Development of DRAM system for m5C detection.

(A) Schematic diagram of the DRAM assay. DRAM, DRAMmut and Deaminase system were transfected into HEK293T cells separately. After DRAM transfection, the deaminase was directed by m5C reader to the vicinity of the m5C site and induce C-to-U/A-to-G mutations, whereas transfection of the DRAMmut or Deaminase system failed to effectively induce similar mutations due to the absence of the m5C-recognition-binding domain.

(B) The overall design of DRAM, DRAMmut and Deaminase system.

DRAM detection system was assayed in an m5C -dependent form.

(A, B) Two m5C sites from RPSA (A) and AP5Z1 (B) mRNA detected by deep sequencing of bisulfite sequencing PCR in HEK293T cells. The m5C sites are highlighted by red color. The m[C fraction of RPSA and AP5Z1 were 75.5% and 27.25% (The number of reads is greater than 1000).

(C, D) Sanger sequencing following RT-PCR verified two m5C sites from RPSA (C) and AP5Z1(D) mRNAs in DRAM-transfected HEK293T cells, respectively. HEK293T cells only expressing DRAMmut or Deaminase were served as negative controls. The left panel illustrates the location of DRAM induced mutation sites, which is highlighted in red asterisk. The right panel shows the corresponding quantification of sanger sequencing.

(E, F) The knockout efficiency of NSUN2 (E) and NSUN6 (F) in HEK293T cell lines verified by Western blotting. The protein level of α-Tubulin and GAPDH were served as loading controls, separately.

(G, H) DRAM induced mutations close to m5C sites in AP5Z1 (G) and RPSA (H) mRNAs after NSUN2 and NSUN6 knockout in HEK293T cells. The left panel illustrates the location of DRAM induced mutation sites, which is highlighted in red asterisk. The right panel shows the corresponding quantification of sanger sequencing.

DRAM enables transcriptome-wide analysis of m5C methylation.

(A) Schematic of the DRAM-seq method.

(B, C) Integrative genomics viewer (IGV) browser traces of DRAM-seq data expressing the indicated constructs in RPSA (B, left panel), TARBP2 (B, right panel), AP5Z1(C, left panel), and TRAF7 (C, left panel) mRNAs. C-to-U or A-to-G mutations found in at least 10% of reads are indicated by coloring. The previously published RNA BS-seq datasets from two individual studies were displayed as panel “Yang et al.” and “Zhang et al.”. (n(DRAM)=3 independent samples, n(Deaminase)=2 independent samples, and n(DRAMmut)=1 independent sample.)

(D, E) Integrative genomics viewer (IGV) browser traces of DRAM-seq data in wildtype and methyltranferases knockout cells in AP5Z1 (D) and RPSA (E) mRNAs. C-to-U or A-to-G mutations were found in at least 10% of reads are indicated by coloring. The previously published RNA BS-seq datasets from two individual studies were displayed as panel “Yang et al.” and “Zhang et al.”. n=3 independent samples.

(F) Screening process for DRAM-seq assays and principles for screening high-confidence genes.

(G) The pie chart shows the distribution of editing sites in different transcript region in cells expressing DRAM (n=3 independent samples).

(H) The density map showing the distribution of editing events across the mRNA transcripts detected by DRAM-seq.

(I) The frequency plot shows the distribution of the distances of edit events in DRAM-seq relative to the m5C sites from the published BS-seq datasets. The position of each m5C site of BS-seq is determined as 0, and the relative distance of each site to the nearest edit event in DRAM-seq is calculated and plotted. The plots are presented separately based on the cutoff of upstream and downstream 3000bp (above) and 80bp (below) windows.

(J, K) Motif analysis discovered within the ±20nt region around the C-to-U or A-to-G editing site in cells expressing DRAM-CBE (J), APOBEC1(J), DRAM-ABE (K) and TadA-8e (K).

Stable and comprehensive cellular identification of m5C loci by DRAM-seq.

(A) Comparison of the overall distribution of genes with m5C modifications detected by DRAM-seq, Yang et al. and Zhang et al. on chromosomes. The mutation sites detected by DRAM-seq on each gene are categorized into dual-colored short lines, with positive strand mutations shown in orange and negative strand mutations in dark green. The line graph and kernel density plot in the inner ring represent the locations and distributions of overlapping genes detected by DRAM-seq (red), Yang et al. (blue) and Zhang et al. (light green).

(B) Venn diagram showing the overlap between DRAM-seq and Yang et al.’s edited genes.

(D) Venn diagram showing the overlap between DRAM-seq and Zhang et al.’s edited genes.

(C, E) Integrative genomics viewer (IGV) browser traces of DRAM-seq data expressing the indicated constructs in the ATG16L1(B), ARHGEF25(B), FANCD2(D), and RPL15(D) mRNAs. C-to-U/A-to-G mutations found in at least 10% of reads are indicated by coloring, and the m5C site found by BS-seq is also labelled.

(F) Genes with DRAM-seq editing events were analyzed for KEGG bioprocess enrichment.

(G) GO biological processes enrichment analysis of genes with DRAM-seq editing events. Statistical analyses were performed using the DAVID tool.

Low-input m5C detection and transfection efficiency of DRAM system.

(A, B) DRAM analysis of RPSA (A) and AP5Z1 (B) mRNAs with 250 ng, 50 ng, and 10 ng of input RNA. Representative Sanger sequencing plots are shown on the left panel, with mutation sites marked with asterisks. The mutation rates are quantified on the right panel.

(C) Flowchart illustrating Cell viability analysis by CCK8 reagent after DRAM transfection in HEK293T cells.

(D) Quantitative comparison of the relative proliferative capacity of DRAM-expressing and untransfected cells.

(E, F) The expression levels of DRAM-CBE (E) and DRAM-ABE (F) systems different plasmid transfection concentrations were verified by Western blotting.

(G, H) Editing of RPSA (G) and AP5Z1 (H) mRNA at varying concentrations of DRAM protein expression. The left panels indicate Sanger sequencing results following RT-PCR, while the corresponding quantifications of DRAM-induced mutations are shown in the right panels.