Crystal Structure of Staphylococcus aureus Cas9

Article

Crystal Structure of Staphylococcus aureus Cas9 Graphical Abstract

Authors Hiroshi Nishimasu, Le Cong, Winston X. Yan, ..., Ryuichiro Ishitani, Feng Zhang, Osamu Nureki

Correspondence [email protected] (F.Z.), [email protected] (O.N.)

In Brief The structure of Staphylococcus aureus Cas9 in complex with sgRNA and its DNA targets is solved and compared to Streptococcus pyogenes Cas9, revealing a different mechanism of PAM recognition and providing new design rationales to expand the CRISPR-Cas9 genome editing toolbox.

Highlights d

Crystal structure of S. aureus Cas9 in complex with sgRNA and target DNA

d

Structural basis for the relaxed recognition of the 50 -NNGRRT-30 PAM

d

Conserved and divergent structural features among orthologous CRISPR-Cas9 systems

d

Rational design of compact transcriptional activators and inducible nucleases

Nishimasu et al., 2015, Cell 162, 1113–1126 August 27, 2015 ª2015 Elsevier Inc. http://dx.doi.org/10.1016/j.cell.2015.08.007

Article Crystal Structure of Staphylococcus aureus Cas9 Hiroshi Nishimasu,1,2 Le Cong,3,4,5,6 Winston X. Yan,3,4,5,6,7 F. Ann Ran,3,8 Bernd Zetsche,3,4,5,6 Yinqing Li,3,4,5,6 Arisa Kurabayashi,1 Ryuichiro Ishitani,1 Feng Zhang,3,4,5,6,* and Osamu Nureki1,* 1Department of Biological Sciences, Graduate School of Science, The University of Tokyo, 2-11-16 Yayoi, Bunkyo-ku, Tokyo 113-0032, Japan 2JST, PRESTO, 2-11-16 Yayoi, Bunkyo-ku, Tokyo 113-0032, Japan 3Broad Institute of MIT and Harvard, Cambridge, MA 02142, USA 4McGovern Institute for Brain Research, Massachusetts Institute of Technology, Cambridge, MA 02139, USA 5Department of Brain and Cognitive Sciences, Massachusetts Institute of Technology, Cambridge, MA 02139, USA 6Department of Biological Engineering, Massachusetts Institute of Technology, Cambridge, MA 02139, USA 7Graduate Program in Biophysics, Harvard-MIT Division of Health Sciences and Technology, Harvard Medical School, Boston, MA 02115, USA 8Society of Fellows, Harvard University, Cambridge, MA 02138, USA *Correspondence: [email protected] (F.Z.), [email protected] (O.N.) http://dx.doi.org/10.1016/j.cell.2015.08.007

SUMMARY

The RNA-guided DNA endonuclease Cas9 cleaves double-stranded DNA targets with a protospacer adjacent motif (PAM) and complementarity to the guide RNA. Recently, we harnessed Staphylococcus aureus Cas9 (SaCas9), which is significantly smaller than Streptococcus pyogenes Cas9 (SpCas9), to facilitate efficient in vivo genome editing. Here, we report the crystal structures of SaCas9 in complex with a single guide RNA (sgRNA) and its doublestranded DNA targets, containing the 50 -TTGAAT-30 PAM and the 50 -TTGGGT-30 PAM, at 2.6 and 2.7 A˚ resolutions, respectively. The structures revealed the mechanism of the relaxed recognition of the 50 -NNGRRT-30 PAM by SaCas9. A structural comparison of SaCas9 with SpCas9 highlighted both structural conservation and divergence, explaining their distinct PAM specificities and orthologous sgRNA recognition. Finally, we applied the structural information about this minimal Cas9 to rationally design compact transcriptional activators and inducible nucleases, to further expand the CRISPR-Cas9 genome editing toolbox. INTRODUCTION Advances in our understanding of the CRISPR (clustered regularly interspaced short palindromic repeat)-Cas (CRISPR-associated) systems (Mojica et al., 2005; Pourcel et al., 2005; Bolotin et al., 2005; Makarova et al., 2006; Barrangou et al., 2007; Brouns et al., 2008; Marraffini and Sontheimer, 2008; Garneau et al., 2010; Deltcheva et al., 2011; Sapranauskas et al., 2011; Jinek et al., 2012; Gasiunas et al., 2012) have propelled the development and applications of Cas9 for genome editing (Cong et al., 2013; Mali et al., 2013; Cho et al., 2013; Hwang et al., 2013; Jiang et al., 2013; Jinek et al., 2013). However, much work remains to understand how Cas9 mediates RNA-guided DNA recognition

and cleavage. Previous studies have shown that Cas9 contains two endonuclease domains, HNH and RuvC, which cleave the DNA strands complementary (target DNA strand) and non-complementary (non-target DNA strand) to the guide RNA, respectively (Sapranauskas et al., 2011; Gasiunas et al., 2012; Jinek et al., 2012), and that Cas9-catalyzed DNA cleavage requires the presence of a short sequence, known as a protospacer-adjacent motif (PAM), located immediately downstream of the target DNA sequence (Bolotin et al., 2005; Deveau et al., 2008; Mojica et al., 2009; Garneau et al., 2010; Sapranauskas et al., 2011; Jinek et al., 2012; Gasiunas et al., 2012). The PAM sequences are diverse among the orthologous CRISPR-Cas systems, and the widely used Cas9 from Streptococcus pyogenes (SpCas9) recognizes a 50 -NGG-30 PAM on the non-target DNA strand. Structural studies of SpCas9 provided insight into the RNAguided DNA cleavage mechanism of the Cas9 enzymes. The crystal structures of SpCas9 in its unbound state (Jinek et al., 2014), SpCas9 in complex with a single guide RNA (sgRNA) (Jiang et al., 2015), and SpCas9 in complex with the sgRNA and its DNA target (Nishimasu et al., 2014; Anders et al., 2014) revealed that SpCas9 undergoes significant structural rearrangement upon association with the sgRNA, and subsequently adopts a bilobed architecture, consisting of a recognition (REC) lobe and a nuclease (NUC) lobe, with a central channel enveloping the RNA–DNA heteroduplex (Jinek et al., 2014; Nishimasu et al., 2014). The PAM is recognized by a PAM-interacting (PI) domain, which facilitates the target DNA unwinding and the heteroduplex formation (Nishimasu et al., 2014; Anders et al., 2014). In addition to SpCas9, the crystal structure of Actinomyces naeslundii Cas9 (AnCas9) is also available, but it lacks the sgRNA and the target DNA (Jinek et al., 2014). Thus, additional crystal structures of Cas9 orthologs (Chylinski et al., 2013; Chylinski et al., 2014) bound to nucleic acids are critical to understand the potential mechanistic and structural conservations underlying RNA-guided DNA targeting by Cas9, and to facilitate the development of new Cas9-based genome engineering technologies. Recently, we harnessed a small Cas9 from Staphylococcus aureus (SaCas9) for eukaryotic genome editing (Ran et al., 2015). Although several Cas9 orthologs can cleave DNA targets Cell 162, 1113–1126, August 27, 2015 ª2015 Elsevier Inc. 1113

in vitro, only SaCas9 and SpCas9 exhibit robust activity when transplanted into mammalian cells. SaCas9 shares only 17% sequence identity with SpCas9, highlighting the structural and functional variations among orthologous CRISPR-Cas9 systems. SaCas9 (1,053 amino acid residues) is significantly smaller than SpCas9 (1,368 amino acid residues), and thus is easier to deliver to somatic tissues for genome editing (Ran et al., 2015). While our previous attempts to create a smaller version of SpCas9 by rationally removing various domains met with limited success (Nishimasu et al., 2014), the structure of the naturally smaller SaCas9 may reveal the minimum and essential components of Cas9 enzymes. In addition, SaCas9 recognizes a 50 -NNGRRT-30 PAM (where R represents a purine [i.e., A or G]), which is distinct from the 50 -NGG-30 PAM for SpCas9, thereby offering an opportunity to understand the diverse PAM specificities of the Cas9 orthologs. Here, we report the crystal structures of SaCas9 in complex with the sgRNA and its target DNA, containing either the 50 -TTGAAT-30 PAM (2.6 A˚ resolution) or the 50 -TTGGGT-30 PAM (2.7 A˚ resolution), to facilitate a comparative analysis of the distinct Cas9 orthologs and to provide the structural foundation for the rational design of SaCas9-based genome engineering tools. Our structural and functional data provide insight into the PAM-dependent, RNA-guided DNA cleavage mechanism of SaCas9, as well as the minimal requirements for general Cas9 activity. Furthermore, our data enable the structural comparison between SaCas9 and SpCas9, revealing notable differences in the interactions in their REC lobe–sgRNA scaffolds and PI domain–PAM sequences. These comparative results also highlighted the flexible nature of the RuvC and HNH nuclease domains, and identified a previously uncharacterized and evolutionarily divergent wedge (WED) domain. Given its compact size, SaCas9 holds great potential as a minimum scaffold for genome engineering. The structural and comparative data presented here advance our understanding of the RNAguided DNA cleavage mechanism and provide a starting point for the future design of Cas9 variants with expanded target space and improved specificity. RESULTS Overall Structure of the SaCas9–sgRNA–Target DNA Complex We solved the crystal structures of full-length SaCas9 (residues 1–1053; N580A/C946A) in complex with a 73-nucleotide (nt) sgRNA, a 28-nt target DNA strand and an 8-nt non-target DNA strand, containing either the 50 -TTGAAT-30 PAM or 50 TTGGGT-30 PAM, at 2.6 and 2.7 A˚ resolutions, respectively (Figures 1A–1D, Figure S1A, and Table S1). Since solvent-exposed cysteine residues may hamper crystallization, we replaced a non-conserved cysteine residue (Cys946) with alanine after confirming that the C946A mutation did not affect the DNA cleavage activity in vivo (Figures S1B–S1D). Based on homology to SpCas9, we mutated one of the putative catalytic residues, Asn580 in the HNH domain, to alanine to prevent the potential cleavage of the target DNA during crystallization. Since the two structures are virtually identical (root-mean-square deviation [rmsd] of 0.2 A˚ for 1,043 equivalent Ca atoms), we will describe 1114 Cell 162, 1113–1126, August 27, 2015 ª2015 Elsevier Inc.

