Refining HIV envelope protein conformational data from different experimental sources using bias-resampling methods and information theory.

Abstract

Human immunodeficiency virus (HIV) remains a global epidemic, and vaccine development has been challenging. This challenge is due in part to conflicting data regarding which conformational states of the envelope glycoprotein (Env) are effectively targeted by neutralizing antibodies. Env is known to undergo conformational change as part of its function. Identifying the conformations best bound by neutralizing antibodies will improve vaccine design. To this end, we use bias resampling ensemble refinement (BRER) to integrate and compare prior spectroscopic measurements of HIV Env conformation. By using both electron paramagnetic resonance and single-molecule FRET data as restraints in conformational ensemble refinement, we generate models for the sets of HIV Env conformations consistent with each set of measurements. Results show that the “hidden” conformational states identified via single-molecule FRET spectroscopy indeed show differences from reference cryo-EM structures. Further analysis shows that the “hidden” conformational states have different distance distributions for some residue pairs than those measured via DEER. We have developed information-theoretic techniques to identify DEER label sites that would best differentiate conformational states defined by single-molecule FRET. We hope that testing of these proposed DEER experiments will help resolve the discrepancy between observations using different spectroscopic techniques and better elucidate the conformational heterogeneity of HIV Env. Extension to datasets in the presence of bound neutralizing antibodies will help elucidate the key conformations recognized by broadly neutralizing anti-HIV antibodies and, we hope, further assist in vaccine design.