Improving synthetic methylotrophy via dynamic formaldehyde regulation of pentose phosphate pathway genes and redox perturbation

Highlights

• Expression of PPP genes rpe and tkt under dynamic formaldehyde control improves methanol carbon assimilation.

• Deletion of NAD-dependent malate dehydrogenase improves biomass generation on methanol and yeast extract co-substrate.

• Short-term ¹³C-methanol assay demonstrates expression of methanol consumption enzymes in Δmaldh strain with rpe and tkt.

• Achieved high carbon and energy utilization from methanol through the P_frm expression of rpe and tkt in the Δmaldh strain.

Abstract

Escherichia coli is an ideal choice for constructing synthetic methylotrophs capable of utilizing the non-native substrate methanol as a carbon and energy source. All current E. coli-based synthetic methylotrophs require co-substrates. They display variable levels of methanol-carbon incorporation due to a lack of native regulatory control of biosynthetic pathways, as E. coli does not recognize methanol as a proper substrate despite its ability to catabolize it. Here, using the E. coli formaldehyde-inducible promoter P_frm, we implement dynamic expression control of select pentose-phosphate genes in response to the formaldehyde produced upon methanol oxidation. Genes under P_frm control exhibited 8- to 30-fold transcriptional upregulation during growth on methanol. Formaldehyde-induced episomal expression of the B. methanolicus rpe and tkt genes involved in the regeneration of ribulose 5-phosphate required for formaldehyde fixation led to significantly improved methanol assimilation into intracellular metabolites, including a 2-fold increase of ¹³C-methanol into glutamate. Using a simple strategy for redox perturbation by deleting the E. coli NAD-dependent malate dehydrogenase gene maldh, we demonstrate 5-fold improved biomass formation of cells growing on methanol in the presence of a small concentration of yeast extract. Further improvements in methanol utilization are achieved via adaptive laboratory evolution and heterologous rpe and tkt expression. A short-term in vivo ¹³C-methanol labeling assay was used to determine methanol assimilation activity for Δmaldh strains, and demonstrated dramatically higher labeling in intracellular metabolites, including a 6-fold and 1.8-fold increase in glycine labeling for the rpe/tkt and evolved strains, respectively. The combination of formaldehyde-controlled pentose phosphate pathway expression and redox perturbation with the maldh knock-out greatly improved both growth benefit with methanol and methanol carbon incorporation into intracellular metabolites.

Introduction

Synthetic methylotrophy, or the engineering of methanol utilization for carbon and energy needs in non-native methylotrophs, has been a highly sought-after goal in metabolic engineering in recent years. The increasing availability and decreasing price of methane from natural gas has renewed interest in utilizing methane, and its oxidized product methanol, as a substrate for microbial fermentations (Whitaker et al., 2015). Methanol has also been argued for as an attractive non-food substrate due to its higher degree of reduction compared to conventional sugar alternatives (Schrader et al., 2009). Renewable methanol, or bio-methanol, sources are being explored by reacting carbon dioxide from geothermal steam, biogas, or municipal solid waste, with hydrogen produced via electrolysis (Olah, 2013). Improving renewable methanol production is an active area of research, including a technoeconomic analysis coupling bio-methanol synthesis with electrolytic hydrogen production powered by wind energy (Matzen et al., 2015), investigating the incorporation of renewable energy into carbon recycling, and reducing the cost of electrolysis for hydrogen production (Leonzio, 2018).

Native methylotrophs, which are capable of consuming single-carbon substrates such as methane and methanol to fulfill their carbon and energy needs, can include yeasts, bacteria, fungi, and archaea. Methanol utilization specifically can be found in bacteria and yeasts. Despite progress in the development of protocols for engineering native methylotrophic bacteria, the genetic toolbox remains relatively undeveloped and the number of potential products limited (Pfeifenschneider et al., 2017). Methylotrophic bacteria such as Bacillus methanolicus and Geobacillus stearothermophilus are valuable resources for the implementation of methylotrophy in the platform organism Escherichia coli (E. coli) (Whitaker et al., 2017). Successfully engineering E. coli for methanol utilization would effectively allow for the methanol-driven production of dozens of industrially relevant products.

