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

Wu, Tong; Gómez-Coronado, Paul A.; Kubis, Armin; Lindner, Steffen N.; Marlière, Philippe; Erb, Tobias J.; Bar-Even, Arren; He, Hai

doi:10.1038/s41467-023-44247-2

Download PDF

Article
Open access
Published: 20 December 2023

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

Nature Communications volume 14, Article number: 8490 (2023) Cite this article

4617 Accesses
1 Citations
25 Altmetric
Metrics details

Subjects

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.

Growth of E. coli on formate and methanol via the reductive glycine pathway

Article 10 February 2020

Engineering a new-to-nature cascade for phosphate-dependent formate to formaldehyde conversion in vitro and in vivo

Article Open access 09 May 2023

An orthogonal metabolic framework for one-carbon utilization

Article 21 October 2021

Introduction

The biotechnological production of fuels and chemicals currently relies on sugars and biomass-derived carbohydrates as raw materials. However, this practice faces various challenges, including competition with human food production, increasing land use, threatening biodiversity, as well as laborious biomass processing, which limits scalability of this approach^1,2. In contrast, one-carbon (C1) compounds, such as methanol or formate, are favorable alternatives, because they do not compete with food production, are widely available, relatively cheap, and can be produced in a sustainable fashion directly from CO₂^3,4. This makes C1 substrates promising feedstocks for a green and circular bioeconomy and biotechnology^5,6,7,8.

In the assimilation of C1 substrates, formaldehyde plays a central role as bridging metabolite, because it is highly reactive and ready to participate in various biochemical reactions with a diverse set of molecules, such as thiols, amines, pterins, and especially (phospho-)sugars^9,10,11. Nature has developed different metabolic pathways for the assimilation of C1 substrates, which can be generally divided in two parts. First, the C1 substrate (i.e., methanol or formate) is converted into formaldehyde, which is then assimilated through downstream reactions. These pathways include the ribulose monophosphate (RuMP) cycle, the xylulose monophosphate (XuMP) cycle, or the serine cycle, which typically occur in specialized microorganisms, so-called methylotrophs. While there has been some progress in metabolic engineering of natural methylotrophs^12,13,14, the lack of genetic tools is still a major constraint in accelerating metabolic engineering efforts in these organisms to broaden their product spectrum and/or improve their bioproduction capabilities.

The lack of genetic accessibility of methylotrophs has motivated researchers to engineer C1-assimilation pathways in established microbial platform organisms, such as Escherichia coli^6,8,15. One example is the RuMP pathway which has been extensively studied and eventually implemented into E. coli to allow synthetic methylotrophy of this organism^{16,17,18,19,20,21}. Other efforts focused on transplanting engineered (natural) pathways, such as a modified serine cycle²², the homoserine cycle²³, or the reductive glycine pathway²⁴ in E. coli. Moreover, synthetic pathways based on new-to-nature enzymatic activities, such as formolase^25,26 or 2-hydroxyacyl-CoA lyase^27,28, were also designed and tested. Finally, several additional pathways have been identified and proposed through computational efforts²⁹.

In this study, we present a synthetic formaldehyde assimilation pathway, the erythrulose monophosphate (EuMP) cycle. The EuMP cycle is devised following a mix-and-match approach³⁰, relies on a promiscuous aldolase reaction with formaldehyde, and is expected to show the same energetic efficiency as the RuMP cycle. We demonstrate activity of the core reactions of the EuMP cycle in E. coli using a modular and selection-based engineering strategy³¹. Applying adaptive laboratory evolution (ALE) allows us to integrate the core modules of the EuMP cycle with endogenously existing reactions of the pentose phosphate pathway and glycolysis. Whole genome sequencing and reverse-engineering reveal the molecular basis for implementation of the EuMP modules in E. coli. Overall, our study demonstrates the feasibility of a new-to-nature formaldehyde assimilation pathway for a future C1-based biomanufacturing.

Results

Design of the erythrulose monophosphate cycle

The EuMP cycle (Fig. 1a) is centered around a dihydroxyacetone phosphate (DHAP) dependent formaldehyde fixation reaction and connects tetritol catabolism^32,33 with the pentose phosphate pathway. The key reaction, catalyzed by erythrulose 1-phosphate synthase (EPS, or erythrulose 1-phosphate formaldehyde-lyase), condenses formaldehyde with DHAP yielding erythrulose 1-phosphate (Eu1P). Depending on the enzyme stereoselectivity, the product could be L-Eu1P or D-Eu1P, the latter one can be converted by D-erythrulose 1-phosphate 3-epimerase (EryC) into L-Eu1P. Following by sequential isomerization reactions catalyzed by L-erythrulose 1-phosphate isomerase (LerI) and D-erythrulose 4-phosphate isomerase (DerI), L-Eu1P is transformed to D-erythrose 4-phosphate (E4P), which enters the pentose phosphate pathway. With reactions of the pentose phosphate pathway and glycolysis, DHAP can be regenerated, thus closing a cyclic pathway. Overall three turns of the cycle provide one molecule of glyceraldehyde 3-phosphate (GAP) from three molecules formaldehyde.

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

The EuMP cycle can operate in four different architectures. Similar to the natural RuMP pathway⁴, the four different architectures are based on four different possibilities of converting E4P to DHAP and GAP (Supplementary Fig. 1). The conversion of E4P to DHAP and GAP can be generally separated in two stages, i.e. a carbon rearrangement stage and a C6-sugar cleavage stage. The carbon rearrangement stage that converts E4P to fructose 6-phosphate (F6P) comes in two variants: a transaldolase (TAL) or sedoheptulose 1,7-bisphosphatase (SBP) route; in the C6-sugar cleavage stage, F6P is subsequently split into GAP and DHAP, which can also proceed through two different routes, either through fructose 1,6-bisphosphate aldolase (FBA) or the Entner-Doudoroff (ED) route. The four different combinations result in different energetic costs, as indicated in Supplementary Fig. 1. Our study focuses on the TAL/FBA variant of EuMP cycle (Fig. 1a and Supplementary Fig. 1a), which is the most energy efficient design. The in vivo thermodynamic feasibility of this design was confirmed with computational modeling by Max-min Driving Force (MDF)³⁴ (Supplementary Fig. 2).

Compared to the three naturally existing formaldehyde assimilation pathways, the EuMP cycle has the same ATP cost as the RuMP cycle, which is the most efficient natural C1 strategy⁴. For one molecule of GAP, the EuMP cycle and the RuMP cycle consume only one ATP molecule and no reducing power. This is substantially less compared to the XuMP pathway and the serine cycle, which require three molecules of ATP per GAP or acetyl-CoA, respectively. The serine cycle needs additional NAD(P)H when producing GAP (Fig. 1b). We further used flux balance analysis to compare the maximal yields of biomass and the 12 building block precursors in bioproduction³⁵ (see “Methods”). As expected, the EuMP cycle has the same yields as the RuMP cycle in biomass as well as precursors, supporting the highest yields in 11 out of 12 precursors (Fig. 1c).

To demonstrate the activity of the EuMP cycle in vivo, we decided to follow a step-wise, modular engineering approach using the model organism E. coli as host³¹. To that end, we split the EuMP cycle into four modules (M1–M4) (Fig. 1a): the formaldehyde assimilation module (M1) that converts formaldehyde and DHAP to L-Eu1P; the erythrulose phosphate isomerization module (M2) that transforms L-Eu1P to E4P; the carbon rearrangement module (M3) in which the non-oxidative pentose phosphate pathway reactions reconfigure E4P to GAP and F6P; and the C6-sugar cleavage module (M4) that cleaves F6P for DHAP regeneration and GAP as product. As the reactions of the latter two modules are part of canonical central carbon metabolism and present in E. coli, the formaldehyde assimilation and erythrulose phosphate isomerization modules (M1 and M2) are key to successfully establish synthetic EuMP in vivo. Hence, we aimed to reconstruct these modules in dedicated selection strains for demonstrating their activities in vivo.

