Structural insight into multistage inhibition of CRISPR-Cas12a by AcrVA4

Significance

Bacteriophage encodes anti-CRISPR (Acr) proteins to inactivate host bacterial CRISPR-Cas systems. These Acrs are also found in bacteria to avoid self-targeting autoimmunity. So far, quite a few Acrs targeting type I and II CRISPR-Cas systems have been well characterized. In contrast, the Acrs inhibiting type V systems remain poorly understood. Both type II (Cas9) and V (Cas12a) CRISPR-Cas systems have been harnessed as powerful tools for genome editing and the latter showed even better efficiency and accuracy. In this work, we report a comprehensive mechanistic insight into a unique multistage inhibitor, AcrVA4, blocking CRISPR-Cas12a activity, different from other characterized single-stage targeting Acrs. This represents a sophisticated mechanism for CRISPR-Cas inhibition and provides clues for developing regulatory tools for genome editing.

Abstract

Prokaryotes possess CRISPR-Cas systems to exclude parasitic predators, such as phages and mobile genetic elements (MGEs). These predators, in turn, encode anti-CRISPR (Acr) proteins to evade the CRISPR-Cas immunity. Recently, AcrVA4, an Acr protein inhibiting the CRISPR-Cas12a system, was shown to diminish Lachnospiraceae bacterium Cas12a (LbCas12a)-mediated genome editing in human cells, but the underlying mechanisms remain elusive. Here we report the cryo-EM structures of AcrVA4 bound to CRISPR RNA (crRNA)-loaded LbCas12a and found AcrVA4 could inhibit LbCas12a at several stages of the CRISPR-Cas working pathway, different from other characterized type I/II Acr inhibitors which target only 1 stage. First, it locks the conformation of the LbCas12a-crRNA complex to prevent target DNA-crRNA hybridization. Second, it interacts with the LbCas12a-crRNA-dsDNA complex to release the bound DNA before cleavage. Third, AcrVA4 binds the postcleavage LbCas12a complex to possibly block enzyme recycling. These findings highlight the multifunctionality of AcrVA4 and provide clues for developing regulatory genome-editing tools.

Prokaryotes utilize a panel of defense systems to fight against the parasitic predators, including phages and mobile genetic elements (MGEs) (1). In addition to the innate immune systems, such as the restriction-modification (R-M) system (2), the toxin-antitoxin abortive infection system (3), DNA interference (4), the bacteriophage exclusion (BREX) system (5), etc., they have also developed a highly specific and prevalent adaptive immune system known as the CRISPR-Cas (clustered regularly interspaced short palindromic repeats [CRISPR]-associated proteins) system (6). It has been estimated that almost all archaea and about 50% of bacteria encode CRISPR-Cas systems in their genomes (7). CRISPR-Cas systems can protect the prokaryotes against phage infection by targeting the foreign nucleic acids in a sequence-specific manner, which works through 3 stages (adaptation, biogenesis, and interference) in the process (8, 9). Initially, the foreign nucleic acid segments are acquired by the host bacteria and incorporated into the CRISPR array as spacers in their genomes. These sequences are transcribed and processed into mature CRISPR RNA (crRNA) that contains a short direct repeat and a spacer. The Cas proteins are then expressed and associate with crRNA to form ribonucleoprotein complexes which detect the foreign nucleic acids by recognizing the protospacer adjacent motif (PAM), followed by base pairing between the spacer and the target sequence, and finally destroy the foreign nucleic acids (8).

CRISPR-Cas systems fall into 2 classes (I and II) and are further classified into 6 types (I–VI) so far, based on their phylogeny and working mechanisms (10, 11). All class I CRISPR-Cas systems, including types I, III, and IV, utilize a multisubunit effector complex, known as the surveillance complex, for target recognition and perform subsequent cleavage by recruiting an additional Cas protein or with resident nuclease subunits in the complex. By contrast, the class II systems, including type II, V, and VI, encode a single multidomain Cas protein to form the effector complex with crRNA, which mediates both the target recognition and degradation processes (10, 11). Due to the high sequence specificity and simple effector composition, 2 class II CRISPR-Cas effectors, Cas9 and Cas12a (Cpf1), have been successfully harnessed as powerful tools for genome editing, providing tremendous promise for therapeutic applications (12, 13). Another newly identified Cas13a (C2c2), which cleaves RNA targets, has also been developed as a rapid and high-sensitivity detection tool for diagnosis of pathogens and genotyping (14). Cas12a is guided by a single crRNA and generates staggered ends in its PAM-distal target site (15), in contrast to the 2-RNA-guided Cas9 enzyme which generates blunt ends at the cleavage site (16). As a next-generation genome-editing tool, Cas12a can simultaneously manipulate multiple target genes with potentially higher precision as compared to Cas9 (13, 17).

