The ability to direct a nuclease to cleave specific sites in the genome, thus triggering gene-altering DNA repair pathways, is being exploited in laboratories around the world to modify the genomes of an ever-expanding list of cells and organisms1.

Among the various types of targeted nucleases, three have been widely utilised for genome engineering approaches, and one has recently entered clinical testing2. Two of these technologies, Zinc Finger Nucleases (ZFNs)3 and Transcriptional Activator-Like Effector Nucleases (TALENS)4, consist of engineered arrays of sequence-specific DNA binding repeats fused to an endonuclease domain derived from the FokI restriction enzyme. The third, and most recently described, technology is derived from the prokaryotic Clustered Regularly Interspersed Short Palindromic Repeat (CRISPR)-CRISPR associated (Cas) system.

Found in bacteria and archaea, CRISPR/Cas functions as an adaptive immune system in which an endonuclease (encoded by the Cas9 gene in the most widely applied Type II CRISPR systems) associates with small RNAs derived from CRISPR transcripts. The small RNAs direct Cas9 to complementary DNA sequences, typically derived from bacteriophages or plasmids, which are then cleaved by the two distinct Cas9 nuclease domains5. To be cleaved, a target site must not only be complementary to the CRISPR RNA, but must also be positioned adjacent to a short sequence motif (the Protospacer Adjacent Motif [PAM]) that is recognised by the particular Cas9. In 2013 it was discovered that a simplified version of the CRISPR/Cas system consisting of a Cas9 protein and a single RNA molecule (referred to as a guide RNA [gRNA]) could be used to induce targeted nucleolytic cleavage of endogenous genomic sites in mammalian cells6-8. The relative simplicity of programming the Cas9 nuclease with short gRNAs targeted to individual or multiple loci has led to its rapid adoption as the method of choice for genome editing in the academic research community9, and further enhances the prospects for expanded clinical application of genome editing technology.

Different types of therapeutic editing
By exploiting endogenous cellular DNA repair pathways following targeted nucleolytic cleavage, it is possible to engineer a variety of genomic alterations in a site-specific manner (Figure 1). DNA breaks are generally repaired via one of two major pathways referred to as non-homologous end-joining (NHEJ) and homology-directed repair (HDR)10. In NHEJ, double-strand breaks (DSBs) are repaired by relegation of the cleaved ends without the involvement of any additional donor or template DNA11. This repair pathway is errorprone resulting in insertions and/or deletions (indels) of various sizes at the site of the DSB. Thus, the most straightforward form of gene editing relies upon indel-mediated disruption of gene function; for example, the introduction of frameshift mutations in the coding sequence of a gene. Alternatively, HDR-mediated mechanisms afford the opportunity to edit genomic sites in a more precise manner by utilising donor DNA as a template for repair12. In its natural setting during DNA replication, the sister chromatid serves as the donor. However, it has been shown that exogenously introduced DNA can function as a donor template, especially in the context of an induced DSB13, allowing the replacement of endogenous nucleotides with any desired sequence. Using this approach, a mutant gene can be converted to its wild-type counterpart.

The simplest use of genome editing is the NHEJmediated introduction of indels to alter the function of targeted genes. When the nuclease target site is located in the coding region of a gene, many studies have shown that the resulting indels can lead to frame-shift mutations disrupting protein expression. This was the approach used in the first and, to date, only reported clinical trial of genome editing technology comprising ZFN-mediated disruption of the CCR5 gene in T cells obtained from HIV-positive individuals. Following infusion back into the patients, evidence for resistance of the altered cells to HIV infection was obtained2. This ground-breaking study bolsters the rationale for further investigation of genome editing to induce clinically beneficial gene mutations in well validated targets. For example, it was shown that adenoviral delivery of CRISPR/Cas components targeting the mouse Pcsk9 gene, a well-established regulator of lipoprotein levels, led to significant rates of gene mutation in the liver and concomitant decreases in plasma cholesterol14. An intriguing application of the NHEJ-mediated editing approach was recently demonstrated in cells harbouring a disease-causing frame-shift mutation in the DMD gene which results in a premature stop codon. TALEN-induced indels were shown to restore the correct reading frame leading to production of a functional version of the DMD protein in some cells15. Evidence that this modality can be extended to infectious agents comes from a recent study demonstrating CRISPR/Cas targeting of the HPV genome in virally transformed cells16.

