Team:Wageningen UR/project/model overview

Modeling Overview

Three aspects of the BananaGuard system have been modeled. (I) Accurate balancing of the different promoters elements in the the kill switch is required in order to maintain a bi-stable system. Using statistical mechanics an estimation for the optimal promoter configuration was made, balancing the amount of different repressors with number of repressor binding sites and their respective position on the promoter. The obtained results have inspired the Design of promoters for the kill-switch. Having designed an optimal system its performance in the soil can be estimated in order to gain insight in its functionality. From a modeling perspective the most interesting questions that can be answered are: the metabolic price that has to be paid for the introduced genetic circuit and the stability and performance of the genetic circuit. (II) A genome scale metabolic model was constructed outlining and answering the Cost query and subsequently the biological control agents ability to not be outcompeted in the soil by other rizhosphere populating micro-organisms. (III) A stochastic model was made in order to quantify and characterize the Stability and Performance of the introduced genetic circuit. The model takes into account rates predicted by the metabolic model, cell division, the anti-gene transfer toxin antitoxin systems effect on cell growth and the kill-switch.

Key results

Design

Before the optimal configuration of repressor binding sites could be determined, the behaviour of the system had to be assigned scores representing a functioning or non-functioning toggle switch. Clustered, three different behaviours could be identified:

In total eight different configurations of repressor binding sites that were tested (Figure 1).Two, F and H, were found to be moderately robust towards variations in binding strengths and production rates. H was chosen as a base to design new promoters for the kill-switch. Biologically it meant there be two TetR repressor binding sites on the LacI promoter, two LacI and one CIλ repressor binding site on the TetR promoter and Two TetR one CIλ repressor binding site in for the toxin promoter

Color maps indicating functioning and non-functioning systems. Each letter represents different repressor binding site configurations. Each small square within the colour maps represents a score for a simulation of the system with a unique set of parameters. The colours correspond to the previously given description

Cost

After changing the carbon composition and quantity to a biological relevant value for the rhizosphere of the banana roots a growth rate of >99% compared to the wild type has been calculated for the resting state and still over 50% for the active state. This indicates that metabolic stress is not a bottleneck for the production of anti-fungals in our activated system.

Performance

Overall the results derived from the stochastic model indicates a large distribution of division times in the population. These distributions can be attributed to the free toxins in the cell caused by stochastic fluctuations in the native toxin antitoxin system and a varying basal production rate on the kill-switch. Figure 3.A indicates that a basal CIλ production rate of 50nM or higher will be devastating to the stability of the genetic circuit in a large portion of the population. The long division times caused by the leaking toxins which in turn is caused by the slow CIλ build-up, result in the kill-switch to change state. Figure 3.B shows that slower growth rates activate the system more efficiently because CIλ is not diluted as much. Given the growth rate obtained from the metabolic model an activation efficiency of 98% is achieved. This means that the vast majority of our reengineered P. putida population will activate in the soil according to the model.

Implications for the experimentalist and the system

Promoter Design: The statistical mechanics model has led to the experimentalists decision to opt for a new set of designed promoters and build two kill-switches in parallel. The model has predicted the newly designed promoters to have a higher stability. These needs to be verified experimentally.

Cost: The results for both the resting and active state show that BananaGuard is not outcompeted by other rhizosphere-populating microorganisms and that metabolic stress is not a bottleneck of the anti-fungal production relatively. This means that for our application it is not necessary to change the proteins in the kill switch or produce less or different anti-fungals. The calculated growth rate is used to calculate the system performance.

Performance: The results have shown that leakiness of promoters on the input/output plasmid (CIλ and Zeta-toxin) can be detrimental to the performance of our reengineered P. putida. A point of attention when the entire system is tested.

The activation of our system is limited by the production of CIλ and not the influx of fusaric acid. The fusaric acid detection experiments has shown the fusaric acid dependent promoter to be activated with a limited amount of fusaric acid present. The activation occurs more efficiently if the cell growth rate is low. This means that it is highly probable that the majority of BananaGuards population will activate given a generally slower growth rate in stressful environments