Validation of the erythrulose phosphate isomerization module (M2) in E. coli

We first aimed at realizing the erythrulose phosphate isomerization module (M2) by employing LerI and DerI that together transform L-Eu1P to E4P. We used a ΔtktAB strain (Fig. 2a and Supplementary Table 1), which is blocked in E4P biosynthesis through the knockout of both transketolase (TKT) genes, tktA and tktB³⁶. To grow on minimal medium, this strain requires E4P supplements (E4P_suppl.: 1 mM shikimic acid, 1 μM pyridoxine, 250 μM tyrosine, 500 μM phenylalanine, 200 μM tryptophan, 6 μM 4-aminobenzoic acid, 6 μM 4-hydroxybenzoic acid and 50 μM 2,3-dihydroxybenzoic acid)¹⁸, because E4P is a precursor of several essential cellular building blocks and cofactors, such as shikimate derived aromatic amino acids and vitamins as well as pyridoxal phosphate. We overexpressed module M2 (i.e., LerI and DerI), together with other tetritol catabolism enzymes, erythritol/L-threitol dehydrogenase (EltD) and L-erythrulose 1-kinase (LerK) from plasmid pKIID in the ΔtktAB strain (Supplementary Table 1). EltD and LerK can convert L-threitol and erythritol, two tetritols, into L-Eu1P and D-erythrulose 4-phosphate, respectively³³. If active, M2 further transforms these products into E4P, which would allow for growth from the two tetritols in the absence of E4P_suppl. in the selection strain (Fig. 2a).

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

Indeed, while both tetritols were not able to directly replace E4P_suppl. to feed the ΔtktAB strain (dashed lines in Fig. 2b), plasmid pKIID enabled growth of the strain on 10 mM L-threitol and erythritol, which were also essential for growth (solid green and pink lines in Fig. 2b). This suggests that L-threitol and erythritol can be converted into E4P with the enzymes from pKIID, indicating that the tetritol catabolism route and particularly the module M2 via LerI and DerI was functional in E. coli. (We note that the lag phase seems to result from tetritol catabolism intermediates, and it appeared only when both the pKIID plasmid and L-threitol or erythritol were presented (Supplementary Fig. 3)).

Establishing erythrulose 1-phosphate synthase (M1) in vivo extends M2

We next aimed at demonstrating the core reaction of EuMP, erythrulose 1-phosphate synthase (EPS) in vivo. This reaction, EC 4.1.2.2, was reported in the 1950s to produce L-erythrulose 1-phosphate (L-Eu1P) by extracts from rat liver³⁷ and Swiss chard leaves³⁸. Later on, this reaction was shown to be catalyzed by class I fructose 1,6-bisphosphate aldolase from rabbit muscle³⁹, Staphylococcus carnosus as well as Staphylococcus aureus^40,41. In addition, other studies reported that L-rhamnulose 1-phosphate aldolase (RhaD) and L-fuculose 1-phosphate aldolase (FucA) from E. coli also showed EPS activity, producing D-erythrulose 1-phosphate (D-Eu1P) from DHAP and formaldehyde^42,43. All these enzymes (i.e., fructose 1,6-bisphosphate aldolase, RhaD and FucA) are DHAP-dependent aldolases and that the reported EPS activity is not their primary function, but rather a side activity. The promiscuity of all these enzymes is in line with the fact that aldolases are generally relatively specific to their donor substrates while being flexible in respect to different aldehyde acceptors⁴⁴.

Apart from RhaD and FucA, E. coli has five more DHAP-dependent aldolases, FbaA, FbaB, GatY, KbaY and YihT. Based on their sequence similarities (Supplementary Fig. 4), we generated a phylogenetic tree to show the sequence-function relationship of all these proteins with each other. As shown in Fig. 3a, the proteins cluster according to their reported functions: FucA and RhaD cluster together and have the same acceptor (S)-lactaldehyde; FbaA, GatY and KbaY accept D-glyceraldehyde 3-phosphate as substrate and derive from a common branch, with FbaA forming the (3 S, 4 R) product and GatY and KbaY forming the (3 S, 4 S) product; YihT and FbaB, finally, are class I aldolases that are distant from all the other enzymes, which are class II aldolases. Even though the different aldolases possess different stereochemical properties (Fig. 3a), the difference at C₄ becomes irrelevant when using formaldehyde as acceptor, because the product Eu1P does not contain a chiral center at C₄. In terms of stereochemistry at the C₃, RhaD and FucA are reported to produce the 3 R product, which would correspond to D-Eu1P when using formaldehyde as substrate, while the other five are reported to produce the 3 S product, which would correspond to L-Eu1P. Since GatY and KbaY are more complex enzymes and require additional, non-catalytic subunits (GatZ and KbaZ, respectively), for full activities and stability⁴⁵, we excluded those candidates for our implementation efforts. We also ruled out class I type FbaB and just kept its class II paralog FbaA, which is generally more stable⁴⁶. Overall, we decided to use E. coli’s RhaD, FucA, FbaA, and YihT as candidates to realize the EPS reaction in vivo.

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

To test the four EPS candidates in vivo, we again used the ΔtktAB strain, which we amended by an additional knockout of the frmRAB operon (ΔfrmTKT, Fig. 3b and Supplementary Table 1). FrmA and FrmB are responsible for the glutathione-dependent oxidation of formaldehyde to formate. The ΔfrmRAB knockout abolishes this route, strongly increasing intracellular formaldehyde concentrations¹⁸. We overexpressed each of the four different EPS candidates together with genes lerI, derI, eryC, and soxA (Supplementary Table 1). LerI and DerI are for the module M2 that we had demonstrated earlier. EryC (D-erythrulose 1-phosphate 3-epimerase) is necessary for RhaD and FucA to convert D-Eu1P into L-Eu1P. Although EryC should not be essential when using FbaA and YihT, we still used it in these constructs, reasoning that the EryC reaction is fully reversible, Δ_rG’ = 0 kJ/mol⁴⁷. Sarcosine oxidase (SoxA), finally, was used for providing formaldehyde via a reliable method from sarcosine in situ^18,48. To focus on the demonstration of the EuMP pathway, we used this established formaldehyde production modular to set aside the inefficient production of formaldehyde from C1 substrates in this study. The overall selection scheme is shown in Fig. 3b. The ΔfrmTKT strain can grow only when M1 (i.e., the different EPS candidates and EryC in case of RhaD and FucA) is active and able to condense sarcosine-derived formaldehyde with DHAP to L-Eu1P. L-Eu1P is further processed downstream by M2 to provide E4P to the selection strain.

Strikingly, when testing the different candidates, only FucA and RhaD (overexpressed from plasmids pFEIIO and pREIIO, respectively, Fig. 3c, d), but neither FbaA nor YihT (Supplementary Fig. 5), enabled growth of the ΔfrmTKT strain on sarcosine as formaldehyde source. In these strains, growth rates and maximal ODs directly correlated with increasing sarcosine concentrations (Fig. 3c, d) and growth could not be restored without EPS (i.e., fucA or rhaD) overexpression (Fig. 3e). Moreover, omitting EryC resulted in no growth (Fig. 3f), supporting the hypothesis that FucA and RhaD yield D-Eu1P and require EryC-catalyzed D-Eu1P - L-Eu1P isomerization for formaldehyde assimilation. We attributed the small growth under sarcosine absent condition (red lines in Fig. 3c and d) to a non-depletable internal source of formaldehyde^9,11, because such growth was FucA/RhaD and EryC dependent (Fig. 3e, f and Supplementary Fig. 5). In contrast to FucA overexpression, overexpression of RhaD did not allow growth at elevated sarcosine concentrations (i.e., > 5 mM), indicating that FucA removes/assimilate formaldehyde more efficiently in vivo, concerning the toxicity of formaldehyde.