Another application of NHEJ-based genome modification is the use of two separate DSBs to induce deletion of the intervening sequence. Recently, it has been shown that simultaneous introduction of two CRISPR/Cas9-induced breaks can give rise to genomic deletions up to several megabases in size17. A therapeutically relevant application of this approach using TALENs was the demonstration that deletion of the 10kb Bcl11A erythroid-specific enhancer element can be achieved in tissue culture cells; an approach that has been proposed for the treatment of hemoglobinopathies18. Introduction of dual targeted DSBs can also lead to chromosomal rearrangements including inversions, duplications and translocations, presenting a potential safety issue for this approach19,20.

Successful application of the more precise HDRmediated form of genome editing to modify targets of therapeutic importance has been achieved mainly in tissue culture cells, including human iPS cells. For example, ZFNs introduced together with donor template DNA into patient-derived iPS cells repaired the sickle cell mutation in the globin gene21 as well as the mutation in the SERPINA1 gene responsible for alpha-1-antitrypsin deficiency22. Similarly, -thalassemia mutations were corrected using CRISPR/Cas plus donor DNA in iPS cells of patient origin23. In intestinal stem cells derived from cystic fibrosis patients, the diseasecausing mutation in the CFTR gene was repaired using CRISPR/Cas components and shown to restore functionality in organoid cultures24.

A variation on the theme of HDR-mediated gene repair is the use of targeted nucleases and DNA donors to effect gene insertion at a precise genomic site; either the endogenous locus or a ‘safe harbour’ site known to support high level gene expression. In this approach, the donor template contains the desired gene insert (eg, a full-length cDNA) flanked by homology arms matching the targeted nuclease cleavage site. In one example, phenotypic correction in a Hemophilia B mouse model was achieved by ZFN-mediated insertion of a fulllength human FIX cDNA sequence25.

In these as well as other published studies, it has been shown that a variety of different types of DNA templates can serve as effective HDR donors. Following early work using plasmid donors with long homology arms, recent studies have shown that single stranded oligonucleotides (ssODNs), with as little as 80 base pairs of homology flanking the alteration site, can serve as efficient donor templates26,27. Viral vectors, such as integrase-deficient lentivirus (IDLV) or adeno-associated virus (AAV), have also proven to be effective sources of donor DNA28,29. The naturally recombinogenic nature of AAV may make it especially well-suited for this purpose30. Still, it remains a challenge to obtain high levels of HDR-mediated genome editing with any of the three nuclease platforms. Optimisation is required before broad clinical applicability can be achieved.

Specificity: A pre-requisite for broad clinical application
The safe application of gene editing as a therapeutic modality ultimately depends on the ability to induce DNA cleavage at a specific target site while minimising off-target activity throughout the genome. Specificity of ZFNs and TALENs has been enhanced by increasing the lengths of the DNA binding domains, as well as by modifying the FokI nuclease dimerisation domain to ensure that it could only function when heterodimerised on the DNA target31. For CRISPR/Cas technology, several significant improvements have gone a long way towards alleviating initial concerns over the specificity of the target cleavage. These include modifications to the Cas9 protein as well as the gRNA. In one improvement, Cas9 is converted to a nickase, capable of cleaving only a single strand of DNA by inactivating either the RuvC or HNH nuclease domain. Combined with a pair of gRNAs, targeted in a specific orientation and spacing, Cas9 nickases can stimulate efficient indel formation at rates comparable to those achieved with a single gRNA and the Cas9 nuclease32,33. Because a single nick is repaired with high fidelity, this dual-nickase approach dramatically increases specificity by requiring two gRNA sites, in close proximity to each other, in order to efficiently induce mutagenesis. As was previously demonstrated with zinc finger nickases, induction of a single strand break through the use of a Cas9 nickase and a single gRNA can induce homologous recombination, albeit at lower rates than a nuclease, while reducing both off-target effects and unwanted indel mutations at the target site34,35. Cas9 nickases have been engineered for both Streptococcus thermophilus and Streptococcus pyogenes Cas9s and sequence homology of the RuvC and HNH domains across species36-38 enables adaptation of this method to a diverse range of CRISPR/Cas systems.

