Relying on the relationship with known disease-causing variants in homologous proteins to predict pathogenicity of SORL1 variants in Alzheimer's disease

SORL1 encodes the retromer-associated receptor SORLA that functions in endosomal recycling. Rare variants in SORL1 have been associated with Alzheimers disease (AD) and rare pathogenic variants are estimated to occur in up to 2.75% of early onset AD patients and in 1.5% of unrelated late onset AD patients. While truncation mutations are observed almost exclusively in AD patients, it is currently unknown which among the hundreds of rare missense variants identified in SORL1, are pathogenic. Here we address this question by relying on SORLAs distinct molecular architecture. First, we completed a structure-guided sequence alignment for all the protein domains. Next, we identified proteins that contain domains homologous to those of SORLA, which include pathogenic variants for monogenic diseases. We identified the analogous domain positions of these variants in the SORLA protein sequence and showed that variants in these positions similarly impair SORL1, and lead to AD. Together, our findings represent a comprehensive compendium on SORLA protein variation and functional effects, which allowed us to prioritize SORL1 genetic variants into high or moderate priority mutations. We envision that this compendium will be used by clinical geneticists for assessing variants they identify in patients, allowing further development of diagnostic procedures and patient counseling strategies. Utimately, this compendium will inform investigations into the molecular mechanisms of endosomal recycling which will support the development of therapeutic treatment strategies for SORL1 variant-carrying patients.