Machine Learning-Optimized Single-Atom Catalysts Enable Microenvironment-Adaptive Chemodynamic-Bioorthogonal Cancer...

Area of Science:

• Biomedical Engineering
• Materials Science
• Catalysis

Background:

• Chemodynamic therapy (CDT) shows promise for cancer treatment by utilizing the tumor microenvironment (TME).
• Catalyst stability and activity in the TME's acidic and reductive conditions are significant challenges.
• Developing robust catalysts is crucial for effective CDT.

Purpose of the Study:

• To design and synthesize a programmable dual-catalytic platform for enhanced cancer therapy.
• To overcome the limitations of traditional catalysts in the TME using machine learning.
• To create stable and highly active iron single-atom catalysts (Fe-N5 SAs).

Main Methods:

• Utilized machine learning for predictive modeling and quantitative structure-performance relationship establishment.
• Employed atomic-level precision for the rational synthesis of Fe-N5 SAs.
• Investigated the dual-catalytic activity (CDT and bioorthogonal) of Fe-N5 SAs in the TME.

Main Results:

• Fe-N5 SAs demonstrated superior chemodynamic reactivity and stability in the TME.
• Efficient conversion of hydrogen peroxide to hydroxyl radicals for tumor ablation was observed.
• In situ prodrug activation and doxorubicin synthesis under physiological conditions were achieved.
• Synergistic CDT-bioorthogonal therapy resulted in enhanced antitumor efficacy.

Conclusions:

• Established a machine learning-guided framework for designing single-atom catalysts.
• Demonstrated the potential of Fe-N5 SAs for stimulus-free, combinatorial cancer therapy.
• Expanded the application of metal catalysis in biomedical fields.