Design of the biosensor-dependent coupling system stabilizes the high-synthesis phenotype of cell factory

Microbial cell factories offer a sustainable route to plant-derived natural products, but yield drift and strain degeneration persist. Growth-coupled biosynthesis can continuously enrich high producers, yet specific product-responsive biosensors remain scarce and their population-level effects are unclear. Here, we describe a rapid transcriptome-mining workflow that, as proof-of-concept, delivers yeast biosensors for glycyrrhetinic acid and medicarpin. By fine-tuning PDR5 promoter, we expand the dynamic range of the glycyrrhetinic acid sensor and wire it to an essential gene, establishing a growth-addiction circuit that increases titer by 46.8 % after subculture. Single-cell transcriptome reveals that the evolved strain population exhibits a completely different division of labor compared to the initial strain. Coupling does not eliminate phenotypic heterogeneity; instead, it amplifies a dedicated sub-population marked by discrete transcriptional signatures. Deletion of genes highly expressed in non-producing cells or enrichment of high-producing cell clusters can further boost population-level production. This study provides both a generalizable biosensor-discovery platform and single-cell-guided strategies for stabilizing and optimizing natural-product cell factories.