the quaternary complex structure containing the 50 -TTGAAT-30 PAM unless otherwise stated. SaCas9 adopts a bilobed architecture consisting of a REC lobe (residues 41–425) and a NUC lobe (residues 1–40 and 435–1053) (Figures 1C and 1D). The two lobes are connected by an arginine-rich bridge helix (residues 41–73) and a linker loop (residues 426–434). The NUC lobe consists of the RuvC (residues 1–40, 435–480 and 650–774), HNH (residues 520– 628), WED (residues 788–909), and PI (residues 910–1053) domains (Figures 1C and 1D). The PI domain can be divided into a Topoisomerase-homology (TOPO) domain and a C-terminal domain, as in SpCas9 (Jinek et al., 2014). The RuvC domain consists of three separate motifs (RuvC-I–III) and interacts with the HNH and PI domains. The HNH domain is connected to RuvCII and RuvC-III by the L1 (residues 481–519) and L2 (residues 629–649) linker regions, respectively. The WED and RuvC domains are connected by a ‘‘phosphate lock’’ loop (residues 775–787), as in SpCas9 (Anders et al., 2014). The active site of the HNH domain is distant from the cleavage site in the target DNA strand (the phosphodiester linkage between dC3 and dA4), indicating that the present structure represents the inactive state, as in the cases of the SpCas9–sgRNA–target DNA complex structures (Nishimasu et al., 2014; Anders et al., 2014). Previous structural studies revealed that SpCas9 undergoes conformational rearrangements upon guide RNA binding, to form the central channel between the REC and NUC lobes (Jinek et al., 2014; Nishimasu et al., 2014; Anders et al., 2014; Jiang et al., 2015). In the absence of the guide RNA, SpCas9 and AnCas9 adopt a closed conformation, where the active site of the HNH domain is covered by the RuvC domain (Figure S2). In contrast, the ternary and quaternary complex structures of SpCas9 adopt an open conformation and have the central channel, which accommodates the guide RNA–target DNA heteroduplex (referred to as the guide:target heteroduplex) (Figures 1E and 1F). The present quaternary complex structure of SaCas9 adopts a similar open conformation to form the central channel, which accommodates the guide:target heteroduplex (Figures 1C and 1D), suggesting that the guide RNA-induced conformational activation is conserved between SaCas9 and SpCas9. A structural comparison between SaCas9 and SpCas9 revealed that, although their overall architectures are similar, there are notable differences in their REC, WED, and PI domains, as described in detail below, thereby explaining the significant sequence and size differences of the two Cas9 orthologs (Figures 1C–1F and Figure S3). Structure of the sgRNA–Target DNA Complex The SaCas9 sgRNA consists of the guide region (G1–C20), repeat region (G21–G34), tetraloop (G35–A38), anti-repeat region (C39–C54), stem loop 1 (A56–G68), and single-stranded linker (U69–U73), with A55 connecting the anti-repeat region and stem loop 1 (Figures 2A–2D). U73 at the 30 end is disordered in the present structure. The guide region (G1–C20) and the target DNA strand (dG1–dC20) form the guide:target heteroduplex, whereas the target DNA strand (dC(8)–dA(1)) and the non-target DNA strand (dT1*–dG8*) form a PAM-containing duplex (referred to as the PAM duplex) (Figures 2A and 2B). The repeat (G21–G34) and anti-repeat (C39–C54) regions form

1053

968

910

775 788

3' DNA strand 5' Target

Target

CTD

TOPO

WED

L2

HNH

5' PAM

DNA strand

3'

RuvC-III

629 650

481

520

L1

426 435

RuvC-II

74

REC lobe

Non-target

B

NUC lobe

REC

RuvC-I

1

41

NUC lobe

BH

A

Phosphate lock loop

Repeat

5' Guide

Tetraloop Anti-repeat

Stem loop 1

PI

Stem loop 2

sgRNA

C

3'

D

Repeat:anti-repeat duplex REC lobe

REC lobe

Repeat:anti-repeat duplex Tetraloop

Tetraloop

Guide:target 3' heteroduplex

Guide:target 3' heteroduplex

HNH

WED

HNH

PAM

5'

L1

RuvC

WED

NUC lobe

NUC lobe

L2 Non-target DNA strand 3' 5' Target DNA strand

PI

Non-target DNA strand 3' 5' Target DNA strand

L1

RuvC

PI

180º Bridge helix

PAM

5'

180º Bridge helix

Stem loop 1

Stem loop 1

3' WED

5'

WED

L2

3'

TOPO

3'

3'

Linker

RuvC

CTD

L2

5'

TOPO

PI

Linker

CTD

Putative binding site of stem loop 2

RuvC

PI

E

F HNH

ns 2 Ins ( EC (REC C2) 2) (REC2)

REC lobe

REC lobe

Repeat:anti-repeat duplex

Repeat:anti-repeat duplex ns 1 Ins nss 3 Ins

L1 Guide:target 3' heteroduplex

Guide:target 3' heteroduplex

5' WED Tetraloop Non-target DNA strand

NUC lobe

5' PAM

3' 5'

PAM

PI

3' 5'

ns (PI) ((PI (P P Ins

Target DNA strand RuvC

WED Tetraloop Non-target DNA strand

5'

NUC lobe

L2

Target DNA strand RuvC

PI

Figure 1. Structure of the SaCas9–sgRNA–Target DNA Complex (A) Domain organization of SaCas9. BH, bridge helix; CTD, C-terminal domain. (B) Schematic of the sgRNA–target DNA complex. The putative stem loop 2 was truncated for crystallization. (C and D) Ribbon (C) and surface (D) representations of the SaCas9–sgRNA–target DNA complex. The active sites of the RuvC (Asp10) and HNH (Asn580) domains are indicated by red circles. Molecular graphics images were prepared using CueMol (http://www.cuemol.org). (E and F) Ribbon (E) and surface (F) representations of the SpCas9–sgRNA–target DNA complex (PDB: 4UN3). The SpCas9-specific insertions in the REC and PI domains are highlighted in pale blue. In (F), the L1 and L2 linker regions and the HNH domain are omitted for clarity. See also Figures S1, S2, and S3.

Cell 162, 1113–1126, August 27, 2015 ª2015 Elsevier Inc. 1115

A

1*

PAM

C

8*

TTGAATAG 3′ Non-target DNA strand |||||||| Target Target AACTTATC-8 5′ DNA strand 20 10 1 -1 CCTTTAATCCACGCGAACCG |||||||||||||||||||| Repeat 29 21 30 31 34 GGAAAUUAGGUGCGCUUGGCGUUUUAGUA CUCUGG 10 1 20 A ||||||||| |||| Guide 60 59 56 A A CGGAACAAAAUCAUCUAAGACA 55 54 46 45 43 42 39 A Anti-repeat | |||| sgRNA A A GCCGUGUUUAUCUCGUCAAC 63 U65 68 69 72 73 U 64 Linker |||||| |||| U 3′ UAGAGCGGUUG

Anti-repeat

5′

3′

U44

C45

Stem loop 2

Stem loop 1

A43

Tetraloop

5′

A42

C30

A62

D A61

Stem loop 1

Repeat

G65

C59

Tetraloop

Stem loop 1

Guide

A63

A60

Repeat:anti-repeat duplex

Guide:target heteroduplex

A29

Repeat

97

B

U46

U31

G58

U64

C66

G57

C67

G68

3′ A56

5′

Target

Linker

sgRNA

A55

Anti-repeat

U69

PAM

3′

G70

5′

Non-target DNA strand

U72

3′

PAM duplex

5′

U71

Linker

Target DNA strand EMX1 indel (%)

E

21

30

40

50

60

70

80

90

0

5

10

15

20

25

GUUUUAGUACUCUGGAAACAGAAUCUACUAAAACAAGGCAAAAUGCCGUGUUUAUCUCGUCAACUUGUUGGCGAGAU ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------CCGG---CCGG----------------------------------------------------------------------------------CCCGCGCCGG----CCGGCGCGGG ---------------------------------------------------------------------------------------------------------------------CGCCC-----------------------GAU +77 del bulge --------- ------------------------------------------------------------------+77 alter bulge #1 ---------A------------------------------------------------------------------+77 alter bulge #2 ----------------------UCA---------------------------------------------------GUUUUAGUACUCUGGAAACAGAAUCUACUAAAACAAGGCAAAAUGCCGUGUUUAUCUCGUCAACUUGUUGGCGAGAU +77 +48 +54 +77 del SL1 +77 alter SL1 +77 alter SL2 +77 del Linker +77 alter Linker

Repeat

F

Tetraloop

Anti-repeat

Repeat:anti-repeat duplex

Stem loop 1

Linker

Repeat:anti-repeat duplex

Guide

Tetraloop

Stem loop 1 Linker

Stem loop 2

Guide

Tetraloop

5′

Stem loop 1 Linker

Stem loop 2

5′

Stem loop 2

SaCas9 sgRNA SpCas9 sgRNA

SaCas9 sgRNA SpCas9 sgRNA Stem loop 3

3′

Stem loop 3

3′

Figure 2. Structure of the sgRNA–Target DNA Complex (A) Nucleotide sequences of the sgRNA and the target DNA. The putative stem loop 2 (dashed box) was truncated for crystallization. U73 is disordered in the structure. (B) Overview of the sgRNA–target DNA complex. (C and D) Close-up views of the repeat:anti-repeat duplex (C) and stem loop 1 (D). Key interactions are shown as dashed lines.

