(182d) Rational Optimization of Synthetic Metabolic Pathways with the Ribosome Binding Site Calculator

Synthetic metabolic pathways can convert common cellular metabolites into a wide variety of valuable chemicals, including fuels, drugs, and materials. To achieve economic viability, the productivity of these pathways must be maximized. The current state-of-the-art relies on trial-and-error mutagenesis, where the genetic elements of the pathway, such as promoters or ribosome binding sites, are randomly mutated until a near-optimal pathway is identified. These techniques can be effective when a high-throughput assay or screen for pathway productivity is available. However, high-throughput assays are not common, creating the laborious challenge of measuring the productivity of an unfeasibly high number of mutated genetic systems to find the best one.

To solve this problem, we have developed a rational optimization method for synthetic metabolic pathways that avoids the need for a high-throughput assay. The method identifies the optimal production rates for each enzyme in the pathway by combining both computational predictions and a small number of experimental assays. The computational predictions are generated by a newly developed method, called the Ribosome Binding Site (RBS) Calculator (http://voigtlab.ucsf.edu/software). The RBS Calculator predicts the sequence of a ribosome binding site that yields a user-selected enzyme production rate in bacteria. By combining the predictions of the RBS Calculator with Newton's method, a classic optimization method, one can perform a small number of rational experimental perturbations to the enzymes' production rates and determine the optimal production rates that maximize the pathway's productivity. The final results of the method are the sequences of the ribosome binding sites that maximize the pathway's productivity. These synthetic DNA sequences are placed upstream of each enzyme's coding sequence. Importantly, the optimization method does not require any knowledge of the enzymes' kinetics, the intracellular metabolite concentrations, or other mechanistic or toxicity information.

We first demonstrate the utility of the method with numerical examples. We then present preliminary experimental data of the method's use on an example metabolic pathway. The numerical examples test the method's ability to optimize a black-boxed differential equation model of a linear, branched, or regulated metabolic pathway. Following these numerical tests, the optimization method is employed on a six-enzyme crtEBIDFC metabolic pathway in Escherichia coli that converts isopentenyl pyrophosphate to a yellow carotenoid (hydroxy-spheroidene).