To further confirm that E4P was produced through condensation of formaldehyde with DHAP via EPS, i.e., module M1, and module M2 of the EuMP cycle, we performed ¹³C-labeling trace experiments (Fig. 3g, h and Supplementary Fig. 6a). When cultivating the FucA and RhaD overexpression strains with unlabeled glycerol and N-methyl-¹³C labeled sarcosine, we observed single-labeled tyrosine and phenylalanine. (The small fractions of double labeling were results of natural abundance of ¹³C (Supplementary Fig. 6a) and the relatively high unlabeled fractions, especially for FucA, were likely from internal source formaldehyde originated from unlabeled glycerol.) The labeling pattern of tyrosine and phenylalanine is in line with the condensation of non-labeled phosphoenolpyruvate (PEP) with single-labeled E4P (derived from DHAP and ¹³C-formaldehyde) during aromatic amino acids biosynthesis. This hypothesis was further confirmed by analysis of the labeling pattern of alanine, which is directly derived from PEP and was indeed unlabeled in these strains. We note that TAL transfers carbons between F6P and GAP, resulting in unlabeled GAP and subsequently PEP. Yet the flux of TAL is net 0 (calculated by flux balance analysis, Supplementary Fig. 7a). Overall, these experiments confirmed that the EuMP core modules M1 and M2 were successfully realized in vivo.

Adaptive laboratory evolution integrates carbon rearrangement module M3

We finally sought to further integrate the EuMP cycle by connecting the two active core modules (M1 and M2) with the naturally existing carbon rearrangement module (M3) in E. coli. To that end, we constructed a ΔFBP/GlpX strain in which the fructose 1,6-bisphosphatases of gluconeogenesis were knocked out, Δfbp ΔglpX (Fig. 4a and Supplementary Table 1). In this strain, glycerol cannot be converted to F6P, and therefore the strain cannot use glycerol alone to produce E4P, ribose 5-phosphate (R5P) and glucose 6-phosphate (G6P)¹⁸. Because these metabolites are essential, this strain becomes dependent on the activities of modules M1, M2 and M3 for growth. Although the enzymes of module M4, the C6-sugar cleavage module, are present, the strain does not rely on it for growth, as shown in Supplementary Fig. 7b as an exemplified flux distribution calculated with flux balance analysis (see Methods). We further deleted the mgsA gene, which encodes methylglyoxal synthase, to eliminate a potential sink lowering DHAP pool⁴⁹.

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

When we transformed the ΔFBP/GlpX strain with plasmids pFEIIO or pREIIO, respectively (Supplementary Table 1), we did not observe growth under selective conditions (i.e., on glycerol and sarcosine). Only when xylose was additionally supplemented, the strains grew (gray and orange lines in Fig. 4b and Supplementary Fig. 8). We reasoned that this was caused by the fact that (i) the ΔFBP/GlpX selection (in total 17% divided from F6P devotes 1% carbons in biomass, G6P 3%, R5P 10% and E4P 3%⁵⁰) requires 6 times higher flux via modules M1 + M2 than the ΔfrmTKT selection (3%), and that (ii) the unique entry of E4P into the pentose phosphate pathway required E. coli to operate the pentose phosphate pathway in an unconventional manner.

Therefore, we started an adaptive laboratory evolution (ALE) experiment to alleviate and remove xylose-dependency of theses strains. At the start of the ALE experiment, we determined that 1 mM xylose was growth-limiting, restricting growth to an OD₆₀₀ around 0.4 (orange line in Fig. 4b and Supplementary Fig. 8). From the same parent ΔFBP/GlpX strain containing plasmid pFEIIO, we started two independent cultures simultaneously with 1 mM xylose in the medium (FUI/FUII). We re-inoculated the cultures into fresh medium after a few days of cultivation at a starting OD₆₀₀ of 0.01. Figure 4c shows the ALE process. After five passages, both cultures, FUI and FUII, were able to reach higher OD₆₀₀ and became able to grow on medium without xylose. We isolated five colonies for further characterization (Supplementary Table 1).

The five isolates were indeed confirmed to grow on xylose-free conditions. We observed identical growth between the three colonies of FUI and between the two colonies of FUII (Supplementary Table 1). While the FUI strains interestingly showed some sarcosine-independent growth (red line in Fig. 4b), sarcosine addition clearly increased growth (Fig. 4b). Overall, these results indicated that evolution had enabled the EuMP to sustain growth under the higher selection regime of ΔFBP/GlpX, and more importantly, the successful cooperation of the modules M1 to M3.

We further subjected the strains for ¹³C labeling experiments and analyzed the ¹³C labeling pattern of alanine, phenylalanine, tyrosine, histidine, methionine and threonine. Upon feeding unlabeled glycerol and N-methyl-¹³C labeled sarcosine, the labeling patterns of alanine, phenylalanine and tyrosine were similar to the ΔfrmTKT strains (see above), non-label in alanine and single-label in phenylalanine and tyrosine (Fig. 4d). However, the FUI strain showed a relatively high fraction of non-labeled aromatic amino acids that was different to the pattern of the FUII strain and higher than expected from natural isotopic distribution (Fig. 4d and Supplementary Fig. 6b). Similar to alanine, threonine and methionine were completely unlabeled, suggesting that label was neither introduced through aspartate (threonine and methionine backbone), nor C1-tetrahydrofolate (methionine methyl group). Finally, as observed for the aromatic amino acids, the labeling pattern of histidine differed between FUI and FUII, indicating some more complex metabolic scenarios. Only 19% of the histidine was single-labeled and 7% was double-labeled in FUI, while 26% and 38% single- and double-label, respectively, were observed in the FUII strain (Fig. 4d). This data confirmed that the EuMP was active in FUI and FUII, but also suggested distinct metabolic adaptations between the two different evolutionary lines.

Mutation characterizations of evolved strains

To analyze the molecular basis of our ALE efforts and understand the differences between the two independent evolutionary lines, we sequenced whole genomes (including plasmids) of the evolved FUI and FUII strains via Illumina sequencing. While the different isolates within the FUI or FUII lines were genetically almost identical (Supplementary Data 1), the FUI and FUII lines accumulated distinct mutations. As shown in Fig. 4e and detailed in Supplementary Data 1, there was only one common mutation in both FUI and FUII strains, a nucleotide transversion in the promoter region of psuK gene. The psuK gene encodes pseudouridine kinase for the degradation of pseudouridine, which is the most abundant modified nucleoside in RNAs⁵¹. This promoter mutation was most likely involved in adapting the ΔFBP/GlpX strain via regulating psuK expression. FUI carried additional mutations, including (1) large deletions of the prophage e14 region and a 49 genes region ranging from rfbD to gatR; (2) frameshifts in genes of allE, ycbM, fabH, and galU; as well as (3) point mutations in promoter regions of gmhB and alsR. In the FUII strain, only two more missense mutations, one in tpiA and one in lerI, could be identified.

These results suggested that the FUI and FUII strains evolved different metabolic solutions under the same ΔFBP/GlpX selection regime, which is in line with the distinct tyrosine, phenylalanine and histidine labeling patterns between FUI and FUII. In FUI, the colanic acid (CA) cluster genes within the rfbD-garR region and galU, gmhB as well as fabH are involved in biosynthesis of nonessential extracellular polysaccharide and lipopolysaccharides (LPS) of the cell outer membrane^52,53,54. Since F6P and G6P are precursors of CA and LPS^50,53, abovementioned null mutations relieved the requirements of F6P and G6P, thus lowering the selection pressure of the ΔFBP/GlpX strain. Note that Fbp and GlpX are the two main proteins carrying fructose-1,6-bisphosphatase (FBP) activity in E. coli, evidenced by that the ΔFBP/GlpX strain required xylose for growth (Fig. 4b and Supplementary Fig. 8). However, E. coli has additional enzymes with FBP activity: a third fructose-1,6-bisphosphatase YggF⁵⁵, haloacid dehalogenase (HAD)-like hydrolases YbhA and GmhB^54,56 as well as alkaline phosphatase (PhoA)⁵⁷. These enzymes are able to substitute the conversion of fructose-1,6-bisphosphate (FDP) to F6P.

