Regulation of the pap operon in uropathogenic Escherichia coli is phase variable. This phase variation arises from competition between regulatory proteins at two sites within the regulatory region, GATC(dist) and GATC(prox). We have used the available literature data to design a stochastic model of the molecular interactions of pap regulation and expression during growth in a non-glucose environment at 37 degrees C. The resulting wild-type model is consistent with reported data. The wild-type model served as a basis for two "in silico" mutant models for investigating the role of key regulatory components, the GATC(dist) binding site and the PapI interaction with Lrp at the GATC(prox) site. Our results show that competition at GATC(dist) is required for phase variation, as previously reported. However, our results suggest that removal of competition at GATC(dist) does not affect initial state dependence. Additionally, the PapI involvement in Lrp translocation from GATC(prox) to GATC(dist) is required for the initial state dependence but not for phase variation. Our results also predict that pap expression is maximized at low growth rates and minimized at high growth rates. These predictions provide a basis for further experimental investigation.

Pyelonephritis-associated pili (Pap) expression in uropathogenic Escherichia coli is regulated by a complex phase variation mechanism involving the competition between leucine-responsive regulatory protein (Lrp) and DNA adenine methylase (Dam). Population dynamics of pap gene expression has been studied extensively and the detailed molecular mechanism has been largely elucidated, providing sufficient information for mathematical modeling. Although the Gillespie algorithm is suited for modeling of stochastic systems such as the pap operon, it becomes computationally expensive when detailed molecular steps are explicitly modeled in a population. Here we developed a Markov Chain model to simplify the computation. Our model is analytically derived from the molecular mechanism. The model presented here is able to reproduce results presented using the Gillespie method, but since the regulatory information is incorporated before simulation, our model runs more efficiently and allows investigation of additional regulatory features. The model predictions are consistent with experimental data obtained in this work and in the literature. The results show that pap expression in uropathogenic E. coli is initial-state-dependent, as previously reported. However, without environment stimuli, the pap-expressing fraction in a population will reach an equilibrium level after approximately 50-100 generations. The transient time before reaching equilibrium is determined by PapI stability and Lrp and Dam copy numbers per cell. This work demonstrates that the Markov Chain model captures the essence of the complex molecular mechanism and greatly simplifies the computation.