High-Fidelity Cytosine Base Editing in a GC-Rich Corynebacterium glutamicum with Reduced DNA Off-Target Editing Effects

Abstract

Genome editing technology is a powerful tool for programming microbial cell factories. However, rat APOBEC1-derived cytosine base editor (CBE) that converts C•G to T•A at target genes induced DNA off-targets, regardless of single-guide RNA (sgRNA) sequences. Although the high efficiencies of the bacterial CBEs have been developed, a risk of unidentified off-targets impeded genome editing for microbial cell factories. To address the issues, we demonstrate the genome engineering of Corynebacterium glutamicum as a GC-rich model industrial bacterium by generating premature termination codons (PTCs) in desired genes using high-fidelity cytosine base editors (CBEs). Through this CBE-STOP approach of introducing specific cytosine conversions, we constructed several single-gene-inactivated strains for three genes (ldh, idsA, and pyc) with high base editing efficiencies of average 95.6% (n = 45, C6 position) and the highest success rate of up to 100% for PTCs and ultimately developed a strain with five genes (ldh, actA, ackA, pqo, and pta) that were inactivated sequentially for enhancing succinate production. Although these mutant strains showed the desired phenotypes, whole-genome sequencing (WGS) data revealed that genome-wide point mutations occurred in each strain and further accumulated according to the duration of CBE plasmids. To lower the undesirable mutations, high-fidelity CBEs (pCoryne-YE1-BE3 and pCoryne-BE3-R132E) was employed for single or multiplexed genome editing in C. glutamicum, resulting in drastically reduced sgRNA-independent off-targets. Thus, we provide a CRISPR-assisted bacterial genome engineering tool with an average high efficiency of 90.5% (n = 76, C5 or C6 position) at the desired targets. IMPORTANCE Rat APOBEC1-derived cytosine base editor (CBE) that converts C•G to T•A at target genes induced DNA off-targets, regardless of single-guide RNA (sgRNA) sequences. Although the high efficiencies of bacterial CBEs have been developed, a risk of unidentified off-targets impeded genome editing for microbial cell factories. To address the issues, we identified the DNA off-targets for single and multiple genome engineering of the industrial bacterium Corynebacterium glutamicum using whole-genome sequencing. Further, we developed the high-fidelity (HF)-CBE with significantly reduced off-targets with comparable efficiency and precision. We believe that our DNA off-target analysis and the HF-CBE can promote CRISPR-assisted genome engineering over conventional gene manipulation tools by providing a markerless genetic tool without need for a foreign DNA donor.

Keywords: CRISPR; Corynebacterium glutamicum; cytosine base editor; gene editing; genome editing; industrial bacteria; nonsense mutation; off-target.

FIG 1

Construction of cytosine base editor (CBE)-STOP for genetically engineering C. glutamicum and the features of pCoryne-BE3. (a) Scheme of CBE using pCoryne-BE3 and pCoryne2-single-guide RNA (sgRNA) in C. glutamicum. After DNA replication or repair, a C•G-to-T•A conversion occurred. CBE-STOP generates stop codons (TGA, TAG, or TAA) in the coding strand sequence by converting the targeted C to T of CAA (Gln), CAG (Gln), or CGA (Arg) in the coding strand and in the noncoding strand of TGG (Trp) at the editing window. The protospacer adjacent motif (PAM) site for the CBE-STOP was 5′-NGG. (b) Targetable coding sequences by CBE-STOP with at least one stop codon in the C4 to C8 position (Fig. S1) using either 5′-NGG PAM or 5′-NG PAM. (c) Mutation frequencies (%) or base editing performance (%) at the editing-window positions. Points in bars represents each mutation frequency for eight different protospacers at editable window (C3 to C12). See the detail data (Table S6). (d) Mutation frequencies (%) for sequence-context dependency. C4 positions were only available with the GC motif. The error bar represents the standard deviation (SD). CDS, coding sequence; UGI, uracil glycosylase inhibitor; C. glutamicum, Coryebacterium glutamicum; rAPOBEC1, rat cytidine deaminase; XTEN, a 16 amino acid flexible linker; nCas9, a DNA nickase; sgRNA, single guide RNA; Spc, spectinomycin; Cm, chloramphenicol.

