Astaxanthin biosynthesis in engineered Escherichia coli was intensified by coupling fermentation optimization with modular enzyme self-assembly. Single-factor screens followed by an L9 (3⁴) 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.
Chen et al. (Thu,) studied this question.