Identification of genome-scale metabolic network models using experimentally measured flux profiles

Genome-scale metabolic network models can be reconstructed for well-characterized organisms using genomic annotation and literature information. However, there are many instances in which model predictions of metabolic fluxes are not entirely consistent with experimental data, indicating that the reactions in the model do not match the active reactions in the in vivo system. We introduce a method for determining the active reactions in a genome-scale metabolic network based on a limited number of experimentally measured fluxes. This method, called optimal metabolic network identification (OMNI), allows efficient identification of the set of reactions that results in the best agreement between in silico predicted and experimentally measured flux distributions. We applied the method to intracellular flux data for evolved Escherichia coli mutant strains with lower than predicted growth rates in order to identify reactions that act as flux bottlenecks in these strains. The expression of the genes corresponding to these bottleneck reactions was often found to be downregulated in the evolved strains relative to the wild-type strain. We also demonstrate the ability of the OMNI method to diagnose problems in E. coli strains engineered for metabolite overproduction that have not reached their predicted production potential. The OMNI method applied to flux data for evolved strains can be used to provide insights into mechanisms that limit the ability of microbial strains to evolve towards their predicted optimal growth phenotypes. When applied to industrial production strains, the OMNI method can also be used to suggest metabolic engineering strategies to improve byproduct secretion. In addition to these applications, the method should prove to be useful in general for reconstructing metabolic networks of ill-characterized microbial organisms based on limited amounts of experimental data.     

Polyamines and root formation in mung bean hypocotyl cuttings : I. Effects of exogenous compounds and changes in endogenous polyamine content

The effect of several polyamines (putrescine, spermidine, and spermine), their precursors (l-arginine and l-ornithine), and some analogs and metabolic inhibitors (l-canavanine, l-canaline, and methylglyoxal-bis [guanylhydrazone]) on root formation have been studied in mung bean (Vigna radiata [L.] Wilczek) hypocotyl cuttings.     

Bacteriocin Protein BacL1 of Enterococcus faecalis Targets Cell Division Loci and Specifically Recognizes l-Ala2-Cross-Bridged Peptidoglycan

Bacteriocin 41 (Bac41) is produced from clinical isolates of Enterococcus faecalis and consists of two extracellular proteins, BacL₁ and BacA. We previously reported that BacL₁ protein (595 amino acids, 64.5 kDa) is a bacteriolytic peptidoglycan d-isoglutamyl-l-lysine endopeptidase that induces cell lysis of E. faecalis when an accessory factor, BacA, is copresent. However, the target of BacL₁ remains unknown. In this study, we investigated the targeting specificity of BacL₁. Fluorescence microscopy analysis using fluorescent dye-conjugated recombinant protein demonstrated that BacL₁ specifically localized at the cell division-associated site, including the equatorial ring, division septum, and nascent cell wall, on the cell surface of target E. faecalis cells. This specific targeting was dependent on the triple repeat of the SH3 domain located in the region from amino acid 329 to 590 of BacL₁. Repression of cell growth due to the stationary state of the growth phase or to treatment with bacteriostatic antibiotics rescued bacteria from the bacteriolytic activity of BacL₁ and BacA. The static growth state also abolished the binding and targeting of BacL₁ to the cell division-associated site. Furthermore, the targeting of BacL₁ was detectable among Gram-positive bacteria with an l-Ala-l-Ala-cross-bridging peptidoglycan, including E. faecalis, Streptococcus pyogenes, or Streptococcus pneumoniae, but not among bacteria with alternate peptidoglycan structures, such as Enterococcus faecium, Enterococcus hirae, Staphylococcus aureus, or Listeria monocytogenes. These data suggest that BacL₁ specifically targets the l-Ala-l-Ala-cross-bridged peptidoglycan and potentially lyses the E. faecalis cells during cell division.     