On the other hand, phages and MGEs, in turn, can encode anti-CRISPR (Acr) proteins to inactivate the CRISPR-Cas systems in host bacteria (18, 19). To date, a panel of Acr proteins have been discovered for type I, II, and V systems, through isolation of CRISPR-resistant phages, or by proximity to anti-CRISPR-associated (aca) genes, or by screening of bacteria genomes with self-targeting spacer sequences (20–29). A total of 5 type V Acrs have been identified and only 3 (AcrVA1, AcrVA4, and AcrVA5) of them can inhibit CRISPR-Cas12a-mediated genome editing in human cells, among which AcrVA4 showed the most potent inhibition activity for Lachnospiraceae bacterium Cas12a (LbCas12a) (28, 29). Recently, Dong et al. reported AcrVA5 inhibits Cas12a activity by acetylation, which renders steric hindrance for PAM recognition to inhibit dsDNA binding (30). Knott et al. found AcrVA1 enables the degradation of spacer sequence in the Cas12a-crRNA complex and AcrVA4 interferes with dsDNA binding by biochemical characterization (31). However, the exact underlying mechanisms for AcrVA4-mediated Cas12a inhibition remain unclear.

In this study, we report the structural basis of AcrVA4 inhibiting LbCas12a-mediated DNA cleavage. We found that AcrVA4 can bind to both crRNA-loaded LbCas12a and LbCas12a-crRNA-dsDNA complex, but not the apo form, with ultrahigh binding affinity. AcrVA4 can lock the conformation of LbCas12a-crRNA complex (crRNA-loaded state) to block target DNA hybridization with the crRNA spacer. It can also interact with the LbCas12a-crRNA-dsDNA complex (full R-loop conformation) to release the bound DNA before cleavage. In addition, it binds the postcleavage R-loop complex to block the recycling usage of the enzyme. These findings expand our understanding on the diverse molecular mechanisms of Acr proteins silencing CRISPR-Cas immunity and offer guidelines for developing anti-CRISPR off-switch tools for genome engineering and related biotechnological applications.

Results

AcrVA4 Homodimer Directly Interacts with Both LbCas12a-crRNA and LbCas12a-crRNA-dsDNA Complexes.

Previous study has shown that AcrVA4 could efficiently inhibit LbCas12a-mediated dsDNA cleavage both in vitro and in vivo (28, 29). To investigate whether AcrVA4 inactivates LbCas12a by direct interaction, we individually expressed and purified AcrVA4 and LbCas12a proteins using the Escherichia coli (E. coli) expression system and tested their binding by biochemical studies. As shown by size-exclusion chromatography (SEC) assay, AcrVA4 was eluted as a monodispersed peak with an estimated molecular weight (MW) of ∼50 kDa, indicating a dimeric form in solution (Fig. 1A). SDS-PAGE analysis under reducing and nonreducing conditions revealed the presence of interchain disulfide (Fig. 1B). Further analytical ultracentrifugation assay showed the MW of soluble AcrVA4 is ∼54 kDa, confirming that AcrVA4 exists as a homodimer cross-linked by interchain disulfide (Fig. 1C).

Fig. 1.

AcrVA4 forms a homodimer and selectively binds the LbCas12a-crRNA binary complex. (A) SEC of AcrVA4 using a Superdex 200 10/300 GL column. (B) SDS-PAGE profiles of AcrVA4 at reduced and nonreduced conditions. The bands of monomeric and dimeric forms are indicated by black and red triangles, respectively. (C) Sedimentation velocity analysis of AcrVA4. The estimated MW corresponds to a homodimer. (D–F) Analytical SEC assays for testing the binding of AcrVA4 to apo LbCas12a (D), LbCas12a-crRNA binary complex (E), and dLbCas12a-crRNA-dsDNA ternary complex (F). (G) EM density and atomic structure of the AcrVA4 dimer. One of the protomers is colored by domains and the other is colored in gray. The interchain disulfide is highlighted by a red dashed oval. The dimeric interface is shown in close-up view and the interacting residues are shown as sticks (colored by atoms). Hydrogen bonds and salt bridges are represented as black dashed lines. (H) Topological diagram of the AcrVA4 structure. As most of the NTD could not be resolved in the density map, it is represented by a dashed oval (cyan).

It has been established that significant conformational changes occur during the transitions from apo Cas12a to crRNA-loaded and dsDNA target-bound states (32–36). We then performed SEC assays to detect the potential interactions between AcrVA4 and LbCas12a at these different conformational states. The apo LbCas12a showed no binding to AcrVA4 (Fig. 1D), whereas the LbCas12a-crRNA complex efficiently interacted with AcrVA4 and formed a stable complex in solution (Fig. 1E). Strikingly, AcrVA4 could also bind to the LbCas12a-crRNA-dsDNA complex in full R-loop conformation, using catalytically dead LbCas12a mutant with an E925Q substitution (dLbCas12a), and induce the release of bound dsDNA to avoid cleavage (Fig. 1F and SI Appendix, Fig. S1 A and B). This interaction could possibly revert the conformation of LbCas12a back to the crRNA-loaded state to deactivate the enzyme. In addition, AcrVA4 could also interact with the postcleavage R-loop complex but the bound DNA did not dissociate (SI Appendix, Fig. S1 C and D). This interaction would probably prevent the new crRNA replacement process for Cas12a resetting (32), which therefore blocks the enzyme recycling for next-round catalysis. These evidences suggest AcrVA4 could potentially inhibit the activity of Cas12a at several states, both before and after dsDNA binding.

