Proxi-RIMS-seq2 applied to native microbiomes uncovers hundreds of known and novel m5C methyltransferase specificities

Abstract

Methylation patterns in bacteria can be used to study restriction-modification or other defense systems with novel properties. While m4C and m6A methylation are well characterized mainly through PacBio sequencing, the landscape of m5C methylation is under-characterized. To bridge this gap, we performed RIMS-seq2 (rapid identification of methyltransferase specificity sequencing) on microbiomes composed of resolved assemblies of distinct genomes through proximity ligation. This high-throughput approach enables the identification of m5C methylated motifs and links them to cognate methyltransferases directly on native microbiomes without the need to isolate bacterial strains. Methylation patterns can also be identified on bacteriophage DNA and compared with host DNA, strengthening evidence for phage-host interactions. Applied to three different microbiomes, the method unveiled over 1900 motifs that were deposited in REBASE. The motifs include a novel eight-base recognition site (CATm5CGATG) that was experimentally validated by characterizing its cognate methyltransferase. Our findings suggest that microbiomes harbor arrays of untapped m5C methyltransferase specificities, providing insights into bacterial biology and biotechnological applications.

Graphical Abstract

Figure 1

(A) Overview of the Proxi-RIMS-seq2 analytic steps: (i) Imbalances in MAGs indicate methylation. (ii) Imbalance at known motifs. (iii) Imbalance is used to de novo identify methylated motifs. (iv) Linking genotype (MAGs) with phenotype (methylation) associates methyltransferases with their predicted recognition motifs. (v) Methyltransferase activities can be cloned and expressed in vivo in strains lacking ^m5C methylation for validation. (B) Relationship between MAG length (in base pairs, x-axis) and read coverage (log₁₀ scale, y-axis) within the genome-resolved vermicompost microbiome. Each data point represents an MAG, distinguished by the presence (green) or absence (purple) of predicted methylated motifs. Notably, MAG 17, belonging to the genus Patulibacter, exhibits the highest read coverage mapped to its consensus genome. (C) Bar plot representing the absolute C-to-T transition rate in R1 (blue) and R2 (red) at all 16 NCN contexts (with N = A,T,C, or G) in MAG 17. (D) Differential C-to-T transition rate between R1 and R2 (imbalance value) at all 16 NCN contexts in MAG 17. (E) PWM found to be most significantly associated with C-to-T imbalance in MAG 17 (* indicates the methylated cytosine).

Figure 2.

Imbalance profiles for selected clusters of MAGs (full dataset in Supplementary Fig. S2) in (A) vermicompost, (B) dental, and (C) fecal microbiome. At an imbalance of 1%, all cytosines at the underlined positions are predicted to be methylated. The x-axis corresponds to the annotated MAGs, y-axis corresponds to known recognition sites of ^m5C methyltransferases (as cataloged in REBASE), and z-axis (gray to red) indicates imbalance values, ranging from 0 to 1%. MAGs are color-coded according to their taxonomic order. Both MAGs and motifs are clustered based on their imbalance profiles. Rows are clustered using the Manhattan distance, and columns are clustered using the Minkowski distance, as implemented in Pheatmap.

Figure 3.

Validation of a new methyltransferase specificity using Tet-assisted PacBio sequencing. (A) IPD ratios at a specific E.colilocus containing both CATCGATG and GATC motifs. (B) Average IPD ratio for all possible 8 mers for A modifications (^m6A, upper panel) and C modifications (^m4C or ^m5C, lower panel). Y-axis corresponds to the first three nucleotides (oriented 3′–5′), x-axis corresponds to the last four nucleotides (orientated 5′–3′), and z-axis (color) corresponds to the average IPD ratio. (C) De novo identification of motifs (output of DiNAMO and PWM).