Predictive β-Sheet Protein Stabilization via Computationally Guided Metal Coordination with Simulations and AlphaFold3

Proteins that withstand mechanical forces in nature possess evolved architectures that resist forced unfolding, yet mimicking this robustness in engineered proteins remains challenging. Although metal coordination bonds can act as powerful molecular clamps, their incorporation has been largely empirical, with no rational method to identify where reinforcement would be most effective. Here, we present a predictive workflow that integrates all-atom molecular dynamics (MD) simulations to identify mechanical soft spots, AlphaFold3 (AF3) to assess the structural feasibility of engineered metal-chelating sites, and single-molecule force spectroscopy for experimental validation. Applying this approach to two previously AI-designed mechanostable β-sheet proteins from the SuperMyo series, we demonstrate that Ni2+ coordination produces site-dependent reinforcement, with single-site variants producing substantial increases in unfolding force. Combining multiple coordination sites produced cumulative reinforcement, with a triple-site variant achieving a >110% gain and an unfolding force exceeding 400 pN. This closed-loop strategy connects computational prediction with experimental validation for the reinforcement of protein mechanical stability. Through programmable, site-specific metal coordination, the workflow enables systematic tuning of mechanical properties in β-sheet protein scaffolds.