Structural Features of AcrVA4.

To elucidate the mechanism of AcrVA4-mediated Cas12a inhibition, we determined the structure of AcrVA4 bound to the LbCas12a-crRNA complex at 3.25-Å resolution by cryo-electron microscopy (cryo-EM) single particle reconstruction (SI Appendix, Figs. S2–S4). This structure corroborated our biochemical evidence that AcrVA4 exists in dimeric form with an interchain disulfide (Fig. 1G). The structure of AcrVA4 is composed of 2 domains, the N-terminal and C-terminal domains (NTD and CTD), connected by a central helix (Fig. 1 G and H). Two AcrVA4 protomers form a dumbbell-shaped structure with a near-parallel orientation in which the central helices stabilize the dimeric interactions. The 2 central helices are not strictly parallel to each other but form a crossover in the middle where an interchain disulfide is observed between C131 and its symmetric partner (Fig. 1G). Apart from the disulfide, the dimeric interface is further stabilized by quite a few polar and nonpolar interactions between the 2 helices (Fig. 1G). To test the role of interchain disulfide for AcrVA4 dimerization, we conducted SEC and sedimentation analyses under reducing conditions and found the dimerization interactions were well maintained (SI Appendix, Fig. S5A). Moreover, the SEC profiles for binding the LbCas12a-crRNA complex were essentially the same in the presence or absence of dithiothreitol (DTT) (SI Appendix, Fig. S5B). Besides, the C131S mutant also displayed homogeneous dimeric form in solution (SI Appendix, Fig. S5C). These observations demonstrated that the interchain disulfide is not essential for AcrVA4 dimerization.

The CTD of AcrVA4 is mainly composed of 6 β-strands connected by flexible loops in between (Fig. 1H) and is responsible for Cas12a interacting (Fig. 2). The NTD of AcrVA4 is located at the distal end of the complex and could not be clearly resolved due to poor density (Fig. 1G). Based on secondary structure predictions, the NTD of AcrVA4 might form a stable domain with ordered structure (SI Appendix, Fig. S5D). The failure of reconstructing this region indicated the connecting loop between NTD and the central helix rendered the entire domain highly flexible. We then performed a DALI (37) search to compare the structure of AcrVA4 with other known protein structures, but no significant hint was obtained, indicating this structure represents a novel fold and further supporting the diverse origins of different Acrs in evolution.

Fig. 2.

AcrVA4 binds to the LbCas12a-crRNA complex with 2 different stoichiometries. (A) Schematic diagram of the LbCas12a domain architecture. Each domain is represented by a unique color and the same color code is used throughout the manuscript. (B) Overall density map of LbCas12a-crRNA-AcrVA4 form A complex in which an AcrVA4 homodimer binds to a single copy of the LbCas12a-crRNA binary complex. The LbCas12a molecule is colored by domains and the crRNA repeat region is colored in black. The AcrVA4 dimer is colored by chains with the main interacting subunit in orange and the other in gray. (C) Representative 2D class averages of LbCas12a-crRNA-AcrVA4 form A and form B complexes. (D) Structure of LbCas12a-crRNA-AcrVA4 form B complex in which an AcrVA4 homodimer binds to 2 copies of the LbCas12a-crRNA complex. One copy of LbCas12a-crRNA is colored by domains and the other is in white.

AcrVA4 Binds LbCas12a-crRNA with Different Stoichiometries.

During cryo-EM image processing, we noticed the LbCas12a-crRNA-AcrVA4 complex exists in 2 forms with different binding stoichiometries. The predominant form (form A, ∼90% of all particles) contains an AcrVA4 homodimer and a single copy of the LbCas12a-crRNA complex, while the minor form (form B, ∼10% of all particles) contains 2 copies of LbCas12a-crRNA binding to an AcrVA4 dimer (Fig. 2C and SI Appendix, Fig. S2 B and C). This observation was consistent with the SEC profile of the LbCas12a-crRNA-AcrVA4 complex that was eluted as a dominant peak with a small shoulder in the front, corresponding to form A and form B complexes, respectively (Fig. 1E). The 2 complex forms were reconstructed at 3.25-Å and 4.10-Å resolutions, respectively (SI Appendix, Fig. S3A and Table S1).

The binding of AcrVA4 to LbCas12a is mediated by its CTD that inserts into a valley formed by the wedge (WED), bridge helix (BH), REC2, and RuvC domains of LbCas12a and the crRNA (Fig. 2 A and B). The interactions mainly involve 1 protomer (AcrVA4.1) within the AcrVA4 dimer and the other protomer (AcrVA4.2) contributes few Van der Waals (VDW) contacts (Fig. 2B and SI Appendix, Table S2). In the form A complex, the vacant Cas12a binding site on AcrVA4.2 is fully exposed and thus offering the opportunity for the binding of a second Cas12a molecule (Fig. 2B). As observed in the form B complex, 2 copies of the LbCas12a-crRNA complex are cross-linked by an AcrVA4 homodimer to form a near-symmetric structure (Fig. 2D and SI Appendix, Fig. S4D), which corroborated the binding interface identified in the form A complex.