Analogous to ZFNs and TALENs it is possible to fuse the cleavage domain of the FokI endonuclease to a catalytically inactive Cas9 (dCas9) to generate a system in which DNA target specificity is dictated by the gRNA-Cas9 complex, but nuclease activity derives from FokI39,40. Because FokI functions as a dimer, the resulting dCas9-FokI fusion protein requires a pair of gRNAs to come together in a strict orientation and spacing in order to generate a double-strand break. This has been shown to dramatically increase specificity, similar to the dual nickase approach, by doubling the target length required for functional nuclease activity. Additionally, it has been suggested that this approach may prove even more specific because while nicks are occasionally repaired with mutagenic effects, FokI is incapable of acting as a monomer and therefore is completely inactive at off-target sites of either single gRNA.

Unlike the two approaches described above which increase specificity by doubling the overall target length, a third approach counter-intuitively found that specificity could be dramatically improved by decreasing the length of a single gRNA41. S. pyogenes truncated gRNAs (trugRNAs) with 17 or 18 nucleotides complementary to the DNA target site, as opposed to the usual 20, showed increased sensitivity to mismatches between the gRNA and target DNA and therefore significantly reduced off-target activity. While this approach can likely be applied to many CRISPR/Cas systems, it remains to be seen whether the optimal length of the truncated gRNAs is consistent or whether it must be empirically determined for each species of Cas9.

The strict requirement for a PAM sequence, recognised by the Cas protein itself, imparts an additional layer of specificity on top of that provided by the gRNA. Cas9 species with longer, more restrictive PAM sequences, such as that of Neisseria meningitides, which recognises a 5’- NNNNGATT-3’ PAM vs 5’-NGG-3’ for S. pyogenes, may therefore be more specific38,42,43. It has been demonstrated that swapping the PAMinteracting domains between Cas9 proteins can alter their PAM specificity44 and it is possible that further structure-function studies and Cas9 protein engineering may further increase the specificity of the CRISPR/Cas system.

Specificity studies to date have largely relied on a priori identification of closely-matched sites in the genome and subsequent sequencing of those off-targets to determine the presence of unintended mutagenesis. Unbiased approaches to identify offtarget sites of ZFNs and TALENs, either by in vitro screening of large target site libraries45,46 or by using uptake of integrase-deficient lenti virus into the genome to mark the site of double strand breaks47 has met with limited success, presumably due to the low sensitivity of these methods48. Whole genome sequencing of clonal populations of modified cells has found few off-target effects of CRISPR/Cas gene editing49-51; however, these methods would be unlikely to detect off-target mutagenesis occurring at rates significantly lower than the on-target modification rate52. A key focus of the gene editing field is the development of an unbiased, genome-wide method for identifying offtarget sites. The ability to detect all sites of nuclease- induced cleavage with high sensitivity will enable selection of nucleases with the most favourable off-target spectrum for therapeutic applications as well as further engineering to increase overall specificity of the system.

Delivering the promise
Therapeutic genome editing approaches can be divided into two categories: 1) ex vivo editing of stem or progenitor cells which are delivered into the patient, and 2) direct in vivo administration, either locally or systemically, of gene editing components into the patient (Figure 2). Given the success of genome modification achieved in a range of stem, progenitor and primary cells, as well as progress of the ongoing HIV trial, it is likely that the former approach will continue to represent the ‘low hanging fruit’ for this technology. While published proof-of-concept studies using the second approach (direct delivery of genome editing components) in animal disease models are few in number, they provide reasons to be optimistic that, with the appropriate delivery technology, these types of therapeutics can be developed. In addition to the previously mentioned FIX25 and Pcsk914 studies, Yin et al recently demonstrated in vivo correction of the causative Fah mutation in a murine model of tyrosinemia. While this study did not use a clinically- relevant delivery method (hydrodynamic injection), another recent report demonstrated in vivo modification of a mouse liver gene by AAV-mediated delivery of CRISPR/Cas components53.

