9, off-target activity being an important one. With an increasing body of knowledge, great progress has been accomplished to increase the specificity of CRISPR/Cas9, although it still needs improvement. It is challenging to decide what is a “tolerable” off-target activity and the tolerance level would depend on the application. For basic science experiments using model organisms or cell lines, offtarget activity might be tolerable. However, for therapeutic applications, technically even a single cell with an oncogenic mutation might cause a risk to the patients and it is hard to decide what off-target mutation frequency is acceptable. Perhaps, an analysis of the potential phenotypic outcomes of the detected off-target mutations can be more informative. Achieving safe therapeutic uses of CRISPR will require continuous improvement of the specificity and the tools for offtarget detection. Ideally, a method which can directly detect CRISPR-induced DSBs and is also applicable to in vivo models would be necessary. Techniques such as GUIDE-Seq and CIRCLE-Seq can detect mutations with 0.1% or less frequency but a continuous effort to improve off-target detection sensitivity and accuracy is necessary for clinical applications. For accurate prediction of CRISPR/Cas9 activity to save time and effort, pooling and careful analysis of on- and offtarget activity data can help training predictive algorithms. To collect consistent and reliable data across different laboratories for such an extensive study, standardization of protocols and data analysis tools would be necessary. For therapeutic applications that require precise gene correction through homologous recombination, preventing random integration of donor DNAs is another important area to be studied.
■
DATA STANDARDIZATION AND POOLING TO IMPROVE COMPUTATIONAL PREDICTION OF OFF-TARGET SITES Although the tools for unbiased detection of genome-wide offtargets are an ideal quality-control, such high-throughput detection methods might not be applicable to every laboratory due to the required expertise and costs associated with the protocols. Especially if an sgRNA will only be used for research purposes, a more ideal case would be using a highly accurate program for off-target site prediction to enable designing sgRNAs with high efficiency and specificity. However, predicting off-target sites might be difficult since in some cases up to seven mismatches, alternative PAMs or small gaps and bulges were tolerated.24 Although some of the current prediction tools can utilize these parameters, it is still not possible to accurately predict which one of these off-target sites are actually cleaved. Training predictive algorithms to their optimum would require a co-operative effort in the CRISPR field by collecting and pooling specificity analysis data in many cell lines and/or animal models using a set of different sgRNAs. Such a detailed study could also benefit from standardization in the CRISPR field and establishment of a genome-editing consortium can accelerate development of standards. If we interpret the specificity data regarding the cell line used for analysis, this examination could potentially identify novel parameters regulating off-target effects. For instance, cells from different origins will have different epigenetic status and as a result, chromatin sites that are accessible to CRISPR/Cas9 might vary. What would be interesting is a comparison of predicted and actual off-target regions as a function of local chromatin marks, DNase hypersensitivity, transcriptional status or subnuclear localization. If any of these traits are found to have an effect on Cas9 cutting efficiency, the regarding data sets can be included in predictive algorithms.
■
AUTHOR INFORMATION
Corresponding Author
*E-mail:
[email protected]. Phone: (217) 333-2631. Fax: (217) 333-5052. ORCID
Huimin Zhao: 0000-0002-9069-6739 Notes
The authors declare no competing financial interest.
■
■
ACKNOWLEDGMENTS We gratefully acknowledge financial support from the U.S. National Institutes of Health (1U54DK107965) and Department of Energy (ER65474) (HZ) for development of new genome engineering tools.
TARGETING SPECIFICITY OF DONOR DNA It is important to remember that many therapeutic applications require gene correction rather than gene knockout. In such cases use of a donor DNA is necessary. For such gene editing applications, another potential threat that can cause off-target modifications is the random integration of the donor. Thus, even if we maximize the specificity of the CRISPR/Cas9 system, many therapeutic applications can still be hampered by insertional mutagenesis created by the donor. Thus, knock-in specificity is another component to be optimized. Novel donor vectors, such as adenoviral donors with protein-protected DNA ends, can reduce off-target insertions by potentially preventing direct ligation of the donor to the genome.28 Inspired by adenoviral donors, future engineering of novel donors with chemically modified ends can be a safer alternative than a virally delivered donor.
