Back

Characterizing 3D RNA structural features from DMS reactivity

Deenalattha, S.; Jurich, C. P.; Lange, B.; Armstrong, D.; Nein, K.; Yesselman, J. D.

2025-02-25 biochemistry
10.1101/2024.11.21.624766 bioRxiv
Show abstract

Dimethyl sulfate (DMS) chemical mapping probes RNA structure, where low reactivity is generally interpreted as Watson-Crick (WC) base pairs and high reactivity as unpaired nucleotides. Studies examining DMS reactivity of RNAs with known 3D structures have identified nucleotides that deviate from this interpretation with distinct solvent accessibility and hydrogen bonding patterns. Understanding the frequency of these outliers and their recurring structural 3D features remains incomplete. To address this knowledge gap, we systematically analyzed DMS reactivity patterns across a library of 7,500 RNA constructs containing two-way junctions with known 3D structures. We observe DMS reactivity exists on a continuum over four orders of magnitude with approximately 10% overlap in reactivity between WC and non-WC nucleotides. We find that non-WC bases with WC-like DMS protection exhibit increased hydrogen bonding and decreased solvent accessibility, whereas WC pairs exhibiting greater DMS reactivity tend to flank junctions, correlating with weaker base stacking and greater junction dynamics. Furthermore, we discover that DMS reactivity values in non-canonical pairs correlate with atomic distances and base pair geometry, enabling discrimination between different 3D conformations. These DMS reactivity patterns indicate that DMS reactivity provides atomic-scale information about RNA 3D conformations, which can be used to model RNA structures and dynamics.

Published in Nucleic Acids Research (predicted rank #1) · training set

Matching journals

The top 4 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.