- Email: [email protected]
CRISPR-Mediated Base Editing without DNA Double-Strand Breaks
Molecular Cell
Technology Preview CRISPR-Mediated Base Editing without DNA Double-Strand Breaks Brian S. Plosky1,* 1Molecular Cell, 50 Hampshire Street, Cambridge, MA 02139, USA *Correspondence: [email protected] http://dx.doi.org/10.1016/j.molcel.2016.05.006
Targeting point mutations using CRISPR/Cas9 so far has required efficient homologous recombination (HR) and donor oligonucleotides. In a recent Nature paper, Komor and colleagues (2016) describe a way to make specific base changes that does not depend on HR or donor DNA and does not involve making double-strand breaks. Genome editing in almost any form (e.g., TALENs, CRISPR, recombineering, etc.) requires making DNA double-strand breaks (DSBs) and involves cellular machinery that repairs DSBs. The outcome can be altered in several ways, including changing the type of cut (blunt versus resected) or providing a template with homology surrounding a specific alteration. While CRISPR-mediated editing has made genome editing easier and more broadly available, there are still issues of low efficiency and undesired editing (Sternberg and Doudna, 2015). In the absence of a donor template, repair of editing sites by nonhomologous end joining (NHEJ) generates insertions or deletions in the DNA sequence known as indels. The presence of a template with homology to both sides of the break can promote the targeted incorporation of a sequence if HR gets to the break before NHEJ. The balance of DSB repair can be tipped toward HR by optimizing how nucleases such as Cas9 cut (Vriend et al., 2016), by modifying the timing of Cas9 delivery (Lin et al., 2014) or expression (Chu et al., 2015; Gutschner et al., 2016), or by inhibiting NHEJ (Maruyama et al., 2015). What if you could precisely edit the genome without introducing DSBs and worrying about the battle between HR and NHEJ? Now, Komor et al. (2016) describe an approach that uses CRISPR/Cas9 to target point mutations without introducing DSBs. The basic concept involves Cas9 fused to a cytosine deaminase. A singleguide RNA (sgRNA) directs Cas9 to the sequence where a base change is needed (perhaps to correct a disease-associated SNP). The sgRNA is bound to the untar-
geted strand, while APOBEC1 can deaminate the targeted strand (Harris et al., 2002). The deaminase acts on any cytosine within reach to generate uracil, which would then lead to a C:G to T:A transition. Using nuclease dead Cas9 (dCas9), it would be possible to generate such mutations without actually nicking the
DNA backbone. However, after several iterations, Komor et al. learned that introducing one nick actually improves efficiency. They started by testing dCas9 fused with one of four different cytosine deaminases and found that rat APOBEC1 fused at the N terminus of Cas9 was the most
Figure 1. A Schematic of the Third-Generation Base Editor, or BE3 The nuclease-deficient dCas9 (in blue) with restored HNH nuclease activity can nick the strand bound to the sgRNA. This nick signals for removal of the unedited strand when the G:U mispair is recognized by mismatch repair. The red ‘‘X’’ indicates the inability to cleave the unbound DNA strand. APOBEC1 (in orange) can edit cytosine to uracil on the unbound strand. UGI blocks the activity of uracil glycosylase to prevent base excision repair and correction of the edit.
Molecular Cell 62, May 19, 2016 ª 2016 Elsevier Inc. 477
Molecular Cell
Technology Preview effective in their in vitro assays. This fusion was named BE1 (first-generation base editor). While the average in vitro editing efficiency was around 44%, in vivo it ranged from 0.7% to 7.8%. To test if cellular uracil DNA glycosylase and base excision repair (BER) were responsible for the relatively error-free repair of uracil, they fused a bacteriophage uracil glycosylase inhibitor to the APOBEC1-dCas9 fusion to generate BE2. While editing efficiency of BE2 is below 5% at some targets, at others, it is as high as 20%. In addition to BER, mismatch repair (MMR) can also correct U:G mismatches. The presence of a nick on one strand is a likely cue for the MMR proteins to discriminate the nascent DNA strand where an incorrect base was inserted from the template strand during DNA replication (Pluciennik et al., 2010). To provide such a strand discrimination signal for their base editor fusions, Komor et al. (2016) restored the nuclease activity of the HNH nuclease domain to generate BE3 (Figure 1). The HNH domain nicks the strand bound by the guide RNA (Jinek et al., 2012). This modification improved editing efficiency up to as high as 37%, and close to 10% for the more refractory sites they tested. So how does this stack up against using donor oligonucleotides and HR to target a mutation to a particular locus? How efficient is the editing, and how often
478 Molecular Cell 62, May 19, 2016
