Cell-Free Mixing of Escherichia coli Crude Extracts to Prototype and Rationally Engineer High-Titer Mevalonate Synthesis

Abstract

Cell-free metabolic engineering (CFME) is advancing a powerful paradigm for accelerating the design and synthesis of biosynthetic pathways. However, as most cell-free biomolecule synthesis systems to date use purified enzymes, energy and cofactor balance can be limiting. To address this challenge, we report a new CFME framework for building biosynthetic pathways by mixing multiple crude lysates, or extracts. In our modular approach, cell-free lysates, each selectively enriched with an overexpressed enzyme, are generated in parallel and then combinatorically mixed to construct a full biosynthetic pathway. Endogenous enzymes in the cell-free extract fuel high-level energy and cofactor regeneration. As a model, we apply our framework to synthesize mevalonate, an intermediate in isoprenoid synthesis. We use our approach to rapidly screen enzyme variants, optimize enzyme ratios, and explore cofactor landscapes for improving pathway performance. Further, we show that genomic deletions in the source strain redirect metabolic flux in resultant lysates. In an optimized system, mevalonate was synthesized at 17.6 g·L^-1 (119 mM) over 20 h, resulting in a volumetric productivity of 0.88 g·L^-1·hr^-1. We also demonstrate that this system can be lyophilized and retain biosynthesis capability. Our system catalyzes ∼1250 turnover events for the cofactor NAD⁺ and demonstrates the ability to rapidly prototype and debug enzymatic pathways in vitro for compelling metabolic engineering and synthetic biology applications.

Keywords: Escherichia coli; cell-free metabolic engineering; cell-free synthetic biology; in vitro; metabolic pathway debugging; mevalonate.

Figure 1.. A cell-free metabolic engineering (CFME) framework for pathway prototyping demonstrated with the mevalonate pathway.

(A) Enzymatic route for isoprenoid synthesis via the mevalonate pathway. Acetyl-CoA acetyltransferase (ACAT, green square), hydroxymethylglutaryl-CoA synthase (HMGS, blue star), and hydroxymethylglutaryl-CoA reductase (HMGR, yellow circle) are selectively overexpressed in E. coli prior to cell lysis and extract preparation. (B) Cell-free metabolic engineering approach. E. coli containing overexpressed enzymes of the mevalonate pathway are lysed and centrifuged to make multiple distinct crude extracts. Mixing of extracts allows testing of enzyme variants and optimization of enzyme ratios; subsequently, cofactors such as ATP, NAD⁺, and CoA can be optimized and genomic modifications made to the source strain. (C) Pathway balancing. Three glucose can be converted to two mevalonate assuming non-pathway enzymes in the lysate convert 6 excess ATP and 8 excess NADH to ADP and NAD⁺, respectively.

Figure 2.. Synthesis of mevalonate from glucose via mixing of three BL21(DE3) extracts each containing a single overexpressed pathway enzyme (ACAT, HMGS, or HMGR), plus native glycolysis enzymes

(A). Enzyme variant prototyping identified active NADPH-preferring HMG-CoA reductases (B), but demonstrated that NADH-dependent reductases (C) produced a higher mevalonate titer. Each extract contributed 3.3 mg·mL⁻¹ of total protein to the reaction. The system containing ACAT from E. coli, HMGS from S. cerevisiae, and HMGR from P. mevalonii was selected for future characterization. Values represent averages (n=3) and error bars represent 1 s.d.

Figure 3.. Modulating extract ratios informs construction of an all-in-one extract.

(A) Varying the mass ratio of the three extracts demonstrates expression of pathway enzymes is not rate-limiting to the system. (B) SDS-PAGE gel (Coomassie stain) shows: overexpression of three pathway enzymes in a single extract, individual enzyme-enriched extracts mixed together in a CFME reaction, and individual enzyme-enriched extracts. (C) The single BL21(DE3) extract with all three pathway enzymes overexpressed performs similarly to three extracts mixed at equal ratios (Insert) Enzymes are active though pH decreases over the course of reaction. Addition of buffers stabilizes the pH but does not affect mevalonate yield. Values represent averages (n=2 for A, n=3 for C) and error bars represent 1 s.d.

Figure 4.. The presence of and concentration of cofactors modulates mevalonate yield.

(A) Cofactors NAD⁺, ATP, and CoA were supplemented to the reaction two-at-a-time, one-at-a-time, and not supplemented at all. Glucose, acetate salts, phosphate, and extract can support ~80% of yield with no supplementary cofactors (red bar). Variation of initial ATP and NAD⁺ concentration with no CoA (B) and 1mM CoA (C) modulates final mevalonate concentration at 20 hours. The black bar in (A) and (C) represents the cofactor concentration (1 mM ATP, 1 mM NAD⁺, 1 mM CoA) used in all other figures. Values represent averages (n=3) and error bars represent 1 s.d.

Figure 5.. Source strain genome engineering can redirect metabolic flux in the cell free reaction.

(A) Simplified metabolic pathway with target enzymes for knockout in color. (B-G) Concentrations of metabolites over time within a cell free reaction: (B) Mevalonate (C) Lactate (D) Pyruvate (E) Glucose (F) Acetate (G) pH. Knockout of ldhA eliminates lactate accumulation and knockout of ackA-pta eliminates acetate consumption. Values represent averages (n=3) and error bars represent 1 s.d.

Figure 6.. Lyophilization of cell-free reactions for mevalonate production.

Complete CFME reactions containing the enzyme-enriched lysate, substrates, cofactors and salts were assembled, immediately flash-frozen, and lyophilized. The rehydrated reactions generated similar mevalonate synthesis rates and titers compared to the control non-freeze dried system. Values represent averages (n=3) and error bars represent 1 s.d.