We speculated that in the evolved FUI strain, with lowered F6P requirements, these enzymes were able to produce sufficient F6P, which was further transformed via the reversible TKT2 reaction into E4P and Xu5P. Such internal E4P source is in line with the slight growth of FUI in the absence of xylose and sarcosine, as well as the better growth of FUI than FUII (Fig. 4b). Notably, it also explains the ¹³C-labeling pattern in FUI, as the internally produced E4P is unlabeled, diluting the EuMP derived ¹³C-labeled E4P in phenylalanine and tyrosine, as well as in histidine in FUI. For histidine, which was only 19% labeled once and 7% labeled twice, the net direction of TKT2, RPI and RPE reactions operated in reverse, so that a major portion of R5P was from Xu5P and a small portion was from TAL and TKT1 reactions (Fig. 5a). The observed mutation in the alsR promoter may have supported this scenario by altering the expression of rpiB. AlsR is a transcriptional repressor of rpiB⁵⁸, the encoded RpiB is the second ribose-5-phosphate isomerase (RPI) and potentially has DerI activity³².

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

In FUII, mutated proteins TpiA and LerI are both EuMP pathway enzymes. These proteins are homologs and may possess activity of each other³². Thus, these mutations may have improved the EuMP flux in FUII. Note that in contrast to FUI, cell envelope biosynthesis was still intact in FUII, so that the F6P and G6P requirements were still high. This apparently precluded reverse operation of TKT2, as evidenced by classical labeling of phenylalanine and tyrosine. Regarding the labeling of histidine, we speculate that both the SBP and the TAL variants operated in FUII (Supplementary Fig. 1a and b). Although the strain does not contain GlpX, an identified sedoheptulose 1,7-bisphosphatase²¹, other phosphatases may possess such activity, similar to the aforementioned FBP. Through the SBP variant (Fig. 5b), unlabeled and single-labeled histidine were derived from R5P and Xu5P, while the double labeled histidine originated from the TAL variant (Fig. 5c).

Indeed, the importance of both LerI and TpiA mutations was confirmed. Plasmid curing then re-transforming the plasmid pFEIIO (with naïve LerI, Supplementary Table 1) abolished growth of FUII strain (Supplementary Fig. 9a). FUII strain reverted the tpiA mutation to wild-type (FUII-tpiA_wt, Supplementary Table 1) did not grow in the presence of the plasmid with a mutated LerI (pFEIIOe, Supplementary Table 1 and Supplementary Fig. 9b). Although introducing a mutation of either LerI or TpiA alone was not sufficient to enable the growth of the parental ΔFBP/GlpX strain, together, they allowed growth of the strain (Supplementary Fig. 9). These results indicate synergy between mutations in LerI and TpiA. Finally, considering the regulation role of PsuK (as mentioned above), the extended lag phase observed in the reverse-engineered ΔFBP/GlpX-tpiA_m strain with pFEIIOe, compared to the FUII strain, may suggest an additional layer of synergy with psuK promoter mutation (Supplementary Fig. 9b and Fig. 4b).

Discussion

One-carbon compounds play a central role in a circular bioeconomy. Formaldehyde is an important intermediate in several natural and synthetic C1-assimilation pathways. In this study, we present a new-to-nature formaldehyde assimilation pathway, the EuMP cycle, following a mix-and-match engineering strategy similar to the homoserine cycle²³. Our theoretical calculations highlight that the EuMP cycle is equally efficient as the most energy efficient natural RuMP cycle, outperforming other natural C1-assimilation pathways. Although the EuMP cycle requires more enzymes than the RuMP cycle, it is hard to compare the enzyme costs of both pathways because of the difficulties of obtaining full kinetics of the core enzymes. Still, the EuMP cycle serves as a complementation to the synthetic homoserine cycle, which has relatively low theoretical yields in biomass and sugar phosphates²³. EuMP can operate in four different modes, offering metabolic flexibility and potentially enhancing adaptability to various requirements. Moreover, the EuMP cycle holds the potential to be further extended by phosphoketolase reaction to further improve its efficiency, as phosphoketolase produces C2 moieties from sugar phosphates without oxidation (releasing CO₂)⁵⁹.

In designated E. coli selection strains, we demonstrated the activity of the core modules of the EuMP cycle. While we established EPS activity through FucA and RhaD from E. coli, enzyme engineering or screening of additional candidates, such as those from Propionibacterium pentosaceum⁶⁰ and Swiss chard leaves³⁸, might provide even more specific and/or active enzymes in the future. An L-Eu1P producing EPS would be especially interesting, as it omits the necessity of EryC overexpression. Such enzymes could also benefit other applications, for instance rare sugars production⁶¹.

Similarly, better enzymes for the isomerases, i.e., EryC, LerI and DerI, can also be identified and substituted. Microorganisms catabolizing tetritols offers a broad mining source. It was suggested that EPS in P. pentosaceum⁶⁰ and Sinorhizobium meliloti⁶² functions in the Eu1P cleavage direction for erythritol catabolism. Further study on the co-occurrence of EPS and these isomerases (as well as the other tetritol catabolism enzymes) is particularly interesting, because it could not only elucidate their physiological functions from evolutionary and ecological aspects, but may also find that the EuMP cycle is already a naturally existing C1-assimilation pathway, similar to the reductive glycine pathway⁶³.

The EuMP pathway assimilates formaldehyde with DHAP and enters central metabolism on the level of E4P. This unique feature of the pathway makes ALE necessary in our (and potentially also future) engineering efforts. Note that the pentose phosphate pathway operates usually from C5 or C6 sugars. E4P is typically “only” an intermediate and a branching point into different biosynthetic pathways, thus rather serving as sink than as source in the classical pentose phosphate pathway. Entering from E4P required rewiring the fluxes of the highly regulated pentose phosphate pathway. In ΔFBP/GlpX with increased selection pressure, ALE was used to establish EuMP dependent growth for the FucA-based overexpression strain. In the FUI strain nutrient requirements from EuMP were lowered by abolishing CA and LPS biosynthesis (a strategy is also used for cell factory optimization⁵³), while in the FUII strain EuMP pathway activity was tuned via mutations in the pathway enzymes TpiA and LerI in a synergistic manner. In summary, two independent evolutionary lines allowed us to establish the first three modules of the EuMP cycle.

The in vivo establishment of the full EuMP cycle relying on formaldehyde solely will require more metabolic engineering and long-term ALE. Module M4 should be active itself as it is part of glycolysis and operates in the same direction as when feeding with glucose; integrating it with the other three modules is still not an easy task. The ΔFBP/GlpX selection strain would still be a good evolution platform, because it contains all four EuMP modules while relying on modules M1, M2 and M3 to grow. Mutations identified from FUI and FUII could be established in the strain. Blocking glycolysis would be beneficial, as it lowers the initial selection pressure⁶⁴ and avoids interference from sarcosine-derived glycine. Such a strain would be subjected to further ALE with decreasing glycerol and increasing sarcosine concentrations. Similar approaches have been used in this study and been applied previously in successfully installing the RuMP cycle^17,19 and the Calvin–Benson–Bassham cycle^64,65 heterologously in E. coli. Although we observed partial bypass in the FUI ALE, similar occurrences restoring growth are unlikely in such a full cycle selection condition which has a selection pressure similar to the abovementioned examples. In silico modeling, such as flux balance analysis¹⁹ and ensemble modeling for robustness analysis¹⁷, may also be conducted to identify potential bottlenecks for rational strain engineering. These strategies could enable the growth of E. coli via the full EuMP cycle first on sarcosine-derived formaldehyde, decoupling the challenge of formaldehyde production from methanol, formate, or other C1 feedstocks, paving the way for the future development of growth solely on C1 feedstocks. While achieving the goal is undoubtedly challenging, it holds the promise as an efficient C1-platform in bioindustry.

Methods

Max-min driving force analysis

Max-min driving force (MDF) analysis³⁴ was applied to evaluate the thermodynamic feasibility of the EuMP cycle for glyceraldehyde 3-phosphate production from formaldehyde. Python packages equilibrator_api (v 0.4.7) and equilibrator_pathway (v 0.4.7) were used for the calculations. The changes in Gibbs energy of the reactions were estimated using the component contribution method⁶⁶. Cofactor and metabolite concentrations were constrained to the range 1 µM to 10 mM³⁴ with a change for formaldehyde upper bound to 0.5 mM¹⁸. pH was assumed to be 7.5, ionic strength was assumed to be 0.25 M, and –log[Mg²⁺] (pMg) was assumed to be 3. The scripts and details are available in the following open-access archive repository: https://github.com/he-hai/PubSuppl, within “2022_EuMP” the directory.

