RNA structure generates natural cooperativity between single-stranded RNA binding proteins targeting 5' and 3'UTRs

TL;DR

This paper introduces a quantitative measure of RNA structure-mediated cooperativity between binding proteins at different mRNA regions, revealing its prevalence and dependence on UTR sequences in human and C. elegans mRNAs.

Contribution

It develops a numerical measure for RNA structure-mediated cooperativity and demonstrates its widespread occurrence, including between 5' and 3' UTRs, based on sequence analysis.

Findings

Cooperativity is a common feature in human and C. elegans mRNAs.

End-to-end cooperativity depends on UTR sequences, not coding regions.

RNA secondary structure influences protein binding interactions.

Abstract

In post-transcriptional regulation, an mRNA molecule is bound by many proteins and/or miRNAs to modulate its function. To enable combinatorial gene regulation, these binding partners of an RNA must communicate with each other, exhibiting cooperativity. Even in the absence of direct physical interactions between the binding partners, such cooperativity can be mediated through RNA secondary structures, since they affect the accessibility of the binding sites. Here we propose a quantitative measure of this structure-mediated cooperativity that can be numerically calculated for an arbitrary RNA sequence. Focusing on an RNA with two binding sites, we derive a characteristic difference of free energy differences, i.e. \Delta\Delta G, as a measure of the effect of the occupancy of one binding site on the binding strength of another. We apply this measure to a large number of human and C. elegans mRNAs, and find that structure-mediated cooperativity is a generic feature. Interestingly, this cooperativity not only affects binding sites in close proximity along the sequence but also configurations in which one binding site is located in the 5'UTR and the other is located in the 3'UTR of the mRNA. Furthermore, we find that this end-to-end cooperativity is determined by the UTR sequences while the sequences of the coding regions are irrelevant.