does each approach lead to undesired edits such as indels? How good is this at correcting disease mutations? To answer each of these questions, they compared BE3 to more typical Cas9-mediated editing at two disease-associated loci, APOE4 (Alzheimer’s) and TP53 (cancer). Both variants that they tested are amenable to correction by C-to-T editing to restore the correct amino acid sequence, and editing of any nearby Cs would not change the amino acid sequence. For APOE4 in astrocytes, BE3 introduced the correct sequence in nearly 75% of total sequencing reads compared to Cas9 and a 200 nt ssDNA donor correcting only 0.3%. Additionally, Cas9 with ssDNA yielded 26.1% indels compared to less than 5% for BE3. The numbers are similarly striking with TP53. BE3 corrected 7.6% of sequence in human breast cancer cells and generated less than 1% indels, while Cas9 and ssDNA was unable to generate any of the desired sequence changes and 6.1% of sequences had indels. This work is exciting both for its potential and for its use of our understanding of so much basic knowledge of DNA repair pathways. While a limitation is the type of changes that can be made, when it can be used, it will be much more efficient and accurate relative to using Cas9 and ssDNA to introduce mutations. Each week, it seems that new modifications
of Cas9 are reported that change the genome-editing landscape and expand the options for and improve the precision with which changes can be made. REFERENCES Chu, V.T., Weber, T., Wefers, B., Wurst, W., Sander, S., Rajewsky, K., and Ku¨hn, R. (2015). Nat. Biotechnol. 33, 543–548. Gutschner, T., Haemmerle, M., Genovese, G., Draetta, G.F., and Chin, L. (2016). Cell Rep. 14, 1555–1566. Harris, R.S., Petersen-Mahrt, S.K., and Neuberger, M.S. (2002). Mol. Cell 10, 1247–1253. Jinek, M., Chylinski, K., Fonfara, I., Hauer, M., Doudna, J.A., and Charpentier, E. (2012). Science 337, 816–821. Komor, A.C., Kim, Y.B., Packer, M.S., Zuris, J.A., and Liu, D.R. (2016). Nature. http://dx.doi.org/10. 1038/nature17946. Lin, S., Staahl, B.T., Alla, R.K., and Doudna, J.A. (2014). eLife 3, e04766. Maruyama, T., Dougan, S.K., Truttmann, M.C., Bilate, A.M., Ingram, J.R., and Ploegh, H.L. (2015). Nat. Biotechnol. 33, 538–542. Pluciennik, A., Dzantiev, L., Iyer, R.R., Constantin, N., Kadyrov, F.A., and Modrich, P. (2010). Proc. Natl. Acad. Sci. USA 107, 16066–16071. Sternberg, S.H., and Doudna, J.A. (2015). Mol. Cell 58, 568–574. Vriend, L.E.M., Prakash, R., Chen, C.-C., Vanoli, F., Cavallo, F., Zhang, Y., Jasin, M., and Krawczyk, P.M. (2016). Nucleic Acids Res. Published online March 21, 2016. http://dx.doi.org/10.1093/nar/ gkw179.
Technology Preview CRISPR-Mediated Base Editing without DNA Double-Strand Breaks Brian S. Plosky1,* 1Molecular Cell, 50 Hampshire Street, Cambridge, MA 02139, USA *Correspondence: [email protected] http://dx.doi.org/10.1016/j.molcel.2016.05.006
Targeting point mutations using CRISPR/Cas9 so far has required efficient homologous recombination (HR) and donor oligonucleotides. In a recent Nature paper, Komor and colleagues (2016) describe a way to make specific base changes that does not depend on HR or donor DNA and does not involve making double-strand breaks. Genome editing in almost any form (e.g., TALENs, CRISPR, recombineering, etc.) requires making DNA double-strand breaks (DSBs) and involves cellular machinery that repairs DSBs. The outcome can be altered in several ways, including changing the type of cut (blunt versus resected) or providing a template with homology surrounding a specific alteration. While CRISPR-mediated editing has made genome editing easier and more broadly available, there are still issues of low efficiency and undesired editing (Sternberg and Doudna, 2015). In the absence of a donor template, repair of editing sites by nonhomologous end joining (NHEJ) generates insertions or deletions in the DNA sequence known as indels. The presence of a template with homology to both sides of the break can promote the targeted incorporation of a sequence if HR gets to the break before NHEJ. The balance of DSB repair can be tipped toward HR by optimizing how nucleases such as Cas9 cut (Vriend et al., 2016), by modifying the timing of Cas9 delivery (Lin et al., 2014) or expression (Chu et al., 2015; Gutschner et al., 2016), or by inhibiting NHEJ (Maruyama et al., 2015). What if you could precisely edit the genome without introducing DSBs and worrying about the battle between HR and NHEJ? Now, Komor et al. (2016) describe an approach that uses CRISPR/Cas9 to target point mutations without introducing DSBs. The basic concept involves Cas9 fused to a cytosine deaminase. A singleguide RNA (sgRNA) directs Cas9 to the sequence where a base change is needed (perhaps to correct a disease-associated SNP). The sgRNA is bound to the untar-
geted strand, while APOBEC1 can deaminate the targeted strand (Harris et al., 2002). The deaminase acts on any cytosine within reach to generate uracil, which would then lead to a C:G to T:A transition. Using nuclease dead Cas9 (dCas9), it would be possible to generate such mutations without actually nicking the
DNA backbone. However, after several iterations, Komor et al. learned that introducing one nick actually improves efficiency. They started by testing dCas9 fused with one of four different cytosine deaminases and found that rat APOBEC1 fused at the N terminus of Cas9 was the most
Figure 1. A Schematic of the Third-Generation Base Editor, or BE3 The nuclease-deficient dCas9 (in blue) with restored HNH nuclease activity can nick the strand bound to the sgRNA. This nick signals for removal of the unedited strand when the G:U mispair is recognized by mismatch repair. The red ‘‘X’’ indicates the inability to cleave the unbound DNA strand. APOBEC1 (in orange) can edit cytosine to uracil on the unbound strand. UGI blocks the activity of uracil glycosylase to prevent base excision repair and correction of the edit.