Biosynthesis of l-tyrosine O-sulphate from the methyl and ethyl esters of l-tyrosine

1. Rat-liver supernatant preparations are capable of achieving the biological sulphation of l-tyrosine methyl ester, the reaction proceeding maximally at a substrate concentration of 30 mm and at pH 7·0. 2. Two sulphated products are formed, one of which has been identified as l-tyrosine O-sulphate. On the basis of indirect evidence the other product can be assumed to be l-tyrosine O-sulphate methyl ester. 3. An enzyme present in rat-liver supernatant preparations is capable of converting l-tyrosine O-sulphate methyl ester into l-tyrosine O-sulphate. This enzyme is inhibited by l-tyrosine methyl ester. 4. l-Tyrosine ethyl ester also yields two sulphated products when used as an acceptor in the liver sulphating system. One of these has been identified chromatographically as l-tyrosine O-sulphate and the other may be presumed to be l-tyrosine O-sulphate ethyl ester.     

Biosynthesis of l-Sorbose and l-Psicose Based on C—C Bond Formation Catalyzed by Aldolases in an Engineered Corynebacterium glutamicum Strain

The property of loose stereochemical control at aldol products from aldolases helped to synthesize multiple polyhydroxylated compounds with nonnatural stereoconfiguration. In this study, we discovered for the first time that some fructose 1,6-diphosphate aldolases (FruA) and tagatose 1,6-diphosphate (TagA) aldolases lost their strict stereoselectivity when using l-glyceraldehyde and synthesized not only l-sorbose but also a high proportion of l-psicose. Among the aldolases tested, TagA from Bacillus licheniformis (BGatY) showed the highest enzyme activity with l-glyceraldehyde. Subsequently, a “one-pot” reaction based on BGatY and fructose-1-phosphatase (YqaB) generated 378 mg/liter l-psicose and 199 mg/liter l-sorbose from dihydroxyacetone-phosphate (DHAP) and l-glyceraldehyde. Because of the high cost and instability of DHAP, a microbial fermentation strategy was used further to produce l-sorbose/l-psicose from glucose and l-glyceraldehyde, in which DHAP was obtained from glucose through the glycolytic pathway, and some recombination pathways based on FruA or TagA and YqaB were constructed in Escherichia coli and Corynebacterium glutamicum strains. After evaluation of different host cells and combinations of FruA or TagA with YqaB and optimization of gene expression, recombinant C. glutamicum strain WT(pXFTY) was selected and produced 2.53 g/liter total ketoses, with a yield of 0.50 g/g l-glyceraldehyde. Moreover, deletion of gene cgl0331, encoding the Zn-dependent alcohol dehydrogenase in C. glutamicum, was confirmed for the first time to significantly decrease conversion of l-glyceraldehyde to glycerol and to increase yield of target products. Finally, fed-batch culture of strain SY14(pXFTY) produced 3.5 g/liter l-sorbose and 2.3 g/liter l-psicose, with a yield of 0.61 g/g l-glyceraldehyde. This microbial fermentation strategy also could be applied to efficiently synthesize other l-sugars.     

Virulence Gene Regulation by l-Arabinose in Salmonella enterica

Invasion of the intestinal epithelium is a critical step in Salmonella enterica infection and requires functions encoded in the gene cluster known as Salmonella Pathogenicity Island 1 (SPI-1). Expression of SPI-1 genes is repressed by l-arabinose, and not by other pentoses. Transport of l-arabinose is necessary to repress SPI-1; however, repression is independent of l-arabinose metabolism and of the l-arabinose-responsive regulator AraC. SPI-1 repression by l-arabinose is exerted at a single target, HilD, and the mechanism appears to be post-translational. As a consequence of SPI-1 repression, l-arabinose reduces translocation of SPI-1 effectors to epithelial cells and decreases Salmonella invasion in vitro. These observations reveal a hitherto unknown role of l-arabinose in gene expression control and raise the possibility that Salmonella may use L-arabinose as an environmental signal.     