Both complexes display highly similar conformations for interacting with AcrVA4, indicating a common mechanism for Cas12a inhibition (SI Appendix, Fig. S3B). This observation also implies that the low abundance of the form B complex compared to form A might result from the excessive amount of AcrVA4 in the preparation. To test this hypothesis, we incubated the LbCas12a-crRNA complex with AcrVA4 at a molar ratio of 2:1 (2 LbCas12a-crRNA complexes versus 1 AcrVA4 dimer) and found that most of the complex exists in form B stoichiometry (SI Appendix, Fig. S5E), in contrast to the form A dominant profile when an excessive amount of AcrVA4 is provided. This evidence suggests that both form A and B binding modes are equivalently functional for Cas12a inhibition, which might be preferentially adopted in response to different AcrVA4 concentrations in the cell.

Interactions between AcrVA4 and the LbCas12a-crRNA Complex.

The binding interface between AcrVA4 and the LbCas12a-crRNA complex could be divided into 3 subregions. The C-D loop of AcrVA4 forms a long protrusion stretching into the valley surrounded by the WED, BH, and REC2 domains of LbCas12a (Fig. 3 A and B). At the bottom of the valley, the tip of the C-D loop also contacts with the crRNA repeat and the RuvC domain. The B/C strands and A-B/E-F loops form 2 wings to cover the valley on both sides (Fig. 3B). The other AcrVA4 protomer (AcrVA4.2) within the homodimer forms several VDW contacts with the WED domain outside the valley (Fig. 3A and SI Appendix, Table S2), contributing almost no effect to the interaction.

Fig. 3.

Contacting interface between AcrVA4 and the LbCas12a-crRNA complex. (A) Atomic structure of the LbCas12a-crRNA-AcrVA4 form A complex. The LbCas12a molecule is shown in surface model and colored by domains. The crRNA is represented by cartoons in which the CRISPR repeat and spacer regions are colored in black and red, respectively. The bound AcrVA4 dimer is shown as ribbons and colored by chains of which the main contacting protomer is colored in orange and the other in gray. (B) Close-up view of the binding interface. (C–F) Interaction details between AcrVA4 and the LbCas12a-crRNA complex. The key interacting residues are shown as sticks and colored by atoms. Hydrogen bonds and salt bridges are represented by black dashed lines.

The B/C strands (residues 183 through 196) contact with the WED domain of LbCas12a mainly through polar interactions (Fig. 3C). Residue E184 within the B strand is interspaced between H759 and K768 of LbCas12a, forming 2 consecutive salt bridges in the interface. The main chain of R187 in the B strand is connected to the main chain of V758 in LbCas12a by 2 hydrogen bonds. Besides, residue M186 of the B strand forms extensive VDW interactions with V758 and H759 of LbCas12a, as well as the nucleotide A-20 within the CRISPR repeat of crRNA. In addition, the connecting loop between the central helix and CTD of AcrVA4 also participates in the interaction (Fig. 3C).

The C-D loop (residues 197 through 208) of AcrVA4 protrudes deep inside the valley and interacts with multiple domains (REC2, WED, and RuvC) of LbCas12a, as well as the crRNA repeat (Fig. 3B). This loop contains quite a few charged residues, forming extensive electrostatic interactions with both LbCas12a and the phosphate of crRNA. The main chain oxygen of K202 hydrogen bonds to the side chain of N895 in the RuvC domain, and its side chain forms a salt bridge with the phosphate of U-14 in the crRNA. Residue R203, E204, and R206 form 3 additional salt bridges with D450 in the REC2 domain and K785 and E755 within the WED domain, respectively (Fig. 3D). Moreover, Y197 also forms a hydrogen bond with K785, and the side chain of T201 forms a hydrogen bond with the phosphate group of A-20 in the crRNA (Fig. 3E). The frequent electrostatic interactions in this region indicate the potential high-affinity binding between AcrVA4 and LbCas12a-crRNA.

On the other side of the valley, the A-B (residues 175 through 182) and E-F (residues 216 through 219) loops form the second wing to cover the BH domain (Fig. 3B). At this interface, residue Y160 in the A strand forms a hydrogen bond to R887 in the BH connecting loop, which concomitantly contacts with the adjacent W178 of AcrVA4 by π-cation interaction. In addition, the side chain of R217 interacts with both the main chain and side chain of F884 by hydrogen bond and π-cation interaction, respectively (Fig. 3F). Collectively, the 3 interacting subregions contact multiple domains of LbCas12a, which would thus lock the conformations of these domains to hinder the structural transition required for target dsDNA unwinding and hybridization with the crRNA.

