Computational design of diverse nuclear factor erythroid 2 activators with cellular antioxidative activity

Oxidative stress disrupts signaling pathways contributing to chronic diseases, while the KEAP1-NRF2 pathway is central to cellular antioxidant defenses. Current synthetic antioxidants struggle to activate this pathway efficiently or selectively. In this study, we employed deep learning algorithms to design miniproteins capable of activating NRF2. Five designed binders potently interfered with the KEAP1-NRF2 complex, exhibiting affinities ranging from 4.4 nM to 53.3 nM toward KEAP1. Two of these binders, designed through the motif scaffolding method, activated NRF2 in eukaryotic cells increasing antioxidant gene expression 3.8-fold and boosting cell survival across oxidative stress levels. Our approach illustrates the potential of integrated deep learning models to develop stable miniproteins that exhibit a variety of structural frameworks and thermodynamic characteristics. These designs hold promise for countering the cumulative effects of oxidative damage and for supporting the establishment of adaptive homeostasis within key antioxidative systems.