As only about 1.5% of the human genome translates into proteins, non-coding RNAs have emerged as promising therapeutic targets for several critical diseases. However, designing small-molecule drugs that bind RNA is more challenging than targeting proteins, primarily due to RNA’s high conformational flexibility and uniform charge density that prevents the formation of a hydrophobic pocket. Molecular dynamics (MD) simulations have provided valuable insights into such biomolecular processes, but they are limited by the timescale problem. Drug binding and unbinding often occur on the millisecond scale, which is difficult to capture at an affordable computational cost. To overcome this challenge, enhanced sampling methods have been developed that apply bias along the potential energy surface to accelerate phase space exploration. In this work, we employed the recently introduced on-the-fly probability enhanced sampling (OPES) method and its flooding variant to calculate the binding affinities and residence times of the prototypical theophylline RNA aptamer with natural ligands (theophylline and caffeine) and a synthetic ligand (TAL1). The calculated binding affinities and residence times reproduced the relative trends observed in experiments. Our simulations also captured the C27 base-flipping transition and the complete ligand binding-unbinding process. For the natural ligands, we identified two distinct binding pathways, (1) an induced-fit mechanism and (2) a conformational selection mechanism, while the synthetic ligand strongly favored the induced-fit pathway. From the pathway-dependent free-energy profiles, we determine that the binding affinity critically determines the choice between the induced fit and conformational selection mechanisms of binding. These mechanistic insights, together with the quantitative agreement in binding affinities and residence times, provide a detailed description of RNA-drug recognition.
Elangovan et al. (Sun,) studied this question.