Highly efficient genome editing by CRISPR-Cpf1 using CRISPR RNA with a uridinylate-rich 3'-overhang

Abstract

Genome editing has been harnessed through the development of CRISPR system, and the CRISPR from Prevotella and Francisella 1 (Cpf1) system has emerged as a promising alternative to CRISPR-Cas9 for use in various circumstances. Despite the inherent multiple advantages of Cpf1 over Cas9, the adoption of Cpf1 has been unsatisfactory because of target-dependent insufficient indel efficiencies. Here, we report an engineered CRISPR RNA (crRNA) for highly efficient genome editing by Cpf1, which includes a 20-base target-complementary sequence and a uridinylate-rich 3'-overhang. When the crRNA is transcriptionally produced, crRNA with a 20-base target-complementary sequence plus a U₄AU₄ 3'-overhang is the optimal configuration. U-rich crRNA also maximizes the utility of the AsCpf1 mutants and multiplexing genome editing using mRNA as the source of multiple crRNAs. Furthermore, U-rich crRNA enables a highly safe and specific genome editing using Cpf1 in human cells, contributing to the enhancement of a genome-editing toolbox.

Fig. 1

Improved dsDNA breakage efficiency of Cpf1 in vitro by U-rich crRNA. a Dependence of the indel efficiency of AsCpf1 on the 3′-overhang sequence of crRNA. Variations of the 3′-terminal three bases of crRNA affected the AsCpf1 indel activity in vivo, and the presence of 3-mers of uridine was the most critical for the improved AsCpf1 activity. b Improved in vitro dsDNA cleavage activity of AsCpf1 by the 3′-end uridine-rich crRNA. Vector constructs (500 ng) were digested in vitro in the presence of a ribonucleoprotein complex of AsCpf1 and crRNA (50 ng) at various incubation times (20–60 min) with variable amounts of AsCpf1 (2–7 μg) in a 20-μL reaction. c Positional ratio of nucleotide counts of crRNA template DNA in the protospacer region as assessed by deep sequencing analysis. The negatively selected E. coli cells carry less-efficient crRNA-encoding plasmid DNA, thereby rendering higher count numbers in deep sequencing. dCpf1 was used to normalize the variations in the crRNA library synthesis and experimental procedures (n = 3). All data are presented as means±standard deviations. d Optimal configuration of AsCpf1-crRNA for yielding the highest indel efficiency. Twenty-nt target sequence should be followed by uridine-rich sequences independently of the subsequent on-target sequences in the crRNA for an optimal AsCpf1 indel activity. e Dependence of the length of the 3′-end uridines of the crRNA on the in vitro AsCpf1 activity. The increment of the length of uridines of the chemically synthesized crRNAs led to the improved AsCpf1 activity in vitro with up to eight bases. f Scheme for an in vitro dsDNA breakage activity assay. Partially digested plasmids were used to transform DH-5α E. coli cells, and the AsCpf1 activity was calculated from the number of colonies formed on ampicillin-containing plates. g Validation of the highly efficient activity of AsCpf1 by the U-rich 3′-overhang in crRNA. Replacements of the uridines with non-uridine bases at any location and at any number resulted in decreased AsCpf1 activity. Data are presented as the mean±standard deviation. *p < 0.05, **p < 0.01, compared with U₈ (n = 3), two-tailed Student’s t test

Fig. 2

Optimized configuration of crRNA for highly efficient genome editing in vivo. a Scheme of the in vivo assay to determine the most efficient configuration of the crRNA. Cpf1-encoding plasmids were co-transfected with crRNA-encoding PCR amplicons into HEK-293T cells, and the Cpf1 activity was assessed by T7E1 indel assays. b Improved indel efficiency of AsCpf1 in vivo by the U-rich 3′-overhang following a 20-nt target-complementary sequence in transcribed crRNAs. This gel image is a representative result of three repeated experiments. Indel values are mean±standard deviation. c Improved indel efficiency by the 3′-end U-rich guide RNA as a unique feature of Cpf1. 3′-Proximal addition of uridinylates did not change the indel efficiency of SpCas9 in vivo. This gel image is representative of three repeated experiments. Indel values are the mean±standard deviation. d Improved indel efficiency of AsCpf1 in vivo by increased uridinylate lengths. The AsCpf1 activity was improved by the increased lengths of 3′-end uridinylates up to 8–10 mers for the chemically synthesized crRNA and up to six bases for the crRNA-encoding PCR amplicons. e Optimized 3′-end configuration of crRNA for highly efficient genome editing. Addition of the U4AU4 3′-overhang in crRNA maximized the indel efficiency of AsCpf1. *p > 0.05, **p < 0.05, ***p < 0.01 (n = 3), two-tailed Student’s t test. f The optimal target length for the use of a U-rich crRNA. A target length of 20 (±1) nt was optimal for the U-rich crRNA. g Validation of the optimal crRNA configuration for highly efficient genome editing using CRISPR-Cpf1. The optimal configuration of the U₄AU₄ 3′-overhang in addition to a 20-nt target-matched sequence was identically applied for LbCpf1 as well as AsCpf1, but not for SpCas9. This gel image is representative of three repeated experiments. Indel values are the mean±standard deviation

