The Effect of DNA Topology on Observed Rates of R-Loop Formation and DNA Strand Cleavage by CRISPR Cas12a

Abstract

Here we explored the mechanism of R-loop formation and DNA cleavage by type V CRISPR Cas12a (formerly known as Cpf1). We first used a single-molecule magnetic tweezers (MT) assay to show that R-loop formation by Lachnospiraceae bacterium ND2006 Cas12a is significantly enhanced by negative DNA supercoiling, as observed previously with Streptococcus thermophilus DGCC7710 CRISPR3 Cas9. Consistent with the MT data, the apparent rate of cleavage of supercoiled plasmid DNA was observed to be >50-fold faster than the apparent rates for linear DNA or nicked circular DNA because of topology-dependent differences in R-loop formation kinetics. Taking the differences into account, the cleavage data for all substrates can be fitted with the same apparent rate constants for the two strand-cleavage steps, with the first event >15-fold faster than the second. By independently following the ensemble cleavage of the non-target strand (NTS) and target strand (TS), we could show that the faster rate is due to NTS cleavage, the slower rate due to TS cleavage, as expected from previous studies.

Keywords: DNA topology; endonuclease mechanism; rapid reaction kinetics.

Figure 1

Domains and ternary structure of Lachnospiraceae bacterium ND2006 Cas12a (PDB: 5xus, TTTA PAM) [11]. Locations of “finger”, “linker” and “lid” from Stella et al. [8]; note that the “lid” is not resolved in PDB: 5xus. The putative path of the non-target strand (NTS) is shown on the structure as a thick dotted line, with the arrowhead pointing towards the RuvC active site.

Figure 2

Measurement of R-loop formation by LbCas12a using a magnetic tweezers assay. (A) DNA protospacer sequence (black) and CRISPR RNA (crRNA) showing the G-residues from in vitro transcription (brown), the pseudoknot (blue) and spacer (red). (B) Principle of the MT assay. See main text. (C) R-loop cycling experiment (1 turn s⁻¹) in the presence of 5 nM Cas12a:crRNA. Raw DNA length taken at 60 Hz (grey). Data smoothed by a 1 Hz moving average (dark colors). DNA is negatively supercoiled at 0.3 pN (red) to induce R-loop formation (in), followed by positive supercoiling to probe R-loop formation (blue), resulting in R-loop dissociation (out). Rot₀ are points where DNA turns are zero. (D) Overlay of R-loop cycles (N = 22) for negative supercoiling (in events) and positive supercoiling (out events). Cycles without Cas12a are in grey. Data was smoothed by a 1 Hz moving average. (i) and (ii) show rotation curve shifts due to captured R-loops. (E) Rotation curve shift due to R-loop events (i). Average = 1.87 ± 0.27 turns (errors = SD). (F) Examples of repetitive R-loop formation cycling (at 10 turns s⁻¹) to measure R-loop formation times. Raw and 1 Hz smoothed data are shown. (G) Mean R-loop formation/dissociation times and standard error (N = 40 to 52) as a function of torque [29]. Solid lines are fits to Equation (1) (Table 1) [27]. (H) Inverted cumulative probability over time for R-loop formation (left) and dissociation (right) used to calculate mean times in panel F.

Figure 3

Apparent DNA cleavage rates are affected by substrate topology. Cleavage of 3 nM supercoiled (A), pre-nicked circular (B) and linear (C) substrates by 50 nM LbCas12a-crRNA. Diagrams represent the substrate, intermediate and product states: supercoiled (SC); nicked (open circle, OC); full length linear (FLL); linear substrate (LIN1); and products of linear DNA cleavage (LIN2). DNA species were separated by native agarose gel electrophoresis. (The OC and LIN1 data shown are from the same gel and thus share the same marker lane). The kinetic models were simultaneously fitted (solid lines) to each of three repeats and the rate constants averaged (Table 2, with SD errors). For panel A, k_rloop was fixed as infinitely fast. For panels (B) and (C), the k₁ and k₂ values were fixed using the fits in panel (A). Plotted points are the average of the 3 repeats, errors bars as SD.

Figure 4

LbCas12a cleaves the non-target strand faster than the target strand. (A) Linear DNA substrates were ³²P-labelled on either the TS (upper panel) or NTS (lower panel). The PAM, approximate positions of NTS and TS cleavage, and sizes of the intact and cut labelled-strands are indicated. DNA species were separated by alkaline denaturing agarose gel electrophoresis. Each lane represents the reaction following LbCas12a addition (left to right): 0, 0.5, 1, 2, 3, 5, 10, 20, 30, 40, 50 and 60 min. (B) Diagrams showing overlap between substrate, intermediates and products depending on the labelled strand. Dotted lines enclose those species which will have the same labelled strand identity on a denaturing gel (C) Ordered kinetic model and (D) random kinetic model, simulated (solid and dashed lines, respectively) using values in Table 2. NTS and TS data points in both panels came from separate gels, as in panel (A), and are the average from 3 repeat experiments, with error bars as SD.