Genetic Engineering of Carbon Monoxide-dependent Hydrogen-producing Machinery in Parageobacillus thermoglucosidasius

The metabolic engineering of carbon monoxide (CO) oxidizers has the potential to create efficient biocatalysts to produce hydrogen and other valuable chemicals. We herein applied markerless gene deletion to CO dehydrogenase/energy-converting hydrogenase (CODH/ECH) in the thermophilic facultative anaerobe, Parageobacillus thermoglucosidasius. We initially compared the transformation efficiency of two strains, NBRC 107763T and TG4. We then disrupted CODH, ECH, and both enzymes in NBRC 107763T. The characterization of growth in all three disruptants under 100% CO demonstrated that both enzymes were essential for CO-dependent growth with hydrogen production in P. thermoglucosidasius. The present results will become a platform for the further metabolic engineering of this organism.

), were transformed into the strains NBRC 107763 T and TG4. The cells were spread onto TGP agar medium containing kanamycin or chloramphenicol and cultured overnight at 52°C. Transformation efficiency was estimated from average colony forming units per microgram of DNA using three and four biological replicates for NBRC 107763 T and TG4, respectively.

Plasmid Constructions
For gene deletions, knockout plasmids were constructed by assembling PCR-amplified fragments using NEBuilder HiFi DNA Assembly (NEB). All plasmids and primers used in the present study are listed in Table S1 and Table S2, respectively. The codh knockout plasmid, pUC18K-CODHdel, consisted of five fragments: pUC18 (generated from pG2K using the primers 23 and 24), 5′-end of cooCSF ('start') (600 bp; generated from genomic DNA using the primers 31 and 32), kanamycin resistance gene (kan R ) (generated from pG2K using the primers 27 and 28), upstream of cooCSF ('up') (600 bp; generated from genomic DNA using primers 25 and 26), and downstream of cooCSF ('down') (600 bp; generated from genomic DNA using primers 29 and 30). It was a suicide plasmid as Geobacillus replication origin (repB) was not included in the PCR fragment. The other two knockout plasmids, pUC18K-ECHdel and pUC18K-CODH/ECHdel, were constructed in a similar manner with the following modifications: the fragment size of downstream of ech was changed to 1,200 bp to increase the chances of recombination, and pUC18 replication origin (ColE1, 840 bp) was PCR-amplified using the primers 37 and 38, instead of amplifying pUC18 (1,184 bp) to facilitate the assembly.

Gene disruptions
codh, ech, and codh-ech were deleted in NBRC 107763 T based on a markerless gene deletion method following the previously adopted strategy (Bacon et al., 2017, Cripps et al., 2009 (Fig. 1). The strategy was based on a two-step homologous recombination. The first step relied on double homologous recombination where the target gene was replaced by kan R . The second step relied on single homologous recombination which pinched off kan R and other plasmid inserts from the genome. It should be noted that, since double crossovers did not occur in the first step, ech was generated by changing the strategy using two steps. The first step was based on single homologous recombination where the whole plasmid was inserted into the genome. The second step relied on single homologous recombination which pinched off plasmid sequence and the target genes.
Our strategy may be one of the simplest strategies for markerless gene deletion, which only involves serial passaging and replica plating methods. NBRC 107763 T was transformed by 3 g of each knockout plasmid and grown overnight on TGP plates containing kanamycin at 52°C. All the colonies were cultured and serially passaged four times in fresh liquid TGP medium containing kanamycin, to increase the chances of recombination, and subsequently grown overnight on TGP agar medium. Correct insertion of the kan R -containing plasmid cassette at the first crossover site were then checked in the resulting transformants by the length of the PCR products using appropriate primers by colony PCR. Then, appropriate transformants were selected and grown in the liquid TGP medium without kanamycin at 65°C for 6 h until kan R dropped off from the genomes by second crossovers. The resulting transformants were screened for kanamycin resistance by replica plating onto TGP plates with and without kanamycin.
Finally, marker-free gene deletions were confirmed by genomic PCR and whole genome shotgun sequencing.

Genome Sequencing and Analysis
Genomic DNA of the strains WT, codh, ech, and codh-ech was extracted using the DNeasy blood and tissue kit (Qiagen, Hilden, Germany), according to the manufacturer's instructions. A DNA library was prepared using the Nextera XT DNA Library Preparation Kit (Illumina, San Diego, CA), according to the manufacturer's instructions.
Sequencing was performed by the Illumina MiSeq platform, which generated total 14,638,574 paired-end reads (Table S4). Quality trimming and adapter removal were performed using Fastp version 0.20.1 (Chen et al., 2018), with the following settings: sliding window with a window size of 4 bases and with an average Phred quality score >Q30, >95% of bases with Phred quality score of >Q30, and minimum length of 50 bases.
After these procedures, over 2 million reads in each strain remained. The read processing and mapping results are summarized in Table S3. The filtered reads were mapped onto the complete genome sequence (NZ_CP012712) of DSM 2542 T using BWA version 0.7.17 (Li & Durbin, 2010) with default settings. The mapping result was viewed using Integrative Genomics Viewer (IGV) (Robinson et al., 2011) (Fig. S1). Single-nucleotide polymorphisms and small insertions and deletions were identified using breseq version 0.35.0 (Deatherage & Barrick, 2014) with default settings. The mutations detected by breseq were checked manually on the mapping results obtained by BWA and listed if the mutations occurred in more than 5% reads (Table S4-S6). Figure   Fig. S1 The genome mapping analysis of the strain WT, codh, ech, and codh-ech in the codh-ech genomic region. The reads were obtained by whole genome shotgun sequencing and mapped onto the reference sequence of DSM 2542 T (NZ_CP012712) using BWA.

Supplementary
The results were shown on IGV in the order of WT, codh, ech, and codh-ech from the top. The read coverage was ~×250 on average in the region. No reads were mapped onto codh, ech, and codh-ech in the strains codh, ech, and codh-ech, respectively, suggesting the genes were knocked out properly in the disruptants.