PACT - Prediction of Amyloid Cross-interaction by Threading

Abstract

Amyloids are protein aggregates usually associated with their contribution to several diseases e.g., Alzheimers and Parkinsons. However, they are also beneficially utilized by many organisms in physiological roles, such as microbial biofilm formation or hormone storage. Recent studies showed that an amyloid aggregate can affect aggregation of another protein. Such cross-interactions may be crucial for understanding the comorbidity of amyloid diseases or the influence of microbial amyloids on human amyloidogenic proteins. However, due to demanding experiments, understanding of interaction phenomena is still limited. Moreover, no dedicated computational method to predict potential amyloid interactions has been available until now. Here, we present PACT - a computational method for prediction of amyloid cross-interactions. The method is based on modeling a heterogenous fibril formed by two amyloidogenic peptides. The stability of the resulting structure is assessed using a statistical potential that approximates energetic stability of a model. Importantly, the method can work with long protein fragments and, as a purely physicochemical approach, it relies very little on training data. PACT was evaluated on data collected in the AmyloGraph database and it achieved high values of AUC (0.88) and F1 (0.82). The new method opens the possibility of high throughput studies of amyloid interactions. We used PACT to study interactions of CsgA, a bacterial biofilm protein from several bacterial species inhabiting human intestines, and human Alpha-synuclein protein which is involved in the onset of Parkinsons disease. We show that the method correctly predicted the interactions, performing experimental validation, and highlighted the importance of specific regions in both proteins. The tool is available as a web server at: https://pact.e-science.pl/pact/. The local version can be downloaded from: https://github.com/KubaWojciechowski/PACT