(legend continued on next page)

1116 Cell 162, 1113–1126, August 27, 2015 ª2015 Elsevier Inc.

a distorted duplex (referred to as the repeat:anti-repeat duplex) via 13 Watson-Crick base pairs (Figures 2A and 2B). The unpaired nucleotides (C30, A43, U44, and C45) form an internal loop (Figure 2C). This distorted structure of the repeat:antirepeat duplex is precisely recognized by the REC and WED domains, as the insertion of GAU into the repeat region, which eliminated the internal loop, drastically reduced the Cas9-mediated DNA cleavage (Figure 2E). Stem loop 1 is formed via three Watson-Crick base pairs (G57:C67–C59:G65) and two non-canonical base pairs (A56: G68 and A60:A63) (Figures 2A and 2D). U64 does not base pair with A60 and is flipped out of the stem loop (Figure 2D). The N1 and N6 of A63 hydrogen bond with the 20 -OH and N3 of A60, respectively. G68 stacks with G57:C67, with the G68 N2 interacting with the backbone phosphate group between A55 and A56. A55 adopts the syn conformation, and its adenine base stacks with U69 (Figure 2D). In addition, the N1 of A55 hydrogen bonds with the 20 -OH of G68, thus stabilizing the basal region of stem loop 1. An adenosine residue immediately after the repeat:anti-repeat duplex is highly conserved among CRISPR-Cas9 systems, and the equivalent adenosine in the SpCas9 sgRNA, A51, also adopts the syn conformation (Nishimasu et al., 2014; Anders et al., 2014) (Figure S4A), suggesting that these adenosine residues play conserved key roles in connecting the repeat:anti-repeat duplex and stem loop 1. The SpCas9 sgRNA contains three stem loops (stem loops 1–3), which interact with Cas9 and contribute to complex formation (Nishimasu et al., 2014) (Figure 2F and Figure S4A). The sgRNA lacking stem loops 2 and 3 supports SpCas9-catalyzed DNA cleavage in vitro but not in vivo, indicating the importance of stem loops 2 and 3 for the cleavage activity in vivo (Cong et al., 2013; Hsu et al., 2013; Nishimasu et al., 2014; Wright et al., 2015). In contrast, the SaCas9 sgRNA is predicted to contain only two stem loops (stem loops 1 and 2), based on its nucleotide sequence (Figure S4B). As in SpCas9, the sgRNA lacking stem loop 2 supported SaCas9-catalyzed DNA cleavage in vitro but not in vivo (Figure 2E and Figure S4C), suggesting that the secondary structures on the 30 tracrRNA tail of the sgRNA are critical for successfully harnessing Cas9 for genome editing applications. Although we failed to obtain diffraction-quality crystals of SaCas9 bound to a version of sgRNA containing the full-length tracrRNA, the truncation of stem loop 2 dramatically improved the quality of the crystals. Thus, it remains unknown how stem loop 2 interacts with SaCas9, although it may bind to the positively charged groove between the RuvC and PI domains, as in SpCas9 (Nishimasu et al., 2014; Anders et al., 2014) (Figure 1D and Figure S4D). Recognition Mechanism of the Guide:Target Heteroduplex The guide:target heteroduplex is accommodated in the central channel formed between the REC and NUC lobes (Figures 1D and 3). The sugar-phosphate backbone of the PAM-distal region

(A3–U6) of the sgRNA interacts with the REC lobe (Thr238, Tyr239, Lys248, Tyr256, Arg314, Asn394, and Gln414) (Figure 3). In SpCas9 and SaCas9, the RNA–DNA base pairing in the 8 bp PAM-proximal ‘‘seed’’ region in the guide:target heteroduplex is critical for Cas9-catalyzed DNA cleavage (Jinek et al., 2012; Hsu et al., 2013; Ran et al., 2015). Consistent with this, the phosphate backbone of the sgRNA seed region (C13–C20) is extensively recognized by the bridge helix (Asn44, Arg48, Arg51, Arg55, Arg59, and Arg60) and the REC lobe (Arg116, Gly117, Arg165, Gly166, Asn169, and Arg209) (Figures 3 and 4A), as in the case of SpCas9 (Nishimasu et al., 2014). In addition, the 20 -OH groups of C15, U16, U17, and G19 interact with the REC lobe (Gly166, Arg208, Arg209, and Tyr211). These structural findings suggest that the sgRNA binds to SaCas9 with its seed region pre-ordered in an A-form conformation for base-pairing with the target DNA strand, as observed in the SpCas9–sgRNA binary complex (Jiang et al., 2015). In addition, the sugar-phosphate backbone of the target DNA strand interacts with the REC lobe (Tyr211, Trp229, Tyr230, Gly235, Arg245, Gly391, Thr392, and Asn419) and the RuvC domain (Leu446, Tyr651 and Arg654) (Figure 3). These structural observations explain the RNA-guided DNA targeting mechanism of SaCas9. The C-terminal region of the REC lobe interacts with the PAMdistal region of the heteroduplex, whereas the N-terminal region of the REC lobe interacts with the repeat:anti-repeat duplex and the PAM-proximal region of the heteroduplex (Figure 3). Notably, the C-terminal region of the REC lobe of SaCas9 shares structural similarity with those of SpCas9 (PDB: 4UN3, 26% identity, rmsd of 1.9 A˚ for 177 equivalent Ca atoms) and AnCas9 (PDB: 4OGE, 16% identity, rmsd of 3.2 A˚ for 167 equivalent Ca atoms) (Figure S5). These structural findings suggested that the Cas9 orthologs recognize the PAM-distal region of the guide:target heteroduplex in a similar manner. Recognition Mechanism of the sgRNA Scaffold The repeat:anti-repeat duplex is recognized by the REC and WED domains, primarily through interactions between the protein and the sugar-phosphate backbone (Figures 3 and 4B). Consistent with our data showing that the distorted repeat:anti-repeat duplex is critical for Cas9-catalyzed DNA cleavage (Figure 2E), the internal loop is recognized by the WED domain (Figure 4B). The 20 -OH of C30 hydrogen bonds with Tyr868, and the backbone phosphate groups of U31, C45, and U46 interact with Lys870, Arg792, and Lys881, respectively (Figure 4B). These structural observations explain the structure-dependent recognition of the repeat:anti-repeat duplex by SaCas9. Stem loop 1 is recognized by the bridge helix and the REC lobe (Figures 3 and 4C). The phosphate backbone of stem loop 1 interacts with the bridge helix (Arg47, Arg54, Arg55, Arg58, and Arg59) and the REC lobe (Arg209, Gly216, and Ser219) (Figure 4C). The 20 -OH of A63 hydrogen bonds with His62. The flipped-out U64 is recognized by Arg209 and Glu213 via stacking and hydrogen-bonding interactions, respectively. A55

(E) Mutational analysis of SaCas9 sgRNA scaffolds. Effects of mutations on the ability to induce indels in the target EMX1 locus were examined. Base changes from the sgRNA (+77) scaffold are shown at the respective positions, with dashes indicating unaltered bases (n = 3, error bars show mean ± SEM). (F) Superimposition of the sgRNAs of SaCas9 and SpCas9 (PDB: 4OO8) (stereoview). See also Figure S4.

Cell 162, 1113–1126, August 27, 2015 ª2015 Elsevier Inc. 1117

Figure 3. Schematic of Nucleic Acid Recognition by SaCas9

PAM duplex

C39

G34

A40

U33

G41

C32

A42

U31

Arg792 (Lys881)

SaCas9 residues that interact with the sgRNA and the target DNA via their main chain are shown in parentheses. Water-mediated hydrogen-bonding interactions are omitted for clarity.

