Cells of Methanococcus aeolicus PL15/H^p were a gift from Prof. Dina Grohmann (Universität Regensburg), grown as previously described (Kendall et al., 2006). All laboratory Escherichia coli strains used in this study are listed in Supplementary Table 1. Cultures were grown in Luria-Bertani (LB) broth or agar on appropriate antibiotics. All plasmids used in this study are listed in Supplementary Table 2. Restriction enzymes, except MaeIII (Roche Diagnostic, GmbH, Mannheim, Germany), DNA substrates, DNA and protein markers were from New England Biolabs (NEB, MA). Q5 “Hot Start” DNA polymerase (M0543, NEB, MA) was used for PCR amplification of Methyltransferase and Restriction endonuclease genes for cloning. All PCR primers were synthesized by IDT, IA and are listed in Supplementary Table 3. The cloning of PCR amplified genes in appropriate vectors were performed using NEBuilder HiFi DNA assembly Master mix (E2621, NEB, MA). The plasmid DNA was purified using a Monarch Plasmid Miniprep kit (T1010, NEB, MA).

Genomic DNA from Methanococcus aeolicus PL15/H^p was purified using a Monarch Genome Purification kit (T3010, NEB, Ipswich, MA, USA) and the DNA sample was sheared to an average size of ∼ 10 kb using the G-tube protocol (Covaris, Woburn, MA, USA). DNA libraries were prepared using a SMRTbell express template prep kit 2.0 (100–938–900, Pacific Biosciences, Menlo Park, CA, USA) and ligated with hairpin barcoded adapter lbc1-lbc1. Incompletely formed SMRTbell templates were removed by digestion with a combination of exonuclease III and exonuclease VII (NEB, Ipswich, MA, USA). The qualification and quantification of the SMRTbell libraries were made on a Qubit fluorimeter (Invitrogen, OR) and a 2,100 Bioanalyzer (Agilent Technologies, Santa Clara, CA, USA). SMRT sequencing was performed using an SQ1 (Pacific Biosciences, Menlo Park, CA, USA) based on the multiplex protocol for 10 kb SMRTbell library inserts. Sequencing reads were collected and de novo assembled using the Microbial Assembly version 10.1.0.1119588 program with default quality and read length parameters. In addition to genome assembly (Chin et al., 2013), the SMRT Analysis pipeline from Pacific Biosciences¹ enables the determination of the epigenetic status of sequenced DNA by identifying the m6A and m4C modified motifs (Flusberg et al., 2010; Clark et al., 2012; Korlach and Turner, 2012).

Following an approach nicknamed the “Hungarian trick” (Szomolányi et al., 1980), 5 μg of M. aeolicus PL15/H^p gDNA was digested with BamHI, BglII and partially digested with Sau3AI restriction enzymes. DNA fragments were cloned into the compatible dephosphorylated BamHI sites of pUC19, pBR322 and pACYC184 vectors yielding 9 shot gun libraries. The foreign restriction and modification genes are usually very toxic for E. coli, therefore in order to clone both we need high level of methylation. These three vectors have different copy numbers pUC19 (high), pBR322 (medium) and pACYC184 (low). Moreover, they utilized different promoters P_lac and P_tet, since native Methanococcus promoters may or may not work in E. coli. Each purified plasmid library (1 μg) was challenged with 10 units of MaeIII restriction endonuclease and the digestion mixtures were transformed into E. coli strain ER2683. Plasmids were recovered from surviving transformants and analyzed by Sanger sequencing and restriction digest.

Homology-based searches of the M. aeolicus genome for putative methyltransferase genes were performed using the Seqware program (Roberts et al., 2022). Additional searches for MTase genes were performed using HMMer (HMMER 3.3.2)² to annotate each predicted protein coding sequence feature (CDS) of the Methanococcus aeolicus PL15/H^p genome. Genes matching MTase sequence profiles were further examined by structure prediction using the ColabFold implementation of AlphaFold2 (Jumper et al., 2021; Mirdita et al., 2022), followed by structure similarity search using predicted MTase models as query inputs to DALI (Holm, 2022).

Gene loci for M.MaeV and M1-M2.MaeIV were PCR amplified with P1-P2 primers and P3-P4 primers, respectively, and assembled into low copy pACYCΔtet vectors under control of a constitutive P_tet promoter and expressed in E. coli strain ER2683 (Supplementary Table 1). The correct clones were verified by Sanger sequencing of the inserts using an ABI DNA sequencer with pTetF and pTetR sequencing primers and transformed into the methylation deficient E. coli strain ER2796 (Supplementary Table 1).