Flux balance analysis

Flux balance analysis modeling was conducted with COBRApy using the most updated E. coli genome-scale metabolic model iML1515⁶⁷. The model was modified as follows: (i) transhydrogenase (THD2pp) translocates one proton instead of two⁶⁸; (ii) homoserine dehydrogenase (HSDy) was set to irreversibly produce homoserine²³; (iii) anaerobic relevant reactions, PFL, OBTFL, FDR2, and FDR3, were removed from the model; (iv) POR5, GLYCK, FDH4pp FDH5pp, GART, DRPA, and PAI2Twere also knocked out to block unrealistic routes; (v) the ATP maintenance reaction (ATPM) was also switched off.

To compare the yields of the 12 building block precursors³⁵ and biomass from formaldehyde assimilation via EuMP and the three natural pathways, the models of the four pathways were integrated in the modified E. coli model. The maximum yields were estimated using formaldehyde as feedstock.

To showcase the flux distributions of the selection strains, gene knock-outs were stimulated within the EuMP E. coli model. Bounds of the biomass reaction were fixed to be 1 and the objective function was changed to minimal the EPS flux.

Strains and genomic manipulation

All strains used in the study are listed in Supplementary Table 1. An E. coli K-12 MG1655 derived SIJ488 strain⁶⁹ was used as host platform. Gene knock-outs were obtained by λ-Red recombination or P1 transduction³¹. Recombineering knock-out cassettes were generated by PCR amplifying the FRT-PGK-gb2-neo-FRT (Km) cassette (Gene Bridges, Germany) using 50 bp homologous arms from the target genes in the primers (Supplementary Data 2). About 300 ng of cassette DNA was transformed into freshly prepared electrocompetent cells, which were induced 45 min prior with 15 mM L-arabinose when OD₆₀₀ was around 0.3. The knock out colony was selected on LB kanamycin plates and verified by colony PCR with “Ver” primers (Supplementary Data 2). Flippase was induced by 50 mM L-rhamnose to remove antibiotic markers. And its removal was confirmed by colony PCR.

CRISPR/Cas9 system was used to reengineer the mutation (G94D) of TpiA into the parent strain (ΔFBP/GlpX, Supplementary Table 1) following the Scar-less method⁷⁰ using a donor oligo (tpiA_m, Supplementary Data 2). The mutation of tpiA in resulting strain, ΔFBP/GlpX-tpiA_m (Supplementary Table 1), was confirmed by Sanger sequencing. To revert the mutation of TpiA in FUII, a chloramphenicol selection marker (Cm) was first inserted downstream of the tpiA gene within the parent strain (ΔFBP/GlpX, Supplementary Table 1) using the λ-Red system mentioned above. P1 phage transduction⁷¹ was then conducted transferring the wild type tpiA to FUII2 strain using the resulting strain (tpiA∷Cm, Supplementary Table 1) as donor. Sequencing confirmed the reversion of tpiA gene.

Plasmid construction

All plasmids are listed in Supplementary Table 1 and their constructions were followed the method in ref. ³¹. Heterologous gene sequences were listed in Supplementary Data 3. fucA, rhaD, fbaA and yihT were cloned from E. coli MG1655 genome with primers listed in Supplementary Data 2. All genes were inserted into pNivC plasmid individually under a ribosome binding site “C” (AAGTTAAGAGGCAAGA). Apart from lerK and eltD, all others have a 6xHis-tag right after the start codon. Genes were then assembled into an operon via BioBrick enzymes: BcuI, SalI, NheI and XhoI (FastDigest, Thermo Scientific). EcoRI and PstI (FastDigest, Thermo Scientific) were used to move the whole operon to an expression vector (pZ plasmid), which has a synthetic promoter pgi-20 or pgi-10, p15A medium copy origin as well as streptomycin selection marker.

Growth media and growth experiments

LB (1% NaCl, 0.5% yeast extract, 1% tryptone) was used in cultivating cells for all gene manipulation e.g., gene cloning and gene knocking out. Antibiotics were used at concentrations of 50 μg/mL kanamycin, 100 μg/mL ampicillin 100 μg/mL streptomycin, or 30 μg/mL chloramphenicol to the medium if needed. M9 minimal medium (Sigma-Aldrich, Germany) supplemented with 2 mM MgSO₄, 100 μM CaCl₂ and trace elements (134 μM EDTA, 31 μM FeCl₃, 6.2 μM ZnCl₂, 0.76 μM CuCl₂, 0.42 μM CoCl₂, 1.62 μM H₃BO₃, 0.081 μM MnCl₂) was used for growth experiments. Carbon sources used were indicated in the main text. For ΔtktAB and ΔfrmTKT strains, E4P supplements (E4P_suppl.) were also added in relaxing medium at final concentration of: 1 mM shikimic acid, 1 μM pyridoxine, 250 μM tyrosine, 500 μM phenylalanine, 200 μM tryptophan, 6 μM 4-aminobenzoic acid, 6 μM 4-hydroxybenzoic acid and 50 μM 2,3-dihydroxybenzoic acid¹⁸.

Growth experiments started with preculturing the cells in 4 mL M9 medium in glass tubes under relaxing conditions. When reaching late exponential phase, the precultures were washed three times with sterilized Mili-Q water by centrifugation and resuspension. Cells were inoculated in dedicated testing conditions in 96-well plates (Nunclon Delta Surface, Thermo Scientific) at starting OD₆₀₀ of 0.01 cultured at 37 °C in microplate reader (BioTek EPOCH 2). Each well contained 150 μL culture and covered with 50 μL sterilized mineral oil (Sigma-Aldrich, Germany). The shaking program cycle (controlled by Gen5 v3) has 4 shaking phases, lasting 60 s each: linear shaking followed by orbital shaking, both at an amplitude of 3 mm, then linear shaking followed by orbital shaking both at an amplitude of 2 mm. The optical density (OD₆₀₀) in each well was monitored and recorded after every three shaking cycles (~16.5 min). Raw data from the plate reader were calibrated to normal cuvette measured OD₆₀₀ values according to Eq. 1.

$${{{{{{\rm{OD}}}}}}}_{{{{{{\rm{cuvette}}}}}}}={{{{{{\rm{OD}}}}}}}_{{{{{{\rm{plate}}}}}}}/0.23$$

(1)

MATLAB (MathWorks) was used for calculating growth parameters based on three technical triplicates – the average values were used to generate the growth curves. Checked in MATLAB, in all cases variability between triplicates measurements were less than 5%.

Molecular phylogenetic analysis

DHAP aldolase sequences from E. coli, FucA P0AB87, RhaD P32169, YihT P32141, FbaA P0AB71, GatY P0C8J6, KbaY P0AB74 and FbaB P0A991, were obtained from UniProt. MAFFT v7.490 was used for multiple sequence alignment with default parameters. MEGA 10.0.5 used the aligned sequences to construct a Neighbor-joining phylogenetic tree, with 1000 Bootstrap replication tests.

