Structural basis of Staphylococcus aureus Cas9 inhibition by AcrIIA14

Abstract Bacteriophages have evolved a range of anti-CRISPR proteins (Acrs) to escape the adaptive immune system of prokaryotes, therefore Acrs can be used as switches to regulate gene editing. Herein, we report the crystal structure of a quaternary complex of AcrIIA14 bound SauCas9–sgRNA–dsDNA at 2.22 Å resolution, revealing the molecular basis for AcrIIA14 recognition and inhibition. Our structural and biochemical data analysis suggest that AcrIIA14 binds to a non-conserved region of SauCas9 HNH domain that is distinctly different from AcrIIC1 and AcrIIC3, with no significant effect on sgRNA or dsDNA binding. Further, our structural data shows that the allostery of the HNH domain close to the substrate DNA is sterically prevented by AcrIIA14 binding. In addition, the binding of AcrIIA14 triggers the conformational allostery of the HNH domain and the L1 linker within the SauCas9, driving them to make new interactions with the target-guide heteroduplex, enhancing the inhibitory ability of AcrIIA14. Our research both expands the current understanding of anti-CRISPRs and provides additional culues for the rational use of the CRISPR-Cas system in genome editing and gene regulation.


INTRODUCTION
The CRISPR-Cas system is an adaptive immune system that bacteria and archaea use to resist the invasion of exogenous nucleic acids. These systems are classified as class 1 and class 2, according to the number of effector proteins involved in immunity (1)(2)(3). In class 1 (includes Type I, III and IV), multiple effector proteins are involved in the effector phase. In class 2 (includes Type II, V and VI), only one effector protein is involved in the effector phase and that makes it more widely used in gene editing since its use is more convenient (4,5). Among these, the most widely used is the Type II Cas9 system (6,7). The CRISPR-Cas9 sys-tem has two functional components: one recognition (REC) lobe and one nuclease (NUC) lobe (8)(9)(10)(11). By targeting a specific genomic sequence complementary to a single guide RNA (sgRNA) and a protospacer adjacent motif (PAM), the HNH domain is primarily responsible for cleaving the complementary strand and the RuvC domain for cleaving non-complementary strands (8)(9)(10)(11)(12). The most widely used Type II Cas9 system is Streptococcus pyogenes Cas9 (Spy-Cas9) (5,9). However, its large molecular weight (150kDa) can hamper cell transfection efficiency and therefore affects the application in gene editing (5). Therefore, many researchers are currently looking for more pocket-sized editing proteins that can replace SpyCas9. A good candidate is Staphylococcus aureus Cas9 (SauCas9), which is ∼300 amino acids smaller than SpyCas9 (13,14). It was confirmed in previous research that SauCas9 can edit the genome of mammalian cells, potentially with high efficiency and specificity (14).
In this work, we investigated the mechanism by which AcrIIA14, an anti-CRISPR protein from Staphylococcus, inhibits SauCas9 activity (29). AcrIIA14 consists of an Nterminal HTH domain and a C-terminal domain which has previously been reported to inhibit SauCas9 activ-ity in vivo and in vitro without affecting the binding of SauCas9 protein to sgRNA and target DNA (29) but its specific inhibitory mechanism is still unclear. We further validated the mechanism of this anti-CRISPR protein inhibitting SauCas9 via biochemical assays and structural biology methods. Our results widen our understanding of the diverse anti-CRISPR inhibitory mechanisms and support the use of SauCas9 as a gene editing tool.
To assemble the SauCas9-sgRNA-AcrIIA14ct-dsDNA complex, SauCas9 (N580A/C946A) protein was incubated with sgRNA (73 nt), AcrIIA14ct and dsDNA at the molar ratio of 1:1.2:5:2. Components were added in the order listed and incubated at room temperature for 5 min and 4 • C for 40 min before adding the next component. The complex was applied onto size-exclusion chromatography (Superose 6 increase 10/300 GL, GE Healthcare) with buffer (10 mM Tris-HCl, pH 8.0, 300 mM NaCl, 3 mM DTT) to remove excess sgRNA, dsDNA and AcrIIA14ct. Finally, complexes were concentrated to an A 280 absorbance to 18, as measured by Nanodrop One, before crystallization.