Previous study has shown that AcrVA4 could efficiently inhibit the activity of LbCas12a but is ineffective to Acidaminococcus sp. Cas12a (AsCas12a), both of which have been utilized for genome editing in cells (29). By comparing the structures of LbCas12a and AsCas12a, we found the WED domain of AsCas12a harbors an insertion of 2 helices, which directly collides with the central helix and CTD of AcrVA4 (SI Appendix, Fig. S6 A and B). Therefore, AsCas12a is resistant to the inhibition by AcrVA4.

We also compared the structures of the LbCas12a-crRNA complex before and after AcrVA4 binding to analyze the potential conformational changes induced by the interaction. No major changes were observed in the structures except some local movements of the REC1 and REC2 domains (SI Appendix, Fig. S7A). The REC1 domain slightly rotates toward the PAM-interacting (PI) domain, and the REC2 domain moves outward to leave the WED domain. This rearrangement displaces the 450-helix by an ∼8.5-Å distance to allow for enough space for accommodating AcrVA4 (SI Appendix, Fig. S7B).

Ultrahigh-Affinity Binding Is Required for LbCas12a Inhibition by AcrVA4.

To verify our structural analysis, we performed mutagenesis studies on the key interacting residues within AcrVA4 to test their effects on Cas12a binding and inhibition (Fig. 4 and SI Appendix, Fig. S8). Among the 3 subregions, the C-D loop protrusion and the wing formed by A-B and E-F loops contribute the majority of strong charged interactions (Fig. 3 C–F). Therefore, we conducted mutations on these 2 subregions which should be crucial for the inhibitory activity of AcrVA4.

Fig. 4.

Mutagenesis verification of the interactions and effects on Cas12a inhibition. (A–C) SPR measurement for the binding affinity between AcrVA4 and LbCas12a at different states. (D–H) The binding kinetics of AcrVA4 mutants to LbCas12a-crRNA binary complex. The raw binding curves (black) and the fitted curves (red) are superimposed in the figures. (I) Dose-dependent inhibition of AcrVA4 for LbCas12a-crRNA-mediated dsDNA cleavage in vitro. The LbCas12a-crRNA complex was used at 100 nM concentration in the reaction. The EC₅₀ concentration of AcrVA4 was calculated based on the fitted curve. The error bars represent SDs of 3 independent experiments. (J) Effects of mutations on the inhibitory activities of AcrVA4 for LbCas12a-crRNA-mediated dsDNA cleavage.

As shown by surface plasmon resonance (SPR) assays, the wild-type (WT) AcrVA4 did not bind apo LbCas12a but showed an ultrahigh affinity to the LbCas12a-crRNA binary complex which did not dissociate after binding (KD < 0.09 pM) (Fig. 4 A and B). This is consistent with the results of SEC assays that AcrVA4 forms a stable complex with the crRNA-loaded but not apo LbCas12a (Fig. 1 D and E). The affinity of AcrVA4 to cleavage-inactivated dLbCas12a-crRNA-dsDNA complex was ∼0.12 pM, similar to that for the LbCas12a-crRNA binary complex (Fig. 4C). This observation suggested that AcrVA4 could efficiently interact with the LbCas12a full R-loop complex and possibly revert its conformation back to the crRNA-loaded state by displacing the bound dsDNA. We also conducted LbCas12a-crRNA-mediated dsDNA cleavage in vitro and tested the inhibitory effect of AcrVA4. The WT AcrVA4 inhibited the dsDNA cleavage in a dose-dependent manner, with an EC₅₀ concentration of ∼65.6 nM, against 100 nM LbCas12a-crRNA complex (Fig. 4I). At a concentration of 400 nM, ∼6 times the EC₅₀ dose, it showed an almost complete inhibition for cleavage (Fig. 4J). Substitutions of K202 and K203 by alanine enabled the AcrVA4 mutant to dissociate after binding, and reduced the affinity by ∼100,000 folds (Fig. 4D). Consistent with this observation, the inhibitory efficiency was obviously compromised with ∼50% and ∼80% inhibitions at 800 and 1,600 nM (12 times and 24 times the EC₅₀) concentrations, respectively (Fig. 4J). The E204A/R206A double mutant showed a 10-fold decrease in binding affinity and still maintained good inhibitory efficiency for dsDNA cleavage (Fig. 4 E and J). Simultaneous substitution of the 4 residues reduced the binding affinity by ∼1,000,000 folds in comparison to the WT AcrVA4, which completely abolished the inhibition at 12-times the EC₅₀ concentration and showed only weak inhibition at 24 times the EC₅₀ concentration (Fig. 4 F and J). On the other hand, replacement of the 3 residues Y160, W178, and R217, which form the right wing to interact with the BH domain, significantly reduced the binding affinity by 1,000,000 folds compared to the WT and lost the inhibitory activity even at 24 times the EC₅₀ concentration (Fig. 4 G and J), comparable to the performance of the quaternary mutant on the C-D loop. Replacing the 7 residues with alanine simultaneously further reduced the binding affinity to show a 10,000,000-fold decrease as compared to the WT AcrVA4 and completely lost the inhibitory effect at 24 times the EC₅₀ concentration (Fig. 4 H and J). These data revealed that high-affinity binding is required for the efficient inhibition which may tightly lock the conformation of multiple LbCas12a domains to disable its enzymatic activity.