Stable isotopic labeling

Sarcosine-(methyl-¹³C) was purchased from Sigma-Aldrich. The strains were cultivated in glass tubes on M9 minimal media with appropriate carbon sources. Cells were harvested at the late exponential phase. The equivalent volume of 1 mL of culture at OD₆₀₀ of 1 was harvested and washed by centrifugation. Experiments were performed with replicates which in all cases showed identical results (± 5%). Proteinogenic amino acids were hydrolyzed from biomass with 6 M HCl at 95 °C for 24 h⁷². The samples were completely dried under a stream of N₂ at 95 °C. Hydrolyzed amino acids were analyzed with UPLC–ESI–MS with a Waters Acquity UPLC system (Waters), using an HSS T3 C₁₈ reversed phase column (100 mm × 2.1 mm, 1.8 μm; Waters)²³. 0.1% formic acid in H₂O (A) and 0.1% formic acid in acetonitrile (B) were the mobile phases. The flow rate was 0.4 mL/min and the gradient was: 0 to 1 min – 99% A; 1 to 5 min – linear gradient from 99% A to 82%; 5 to 6 min – linear gradient from 82% A to 1% A; 6 to 8 min – kept at 1% A; 8-8.5 min—linear gradient to 99% A; 8.5-11 min – re-equilibrate. Mass spectra were acquired using an Exactive mass spectrometer (Thermo Scientific) in positive ionization mode, with a scan range of 50.0 to 300.0 m/z. The spectra were recorded during the first 5 min of the LC gradients. Data analysis was performed using Xcalibur (Thermo Scientific). Determination of retention times was performed by analyzing amino-acid standards (Sigma-Aldrich) under the same conditions.

Adaptive laboratory evolution

To evolve ΔFBP/GlpX pFEIIO strains (Supplementary Table 1) for formaldehyde dependent growth, xylose concentration was tested firstly. The resulting limiting condition of 1 mM xylose was used as a starting point. The strains were first inoculated from cyro-glycerol stocks onto LB streptomycin plates, then inoculated to M9 liquid medium with 10 mM glycerol, 1 mM xylose and 1 mM sarcosine in two glass tubes of each strain. The two independent cultures were subcultured to fresh media at initial OD₆₀₀ of 0.01. After seven passages, both pFEIIO cultures were then cultivated at no xylose conditions for another two passages. The cultures were then streaked out onto agar plates and single colonies were isolated (Supplementary Table 1).

Whole genome sequencing and analysis

NucleoSpin Microbial DNA kit (MACHERY-NAGEL, Düren, Germany) was used for genomic DNA extraction. Library construction and genome sequencing were performed by Novogene (Cambridge, United Kingdom) using the paired-end Illumina sequencing platform. Analysis of the sequencing data was carried out using breseq (v0.31.0)⁷³ with MG1655 (GenBank: NC_000913) and pFEIIO as references.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Data availability

Raw reads of NGS are deposited at NCBI and can be accessed under BioProject PRJNA895983. MG1655 genome is available at Genbank under accession NC_000913. Protein sequences of DHAP aldolases from E. coli are obtained from UniProt: FucA P0AB87, RhaD P32169, YihT P32141, FbaA P0AB71, GatY P0C8J6, KbaY P0AB74, and FbaB P0A991. Source data are provided with this paper.

Code availability

Full code and models of MDF and flux balance analysis are available at Github [https://github.com/he-hai/PubSuppl] within the “2022_EuMP” directory. They can also be found under https://doi.org/10.5281/zenodo.10207936⁷⁴.