Amino Acid Isomerization in the Production of l-Phenylalanine from d-Phenylalanine by Bacteria

To establish an advantageous method for the production of l-amino acids, microbial isomerization of d- and dl-amino acids to l-amino acids was studied. Screening experiments on a number of microorganisms showed that cell suspensions of Pseudomonas fluorescens and P. miyamizu were capable of isomerizing d- and dl-phenylalanines to l-phenylalanine. Various conditions suitable for isomerization by these organisms were investigated. Cells grown in a medium containing d-phenylalanine showed highest isomerization activity, and almost completely converted d- or dl-phenylalanine into l-phenylalanine within 24 to 48 hr of incubation. Enzymatic studies on this isomerizing system suggested that the isomerization of d- or dl-phenylalanine is not catalyzed by a single enzyme, “amino acid isomerase,” but the conversion proceeds by a two step system as follows: d-pheylalanine is oxidized to phenylpyruvic acid by d-amino acid oxidase, and the acid is converted to l-phenylalanine by transamination or reductive amination.     

Bacterial catabolism of threonine. Threonine degradation initiated by l-threonine acetaldehyde-lyase (aldolase) in species of Pseudomonas

1. The route of l-threonine degradation was studied in four strains of the genus Pseudomonas able to grow on the amino acid and selected because of their high l-threonine aldolase activity. Growth and manometric results were consistent with the cleavage of l-threonine to acetaldehyde+glycine and their metabolism via acetate and serine respectively. 2. l-Threonine aldolases in these bacteria exhibited pH optima in the range 8.0–8.7 and K_m values for the substrate of 5–10mm. Extracts exhibited comparable allo-l-threonine aldolase activities, K_m values for this substrate being 14.5–38.5mm depending on the bacterium. Both activities were essentially constitutive. Similar activity ratios in extracts, independent of growth conditions, suggested a single enzyme. The isolate Pseudomonas D2 (N.C.I.B. 11097) represents the best source of the enzyme known. 3. Extracts of all the l-threonine-grown pseudomonads also possessed a CoA-independent aldehyde dehydrogenase, the synthesis of which was induced, and a reversible alcohol dehydrogenase. The high acetaldehyde reductase activity of most extracts possibly resulted in the underestimation of acetaldehyde dehydrogenase. 4. l-Serine dehydratase formation was induced by growth on l-threonine or acetate+glycine. Constitutively synthesized l-serine hydroxymethyltransferase was detected in extracts of Pseudomonas strains D2 and F10. The enzyme could not be detected in strains A1 and N3, probably because of a highly active `formaldehyde-utilizing' system. 5. Ion-exchange and molecular exclusion chromatography supported other evidence that l-threonine aldolase and allo-l-threonine aldolase activities were catalysed by the same enzyme but that l-serine hydroxymethyltransferase was distinct and different. These results contrast with the specificities of some analogous enzymes of mammalian origin.     

l-Lactate Production from Biodiesel-Derived Crude Glycerol by Metabolically Engineered Enterococcus faecalis: Cytotoxic Evaluation of Biodiesel Waste and Development of a Glycerol-Inducible Gene Expression System

Biodiesel waste is a by-product of the biodiesel production process that contains a large amount of crude glycerol. To reuse the crude glycerol, a novel bioconversion process using Enterococcus faecalis was developed through physiological studies. The E. faecalis strain W11 could use biodiesel waste as a carbon source, although cell growth was significantly inhibited by the oil component in the biodiesel waste, which decreased the cellular NADH/NAD⁺ ratio and then induced oxidative stress to cells. When W11 was cultured with glycerol, the maximum culture density (optical density at 600 nm [OD₆₀₀]) under anaerobic conditions was decreased 8-fold by the oil component compared with that under aerobic conditions. Furthermore, W11 cultured with dihydroxyacetone (DHA) could show slight or no growth in the presence of the oil component with or without oxygen. These results indicated that the DHA kinase reaction in the glycerol metabolic pathway was sensitive to the oil component as an oxidant. The lactate dehydrogenase (Ldh) activity of W11 during anaerobic glycerol metabolism was 4.1-fold lower than that during aerobic glycerol metabolism, which was one of the causes of low l-lactate productivity. The E. faecalis pflB gene disruptant (Δpfl mutant) expressing the ldhL1_LP gene produced 300 mM l-lactate from glycerol/crude glycerol with a yield of >99% within 48 h and reached a maximum productivity of 18 mM h⁻¹ (1.6 g liter⁻¹ h⁻¹). Thus, our study demonstrates that metabolically engineered E. faecalis can convert crude glycerol to l-lactate at high conversion efficiency and provides critical information on the recycling process for biodiesel waste.     

