An engineered production strain carries its therapeutic/enzyme gene on a plasmid, not the chromosome. Two forces erode the producing population every generation in a bioreactor:
Segregational loss — at division, a plasmid-bearing
cell has probability p of producing a plasmid-free
daughter (plasmids partition imperfectly).
Metabolic burden — carrying and expressing the
plasmid slows growth: μ+ = μ0·(1 − burden),
while plasmid-free "cheaters" grow at the full
unburdened rate μ0.
dF/dt = (μ+ − μ−)·F·(1−F) − p·μ+·F
F is the fraction of the population still carrying the plasmid. The first term is the competitive disadvantage from burden; the second is the constant leak from segregation. Both terms drain F — burden alone would settle into slow decline, but combined with segregational loss the cheaters compound their advantage every generation.
- Segregation loss rate p — how imperfectly the plasmid partitions at division (worse for low-copy plasmids, better for actively-partitioned ones).
- Metabolic burden — the growth-rate penalty from expressing the payload gene; sets how much faster cheaters divide.
- Selection pressure — an antibiotic (or auxotrophic complementation) in the medium that only plasmid-bearing cells survive, preferentially removing plasmid-free cells from the culture on each dilution/division event — the standard industrial countermeasure.
- Reseed — restart a fresh fermentation run at 100% plasmid-bearing.
Real-world relevance: this is exactly why large-scale biotech manufacturing runs are capped in duration, why continuous antibiotic selection (or plasmid-free-lethal auxotrophy) is standard practice, and why a strain that looks stable in a small flask can still collapse to near-zero titer after enough generations in a production-scale fermenter — the core "manufacturing" and "stability" risk in engineering a microbial consortium or single production strain.