Attempts to create an E. coli-based synthetic methylotroph have achieved significant progress in the past 5 years (Bennett et al., 2018b) (Antoniewicz, 2019). Groups have demonstrated ¹³C-labeled methanol incorporation in resting cells (Müller et al., 2015a), methanol incorporation into biomass using a yeast extract co-substrate (Whitaker et al., 2017) (Gonzalez et al., 2018), and significantly improved methanol carbon utilization in combination with co-substrates through methanol auxotrophy strain engineering (Chen et al., 2018) (Meyer et al., 2018). These advances in E. coli-based synthetic methylotrophy have relied on NAD-dependent methanol dehydrogenases for initial methanol oxidation and the ribulose monophosphate (RuMP) pathway for subsequent formaldehyde fixation. Expression of three genes encoding native methylotrophic enzymes are utilized to convert methanol to the central carbon intermediate fructose 6-phosphate. An NAD-dependent methanol dehydrogenase, Mdh, converts methanol to formaldehyde, which is fixed with ribulose 5-phosphate (Ru5P) via 3-hexulose-6-phosphate synthase, Hps, to form hexulose-6-phosphate. Hexulose-6-phosphate is subsequently isomerized by 6-phospho-3-hexuloisomerase, Phi, to form fructose 6-phosphate (Fig. 1A). Research on the Mdh, Hps, and Phi enzymes includes the investigation of enzymes from different methylotrophs (Müller et al., 2015a), enzyme evolution (Wu et al., 2016b), and protein fusion approaches (Price et al., 2016). However significant challenges need to be overcome for E. coli to grow on methanol as a sole carbon and energy source. First, unlike native substrates, E. coli has no mechanism to recognize methanol, or regulatory networks in place to upregulate catabolic pathways. Second, the consumption of formaldehyde requires the continuous regeneration of the intermediate Ru5P at unusually high levels. And third, formaldehyde production through the heterologous NAD-dependent Mdh could be limited by a suboptimal redox balance. We focus here on maximizing methanol carbon incorporation and the methanol growth benefit by tackling these challenges with three approaches: (1) implementing dynamic regulation machinery in response to formaldehyde, (2) expressing heterologous non-oxidative pentose phosphate pathway genes for the regeneration of Ru5P, and (3) strain engineering to increase reliance on reducing equivalents from methanol for biomass production.

Implementing dynamic regulation mechanisms in the cell should allow E. coli to effectively recognize methanol as a substrate. In this respect, the E. coli formaldehyde-inducible promoter, P_frm, is particularly useful. The P_frm promoter, found upstream of the formaldehyde-detoxification frmRAB operon in E. coli, has been found to respond with high specificity to formaldehyde, which disrupts the DNA binding of the transcriptional repressor FrmR (Fig. 1B) (Denby et al., 2016). Using the P_frm promoter, the expression of genes involved in continued methanol and formaldehyde assimilation can be regulated directly in response to cell needs. Formaldehyde induction also prevents the accumulation of the toxic intermediate, which can lead to slowed growth or cell death (Yu et al., 2015).

In addition to mdh, hps, and phi, heterologous non-oxidative pentose phosphate pathway (PPP) genes from the methylotroph B. methanolicus MGA3 have also been investigated for their ability to bypass or alleviate metabolic bottlenecks. The expression of five B. methanolicus genes: rpe, tkt, fba, glpX, and pfk, which were integrated into the chromosome of E. coli and constitutively expressed, were shown to significantly improve ¹³C-labeled methanol incorporation in a range of intracellular metabolites and amino acids, including a 41% improvement in labeled glycine and a 17% improvement in labeled 3-phosphoglycerate (3PG) (Bennett et al., 2018a). We hypothesize that further investigation of these five genes and optimization of their expression can lead to further improvements in methanol utilization.

In an effort to generate gluconate- and methanol-essential growth in E. coli, flux balance calculations predicted that most cellular NADH was produced from the heterologous Mdh (Meyer et al., 2018), and it was thus hypothesized that lower TCA cycle activity, and therefore lower NADH production via the TCA cycle, would improve growth on methanol and gluconate. This hypothesis is supported by metabolic flux analysis on the Δmaldh strain during growth on glucose, which predicts significantly reduced TCA cycle activity leading to high acetate overflow production (Long, 2018). Native methylotrophs utilizing the RuMP pathway often have incomplete TCA cycles, including the obligate methylotroph Methylobacillus flagellatus, which is missing the α–ketoglutarate, malate, and succinate dehydrogenase enzymes (Chistoserdova et al., 2007). In an attempt to mimic methylotrophs and re-calibrate the redox balance, Meyer et al. generated a knock-out of the NAD-dependent malate dehydrogenase, encoded by maldh, which led to major improvements in growth with methanol and gluconate (Meyer et al., 2018). Manipulation of the redox balance to make the oxidation of methanol more thermodynamically favorable can increase methanol assimilation and its proportional contribution to cellular reducing equivalents.