Elsevier

Metabolic Engineering

Volume 56, December 2019, Pages 142-153
Metabolic Engineering

Systematic design and in vitro validation of novel one-carbon assimilation pathways

https://doi.org/10.1016/j.ymben.2019.09.001Get rights and content

Highlights

  • A combinatorial algorithm calculating multiple optimal pathways in metabolic networks.

  • 59 designed C1 assimilation pathways for AcCoA derived products.

  • Criteria to choose the most promising C1 assimilation pathways.

  • In vitro construction of a novel designed pathway reaching 88% carbon yield.

Abstract

The utilization of one-carbon (C1) assimilation pathways to produce chemicals and fuels from low-cost C1 compounds could greatly reduce the substrate-related production costs, and would also alleviate the pressure of the resource supply for bio-manufacturing. However, the natural C1 assimilation pathways normally involve ATP consumption or the loss of carbon resources as CO2, resulting in low product yields, making the design of novel pathways highly pertinent. Here we present several new ATP-independent and carbon-conserving C1 assimilation cycles with 100% theoretical carbon yield, which were discovered by computational analysis of metabolic reaction set with 6578 natural reactions from MetaCyc database and 73 computationally predicted aldolase reactions from ATLAS database. Then, kinetic evaluation of these cycles was conducted and the cycles without kinetic traps were chosen for further experimental verification. Finally, we used the two engineered enzymes Gals and TalBF178Y for the artificial reactions to construct a novel C1 assimilation pathway in vitro and optimized the pathway to achieve 88% carbon yield. These results demonstrate the usefulness of computational design in finding novel metabolic pathways for the efficient utilization of C1 compounds and shedding light on other promising pathways.

Introduction

The assimilation of one-carbon (C1) compounds is one of the fundamental biological processes in most methylotrophs (Kalyuzhnaya et al., 2015). Different C1 substrates such as methanol, methane and dichloromethane are all converted to formaldehyde (FALD) and further metabolized, which makes the FALD consumption pathways into a central part of the metabolism of methylotrophic bacteria (Vorholt, 2002). Methanol is a highly reduced renewable feedstock that can be derived from natural gas. It is desirable to utilize methanol for the production of value-added platform chemicals and polymers (Fan et al., 2018; Tuyishime et al., 2018; Wang et al., 2017b). Improving FALD fixation is a major way to drive the assimilation of methanol (Woolston et al., 2018). Two main natural FALD fixing pathways, the ribulose monophosphate (RuMP) and xylulose monophosphate (XuMP), have shown great promise for C1 fixing (Zhang et al., 2017; Bennett et al., 2018). The RuMP and XuMP convert one FALD (C1) and one ribulose 5-phosphate (Ru5P, C5) or xylulose 5-phosphate (Xu5P, C5) to two glyceraldehyde 3-phosphate (G3P, the C3 building block) molecules (Fig. 1). However, a major problem in using the two natural pathways is that one carbon is lost as CO2 when G3P is converted to acetyl-CoA (AcCoA, the C2 building block), leading to low carbon yield when producing AcCoA derived chemicals. Although the natural serine cycle can convert C1 to AcCoA without carbon loss, FALD assimilation requires 2 ATP and 2 NADH (Fig. 1). Non-natural FALD fixing pathways have also been developed by hijacking the natural serine pathway (Yu and Liao, 2018). However, the modified serine cycle simplifies the enzymatic steps but does not solve the problem of high energy consumption.

Efforts have been taken to avoid CO2 loss and circumvent ATP consumption. Bogorad et al. designed a novel pathway named methanol condensation cycle (MCC) by combining the RuMP pathway and the non-oxidative glycolysis (NOG) pathway for carbon-conserving, ATP-independent one-carbon assimilation (Bogorad et al., 2014) (Fig. 1). They constructed the MCC pathway in vitro and optimized it to produce the AcCoA derived products ethanol and butanol. Finally, the carbon yield of MCC reached 80%, exceeding the 67% theoretical yield of native RuMP and XuMP pathways coupled with the Embden-Meyerhof-Parnas (EMP) pathway. MCC pathway shows the possibility of improving the product yields by constructing new non-natural pathways. Trudeau et al. systematically searched for photorespiration bypass routes that recycle 2-phosphoglycolate into the Calvin cycle without loss of CO2. Combining natural and artificially designed enzymes, they established a carbon-conserving photorespiration bypass in vitro (Trudeau et al., 2018).

