Structural and functional insights into the bona fide catalytic state of Streptococcus pyogenes Cas9 HNH nuclease domain

Abstract

The CRISPR-associated endonuclease Cas9 from Streptococcus pyogenes (SpyCas9), along with a programmable single-guide RNA (sgRNA), has been exploited as a significant genome-editing tool. Despite the recent advances in determining the SpyCas9 structures and DNA cleavage mechanism, the cleavage-competent conformation of the catalytic HNH nuclease domain of SpyCas9 remains largely elusive and debatable. By integrating computational and experimental approaches, we unveiled and validated the activated Cas9-sgRNA-DNA ternary complex in which the HNH domain is neatly poised for cleaving the target DNA strand. In this catalysis model, the HNH employs the catalytic triad of D839-H840-N863 for cleavage catalysis, rather than previously implicated D839-H840-D861, D837-D839-H840, or D839-H840-D861-N863. Our study contributes critical information to defining the catalytic conformation of the HNH domain and advances the knowledge about the conformational activation underlying Cas9-mediated DNA cleavage.

Keywords: CRISPR-Cas9; HEK293T cell; catalysis model; human; molecular biophysics; structural biology.

Figure 1.. Architecture of the HNH domain ββα-Me fold in different binding forms of Cas9 (a–d) and site-directed mutagenesis experiments identifying potential catalytic residues (e–f).

(a–d) ββα-Me fold in the sgRNA-bound state of Cas9 (a), in the intermediate state (b), in the pre-catalytic state (c), and in the pseudoactive state (d). The ββα-Me fold is represented as pink ribbons, and the residues are shown in stick models and colored by atom type (C, dark green; N, blue; O, red). If present, the bound Mg²⁺ ion is depicted as a magenta sphere, and only the tDNA phosphate-sugar backbone is displayed for clarity. The location of the Cas9 D861 is highlighted by an arrow, and the dashed lines denote hydrogen bonds or coordinative bonds. (e) The expression and DNA-editing activity of the wild-type and D861A variants of Cas9 paired with an sgRNA sequence that targets the egfp gene in HEK293T-EGFP cells. (f) The expression and DNA-editing activity of the wild-type and indicated variants of Cas9 paired with an sgRNA sequence that targets the egfp gene in HEK293T-EGFP cells. The retention of EGFP expression reflected the loss of activity of Cas9 protein in the cells.

Figure 1—figure supplement 1.. The DNA sequencing analysis of vectors for expressing different Cas9 variants and the flow cytometry analysis of HEK293T-EGFP cells with different Cas9-sgRNA expression vectors.

(a) Upper panel: A DNA sequencing chromatogram showing the D861A mutation of the Cas9 gene open reading frame in the lentiCRISPR expression vector of Cas9 with the EGFP sgRNA1 sequence. Lower panel: Representative histograms of flow cytometry analysis in HEK293T-EGFP cells with the expression vectors of wild-type (WT) and D861A Cas9. (b) Upper panel: DNA sequencing chromatograms showing the D837A, D839A, N863A and D861A/N863A mutations of the Cas9 gene open reading frames in the lentiCRISPR expression vectors of Cas9 with the EGFP sgRNA1 sequence. Lower panel: Representative histograms of flow cytometry analysis in HEK293T-EGFP cells with the expression vectors of WT, D837A, D839A, N863A and D861A/N863A Cas9.

Figure 1—figure supplement 2.. Plasmid cleavage activity of SpyCas9^WT.

(a) SpyCas9^D861A (b) and SpyCas9^N863A (c) Left panels: The gel images of DNA electrophoresis showing the bands of different plasmid conformations due to the cleavage activity of Cas9. Right panels: The quantitative results of cleaved DNA over the indicated reaction time. Each data point presents the average value of three replications. Error bars represents standard error of mean. The results indicate that SpyCas9^D861A has a similar activity profile as SpyCas9^WT, expect for a slower rate. SpyCas9^N863A produces only nicked products even after one hour of reaction. [N: nicked, L: linear, SC: supercoiled].