Genome editing technology affords the potential to develop novel therapeutics that could transform the lives of patients suffering from a myriad of diseases. The extraordinary progress made over the past few years to improve efficacy, specificity and delivery across the major genome editing platforms provides hope that the true promise of this groundbreaking technology can be realised.

Dr Morgan Maeder is a Research Scientist at Editas Medicine, Inc. She has extensive experience with engineering targeted nucleases, including ZFNs, TALENs and the CRISPR/Cas system, and applying them in human cells for genetic and epigenetic engineering. Following the completion of her PhD in the laboratory of Dr Keith Joung (Harvard) she joined Editas, a Cambridge, MA-based company focused on the development of novel therapeutics based on genome editing technology.

Dr David Bumcrot is Senior Director of Molecular & Cell Biology at Editas Medicine, Inc. During his 18-year career he has worked at a number of innovative biotechnology companies developing novel therapeutic technologies. Prior to Editas, he was a Director of Research at Alnylam Pharmaceuticals where he worked to advance several novel RNAibased drugs into clinical testing.

Hsu, PD, Lander, ES and Zhang, F. Development and applications of CRISPR-Cas9 for genome engineering. Cell 157, 1262-1278, doi:10.1016/j.cell.2014.05.010 (2014).

2 Tebas, P et al. Gene editing of CCR5 in autologous CD4 T cells of persons infected with HIV. The New England journal of medicine 370, 901-910, doi:10.1056/NEJMoa1300662 (2014).

3 Carroll, D. Genome engineering with zinc-finger nucleases. Genetics 188, 773-782, doi:10.1534/genetics.111.131433 (2011).

4 Christian, M et al. Targeting DNA double-strand breaks with TAL effector nucleases. Genetics 186, 757-761, doi:10.1534/genetics.110.120717 (2010).

5 Jinek, M et al. A programmable dual-RNA-guided DNA endonuclease in adaptive bacterial immunity. Science 337, 816-821, doi:10.1126/science.1225829 (2012).

6 Cong, L et al. Multiplex genome engineering using CRISPR/Cas systems. Science 339, 819-823, doi:10.1126/science.1231143 (2013).

7 Mali, P et al. RNA-guided human genome engineering via Cas9. Science 339, 823-826, doi:10.1126/science.1232033 (2013).

8 Jinek, M et al. RNA-programmed genome editing in human cells. eLife 2, e00471, doi:10.7554/eLife.00471 (2013).

9 Pennisi, E. The CRISPR craze. Science 341, 833-836, doi:10.1126/science.341.6148.833 (2013).

10 Takata, M et al. Homologous recombination and nonhomologous end-joining pathways of DNA doublestrand break repair have overlapping roles in the maintenance of chromosomal integrity in vertebrate cells. The EMBO journal 17, 5497-5508, doi:10.1093/emboj/17.18.5497 (1998).

11 Lieber, MR, Ma, Y, Pannicke, U and Schwarz, K. Mechanism and regulation of human non-homologous DNA end-joining. Nature reviews. Molecular cell biology 4, 712-720, doi:10.1038/nrm1202 (2003).

12 Szostak, JW, Orr-Weaver, TL, Rothstein, RJ and Stahl, FW. The double-strand-break repair model for recombination. Cell 33, 25-35 (1983).

13 Rouet, P, Smih, F and Jasin, M. Expression of a sitespecific endonuclease stimulates homologous recombination in mammalian cells. Proceedings of the National Academy of Sciences of the United States of America 91, 6064-6068 (1994).

14 Ding, Q et al. Permanent Alteration of PCSK9 With In Vivo CRISPR-Cas9 Genome Editing. Circulation research 115, 488-492, doi:10.1161/circresaha.115.304351 (2014).

15 Ousterout, DG et al. Reading frame correction by targeted genome editing restores dystrophin expression in cells from Duchenne muscular dystrophy patients. Molecular therapy : the journal of the American Society of Gene Therapy 21, 1718-1726, doi:10.1038/mt.2013.111 (2013). 

16 Kennedy, EM et al. Inactivation of the human papillomavirus E6 or E7 gene in cervical carcinoma cells using a bacterial CRISPR/Cas RNA-guided endonuclease. Journal of virology, doi:10.1128/jvi.01879-14 (2014).

