Mapping the Conformational Heterogeneity Intrinsic to the Protein Native Ensemble

In the AlphaFold era, there is a significant momentum in predicting protein structures, functionality, and mutational hotspots from deep learning approaches. In this review, we highlight how structural information is only a starting point in understanding function and why a single structure for a given sequence rarely captures the true picture. We provide an overview of selected experimental and computational techniques that can be employed to delineate the conformational landscapes of proteins at different levels of resolution. An integrative approach has often led to the identification of considerable heterogeneity in the native ensemble, with high-resolution methods revealing proportionately larger complexity. Partial structure in the native ensemble appears to be the norm, which typically appears as excited or intermediate states in the folding conformational landscape. A more nuanced approach mapping the ensemble of states, their relative populations, associated time scale of interconversion, and their sensitivity to different physical perturbations is therefore necessary. Thus, "sequence-ensemble-function" paradigm is the way forward even for apparently well-folded proteins, with multiprobe experiments and physically grounded models providing an intimate and intuitive understanding of this connection.