In vitro transcription and purification of sgRNA
In vitro transcription with T7 RNA polymerase and template (dsDNA) for sgRNA was generated by PCR. All relevant sequences are listed in Supplementary Table S1.Transcription reactions were performed at 37 • C for 4 h in buffer containing 0.1 M HEPES-K, pH 7.9, 12 mM MgCl 2 , 30 mM DTT, 2 mM spermidine, 2 mM each NTP (ATP, UTP, GTP, CTP), 100 g/ml T7 polymerase and 500 nM transcription template. The sgRNA was purified by gel electrophoresis on a 8% denatured (8 M urea) polyacrylamide gel and followed by running Elutrap system. Finally, the sgRNA was resuspended in DEPC (diethylpyrocarbonate) H 2 O, stored at -80 • C.

Crystallization and structure determination and refinement
The hanging-drop vapor-diffusion method was used for crystal growth. Crystallization conditions were screened on a large scale with mosquito robot and then optimized according to the results of the primary screening. All crystals used for diffraction were obtained by mixing 1 l SauCas9 complex with 1 l reservoir solution and incubated at 20 • C for 7-14 days. Crystals of the SauCas9-sgRNA-AcrIIA14ct-dsDNA complex were grown from 1.0 M sodium citrate dihydrate, 0.3 M imidazol-HCl, pH 10.0, 0.02 M sodium malonate, pH 7.0.
Before data collection, the crystals were transferred into cryo-protectant buffer (the crystallization buffer containing 20% (w/v) glycerol) and flash-cooled in liquid nitrogen. Xray diffraction data were collected at beamline BL-17U1 at Shanghai Synchrotron Radiation Facility (SSRF). All images were collected at the wavelength of 0.9791Å with 0.5 • rotation. The diffraction data for SauCas9-sgRNA-AcrIIA14ct-dsDNA was processed with HKL2000. The structure was determined by molecular replacement (MR) with the program PHASER. Each domain of the SauCas9 (PDB 5CZZ) was used as an individual search model for MR. The initial model was improved by several rounds of structural refinement in PHENIX. The final model was validated through MOLPROBITY and PROCHECK. Data collection and structural refinement statistics were listed in supplementary information, Supplementary Table S2. All of the structural figures were prepared using PyMOL.

In vitro DNA cleavage assays
The in vitro DNA cleavage assays were performed in 20 l system containing 500 nM Cas9 RNP complex, 25 nM pUC19 plasmid in the buffer (20 mM HEPES, pH 7.55, 100 mM KCl, 10 mM MgCl 2 , 1 mM DTT and 5% (w/v) glycerol). The pUC19 target DNA plasmid (target DNA with PAM sequence cloned into the pUC19 vector in Ssp1 restriction site) was linearized by EcoR1 digestion before the cleavage reaction. To test AcrIIA14 mediated inhibition of Cas9 to cleave target DNA, pre-incubated of AcrIIA14 with Cas9-sgRNA complex for 5 min at 37 • C and molar ratios of Cas9 to AcrIIA14 ranging from 1:0 to 1:4. The cleavage assays were performed at 37 • C for 30 min. Reactions were quenched by adding 6× TBE-urea gel loading and ran on 1% agarose gels stained with ethidium bromide for production detection.

Biolayer interferometry binding assays
All ffinities were measured by biolayer interferometry (BLI) using the ctet RED96 system (FortéBio). All experiments were performed at 20 • C with buffer containing 20 mM HEPES, pH 7.55, 100 mM KCl, 10 mM MgCl 2 , 1 mM DTT and 5% (w/v) glycerol. Streptavidin sensors were preequilibrated in the buffer for at least 10 min before used in experiments. Biotinylated SauCas9 protein (or SauCas9-sgRNA complex) were loaded onto streptavidin biosensors for 60 s. Binding kinetics were determined from binding data obtained with four or five concentrations of sample. The interference from the biotinylated protein with buffer were analysised as a control. Finally, we calculated the binding constant using Octet data analysis software.