Inhibition Mechanism of AcrVA4.

Previous studies have shown that crRNA loading transforms the structure of Cas12a from extended to compact conformation (36); the AcrVA4 binding valley located at the junction region between 2 Cas12a lobes (REC and NUC lobes) thus could not be formed in the extended apo LbCas12a. The binding of the dsDNA target is coupled with further conformational changes in the Cas12a-crRNA binary complex to allow hybridization between the crRNA spacer and target strand (TS) of dsDNA and to expose the nuclease active site between Nuc and RuvC nuclease domains (Fig. 5A) (32, 33, 35). In this transition, the conformation of the NUC lobe does not substantially change except the BH and PI domains, whereas the domains within the REC lobe are significantly reoriented. The REC1 domain rotates toward the PI domain and the REC2 domain flips upward to detach from the NUC lobe, thus creating a central channel to accommodate the crRNA-TS heteroduplex. In addition, the BH helix is elongated by the folding of connecting loop and moves toward the REC lobe to make contacts with the REC2 domain and crRNA-TS heteroduplex (Fig. 5A), which has been shown essential for dsDNA cleavage (35).

Fig. 5.

AcrVA4 binding locks the conformation of the LbCas12a-crRNA binary complex and interferes with dsDNA binding. (A) The conformational changes of LbCas12a-crRNA before (Left, in dark color) and after (Right, in light color) dsDNA binding. The conformation of the NUC lobe does not significantly change (shown as transparent surface), whereas the REC lobe is substantially reorganized (in cartoon model, colored by domains). The key motifs involved in the transition are shown as cartoons. The directions for structural transition are indicated by arrows with the same colors of corresponding domains. The spacer of crRNA is colored in red, and the TS and NTS of dsDNA are colored in purple and cyan, respectively. The catalytic site is highlighted with a red asterisk. (B) AcrVA4 (orange, transparent) binds to the LbCas12a-crRNA binary complex and interacts with the BH domain, 895-helix, and the REC2 domain to prevent their structural rearrangement for dsDNA binding. (C) Superposition of the LbCas12a-crRNA-AcrVA4 with LbCas12a-crRNA-dsDNA complex. The close-up view at the AcrVA4-interacting interface is shown to reveal the conformational changes of the BH domain, the 895-helix of RuvC domain, and the 450-helix of the REC2 domain, which are locked by AcrVA4 binding. The key interacting residues in LbCas12a are shown in sticks to highlight the deformation of the AcrVA4 binding site after dsDNA binding. (D and E) EMSA assay to test the binding of dsDNA to the dLbCas12a-crRNA complex in the absence (D) or presence (E) of AcrVA4. The NTS was labeled with Cy5 fluorophore at the 5′-end. Protein complexes were used with gradient concentrations.

We then compared the structure of AcrVA4-bound LbCas12a-crRNA with the ternary complex of LbCas12a-crRNA-dsDNA by superimposing the NUC lobe. Despite that the conformation of the WED domain is well maintained after dsDNA binding, the 450-helix in REC2 domain, the 895-helix in the RuvC domain, and the BH helix are substantially rearranged, which are key structural elements for AcrVA4 binding and inhibition (Fig. 5 B and C). The conformation of the WED domain is preserved in the LbCas12a-crRNA-dsDNA ternary complex (SI Appendix, Fig. S6C), which might provide an initial binding site for AcrVA4 to release the bound DNA and possibly revert the conformation of LbCas12a back to the crRNA-loaded state (SI Appendix, Fig. S1B). In addition, the deformation of the AcrVA4-contacting interface in the REC2, RuvC, and BH domains in the LbCas12a-crRNA-dsDNA complex indicates the ultrahigh-affinity binding of AcrVA4 to the LbCas12a-crRNA binary complex would indeed prevent conformational changes of these domains, which as a result would block dsDNA binding (Fig. 5 B and C). This is supported by the observation that unlocking the interactions by mutations in the C-D loop protrusion and the A-B/E-F loop wing impaired the inhibitory activity (Fig. 4J).

To further test the hypothesis, we performed electrophoretic mobility shift assays (EMSA) to test the binding between LbCas12a-crRNA-AcrVA4 and dsDNA. As the positive control, the dLbCas12a-crRNA complex efficiently interacted with dsDNA and showed a dose-dependent dsDNA retardation profile (Fig. 5D). In contrast, the presence of AcrVA4 significantly impaired the dsDNA binding activity of the dLbCas12a-crRNA complex and displayed very weak binding at high concentration of the complex (Fig. 5E). This result indicated that AcrVA4 binding hinders the full R-loop formation but possibly allows PAM recognition at the early stage of dsDNA binding, which does not require structural rearrangement in the AcrVA4-binding interface of LbCas12a.

