A structured, segregated model for genetically modified Escherichia coli cells and its use for prediction of plasmid stability.

A structured, segregated model is presented for an asynchronously growing population of genetically modified Escherichia coli cells. A finite representation method was modified so that 272 cells could be used to represent a microbial population. The concept of a "limbo" compartment was introduced to allow random plasmid distribution to daughter cells upon cell division while restricting the number of computer cells included in the calculation. This scheme enabled us to predict plasmid instability and distribution of plasmid-originated properties in a population without a priori determination of growth rates or probability of forming plasmid-free cells from plasmid-containing cells. Predictions of population behavior using a single-cell model requires no adjustable parameters. The results comparing different induction strategies suggest that in continuous culture, there exists an optimum efficiency of partial induction that maximizes the long-term productivity of the gene product due to plasmid stability. With the optimum efficiency of partial induction, constant induction appears to prove more stable than cycling induction.