Overall structure of AcrIIA14ct bound SauCas9-sgRNA-dsDNA complex
It was previously reported that three inhibitors of SauCas9, AcrIIA13-AcrIIA15, that share a conserved N-terminal HTH domain which can interact directly with the promoterproximal sequences of the anti-CRISPR protein locus and plays a role in regulating protein expression, but is not necessary for inhibiting the cleavage activity of SauCas9 (Figure S1A). The C-terminal domain in contrast, here referred to as AcrIIA14ct, is required and spans residues 60-159 ( Figure 1A) (29).
In order to elucidate the mechanism of SauCas9 inhibition by AcrIIA14, we determined the crystal structure of SauCas9-sgRNA-AcrIIA14ct-dsDNA quaternary complex at 2.22Å resolution (Figure 1 and Supplementary  Table S2). In the complex, SauCas9 adopts a bi-lobed structure, comprising recognition (REC) and nuclease (NUC) lobes. The channel formed by these two lobes binds the sgRNA-target DNA heteroduplex, as described previously (14). AcrIIA14ct binds to the SauCas9 complex with a 1:1 stoichiometry partially interacting with the L2 linker and making extensive contacts with the HNH domain ( Figure  1B), which suggests that this interaction is responsible for SauCas9 inhibition.
Close inspection of the interface between AcrIIA14ct and the HNH domain shows that ␤2, loop ␣2-␤4, loop ␤4-␣3 and C-terminus of AcrIIA14ct make extensive intermolecular hydrogen bond contacts with ␣4-␣5 of the HNH domain and the loop between the HNH domain and the L2 linker ( Figure 2A). Specifically, residues Arg18, Tyr20 and Asn22 from the ␤2 sheet of AcrIIA14ct make hydrogen bonds with residues Ser593, Ser594, Ser595 and Glu602 from ␣4-␣5 of SauCas9 HNH domain ( Figure 2B). In addition, two residues in ␣2-␤4 loop of AcrIIA14ct, His56 and Lys57, make four hydrogen bonds with ␣4 Gln590 in the HNH domain and Glu629, Asp631 in the loop between the HNH domain and the L1 linker, respectively ( Figure 2D). This implies that His56 may be important for AcrIIA14ct recognition of SauCas9. Notably, the protruding ␤4-␣3 loop in AcrIIA14ct is close to the loop ␣4-␣5 in the HNH domain. And residues Gln61, Lys64, Gly71 and Ile72 in AcrIIA14ct interact with residues Ser595, Asp596 and Ser597 of ␣4-␣5 in the HNH domain ( Figure 2E). Moreover, the Cterminus of AcrIIA14ct further strengthens the contact through the backbone between residues Asn97 and Leu98 interacting with Ser600 and Thr603 of ␣5 in the HNH domain ( Figure 2C). Consistent with these observations, we found that mutations around the AcrIIA14ct-SauCas9 interface hampered interaction and reducded the effectiveness of AcrIIA14ct as a SauCas9 inhibitor ( Figure 2F-I and Supplementary Figure S1C). Notably, AcrIIA14ct mutant H56K significantly reduced the affinity to SauCas9 by four orders of magnitude (K D = 58.0 nM versus 5.58 pM for mutant and wild type, respectively) ( Supplementary Figure S1C), implying a destabilization of the AcrIIA14ct and Saucas9 complex. Consistent with this, in vitro cleavage assays showed the mutant H56K of AcrIIA14ct almost abolished inhibition ( Figure 2F and G). Overall, these data indicate that inhibition of SauCas9 is caused by direct interaction between AcrIIA14ct and the HNH domain of SauCas9.

AcrIIA14ct binding induces SauCas9 allostery
To explore how AcrIIA14 binding affects the conformation of SauCas9, we overlapped the structures of SauCas9 when bound to AcrIIA14ct and when bound to dsDNA (PDB 5CZZ). This revealed that AcrIIA14 binding only induces significant conformational changes in the HNH domain and the L1 linker ( Figure 3A). In detail, binding of AcrIIA14ct induces HNH domain and L1 linker to approach the target-guide heteroduplex. As a result, the L1 linker becomes two short ␣ helices, whereas the former helix, together with the HNH domain, become closer to the target-guide heteroduplex ( Figure 3B). We also found that allostery between these two domains allows the formation of two new interaction surfaces inside the SauCas9 protein complex. First, residue Arg617 in the loop between ␣5 and ␣6 of HNH domain hydrogen binds the sugarphosphate backbone of the target DNA strand (fifteenth nucleotide upstream of the PAM sequence) ( Figure 3C). Second, residues Lys485 and Lys489 in the L1 linker interact with the phosphate backbone between the 11th and 12th nucleotides of the sgRNA ( Figure 3D). Accordingly, we speculated that these novel intramolecular interactions induced by AcrIIA14ct binding may act to stabilize the inhibitory conformation. To test this, we mutated these three residues separately. As expected, an in vitro DNA cleavage assay showed that SauCas9 mutants R617G, K485G and K489G reduced sensitivity to AcrIIA14ct inhibition (Figure 2H and I). In addition, analysis of our structure reveals that residue Asp486 is very important for maintaining the spatial orientation of Lys485 and Lys489 on the L1 linker ( Figure 3D). As expected, the inhibitory effect of AcrIIA14ct was also reduced when we introduced the mutation D486A ( Figure 2H and I). Taken together, these results suggest that AcrIIA14ct binding induces SauCas9 allostery and that formation of two novel intramolecular interactions further enhance AcrIIA14ct inhibition of SauCas9.

