Identification of Methylomonas sp. DH-1 methylation site
To identify the RM system, the genome of Methylomonas sp. DH-1 was analyzed by whole-genome bisulfite sequencing (WGBS). Interestingly, only the TGGCCA motif was identified (Fig. 1a). In the REBASE database [30, 31], Methylomonas sp. DH-1 contains twelve RM systems in its genome and two in its native plasmid (Fig. 1b). According to REBASE, it was predicted that the cytosine methyltransferase AYM39_01025 would recognize the GGCC sequence for methylation, which is similar to the identified methylation site TGGCCA, in which the fourth nucleotide (C) was methylated in our results. Therefore, this cytosine methyltransferase was selected as a potential methylase for TGGCCA.
Digestion protection assay
To investigate whether the selected cytosine methyltransferase (AYM39_01025) was able to recognize the identified sequence (TGGCCA) instead of the predicted sequence (GGCC), we conducted a DNA protection assay against digestion, using several restriction enzymes. When the cytosine methyltransferase protein was over-expressed in E. coli BL21 (DE3), the protein formed an inclusion body even though it was co-expressed with chaperones (pGro7 and pTf16). Thus, we could not perform the in vitro assay requiring purified methyltransferase. Instead, we introduced a plasmid harboring the methyltransferase gene and TGGCCA sites into the E. coli JM110 strain (dam and dcm methylase genes were deleted). Since the cytosine methyltransferase was under the control of the T5 promoter with a lac operator, we could obtain a non-methylated or methylated plasmid by IPTG. For further analysis, the plasmid was extracted from E. coli JM110 strain.
According to the REBASE annotations, the cytosine methyltransferase of Methylomonas sp. DH-1 was predicted to methylate the GGCC sequence, while the only methylation site identified in Methylomonas sp. DH-1 by WGBS was TGGCCA. To confirm that the cytosine methyltransferase recognized TGGCCA instead of GGCC, several restriction enzymes that contain GGCC in their restriction sites were used: MscI (TGGCCA), ApaI (GGGCCC), and NotI (GCGGCCGC). We also used EcoRl (GAATTC) and Xbal (TCTAGA) restriction enzymes as negative controls. The plasmid harboring the cytosine methyltransferase gene contained all of the above-mentioned restriction sites, as well. If the methylation site was GGCC, the restriction enzymes (MscI, ApaI, and NotI) would not be able to cleave the plasmid DNA. As shown in Fig. 2a, most restriction enzymes were able to cleave both the non-methylated and methylated plasmids, but MscI failed to cleave the methylated plasmid. This result indicated that the cytosine methyltransferase recognized TGGCCA and not GGCC.
To identify the cytosine nucleotide methylated by the cytosine methyltransferase, the methylated plasmid was analyzed by bisulfite sequencing. In bisulfite sequencing, only non-methylated cytosines are converted to uracil, and during PCR, the uracil is converted to T. Methylated cytosines are not changed by bisulfite sequencing. As shown in Fig. 2b, TGGCCA in the non-methylated plasmid was converted to TGGTTA, indicating that the cytosines were non-methylated, as expected. In the methylated plasmid, only the fifth cytosine in TGGCCA was converted to T, indicating that the fourth cytosine was methylated by the cytosine methyltransferase.
Methylation of plasmid DNA increased transformation efficiency
The plasmid harboring the psy (phytoene synthase) gene was constructed (Fig. 3a) and co-transformed into E. coli JM110 with the plasmid harboring the cytosine methyltransferase gene psy. The psy gene is involved in the biosynthetic pathway that produces carotenoids. For the methylation of the plasmid containing psy, the media were supplemented with 0.1 mM IPTG to induce the expression of the cytosine methyltransferase. Since E. coli contains two plasmids (psy plasmid + cytosine methyltransferase plasmid), the plasmids were separated by gel electrophoresis, and the psy plasmid was extracted from the gel (Fig. 3b). The non-methylated plasmid was also extracted from the cell without IPTG to create a control sample in which the expression of the cytosine methyltransferase was not induced.
