Correlations of 276 missense variants of PSEN1, PSEN2 , and APP on the production of Aβ peptides against variant effect predictors and biophysical structures

Abstract INTRODUCTION Early-onset Alzheimer’s disease (EOAD) is mostly caused by mutations in Presenilin-1,2 ( PSEN1, PSEN2) , and amyloid precursor protein ( APP) genes. Both genetic and experimental evidence has supported the PSEN-APP amyloid hypothesis as the leading pathogenesis of AD. Thus far, 276 missense variants from PSEN1,2 , and APP have been identified. Herein, we report the development of algorithm platforms for predicting the degrees of pathogenic potentials of these various missense mutations in AD. METHODS We performed extensive pairwise correlations between 276 in vitro functional assays of missense variants from PSEN1, PSEN2 , and APP and clinical data among EOAD patients against 37 variant effect predictors (VEPs). Furthermore, we used the structural models of all three proteins to refine understanding of the molecular basis of missense variants in dominant negative (DN) or loss-of-function (LoF) effects. RESULTS We found that VEPs were consistently correlated with age of onset (AAO) among EOAD patients and the Aβ42/Aβ40 ratio biomarker. This study also identified discrepancies in the predictor variable leading to an increase in the Aβ42/Aβ40 ratio; The increased Aβ42/Aβ40 ratio was due to decreased Aβ40 levels by Sun et al., (2016), while Petit et al., (2022) argued that increased Aβ42 level was the culprit. We also identified that with increased predicted pathogenicity, there were decreased Aβ38 levels. Conversely, this study found no direct correlations with Aβ37 and Aβ43. Lastly, structural studies show characteristics of either DN and LoF mechanisms, as there were numerous variants located in buried hydrophobic domains and those on charged surfaces of the proteins. DISCUSSION Currently, a major limitation in the field is the lack of reliable approaches for predicting the pathogenicity of variants in EOAD. We show that VEPs are promising in deciphering the effects of variants in PSEN1, PSEN2 , and APP genes, especially between multiple pathogenic mutations. Our study also illuminates the need for experimental studies to identify the predictor variable causing the increased Aβ42/Aβ40 biomarker. Our biophysical studies also identify the need to characterize whether the mechanism of pathogenesis caused by DN or LoF effects is in scope of the presenilin hypothesis. Research in Context Systematic review: The authors reviewed experimental evidence of pathogenic PSEN1, PSEN2 , and APP variants in scope of the presenilin hypothesis. To our knowledge, the pathogenesis of EOAD remains incompletely understood. There are also no predictive tools/models available for informing clinical translation. Interpretation: VEPs demonstrate strong correlations with AAO and Aβ levels from pairwise correlations with 276 missense variants. Structurally, it appears that mutations demonstrate characteristics of dominant negative and loss-of-function effects. Future Direction: Larger studies are necessary of VEPs for realistic clinical usage. Highlights We performed correlations of 37 VEPs against in vitro data of EOAD variants. VEPs demonstrate potential as diagnostic tools for early onset Alzheimer’s disease. Pathogenic missense variants show characteristics of DN and LoF effects.