Rational Design of Small Molecules to Enhance Genome Editing Efficiency by Selectively Targeting Distinct Functional States of CRISPR-Cas12a

Abstract

CRISPR-Cas12a, a type-V CRISPR-Cas endonuclease, is an effective genome editing platform. To improve the gene editing efficiency of Cas12a, we rationally designed small molecule enhancers through a combined computational approach. First, we used extensive molecular dynamics (MD) simulations to explore the conformational landscape of Cas12a from Acidaminococcus (AsCas12a), revealing distinct conformational states that could be targeted by small molecules to modulate its genome editing function. We then identified 57 compounds that showed different binding behavior and stabilizing effects on these distinct conformational states using molecular docking. After experimental testing 6 of these 57 compounds, compound 1, quinazoline-2,4(1H,3H)-dione, was found particularly promising in enhancing the AsCas12a-mediated genome editing efficiency in human cells. Compound 1 was shown to act like a molecular “glue” at the interface between AsCas12a and crRNA near the 5′-handle region, thus specifically stabilizing the enzyme–crRNA complex. These results provide a new paradigm for future design of small molecules to modulate the genome editing of the CRISPR-Cas systems.

Graphical Abstract

The CRISPR-Cas systems are a complex molecular machinery consisting of multiple components including the Cas protein, CRISPR guide RNA (gRNA), and protospacer adjacent motif (PAM) in the targeting DNA molecules of the genome.^1–7 Moreover, genome editing involves multistages of actions such as recruitment of gRNA, PAM detection, formation of a gRNA-DNA duplex inside the Cas protein, not to mention myriads of correction steps.^7–14 Despite this complexity, the genome editing function is to a large extent governed by the three-dimensional structure and dynamics of the enzymes.^{8,12,13,15,16} Thus, modulation of the Cas protein structure and/or Cas protein–nucleic acid interactions with small molecules is a promising strategy for the improvement of the CRISPR-Cas mediated genome editing efficiency.

CRISPR-Cas12a (also known as Cpf1), a type-V CRISPR-Cas effector endonuclease, exhibits effective genome-editing capability in multiple cell and animal models.^{7,13,14,17–20} The guide RNA for Cas12a is a single CRISPR-RNA (crRNA) with a length of 44 nucleotides.⁶ Previously, several small molecules have been identified to improve the DNA repair process through blocking the nonhomologous end joining (NHEJ) or activating the abundance of proteins involved in homology directed repair (HDR)/damage-dependent signaling, after the Cas12a-mediated genome editing.^21,22 Building upon the conventional structure-based drug design approach, we present here a strategy to discover genome editing enhancers by rationally designing small molecules that can bind AsCas12a and differentially stabilize specific functional conformations or states of AsCas12a.

RESULTS AND DISCUSSION

To thoroughly explore the conformational space of the protein in the different functional states, e.g., in the presence and the absence of crRNA and DNA, we performed microsecond molecular dynamics (MD) simulations of the AsCas12a ternary (T-State), binary (B-State), and apo (A-State) systems (Figure 1B), respectively (3 μs in total) (Table 1). We then clustered each simulation trajectory according to root-mean-squared deviation (RMSD) of Cα positions of the protein, from which 11 representative conformations were chosen, one from each of the most populated clusters (1 ternary, 6 binary, and 4 apo structures). These structures are expected to represent AsCas12a in its different functional states. We hypothesized that if a small molecule could bind AsCas12a and differentially stabilize these conformations along the genome cleavage pathway (Figure 1A), then this molecule would be able to affect AsCas12a’s activity through shifting the free energy equilibria among these conformations, thus modulating the enzyme’s genome editing efficiency.

Figure 1.

Rational design of small molecules differentially stabilizing AsCas12a at different functional states. (A) Conformational landscape of CRISPR-AsCas12a projected along the first two principal component (PC) modes sampled by microseconds of molecular dynamics simulations. The PCs of Cα atom fluctuations were computed using principal component analysis (PCA). (B) Snapshots of different functional states of CRISPR-AsCas12a. T: ternary; B: binary, and A: apo. (C) Scatter plot showing the docking score differences ΔScore vs the docking scores for all ~220,000 compounds. Ability of a small molecule to distinguish different functional states of the protein is reflected by a lower ΔScore. Compounds 1 and 2 are labeled. (D) Compound 1 bound at the protein–crRNA interface in the ternary AsCas12a system. Interacting residues were shown sticks. crRNA was shown in pink. Illustrations were made using VMD.²³

Table 1.

Simulation Summary

index	system	Time (μs)	size	atom number	clusters identified
1	Ternary	1.0	216 × 163 × 163	520,620	1
2	Binary	1.0	203 × 153 × 153	430,664	6
3	Apo	1.0	203 × 153 × 153	430,624	4

We obtained 11 representative structures corresponding to the different functional states of AsCas12a from the above MD simulations. Subsequently, a virtual screening of ~220,000 compounds against all these 11 AsCas12a structures was carried out, which yielded 57 compounds that showed reasonably strong binding to at least one functional state of AsCas12a (i.e., binding score ≤ −9.0 kcal/mol) as well as preferential binding for the ternary or binary structure vs the apo structure (i.e., ΔScore ≥ 2.7 kcal/mol). By performing a global docking, we were able to identify small molecules that bind to unique sites only present in distinct functional states of AsCas12a. Many small molecules identified appear to act as a molecular “glue”, which enhances the interactions between individual components of the system, i.e., between the AsCas12a and the crRNA in the ternary and binary systems. Finally, based on some practical considerations (e.g., molecular weight < 500 g/mol, 0 < CLogP < 2.0 and availability), six compounds were selected for experimental validation (Table 2).

