Rational design of dual-atom nanozymes for deep infection healing guided by an interpretable intrinsic descriptor.

Deep infections (DIs) continue to pose substantial threats to global public health and represent a critical challenge requiring a new generation of antibacterial agents. Oxidase (OXD)-like nanozymes, which directly activate oxygen, offer a promising approach for treating DIs. However, the vast design space has limited progress in discovering efficient OXD-like nanozymes. In this study, we developed a simple, interpretable intrinsic descriptor φABC = minθ0.2d × minχ0.3 × (θd·1 × θd·2)0.1 by integrating density functional theory calculations, machine learning, and experimental validation. The descriptor φABC offered deep physical insights by incorporating atomic properties (A), bimetallic synergistic effects (B), and coordination environments (C). Guided by this descriptor, we discovered FeZn dual-atom nanozymes (DAzymes) exhibiting a kcat value of 0.59 s-1 and a kcat/Km value of 1.43 × 1010 mM-1 s-1 for OXD-like reactions. This performance surpassed that of previously reported state-of-the-art nanozymes by three orders of magnitude. With the aid of near-infrared (NIR) irradiation, FeZn DAzyme achieved a methicillin-resistant Staphylococcus aureus (MRSA) clearance rate of 99.99% and exhibited a superior healing effect compared with vancomycin, effectively overcoming bacterial resistance, eliminating bacterial biofilms, and promoting DI recovery. Our work integrates descriptor-guided design with biological applications, which ultimately provides a new paradigm for screening nanozymes and elucidating structure-mechanism-activity-therapeutic relationships.