References

Liao, J. C., Mi, L., Pontrelli, S. & Luo, S. Fuelling the future: microbial engineering for the production of sustainable biofuels. Nat. Rev. Microbiol. 14, 288–304 (2016).
Article CAS PubMed Google Scholar
Sanderson, K. Lignocellulose: A chewy problem. Nature 474, S12–S14 (2011).
Article CAS PubMed Google Scholar
Clomburg, J. M., Crumbley, A. M. & Gonzalez, R. Industrial biomanufacturing: The future of chemical production. Science 355, aag0804 (2017).
Article PubMed Google Scholar
Cotton, C. A. R., Claassens, N. J., Benito-Vaquerizo, S. & Bar-Even, A. Renewable methanol and formate as microbial feedstocks. Curr. Opin. Biotechnol. 62, 168–180 (2020).
Article CAS PubMed Google Scholar
Gregory, G. J., Bennett, R. K. & Papoutsakis, E. T. Recent advances toward the bioconversion of methane and methanol in synthetic methylotrophs. Metab. Eng. 71, 99–116 (2022).
Article CAS PubMed Google Scholar
Jiang, W. et al. Metabolic engineering strategies to enable microbial utilization of C1 feedstocks. Nat. Chem. Biol. 17, 845–855 (2021).
Article CAS PubMed Google Scholar
Peiro, C., Vicente, C. M., Jallet, D. & Heux, S. From a hetero- to a methylotrophic lifestyle: Flash back on the engineering strategies to create synthetic methanol-user strains. Front. Bioeng. Biotechnol. 10, 907861 (2022).
Article PubMed PubMed Central Google Scholar
Wang, Y., Fan, L., Tuyishime, P., Zheng, P. & Sun, J. Synthetic methylotrophy: A practical solution for methanol-based biomanufacturing. Trends Biotechnol. 38, 650–666 (2020).
Article CAS PubMed Google Scholar
Klein, V. J., Irla, M., Gil López, M., Brautaset, T. & Fernandes Brito, L. Unravelling formaldehyde metabolism in bacteria: Road towards synthetic methylotrophy. Microorganisms 10, 220 (2022).
Article CAS PubMed PubMed Central Google Scholar
Yurimoto, H., Kato, N. & Sakai, Y. Assimilation, dissimilation, and detoxification of formaldehyde, a central metabolic intermediate of methylotrophic metabolism. Chem. Rec. 5, 367–375 (2005).
Article CAS PubMed Google Scholar
Chen, N. H., Djoko, K. Y., Veyrier, F. J. & McEwan, A. G. Formaldehyde stress responses in bacterial pathogens. Front. Microbiol. 7, 257 (2016).
CAS PubMed PubMed Central Google Scholar
Chen, A. Y. & Lan, E. I. Chemical production from methanol using natural and synthetic methylotrophs. Biotechnol. J. 15, e1900356 (2020).
Article PubMed Google Scholar
Gao, J., Li, Y., Yu, W. & Zhou, Y. J. Rescuing yeast from cell death enables overproduction of fatty acids from sole methanol. Nat. Metab. 4, 932–943 (2022).
Article CAS PubMed Google Scholar
Quynh Le, H. T., Anh Mai, D. H., Na, J.-G. & Lee, E. Y. Development of Methylorubrum extorquens AM1 as a promising platform strain for enhanced violacein production from co-utilization of methanol and acetate. Metab. Eng. 72, 150–160 (2022).
Article CAS PubMed Google Scholar
Pfeifenschneider, J., Brautaset, T. & Wendisch, V. F. Methanol as carbon substrate in the bio‐economy: Metabolic engineering of aerobic methylotrophic bacteria for production of value‐added chemicals. Biofuels Bioprod. Bioref. 11, 719–731 (2017).
Article CAS Google Scholar
Bennett, R. K. et al. Engineering Escherichia coli for methanol-dependent growth on glucose for metabolite production. Metab. Eng. S1096717620300525 https://doi.org/10.1016/j.ymben.2020.03.003 (2020).
Chen, F. Y.-H., Jung, H.-W., Tsuei, C.-Y. & Liao, J. C. Converting Escherichia coli to a synthetic methylotroph growing solely on methanol. Cell 182, 933–946.e14 (2020).
Article CAS PubMed Google Scholar
He, H., Edlich-Muth, C., Lindner, S. N. & Bar-Even, A. Ribulose monophosphate shunt provides nearly all biomass and energy required for growth of E. coli. ACS Synth. Biol. 7, 1601–1611 (2018).
Article CAS PubMed Google Scholar
Keller, P. et al. Generation of an Escherichia coli strain growing on methanol via the ribulose monophosphate cycle. Nat. Commun. 13, 5243 (2022).
Article ADS CAS PubMed PubMed Central Google Scholar
Tuyishime, P. et al. Engineering Corynebacterium glutamicum for methanol-dependent growth and glutamate production. Metab. Eng. 49, 220–231 (2018).
Article CAS PubMed Google Scholar
Woolston, B. M., King, J. R., Reiter, M., Van Hove, B. & Stephanopoulos, G. Improving formaldehyde consumption drives methanol assimilation in engineered E. coli. Nat. Commun. 9, 2387 (2018).
Article ADS PubMed PubMed Central Google Scholar
Yu, H. & Liao, J. C. A modified serine cycle in Escherichia coli coverts methanol and CO₂ to two-carbon compounds. Nat. Commun. 9, 3992 (2018).
Article ADS PubMed PubMed Central Google Scholar
He, H., Höper, R., Dodenhöft, M., Marlière, P. & Bar-Even, A. An optimized methanol assimilation pathway relying on promiscuous formaldehyde-condensing aldolases in E. coli. Metab. Eng. 60, 1–13 (2020).
Article CAS PubMed Google Scholar
Kim, S. et al. Growth of E. coli on formate and methanol via the reductive glycine pathway. Nat. Chem. Biol. 16, 538–545 (2020).
Article CAS PubMed Google Scholar
Lu, X. et al. Constructing a synthetic pathway for acetyl-coenzyme A from one-carbon through enzyme design. Nat. Commun. 10, 1378 (2019).
Article ADS PubMed PubMed Central Google Scholar
Wang, X. L. et al. Biological conversion of methanol by evolved Escherichia coli carrying a linear methanol assimilation pathway. Bioresour. Bioprocess. 4, 41 (2017).
Article Google Scholar
Chou, A., Lee, S. H., Zhu, F., Clomburg, J. M. & Gonzalez, R. An orthogonal metabolic framework for one-carbon utilization. Nat. Metab. 3, 1385–1399 (2021).
Article CAS PubMed Google Scholar
Chou, A., Clomburg, J. M., Qian, S. & Gonzalez, R. 2-Hydroxyacyl-CoA lyase catalyzes acyloin condensation for one-carbon bioconversion. Nat. Chem. Biol. 15, 900–906 (2019).
Article CAS PubMed Google Scholar
Yang, X. et al. Systematic design and in vitro validation of novel one-carbon assimilation pathways. Metab. Eng. 56, 142–153 (2019).
Article CAS PubMed Google Scholar
Erb, T. J., Jones, P. R. & Bar-Even, A. Synthetic metabolism: metabolic engineering meets enzyme design. Curr. Opin. Chem. Biol. 37, 56–62 (2017).
Article CAS PubMed PubMed Central Google Scholar
Wenk, S., Yishai, O., Lindner, S. N. & Bar-Even, A. An engineering approach for rewiring microbial metabolism. in Methods in Enzymology vol. 608 329–367 (Academic Press), (2018).
Barbier, T. et al. Erythritol feeds the pentose phosphate pathway via three new isomerases leading to D-erythrose-4-phosphate in Brucella. Proc. Natl Acad. Sci. 111, 17815–17820 (2014).
Article ADS CAS PubMed PubMed Central Google Scholar
Huang, H. et al. A general strategy for the discovery of metabolic pathways: D‑threitol, L‑threitol, and erythritol utilization in Mycobacterium. smegmatis. J. Am. Chem. Soc. 137, 14570–14573 (2015).
Article CAS PubMed Google Scholar
Noor, E. et al. Pathway thermodynamics highlights kinetic obstacles in central metabolism. PLoS Comput. Biol. 10, e1003483 (2014).
Article PubMed PubMed Central Google Scholar
Nielsen, J. & Keasling, J. D. Engineering cellular metabolism. Cell 164, 1185–1197 (2016).
Article CAS PubMed Google Scholar
Krusemann, J. L. et al. Artificial pathway emergence in central metabolism from three recursive phosphoketolase reactions. FEBS J. 285, 4367–4377 (2018).
Article PubMed Google Scholar
Charalampous, F. C. & Mueller, G. C. Synthesis of erythrulose phosphate by a soluble enzyme from rat liver. J. Biol. Chem. 201, 161–173 (1953).
Article CAS PubMed Google Scholar
Mueller, G. C., Quinn, E. M. & Rueckert, R. R. The formation of 1-phosphoerythrulose-4-C¹⁴ by homogenates of Swiss chard leaves. Arch. Biochem. Biophys. 55, 408–414 (1955).
Article CAS Google Scholar
Clark, M. G., Blackmore, P. F. & Williams, J. F. Use of L-[¹⁴C] glycerotetrulose for the enzymatic preparation of ¹⁴C-labelled intermediates of the pentose phosphate pathway. J. Label. Compd. 8, 637–652 (1972).
Article CAS Google Scholar
Brockamp, H. P. & Kula, M. R. Purification and characterization of a class I fructose 1,6-bisphosphate aldolase from Staphylococcus carnosus. Appl. Microbiol. Biotechnol. 34, 287–291 (1990).
Article CAS Google Scholar
Schoevaart, R., van Rantwijk, F. & Sheldon, R. A. Class I fructose-1,6-bisphosphate aldolases as catalysts for asymmetric aldol reactions. Tetrahedron Asymmetry 10, 705–711 (1999).
Article CAS Google Scholar
Chiu, T. H. & Feingold, D. S. L-rhamnulose 1-phosphate aldolase from Escherichia coli. Crystallization Prop. Biochem. 8, 98–108 (1969).
CAS Google Scholar
Fessner, W.-D. et al. Diastereoselective enzymatic aldol additions: L-rhamnulose and L-fuculose 1-phosphate aldolases from E. coli. Angew. Chem. Int. Ed. Engl. 30, 555–558 (1991).
Article Google Scholar
Hélaine, V., Gastaldi, C., Lemaire, M., Clapés, P. & Guérard-Hélaine, C. Recent advances in the substrate selectivity of aldolases. ACS Catal. 733–761 https://doi.org/10.1021/acscatal.1c04273 (2021).
Brinkkötter, A., Shakeri-Garakani, A. & Lengeler, J. W. Two class II D-tagatose-bisphosphate aldolases from enteric bacteria. Arch. Microbiol. 177, 410–419 (2002).
Article PubMed Google Scholar
Brovetto, M., Gamenara, D., Saenz Méndez, P. & Seoane, G. A. C−C bond-forming lyases in organic synthesis. Chem. Rev. 111, 4346–4403 (2011).
Article CAS PubMed Google Scholar
Flamholz, A., Noor, E., Bar-Even, A. & Milo, R. eQuilibrator-the biochemical thermodynamics calculator. Nucleic Acids Res. 40, D770–D775 (2012).
Article CAS PubMed Google Scholar
He, H. et al. In vivo rate of formaldehyde condensation with tetrahydrofolate. Metabolites 10, 65 (2020).
Article CAS PubMed PubMed Central Google Scholar
Iacometti, C. et al. Activating silent glycolysis bypasses in Escherichia coli. BioDesign Res. 2022, 1–17 (2022).
Article Google Scholar
Neidhardt, F. C., Ingraham, J. L. & Schaechter, M. Building blocks needed to produce 1g of E. coli protoplasm. in Physiology of the Bacterial Cell: A Molecular Approach (eds. Neidhardt, F. C., Ingraham, J. L. & Schaechter, M.) 134–143 (Sinauer Associates Inc), (1990).
Preumont, A., Snoussi, K., Stroobant, V., Collet, J.-F., & Van Schaftingen, E. Molecular identification of pseudouridine-metabolizing. Enzymes. J. Biol. Chem. 283, 25238–25246 (2008).
Article CAS PubMed Google Scholar
Stevenson, G., Andrianopoulos, K., Hobbs, M. & Reeves, P. R. Organization of the Escherichia coli K-12 gene cluster responsible for production of the extracellular polysaccharide colanic acid. J. Bacteriol. 178, 4885–4893 (1996).
Article CAS PubMed PubMed Central Google Scholar
Wang, J., Ma, W. & Wang, X. Insights into the structure of Escherichia coli outer membrane as the target for engineering microbial cell factories. Microb. Cell Factories 20, 73 (2021).
Article Google Scholar
Wang, L. et al. Divergence of biochemical function in the HAD superfamily: D-glycero-D-manno-Heptose-1,7-bisphosphate phosphatase (GmhB). Biochemistry 49, 1072–1081 (2010).
Article CAS PubMed Google Scholar
Brown, G. et al. Structural and Biochemical Characterization of the Type II Fructose-1,6-bisphosphatase GlpX from Escherichia coli. J. Biol. Chem. 284, 3784–3792 (2009).
Article CAS PubMed PubMed Central Google Scholar
Kuznetsova, E. et al. Genome-wide analysis of substrate specificities of the Escherichia coli haloacid dehalogenase-like phosphatase. Fam. J. Biol. Chem. 281, 36149–36161 (2006).
Article CAS Google Scholar
Reid, T. W. & Wilson, I. B. 17 E. coli Alkaline Phosphatase. in The Enzymes (ed. Boyer, P. D.) vol. 4 373–415 (Academic Press), (1971).
Sorensen, K. I. & Hove-Jensen, B. Ribose catabolism of Escherichia coli: characterization of the rpiB gene encoding ribose phosphate isomerase B and of the rpiR gene, which is involved in regulation of rpiB expression. J. Bacteriol. 178, 1003–1011 (1996).
Article CAS PubMed PubMed Central Google Scholar
Bogorad, I. W., Lin, T. S. & Liao, J. C. Synthetic non-oxidative glycolysis enables complete carbon conservation. Nature 502, 693–697 (2013).
Article ADS CAS PubMed Google Scholar
Wawszkiewicz, E. J. Erythritol metabolism by Propionibacterium pentosaceum. Role of L-erythrulose 1-phosphate. Biochemistry 7, 683–687 (1968).
Article CAS PubMed Google Scholar
Yang, J. et al. Biosynthesis of rare ketoses through constructing a recombination pathway in an engineered Corynebacterium glutamicum. Biotechnol. Bioeng. 112, 168–180 (2015).
Article CAS PubMed Google Scholar
Geddes, B. A. et al. A locus necessary for the transport and catabolism of erythritol in Sinorhizobium meliloti. Microbiology 156, 2970–2981 (2010).
Article CAS PubMed Google Scholar
Sánchez-Andrea, I. et al. The reductive glycine pathway allows autotrophic growth of Desulfovibrio desulfuricans. Nat. Commun. 11, 5090 (2020).
Article ADS PubMed PubMed Central Google Scholar
Antonovsky, N. et al. Sugar synthesis from CO₂ in Escherichia coli. Cell 166, 115–125 (2016).
Article CAS PubMed PubMed Central Google Scholar
Gleizer, S. et al. Conversion of Escherichia coli to generate all biomass carbon from CO₂. Cell 179, 1255–1263.e12 (2019).
Article CAS PubMed PubMed Central Google Scholar
Noor, E., Haraldsdottir, H. S., Milo, R. & Fleming, R. M. Consistent estimation of Gibbs energy using component contributions. PLoS Comput. Biol. 9, e1003098 (2013).
Article ADS CAS PubMed PubMed Central Google Scholar
Monk, J. M. et al. iML1515, a knowledgebase that computes Escherichia coli traits. Nat. Biotechnol. 35, 904–908 (2017).
Article CAS PubMed PubMed Central Google Scholar
Bizouarn, T., van Boxel, G. I., Bhakta, T. & Jackson, J. B. Nucleotide binding affinities of the intact proton-translocating transhydrogenase from Escherichia coli. Biochim Biophys. Acta 1708, 404–410 (2005).
Article CAS PubMed Google Scholar
Jensen, S. I., Lennen, R. M., Herrgard, M. J. & Nielsen, A. T. Seven gene deletions in seven days: Fast generation of Escherichia coli strains tolerant to acetate and osmotic stress. Sci. Rep. 5, 17874 (2015).
Article ADS CAS PubMed PubMed Central Google Scholar
Reisch, C. R. & Prather, K. L. J. Scarless Cas9 assisted recombineering (no-SCAR) in Escherichia coli, an easy-to-use system for genome editing. Curr. Protoc. Mol. Biol. 117, 31 8 1–31 8 20 (2017).
Article PubMed Google Scholar
Thomason, L. C., Costantino, N. & Court, D. L. E. coli genome manipulation by P1 transduction. Curr. Protoc. Mol. Biol. Chapter 1, Unit 1 17 (2007).
Google Scholar
You, L. et al. Metabolic pathway confirmation and discovery through ¹³C-labeling of proteinogenic amino acids. J. Vis. Exp. 26, e3583 (2012).
Google Scholar
Deatherage, D. E. & Barrick, J. E. Identification of Mutations in Laboratory-Evolved Microbes from Next-Generation Sequencing Data Using breseq. in Engineering and Analyzing Multicellular Systems (eds. Sun, L. & Shou, W.) vol. 1151 165–1188 (Springer New York), (2014).
Wu, T. et al. Engineering a synthetic energy-efficient formaldehyde assimilation cycle in Escherichia coli. Zenodo https://doi.org/10.5281/zenodo.10207936. (2023).

