Sensor-Selector Strategy for Directed Evolution of Biosynthetic Pathways

Metabolites biosynthetically produced in nature hold enormous potential for the
sustainable production of chemicals, materials and therapeutics. This potential remains
largely unfulfilled because technological limitations severely restrict the types of
metabolites that can be readily detected and the number of biosynthetic pathway
prototypes that can be evaluated. Thus, despite advances in genetic diversification
methods, phenotype evaluation remains the major rate-limiting step. Here we present a
broadly extensible synthetic selection system, known as sensor-selector, that
simultaneously expands our phenotype evaluation capabilities in two ways: 1) a wide
range of metabolites are detectable by drawing on the diversity of natural protein and
RNA sensors and 2) directed evolution is harnessed to select the highest-producing
biosynthetic pathways from libraries orders of magnitude larger than current screening
methods permit. To overcome the problem of unproductive escape mutants that has
previously plagued selection-based pathway optimizations, we employ a toggled
selection scheme with dual selective markers to robustly eliminate cells capable of
circumventing selection. We characterize sensor-selectors for 15 metabolites belonging to
diverse classes – polyketide antibiotics, flavonoids, polymer-precursor diacids, vitamins,
long-chain alkanes and sugars. We have developed a general suite of genetic
interventions capable of expanding library sizes by four orders of magnitude and
programming sensor-selectors to function across user-specified metabolite concentration
ranges. Evaluating more than a billion cells per day, we deploy sensor-selector to 

optimize biosynthesis of the industrially useful metabolite, naringenin, yielding nearly 50
fold improvement, through four iterations of directed evolution. Sensor-selectors allow
multiplexed phenotype evaluation, obviating the need to characterize designs individually,
thereby facilitating rapid design-build-test cycles in biological engineering. We envision
further application of sensor-selectors to metagenomic pathway discovery, enzyme
engineering, and the evolution of robust genetic circuits.