RNA sequence elements such as those within 3’UnTranslated Regions (3’UTRs) are involved in regulating protein expression and localization. The function of such elements is also often dependent on their folding into secondary and tertiary structures. However, there is limited knowledge about 3’UTR structural motifs functionality. Proximity labelling coupled with mass spectrometry has generated protein co-localization networks that can help tackle this issue. Proteins that are densely connected in such networks likely localize to the same cellular compartments. If the transcripts of such proteins share a common 3’UTR structural motif, this motif could be directly or indirectly related to the proteins’ regulation or localization. We therefore developed a novel graph theory-based algorithm, named LESuMoN (Local Enrichment of Structural Motifs in biological Networks) that detects RNA structural motifs in 3’UTRs associated with proteins that are significantly clustered in a co-localization network. A Monte Carlo sampling approach is employed to determine the clustering statistical significance of proteins sharing a given 3’UTR structural motif and to estimate a false discovery rate. LESuMoN mined the Human Cell Map protein co-localization network and identified four putative functional 3’UTR structural motifs. These motifs were not detected by a standard network clustering strategy. The motifs are evolutionary conserved and show preferential positioning along 3’UTRs. Notably, three motifs were associated with proteins enriched for cellular compartments and biological processes, suggesting a putative role for the motifs in protein regulation or localization. Overall, we demonstrate that leveraging protein co-localization information can shed light on putative functional elements within 3’UTRs.
Nadeau et al. (2026) studied this question.