Since bacterial cells synthesize valuable metabolites as encoded by their genomes, precise editing of microbial genomes is indispensable for the design of microbial cell factories. CRISPR/Cas9 (or Cpf1) technologies have been recently developed to edit genome sequences in numerous cellular platforms including C. glutamicum [35]. The PAM sequence (5′-NYTV) of CRISPR/Cpf1 [36] does not restrict editing of the genome of industrial C. glutamicum strains with G + C contents of 53.8% [37]. Moreover, owing to the potential toxicity of CRISPR/Cas9, the CRISPR/Cpf1 system of C. glutamicum has received increasing attention [12, 18, 38, 39].
Herein, we integrated cpf1 in the C. glutamicum genome because frequent transformations were laborious for introducing a genome editing tool into C. glutamicum with a low transformation efficiency. First, cpf1 was introduced into the C. glutamicum ATCC13869 genome via a single-crossover to generate strain HK1220 (Fig. 1a). recT products facilitate the oligonucleotide-directed homologous recombination in C. glutamicum (Fig. 1b) [40, 41]. After genome engineering, plasmids can be eliminated through high temperature treatment, and cpf1 can be excised from the HK1220 genome, using a counter-selectable sacB marker (Fig. 1c).
Negative selection using CRISPR/Cpf1 facilitates the survival of genome-edited cells; however, unedited cells are eliminated through double-stranded breaks at target DNA sequences. Therefore, CRISPR/Cpf1 can increase the editing efficiency of surviving cells [42]. After transformation of IPTG-induced HK1220 cells harboring the RecT plasmid (pHK489) with single-stranded mutagenic oligonucleotides and crRNA plasmids, the surviving cells putatively harboring the desired mutations were obtained through negative selection (Fig. 3a). The use of double, triple, and quadruple mutagenic oligonucleotides successfully introduced the TAG stop codon in the middle of crtEb with an editing efficiency (i.e., the proportion of pink colonies) of 91.5–95.3%. However, we rarely obtained pink colonies (0.6%) when using single-base-mutagenic oligonucleotides (Fig. 3b). Moreover, Sanger sequencing revealed successful single-base edits in 2 of 9 selected pink colonies generated through target-matched crRNA (pHK493) (Fig. 5), probably owing to mismatch tolerance, whereby the Cpf1/crRNA complex can recognize and cleave both single-base-edited and unedited targets (Fig. 3c), which have been assessed to resolve off-target effects [34, 43, 44].
Even upon transformation of only crRNA plasmids into cells without oligonucleotides, we still observed numerous surviving cells (~ 102 CFU/µg DNA of pHK493), probably owing to null cpf1 mutations or the repair of double-strand breaks in target DNA sequences. Furthermore, heterogeneity in colony shape and size was observed primarily in cases of failed negative selection (Additional file: Figure S1 and S2). Accurately edited colonies were larger than unedited colonies in our Cpf1-mediated study, while larger colonies were false-positive during Cas9-mediated genome editing of C. glutamicum [38]. Therefore, colony size does not reflect successful genome editing on using CRISPR-mediated negative selection.
To differentiate single-base-edited targets from unedited targets, mismatched crRNAs were designed and used for precise CRISPR/Cpf1-mediated negative selection (Fig. 4a). With single-base-mutagenic oligonucleotides, different target-mismatched crRNA plasmids were transformed for single-base editing of T150G (i.e., introduction of TAG stop codon) in crtEb. In cases of single-mismatched crRNAs (from pHK494 and pHK497), and one of double-mismatched crRNAs (from pHK495), 14.9%, 99.7%, and 91.5% of surviving colonies were pink owing to intracellular lycopene accumulation (Fig. 4b). Subsequently, pink colonies were randomly selected from each agar plate and their genomes were subjected to Sanger sequencing. Three of nine colonies were correctly edited among 14.9% pink colonies. The DNA sequences were accurately edited in all colonies among 99.7% and 91.5% of surviving colonies (Fig. 5), indicating that even if mutants harboring the desired phenotypes were obtained among the colonies obtained through negative selection, using the CRISPR/Cpf1 system, the efficiency of harboring a genotype that is precisely altered to the base sequence can be much lower.
The transformation efficiencies reflected between 102 and 104 CFU/µg crRNA plasmid among genome-edited cells. However, in one case of double- and two cases of triple-mismatched crRNAs (from pHK498, pHK496, and pHK499), no pink colonies were observed. Moreover, the number of surviving colonies increased to 105–106 (CFU/µg crRNA plasmid). The increased number of surviving colonies indicates that Cpf1/target-mismatched crRNAs could not accurately recognize the targets, and consequently, improper negative selection facilitated the survival of all transformants on agar plates.