Experience-based pathway design has produced better C1 assimilation cycles than natural ones, such as MCC and the modified serine cycle (Bogorad et al., 2014; Yu and Liao, 2018), yet rich experience is needed in designing these cycles. Computer-based pathway design is gradually favored for novel pathways mining due to their rational approach and systematic advantages (Bar-Even et al., 2010; Trudeau et al., 2018). Here we report our work on a systematic study of the possible C1 assimilation pathways by computational analysis of a metabolic reaction set containing reactions from the MetaCyc and ATLAS databases. MetaCyc is a curated database containing experimentally elucidated metabolic pathways spanning all groups of organisms (Caspi et al., 2016) and ATLAS contains more than 130,000 theoretically possible enzymatic reactions designed according to known biotransformation rules (Hadadi et al., 2016). To predict new FALD utilization pathways and ensure the reliability of the designed pathways, the entire MetaCyc database and all the possible aldolase (ALS, the C–C bonding enzyme) reactions extracted from ATLAS database were used as the reaction set for pathways design. By performing comb-FBA (combination of combinatorial algorithm and pFBA), we discovered 59 ATP and NAD(P)H independent and carbon conserving FALD assimilation cycles with 100% theoretical carbon yield for producing acetate, the simplest AcCoA derived product.

In vitro synthetic biology is emerging as a promising bioconversion platform (Opgenorth et al., 2016; Rollin et al., 2013). Compared to the cell-based counterparts, in vitro synthetic systems are superior in many aspects, including fast reaction rates (Li et al., 2017), high product yields (Rollin et al., 2015) and easy manipulation due to high tolerance to toxic chemicals (Guterl et al., 2012). Pathway discovery and enzyme engineering are mutually reinforcing each other. The roaring development of enzyme engineering enables the realization of artificial pathways with non-natural reactions (Wang et al., 2017a), while the computationally designed pathways also provide guidance on target reactions for enzyme engineering. Computationally designed pathways are often first tested clearly in vitro by choosing and designing proper enzymes, then the in vitro realization provides the foundation for eventual application in vivo (Bogorad et al., 2014; Schwander et al., 2016; Trudeau et al., 2018).

How to choose suitable pathways for further experimental tests from a large set of computationally predicted pathways is challenging. Pathway simplicity, thermodynamics and pathway kinetics are often the key factors to be considered in choosing suitable pathways (Bar-Even et al., 2013; Schwander et al., 2016). In the pathways with metabolite branching, improper amounts of enzymes can lead to very low conversion rates and even systemic instability, whereby certain metabolites may accumulate or be depleted, leading to the irreversible loss of the desired operating point (Theisen et al., 2016) or a lower utilization efficiency of enzymes (Barenholz et al., 2017). We narrowed down the options by limiting the numbers of new reactions, kinetic trap analysis and enzyme specificity assay of xylulose 5-phosphate/fructose 6-phosphate phosphoketolase (X/Fpk). Finally, a novel pathway for FALD assimilation was constructed in vitro and was named the glycolaldehyde assimilation (GAA) pathway. Through enzymes selection and process optimization, the product carbon yield of the GAA pathway reached 88%.