(Tyr868)

A43 U44

Repeat

Anti-repeat

Repeat:anti-repeat duplex

Tetraloop

C30

(Lys870)

C45

WED domains recognize distinct structural features of the repeat:anti-repeat A47 U28 T-7 A7 (Val164) duplex, allowing the cognate sgRNAs to G27 C48 T6 A-6 (Arg165) Arg991 (Gly162) be distinguished in a highly specific (Lys886) U49 A26 A5 T-5 Asn888 His111 Asn986 manner (Figures 4E and 4F and Fig(Ala889) U73 A50 U25 A4 T-4 Arg165 Asn985 Lys459 (Arg901) Lys114 ure S6). The SaCas9 WED domain has a U72 A51 U24 G3 C-3 (Leu909) Arg774 Tyr882 Ser790 Arg1015 Arg61 Arg452 new fold comprising a twisted fiveU71 A52 U23 A-2 T2 Tyr789 Tyr897 (Leu788) Arg781 Asn780 stranded b sheet flanked by four a G70 A53 U22 T1 A-1 Asn804 Lys50 (Arg781) Arg47 Lys57 (Leu783) U69 helices, and is responsible for the recogC54 G21 (Asp786) Lys906 Asn780 (Thr787) Thr787 A55 nition of the distorted repeat:anti-repeat (Gly166) Leu783 G1 C20 duplex, as described above (Figures 4B Arg165 Leu891 A56 G68 Arg54 Asn169 C2 G19 and 4E and Figures S6A and S6B). In G57 C67 Ser219 Arg60 (Gly117) C3 G18 contrast, the SpCas9 WED domain Arg208 Arg58 G58 C66 Arg116 A4 U17 (Gly216) adopts a compact loop conformation (Arg209) (Tyr211) C59 G65 Arg209 A5 U16 Arg59 and interacts with the repeat:anti-repeat Glu213 U64 Tyr211 G6 C15 Arg55 duplex, which is structurally different His62 A60 A63 Trp229 C7 G14 Arg51 from that of the SaCas9 sgRNA (NishiA61 Tyr230 G8 C13 Arg48 masu et al., 2014; Anders et al., 2014) (Gly391) C9 G12 Asn44 (Gly235) (Figure 4F and Figures S6A and S6C). Thr392 A10 U11 The AnCas9 WED domain has yet Stem loop 1 Arg245 C11 G10 another different fold containing three Leu446 C12 G9 REC lobe Tyr651 antiparallel b-hairpins (Jinek et al., 2014) T13 A8 Arg654 Bridge helix (Figure S6A). These structural differences A14 U7 RuvC domain Arg314 in the WED domains are consistent with A15 U6 Lys248 Phosphate lock loop the variations in the sgRNA scaffolds Tyr256 T16 A5 Asn394 among the CRISPR-Cas9 systems (FonWED domain T17 A4 Tyr239 fara et al., 2014; Briner et al., 2014; Ran PI domain Thr238 T18 A3 Gln414 et al., 2015). Watson-Crick base pair C19 G2 The REC lobes also contribute to the Non Watson-Crick base pair C20 G1 Asn419 orthogonal recognition of the sgRNA Hydrogen bond/salt bridge scaffolds. While the REC lobes of Hydrophobic interaction SaCas9 and SpCas9 share structural Water molecule similarity, the SpCas9 REC lobe has four characteristic insertions (Ins 1–4), which are absent in the SaCas9 REC is extensively recognized by the phosphate lock loop (Figure 4D). lobe (Figures 4E and 4F and Figures S6B and S6C). Ins 2 (also The N6, N7, and 20 -OH of A55 hydrogen bond with Asn780/ known as the REC2 domain) does not contact the nucleic acids Arg781, Leu783, and Lys906, respectively. Lys57 interacts with in the SpCas9 structures and is dispensable for the DNA cleavthe backbone phosphate group between C54 and A55, and age activity (Nishimasu et al., 2014), consistent with the absence the side chain of Leu783 forms hydrophobic contacts with the of Ins 2 in SaCas9 (Figures 4E and 4F). Ins 1 and 3 recognize the nucleobases of A55 and A56. The phosphate backbone of the SpCas9-specific internal loop in the repeat:anti-repeat duplex linker region electrostatically interacts with the RuvC domain (Figure 4F and Figure S6C). In particular, Ins 3 interacts with (Arg452, Lys459, and Arg774) and the phosphate lock loop the flipped-out G43 and U44 in the repeat:anti-repeat duplex in (Arg781), and the nucleobase of G70 stacks with the side chain base-specific manners (Figure S6C). In addition, Ins 4 interacts of Arg47 on the bridge helix (Figure 4D). with stem loop 1 of the SpCas9 sgRNA, which is shorter than that of the SaCas9 sgRNA (Figures 4E and 4F and Figures S6B Structural Basis for the Orthogonal Recognition of and S6C). Together, these structural observations demonstrate sgRNA Scaffolds how the Cas9 orthologs recognize their cognate sgRNAs in A comparison of the quaternary complex structures of SaCas9 orthogonal manners, using specific combinations of the structurand SpCas9 revealed that the structurally diverse REC and ally diverse REC and WED domains. C-8

(Lys878) (Lys879)

U46

A29

sgRNA

Target DNA strand

Target

Guide

A62

Guide:target heteroduplex

Non-target DNA strand

PAM

Linker

G8

1118 Cell 162, 1113–1126, August 27, 2015 ª2015 Elsevier Inc.

Recognition Mechanism of the 50 -NNGRRT-30 PAM SaCas9 recognizes the 50 -NNGRRN-30 PAM, with a preference for a thymine base at the 6th position (Ran et al., 2015), which is distinct from the 50 -NGG-30 PAM of SpCas9. In the present structures containing either the 50 -TTGAAT-30 PAM or the 50 -TTGGGT-30 PAM, the PAM duplex is sandwiched between the WED and PI domains, and the PAM in the non-target DNA strand is read from the major groove side by the PI domain (Figures 5A and 5B). dT1* and dT2* do not directly contact the protein (Figures 5A and 5B). Consistent with the observed requirement for the 3rd G in the 50 -NNGRRT-30 PAM, the O6 and N7 of dG3* form bidentate hydrogen bonds with the side chain of Arg1015, which is anchored via salt bridges with Glu993 in both complexes (Figures 5A and 5B). In the 50 -TTGAAT-30 PAM complex, the N7 atoms of dA4* and dA5* form direct and water-mediated hydrogen bonds with Asn985 and Asn985/Asn986/Arg991, respectively (Figure 5A). In addition, the N6 of dA5* forms a water-mediated hydrogen bond with Asn985. Similarly, in the 50 -TTGGGT-30 PAM complex, the N7 atoms of dG4* and dG5* form direct and water-mediated hydrogen bonds with Asn985 and Asn985/Asn986/Arg991, respectively (Figure 5B). The O6 of dG5* forms a water-mediated hydrogen bond with Asn985. These structural features explain the ability of SaCas9 to recognize the purine nucleotides at positions 4 and 5 in the 50 -NNGRRT-30 PAM. The O4 of dT6* hydrogen bonds with Arg991 (Figures 5A and 5B), explaining the preference of SaCas9 for the 6th T in the 50 -NNGRRT-30 PAM. Single alanine mutations of these PAM-interacting residues reduced the cleavage activity in vivo, and double mutations abolished the activity (Figure 5C), confirming the importance of Asn985, Asn986, Arg991, Glu993, and Arg1015 for PAM recognition. In addition, the phosphate backbone of the PAM duplex is recognized from the minor groove side by the WED domain (Tyr789, Tyr882, Lys886, Ans888, Ala889, and Leu909), in a distinct manner from that in SpCas9 (Figure 3). Together, our structural and functional data have revealed the mechanism underlying the relaxed recognition of the 50 -NNGRRT-30 PAM by SaCas9. Structural Basis for the Distinct PAM Specificities A structural comparison of SaCas9, SpCas9, and AnCas9 revealed that, despite the lack of sequence homology, their PI domains share a similar protein fold (Figures 5D and 5E, and Figure S6A). The PI domains consist of the TOPO domain, comprising a three-stranded anti-parallel b sheet (b1–b3) flanked by several a helices, and the C-terminal domain, comprising a twisted six-stranded anti-parallel b sheet (b4–b9) (b7 in SpCas9 adopts a loop conformation) (Figures 5D and 5E and Figure S6A). In both SaCas9 and SpCas9, the major groove of the PAM duplex is read by the b5–b7 region in their PI domains (Figures 5D and 5E). The 3rd G in the 50 -NNGRRT PAM-30 is recognized by Arg1015 in SaCas9 (Figure 5D), whereas the 3rd G in the 50 -NGG-30 PAM is recognized by Arg1335 in SpCas9 in a similar manner (Figure 5E). However, there are notable differences in the PI domains of SaCas9 and SpCas9, consistent with their distinct PAM specificities. Arg1333 of SpCas9, which recognizes the 2nd G in the 50 -NGG-30 PAM, is replaced with Pro1013 in SaCas9 (Figures 5D and 5E and Figure S3). In addition, SpCas9 lacks