Collectively, these structural and biochemical evidences demonstrate that AcrVA4 could disable LbCas12a-mediated CRISPR-Cas immunity at different stages of the working pathway (Fig. 6). First, AcrVA4 locks the conformation of crRNA-loaded LbCas12a with ultrahigh-affinity binding, preventing the dsDNA target engagement and cleavage. Second, AcrVA4 can interact with the LbCas12a-crRNA-dsDNA (full R-loop) complex to release the bound DNA before cleavage. In addition, it can also bind the postcleavage complex (cleaved R-loop) and probably block the recycling usage of the enzyme in the next round of catalysis.

Fig. 6.

Schematic model of AcrVA4 inhibiting Cas12a-mediated dsDNA cleavage. The models of Cas12a are colored by domains as annotated and the AcrVA4 dimer is shown with 2 protomers in different colors. AcrVA4 could inhibit the activity of Cas12a at several steps in the catalytic cycle. It binds the LbCas12a-crRNA binary complex to prevent dsDNA engagement. Despite that PAM recognition is not prohibited, further dsDNA unwinding and base pairing with the crRNA spacer are perturbed to block full R-loop formation. Moreover, it could also interact with the LbCas12a-crRNA-dsDNA ternary complex (full R-loop state) to release the bound dsDNA before cleavage. This process could possibly revert the conformation of LbCas12a back to the pre-dsDNA binding state. In addition, AcrVA4 also binds to the cleaved R-loop complex to block the recycling of the enzyme. Thus, AcrVA4 utilizes different mechanisms for inhibiting LbCas12a activity at multiple stages of CRISPR-Cas immunity.

Discussion

So far, quite a few Acr proteins targeting different CRISPR-Cas systems have been identified and characterized, and the inhibition mechanisms vary for different types of CRISPR-Cas systems (19). For example, AcrIF1 targets the type I-F surveillance complex to block target DNA recognition by interfering with the base pairing between spacer and target (38, 39); AcrIF2 and AcrIF10 compete with target dsDNA for the binding site on the surveillance complex (38, 40); AcrIF3 binds to Cas3 endonuclease to prevent being recruited by the surveillance complex thereby blocking target DNA access (41, 42); AcrIIA2 and AcrIIA4 prevent target DNA recognition of Cas9-guide RNA (gRNA) complex by dsDNA mimicking (43–45); AcrIIC1 inactivates the Cas9 activity by blocking the catalytic site (46); and AcrIIC2 interferes with the guide RNA loading process of Cas9 which consequently impairs target DNA binding (47). AcrIIC3 dimerizes Cas9-gRNA complexes and inhibits target DNA binding (46, 47).

In this work, we describe a comprehensive mechanistic insight into a Cas12a-targeted Acr protein, AcrVA4, which inhibits CRISPR-Cas immunity at multiple stages of the interference process. It not only interferes with target DNA binding, but also releases the bound dsDNA to revert its fate before cleavage. In addition, it may block the recycling of the Cas12a enzyme after dsDNA cleavage. Thus, AcrVA4 utilizes a unique multistage inhibition mechanism for silencing CRISPR-Cas immunity, which differs from other known Acr proteins with mainly only 1 stage interference activity. The multistage interference of AcrVA4 ensures high efficiency of CRISPR-Cas inhibition, which might represent a more sophisticated strategy selected after the long-term evolution. Of note, most of the Acrs represent individually novel protein folds, suggesting the diverse origin of different Acr proteins. Based on sequence analysis, AcrVA4 could only be found in the genome of Moraxella bovoculi and no other homologs could be detected. Interestingly, we identified a hypothetical protein in the same bacterium with ∼30% sequence identity, which potentially represents an AcrVA4 isoform or a novel Acr protein (SI Appendix, Fig. S9).

Among all these Acr proteins characterized so far, quite a number of them intercept the recognition and binding of target DNA, which thus represents an efficient and ubiquitous strategy to silence CRISPR-Cas immunity. However, this could be achieved through various mechanisms by different Acr proteins. AcrIF1 hinders the base pairing between the spacer and TS by steric hindrance and inducing local conformational changes of the surveillance complex (38, 39). AcrIF2, AcrIF10, AcrIIA2, and AcrIIA4 mimic the structure of dsDNA to directly compete for the binding site of dsDNA target on the effector complexes (38, 40, 43–45). A recent work showed that AcrVA5 acetylated lysine in the PI domain of Cas12a, thus conferring steric hindrance for dsDNA binding (30). The observation that AcrVA4 binds at the opposite side of the dsDNA binding site without direct spatial competition is quite remarkable, which adopts a noncompetitive mechanism to inhibit target DNA binding by locking the conformation of the CRISPR effector complex. During the reviewing process of our manuscript, Zhang et al. (48) reported similar results that AcrVA4 prevents the conformational changes of Cas12a for R-loop formation. This unique inhibition mechanism is achieved by the ultrahigh binding affinity that results from multiple contacting interfaces and the occurrence of frequent salt bridge interactions.