FIG 2

Application of CBE-STOP for single or multiple gene inactivation in C. glutamicum. (a) Phenotypic results of WT and ldh-STOP mutants (ldh-W57Ter, ldh-W134Ter, and ldh-R155Ter). Each ldh mutant with a different genotype constructed with CBE-STOP showed comparable growth rate and glucose consumption capacity compared to WT (left). Solid circles, optical density at 600 nm (O.D.₆₀₀); squares, glucose; black, WT; red, ldh-W57Ter; blue, ldh-W134Ter; purple, ldh-R155Ter. Cultivations were done in at least triplicate. The data represent mean values, and the error bars represent the standard deviations. ldh-STOP mutant showed no lactate product 12 h after transformation (right). (b) Phenotypic results of WT (black), Δpyc (red), pyc-Q279Ter (green), and pyc-W928Ter (blue). (c) Phenotypic results of the cell pellets of wild type (WT) (right) and idsA-W178Ter (left). (d) A reconstruction of the metabolic pathway for succinate production in BOL-1-STOP. (e) Flow diagram for a single CBE-STOP and multiple CBE-STOPs. (f) Mutation frequencies of the multiplexed CBE-STOP mutants to construct BOL-1-STOP in a sequential order (ldh → pqo → pta → ackA → actA). (g) Production of organic acids (succinate, lactate, adn acetate) by WT, BOL-1, BOL-1-STOP, and ldh-W134Ter. The error bars represent the standard deviation (SD).

FIG 3

Genome-wide WGS analysis of the off-targets induced by CBE. (a) The number of substitutions in ldh-W134Ter mutant sequenced by either PacBio or Illumina. The number of substitutions confirmed by deep sequencing for ldh-W134Ter mutant and BOL-1-STOP. Whole-genome sequencing (WGS) was duplicated. (b) Mutation frequencies (%) of the bases for ldh-W134Ter mutant and BOL-1-STOP (c) Substitution frequencies (%) in either coding sequences or promoter regions in ldh-W134Ter mutant and BOL-1-STOP. (d) DNA logo for sequence context of the off-targets for either ldh-W134Ter mutant or BOL-1-STOP. The flanking sequences (2 bp on either side) were aligned at target cytosine positions. SNV, single-nucleotide variant.

FIG 4

Genome-wide WGS analysis of the off-targets induced by high-fidelity (HF) CBE. (a) A scheme of the HF-CBE-STOP strain development of ldh-W134Ter using either YE1-BE3 or BE3-R132E. The WGS experiments were duplicated using Illumina. (b) Number of substitutions in ldh-W134Ter mutant using HF-CBEs, compared to original BE3. (c) Mutation frequencies (%) of the bases for ldh-W134Ter mutant using either BE3, YE1-BE3, or BE3-R132E. (d) Frequency of C-to-T mutations (%) for all sgRNAs tested using either YE1-BE3 (n = 94, colony), or BE3-R132E (n = 94, colony) (Table S8). (e) Mutation frequencies (%) or base editing performance (%) in arbitrary mutants at the editing-window positions using HF-CBEs. See the detail data (Table S8). (f) Mutation frequencies (%) for sequence-context dependency using HF-CBE (pCoryne-BE3-R132E). C4 and C7 positions were only available with the GC motif. The error bars represent the SD. (g) Mutation frequencies (%) for sequence-context dependency using HF-CBE (pCoryne-YE1-BE3) (Table S9). The C4 and C7 positions were only available with the GC motif.

FIG 5

Multiplexed base editing by high-fidelity (HF) CBE. (a) Serial multiplex base editing using pCoryne-BE3-R132E (HF-CBE) and various sgRNA vectors targeting ldh, pqo, pta, ackA, or actA genes. For each round, competent C. glutamicum harboring pCoryne-BE3-R132E cells and a sgRNA vector were used for base editing. The sgRNA vectors were cured for the next rounds. Mutation analysis was shown for the C positions with the motif and mutation efficiencies (the numbers of edited colonies among picked colonies). (b) Simultaneous multiplex base editing using pCoryne-BE3-R132E (HF-CBE) and sgRNA vectors targeting both sdhCD, sdhA, and sdhB genes (round 1). The multiple targets for the second round were the pqo, pta, and ackA genes. Mutation analysis was shown for the C positions with mutation efficiencies (the numbers of edited colonies among picked colonies). Strains with asterisks were used for WGS analysis. PTC, premature termination codon.

FIG 6

Genome-wide WGS analysis of the off-targets induced by HF-CBE (pCoryne-BE3-R132) for the multiplexed base editing. (a) The number of substitutions in ldh, ldh-pqo-pta, sdhCD-sdhA-sdhB, and sdhCD-sdhA-sdhB-pta mutants using pCoryne-BE3-R132E. WGS was duplicated. (b) Mutation frequencies (%) of the bases for those MGE mutants using pCoryne-BE3-R132E. (c) DNA logo for sequence context of the off-targets for those MGE mutants. The flanking sequences (2 bp on either side) were aligned at target cytosine positions. MGE, multiplexed genome editing; SNP, single-nucleotide polymorphism.