the amino acid residues equivalent to Asn985/Asn986 (b5) and Arg991 (b6) of SaCas9, because the b5–b6 region of SpCas9 is shorter than that of SaCas9 (Figures 5D and 5E and Figure S3). Moreover, Asn985, Asn986, Arg991, and Arg1015 in SaCas9 are replaced with Asp1030, Thr1031, Lys1034, and Lys1061 in AnCas9, respectively (Figure S6A), suggesting that the PAM of AnCas9 is different from those of SaCas9 and SpCas9 (although the sequence remains unknown). Together, these structural findings demonstrate that the distinct PAM specificities of the Cas9 orthologs are primarily defined by the specific differences in the PAM-interacting residues in the PI domains. Mechanism of Target DNA Unwinding In SpCas9, Glu1108 and Ser1109, in the phosphate lock loop, hydrogen bond with the phosphate group between dA(1) and dT1 in the target DNA strand (referred to as the +1 phosphate), thereby contributing to the target DNA unwinding (Anders et al., 2014) (Figure 5F). The present structure revealed that SaCas9 also has the phosphate lock loop, although it shares limited sequence similarity to that of SpCas9 (Figure 5G and Figure S3). In SaCas9, the +1 phosphate between dA(1) and dG1, in the target DNA strand, hydrogen bonds with the main-chain amide groups of Asp786 and Thr787 and the side chain of Thr787 in the phosphate lock loop (Figure 5G). These interactions result in the rotation of the +1 phosphate, thereby facilitating base-pairing between dG1 in the target DNA strand and C20 in the sgRNA. Indeed, the SaCas9 T787A mutant showed reduced DNA cleavage activity (Figure 5C), confirming the functional significance of Thr787 in the phosphate lock loop. These observations indicated the conserved molecular mechanism of target DNA unwinding in SaCas9 and SpCas9. RuvC and HNH Nuclease Domains The RuvC domain of SaCas9 has an RNase H fold, and shares structural similarity with those of SpCas9 (PDB: 4UN3, 26% identity, rmsd of 2.0 A˚ for 179 equivalent Ca atoms) and AnCas9 (PDB: 4OGE, 17% identity, rmsd of 3.0 A˚ for 169 equivalent Ca atoms) (Figure 6A). Asp10, Glu477, His701, and Asp704 of SaCas9 are located at positions similar to those of the catalytic residues of SpCas9 (Asp10, Glu762, His983, and Asp986) and AnCas9 (Asp17, Glu505, His736, and Asp739) (Figure 6A and Figure S3). Indeed, the D10A, E477A, H701A, and D704A mutants of SaCas9 exhibited almost no DNA cleavage activity (Figures S7A and S7B), suggesting that the SaCas9 RuvC domain cleaves the non-target DNA strand through a two-metal ion mechanism, as in other RNase H superfamily endonucleases (Go´recka et al., 2013). The HNH domain of SaCas9 has a bba-metal fold, and shares structural similarity with those of SpCas9 (27% identity, rmsd of 1.8 A˚ for 93 equivalent Ca atoms) and AnCas9 (18% identity, rmsd of 2.6 A˚ for 98 equivalent Ca atoms) (Figure 6B). Asp556, His557, and Asn580 of SaCas9 are located at positions similar to those of the catalytic residues of SpCas9 (Asp839, His840, and Asn863) and AnCas9 (Asp581, His582, and Asn606) (Figure 6B and Figure S3). Indeed, the H557A and N580A mutants of SaCas9 almost completely lacked DNA cleavage activity (Figures S7A and S7B), suggesting that the SaCas9 HNH domain cleaves the target DNA strand through a one-metal ion Cell 162, 1113–1126, August 27, 2015 ª2015 Elsevier Inc. 1119

A

B

REC lobe

REC lobe Bridge helix

His111

Anti-repeat

Lys11 14 Lys114 Arg116

Arg208

Arg209

38 Arg61 Arg61

Tyr211

6 26 2

50 0

40 40

7 27

3 39

Gly117 Tyr897

Asn169

16 17 18

Arg59

15

51 1

52 Arg165

49 4

34

24 2

Gly166

4 44

Arg51

29 2

Gly1 Gly y1 y 162 2 Gly162

23 3

21

30

20

13

47 47

Ser790 Lys 881

21

12

32

4 45

Leu788

Arg48

33

43 43

Arg792

22

Repeat

1 31

4 46

22 Lys870

Repeat Asn44

C

7 67

7 57

Stem loop 1

8 68 Arg5 Arg A g 4 Arg54

56 63

1 61

Lys9 Lys906

His62 60 65

u213 Glu213

Bridge helix 6 69 Leu783

B

A

Arg47 Asn 780

Gly216

7 70

Link r Linker

Arg781

66

180º

A Arg2 Arg209 71

58 8 Arg58 67 7 7 57 68 8

Stem loop 1

55

WED domain

64 4

59

Tyr868

Lys879

D 2 62

Lys878

Asn8 4 Asn804

WED domain

Bridge helix

Arg54

Arg59

Arg55

72

Phosphate lock loop

Ser219

C

REC lobe

D

56 6 Arg4 452 Arg452

55

Bridge helix

Arg774

69

Ly 459 Lys4 L Lys459

RuvC domain

E

3 35

42

V 64 Val1 Val164

4 48 8 Arg55

14

Guide

41 4 1

28

A 165 Arg1 Arg165

53

4 54

19

25 2

SaCas9

REC (C)

F

REC (N)

SpCas9

REC (C)

REC (N) Repeat:anti-repeat duplex

REC lobe

REC lobe

Repeat:anti-repeat duplex

90º Guide:target heteroduplex

WED

90º Guide:target heteroduplex Ins 4

Stem loop 1 Bridge helix

WED Stem loop 1 Bridge helix Ins 1

Ins 2 (REC2)

Ins 3

(legend on next page)

1120 Cell 162, 1113–1126, August 27, 2015 ª2015 Elsevier Inc.

mechanism, as in other bba-metal endonucleases (Biertu¨mpfel et al., 2007). A structural comparison of SaCas9 with SpCas9 and AnCas9 revealed that the RuvC and HNH domains are connected by a-helical linkers, L1 and L2, and that notable differences exist in the relative arrangements between the two nuclease domains (Figure 6C). A biochemical study suggested that PAM duplex binding to SpCas9 facilitates the cleavage of the target DNA strand by the HNH domain (Sternberg et al., 2014). However, in the PAM-containing quaternary complex structures of SaCas9 and SpCas9, the HNH domains are distant from the cleavage site of the target DNA strand (Figure 6C). A structural comparison of SaCas9 with Thermus thermophilus RuvC in complex with a Holliday junction substrate (Go´recka et al., 2013) indicated steric clashes between the L1 linker and the modeled non-target DNA strand, bound to the active site of the SaCas9 RuvC domain (Figures S7C and S7D). These observations suggested that the binding of the non-target DNA strand to the RuvC domain may facilitate a conformational change of L1, thereby bringing the HNH domain to the scissile phosphate group in the target DNA strand. Structure-Guided Engineering of SaCas9 Transcription Activators and Inducible Nucleases Using the crystal structure of SaCas9, we sought to conduct structure-guided engineering to further expand the CRISPRCas9 toolbox, as we have done previously using the SpCas9 crystal structure. Given the similarities in the overall domain organizations of SaCas9 and SpCas9, we initially explored the feasibility of engineering the SaCas9 sgRNA, to develop robust transcription activators. In the SpCas9 structure, the tetraloop and stem loop 2 of the sgRNA are exposed to the solvent (Nishimasu et al., 2014; Anders et al., 2014) (Figure S4D), and permitted the insertion of RNA aptamers into the sgRNA to create robust RNA-guided transcription activators (Konermann et al., 2015). To generate the SaCas9-based activator system, we created a catalytically inactive version of SaCas9 (dSaCas9) by introducing the D10A and N580A mutations to inactivate the RuvC and HNH domains, respectively, and attached VP64 to the C terminus of dSaCas9 (Figures 7A and 7B). The sgRNA scaffold was modified by the insertion of the MS2 aptamer stem loop (MS2-SL), to allow the recruitment of MS2-p65-HSF1 transcriptional activation modules (Figure 7A). To evaluate the dSaCas9based activator design, we constructed a transcriptional activation reporter system, consisting of tandem sgRNA target sites upstream of a minimal CMV promoter driving the expression of the fluorescent reporter gene mKate2 (Zhang et al., 2011) (Figure 7B). We included an additional transcriptional termination signal upstream of the reporter cassette, to reduce the background previously observed in a similar reporter (Cong et al., 2012) (Figure 7B). We observed robust activation of

mKate2 transcription when we expressed the engineered sgRNA complementary to the target sites, whereas the non-binding sgRNA had no detectable effect (Figure 7C). Based on a screening of different sgRNA designs with this reporter assay, we found that the insertions of MS2-SL into the tetraloop and putative stem loop 2 induced strong activation in our reporter system, whereas the insertion of MS2-SL into stem loop 1 yielded modest activation, consistent with the structural data (Figure 7D). The single insertion of MS2-SL into the tetraloop was the simplest design that yielded strong transcriptional activation. Using this optimal sgRNA design, we further tested the activation of endogenous targets in the human genome. We selected two guides each for the human ASCL1 and MYOD1 genomic loci, and demonstrated that the dSaCas9-based activator system activated both genes to levels comparable to those of the dSpCas9-based activator (Konermann et al., 2013) (Figure 7E). Given that the sgRNAs for SaCas9 and SpCas9 are not interchangeable, the SaCas9-based transcription activator platform complements the SpCas9-based activator systems, by allowing the independent activation of different sets of genes. The SpCas9 structure also facilitated the rational design of split-Cas9s (Zetsche et al., 2015; Wright et al., 2015), which can be further engineered into an inducible system (Zetsche et al., 2015). Our SaCas9 structure revealed several flexible regions in SaCas9 that could likewise serve as potential split sites (Figure 7F). We created three versions of a split-SaCas9, and two of them showed robust cleavage activity at the endogenous EMX1 target locus (Figure 7G). Using the best split design, we then tested inducible schemes based on the abscisic acid (ABA) sensing system (Liang et al., 2011), as well as two versions of the rapamycin-inducible FKBP/FRB system (Banaszynski et al., 2005) (Figures 7H and 7I). All three systems were able to support inducible SaCas9 cleavage activity, demonstrating the possibility of an inducible, split-SaCas9 design; however, further optimization is required to increase its efficiency and reduce its background activity (Figure 7J). DISCUSSION The present SaCas9 complex structures allow a detailed structural comparison of Cas9 orthologs bound to nucleic acids, thereby illuminating the conserved structural features. SaCas9 and SpCas9 both adopt a bilobed architecture consisting of the REC and NUC lobes, with the guide:target heteroduplex accommodated between these lobes. In addition, both Cas9 orthologs have a phosphate lock loop, which participates in target DNA unwinding and heteroduplex formation. We also found that the HNH and RuvC domains are connected by the L1 and L2 linkers in both SaCas9 and SpCas9. These flexible linker regions appear to play a role in the inactive-to-active conformational

Figure 4. sgRNA Recognition Mechanism (A–D) Recognition of the seed region (A), the repeat:anti-repeat duplex (B), stem loop 1 (C), and the basal region of stem loop 1 (D). Hydrogen bonds and salt bridges are shown as dashed lines. In (A), the target DNA strand is omitted for clarity. (E and F) Recognition of the sgRNA and target DNA by the REC and WED domains of SaCas9 (E) and SpCas9 (PDB: 4UN3) (F). The SpCas9-specific insertions are highlighted in pale blue. See also Figures S5 and S6.

