Dencichine—chemists call it β-ODAP—is the compound that makes the traditional Chinese herb Panax notoginseng stop bleeding. It also shows neuroprotective activity. Extracting it from plants is slow and inefficient, so metabolic engineers have tried growing it in Escherichia coli instead. The first attempt stalled at 1.29 grams per liter, crippled by a familiar trade-off: the engineered pathway poisoned the cells that were supposed to produce it. A team at Beijing University of Chemical Technology, writing in The Innovation Life , has now untangled that toxicity and pushed production to 13.46 grams per liter—a roughly tenfold gain and the highest reported titer—at a yield of 0.32 grams per gram in 3-liter bioreactors. The work started conventionally. Flux-balance analysis picked the oxaloacetate-cleavage route as the most efficient of several candidate pathways. Multi-copy integration of key genes plus deletion of the competing serB gene raised the limiting precursor L-DAP to 3.07 grams per liter, and iterative tuning of the two acyl-transfer steps—ScAAE and LsBAHD—lifted output another 223%. Balancing flux at the PEP–OAA–pyruvate node brought the titer to 2.3 grams per liter. Then growth collapsed, and adding back L-serine or L-glycine did not fix it. Rather than pushing flux harder, the team ran adaptive laboratory evolution on the stunted strain and sequenced the survivors. The evolved lines carried mutations in glpK and serA. Reverse engineering showed what serA was hiding: the enzyme it encodes, 3-phosphoglycerate dehydrogenase, is promiscuous enough to convert α-ketoglutarate into 2-hydroxyglutarate, a toxic byproduct best known as an oncometabolite. Serine-pathway engineering had been quietly poisoning its own host. The fix was a growth-phase-dependent dual-dynamic circuit with no external inducer. A growth-dependent promoter keeps serB active early—avoiding serine starvation—then gradually shuts it off as cells enter stationary phase, when flux should serve the product. A stationary-phase promoter does the opposite to serA: repressed during exponential growth to keep 2-hydroxyglutarate at bay, activated later to feed precursor supply. The resulting strain, ODAP28, cut the toxin to trace levels and reached 4.43 grams per liter with biomass largely restored. One constraint remained. The pathway generates surplus NADH, so the team introduced an NADH-dependent glutamate dehydrogenase (BsuRocG), building a cofactor-recycling loop that lowers the NADH/NAD+ ratio while supplying glutamate as a substrate. The final strain, ODAP32, delivered 13.46 grams per liter. The authors frame the finding as a general lesson: substrate promiscuity in engineered enzymes can create hidden cellular toxins, and adaptive evolution plus reverse engineering is an efficient way to find them. The growth-coupled circuit needs no inducers or heterologous regulators, and the team says the strategy extends to other serine-pathway products such as L-cysteine and L-tryptophan. How 2-hydroxyglutarate actually arrests growth, and whether the circuit can be made more responsive, remain open questions—and industrial scale-up beyond the 3-liter reactor is still ahead.
The Innovation Life
10.59717/j.xinn-life.2026.100242
Experimental study
Not applicable
Dual-dynamic control and metabolic rebalancing guided by toxic target discovery for efficient dencichine production
18-Aug-2026
The authors declare no competing interests.