BPS2026 – Computational design of ultrastable protein via hydrogen bond maximization for extreme environments

Hydrogen bonds are fundamental chemical interactions that stabilize protein structures, particularly in β-sheets, enabling resistance to mechanical stress and environmental extremes. Inspired by natural mechanostable proteins with shearing hydrogen bonds, such as titin and silk fibroin, we de novo designed superstable proteins by maximizing hydrogen bond networks within force-bearing β-strands. Using a computational framework combining AI-guided structure and sequence design with all-atom molecular dynamics (MD) simulations, we systematically expanded protein architecture, increasing backbone hydrogen bonds from four to thirty-three. The resulting proteins exhibited unfolding forces exceeding 1,000 picoNewtons, about 400% stronger than the natural titin Ig domain. Our work introduces a scalable and efficient computational strategy for engineering robust proteins, offering a generalizable approach for the rational design of resilient protein systems for extreme environments.