Cell 162, 1113–1126, August 27, 2015 ª2015 Elsevier Inc. 1121

T et DNA strand Targ Target

A

T et DNA strand Targ Target

Non-target DNA strand

Non-target DNA strand

5'-TTGAAT-3' PAM

5'-TTGAAT-3' PAM

F

A1 A19

dA4

Guide A20 A2

dT3 dT3 Glu1108 u1108

dG3* dG3* dA dA4* A4* A4 4 4**

dT1 dT dT1 A4 * A4* A4 dA4*

Asn986 Glu993

dT2 T

dG3* dG3*

dA5* dA5*

dT T6 dT6*

Asn986 Glu993

Asn985

dA5* dA5*

Phosphate lock loop

dT T dT6*

Ser1109 er11

Target DNA stran strand rand ran

Asn985

Arg1015

d dA dA(-1 dA( dA(-1) A(-1 A(-1 A( 1

Arg1015 Arg1015

Non-target DNA strand

5'-NGG-3' PAM Arg991

Arg991

dC(-2) dC(dC(-2 C(C(-2 dG2* d

dT1* dT1* T1 1

Target DNA strand

B

Target DNA strand

dA4 4

G

G19 1 C20 C 2

Non-target DNA strand

Non-target DNA strand

5'-TTGGGT-3' PAM

5'-TTGGGT-3' PAM

dC3 dC C3

dC2 dG3* dG3*

dG3* 3 dG3* d 4 dG4* 4**

Phosphate lock loop Thr787

dT T dT6*

Asn986

dG5*

Target DNA strand

dG5*

Glu993

Asn98 Asn985 Arg1015

Asp786

dG1 d 1 4** 4 dG4*

dT6* dT

Asn986 Glu993

Guide

Asn98 Asn985 Arg1015

Arg991

5'-NNGRRT-3' PAM

dA dA( dA(-1 dA(-1) A(-1 A(-1 1

Arg99 91 Arg991

dA(-2) dA(-2 d A(-2 A( )

Non-target DNA strand

dT1* dT2*

C

EMX1-sg1

EMX1-sg2 Single mutants

Relative indel (%)

100

Double mutants

PLL

50

0

WT

N985A

N986A

R991A

E993A

R1015A

D

N985A N986A

N985A R991A

N986A R991A

N985A R1015A

N986A R1015A

R991A R1015A

E993A R1015A

T787A

E TOPO

SaCas9 PI

SpCas9 PI

TOPO

5'-NGG-3' PAM

5'-NNGRRT-3' PAM

2

3

1

CTD 5'

2

5

1 CTD

8 9

3

9

4

5

6 Arg Arg1 Arg rg1015 Arg1015

Arg1333 Arg13 13 33 33 3

6

Arg1 Ar Arg1335 rg13 g13 g 1335

Non-target DNA strand

(7)

Non-target DNA strand

5'

8

4

Arg991

7

3'

Asn985

Pro1013 1

CTD

3' 1053

1138

1368

Insertion

Ins

CTD

PI

1200

TOPO

968

TOPO

910

PI

Figure 5. PAM Recognition Mechanism (A and B) Recognition of the 50 -TTGAAT-30 PAM (A) and the 50 -TTGGGT-30 PAM (B) (stereoview). The PAM sequences are highlighted in purple. Water molecules are shown as red spheres. Hydrogen bonds are shown as dashed lines. (C) Mutational analysis of the PAM-interacting residues and the phosphate lock loop (PLL), measured by indel rates at two EMX1 targets (n = 3, error bars show mean ± SEM). (D and E) PAM recognition by the PI domains of SaCas9 (D) and SpCas9 (PDB: 4UN3) (E). The PAM sequences are highlighted in purple. Hydrogen bonds are shown as dashed lines. The target DNA strands are omitted for clarity. In (E), the SpCas9-specific insertion is highlighted in pale blue. (F and G) +1 phosphate recognition by the phosphate lock loop in SpCas9 (PDB: 4UN3) (F) and SaCas9 (G). See also Figure S6.

1122 Cell 162, 1113–1126, August 27, 2015 ª2015 Elsevier Inc.

A

58 (BH) 650 (L2)

925 (L2)

40 0 (B (BH) BH

2 5

4

508 (L1)

1

2

4 1 2

3

435 (REC) 774 (PLL)

3

His701 His s7 1 s7

5 719 (REC)

Glu762 76 2 Asp10 Asp1 sp1 p10 Asp986 Asp98 Asp9 sp98

Glu505

1 1097 (PLL)

3

B

2 His736 His7 s73 s7 73

4

815 (PLL)

3 6

5

5

SaCas9 RuvC

468 (REC) 48 (BH)

1

Asp739 As A Asp Asp73 9

6

6

1

4

Asp Asp1 As Asp17 sp17 p

3

2 His983 H is 83

4

5

2

1

4

5

Glu477 u47 77 Asp1 Asp10 Asp704 A Asp70 Asp sp7 p 4

3

679 (L2)

765 (L1)

480 (L1)

SpCas9 RuvC

AnCas9 RuvC

Ala580 Ala58 (Asn580) Asn580

3 Asn863 Asn8

3

902 (L2)

4

628 (L2)

3

2

2

4

2

4 Asn606 Asn60

663 (L2)

2

2

5

1

5

Asp556 His557

2 5

1

1

1

1

Asp839

Asp581 His582

1

Ala840 (His840) His840

520 (L1)

808 (L1)

SaCas9 HNH

549 (L1)

SpCas9 HNH

AnCas9 HNH

HNH

C

HNH

Target DNA strand

HNH

L1

Target DNA strand 4

4 3

L1

L L2

2 L2

His557

1

2 5

His582

2

His840 0

5'

1 2

1

3

L1

3

5' L2

1

3 4

1

4

3

4

1

5

2 3

4 5 4

6 5

5

1

2 3 6

RuvC

RuvC

SaCas9

2 3

4 6

5

RuvC

2

1

SpCas9

AnCas9

Figure 6. RuvC and HNH Nuclease Domains (A and B) Structural comparison of the RuvC (A) and HNH (B) domains of SaCas9, SpCas9 (PDB: 4UN3), and AnCas9 (PDB: 4OGE). The conserved a helices and b strands are numbered. The catalytic residues are shown as stick models. BH, bridge helix; PLL, phosphate lock loop. (C) Comparison of the spatial arrangements between the RuvC and HNH domains in SaCas9, SpCas9 (PDB: 4UN3), and AnCas9 (PDB: 4OGE). The Ca atoms of the catalytic histidine residues in the HNH domains are shown as spheres. The cleavage sites of the target DNA strands are indicated by magenta circles. In SpCas9, the disordered region in the L1 linker is indicated by dashed lines. See also Figure S7.

transition of the HNH domain, although further structural and functional studies are required to elucidate the activation mechanism of Cas9 enzymes. A comparison of the Cas9 orthologs also revealed the structural diversity among the CRISPR-Cas9 systems. In both SaCas9 and SpCas9, the structurally diverse REC and WED domains are responsible for the recognition of sgRNA scaffolds, which have diverse sequences and structures among the CRISPR-Cas9 systems, thereby enabling the orthogonal, species-specific recognition of the sgRNA scaffolds. In addition, the PI domains of SaCas9, SpCas9, and AnCas9 share a similar

core fold, but possess different PAM-interacting residues, corresponding to their distinct PAM specificities. We leveraged our newly obtained structural knowledge to develop SaCas9-based transcriptional activators and inducible SaCas9 systems. This opens the door to new possibilities, including the combinatorial use of SaCas9- and SpCas9-based genome editing and transcriptional regulation systems, to enable simultaneous, inducible editing, activation or repression of multiple endogenous loci. Further applications include the design of a minimal Cas9 enzyme with tailored PAM specificities as well as increased specificity for more versatile genome editing. Cell 162, 1113–1126, August 27, 2015 ª2015 Elsevier Inc. 1123

A

B

sgRNA (with MS2-SL)

dSaCas9

NLS

SaCas9 activator system

D10A

dSaCas9

N580A 3x SV40 PolyA signal

HSF1 VP64

NLS VP64

NLS MS2

p65

Guide U6

2A-GFP

HSF1

Linker

sgRNA scaffold

8x SaCas9 guide binding sites mKate2

p65

Transcriptional activation reporter

MS2

PolyA signal

Minimal CMV

C

GFP Assembled SaCas9 activator complex

mKate2

GFP

mKate2

Non-binding control + Reporter

SaCas9 activator + Reporter

Fold reporter activation

D

100 21

30

40

50

60

70

80

101

102

103

104

90

No insertion (WT) GUUUUAGUACUCUGGAAACAGAAUCUACUAAAACAAGGCAAAAUGCCGUGUUUAUCUCGUCAACUUGUUGGCGAGAU MS2-SL in TL --------------[IS]----------------------------------------------------------MS2-SL in SL1 ---------------------------------------[IS]---------------------------------MS2-SL in SL2 ---------------------------------------------------------------[IS]---------MS2-SL in TL+SL1 --------------[IS]---------------------[IS]---------------------------------MS2-SL in TL+SL2 --------------[IS]---------------------------------------------[IS]----------

GUUUUAGUACUCUGGAAACAGAAUCUACUAAAACAAGGCAAAAUGCCGUGUUUAUCUCGUCAACUUGUUGGCGAGAU Repeat

E Relative mRNA level

10

Tetraloop

ASCL1

4

Anti-repeat

Stem loop 1

Linker

F

10

Control

G

MYOD1 10

SaCas9 activator

Stem loop 2

Split #1

3

(bp) 650

2

500

Split #2

Split #3

WT

(–)

17

n.d.