Fig. 3

Large-scale comparison of the genome-editing efficiencies of CRISPR-AsCpf1 and CRISPR-SpCas9. a A dot plot of the indel efficiencies of AsCpf1 and SpCas9 in HEK-293T cells. The indel efficiencies of AsCpf1 and SpCas9 were compared on common targets with a sequence of 5′-TTTV(N)₂₀NGG-3′, where V is A, C, or G. For the AsCpf1 activity, the conventional crRNA (for con-AsCpf1) configuration was compared with our optimized U-rich crRNA (for Opt-AsCpf1). b A box-and-whisker plot for the indel efficiencies of AsCpf1 and SpCas9. The graph consists of median values, lower and upper quartile, and outer standard deviations. *p = 0.003, **p = 0.00003, ***p = 0.29, two-tailed Student’s t test

Fig. 4

No influence of the U-rich crRNA on off-target effects. a Comparison of the off-target activity of conventional and U-rich crRNA-guided AsCpf1 at potential off-target sites. Off-target activity was measured by deep sequencing at potential off-target sites with <2 bulges and 2 mismatches against an on-target sequence of phospho-tyrosine kinase 6 (PTK6). The use of U-rich crRNA did not harm the specificity of AsCpf1, whereas it contributed to the increased on-target activity of Cpf1. b Comparison of the off-target activity of AsCpf1 between the conventional and U-rich crRNA with a one-base mismatch against an on-target sequence. The indel efficiency was calculated from the T7E1 indel assays. The use of a U-rich crRNA effectively lowered the tolerance of a single base mismatch at several positions, thereby contributing to the enhanced AsCpf1 specificity. *p < 0.05, compared with U-rich crRNA, two-tailed Student’s t test. c The genome-wide Circos plot for the off-target sites identified by Digenome-seq analysis with a DNMT1 on-target site. d The number of off-target sites for canonical and U-rich crRNA. e The sequence logo analysis for recognition of the target sequence. The recognition pattern was equally conserved for the both crRNAs

Fig. 5

Applications of the U-rich crRNA to multiplexed genome editing and PAM-divergent AsCpf1 variants. a Simultaneous improvements of the indel efficiencies of multiple targets by an array of U-rich crRNAs. crRNA-encoding sequences with three different targets and one scrambled one were cloned into the 3′-UTR region of the eGFP gene in the pEGFP-C1 vector (Clontech). The U-rich crRNAs have target sequences with a 20-base match plus an additional T4AT6, while each control crRNA only has 23-base match target sequences. HEK-293T cells were transfected with 5 μg each of Cpf1-encoding and crRNA-encoding vectors. The indel efficiency was calculated after normalization with the transfection efficiency as assessed by the green-fluorescent cell counts. b, c Application of U-rich crRNA to AsCpf1 PAM variants. *p < 0.001, **p < 0.01 (n = 3), two-tailed Student’s t test. b Three targets were selected as common targets for the WT and RR variant of AsCpf1, which have TTTA and TYCC PAM sequence in each strand. The indel efficiency of WT AsCpf1 or the RR variant was investigated in the presence of the canonical or U-rich crRNA. *p < 0.001, **p < 0.01 (n = 3), two-tailed Student’s t test. c Three different targets with a TTTA PAM sequence were subjected to indel mutation by transfecting HEK-293T cells with the WT or the RVR variant of AsCpf1, each of which was guided by either canonical or U-rich crRNA. *p < 0.001, (n = 3), two-tailed Student’s t test

Fig. 6

Improved binding affinity of the crRNA-AsCpf1 complex by the U-rich 3′-overhang. a The endogenous levels of the crRNAs were assessed by northern blot analysis following transfection of the crRNA template PCR amplicons into HEK293-T cells. The intracellular levels of 38s/18s RNA were used as a loading and quality control of the endogenous RNAs. The transfection efficiency was also controlled by the PCR intensities of the crRNA template DNA isolated from the transfected HEK293-T cells. This result is representative of three repeated experiments. b The indel efficiency was monitored at a DNMT1 locus following transfection of the AsCpf1 plasmid constructs and synthetic guide RNAs into HEK293-T cells. The 3′-terminal four nucleotides of crRNA and sgRNA for AsCpf1 and SpCas9, respectively, were covalently linked with the phosphorothiate group. *p = 0.008, **p = 0.39 (n = 3), two-tailed Student’s t test. c The indel efficiency was measured using the crRNA and tracrRNA of the CRISPR-Cas9 system. The truncated crRNAs with a length of 53 nt and 63 nt were compared with those with the U4AU4 3′-overhang. This gel image is representative of three repeated experiments. d Binding experiments were conducted using a microscale thermophoresis (MST) system. The binding affinities of guide RNAs and the effector proteins (SpCas9 and AsCpf1) were measured using Monolith NT.115 (NanoTemper Technologies GmbH). All samples were loaded into NanoTemper standard capillaries and repeated three times for each measurement. e Isothermal titration calorimetry (ITC) measurements were conducted on an Auto-iTC200 Microcalorimeter (GE Healthcare) at 25 °C in PBS buffer (pH 7.4). The purified recombinant AsCpf1 proteins (5 μM) were titrated with 50 μM synthetic crRNAs with 2 μL injections. K_a = (1.90±0.87) × 10⁸ M⁻¹ for the U-rich crRNA and (1.15±0.54) × 10⁷ M⁻¹ for the canonical crRNA. ΔH = –31.92±1.79 kcal/mol and −22.86±1.86 kcal/mol for the U-rich and canonical crRNA, respectively. The values are the average of three independent experiments