Figure 1—figure supplement 3.. Cleavage of radioisotope-labeled oligo DNA substrate by SpyCas9^WT, SpyCas9^D861A and SpyCas9^N863A.

(a) A schematic illustration of radioisotope-labeled oligo DNA substrate used in this study. The duplex is labeled with ³²P at the 5’-end of both strands. The sequences in bold are the protospacer; the PAM sequence is in red. Black triangles indicate the anticipated cleavage sites by HNH and RuvC. The size of denatured DNA products as visible in electrophoresis followed by film exposure is indicated for each DNA strand. (b) A representation of gel image showing the products of oligo DNA cleavage. Reactions were performed for different time periods, and the products were resolved on a denaturing 16% urea-formamide gel. The NT-strand cleavage product and T-strand cleavage product are produced respectively by the RuvC and HNH domains. SpyCas9^N863A lacks T-strand cleavage product, suggesting that the HNH activity is eliminated in this variant. In summary, our in vitro cleavage reactions of plasmid and oligo DNAs clearly indicate that N863 is indispensable for the HNH nuclease activity, whereas D861 appears to provide a supporting role to enable a faster reaction rate but is nonessential for target DNA strand cleavage. A close inspection of our simulation trajectories led us to propose that D861 might have a role in stabilizing the catalytic ββα-Me motif by forming an intra-molecular salt bridge (Figure 2d), or aid initial recruitment of metal ions around the active center with other negative species like D839 and D837 (Figure 2a).

Figure 1—figure supplement 4.. The architecture of the HNH domain ββα-Me motif in the apo structure of SpyCas9 (a) and structural modeling of SpyCas9 with Mg²⁺ ion bound at the catalytic center (b–c).

The ββα fold is depicted as a pink ribbon diagram, and the residues at and around the fold core are shown as a stick model and colored by atom types (C, dark green; N, blue; O, oxygen). The disordered loop in apo-SpyCas9 is indicated by a chain of beads. The magenta sphere represents the bound Mg²⁺, and the dashed lines denote hydrogen bonds or coordination bonds.

Figure 1—figure supplement 5.. The architecture of the HNH domain ββα-Me motif in the apo structure of Actinomyces naeslundii Cas9 (AnaCas9) (a) and in that of partial dsDNA-bound Staphylococcus aureus Cas9 (SauCas9) (b).

The ββα fold is depicted as a pink ribbon diagram, and the residues at and around the fold core are shown as a stick model and colored by atom types (C, dark green; N, blue; O, oxygen). The magenta sphere represents the bound Mg²⁺, and the dashed lines denote hydrogen bonds or coordination bonds.

Figure 2.. Comparison of the active and pseudoactive Cas9-nucleic acid complex structures and proposed mechanism for DNA cleavage activation of Cas9.

(a–b) Zoomed-in view (a) and zoomed-out view (b) of the HNH domain docked onto the tDNA and REC lobe in the optimized pseudoactive state. (c) Close-up view of the catalytic configuration of the T4 Endo VII (N62D) ββα-Me motif complexed with a DNA substrate. (d–f) Zoomed-in views (d, f) and zoomed-out view (e) of the HNH domain docked onto the tDNA and REC lobe in the catalytically active state. (g) Schematic diagram of the proposed mechanism underlying Cas9 HNH domain conformational activation.

Figure 2—figure supplement 1.. Close-up view of the catalytic centers of some representative ββα-Me superfamily members beyond Cas9.

(a) Type II restriction endonuclease Hpy99I; (b–c) Homing endonucleases I-HmuI (b) and I-PpoI (c); (d) non-specific periplasmic nuclease from Vibrio vulnificus (Vvn); and (e) Endonuclease I from Vibrio cholerae (Vcl). See also Figure 2c for the Holliday Junction resolvase phage T4 Endo VII.

Figure 2—figure supplement 2.. Multiple sequence alignment of Type II-A and Type II-C Cas9 orthologs focusing on the ββα motif regions.

