Engineering a synthetic energy-efficient formaldehyde assimilation cycle in Escherichia coli

Abstract

One-carbon (C1) substrates, such as methanol or formate, are attractive feedstocks for circular bioeconomy. These substrates are typically converted into formaldehyde, serving as the entry point into metabolism. Here, we design an erythrulose monophosphate (EuMP) cycle for formaldehyde assimilation, leveraging a promiscuous dihydroxyacetone phosphate dependent aldolase as key enzyme. In silico modeling reveals that the cycle is highly energy-efficient, holding the potential for high bioproduct yields. Dissecting the EuMP into four modules, we use a stepwise strategy to demonstrate in vivo feasibility of the modules in E. coli sensor strains with sarcosine as formaldehyde source. From adaptive laboratory evolution for module integration, we identify key mutations enabling the accommodation of the EuMP reactions with endogenous metabolism. Overall, our study demonstrates the proof-of-concept for a highly efficient, new-to-nature formaldehyde assimilation pathway, opening a way for the development of a methylotrophic platform for a C1-fueled bioeconomy in the future.

Fig. 1. The synthetic erythrulose monophosphate (EuMP) cycle.

a A schematic representation of the EuMP cycle and its modularization. Pathway enzyme names are shown in red. The four modules for metabolic engineering are highlighted in different color codes. Overall, the pathway transforms three molecules formaldehyde (FALD) into a molecule of glyceraldehyde 3-phosphate (GAP). b ATP and reducing equivalent costs of the EuMP cycle and the three natural formaldehyde assimilation pathways, RuMP (ribulose monophosphate pathway), XuMP (xylulose monophosphate pathway) and serine cycle. The costs are for production of acetyl-CoA in terms of serine cycle and for GAP in terms of other pathways. c The EuMP pathway is equally efficient as the natural RuMP for most precursor production. See Methods for the flux balance analysis. The values were normalized to the yields via the RuMP cycle. 3PG 3-phosphoglycerate, AcCoA acetyl-coenzyme A, akg 2-ketoglutarate, DHAP dihydroxyacetone phosphate, DerI D-erythrulose 4-phosphate isomerase, E4P D-erythrose 4-phosphate, EPS D-erythrulose 1-phosphate synthase, EryC D-erythrulose 1-phosphate 3-epimerase, D-Eu1P D-erythrulose 1-phosphate, L-Eu1P L-erythrulose 1-phosphate, D-Eu4P D-erythrulose 4-phosphate, F6P fructose 6-phosphate, FALD formaldehyde, FBA fructose-bisphosphate aldolase, FDP fructose-1,6-bisphosphate, G6P glucose 6-phosphate, GAP glyceraldehyde 3-phosphate, LerI L-erythrulose-1-phosphate isomerase, OAA oxaloacetate, PEP phosphoenolpyruvate, PFK phosphofructokinase, R5P ribose 5-phosphate, RPE ribulose-5-phosphate 3-epimerase, RPI ribose-5-phosphate isomerase, Ru5P ribulose 5-phosphate, S7P sedoheptulose 7-phosphate, succoa succinyl-coenzyme A, TAL transaldolase, TKT1 and TKT2 transketolase, TPI triosephosphate isomerase, Xu5P xylulose 5-phosphate.

Fig. 2. The erythrulose phosphate isomerization module (M2) is active in E. coli.

a A schematic representation of the ΔtktAB selection scheme. Gene deletions are in red. Such a strain cannot synthesize the essential metabolite E4P, highlighted in yellow. Enzymes expressed from a plasmid, pKIID (Supplementary Table 1), are in brown. Carbon sources are shown in purple. D-Eu is D-erythrulose, EltD is erythritol/L-threitol dehydrogenase, FBP is fructose-1,6-bisphosphatase, L-Eu is L-erythrulose, and LerK is L-erythrulose 1-kinase. Other abbreviations are the same as Fig. 1. b L-threitol and erythritol can replace E4P supplements (E4P_suppl., see “Methods”) only when pKIID presents. PC represents positive control, NC represents negative control of no L-threitol or erythritol, NP represents no plasmid control. Source data are provided as a Source Data file.

Fig. 3. In vivo selection of EPS in the formaldehyde assimilation module (M1).

a EPS candidates from E. coli that are known to be DHAP dependent aldolases. Phylogenetic relationship of the proteins is based on their sequence similarity (Supplementary Fig. 4 and “Methods”). FbaB is used as an out group. b A schematic representation of the ΔfrmTKT selection. Gene deletions are in red. Such a strain cannot synthesize the essential metabolite E4P, highlighted in yellow. Enzymes expressed from plasmid are in brown. Carbon sources are shown in purple. SoxA, sarcosine oxidase. Other abbreviations are the same as Fig. 2. c FucA and (d) RhaD, once overexpressed together with other EuMP enzymes, restored growth of the ΔfrmTKT strain in a sarcosine dependent manner. Strains omitted EPS (e) or EryC (f) overexpression failed to grow. Expressed plasmids (detailed in Supplementary Table 1) are shown at the top left corner of the curves. FbaA and YihT failed to support the growth, the results are shown in Supplementary Fig. 5. g Trace of ¹³C labeling from formaldehyde. h Labeling pattern of proteinogenic alanine (Ala), phenylalanine (Phe), and tyrosine (Tyr), within the strains ΔfrmTKT overexpressed FucA (pFEIIO) and RhaD (pREIIO), upon feeding with unlabeled glycerol and sarcosine-(methyl-¹³C). The accompanied results with unlabeled sarcosine are shown in Supplementary Fig. 6a. Source data are provided as a Source Data file.

Fig. 4. ALE enables cooperation of modules M1, M2 and M3.

a A schematic representation of the ΔFBP/GlpX selection. Gene deletions are in red. Such a strain cannot synthesize the essential metabolites E4P, R5P and F6P, highlighted in yellow. Enzymes expressed from plasmid are in brown. Carbon sources are shown in purple. Abbreviations are the same as Fig. 3. b The growth behaviors of the unevolved parent strain and the evolved strains. xyl xylose, sarc sarcosine, PC positive control. c Adaptive laboratory evolution process of ΔFBP/GlpX pFEIIO. Two independent cultures were cultivated at xylose limiting condition and eventually eliminated xylose requirement for growth. d Labeling pattern of proteinogenic alanine (Ala), phenylalanine (Phe), tyrosine (Tyr), histidine (His), methionine (Met) and threonine (Thr), within the evolved strains FUI and FUII, upon feeding with unlabeled glycerol and sarcosine-(methyl-¹³C). The accompanied results with unlabeled sarcosine are shown in Supplementary Fig. 6b. e Mutations emerged from ALE within FUI and FUII. The mutations are detailed in Supplementary Data 1. Source data are provided as a Source Data file.

Fig. 5. Possible histidine labeling traces of FUI and FUII strains.

From ¹³C-labeled formaldehyde and unlabeled DHAP, fructose-1,6-bisphosphatase activity from YggF, YbhA, GmhB, or PhoA generates unlabeled F6P, resulting in histidine (His) unlabeled and once labeled (a). Histidine would be in unlabeled and once labeled when EuMP operates via the sedoheptulose 1,7-bisphosphatase (SBP) route (b). And it would be twice ¹³C-labeled when EuMP operates in the transaldolase (TAL) manner (c).