DNA capture by a CRISPR-Cas9-guided adenine base editor

Abstract

CRISPR-Cas-guided base editors convert A•T to G•C, or C•G to T•A, in cellular DNA for precision genome editing. To understand the molecular basis for DNA adenosine deamination by adenine base editors (ABEs), we determined a 3.2-angstrom resolution cryo-electron microscopy structure of ABE8e in a substrate-bound state in which the deaminase domain engages DNA exposed within the CRISPR-Cas9 R-loop complex. Kinetic and structural data suggest that ABE8e catalyzes DNA deamination up to ~1100-fold faster than earlier ABEs because of mutations that stabilize DNA substrates in a constrained, transfer RNA-like conformation. Furthermore, ABE8e's accelerated DNA deamination suggests a previously unobserved transient DNA melting that may occur during double-stranded DNA surveillance by CRISPR-Cas9. These results explain ABE8e-mediated base-editing outcomes and inform the future design of base editors.

Fig. 1.. Cryo-EM structure of ABE8e in a substrate-bound state.

A) Schematic representation of RNA guided DNA adenosine deamination. B) Mechanism of E. coli TadA catalysed tRNA adenosine deamination. C) Domain architecture of four generations of ABEs encoding SpCas9 N-terminally linked to either wild-type (WT) TadA, evolved TadA7.10 (TadA*, yellow) or evolved TadA8e (TadA*, red). D) Single-turnover kinetics of targeting ABE RNPs measured using dsDNA containing a single adenine. The fraction of deaminated dsDNA is shown plotted as a function of time and fitted to a single phase exponential equation. The extracted apparent deamination rates of ABE7.10 (black), miniABEmax (orange) and ABE8e (red) are 0.0010 ± 3×10⁻⁴ min⁻¹, 0.0005 ± 1×10⁻⁴ min⁻¹, and 0.585 ± 0.034 min⁻¹, respectively. Data are represented as the mean ± SD from three independent experiments. The deamination data of ABE8e and ABE7.10 are reproduced and were originally published in ref. 15. E) Mechanism of inhibition of adenosine deamination by 8-azanebularine that mimics adenosine deamination reaction intermediate. F) 3.2 Å resolution cryo-EM structure of the SpCas9-ABE8e complex. Subunits are colored: SpCas9 (white), single-guide RNA (sgRNA, purple), target-strand DNA (TS, teal), non-target strand DNA (NTS, blue) and the TadA-8e dimer (red and pink). The theoretical connectivity of the linker region is shown as a dashed orange line between the Cas9 N-terminus and either TadA-8e C-terminus.

Fig. 2.. ABE8e is a multiple-turnover enzyme with a wide editing window.

A) Surface exposed and single-stranded topology of the non-target strand (NTS) (cartoon, blue) engaged with the TadA8e domain (surface, red). B) Schematic representation of ABE8e RNP complex acting on radiolabeled trans-ssDNA or trans-dsDNA substrate containing a single adenine. (right) A representative gel showing ABE8e RNP single-turnover deamination of radio-labeled ssDNA and dsDNA over time (min) in trans. C) Single-turnover kinetics of ABE8e RNP measured using either dsDNA (deamination in cis) or ssDNA (deamination in trans) containing single adenine. The fraction of deaminated DNA plotted as a function of time, and fitted to a single exponential equation. Data are represented as the mean ± SD from three independent experiments.The apparent rate of DNA editing in trans is ~3.7 fold lower than the apparent rate of DNA editing in cis (k_app = 0.16 ± 0.01 min⁻¹ vs. k_app = 0.59 ± 0.03 min⁻¹). D) DNA deamination assay in multiple-turnover conditions using ABE8e RNP and either ssDNA (trans) or dsDNA (cis) substrate. Turnover number (a ratio of the deaminated DNA concentration and total concentration of ABE8e) was plotted as a function of time. The extracted turnover number for ssDNA deamination in trans is 4.5 ± 0.1 while for DNA in cis it is 0.97 ± 0.03. E) Gel representing single-turnover kinetics of ABE0, ABE7.10, miniABEmax, and ABE8e measured using dsDNA (cis) containing multiple adenines. Assays were performed in three independent replicates, and time points for ABE0, ABE7.10 and miniABEmax assays were taken at 0, 1, 3, 8, 24 and 32 hours, while for ABE8e at 0, 1, 5, 10, 20, 60 and 180 min. Concentrations of ABE RNPs were 1 μM for single-turnover assays and 25 nM for multiple-turnover assays. Concentrations of DNA were 2.5 nM for single-turnover assays and 250 nM for multiple-turnover assays.

