Statistical modeling aids in the development of high-performance whole cell biosensors

Abstract

Whole cell biosensors are genetic systems that, for sensing and control applications, correlate the
presence of a chemical or other stimulus with a user-defined gene expression output. However, the
degree of gene expression of biosensor regulatory components needed for optimal performance is
not intuitive, and multidimensional experimental space is not effectively explored by standard
iterative procedures. In order to overcome these obstacles, we effectively mapped gene expression
levels and improved the performance of biosensors using a design of experiments (DoE)
methodology. This approach was used on two biosensors that react to protocatechuic acid and
ferulic acid, byproducts of the catabolic degradation of lignin biomass. In order to create biosensor
designs with both digital and analog dose-response behavior, we systematically changed the
biosensor's dose-response behavior using DoE. We did this by increasing the maximum signal output
by up to 30 times, improving the dynamic range by more than 500 times, increasing the sensing
range by approximately 4 orders of magnitude, increasing the sensitivity by more than 1500 times,
and modulating the slope of the curve. This DoE approach has potential for optimizing metabolic
pathways and regulatory systems built from new, poorly understood components.