Table 2.

Chemical Compound Information

ZINC	CAS No.	Chemical structure	Compound code
ZINC00150158	86-96-4		1
ZINC04430890	13548-68-0		2
ZINC00155806	17647-60-8		3
ZINC45373881	1038549-25-5		4
ZINC04764008	1009315-76-7		5
ZINC09600261	868981-42-4		6

We first examined the cytotoxicity of these compounds in human HEK293T cells by an MTT assay. Minimal cytotoxicity was observed in cells treated with 5 μM and 10 μM of each compound (Figure S1). The increasing concentrations of compounds 3 and 6 led to higher cytotoxicity such as 69% and 73% cell death, respectively, at the concentration of 50 μM. Then, we used human HEK293T cells to evaluate these six compounds for their effects on genome editing efficiency of crRNA targeting DNMT1 (crDNMT1). Compounds 1, 2, 4, and 6 significantly increased genome editing efficiency (Table 2, Figures 2A and S2A). Then, we studied their effects on the HEK293-AsCas12a cells, stably expressing AsCas12a. Compound 1 significantly enhanced the gene-editing efficiency of crDNMT1, while compound 2 did not (Figures 2B and S2B). Similarly, compound 1 increased the gene-editing efficiency of crDNMT1 in U87 cells (Figures 2C and S3A). We then investigated the crRNAs targeting FANCF-2 (crFANCF-2). Consistent enhancement of gene editing was found with the treatment of compound 1 in HEK293-AsCas12a cells (Figures 2D, S3B, and S3C). Increase of compound 1 concentrations further improved the gene-editing efficiency of crDNMT1 in 293T-AsCas12a cells in comparison to the control group (Figure 2E). To further characterize on-target and off-target effects, we conducted targeted deep sequencing of the DNA samples isolated from the experiments in Figure 2E. The results showed that compound 1 increased the gene-editing efficiency by 25%, without affecting off-target sites compared with the control group (Figure S4). Taken together, compound 1 was found to particularly enhance the AsCas12a-mediated genome editing efficiency in human cells. It enhances the genome editing directly without affecting the DNA repair process, which may also decrease the off-targeting effect.

Figure 2.

Chemical compounds affect the genome editing efficiency of the AsCas12a-crRNAs system. (A) Effects of chemical compounds on gene-editing efficiency of AsCas12a-crDNMT1–3 in 293T cells. (B) Effects of compounds 1 and 2 on gene-editing efficiency of crDNMT1–3 in HEK293-AsCas12a cells. (C) Compound 1 increased gene-editing efficiency of AsCas12a-crDNMT1–3 in U87 cells. (D) Compound 1 increased the gene-editing efficiency of crFANCF-2 in HEK293-AsCas12a cells. (E) Effects of increased concentrations of compound 1 on gene-editing efficiency of crDNMT1–3 in HEK293-AsCas12a cells. Data are shown as mean ± SD (n = 3); *P < 0.05; **P < 0.01; ***P < 0.001, two-tailed t test.

Disrupting protein–protein interactions by small molecules has become an increasingly popular strategy in drug discovery. However, the design of small molecules that stabilize protein–protein interactions is under appreciated, whereas the proteolysis targeting chimera (PROTAC) technique that uses heterobifunctional small molecules to promote transient interactions between a target protein and an E3 ligase to degrade the protein has recently attracted a lot of attention.^24,25 An essential premise of this work is that the AsCas12a function can be modulated by the design of small molecules that selectively stabilize the different functional states of the enzyme complex.

We first employed large-scale MD simulations to thoroughly explore the conformational space of AsCas12a in the different functional states. The enzyme was found to adopt distinct conformations, which are believed to correspond to different functional states of the enzyme along the genome cleavage pathway (Figure 1A). We thus obtained representative structures of AsCas12a from the simulations and then computationally docked two libraries of ~220,000 compounds to these selected structures. The “hit” compounds were chosen based on the docking score difference between the binding of a molecule to two different functional states of AsCas12a in addition to the conventional scoring criteria. This filter resulted in 57 small molecules that can bind differentially to the different states of AsCas12a. After experimental testing of 6 of these compounds, we confirmed compound 1, quinazoline-2,4(1H,3H)-dione, as a small molecule enhancer of AsCas12a. Compound 1 was found to act like a molecular “glue” at the interface between AsCas12a and crRNA near the 5′-handle region (Figure 1D) to stabilize the complex structure. It forms extensive electrostatic interactions with positively charged residues Lys748, Lys1029 and polar residue Asn759 (Figure 1D). Its aromatic rings stacking against the nearby nucleotides may further strengthen the “gluing” effect.

CONCLUSION

Overall, our molecular dynamics simulations revealed distinct conformations in AsCas12a that could be targeted by small molecules to modulate its genome editing function (Figure 1C), which provided a unique approach to understand the CRISPR-Cas systems and rationally design new agents to modulate the dynamic function of the CRISPR-Cas systems.

ABBREVIATIONS

CRISPR

clustered regularly interspaced short palindromic repeats

NHEJ

homologous end joining

HDR

directed repair

MD

molecular dynamics

RMSD

root-mean squared deviation

PC

principal component

PROTAC

proteolysis targeting chimera