Purified MaeIII endonuclease was subjected to 10–20% Tris-Glycine SDS-PAGE and stained with SimplyBlue SafeStain (Invitrogen, OR). The major band at approximately 35 kDa was excised for in-gel digestion (Figure 4A). The gel band was sliced into 1 mm³ pieces and the pieces were divided into two aliquots for multiple enzymatic digestions (Wiśniewski et al., 2019). The gel pieces were destained, and the protein was reduced with DTT, alkylated with iodoacetamide, and then digested with either Trypsin-ultra, Mass Spectrometry Grade (P8101; NEB, Ipswich, MA, USA) or subtilisin (Sigma-Aldrich, St. Louis, MO, USA) at a protease concentration of 2 ng/μL. Digestions were conducted for 1 h (50°C for trypsin, 37°C for subtilisin) and were quenched with trifluoroacetic acid at a final concentration of 0.5% for downstream analysis.

The resulting peptides were loaded via Proxeon Easy-nLC 1,200 (ThermoFisher, Waltham, MA, USA) onto a reversed phase analytical column 25 cm × 100 μm ID, Reprosil 3 μm C18 (New Objective, Littleton, MA, USA) and eluted at a flow rate of 0.3 μL/min over a 35 min gradient of organic solvent (mobile phase A: 0.1% formic acid in water; mobile phase B: 90% acetonitrile/0.1% formic acid) as follows: 1-min equilibration at 10% B, 10–40% B in 25 min, 40–50% B in 4 min, and 50–78% B in 5 min. Peptides were electro-sprayed into a LTQ Orbitrap XL with a Nanospray Flex Ion Source (Thermo Fisher, Waltham, MA, USA) and fragmentation spectra were generated by collision-induced dissociation (CID). Full scans between 400–1,600 m/z were acquired with 30 k resolution. MS/MS spectra were generated in a data-dependent manner, with a normalized collision energy of 35.0 and dynamic exclusion of 30 s. The five most intense ions (including + 1 charge states for subtilisin-generated peptides) were selected for fragmentation.

Acquired MS/MS spectra were analyzed using PEAKS Studio Xpro (Bioinformatics Solutions, Inc., Waterloo, ON, USA). The data were searched against Methanococcus aeolicus PL15/H^p translated ORFs from the genome sequence, allowing for a fixed carbamidomethyl modification on cysteines, variable oxidation of methionine, and two missed cleavages (tryptic peptides). The precursor and fragment mass tolerances were set to 20 ppm and 0.6 Da, respectively. Identified proteins were filtered by a 1% false discovery rate and required at least two unique peptides per protein.

The recombinant MaeIII restriction enzyme was partially purified on Ni-NTA magnetic beads (S1423L, NEB, Ipswich, MA, USA) according to the manufacturer’s protocol. DNA digestion patterns obtained by titration of restriction endonuclease activity from 250 mM imidazole Ni-NTA column elution fraction from E. coli clone 3081[pM.MaeIII + pET21bMaeIII] expressing MaeIII enzyme compared to MaeIII restriction enzyme from Roche on the same DNA substrate on λ DNA substrate. 1 μg of λ DNA incubated with Ni-NTA imidazole elution fraction of 6×His:MaeIII serial 3× dilution (from 9 to 0.1 μl) and serial 3× dilution of commercial MaeIII enzyme (from 9 to 0.1 units) for 1 h at 55°C in 1× MaeIII commercial buffer (Roche Diagnostic, GmbH, Mannheim, Germany). 2× Buffer contains 40 mM Tris–HCl, 550 mM NaCl, 12 mM MgCl₂, 14 mM 2-mercaptoethanol, pH 8.2 (+55°C). The reaction mixtures were separated on 0.8% agarose gel in TBE buffer and visualized by staining with ethidium bromide.

Monarch purified gDNA (2 μg) from a Methanococcus aeolicus PL15/H^p culture was used to make one barcoded (lbc1-lbc1) 10 kb SMRT bell library sequenced on an SQ1 instrument for 20 h. 418,400 long continuous sequencing subreads with 2,583 bp mean subread lengths were converted to 12,577 HiFi subreads with 5,328 bp mean subread length, yielding 43.7 Mb of high-quality sequencing data. The polished assembly generated one closed circular genomic element of 1,677,738 bp with 26-fold genome coverage and 29.93% G + C content for the chromosome (Figure 1A). The assembled sequence was annotated using the National Center for Biotechnology Information (NCBI) Prokaryotic Genome Annotation Pipeline (PGAP) (Tatusova et al., 2016; Haft et al., 2018) which identified 1,615 protein coding features and 38 tRNA genes. The annotated genome sequence of M. aeolicus PL15/H^p is available from NCBI via the Bioproject PRJNA622823; BioSample: SAMN30825808, SRA: SRS15238210, and genome sequence: NZ_CP104873.