Download references

Acknowledgements

We thank Änne Michaelis for experimental assistance. We thank Shanshan Luo for critical reading of the manuscript. This study was funded by the Max Planck Society. Tong Wu and Hai He were also funded by the China Scholarship Council.

Funding

Open Access funding enabled and organized by Projekt DEAL.

Author information

Authors and Affiliations

Max Planck Institute of Molecular Plant Physiology, Am Mühlenberg 1, 14476, Potsdam-Golm, Germany
Tong Wu, Paul A. Gómez-Coronado, Armin Kubis, Steffen N. Lindner, Arren Bar-Even & Hai He
Charité—Universitätsmedizin Berlin, corporate member of Freie Universität Berlin and Humboldt Universität zu Berlin, Institute of Biochemistry, Charitéplatz 1, 10117, Berlin, Germany
Tong Wu & Steffen N. Lindner
Max Planck Institute for Terrestrial Microbiology, Karl-von-Frisch-Str. 10, 35043, Marburg, Germany
Paul A. Gómez-Coronado, Tobias J. Erb & Hai He
TESSSI, The European Syndicate of Synthetic Scientists and Industrialists, 81 rue Réaumur, 75002, Paris, France
Philippe Marlière

Authors

Tong Wu
View author publications
You can also search for this author in PubMed Google Scholar
Paul A. Gómez-Coronado
View author publications
You can also search for this author in PubMed Google Scholar
Armin Kubis
View author publications
You can also search for this author in PubMed Google Scholar
Steffen N. Lindner
View author publications
You can also search for this author in PubMed Google Scholar
Philippe Marlière
View author publications
You can also search for this author in PubMed Google Scholar
Tobias J. Erb
View author publications
You can also search for this author in PubMed Google Scholar
Arren Bar-Even
View author publications
You can also search for this author in PubMed Google Scholar
Hai He
View author publications
You can also search for this author in PubMed Google Scholar

Contributions

Conceptualization, P.M., A.B.-E. and H.H.; Methodology, T.W., P.A.G.C, A.K. and H.H.; Software, H.H.; Validation, T.W, P.A.G.C and A.K.; Formal analysis, T.W., P.A.G.C, A.K. and H.H.; Investigation, T.W, P.A.G.C and A.K.; Resources, T.J.E. and A.B.-E.; Writing – Original Draft, H.H.; Writing—Review & Editing, T.W., P.A.G.C., S.L., P.M., T.J.E. and H.H.; Visualization, H.H.; Supervision, A.B.-E. and H.H.; Funding Acquisition, T.J.E. and A.B.-E.

Corresponding author

Correspondence to Hai He.

Ethics declarations

Competing interests

The authors declare no competing interests.

Peer review

Peer review information

Nature Communications thanks Stéphanie Heux and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Additional information

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary information

Supplementary Information

Peer Review File

Description of Additional Supplementary Files

Supplementary Data 1

Supplementary Data 2

Supplementary Data 3

Reporting Summary

Source data

Source Data

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.

Reprints and permissions

About this article

Cite this article

Wu, T., Gómez-Coronado, P.A., Kubis, A. et al. Engineering a synthetic energy-efficient formaldehyde assimilation cycle in Escherichia coli. Nat Commun 14, 8490 (2023). https://doi.org/10.1038/s41467-023-44247-2

Download citation

Received: 12 November 2022
Accepted: 05 December 2023
Published: 20 December 2023
DOI: https://doi.org/10.1038/s41467-023-44247-2

Comments

By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.