Label-free Identification of Protein Aggregates Using Deep Learning

Abstract

AbstractProtein misfolding and aggregation play central roles in the pathogenesis of various neurodegenerative diseases (NDDs), including Huntington’s disease, which is caused by a genetic mutation that leads to a polyglutamine repeat length > 35 in exon 1 of the Huntingtin protein (Httex1). Current research on protein aggregation often involves the use of fluorescent labels to visualize and monitor the dynamics of protein expression, which can alter the biophysical properties of proteins and the final ultrastructure, composition, and toxic properties of the formed aggregates. To overcome this limitation, we present a method for label-free identification of NDD-associated aggregates (LINA). Our approach utilizes deep learning to detect unlabeled and unaltered Httex1 aggregates in living cells from transmitted-light images, without the need for fluorescent labeling. We developed pixel-classification and pixel-regression models, which are robust across imaging conditions, and validated them on aggregates formed by different constructs of Httex1. Our results reveal that Httex1 proteins with shorter polyglutamine repeat lengths form aggregates with a higher average dry mass and area, highlighting the differences in their ultrastructure and aggregation mechanisms. LINA enables the dynamic identification of label-free aggregates and measurement of their dry mass and area changes during their growth process. Our highly-robust models offer high speed, specificity, and simplicity to analyze label-free protein aggregation dynamics and obtain high-fidelity information.