Optimization of a bacterial bioluminescent biosensor through experimental design

A new study has investigated the usefulness of experimental designs to optimize a liquid phase bacterial bioluminescent biosensor while aiming at modelling the effects of six selected growth factors, e.g., dilution rate (D), glucose medium concentration (Cg), oxygen (O2), decanal concentration (d), temperature (T) and the buffering capacity (Bc), on the performances of the biosensor, that is, induction ratio (IR), response time (Tres) and recovery time (Trec). eneral statement it was demonstrated that the above-mentioned growth factors explained most of the variability in the biosensorʹs performances. Linear regression models, which were generated by the statistical analysis allowed the definition of a domain consisting of three distinct zones. Two of the zones delimited a spatial area in which all the combinations of the growth factors led to very low IR (less than 2). The third zone was characterized by the presence of local optimums corresponding to the concomitant increase of IR, Tres and Trec. y, the three main performance criteria were optimized simultaneously on the basis of the experimental data and the corresponding combination of values of the growth factors was defined. The modelling procedure highlighted the fact that more than one set of values of the growth factors could be isolated to optimize the biosensor performances. Consequently, the end user(s) will have to choose between a fast and a slow biosensor, according to whether a low or a high level of global sensitivity is preferred. For example, in order to select a sensitive, but slow biosensor (e.g., IR = 12.5, Tres = 80.5 min, Trec = 624 min) the following values of the growth factors have to be imposed on the biosensor (D = 0.9 h−1, Cg = 1×, O2 low, d = 30 μM, T = 30 °C, pH buffered). Conversely, the following values (D = 0.78 h−1, Cg = 0.85, O2 high, d = 212 μM, T = 34.6 °C, pH buffered) will provide a fast, but less sensitive biosensor (e.g., IR = 2, Tres = 50 min, Trec = 250 min). atistical approach demonstrated that several optimums may be considered and can therefore be reached. However, a compromise between two antagonistic options has to be found in relation to the performance optimization of such a bioluminescent biosensor.