■
REFERENCES
(1) Schaefer, K. A., Wu, W. H., Colgan, D. F., Tsang, S. H., Bassuk, A. G., and Mahajan, V. B. (2017) Unexpected mutations after CRISPRCas9 editing in vivo. Nat. Methods 14, 547−548. (2) Kim, S. T., Park, J., Kim, D., Kim, K., Bae, S., Schlesner, M., and Kim, J. S. (2017) Questioning unexpected CRISPR off-target mutations in vivo. bioRxiv, 157925. (3) Lareau, C. A., Clement, K., Hsu, J. Y., Pattanayak, V., Joung, J. K., Aryee, M. J., and Pinello, L. (2017) “Unexpected mutations after CRISPR-Cas9 editing in vivo” are most likely pre-existing sequence variants and not nuclease-induced mutations. bioRxiv, 159707.
1612
DOI: 10.1021/acssynbio.7b00270 ACS Synth. Biol. 2017, 6, 1609−1613
Viewpoint
ACS Synthetic Biology (4) Wilson, C. J., Fennell, T., Bothmer, A., Maeder, M. L., Reyon, D., Cotta-Ramusino, C., Fernandez, C. A., Marco, E., Barrera, L. A., Jayaram, H., Albright, C. F., Cox, G. F., Church, G. M., and Myer, V. E. (2017) The experimental design and data interpretation in “Unexpected mutations after CRISPR Cas9 editing in vivo” by Schaefer et al. are insufficient to support the conclusions drawn by the authors. bioRxiv, 153338. (5) Kleinstiver, B. P., Pattanayak, V., Prew, M. S., Tsai, S. Q., Nguyen, N. T., Zheng, Z., and Joung, J. K. (2016) High-fidelity CRISPR−Cas9 nucleases with no detectable genome-wide off-target effects. Nature 529, 490−495. (6) Slaymaker, I. M., Gao, L., Zetsche, B., Scott, D. A., Yan, W. X., and Zhang, F. (2016) Rationally engineered Cas9 nucleases with improved specificity. Science 351, 84−88. (7) Kleinstiver, B. P., Prew, M. S., Tsai, S. Q., Nguyen, N. T., Topkar, V. V., Zheng, Z., and Joung, J. K. (2015) Broadening the targeting range of Staphylococcus aureus CRISPR-Cas9 by modifying PAM recognition. Nat. Biotechnol. 33, 1293−1298. (8) Kleinstiver, B. P., Prew, M. S., Tsai, S. Q., Topkar, V. V., Nguyen, N. T., Zheng, Z., Gonzales, A. P., Li, Z., Peterson, R. T., Yeh, J. R., Aryee, M. J., and Joung, J. K. (2015) Engineered CRISPR-Cas9 nucleases with altered PAM specificities. Nature 523, 481−485. (9) Ran, F. A., Cong, L., Yan, W. X., Scott, D. A., Gootenberg, J. S., Kriz, A. J., Zetsche, B., Shalem, O., Wu, X., Makarova, K. S., Koonin, E. V., Sharp, P. A., and Zhang, F. (2015) In vivo genome editing using Staphylococcus aureus Cas9. Nature 520, 186−191. (10) Zetsche, B., Gootenberg, J. S., Abudayyeh, O. O., Slaymaker, I. M., Makarova, K. S., Essletzbichler, P., Volz, S. E., Joung, J., van der Oost, J., Regev, A., Koonin, E. V., and Zhang, F. (2015) Cpf1 is a single RNA-guided endonuclease of a class 2 CRISPR-Cas system. Cell 163, 759−771. (11) Lee, C. M., Cradick, T. J., and Bao, G. (2016) The Neisseria meningitidis CRISPR-Cas9 system enables specific genome editing in mammalian cells. Mol. Ther. 24, 645−654. (12) Müller, M., Lee, C. M., Gasiunas, G., Davis, T. H., Cradick, T. J., Siksnys, V., Bao, G., Cathomen, T., and Mussolino, C. (2016) Streptococcus thermophilus CRISPR-Cas9 systems enable specific editing of the human genome. Mol. Ther. 24, 636−644. (13) Cho, S. W., Kim, S., Kim, Y., Kweon, J., Kim, H. S., Bae, S., and Kim, J. S. (2014) Analysis of off-target effects of CRISPR/Cas-derived RNA-guided endonucleases and nickases. Genome Res. 24, 132−141. (14) Guilinger, J. P., Thompson, D. B., and Liu, D. R. (2014) Fusion of catalytically inactive Cas9 to FokI nuclease improves the specificity of genome modification. Nat. Biotechnol. 