Abstract
The conformational landscapes of intrinsically disordered proteins (IDPs) remain obscured by the ensemble averaging inherent to experimental observables. Here, we present Multi-replica Averaged Restraint Simulation (MARS), a data-driven modeling framework that reconstructs these landscapes by enforcing ensemble-averaged restraints across multiple replicas without imposing prior structural knowledge. Using the N-terminal domain of estrogen receptor alpha (ERα-NTD) as a model system, MARS simultaneously integrates small-angle X-ray scattering (SAXS) and six paramagnetic relaxation enhancement (PRE) profiles, comprising over 600 pairwise restraints, to generate ensembles that quantitatively fit all input structural restraints. The resulting ensemble is consistent with independent backbone relaxation measurements and reveals two major conformational states: a dominant extended state and a low-populated yet functionally relevant compact state whose structural features align with prior mutagenesis studies. Systematic benchmarking demonstrates that SAXS and PRE provide orthogonal global and local constraints, that each PRE profile contributes non-redundant structural information, and that multi-replica sampling is essential for preserving conformational heterogeneity. MARS offers a scalable framework for integrating orthogonal biophysical measurements to resolve both dominant and rare functional states in IDPs.