Mechanism of AcrIIA14 inhibition of SauCas9
In the crystal structure of SauCas9-sgRNA-AcrIIA14ct-dsDNA, the distance between the catalytic center of the HNH domain (D556, H557) and the scissile phosphodiester linkage in the target DNA strand (between the third and fourth nucleotides upstream of the PAM sequence) is ∼33.3Å ( Figure 4A). This suggests that, in this structure, SauCas9 is in the precatalytic state. However, in the SauCas9-sgRNA-dsDNA crystal structure (PDB: 5CZZ) the distance between the HNH domain and the cleavage site is about 46.2Å ( Figure 4A), indicating that the binding of AcrIIA14ct brings the HNH domain ∼13Å closer to the cleavage site.
According to the previously reported Cas9 activation model, sequential binding of sgRNA and target DNA, the interaction between non-target strand and RuvC domain destroy the fragile stability of the original conformation between the RuvC, L2, L1 and HNH domain, which drives the movement of HNH domain to the cleavage site (30,31). In the AcrIIA14ct bound SauCas9 structure, AcrIIA14ct mainly binds to the region close to the RuvC domain on the HNH domain and makes also some interaction with the anterior segment of the L2 linker ( Figure 1B). Therefore, we speculated that the binding of AcrIIA14ct interferes the interaction between the HNH domain and the RuvC domain, L1 and L2 linker. This leads to the proximity of HNH domain to the target-guide heteroduplex, just like the movement of HNH domain during activation. This large movement further facilitates the allostery of the L1 linker.
As described above, during Cas9 activation the HNH domain undergoes a large-scale rotation and translocation. Based on the previously solved structure of Cas9 in the activated state (11,30), we simulated the approximate conformation and location of the HNH domain after SauCas9 activation ( Figure 4B). In this simulated structure, there is a steric clash between AcrIIA14ct and the WED domain, indicating that AcrIIA14ct binding to HNH domain sterically prevents the HNH domain from transitioning to the catalytic state. It is known that the L1 linker plays a very critical role in the movement of the HNH domain to the cleavage site (11,14), therefore the change in the position and conformation of the L1 linker caused by the binding of AcrIIA14ct also prevents activation of SauCas9.
From the above structural analysis, AcrIIA14ct binding leads to the formation of two new interactions between the HNH domain and L1 linker and the target-guide heteroduplex in the SauCas9 complex. Mutating the residues that mediate these interactions reduced the effectiveness of AcrIIA14ct (Figure 2H and I). Therefore, we speculate that these novel interactions stabilize the inhibitory state of SauCas9, further improving the inhibitory activity of AcrIIA14ct. This is consistent with its robust inhibitory capacity (Supplementary Figure S1A). Of course, more direct assays are needed to verify these conclusions. Together, our structural and biochemical results demonstrate that AcrIIA14ct inhibits SauCas9 activity by directly binding to its HNH domain and preventing its movement towards the cleavage site ( Figure 4B and C). In addition, we speculate that the inhibitory activity of AcrIIA14 is enhanced by stabilizing a SauCas9 protein inhibited state.