The primary sequences of Type II-A Cas9 orthologs from Streptococcus pyogenes (Spy, GI 15675041), Streptococcus thermophilus LMD-9 (Sth, GI 11662823), Listeria innocua Clip 11262 (Lin, GI 16801805), Streptococcus agalactiae A909 (Sag, GI 76788458), Streptococcus mutans UA159 (Smu, GI 4379809), Enterococcus faecium 1231408 (Ffa, GI 257893735), Treponema denticola (Tde, WP_002676671.1), and Staphylococcus aureus (Sau, GI 1027923597), together with Type II-C Cas9 orthologs from Neisseria meningitidis (Nme, WP_019742773.1), Campylobacter jejuni (Cje, WP_002876341.1) and Actinomyces naeslundii str. Howell 279 (Ana, EJN84392.1) were aligned using Cluster Omega. The alignment was illustrated by MSAViewer with default settings. The secondary structures of SpyCas9 from its apo-state crystal structure (PDB code: 4CMP) are represented at the top of the sequence alignment diagram, with the residue numbers indicated below.

Figure 2—figure supplement 3.. Lys862 making interactions with Asp837 and/or tDNA as captured in another simulation trajectory of the active state Cas9 complex.

The ββα-Me motif is represented by pink ribbons, and the residues is depicted as stick models and colored by atom types (C, dark green; N, blue; O, oxygen). The Mg²⁺ is rendered as a magenta sphere. For the tDNA, only the phosphate-sugar backbone is shown for clarity. The dashed lines denote coordination bonds.

Figure 2—figure supplement 4.. Interactions between the REC2 domain (colored cyan) and the HNH domain formed in the pseudo-active Cas9 complex structure.

The residues on the REC2 and HNH domains are colored by green and yellow, respectively. The dashed lines indicate hydrogen bonds or salt bridges.

Figure 2—figure supplement 5.. Free energy landscapes of the α structure element (residues 859 to 872) in the ββα-Me motif against different sets of reactions coordinates.

(a)-(b) Free energy landscapes of the α structure element as obtained based on the principal component analysis of the replicated simulations initiated from the N863-IN (a) and N863-OUT (b) state. (c) Free energy landscape of the α structure element derived from the combined sets of simulations. The total sampling time adds up to 50 μs and the Cα atoms are selected for the present calculations. In deriving the landscapes in (a) and (b), the simulation trajectories (500,000 snapshots in total) of each state are projected onto the same subspace defined by the two principal components (i.e., PC1 and PC2) that are calculated over the combined trajectories (1000,000 snapshots). The starting structure for each system is mapped on the PC1-PC2 plane, highlighted with a circle filled in magenta. The free landscape in (c) are constructed from the combined sets of simulations for the N863-IN and N863-OUT state, against the distance RMSD (dRMSD) with reference to the N863-IN simulation starting structure (marked with a magenta circle) and the difference in distances of D839-D861 (d_D839-D861) and D839-N863 (d_D839-N863). Note that dRMSD compares pairs of internal distances (not absolute coordinates) and is thus insensitive to translations and rotations. The representative structures at the prominent minima are illustrated below panel (c).

Figure 2—figure supplement 6.. The MD-derived metal center configuration in the active (a) and psuedoactive state (b) optimized by DFTB3 QM/MM simulations.

The drawing style and coloring scheme are as in Figure 2a and Figure 2d. Remarkable similarities have been observed at the metal center by the QM/MM simulations, confirming the desirable accuracy of the current metal ion model in MD simulations.

Figure 2—figure supplement 7.. The coordination configuration at the HNH domain active center from our original pseudo-active Cas9 complex structure derived by use of the 12–6 point-charge Mg²⁺ model (viz. the normal usage set in AMBER force field).

In this structure, the active center Mg²⁺ lost one critical coordination bond with the leaving group O3’ as compared to the catalytic configuration captured in the homologous T4 Endo VII in complex with a DNA substrate (Figure 2c). Here we refined this structure using an advanced multisite Mg²⁺ model based on a 12-6-4 LJ potential, and the resulting active center configuration (Figure 2a) well reproduced that in the T4 Endo VII system (Figure 2c).