Molecular Cell 62, May 19, 2016 ª 2016 Elsevier Inc. 477
Molecular Cell
Technology Preview effective in their in vitro assays. This fusion was named BE1 (first-generation base editor). While the average in vitro editing efficiency was around 44%, in vivo it ranged from 0.7% to 7.8%. To test if cellular uracil DNA glycosylase and base excision repair (BER) were responsible for the relatively error-free repair of uracil, they fused a bacteriophage uracil glycosylase inhibitor to the APOBEC1-dCas9 fusion to generate BE2. While editing efficiency of BE2 is below 5% at some targets, at others, it is as high as 20%. In addition to BER, mismatch repair (MMR) can also correct U:G mismatches. The presence of a nick on one strand is a likely cue for the MMR proteins to discriminate the nascent DNA strand where an incorrect base was inserted from the template strand during DNA replication (Pluciennik et al., 2010). To provide such a strand discrimination signal for their base editor fusions, Komor et al. (2016) restored the nuclease activity of the HNH nuclease domain to generate BE3 (Figure 1). The HNH domain nicks the strand bound by the guide RNA (Jinek et al., 2012). This modification improved editing efficiency up to as high as 37%, and close to 10% for the more refractory sites they tested. So how does this stack up against using donor oligonucleotides and HR to target a mutation to a particular locus? How efficient is the editing, and how often
478 Molecular Cell 62, May 19, 2016
does each approach lead to undesired edits such as indels? How good is this at correcting disease mutations? To answer each of these questions, they compared BE3 to more typical Cas9-mediated editing at two disease-associated loci, APOE4 (Alzheimer’s) and TP53 (cancer). Both variants that they tested are amenable to correction by C-to-T editing to restore the correct amino acid sequence, and editing of any nearby Cs would not change the amino acid sequence. For APOE4 in astrocytes, BE3 introduced the correct sequence in nearly 75% of total sequencing reads compared to Cas9 and a 200 nt ssDNA donor correcting only 0.3%. Additionally, Cas9 with ssDNA yielded 26.1% indels compared to less than 5% for BE3. The numbers are similarly striking with TP53. BE3 corrected 7.6% of sequence in human breast cancer cells and generated less than 1% indels, while Cas9 and ssDNA was unable to generate any of the desired sequence changes and 6.1% of sequences had indels. This work is exciting both for its potential and for its use of our understanding of so much basic knowledge of DNA repair pathways. While a limitation is the type of changes that can be made, when it can be used, it will be much more efficient and accurate relative to using Cas9 and ssDNA to introduce mutations. Each week, it seems that new modifications
of Cas9 are reported that change the genome-editing landscape and expand the options for and improve the precision with which changes can be made. REFERENCES Chu, V.T., Weber, T., Wefers, B., Wurst, W., Sander, S., Rajewsky, K., and Ku¨hn, R. (2015). Nat. Biotechnol. 33, 543–548. Gutschner, T., Haemmerle, M., Genovese, G., Draetta, G.F., and Chin, L. (2016). Cell Rep. 14, 1555–1566. Harris, R.S., Petersen-Mahrt, S.K., and Neuberger, M.S. (2002). Mol. Cell 10, 1247–1253. Jinek, M., Chylinski, K., Fonfara, I., Hauer, M., Doudna, J.A., and Charpentier, E. (2012). Science 337, 816–821. Komor, A.C., Kim, Y.B., Packer, M.S., Zuris, J.A., and Liu, D.R. (2016). Nature. http://dx.doi.org/10. 1038/nature17946. Lin, S., Staahl, B.T., Alla, R.K., and Doudna, J.A. (2014). eLife 3, e04766. Maruyama, T., Dougan, S.K., Truttmann, M.C., Bilate, A.M., Ingram, J.R., and Ploegh, H.L. (2015). Nat. Biotechnol. 33, 538–542. Pluciennik, A., Dzantiev, L., Iyer, R.R., Constantin, N., Kadyrov, F.A., and Modrich, P. (2010). Proc. Natl. Acad. Sci. USA 107, 16066–16071. Sternberg, S.H., and Doudna, J.A. (2015). Mol. Cell 58, 568–574. Vriend, L.E.M., Prakash, R., Chen, C.-C., Vanoli, F., Cavallo, F., Zhang, Y., Jasin, M., and Krawczyk, P.M. (2016). Nucleic Acids Res. Published online March 21, 2016. http://dx.doi.org/10.1093/nar/ gkw179.