Industrial biomanufacturing, using well-tailored microbial cell factories with economically competitive titers, synthesis rates and yields (TRY), offers a potentially green and economical alternative to current petroleum-based chemical synthesis (Clomburg et al., 2017). Corynebacterium glutamicum, the famous industrial workhorse for amino acid production with a current output of over 6 million tons per year (Lee et al., 2016), is increasingly being adapted as a promising chassis for the biosynthesis of other compounds (Becker et al., 2018). However, most microorganisms, including the industrial C. glutamicum, strains have not evolved to naturally and/or efficiently produce the majority of petrochemical compounds (Lee et al., 2012). Despite the substantial rational engineering efforts devoted to developing efficient cell factories, it is still arduous to achieve competitive TRY values due to our current limited understanding of cellular metabolism (Nielsen and Keasling, 2016; Ding et al., 2020). Instead, most industrial workhorses are developed without in-depth genetic knowledge by random mutagenesis strategies such as adaptive laboratory evolution and chemical/physical mutagenesis (Ikeda, 2003; Zhang et al., 2014; Sandberg et al., 2019). Over the past few decades, random mutagenesis strategies combined with applicable selection methods have led to the development of various industrial C. glutamicum strains as well as a valuable trove of genetic diversity (Ikeda, 2003; Zhang et al., 2018; Stella et al., 2019). With the development of genome sequencing and genetic engineering tools, novel synthetic biology elements such as enzyme variants have been identified through reverse engineering (Ikeda and Takeno, 2020), which can motivate further innovation in the development of industrial C. glutamicum strains. Since point mutations (single nucleotide substitutions, insertions or deletions) are the predominant mutation type identified in industrial/evolved strains (Kvitek and Sherlock, 2013; Lang et al., 2013), large-scale point mutation analysis is highly desired for further understanding the genetic basis responsible for the evolved C. glutamicum phenotypes (Bailey et al., 2002; Nielsen and Keasling, 2016). However, one major issue in the current point mutation analysis is the genetic modification of cells to introduce the enormous numbers of point mutations needed for high-throughput screening, which appears to be an impossible task for laboratory biologists. For the case of a saturation mutation library targeting the 10 bp sequence of the core region of a bacterial promoter (−35/−10 region), the strain library size can reach 4^10, nearly one million, which is impossible to accomplish with human labor alone. Recently, a robotic system-assisted C. glutamicum automation genome editing platform (MACBETH) has been developed, with the capacity to generate thousands of single nucleotide mutant strains per month (Wang Y. et al., 2018), which opens the possibility for future robot-assisted large-scale point mutation editing. Unfortunately, as a CRISPR/Cas-deaminase-mediated base editing platform, MACBETH was unable to edit all the nucleotides of interest due to the limitation of genome-targeting scope, editing window, and base transition capability (Wang Y. et al., 2019).

Genome point mutation editing is a much more precise genetic modification than gene deletion or insertion. Although there are many genome editing techniques, which can be simply divided into ones based on homologous-recombination (HR) and non-homologous end-joining (NHEJ), few are suitable for strictly precise genomic point mutation editing (Wang et al., 2021). The nuclease/integrase/transposon-mediated NHEJ system, which requires a specific recognition site at the targeted chromosomal locus in advance and will inevitably generate chromosomal scars (such as the loxP or attB sites) after genome editing (Hu et al., 2014; Marques et al., 2020), is obviously inapplicable for point mutation editing. The RNA-guided CRISPR/Cas-based HR systems can provide scarless chromosomal modifications (Jiang et al., 2017; Wang et al., 2018a). However, additional mutations in protospacer and PAM regions are usually prerequisite for avoiding re-cutting by endonucleases, and editable genome regions are restricted due to the limited availability of guide RNAs (Wang T. et al., 2019), which also limits its application in point mutation editing. The RecT-mediated ssDNA/dsDNA HR system, which can support the editing of any nucleotide of interest without the need to repeatedly construct editing vectors (Binder et al., 2013), seems like an ideal technique. However, the ssDNA/dsDNA electro-transformation efficiency in Gram-positive C. glutamicum might be problematic (Ruan et al., 2015), especially for the industrial recombinant strains. In addition, due to the relatively short homologous arms in ssDNA/dsDNA, the off-target risk is high, especially for point mutations inside sequences with additional copies in the chromosome. In the counter-marker-assisted HR system, which is based on two rounds of single crossover HR (Schäfer et al., 1994), any nucleotide in the genome is theoretically editable without risk of introducing additional mutations/scars. The application of an editing vector carrying relatively long homologous arms can not only significantly improve the electro-transformation efficiency, but also lower the off-target risk caused by mismatching. In addition, the editing vector can be used repeatedly, which reduces the cost of vector DNA assembly. Although the efficiency of HR is relatively low, conditional lethality mediated by counter-selectable markers, such as the sucrose-lethal gene sacB (Schäfer et al., 1994) or streptomycin-sensitive gene rpsL (Kim et al., 2011), can ensure the occurrence of two rounds of single-crossover HR. Therefore, the counter-marker-assisted HR system is the most promising chromosomal point mutation technique for C. glutamicum.

