Transcriptome-wide mapping of 5-methylcytidine RNA modifications in bacteria, archaea, and yeast reveals m5C within archaeal mRNAs

Abstract

The presence of 5-methylcytidine (m(5)C) in tRNA and rRNA molecules of a wide variety of organisms was first observed more than 40 years ago. However, detection of this modification was limited to specific, abundant, RNA species, due to the usage of low-throughput methods. To obtain a high resolution, systematic, and comprehensive transcriptome-wide overview of m(5)C across the three domains of life, we used bisulfite treatment on total RNA from both gram positive (B. subtilis) and gram negative (E. coli) bacteria, an archaeon (S. solfataricus) and a eukaryote (S. cerevisiae), followed by massively parallel sequencing. We were able to recover most previously documented m(5)C sites on rRNA in the four organisms, and identified several novel sites in yeast and archaeal rRNAs. Our analyses also allowed quantification of methylated m(5)C positions in 64 tRNAs in yeast and archaea, revealing stoichiometric differences between the methylation patterns of these organisms. Molecules of tRNAs in which m(5)C was absent were also discovered. Intriguingly, we detected m(5)C sites within archaeal mRNAs, and identified a consensus motif of AUCGANGU that directs methylation in S. solfataricus. Our results, which were validated using m(5)C-specific RNA immunoprecipitation, provide the first evidence for mRNA modifications in archaea, suggesting that this mode of post-transcriptional regulation extends beyond the eukaryotic domain.

Figure 1. Flow of the bisulfite-treatment/RNA-seq experiment.

(A) Shown is a schematic representation of RNA molecules; colored boxes represent single bases. Red, blue, green and black represent C, U/T, A, and G residues, respectively. m⁵C modified cytosine bases are marked by a black circle. Total RNA is bisulfite-treated, leading to deamination of ‘C’ residues into ‘U’, except for methylated residues. Bisulfite-treated RNA is reverse transcribed into cDNA, which is then sequenced via an Illumina HiSeq system. This yields short sequence reads, representing random fragments of the sequenced RNAs. Resulting reads are mapped to the genome using an algorithm that allows mapping of ‘T’ residues in the sequenced cDNAs onto genomic ‘C's. Residues in which ‘C's are found to be consistently non-modified are declared as m⁵C (red arrow). (B) Flowchart of data analysis and artifact filtering (Methods).

Figure 2. Sites of m⁵C identified in rRNA molecules of the studied organisms.

(A–F) Each panel presents a sequence window around a detected methylated site. X-axis, position along the rRNA molecule; Y-axis, number of supporting reads per position. Positions highlighted in yellow are the methylated ones, with the methylation ratio indicated by the fraction of residues in which a “C” residue was not converted into “U”.

Figure 3. Sanger-based verification of two novel methylated positions in S. solfataricus rRNA identified using RNA-seq.

Bisulfite-converted sequences were amplified using amplicon-specific primers and sequenced. (A) Position C1369 in the S. solfataricus 16S rRNA. (B) Position C2643 in the S. solfataricus 23S rRNA.

Figure 4. Sites of m⁵C identified in tRNA molecules of the studied organisms.

Each line denotes a tRNA sequence, identified (left) by its accession number in tRNAdb . Positions along the consensus are indicated below the sequences, numbers based on E. coli tRNA positions as in . The relevant anticodon is given for each tRNA. Methylated positions are color coded by their methylation ratio, according to the presented color gradient (bottom). For positions 48–50, numbers indicate the proportion of reads supporting methylation out of total reads (e.g. 52/56 indicated 52 non-converted reads out of total of 56 covering reads at the specific position). Top-block, tRNAs of S. solfataricus; bottom block, tRNAs of S. cerevisiae.

Figure 5. Methylated positions in mRNAs of S. solfataricus.

(A) Representative position in S. solfataricus oxidoreductase gene (locus: SSO3054). Colors and axes are as in Fig. 2. (B) Sanger-based verification of the position in panel A, on bisulfite-treated reverse transcribed RNA (C) Sanger-based sequencing of the same position in panel B, on bisulfite-treated DNA, as a negative control (D) Methylated positions in mRNAs of S. solfataricus show a consensus motif. Shown are the sequences flanking the methylated positions (red) identified in mRNAs of S. solfataricus (Table 3). Recurring sequence motif is underlined. (E) Visual representation of consensus sequence motif, prepared using weblogo .

Figure 6. RNA immunoprecipitation with modification-specific antibodies.

Shown is the coverage of Illumina-sequenced cDNA following RNA fragmentation, antibody pulldown, reverse transcription and sequencing. Black line, pulldown performed with an anti-5-methylcitosine (hm⁵C) antibody; green line, pulldown performed with an anti-5-hydroxy-methylcitosine antibody; red line, input RNA (no antibody applied). X-axis, position along the genome; Y-axis (right), read coverage of the sequenced anti-m⁵C library; Y-axis (left), fold enrichment of peaks related to median coverage along the gene. The coverage of the anti-hm⁵C and input libraries was normalized using the median of the anti-m⁵C library as a reference point. (A) The 23S gene of S. solfataricus. Peaks corresponding to positions 2121 and 2643 in the gene (875,473 and 875,995 relative to the S. solfataricus genome, respectively) are marked. Another peak, which did not come up in our bisulfite-based analysis, is observed around position 2760. (B) The 16S gene of S. solfataricus. A single peak corresponding to position 1369 in the gene (position 873,040 relative to the genome) is marked. (C–E) Antibody pulldown of m⁵C modifications in protein-coding genes from Table 3.