Section snippets

Metabolic reaction set preprocessing

We downloaded all reactions from MetaCyc database (version 20.0) (https://metacyc.org/), which contains most known biochemical reactions. Exchange reactions are employed in networks to define in silico medium, the boundaries of which reflect that certain metabolite can be consumed or secreted by the model. As the objective is to find novel FALD assimilation pathways, exchange reactions for FALD were added and the maximal input rate was set at 10 mmol gDCW−1 h−1 for pathway calculation. Exchange

Construction of reaction sets for pathways design

FALD assimilation pathways were calculated using the process shown in Fig. 2. We began by establishing metabolic reaction sets based on MetaCyc and ATLAS databases (Fig. 2A). A total of 19449 reactions were downloaded from the MetaCyc database. Exchange reactions and objective function were added to the initial data set. The computational complexity scales with the search space, which can be decreased by removing blocked reactions that do not participate in any steady state pathway due to the

Discussion

By including a small set of computationally predicted reactions from ATLAS, we obtained much more novel pathways (P4–P76) than those based only on known reactions from MetaCyc (P1–P3). Although the reactions from ATLAS are not known to occur in natural organisms, they are likely to be feasible because their enzymatic reaction mechanisms are based on known biochemistry, which provides more possibilities for designing de novo biosynthesis pathways (Hadadi et al., 2016). The adoption of the

Conclusions

Using a comb-FBA, we designed 59 carbon conserved and ATP/NAD(P)H independent FALD assimilation pathways from a large metabolic network model integrating the known reactions from MetaCyc and the hypothetical aldolase reactions from ATLAS. By applying several criteria for pathway selection, we choose three pathways for further experimental verification. By careful experimental design, we confirmed the novel reaction activity of an engineered aldolase by GC-MS data and constructed the GAA pathway

Conflicts of interest

The authors declare no competing financial interests.

Author contributions

HM conceived the project. QY and HL developed the comb-FBA. QY, HL, FL, JD, PL and YfM reconstructed and revised the metabolic network model. QY performed pathways calculation and analysis. XY performed the pathways kinetic models analysis. XY, XZ, YZ, XJ and YC performed enzymes activities characterization and in vitro pathway experiments. XY and JD performed the pathway thermodynamic analysis. YL and HJ conducted the enzymes engineering experiments, and YY assisted in GC-MS analysis. HM, XY,

Additional information competing interests

The authors declare no competing financial interests.

Acknowledgments

This work was funded by the National Key Research and Development Program of China (2018YFA0900301, 2015CB755704); the Key Research Program of the Chinese Academy of Sciences (ZDRW-ZS-2016-3); the International Partnership Program of Chinese Academy of Sciences (153D31KYSB20170121); the National Natural Science Foundation of China (21908239). We thank Dr. Ting Shi, Dr. Chaoyou Xue, Prof. Tao Chen, Dr. Jiangang Yang and Dr. Wan Yu for assistance in revising the manuscript.

References (76)

  • M.G. Kalyuzhnaya et al.

    Metabolic engineering in methanotrophic bacteria

    Metab. Eng.

    (2015)
  • A. Marsol-Vall et al.

    Injection-port derivatization coupled to GC-MS/MS for the analysis of glycosylated and non-glycosylated polyphenols in fruit samples

    Food Chem.

    (2016)
  • P. Mendes et al.

    Enzyme kinetics and computational modeling for systems biology

    Methods Enzymol.

    (2009)
  • S. Schneider et al.

    Replacement of a phenylalanine by a tyrosine in the active site confers fructose-6-phosphate aldolase activity to the transaldolase of Escherichia coli and human origin

    J. Biol. Chem.

    (2008)
  • M. Schramm et al.

    Phosphorolytic cleavage of fructose-6-phosphate by fructose-6-phosphate phosphoketolase from Acetobacter xylinum

    J. Biol. Chem.

    (1958)
  • W. Schröter et al.

    On the mechanism of phosphoketolase II. Conversion of “thiamine pyrophosphate activated glyeolaldehyde”

    Biochim. Biophys. Acta

    (1963)
  • M. Schurmann et al.

    Fructose-6-phosphate aldolase is a novel class I aldolase from Escherichia coli and is related to a novel group of bacterial transaldolases

    J. Biol. Chem.

    (2001)
  • J. Snider et al.