The counter-marker HR mediated genome editing system is mostly implemented using non-replicating suicide plasmids (Wang et al., 2021), among which the suicide vector pK18mobsacB (GenBank: FJ437239.1) is the most widely used in C. glutamicum (Schäfer et al., 1994). For point mutation editing based on the pK18mobsacB vector, a pK18mobsacB-derived vector and a total of two rounds of crossover HR and mutant isolation are required (Figure 1). Firstly, the upstream homologous arm (UHA) and the downstream homologous arm (DHA), which contain the corresponding point mutation, are amplified from the chromosome of C. glutamicum by primer pairs 1/2 and 3/4 (named as primer-1/2/3/4), respectively. Then, the homologous arms as well as the linearized pK18mobsacB vector fragment are assembled into the pK18mobsacB-derived suicide vector, containing homologous arms and the desired point mutation. Subsequently, the resulting vector is electroporated into the cells and integrated into the chromosome in the 1^st-round of single crossover, and the resulting transformants are selected in the 1^st-round isolation based on the positive selection marker, the kanamycin resistance gene kan ^R . Next, the integrated vector is eliminated in the 2^nd-round of single crossover and the transformants are selected in the 2^nd-round of isolation based on a negative selection marker, the sucrose-lethal gene sacB. Finally, the correct mutants containing the point mutation are confirmed by sequencing. Thus, a total of two homologous arms and at least two pairs of primers (primer-1/2/3/4) are required for the whole process of point mutation editing.

Despite the development of computer-aided design (CAD) tools for the design of genetic modifications (Kalendar et al., 2011; Appleton et al., 2017a; Appleton et al., 2017b), no CAD tool is currently available for the design of counter-marker-assisted HR editing. Although some tools such as j5 (Hillson et al., 2012), Raven (Appleton et al., 2014) and FastPCR (Kalendar et al., 2017) can assist the design of optimal primers for amplifying UHA/DHA, laboratory biologists still have to manually extract and input optional templates, which is laborious, time-consuming and error-prone if it needs to be done on a large scale. In addition, the design of primers for sequence verification, one of the most important experimental steps, is often neglected by laboratory beginners, which limits the ability to correct unexpected misediting/non-editing. For instance, if adding one pair of verification primers (test-primer-1/2, Figure 1), non-editing failures caused by false-positive transformants (Ma et al., 2015) during the 1^st-round of single crossover and isolation can be avoided. In addition, unexpected mutations located around the ends (∼100 bp) of UHA-DHA cannot be precisely sequenced if simply using primer-1/2 rather than test-primer-1/3 (Figure 1) for the final sequence verification. Thus, a one-stop, comprehensive CAD tool for the whole design process of counter-marker HR mediated genome editing is highly desirable for automated and high-throughput point mutation editing in C. glutamicum.

In this study, to reduce the effort and time needed for point mutation editing design and provide a comprehensive packaged result for laboratory biologists, we developed a user-friendly online tool (Figure 1), named the Genome Editing automated Design platform for point mutation construction in Corynebacterium glutamicum (GEDpm-cg, https://gedpm-cg.biodesign.ac.cn/). The counter-selection HR system (Schäfer et al., 1994; Tan et al., 2012) and the overlap-based assembly method (Casini et al., 2015) were chosen as the loading techniques. Homologous arms and primers required for genetic modification, vector DNA assembly and sequencing verification were provided as design results. Moreover, the GEDpm-cg was built in a novel, entirely serverless architecture, with computing, as well as data storage, done in a serverless manner, which ensures flexibility in allocating computing resources. Finally, to verify the accuracy of design results generated by GEDpm-cg, three independent point mutations were experimentally implemented.