Comparison of AcrIIA14 with AcrIIC1 and AcrIIC3
Like AcrIIA14, AcrIIC1 and AcrIIC3 prevent Cas9 protein from cleaving target DNA via direct binding to the HNH domain, whereas their action does not interfere with sgRNA and target DNA binding (11,26). To compare their Cas9 inhibition mechanisms, we superimposed structures of SauCas9-sgRNA-AcrIIA14ct-dsDNA, NmeCas9 HNH domain-AcrIIC1 and NmeCas9-sgRNA-AcrIIC3 ( Figure 5A). Three Acr proteins display distinct overall architectures and bind at different positions of the HNH domain. For example, the ␤ sheet of AcrIIC1 inserts into the HNH domain active groove ( Figure 5A) preventing the catalytic center from accessing the target DNA and likely excluding the divalent cation. In contrast, AcrIIC3 binds to a surface of the HNH domain close to the target-guide heteroduplex ( Figure 5A). According to a previous structure of Nme1Cas9-sgRNA-AcrIIC3, two AcrIIC3 proteins aggregation two Nme1Cas9 proteins together, each AcrIIC3 protein interacts with the HNH domain of one Nme1Cas9 and the REC2 domain of the other. Thus, the two AcrIIC3 proteins lock the position of the HNH domain by bridging the two Cas9 proteins in a quaternary complex, hindering rotation towards the target DNA strand(Supplementary Figure  S2A).
In contrast to AcrIIC3, AcrIIA14 binds on a surface of the HNH domain far from the nucleic acid heteroduplex, interacting with the protrusion formed by the two ␣ helices of the HNH domain ( Figure 5A). This sterically prevents the HNH domain from moving to the cleavage site of target DNA. In addition, it drives a significant change in the conformation of SauCas9 HNH domain and L1 linker, enhancing its inhibitory ability, as described above. However, structural superposition of Cas9 complexes bound to AcrIIC3 or sgRNA show nearly identical structures of NmeCas9-sgRNA in these two complexes (Supplementary Figure S2B), indicating that AcrIIC3 binding does not induce a conformational change of NmeCas9 and sgRNA. Since the structure of intact Cas9 protein bound to AcrIIC1 is not available, it is unclear whether AcrIIC1 binding can also induce changes in the Cas9 protein. Taken together, comparison of the structures of AcrIIA14, AcrIIC1 and AcrIIC3 bound Cas9 complex, shows that AcrIIA14 binds to a different surface of the HNH domain, enhancing its inhibitory ability by inducing Cas9 protein allostery, in contrast to AcrIIC1 and AcrIIC3.
Further, we also compared the surfaces of the HNH domain in the three Type II Cas9 proteins, SauCas9, NmeCas9 and SpyCas9, recognized by the three Acr proteins. Sequence conservation analysis shows that only the residues bound by AcrIIC1 are highly conserved ( Figure  5C). Structural comparison of these three HNH domains also show that the catalytic center bound by AcrIIC1 is conserved, whereas the regions bound by AcrIIA14 and AcrIIC3 are not ( Figure 5B and Supplementary Figure  S2C). This is consistent with our in vitro cleavage assays (Supplementary Figure S1A and S1B) and previous reports that show that AcrIIA14 and AcrIIC3 specifically inhibit SauCas9 and NmeCas9, respectively, whereas AcrIIC1 can inhibit diverse Cas9 orthologs. In summary, in contrast with AcrIIC1 and AcrIIC3, AcrIIA14 identifies a novel nonconserved HNH surface and inhibits specifically SauCas9 activity.

DISCUSSION
In the present work, we have solved the crystal structure of the quaternary complex formed by SauCas9-sgRNA-AcrIIA14ct-dsDNA at 2.22Å resolution. This high-resolution structure allows us to clearly describe the mechanism of SauCas9 inhibition by AcrIIA14. Our results indicate that AcrIIA14ct binds to SauCas9 HNH domain. This binding sterically prevents the movement of the HNH domain toward the scissile phosphodiester linkage in the target DNA strand ( Figure 4B and C). At the same time, it positions the HNH domain and the L1 linker closer to the target-guide heteroduplex, further improving its inhibitory ability. In contrast to AcrIIC1 and AcrIIC3, AcrIIA14ct binds to a novel non-conserved region of the HNH domain, which results in AcrIIA14ct inhibition being specific to SauCas9. The fact that diverse Acr proteins can target different surfaces of the HNH domain suggests that HNH allostery is very important during Cas9 protein activation. Additionally, it may be a mechanism used by phages to evade bacterial anti-anti-CRISPR strategies in this ongoing arms race.
AcrIIA14ct is able to bind SauCas9 apo, sgRNA-bound or DNA-bound precatalytic SauCas9 ( Figure 4C). This is consistent with the fact that the HNH domain is not involved in sgRNA or target DNA binding (9,31,32). The ability of AcrIIA14 to bind Cas9 protein in multiple states may enhance the inhibition of bacterial CRISPR systems during phage invasion. We also found that the affinity of SauCas9 to AcrIIA14ct (K D = 5.58 pM) is significantly higher than to sgRNA (K D = 473 pM) or to target DNA (535 pM) (Supplementary Figure S1C). This allows AcrIIA14 to inhibit Cas9 protein before its activation by sgRNA and target DNA, ensuring that phages win the arms race with bacteria. In a wider context, our studies expand the understanding of the diverse anti-CRISPR inhibitory mechanisms.

DATA AVAILABILITY
The atomic coordinates of the crystal structure of SauCas9-sgRNA-AcrIIA14ct-dsDNA have been deposited in the Protein Data Bank (PDB ID 7EL1).