Abstract
Whole-cell biosensors are powerful tools for metabolite monitoring, yet challenges such as narrow dynamic range and high leaky expression limit their broader applications. Here, we present a systematic workflow based on two Design-Build-Test-Learn (DBTL) cycles to develop and optimize a transcription factor-based pyruvate biosensor in Escherichia coli. In the first iteration of the cycle, we constructed a biosensor that responded to intracellular pyruvate levels within the 0.05-10 mM range. In the second cycle, we implemented the design of experiments (DoE) to systematically explore combinatorial effects of promoters and ribosome-binding sites (RBSs). A first set of experiments was designed to identify factors with a significant effect on biosensor performance. The results showed that the RBS of the reporter gene significantly influenced the dynamic range by modulating basal and maximum expression, while the RBS of the transcription factor affected the signal span. The Akaike Information Criterion was used to select a model incorporating two main effects and one interaction effect. The best-performing strain exhibited an 18.54-fold increase in the dynamic range and a 37.22-fold reduction in leaky expression. Quantification of intracellular pyruvate confirmed an operational range of 1.23-6.81 μmol/g DCW. Our work demonstrates the power of DBTL cycles with statistical modeling for biosensor engineering, offering potential applications in precise metabolic regulation and screening applications.