Fig. 3.. Fast ABE8e kinetics reports on a new step in the SpCas9 target search pathway.

Model for SpCas9 target search mechanism: (A) apo-Cas9 – dsDNA interaction is a random collision; Cas9 bound to gRNA gets primed for a target search which initiates via random interaction with the DNA, followed by unidirectional (NTS-3’ to 5’) Cas9 RNP diffusion along dsDNA (for ~27 bp) in search of PAM and rapid dissociation from DNA unless PAM is encountered; (B) once Cas9 RNP comes across PAM its residence time on DNA increases and it unwinds first few nucleotides adjacent to PAM to probe for guide RNA complementarity; Cas9 RNP at correct target sites initiates an R-loop formation via sequential unwinding (9,10). C) Gel representing single-turnover kinetics of apo-ABE8e acting on either ssDNA (trans) or dsDNA (trans), of ABE8e RNP programmed with targeting gRNA deaminating ssDNA (trans) or dsDNA (cis), and of ABE8e RNP programmed with non-targeting gRNA deaminating ssDNA (trans) or dsDNA (trans). Time points were taken at 0, 1, 10 and 60 min. D) Gel representing single-turnover kinetics of apo-ABE8e, ABE8e targeting RNP and ABE8e non-targeting RNP acting on dsDNA that lacks PAM sequence. Time points were taken at 0, 1, 10 and 60 min. Concentrations of ABE RNPs were 1 μM and concentrations of DNA were 1 nM.

Fig. 4.. Mechanism of ABE8e substrate specificity.

A) Superposition of TadA-8e (cartoon, red) and TadA-WT (cartoon, yellow) showing the altered disposition of ɑ5-helix. B) Single-turnover kinetics of ABE8e and four ABE8e variants using dsDNA (containing a single adenosine in the NTS). The apparent deamination rates are: ABE8e 0.59 ± 0.04 min⁻¹; ABE8e-P152R 0.11 ± 0.02 min⁻¹; ABE8e-Y149F 0.05 ± 0.02 min⁻¹; ABE8e-R111T 0.05 ± 0.04 min⁻¹. and N.D. for ABE8e-REIK. Concentrations of ABE RNPs were 1 μM and concentrations of DNA were 2.5 nM. C) TadA-8e active site showing the non-target strand (NTS; sticks, blue) entering the TadA-8e substrate binding pocket (surface, red) with evolved residue R111 and Y149 shown as sticks (brown). D) TadA-8e active-site proximal pocket (surface, red) with evolved residues F84 and V106 shown as sticks (white). The E. coli tRNA U(33) is shown in yellow. E) Single-turnover RNA deamination kinetics of all four ABEs programmed with targeting sgRNA (Table S2) using hpRNA as a substrate which contains single adenine. The fraction of deaminated hpRNA plotted as a function of time and fitted to a single exponential equation. The extracted apparent deamination rates of ABE0 (blue) and ABE7.10 (black) are 1.38 ± 0.28 min⁻¹ and 0.41 ± 0.08 min⁻¹, respectively. The kinetics of miniABEmax and ABE8e were much slower, and non-exponential. Data are represented as the mean ± SD from three independent experiments. Concentrations of ABE RNPs were 1 μM and concentrations of RNA were 1 nM. F) Single-turnover kinetics of all four ABEs programmed with targeting sgRNA (Table S2) using ssRNA as a substrate which contains single adenine. The fraction of deaminated hpRNA plotted as a function of time and fitted to a single exponential equation. The extracted apparent deamination rate of ABE8e (red) is 0.02 ± 0.01 min⁻¹. The kinetics of ABE0, ABE7.10, and miniABEmax were much slower, and non-exponential. Data are represented as the mean ± SD from three independent experiments. Concentrations of ABE RNPs were 1 μM and concentrations of RNA were 1 nM.