Hao‐Hong Chen, Qian‐Xi Zheng, Jv-Liang Dai, Jia‐Yuan Luo, Ming‐Hua Liang, Jian‐Guo Jiang
ABSTRACT Astaxanthin biosynthesis in engineered Escherichia coli was intensified by coupling fermentation optimization with modular enzyme self‐assembly. Single‐factor screens followed by an L9 (3^4) orthogonal design quantified how cultivation time, temperature, agitation, and IPTG concentration shape astaxanthin accumulation. A back‐propagation neural network (BPNN) trained on these data sets accurately predicted titres ( R² = 0.98). Range analysis of the orthogonal array identified an optimal discrete operating point (36 h, 18°C, 150 rpm, 4 mM IPTG), which was further supported by the BPNN landscape and validated experimentally to yield 597.06 ± 21.32 µg g −1 dry cell weight (DCW). We then repurposed orthogonal docking‐domain pairs from type I trans ‐AT polyketide synthases to co‐localize terminal enzymes in the carotenoid pathway, increasing astaxanthin production to 1.3 mg g −1 DCW in strain sE. HC073 (2.2‐fold over the fermentation‐optimized control). Together, these results show that data‐driven process selection and selective spatial organization provide complementary levers for building higher‐yield carotenoid cell factories.