Predicting the impact of missense mutations on an unresolved protein's stability, structure, and function: A case study of Alzheimer's disease-associated TREM2 R47H variant.

AlphaFold2 (AF2) has spurred a revolution in predicting unresolved structures of wild-type proteins with high accuracy. However, AF2 falls short of predicting the effects of missense mutations on unresolved protein structures that may be informative to efforts in personalized medicine. Over the last decade, countless in-silico methods have been developed to predict the pathogenicity of point mutations on resolved structures, but no studies have evaluated their capabilities on unresolved protein structures predicted by AF2. Herein, we investigated Alzheimer's disease (AD)-causing coding variants of the triggering receptor expressed on myeloid cells 2 (TREM2) receptor using in-silico mutagenesis techniques on the AF2-predicted structure. We first demonstrated that the predicted structure retained a high accuracy in critical regions of the extracellular domain and subsequently validated the in-silico mutagenesis methods by evaluating the effects of the strongest risk variant R47H of TREM2. After validation of the R47H variant, we predicted the molecular basis and effects on protein stability and ligand-binding affinity of the R62H and D87N variants that remain unknown in current literature. By comparing it with the R47H variant, our analysis reveals that R62H and D87N variants exert a much less pronounced effect on the structural stability of TREM2. These in-silico findings show the possibility that the R62H and D87N mutations are likely less pathogenic than the R47H AD. Lastly, we investigated the Nasu-Hakola (NHD)-causing Y38C and V126G TREM2 as a comparison and found that they imposed greater destabilization compared to AD-causing variants. We believe that the in-silico mutagenesis methods described here can be applied broadly to evaluate the ever-growing numbers of protein mutations/variants discovered in human genetics study for their potential in diseases, ultimately facilitating personalized medicine.