17 Canver, MC et al. Characterization of genomic deletion efficiency mediated by clustered regularly interspaced palindromic repeats (CRISPR)/Cas9 nuclease system in mammalian cells. The Journal of biological chemistry 289, 21312-21324, doi:10.1074/jbc.M114.564625 (2014).

18 Bauer, DE et al. An erythroid enhancer of BCL11A subject to genetic variation determines fetal hemoglobin level. Science 342, 253-257, doi:10.1126/science.1242088 (2013).

19 Brunet, E et al. Chromosomal translocations induced at specified loci in human stem cells. Proceedings of the National Academy of Sciences of the United States of America 106, 10620-10625, doi:10.1073/pnas.0902076106 (2009).

20 Lee, HJ, Kweon, J, Kim, E, Kim, S and Kim, JS. Targeted chromosomal duplications and inversions in the human genome using zinc finger nucleases. Genome research 22, 539-548, doi:10.1101/gr.129635.111 (2012).

21 Sebastiano, V et al. In situ genetic correction of the sickle cell anemia mutation in human induced pluripotent stem cells using engineered zinc finger nucleases. Stem cells (Dayton, Ohio) 29, 1717-1726, doi:10.1002/stem.718 (2011).

22 Yusa, K et al. Targeted gene correction of alpha1-antitrypsin deficiency in induced pluripotent stem cells. Nature 478, 391-394, doi:10.1038/nature10424 (2011).

23 Xie, F et al. Seamless gene correction of beta-thalassemia mutations in patient-specific iPSCs using CRISPR/Cas9 and piggyBac. Genome research 24, 1526-1533, doi:10.1101/gr.173427.114 (2014).

24 Schwank, G et al. Functional repair of CFTR by CRISPR/Cas9 in intestinal stem cell organoids of cystic fibrosis patients. Cell stem cell 13, 653-658, doi:10.1016/j.stem.2013.11.002 (2013).

25 Anguela, XM et al. Robust ZFN-mediated genome editing in adult hemophilic mice. Blood 122, 3283-3287, doi:10.1182/blood-2013-04- 497354 (2013).

26 Chen, F et al. High-frequency genome editing using ssDNA oligonucleotides with zinc-finger nucleases. Nature methods 8, 753-755, doi:10.1038/nmeth.1653 (2011).

27 Bedell, VM et al. In vivo genome editing using a high-efficiency TALEN system. Nature 491, 114-118, doi:10.1038/nature11537 (2012).

28 Hirsch, ML, Green, L, Porteus, MH and Samulski, RJ. Self-complementary AAV mediates gene targeting and enhances endonuclease delivery for double-strand break repair. Gene therapy 17, 1175-1180, doi:10.1038/gt.2010.65 (2010).

29 Handel, EM et al. Versatile and efficient genome editing in human cells by combining zinc-finger nucleases with adeno-associated viral vectors. Human gene therapy 23, 321-329, doi:10.1089/hum.2011.140 (2012).

30 Miller, DG et al. Gene targeting in vivo by adeno-associated virus vectors. Nature biotechnology 24, 1022-1026, doi:10.1038/nbt1231 (2006).

31 Miller, JC et al. An improved zinc-finger nuclease architecture for highly specific genome editing. Nature biotechnology 25, 778-785, doi:10.1038/nbt1319 (2007).

32 Ran, FA et al. Double nicking by RNA-guided CRISPR Cas9 for enhanced genome editing specificity. Cell 154, 1380-1389, doi:10.1016/j.cell.2013.08.021 (2013).

33 Mali, P et al. CAS9 transcriptional activators for target specificity screening and paired nickases for cooperative genome engineering. Nature biotechnology 31, 833-838, doi:10.1038/nbt.2675 (2013).

34 Kim, E et al. Precision genome engineering with programmable DNA-nicking enzymes. Genome research 22, 1327-1333, doi:10.1101/gr.138792.112 (2012).

35 Hsu, PD et al. DNA targeting specificity of RNA-guided Cas9 nucleases. Nature biotechnology 31, 827-832, doi:10.1038/nbt.2647 (2013).

