De novo designed binders suppress aggregation of Cu/Zn-superoxide dismutase implicated in amyotrophic lateral sclerosis

Le résumé fourni par la source

Protein aggregation represents a pathological hallmark of many neurodegenerative disorders and constitutes a therapeutic target. However, the development of molecules capable of modulating pathological aggregation has remained elusive. Recent advances in computational protein engineering, particularly using machine-learning frameworks, have enabled the de novo design of proteins that specifically bind target proteins. Here, we applied this strategy to design de novo binders against Cu/Zn-superoxide dismutase (SOD1), whose aberrant aggregation is implicated in amyotrophic lateral sclerosis caused by SOD1 mutations. SOD1 functions as a homodimer, whereas disease-linked mutations shift the equilibrium toward an aberrant monomer prone to aggregation; accordingly, we sought binders that recognize monomeric SOD1. Among the designed candidates, binder #313 displayed selective binding to monomeric SOD1. Through iterative optimization, we improved its binding affinity to yield an enhanced variant, #313-604. The crystal structure of the complex between #313-604 and SOD1 monomer closely matched the model predicted by AlphaFold2, confirming the designed interface. Importantly, whereas a monomeric SOD1 variant spontaneously forms oligomers and amyloid-like fibrils over time, #313-604 suppressed the SOD1 aggregation in vitro. Taken together, these results demonstrate that machine-learning-based de novo protein design can generate functional binders capable of directly modulating disease-linked aggregation processes.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.