An automated data-driven pipeline for improving heterologous enzyme expression

A key challenge in cellular biomanufacturing of fuels, chemicals, and pharmaceuticals is that many pathway enzymes have very low activity, limiting overall titers and productivities. One reason is that enzymes are marginally stable under their native conditions, and expression in a different environment can thermodynamically favor the unfolded state. Additionally, overexpression can result in aggregation because natively expressed proteins are close to their solubility limit. This challenge suggests an engineering solution: engineer pathways enzymes to be stable in their biomanufacturing chassis. However, this is difficult because: (a.) many enzymes do not have high-throughput activity screens needed for directed evolution; (b.) there are few or no structures available; (c.) there are often multiple limiting enzymes; (d.) most mutations confer small benefits to stability; and (e.) the plurality of stability-enhancing mutations decrease catalytic efficiency. I will present a culmination of my group\u27s approach to solve the above challenges, in effect automating the design of stable, active enzymes from limited combinatorial datasets. This engineering strategy involves user-defined precise mutagenesis1,2, deep sequencing to evaluate the functional effect of nearly all possible single point mutants on solubility3, Bayesian methods to discriminate stable, catalytically neutral from deleterious mutations3, and computational design to combine up to 50 mutations at once4. I will show recently published work on application of this method to improve the pathway productivity of a medicinal alkaloid pathway in Saccharomyces cerevisiae5, and end with the description of a computational pipeline to automate our process for any enzyme of interest. References 1. Wrenbeck EE, KlesmithJR, AdeniranA, StapletonJA, Tyo KJ, Whitehead TA, (2016) Plasmid-based single-pot saturation mutagenesis, Nature Methods 13(11): 928-930 doi:10.1038/nmeth.4029 2. Medina-Cucurella A, Steiner PJ, Faber MS, Beltran J, Borelli A, Kirby M, Cutler S, Whitehead TA, (2019) “User-defined single pot mutagenesis using unpurified oligo pools”, under review 3. Klesmith JR, Bacik JP, Wrenbeck EE, Michalczyk R, Whitehead TA (2017) Trade-offs between enzyme fitness and solubility illuminated by deep mutational scanning, PNAS 114:2265-2270 doi: 10.1073/pnas.1614437114 4. Klesmith JR, Bacik JP, Michalczyk R, Whitehead TA (2015) High-resolution sequence function mapping of a levoglucosan utilization pathway in E. coliâ, ACS Synthetic Biology 4 (11), 1235-1243 DOI: 10.1021/acssynbio.5b001 5. Wrenbeck EE, Bedewitz MA, Klesmith JR, Noshin S, Barry CS, Whitehead TA* (2019) “An automated data-driven pipeline for improving heterologous enzyme expression” ACS Synthetic Biology accepte