21

n.d.

400 10

10

Split #1 430–431 (Linker)

1

0

300 Split #2 739–740 (RuvC)

200

-1

Split #3 801–802 (WED)

100 Indel (%) 10

I

– inducer (e.g. ABA)

+ inducer (e.g. ABA)

J

25

DA NES

Nucleus

DB NLS

NLS

Cytoplasm Nucleus

DB NLS

EMX1 indel (%)

Cytoplasm

DA NES

Assembly of split SaCas9

DB NLS

NLS

AC

Rapa

20

SaCas9(N)-DA-NES SaCas9(C)-DB-2xNLS

ABA

+ AC – AC

H

+ ABA – ABA

Sa

_ Sa ASC _A L Sp SC 1-s _A L1 g1 SC -s L1 g2 Sa (+ _M ) Sa Y _M OD Sp YO 1-s _M D g1 YO 1-s D g2 1 (+ ) G FP (– )

10

15 10 5

NLS

PAM

NLS

PAM

0

WT Control

DB NLS

+ Rapa – Rapa

DA NES

Genomic target

(legend on next page)

1124 Cell 162, 1113–1126, August 27, 2015 ª2015 Elsevier Inc.

EXPERIMENTAL PROCEDURES

AUTHOR CONTRIBUTIONS

Detailed experimental procedures are described in the Supplemental Experimental Procedures. The full-length S. aureus Cas9 N580A/C946A mutant (residues 1–1053) was expressed in Escherichia coli Rosetta 2 (DE3) (Novagen) and purified by chromatography on Ni-NTA Superflow (QIAGEN), Mono S (GE Healthcare), and HiLoad Superdex 200 16/60 (GE Healthcare) columns. The SeMet-labeled protein was expressed in E. coli B834 (DE3) (Novagen) and purified using a similar protocol as that for the native protein. The sgRNA was transcribed in vitro with T7 RNA polymerase and purified by 8% denaturing (7 M urea) PAGE. The target DNAs were purchased from Sigma-Aldrich. The purified SaCas9 protein was mixed with the sgRNA, target DNA strand, and non-target DNA strand (molar ratio, 1:1.5:2.3:3.4), and the SaCas9–sgRNA–target DNA complex was purified by gel filtration chromatography on a Superdex 200 Increase column (GE Healthcare). The purified SaCas9–sgRNA–target DNA complex (containing either the 50 -TTGAAT-30 PAM or the 50 -TTGGGT-30 PAM) was crystallized at 20 C by the hanging-drop vapor diffusion method. Crystals were obtained by mixing 1 ml of the complex solution (A260 nm, 15) and 1 ml of the reservoir solution (10%–12% PEG 4,000, 0.75 M NaCl, 0.15 M Na2HPO4 and 0.15 M NaN3). The SeMet-labeled complex (containing the 50 -TTGGGT-30 PAM) was crystallized under similar conditions. X-ray diffraction data were collected at 100 K on the beamlines BL32XU and BL41XU at SPring-8 (Hyogo, Japan). The structure was determined by the Se-SAD method, using the 3 A˚ resolution dataset from the SeMet-labeled crystal. The final models of the 50 -TTGAAT-30 PAM complex (2.6 A˚ resolution) and the 50 -TTGGGT-30 PAM complex (2.7 A˚ resolution) were refined using the native datasets. Molecular graphics images were prepared using CueMol (http://www.cuemol.org). Around 24 hr prior to transfection, HEK293FT cells (Life Technologies) were seeded into 24-well plates (Corning) at a density of 2.5 3 105 cells/well and transfected at 70%–80% confluency using Lipofectamine 2000 (Life Technologies), according to the manufacturer’s recommended protocol. A total of 600 ng DNA was used for each well of the 24-well plate. About 72 hr after transfection, the genomic DNA was extracted, and then the genomic modifications were examined using the SURVEYOR assay and targeted deep sequencing, as previously described (Cong et al., 2013).

H.N. solved the crystal structures and performed in vitro cleavage experiments; L.C., W.X.Y., F.A.R., B.Z., and Y.L. performed in vivo functional analyses; A.K. assisted with the plasmid construction; R.I. assisted with the structural analysis; and H.N., L.C., W.X.Y., F.A.R., F.Z., and O.N. wrote the manuscript with help from all authors. H.N., F.Z., and O.N. directed and supervised all of the research.

ACCESSION NUMBERS The accession numbers for the atomic coordinates of the SaCas9–sgRNA– target DNA complexes reported in this paper are Protein Data Bank: 5CZZ (50 -TTGAAT-30 PAM) and 5AXW (50 -TTGGGT-30 PAM). SUPPLEMENTAL INFORMATION Supplemental Information includes Supplemental Experimental Procedures, seven figures, and one table and can be found with this article online at http://dx.doi.org/10.1016/j.cell.2015.08.007.

ACKNOWLEDGMENTS We thank Tomohiro Nishizawa, Takanori Nakane, Kazuki Kato, Naoki Matsumoto, and Rhiannon Macrae for helpful comments on the manuscript. We thank the beam-line staffs at BL32XU and BL41XU of SPring-8, Japan, for assistance with data collection. H.N. is supported by JST, PRESTO, JSPS KAKENHI Grant Numbers 26291010 and 15H01463 and the Platform for Drug Discovery, Informatics, and Structural Life Science from the Ministry of Education, Culture, Sports, Science and Technology. W.X.Y. is supported by T32GM007753 from the National Institute of General Medical Sciences and a Paul and Daisy Soros Fellowship. F.A.R. is a Junior Fellow at the Harvard Society of Fellows. F.Z. is supported by an NIH Director’s Pioneer Award (1DP1-MH100706), the Keck, McKnight, Poitras, Merkin, Vallee, Damon Runyon, Searle Scholars, Klingenstein, and Simons Foundations, Bob Metcalfe, and Jane Pauley. O.N. is supported by the Basic Science and Platform Technology Program for Innovative Biological Medicine from the Japan Agency for Medical Research and Development, AMED, and the Council for Science, Technology and Innovation (CSTI), Cross-ministerial Strategic Innovation Promotion Program (SIP), ‘‘Technologies for creating next-generation agriculture, forestry and fisheries’’ (funding agency: Bio-oriented Technology Research Advancement Institution, NARO), and the Platform for Drug Discovery, Informatics, and Structural Life Science from the Ministry of Education, Culture, Sports, Science and Technology. A patent application has been filed related to this work, and the authors plan to make the reagents widely available to the academic community through Addgene and to provide software tools via the Zhang lab web site (www.genome-engineering.org). Received: July 6, 2015 Revised: August 1, 2015 Accepted: August 6, 2015 Published: August 27, 2015 REFERENCES Anders, C., Niewoehner, O., Duerst, A., and Jinek, M. (2014). Structural basis of PAM-dependent target DNA recognition by the Cas9 endonuclease. Nature 513, 569–573. Banaszynski, L.A., Liu, C.W., and Wandless, T.J. (2005). Characterization of the FKBP.rapamycin.FRB ternary complex. J. Am. Chem. Soc. 127, 4715–4721.

Figure 7. Structure-Guided Engineering of the SaCas9 System and Its Applications (A) Schematic of the three-component SaCas9 activator system. (B) Design of the dSaCas9-based transcriptional activator and the reporter system. NLS, nuclear localization signal; 2A, 2A self-cleaving peptide; CMV, cytomegalovirus. (C) Representative fluorescent microscopy images, showing the activation of the reporter gene and the low background level of the newly designed reporter. Scale bar, 100 mm. (D) Optimization of different fusion scaffolds for the dSaCas9-based activator, using the reporter system. MS2-SL, MS2-binding stem loop. ‘‘IS’’ denotes the insertion site of the MS2-SL. (E) Transcriptional activation of the endogenous target ASCL1 and MYOD1 genes by dSaCas9-based activators, with previously described dSpCas9-based activators as positive controls. (F) Structure-guided design of split-SaCas9 systems. Split sites 1 and 2 were designed at flexible linker regions, whereas split site 3 was designed at the b strand in the WED domain, as a negative control. The sgRNA and target DNA molecules were removed from the original structure for clarity. (G) Cleavage activity of wild-type (WT) SaCas9 and the three different auto-assembled split-SaCas9 designs. (H and I) Schematic of the inducible SaCas9 system. NES, nuclear export signal; ABA, abscisic acid; DA, dimerization domain A; DB, dimerization domain B. (J) Cleavage activity of the wild-type (WT) SaCas9 and three different inducible SaCas9 systems. ‘‘ABA’’ denotes the inducible SaCas9 design based on the abscisic acid sensing system; ‘‘AC’’ denotes the design based on the A/C heterodimerizer; ‘‘Rapa’’ denotes the design based on the FRB/FKBP system (see Experimental Procedures section for additional details).