The extracted plasmids were transformed into Methylomonas sp. DH-1 by electroporation. Since there are no artificial plasmids that exist separate from the genome of Methylomonas sp. DH-1, we measured the genome integration efficiencies of the psy gene involved in the carotenoid biosynthetic pathway (Fig. 4a) to deduce the transformation efficiency. The transformation efficiency of the methylated DNA of the psy plasmid was 2.5 × 103 CFU/μg of DNA. The efficiency was increased by 124 % compared with that of the non-methylated plasmid DNA (Fig. 4b). Despite of methylation, the efficiency increase was not dramatic. We think that the introduced plasmids were easily integrated into the genome of Methylomonas sp. DH-1 by recombinases and thus the protection of plasmids by methylation was not essential in genome integration experiments. However, to date there are no artificial plasmids available for Methylomonas sp. DH-1, which is independent from its genome. For the development of artificial plasmids, the protection of plasmids by methylation is essential and the identification of a cytosine methylation system is the first step for the development. In this regard, the newly identified methylation system would facilitate the development of artificial plasmids as well as other biotechnological techniques based on plasmids.
For further evaluation of methylation effect on transformation, we removed the three methylation sites in the psy plasmid by mutating nucleotides: one in an intergenic region and two in the coding region of psy. The former site was converted from TGGCCA to TGTCCA, and the latter two were converted from GTG GCC AAT to GTA GCG AAT and from CTG GCC AAA to CTA GCG AAA based on codon degeneracy in order not to mutate amino acids (Fig S1a). As shown in Fig. S1b, the deletion of the methylation sites increased transformation efficiency similar to that of the methylated psy plasmid. This indicates that methylation of plasmid DNA by the identified cytosine methyltransferase protects plasmid DNAs from the RM system of Methylomonas sp. DH-1 and increases the transformation efficiency, which may facilitate the genetic manipulation of Methylomonas sp. DH-1.
To investigate the effects of plasmid size and methylation on transformation efficiency, we constructed three different plasmids with a different length (5 – 7 kb). As shown in Fig. S2, methylation increased transformation efficiency while plasmid length did not show any significant effect on transformation efficiency. This indicates that transformation efficiency is dependent on methylation, not plasmid size. Furthermore, we measured the growths of non-transformed wild type cells and transformed cells with methylated plasmids (the psy plasmid or anti-psy sRNA plasmid). As shown in Fig. S3, their growth rates are very similar, showing that there could be no significant changes in cellular physiology.
Methylomonas sp. DH-1 carries a complete MEP pathway for carotenoid production . The selected gene, psy, is involved in the carotenoid biosynthetic pathway. The gene was designed to be expressed by the promoter of the mxaF gene  (Fig. 4a). When the plasmid containing the psy gene was introduced into Methylomonas sp. DH-1, the psy gene was integrated into the genome by homologous recombination. The additional copy of the psy gene increased carotenoid biosynthesis by 26 % (Fig. 4c).
Synthetic sRNAs have been utilized to increase the production of desired substances by regulating gene expression . Synthetic sRNAs were designed to bind to the nucleotides in the translation initiation regions of mRNAs, and thereby they repressed the translation of mRNAs by preventing the binding of ribosomes in assistance of Hfq protein . In this study, we constructed a plasmid containing an anti-psy synthetic sRNA gene to investigate the methylation effect on transformation efficiency and also to investigate the knock-down effect of the psy gene on carotenoid production (Fig.5a).
Synthetic sRNAs are composed of two elements: a target binding region and a scaffold. In previous studies, various scaffolds originated from E. coli were used including MicC and SgrS [33, 35]. Of many inherent sRNA genes, RyhB was known to operate in the absence of Hfq protein  and thus was expected to work in various bacterial species. We designed an anti-psy synthetic sRNA using the scaffold of RyhB sRNA. The anti-psy synthetic sRNA gene was under the control of tac promoter.
When the plasmid harboring the anti-psy synthetic sRNA gene was transformed after methylation, its transformation efficiency was enhanced by 70 % compared with the non-methylated plasmid (Fig. 5b). In addition, when the synthetic sRNA gene was integrated into the genome of Methylomonas sp. DH-1, the synthetic sRNA decreased carotenoid production by 40 % compared with that of wild type Methylomonas sp. DH-1 (Fig.5c). These results indicate that methylation of plasmids can improve transformation efficiency as well as that synthetic sRNAs based on a RyhB scaffold can be used to regulate the expression of genes in Methylomonas sp. DH-1.