    Formation of a distinctive complex between the inducible bacterial lysine decarboxylase and a novel AAA+ ATPase

    J. Biol. Chem.

    (2006)
  • R. Suzuki et al.

    Crystal structures of phosphoketolase: thiamine diphosphate-dependent dehydration mechanism

    J. Biol. Chem.

    (2010)
  • P. Tuyishime et al.

    Engineering Corynebacterium glutamicum for methanol-dependent growth and glutamate production

    Metab. Eng.

    (2018)
  • L. Wang et al.

    A review of computational tools for design and reconstruction of metabolic pathways

    Synth. Syst. Biotechnol.

    (2017)
  • X. Wang et al.

    Biological conversion of methanol by evolved Escherichia coli carrying a linear methanol assimilation pathway

    Bioresour. Bioprocess.

    (2017)
  • A. Yevenes et al.

    Cloning, expression, purification, cofactor requirements, and steady state kinetics of phosphoketolase-2 from Lactobacillus plantarum

    Bioorg. Chem.

    (2008)
  • Y. Zhang et al.

    Determination of key enzymes for threonine synthesis through in vitro metabolic pathway analysis

    Microb. Cell Factories

    (2015)
  • G.E. Anthon et al.

    Comparison of three colorimetric reagents in the determination of methanol with alcohol oxidase. Application to the assay of pectin methylesterase

    J. Agric. Food Chem.

    (2004)
  • U. Barenholz et al.

    Design principles of autocatalytic cycles constrain enzyme kinetics and force low substrate saturation at flux branch points

    Elife

    (2017)
  • A. Bar-Even et al.

    Design and analysis of synthetic carbon fixation pathways

    Proc. Natl. Acad. Sci. U.S.A.

    (2010)
  • E.C. Bigham et al.

    Inhibition of arabinose 5-phosphate isomerase. An approach to the inhibition of bacterial lipopolysaccharide biosynthesis

    J. Med. Chem.

    (1984)
  • I.W. Bogorad et al.

    Building carbon-carbon bonds using a biocatalytic methanol condensation cycle

    Proc. Natl. Acad. Sci. U.S.A.

    (2014)
  • I.W. Bogorad et al.

    Synthetic non-oxidative glycolysis enables complete carbon conservation

    Nature

    (2013)
  • M. Brovetto et al.

    C-C bond-forming lyases in organic synthesis

    Chem. Rev.

    (2011)
  • R. Caspi et al.

    The MetaCyc database of metabolic pathways and enzymes and the BioCyc collection of pathway/genome databases

    Nucleic Acids Res.

    (2016)
  • J.A. Castillo et al.

    Fructose-6-phosphate aldolase in organic synthesis: preparation of D-fagomine, N-alkylated derivatives, and preliminary biological assays

    Org. Lett.

    (2006)
  • L. Chindelevitch et al.

    An exact arithmetic toolbox for a consistent and reproducible structural analysis of metabolic network models

    Nat. Commun.

    (2014)
  • A. Chowdhury et al.

    Designing overall stoichiometric conversions and intervening metabolic reactions

    Sci. Rep.

    (2015)
  • A.L. Concia et al.

    D-fructose-6-phosphate aldolase in organic synthesis: cascade chemical-enzymatic preparation of sugar-related polyhydroxylated compounds

    Chemistry

    (2009)
  • Z. Cui et al.

    Concomitant cell-free biosynthesis of optically pure D-(−)-acetoin and xylitol via a novel NAD+ regeneration in two-enzyme cascade

    J. Chem. Technol. Biotechnol.

    (2018)
  • M.K. Dudek et al.

    Spontaneous keto-enol tautomerization in the crystal lattice visualized with the help of water encapsulated in hydrophilic reservoirs

    ChemPhysChem

    (2017)
  • Cited by (58)

    View all citing articles on Scopus
    1

    These authors contributed equally to this work.

    View full text