Simultaneous, high dimensional strain engineering and process optimisation for increased productivity and robustness

The standard model of bioprocess development is to establish genetics/strain first using screens at the lab scale, then take a set of high producing strains into process development and scaleup.  However, the amount of product a biological system produces is a function of its genetics, the process conditions and medium that we subject it to.  Separating the optimisation genetics and environment relies on the erroneous assumption that there are no significant interactions between genetic and environmental factors. 

Here we present data from high dimensional bioprocess optimisations that investigated up to 30 genetic and process factors simultaneously: rapidly identifying and optimising complex interactions that underlie bioprocess performance.  Optimising these interactions resulted in 10 to 200 fold improvements in active product formation in short timeframes.  Our methodologies take advantage of advances in synthetic biology in conjunction with automation, which we drive with Antha, our high level programming language for biology.  Antha enables rapid programming of lab equipment to execute our cutting edge multifactorial experimental designs.  With these, we have regularly observed interactions between as many as four genetic and process factors, whose optimisation has allowed us harness the underlying complexity of the biology and maximise productivity and robustness.

For scaleup, we transfer our multidimensional models into parallel stirred tank bioreactors (12 reactors run in parallel), before scaling into larger steam in place tanks.  Our multifactorial approach ensures that technology transfer is based on sophisticated models of the bioprocess design space, defining a robust  window in which quality and productivity metrics are maximised.