Fig. 3

Local secondary structure around the consensus DRAC motif impedes methylation. a Boxplot displaying the difference in the minimum free energies values (MFE) for predicted structures between S. cer and S. par in the Invariable m6A sites (both alleles are methylated) and DM-cis sites. The prediction is made for 61-nt long window around the methylated adenosine. Sites exclusive to S. par are more structured in S. cer, whereas ones exclusive to S.cer are more structured in S. par. Sites with no variation in methylation don’t show differences in predicted structure between the alleles. To eliminate the effect of sequence on methylation and to increase the prediction tool accuracy, only sites without motif-damaging mutations and at least one allele with a predicted solid structure (MFE < -8 kcal/mol) are displayed. The whiskers correspond to the value no further than 1.5 × the interquartile range (n = 360). b mRNA secondary structure affects methylation independently of relative position within the gene. The percentages of methylated DRAC motifs are displayed along the transcriptome, binned along two dimensions: relative localization and propensity for structure. Relative localization binning was conducted on the basis of five bins: 5’ UTR, within the first 10% of CDS (CDS Adjacent to 5’ UTR), CDS, within the last 10% of CDS (CDS Adjacent to 3’ UTR), and 3’ UTR. The spectrum of MFEs was divided into four quartiles, which were used for binning sites based on structural predictions. c Validation of causal role on methylation played by secondary structure, based on CRISPR-based perturbations conducted in the vicinity of an m6A site in the bub3 gene. For all WT and mutant sequences, m6A-seq2 IP coverage plots are shown on the top, with green bars indicating the m6A locus and its homologous locus in the counterpart allele. The bottom set of panels depicts RNAfold structural predictions around the DRAC motif. The DRAC motif is embedded within a strong stem in the non-methylated allele (S. par WT, left panel) but not in the methylated one (S.cer WT, center-left panel). Mutating three nucleotides (depicted in red), all residing within a distance > 12 nt from the DRAC motif in S.par, led to a predicted open structure (center-right panel), and led to methylation of the DRAC motif in the S.par allele. Re-introducing a stem structure via additional compensatory mutations (depicted in red) led to elimination of the m6A signal (right panel). The methylated adenosine is depicted in green across all strains. d Quantification of m6A site score for the alleles depicted in (c). e–f similar analysis (as in c-d) for a site detected in the ADP1 gene in S.cer. g m6A IDENT-score for the m6A sites at BUB3 and ADP1. The read coverage in both IP and Input samples only considers sequencing reads that come from identical RNA fragments between the WT strain and the two mutated strains in (c)