Thermodynamic prediction of RNA cellular activity from sequence via conformational ensembles

Despite advances in structure prediction from sequence, predicting cellular activity requires conformational ensembles that capture propensities to form functionally active states. Such ensembles remain difficult to measure and even harder to predict. Here, we systematically altered the HIV-1 transactivation response element (TAR) RNA sequence to change its propensity to adopt a functional versus inactive secondary structure and quantified these propensities using proton chemical exchange saturation transfer (¹H CEST) NMR without isotopic labeling. Minor sequence changes shifted the active-state propensity by ∼500-fold, quantitatively predicting 125- to 300-fold changes in binding to the RNA-binding region of Tat and cellular transactivation. These propensities could be inferred from secondary-structure prediction algorithms and incorporated into a thermodynamic framework to quantitatively predict how sequence changes alter protein-binding affinity and cellular activity in this well-characterized system. Our findings establish a quantitative thermodynamic framework that links the RNA sequence to cellular activity through conformational ensembles, setting the stage for more generalized predictions as computational ensemble modeling continues to advance.