The development of microbial cell factories has been greatly facilitated by computer-aided design tools (Appleton et al., 2017b; Hillson et al., 2019; Carbonell et al., 2020), among which design tools for genome editing play an important role in liberating biologists from laborious, repetitive and error-prone design work (Montague et al., 2014; Quintin et al., 2016; Wang et al., 2019b). However, most genetic modification design tools were specifically developed to handle a specific module for a single phase of the editing process that will be more programmable, such as CHOPCHOP (Montague et al., 2014) for designing CRISPR guide RNAs, PrimeDesign (Hsu et al., 2021) for designing specifically engineered guide RNAs (pegRNAs), Merlin (Quintin et al., 2016) for designing ssDNAs, as well as j5 (Hillson et al., 2012) and Raven (Appleton et al., 2014) for designing DNA assembly primers. To further improve the automaticity of MACBETH and evaluate the off-target risk, our group previously developed an online tool (gBIG, http://gbig.ibiodesign.net/) for the high-throughput design of guide RNAs, which allowed sequence design for the base editing-mediated inactivation of over 3,000 target genes within minutes (Wang Y. et al., 2019). However, no CAD tool is currently available for one-stop design covering all the experimental steps required for genetic modification. Although laboratory biologists can employ these specific CAD tools in a stepwise manner to assist their genetic modification design, truly automated and high-throughput design is still limited by the non-standardized data exchange and input/output formats (Carbonell et al., 2020). Specifically, users have to manually extract and upload numerous templates (the optional up- and downstream homologous arms) for primer design using j5/Raven or other CAD DNA assembly tools (Hillson et al., 2012; Appleton et al., 2014), which is cumbersome in high-throughput approaches and error-prone. In this study, we developed the online tool GEDpm-cg for the automated, rapid and precise design of genomic point mutation editing in C. glutamicum. For the first time, the design of functional elements (homologous arms required for the counter-marker-assisted HR system), the vector DNA assembly (primers design for vector construction) and the sequencing verification are integrated in the single CAD tool GEDpm-cg (Figure 2). As a result, it can provide automated and high-throughput design results covering all the experimental elements required for the constructing and verification of point mutations (Figures 4C,D). To be biologist-friendly, GEDpm-cg provides an open and free web-service, and the step-by-step how-to examples as well as the in-depth descriptions of input and output files (see in the online user’s manual) are all developed to suit the needs of our biologist colleagues. Moreover, to further ensure that the point mutation is being introduced as the user desires without off-target mutations, the alignment between the targeted sequences and targeted genome is checked in advance and the possibility of off-target events is also evaluated. Finally, a testing simulation of over 10,000 single point mutations could be completed within only 5 min, and three point-mutations in the genome of C. glutamicum were experimentally constructed guided by GEDpm-cg (Figure 4G). Thus, the in-silico design results were seamlessly bridged with in-vitro vector construction and in-vivo C. glutamicum point mutation editing.

Although the emerging CRISPR/Cas genome editing systems are increasingly favored for genetic manipulation in C. glutamicum (Wang et al., 2021), the counter-selection-based system is still a reliable genome editing technique, especially for the construction of mutations that require strictly precise nucleotide editing (Stella et al., 2019). To improve the editing efficiency of this technique, various variants were developed, for example by replacing the native promoter of sacB in classical pK18mobsacB with the 18-times stronger P _lacM promoter (Tan et al., 2012), replacing the negative selection marker sacB with the novel streptomycin-sensitive gene rpsL (Kim et al., 2011) or 5-fluorouracil-lethal gene upp (Ma et al., 2015), as well as replacing single-copy and non-replicating pK18-derived vectors with multi-copy and temperature-sensitive pCGR2-derived vectors (Okibe et al., 2011). Notably, GEDpm-cg is able to flexibly support all these counter-selection-based variants upon the uploading of specific linear vector sequences by the user. What’s more, GEDpm-cg can also be used for genomic point mutation editing in other species beyond C. glutamicum based on this same approach if the users upload corresponding genome and vector sequences. Nevertheless, the upgrading of GEDpm-cg to support point mutation editing by other genome editing techniques such as CRISPR/Cas system and/or other editing types for fragment sequence editing is still expected to fulfil different users’ preferences in our future work.

With the recent technology advances in robotic/software-assisted strain engineering system, it has become feasible to enable an ultra-efficient turnover rate of design-build-test-learn synthetic biology cycle (Chao et al., 2017; Hillson et al., 2019). The robotic system-assisted CRISPR/Cas-deaminase-mediated C. glutamicum genome base editing platform, MACBETH, was developed by our colleagues in 2018. The MACBETH enables a maximal editing capacity of up to 9,000 single nucleotide mutant strains within 1 month (Wang Y. et al., 2018), which obviously exceeds the ability of human labor alone to construct no more than one hundred mutant strains per month. Since the basic experimental operations will not be beyond vector construction, plating, cultivating and screening, the MACBETH platform, based on the CRISPR/Cas9 mediated base editing, can be feasibly modified to support automated point mutation editing based on the counter-marker homologous-recombination. The combination of design automation based on GEDpm-cg and experiment operation automation based on MACBETH platform will be a superior tool for high-throughput point mutation editing of C. glutamicum.

In conclusion, we developed GEDpm-cg with superior efficiency, user-friendliness and flexibility for the design of genomic point mutation editing in C. glutamicum, which can liberate biologists from laborious, repetitive and error-prone experimental design. We believe our platform can open the possibility for large-scale mutation mining via robotic/software-assisted systems and consequently lead to a better understanding/engineering of cellular metabolism in the near future.