36 Chylinski, K, Le Rhun, A and Charpentier, E. The tracrRNA and Cas9 families of type II CRISPR-Cas immunity systems. RNA biology 10, 726-737, doi:10.4161/rna.24321 (2013).

37 Chylinski, K, Makarova, KS, Charpentier, E and Koonin, EV. Classification and evolution of type II CRISPR-Cas systems. Nucleic acids research 42, 6091-6105, doi:10.1093/nar/gku241 (2014).

38 Fonfara, I et al. Phylogeny of Cas9 determines functional exchangeability of dual- RNA and Cas9 among orthologous type II CRISPR-Cas systems. Nucleic acids research 42, 2577-2590, doi:10.1093/nar/gkt1074 (2014).

39 Guilinger, JP, Thompson, DB and Liu, DR. Fusion of catalytically inactive Cas9 to FokI nuclease improves the specificity of genome modification. Nature biotechnology 32, 577-582, doi:10.1038/nbt.2909 (2014).

40 Tsai, SQ et al. Dimeric CRISPR RNA-guided FokI nucleases for highly specific genome editing. Nature biotechnology 32, 569-576, doi:10.1038/nbt.2908 (2014).

41 Fu, Y, Sander, JD, Reyon, D, Cascio, VM and Joung, JK. Improving CRISPR-Cas nuclease specificity using truncated guide RNAs. Nature biotechnology 32, 279-284, doi:10.1038/nbt.2808 (2014).

42 Esvelt, KM et al. Orthogonal Cas9 proteins for RNA-guided gene regulation and editing. Nature methods 10, 1116-1121, doi:10.1038/nmeth.2681 (2013).

43 Hou, Z et al. Efficient genome engineering in human pluripotent stem cells using Cas9 from Neisseria meningitidis. Proceedings of the National Academy of Sciences of the United States of America 110, 15644-15649, doi:10.1073/pnas.1313587110 (2013).

44 Nishimasu, H et al. Crystal structure of Cas9 in complex with guide RNA and target DNA. Cell 156, 935-949, doi:10.1016/j.cell.2014.02.001 (2014).

45 Pattanayak, V, Ramirez, CL, Joung, JK and Liu, DR. Revealing off-target cleavage specificities of zinc-finger nucleases by in vitro selection. Nature methods 8, 765-770, doi:10.1038/nmeth.1670 (2011).

46 Guilinger, JP et al. Broad specificity profiling of TALENs results in engineered nucleases with improved DNA-cleavage specificity. Nature methods 11, 429-435, doi:10.1038/nmeth.2845 (2014).

47 Gabriel, R et al. An unbiased genome-wide analysis of zinc-finger nuclease specificity. Nature biotechnology 29, 816-823, doi:10.1038/nbt.1948 (2011).

48 Sander, JD et al. In silico abstraction of zinc finger nuclease cleavage profiles reveals an expanded landscape of off-target sites. Nucleic acids research 41, e181, doi:10.1093/nar/gkt716 (2013).

49 Smith, C et al. Whole-genome sequencing analysis reveals high specificity of CRISPR/Cas9 and TALEN-based genome editing in human iPSCs. Cell stem cell 15, 12-13, doi:10.1016/j.stem.2014.06.011 (2014).

50 Suzuki, K et al. Targeted gene correction minimally impacts whole-genome mutational load in human-disease-specific induced pluripotent stem cell clones. Cell stem cell 15, 31-36, doi:10.1016/j.stem.2014.06.016 (2014).

51 Veres, A et al. Low incidence of off-target mutations in individual CRISPR-Cas9 and TALEN targeted human stem cell clones detected by whole-genome sequencing. Cell stem cell 15, 27-30, doi:10.1016/j.stem.2014.04.020 (2014).

52 Tsai, SQ and Joung, JK. What’s changed with genome editing? Cell stem cell 15, 3-4, doi:10.1016/j.stem.2014.06.017 (2014).

53 Senis, E et al. CRISPR/Cas9-mediated genome engineering: An adeno-associated viral (AAV) vector toolbox. Biotechnology journal, doi:10.1002/biot.201400046 (2014).