Cell 162, 1113–1126, August 27, 2015 ª2015 Elsevier Inc. 1125

Barrangou, R., Fremaux, C., Deveau, H., Richards, M., Boyaval, P., Moineau, S., Romero, D.A., and Horvath, P. (2007). CRISPR provides acquired resistance against viruses in prokaryotes. Science 315, 1709–1712.

Jiang, F., Zhou, K., Ma, L., Gressel, S., and Doudna, J.A. (2015). A Cas9-guide RNA complex preorganized for target DNA recognition. Science 348, 1477– 1481.

Biertu¨mpfel, C., Yang, W., and Suck, D. (2007). Crystal structure of T4 endonuclease VII resolving a Holliday junction. Nature 449, 616–620.

Jinek, M., Chylinski, K., Fonfara, I., Hauer, M., Doudna, J.A., and Charpentier, E. (2012). A programmable dual-RNA-guided DNA endonuclease in adaptive bacterial immunity. Science 337, 816–821.

Bolotin, A., Quinquis, B., Sorokin, A., and Ehrlich, S.D. (2005). Clustered regularly interspaced short palindrome repeats (CRISPRs) have spacers of extrachromosomal origin. Microbiology 151, 2551–2561. Briner, A.E., Donohoue, P.D., Gomaa, A.A., Selle, K., Slorach, E.M., Nye, C.H., Haurwitz, R.E., Beisel, C.L., May, A.P., and Barrangou, R. (2014). Guide RNA functional modules direct Cas9 activity and orthogonality. Mol. Cell 56, 333–339. Brouns, S.J., Jore, M.M., Lundgren, M., Westra, E.R., Slijkhuis, R.J., Snijders, A.P., Dickman, M.J., Makarova, K.S., Koonin, E.V., and van der Oost, J. (2008). Small CRISPR RNAs guide antiviral defense in prokaryotes. Science 321, 960–964. Cho, S.W., Kim, S., Kim, J.M., and Kim, J.S. (2013). Targeted genome engineering in human cells with the Cas9 RNA-guided endonuclease. Nat. Biotechnol. 31, 230–232. Chylinski, K., Le Rhun, A., and Charpentier, E. (2013). The tracrRNA and Cas9 families of type II CRISPR-Cas immunity systems. RNA Biol. 10, 726–737. Chylinski, K., Makarova, K.S., Charpentier, E., and Koonin, E.V. (2014). Classification and evolution of type II CRISPR-Cas systems. Nucleic Acids Res. 42, 6091–6105. Cong, L., Zhou, R., Kuo, Y.C., Cunniff, M., and Zhang, F. (2012). Comprehensive interrogation of natural TALE DNA-binding modules and transcriptional repressor domains. Nat. Commun. 3, 968. Cong, L., Ran, F.A., Cox, D., Lin, S., Barretto, R., Habib, N., Hsu, P.D., Wu, X., Jiang, W., Marraffini, L.A., and Zhang, F. (2013). Multiplex genome engineering using CRISPR/Cas systems. Science 339, 819–823. Deltcheva, E., Chylinski, K., Sharma, C.M., Gonzales, K., Chao, Y., Pirzada, Z.A., Eckert, M.R., Vogel, J., and Charpentier, E. (2011). CRISPR RNA maturation by trans-encoded small RNA and host factor RNase III. Nature 471, 602–607. Deveau, H., Barrangou, R., Garneau, J.E., Labonte´, J., Fremaux, C., Boyaval, P., Romero, D.A., Horvath, P., and Moineau, S. (2008). Phage response to CRISPR-encoded resistance in Streptococcus thermophilus. J. Bacteriol. 190, 1390–1400. Fonfara, I., Le Rhun, A., Chylinski, K., Makarova, K.S., Le´crivain, A.L., Bzdrenga, J., Koonin, E.V., and Charpentier, E. (2014). Phylogeny of Cas9 determines functional exchangeability of dual-RNA and Cas9 among orthologous type II CRISPR-Cas systems. Nucleic Acids Res. 42, 2577–2590. Garneau, J.E., Dupuis, M.E., Villion, M., Romero, D.A., Barrangou, R., Boyaval, P., Fremaux, C., Horvath, P., Magada´n, A.H., and Moineau, S. (2010). The CRISPR/Cas bacterial immune system cleaves bacteriophage and plasmid DNA. Nature 468, 67–71. Gasiunas, G., Barrangou, R., Horvath, P., and Siksnys, V. (2012). Cas9-crRNA ribonucleoprotein complex mediates specific DNA cleavage for adaptive immunity in bacteria. Proc. Natl. Acad. Sci. USA 109, E2579–E2586.

Jinek, M., East, A., Cheng, A., Lin, S., Ma, E., and Doudna, J. (2013). RNA-programmed genome editing in human cells. eLife 2, e00471. Jinek, M., Jiang, F., Taylor, D.W., Sternberg, S.H., Kaya, E., Ma, E., Anders, C., Hauer, M., Zhou, K., Lin, S., et al. (2014). Structures of Cas9 endonucleases reveal RNA-mediated conformational activation. Science 343, 1247997. Konermann, S., Brigham, M.D., Trevino, A.E., Hsu, P.D., Heidenreich, M., Cong, L., Platt, R.J., Scott, D.A., Church, G.M., and Zhang, F. (2013). Optical control of mammalian endogenous transcription and epigenetic states. Nature 500, 472–476. Konermann, S., Brigham, M.D., Trevino, A.E., Joung, J., Abudayyeh, O.O., Barcena, C., Hsu, P.D., Habib, N., Gootenberg, J.S., Nishimasu, H., et al. (2015). Genome-scale transcriptional activation by an engineered CRISPRCas9 complex. Nature 517, 583–588. Liang, F.S., Ho, W.Q., and Crabtree, G.R. (2011). Engineering the ABA plant stress pathway for regulation of induced proximity. Sci. Signal. 4, rs2. Makarova, K.S., Grishin, N.V., Shabalina, S.A., Wolf, Y.I., and Koonin, E.V. (2006). A putative RNA-interference-based immune system in prokaryotes: computational analysis of the predicted enzymatic machinery, functional analogies with eukaryotic RNAi, and hypothetical mechanisms of action. Biol. Direct 1, 7. Mali, P., Yang, L., Esvelt, K.M., Aach, J., Guell, M., DiCarlo, J.E., Norville, J.E., and Church, G.M. (2013). RNA-guided human genome engineering via Cas9. Science 339, 823–826. Marraffini, L.A., and Sontheimer, E.J. (2008). CRISPR interference limits horizontal gene transfer in staphylococci by targeting DNA. Science 322, 1843– 1845. Mojica, F.J., Dı´ez-Villasen˜or, C., Garcı´a-Martı´nez, J., and Soria, E. (2005). Intervening sequences of regularly spaced prokaryotic repeats derive from foreign genetic elements. J. Mol. Evol. 60, 174–182. Mojica, F.J., Dı´ez-Villasen˜or, C., Garcı´a-Martı´nez, J., and Almendros, C. (2009). Short motif sequences determine the targets of the prokaryotic CRISPR defence system. Microbiology 155, 733–740. Nishimasu, H., Ran, F.A., Hsu, P.D., Konermann, S., Shehata, S.I., Dohmae, N., Ishitani, R., Zhang, F., and Nureki, O. (2014). Crystal structure of Cas9 in complex with guide RNA and target DNA. Cell 156, 935–949. Pourcel, C., Salvignol, G., and Vergnaud, G. (2005). CRISPR elements in Yersinia pestis acquire new repeats by preferential uptake of bacteriophage DNA, and provide additional tools for evolutionary studies. Microbiology 151, 653–663. 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., et al. (2015). In vivo genome editing using Staphylococcus aureus Cas9. Nature 520, 186–191. Sapranauskas, R., Gasiunas, G., Fremaux, C., Barrangou, R., Horvath, P., and Siksnys, V. (2011). The Streptococcus thermophilus CRISPR/Cas system provides immunity in Escherichia coli. Nucleic Acids Res. 39, 9275–9282.

Go´recka, K.M., Komorowska, W., and Nowotny, M. (2013). Crystal structure of RuvC resolvase in complex with Holliday junction substrate. Nucleic Acids Res. 41, 9945–9955.

Sternberg, S.H., Redding, S., Jinek, M., Greene, E.C., and Doudna, J.A. (2014). DNA interrogation by the CRISPR RNA-guided endonuclease Cas9. Nature 507, 62–67.

Hsu, P.D., Scott, D.A., Weinstein, J.A., Ran, F.A., Konermann, S., Agarwala, V., Li, Y., Fine, E.J., Wu, X., Shalem, O., et al. (2013). DNA targeting specificity of RNA-guided Cas9 nucleases. Nat. Biotechnol. 31, 827–832.

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. USA 112, 2984–2989.

Hwang, W.Y., Fu, Y., Reyon, D., Maeder, M.L., Tsai, S.Q., Sander, J.D., Peterson, R.T., Yeh, J.R., and Joung, J.K. (2013). Efficient genome editing in zebrafish using a CRISPR-Cas system. Nat. Biotechnol. 31, 227–229.

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.

Jiang, W., Bikard, D., Cox, D., Zhang, F., and Marraffini, L.A. (2013). RNAguided editing of bacterial genomes using CRISPR-Cas systems. Nat. Biotechnol. 31, 233–239.

Zhang, F., Cong, L., Lodato, S., Kosuri, S., Church, G.M., and Arlotta, P. (2011). Efficient construction of sequence-specific TAL effectors for modulating mammalian transcription. Nat. Biotechnol. 29, 149–153.

1126 Cell 162, 1113–1126, August 27, 2015 ª2015 Elsevier Inc.