Designing RNA folding cooperativity

Abstract

We have developed a new method for computing the partition functions of RNA secondary structures. Previous work has shown that it is accurate when tested against exact lattice simulations, and that it predicts experimental melting curves on molecules smaller than about 100 base-pairs. The method can also predict RNA folding energy landscapes. Here, we show that this method predicts unexpected sequence-dependent contributions to melting cooperativity. GC pairs are more stable than AU pairs. We find that shifting GC positions relative to AU pairs in a simple hairpin can cause the hairpin either to melt, loop-end first, or open-end first, depending on the GC placements. Such sequence dependence would not be predicted by gaussian chain and Jacobsen-Stockmayer treatments that neglect excluded volume, as is the case with most current treatments of RNA stabilities. No experimental tests are yet available.