Leveraging Epistatic Interactions to Model the Structural Ensemble of a Misfolded Rhodopsin Variant

The misfolding of integral membrane proteins plays a prominent role in the molecular etiology of a wide variety of diseases. These misfolding reactions can occur in many ways and deviations in the conformational properties of individual pathogenic variants have undermined efforts to develop broadly efficacious therapeutics. However, the structural basis of membrane protein misfolding remains poorly understood. In this work, we develop a hybrid computational/ experimental approach to develop and refine structural models of S297R rhodopsin- a poorly expressed disease variant that promotes misfolding by stabilizing an alternative membrane topology. By shifting the sequence alignments used for comparative homology modeling, we generated an ensemble of structural models for this aberrant topomer. Models within this ensemble suggest misfolding should result in the formation of several non-native helical interfaces. To probe these contacts, we compared the context-dependent effects of 456 sequence-distant mutations on the plasma membrane expression of WT and S297R rhodopsin by deep mutational scanning (DMS). Our results reveal that the S297R mutation generates widespread folding-mediated epistasis that potentially reflect context-dependent interactions formed by these mutations in the misfolded state. To identify a subset of misfolded models that may account for these interactions, we used in silico DMS profiles of non-native models to iteratively refine a weighted subset of models that best recapitulate the observed mutagenic trends. Our best models suggest conformational defects generated by topological perturbations are propagated across the rhodopsin structure and suggest how certain mutations may rescue the expression of the misfolded S297R variant. Together, these investigations outline a new approach to map denatured state ensembles while providing general insights into the molecular basis of membrane protein misfolding.