The binding affinity of AcrVA4 to Cas12a is extraordinarily high among all of the Acr proteins characterized so far, even though the affinities of these Acrs are universally as high as nanomolar range (38, 45–49). As the SPR assays revealed nondissociative binding kinetics for AcrVA4, the affinity value could not be precisely calculated but indeed was beyond nanomolar scale. We further verified the binding affinities by biolayer interferometry (BLI) analysis and similar nondissociative kinetics were observed (SI Appendix, Fig. S10), suggesting the interactions between AcrVA4 and Cas12a might be irreversible at any one stage of the working pathway. This highlights the fact that the conformational locking mechanism requires ultrahigh binding affinity of the inhibitor, as binding itself is not sufficient for the inhibition. These unique noncompetitive features offer the guidelines for designing dedicate “off-switch” tools for Cas12a-based genome-editing and DNA-targeting platforms. In addition, it could be utilized for programming specific phages to treat challenging infections of bacteria with CRISPR-Cas12a systems.

It is noteworthy that most Acr proteins characterized so far directly interact with CRISPR effector complexes. However, there are other checkpoints in the CRISPR-Cas working pathway that could be targeted to block the immunity. It has been shown that AcrIIA1 interacts with RNA but the extract working mechanism is unclear (50). Possibly it affects the crRNA loading into the Cas9 effector and inhibits the assembly of the functional CRISPR-Cas effector complex. Besides, suppression of pre-crRNA transcription/processing and Cas proteins expression/recycling, or target nucleic acid shielding, could also be feasible strategies to achieve CRISPR-Cas inhibition. Further work in the field would greatly expand our understanding of the functionality of these systems and stimulate new applications.

Materials and Methods

More detailed descriptions of the materials and methods used in this study are provided in SI Appendix. A brief summary is provided here.

Protein Expression and Purification.

The coding sequences for AcrVA4 (NCBI Entry: WP_046699156) and LbCas12a (NCBI Entry: WP_035635841) were synthesized and codon optimized for expression in E. coli. AcrVA4 was expressed with an N-terminal 6×His tag, and LbCas12a was fused with a maltose binding protein (MBP) tag, an N-terminal 6×His, and C-terminal 2×Strep tag to facilitate purification. The crRNA was produced with an additional plasmid which was cotransformed with Cas12a to enable self-processing into mature crRNA. All proteins were solubly expressed in E. coli BL21 (DE3) at 16 °C and purified by tandem affinity chromatography, ion-exchange chromatography, and size-exclusion chromatography (SI Appendix, SI Materials and Methods). All protein samples were analyzed by SDS-PAGE and reached a purity of ∼95%.

Cryo-EM Data Collection, Image Processing, and Model Building.

The LbCas12a-crRNA-AcrVA4 complex was prepared and vitrified using a FEI Vitrobot Mark IV. Cryo-EM data were collected with a 300 kV Titan Krios transmission electron microscope equipped with a GIF-Quantum energy filter and a Gatan K2-summit direct electron detector (SI Appendix, SI Materials and Methods). Images of processing and reconstruction were performed using RELION-3.0 (51). Model building and refinement were conducted with CHIMERA (52), COOT (53), and PHENIX (54) (SI Appendix, SI Materials and Methods). Structural figures were rendered by PyMOL (https://pymol.org/) or CHIMERA (52). Representative densities are shown in SI Appendix, Fig. S4. The statistics for image processing and model refinement are summarized in SI Appendix, Table S1.

SPR Assay.

SPR experiments were performed at room temperature (r.t.) using a Biacore 8K system with SA chips (GE Healthcare). LbCas12a at different states (apo LbCas12a, LbCas12a-crRNA complex, dLbCas12a-crRNA-dsDNA complex, and LbCas12a-crRNA-dsDNA complex) was immobilized on the chip by biotin affinity tag. The single-cycle binding kinetics for WT AcrVA4 and mutants was analyzed using a 1:1 (a copy of LbCas12a-crRNA versus an AcrVA4 dimer) binding model (SI Appendix, SI Materials and Methods).

In Vitro Cleavage Assay.

The in vitro dsDNA cleavage assays were conducted in a cleavage buffer consisting of 20 mM Hepes-NaOH, pH 7.5, 150 mM KCl, 10 mM MgCl₂, and 0.5 mM Tris(2-carboxyethyl)phosphine (TCEP). The linearized plasmid and serially diluted AcrVA4 protein were preincubated at 37 °C for 10 min, and the LbCas12a-crRNA binary complex was then added to start the reaction. After an additional 30-min incubation at 37 °C, the reaction system was resolved by electrophoresis using a 1% agarose gel and visualized by staining with SYBR Green dyes (Invitrogen) (SI Appendix, SI Materials and Methods). The images were quantified by integrating the intensity of each band using ImageJ software. The dose-dependent inhibition curve was generated with GraphPad Prism 5 to estimate the EC₅₀ value.

Data Availability.

The cryo-EM density maps of form A and form B LbCas12a-crRNA-AcrVA4 complexes have been deposited to the Electron Microscopy Data Bank (EMDB) with the accession codes EMD-0704 and EMD-0705, respectively. The coordinates of corresponding atomic models have been deposited to the Protein Data Bank (PDB) under the entries 6KL9 and 6KLB, respectively.