One advantage of the SMRT sequencing platform is its ability to directly detect the epigenetic state of sequenced DNA. As the DNA sequencing polymerase traverses methylated nucleobases in the sequencing template, changes in the amount of time between the nucleotide incorporation events (also known as the Interpulse Duration, or IPD) are recorded and analyzed by the SMRT-Link software yielding m6A and m4C modification motifs (Flusberg et al., 2010; Clark et al., 2012). The complete motif and modification analysis of assembled Methanococcus aeolicus PL15/H^p using these kinetic signatures detected three m6A and one m4C modified motifs (Figure 1B). Two of the motifs with m4C CTAG and m6A CTNAG were recognized as signatures of the previously discovered MaeI and MaeIII RM systems (Schmid et al., 1984). The m5C motif of MaeII system ACGT cannot be directly detected by SMRT sequencing due to the weak IPD ratio signal generated by DNA polymerase on m5C modified base. Two additional m6A motifs with single DNA strand modification GCACY and CGAG were also detected by SMRT motif and modification analysis.

The primary scanning of the M. aeolicus PL15/H^p assembled genome using Seqware identified 4 putative methyltransferase genes. Two of them for M.MaeI and M.MaeII where assigned bioinformatically based on homology search against methyltransferase genes in REBASE with known specificities (Figures 2A, B). The two other putative methyltransferase genes were assigned experimentally. Gene loci for M.MaeV and M1-M2.MaeIV were cloned into low copy pACYCΔtet vectors under the control of a constitutive P_tet promoter and their expression induced in E. coli strain ER2683 (Supplementary Table 1). SMRT sequencing and methylation analysis of E. coli gDNA recovered from pM1-2.MaeIV and pM.MaeV expressing clones revealed that M.MaeV recognized the m6A modified motif CGAG. M1.MaeIV and M2.MaeIV belong to a family of so called split methylase/RM genes similar to DrdVI (Fomenkov et al., 2019) and recognize the m6A GCACY modified motif (Figures 3A, B). However, at that point, we were not able to assign the m6A GTNAC motif, corresponding to M.MaeIII, as there was no candidate gene identified by the PGAP annotation.

To find the missing gene for M.MaeIII we decided to apply an old methylation selection method known as the “Hungarian trick” (Szomolányi et al., 1980). Basically, this method is based on selection of clone(s) expressing the methyltransferase by treatment of the library with the cognate restriction endonuclease to destroy unmodified plasmid DNAs, leaving only plasmids that contain an active methyltransferase gene. In addition, we made a fosmid library in the pCC1Fos vector (Epicenter, WI). The plasmid and fosmid DNAs from survival clones were tested using a methylation protection assay for MaeIII resistance to digestion. Unfortunately, after selection and screening we were not able to find any MaeIII resistant clones (data not shown). Therefore, we decided to use a different approach to find the missing genes.

We used an MS/MS method to identify the protein sequence in functionally purified MaeIII from Roche (Figure 4A). This yielded a top target hit corresponding to locus_tag N6C89_05690 from M. aeolicus PL15/H^p (Figure 4B). In addition, the M. aeolicus PL15/H^p assembled genome was scanned for additional DNA methyltransferase genes using an automated implementation of HMMer, which produced another potential hit corresponding to locus_tag N6C89_5685. The best scoring match to an AlphaFold2 predicted model of N6C89_5685 was 7lt5, the PDB structure of CamA, an adenine methyltransferase of Clostridioides difficile (Figure 4C; Zhou et al., 2021). Thus, the close proximity of the two identified target genes strongly suggested that they corresponded to the MaeIII RM system.

The genes corresponding to locus_tags 5,685 and 5,690 (Figure 5A) were cloned into the low copy pACYCΔtet vector under control of a constitutive P_tet promoter for expression in E. coli T7 expression strain, ER3081. The plasmid and gDNA from this strain were purified and challenged with 10 units of the MaeIII restriction enzyme to test for methylation protection (Figure 5C). In addition, the DNA obtained from E. coli expressing this putative M.MaeIII methyltransferase was sequenced using the SMRT analysis platform, identifying the sequence GTNAC as modified (Figure 5B), consistent with the M.MaeIII modification motif previously reported (Schmid et al., 1984). These results clearly indicated that locus_tag N6C89_5685 did indeed code for M.MaeIII. Therefore the strain was designated as ER3081[pM.MaeIII].

A PCR amplicon spanning the presumptive endonuclease gene, locus_tag 5690, was cloned in-frame with the N-terminal 6× His tag and linker sequence of the pET21b expression vector under the control of the pT7 promoter and transformed into an E. coli strain previously transformed with a compatible M.MaeIII expressing plasmid. The resulting two-plasmid strain encoding M.MaeIII and the putative MaeIII endonuclease gene was induced and the his-tagged putative endonuclease purified using Ni-NTA beads as described. MaeIII restriction endonuclease activity was subsequently detected in a 250 mM imidazole elution fraction (Figure 5D).