Christoph Hethey (1,2), Charlotte Kloft (2,3), Wilhelm Huisinga (2,4)
(1) Institut für Biochemie und Biologie, Universität Potsdam; (2) Graduate Research Training Program PharMetrX: Pharmacometrics & Computational Disease Modelling; (3) Institute of Pharmacy, Dept. Clinical Pharmacy & Biochemistry, Freie Universität Berlin and (4) Institut für Mathematik, Universität Potsdam
Objectives: Typical drug-effect models directly link bacterial population growth and exposure to antibiotics. They rarely account for known mechanisms of action of the drug – which are particularly relevant for the analysis of synergistic or antagonistic effects of drug combinations[1]. Our aim was to develop a generic pharmacodynamic model which allows for mechanistic integration of antimicrobial drug effects on the cellular level to predict the impact on bacterial growth.
Methods: Control bacterial growth experiments without drug resulted in baseline values for population growth. An established single cell model was extended to predict cell-level parameters from this growth rate[2]. Instantaneous drug effects were integrated on the cellular level and exemplified for protein synthesis inhibitors. Time dependent cellular responses to this inhibition were predicted by a transit compartment cell-cycle model. Parameter estimation and model assessment for time-kill curves (TKC) were based on training and validation data sets.
Results: The model successfully predicts data for diverse experimental observations: (i) TKC data (E. coli, tetracycline) for constant drug exposure; (ii) septation dynamics during shift from exponential into stationary phase (B. subtilis, no drug); (iii) impact of drug and growth medium on cellular RNA concentrations (E. coli, chloramphenicol) and (iv) lag times between increase of cell number and population mass after change of growth medium (E. coli, no drug). The peptide chain elongation rate turns out to be a crucial predictor for bacterial cell composition during drug exposure. Since all scenarios show good agreement between predicted and experimental data, these promising results are a first step to mechanistically model bacterial growth during exposure to multiple antibiotics.
Conclusions: Our model allows to quantify the impact of strain, growth media and pre-experiment history on TKC and other readouts of antibiotic in vitro assays. Typically, a direct comparison between different experiments is not possible. Mechanistic models, as presented here, can fill this gap – they explicitly consider parameters like control growth rate or persister fractions and extract relevant information. This is critical when comparing the in vitro effects of different antibiotics, for example to ensure optimal combination therapy.
References:
[1] Yeh, P., Tschumi, A. I., & Kishony, R. (2006). Functional classification of drugs by properties of their pairwise interactions. Nature Genetics, 38(4), 489–494.
[2] Bremer, H., & Dennis, P. P. (1996). Modulation of Chemical Composition and Other Parameters of the Cell by Growth Rate. In F. C. Neidhardt (Ed.), Escherichia Coli and Salmonella Typhimurium: Vol 2: Cellular and Molecular Biology (2. ed., pp. 1527–1540).
Reference: PAGE 24 (2015) Abstr 3464 [www.page-meeting.org/?abstract=3464]
Poster: Drug/Disease modeling - Infection