32, 577−582. (15) Fu, Y., Sander, J. D., Reyon, D., Cascio, V. M., and Joung, J. K. (2014) Improving CRISPR-Cas nuclease specificity using truncated guide RNAs. Nat. Biotechnol. 32, 279−284. (16) Chen, B., Gilbert, L. A., Cimini, B. A., Schnitzbauer, J., Zhang, W., Li, G. W., Park, J., Blackburn, E. H., Weissman, J. S., Qi, L. S., and Huang, B. (2013) Dynamic imaging of genomic loci in living human cells by an optimized CRISPR/Cas system. Cell 155, 1479−1491. (17) Liang, X., Potter, J., Kumar, S., Zou, Y., Quintanilla, R., Sridharan, M., Carte, J., Chen, W., Roark, N., Ranganathan, S., Ravinder, N., and Chesnut, J. D. (2015) Rapid and highly efficient mammalian cell engineering via Cas9 protein transfection. J. Biotechnol. 208, 44−53. (18) Wright, A. V., Sternberg, S. H., Taylor, D. W., Staahl, B. T., Bardales, J. A., Kornfeld, J. E., and Doudna, J. A. (2015) Rational design of a split-Cas9 enzyme complex. Proc. Natl. Acad. Sci. U. S. A. 112, 2984−2989. (19) Zetsche, B., Volz, S. E., and Zhang, F. (2015) A split-Cas9 architecture for inducible genome editing and transcription modulation. Nat. Biotechnol. 33, 139−142. (20) Nihongaki, Y., Kawano, F., Nakajima, T., and Sato, M. (2015) Photoactivatable CRISPR-Cas9 for optogenetic genome editing. Nat. Biotechnol. 33, 755−760. (21) Davis, K. M., Pattanayak, V., Thompson, D. B., Zuris, J. A., and Liu, D. R. (2015) Small molecule-triggered Cas9 protein with improved genome-editing specificity. Nat. Chem. Biol. 11, 316−318.
(22) Shin, J., Jiang, F., Liu, J. J., Bray, N. L., Rauch, B. J., Baik, S. H., Nogales, E., Bondy-Denomy, J., Corn, J. E., and Doudna, J. A. (2017) Disabling Cas9 by an anti-CRISPR DNA mimic. Sci. Adv. 3, e1701620. (23) Wang, X., Wang, Y., Wu, X., Wang, J., Wang, Y., Qiu, Z., Chang, T., Huang, H., Lin, R. J., and Yee, J. K. (2015) Unbiased detection of off-target cleavage by CRISPR-Cas9 and TALENs using integrasedefective lentiviral vectors. Nat. Biotechnol. 33, 175−178. (24) Tsai, S. Q., Zheng, Z., Nguyen, N. T., Liebers, M., Topkar, V. V., Thapar, V., Wyvekens, N., Khayter, C., Iafrate, A. J., Le, L. P., Aryee, M. J., and Joung, J. K. (2015) GUIDE-seq enables genome-wide profiling of off-target cleavage by CRISPR-Cas nucleases. Nat. Biotechnol. 33, 187−197. (25) Frock, R. L., Hu, J., Meyers, R. M., Ho, Y. J., Kii, E., and Alt, F. W. (2015) Genome-wide detection of DNA double-stranded breaks induced by engineered nucleases. Nat. Biotechnol. 33, 179−186. (26) Kim, D., Bae, S., Park, J., Kim, E., Kim, S., Yu, H. R., Hwang, J., Kim, J. I., and Kim, J. S. (2015) Digenome-seq: genome-wide profiling of CRISPR-Cas9 off-target effects in human cells. Nat. Methods 12, 237−243. (27) Tsai, S. Q., Nguyen, N. T., Malagon-Lopez, J., Topkar, V. V., Aryee, M. J., and Joung, J. K. (2017) CIRCLE-seq: a highly sensitive in vitro screen for genome-wide CRISPR-Cas9 nuclease off-targets. Nat. Methods 14, 607−614. (28) Holkers, M., Maggio, I., Henriques, S. F., Janssen, J. M., Cathomen, T., and Goncalves, M. A. (2014) Adenoviral vector DNA for accurate genome editing with engineered nucleases. Nat. Methods 11, 1051−1057.
1613
DOI: 10.1021/acssynbio.7b00270 ACS Synth. Biol. 2017, 6, 1609−1613
