Abstract

Autonomous bioluminescence systems enable continuous light emission in engineered organisms by genetically encoding both luciferase enzymes and their substrate biosynthetic pathways, offering a powerful platform for non-invasive and long-term monitoring of biological processes. However, bioluminescence output is highly sensitive to substrate availability and host metabolic state, often leading to signal instability under energy-limited conditions. Here, we report an alternative luciferin biosynthetic strategy for the bacterial Lux bioluminescence system in which the adenosine triphosphate (ATP) dependent LuxEC complex is replaced by dioxygenase (DOX), an enzyme that directly converts fatty acids into fatty aldehydes without consuming ATP. Using machine learning guided directed evolution, we engineered DOX variants that markedly enhanced bioluminescence intensity when coupled with bacterial luciferase. The resulting ATP-free Lux bioluminescence system enabled single cell level bioluminescence imaging and maintained stable light emission under diverse antibiotic treatments, demonstrating enhanced robustness against metabolic perturbations.