Fast Reconstruction of Compact Context-Specific Metabolic Network Models

Systemic approaches to the study of a biological cell or tissue rely increasingly on the use of context-specific metabolic network models. The reconstruction of such a model from high-throughput data can routinely involve large numbers of tests under different conditions and extensive parameter tuning, which calls for fast algorithms. We present fastcore, a generic algorithm for reconstructing context-specific metabolic network models from global genome-wide metabolic network models such as Recon X. fastcore takes as input a core set of reactions that are known to be active in the context of interest (e.g., cell or tissue), and it searches for a flux consistent subnetwork of the global network that contains all reactions from the core set and a minimal set of additional reactions. Our key observation is that a minimal consistent reconstruction can be defined via a set of sparse modes of the global network, and fastcore iteratively computes such a set via a series of linear programs. Experiments on liver data demonstrate speedups of several orders of magnitude, and significantly more compact reconstructions, over a rival method. Given its simplicity and its excellent performance, fastcore can form the backbone of many future metabolic network reconstruction algorithms.     

Bonded Cumomer Analysis of Human Melanoma Metabolism Monitored by 13C NMR Spectroscopy of Perfused Tumor Cells

A network model for the determination of tumor metabolic fluxes from ¹³C NMR kinetic isotopomer data has been developed and validated with perfused human DB-1 melanoma cells carrying the BRAF V600E mutation, which promotes oxidative metabolism. The model generated in the bonded cumomer formalism describes key pathways of tumor intermediary metabolism and yields dynamic curves for positional isotopic enrichment and spin-spin multiplets. Cells attached to microcarrier beads were perfused with 26 mm [1,6-¹³C₂]glucose under normoxic conditions at 37 °C and monitored by ¹³C NMR spectroscopy. Excellent agreement between model-predicted and experimentally measured values of the rates of oxygen and glucose consumption, lactate production, and glutamate pool size validated the model. ATP production by glycolytic and oxidative metabolism were compared under hyperglycemic normoxic conditions; 51% of the energy came from oxidative phosphorylation and 49% came from glycolysis. Even though the rate of glutamine uptake was ∼50% of the tricarboxylic acid cycle flux, the rate of ATP production from glutamine was essentially zero (no glutaminolysis). De novo fatty acid production was ∼6% of the tricarboxylic acid cycle flux. The oxidative pentose phosphate pathway flux was 3.6% of glycolysis, and three non-oxidative pentose phosphate pathway exchange fluxes were calculated. Mass spectrometry was then used to compare fluxes through various pathways under hyperglycemic (26 mm) and euglycemic (5 mm) conditions. Under euglycemic conditions glutamine uptake doubled, but ATP production from glutamine did not significantly change. A new parameter measuring the Warburg effect (the ratio of lactate production flux to pyruvate influx through the mitochondrial pyruvate carrier) was calculated to be 21, close to upper limit of oxidative metabolism.     

Manipulation of the Anoxic Metabolism in Escherichia coli by ArcB Deletion Variants in the ArcBA Two-Component System

Bioprocesses conducted under conditions with restricted O₂ supply are increasingly exploited for the synthesis of reduced biochemicals using different biocatalysts. The model facultative anaerobe Escherichia coli has elaborate sensing and signal transduction mechanisms for redox control in response to the availability of O₂ and other electron acceptors. The ArcBA two-component system consists of ArcB, a membrane-associated sensor kinase, and ArcA, the cognate response regulator. The tripartite hybrid kinase ArcB possesses a transmembrane, a PAS, a primary transmitter (H1), a receiver (D1), and a phosphotransfer (H2) domain. Metabolic fluxes were compared under anoxic conditions in a wild-type E. coli strain, its ΔarcB derivative, and two partial arcB deletion mutants in which ArcB lacked either the H1 domain or the PAS-H1-D1 domains. These analyses revealed that elimination of different segments in ArcB determines a distinctive distribution of d-glucose catabolic fluxes, different from that observed in the ΔarcB background. Metabolite profiles, enzyme activity levels, and gene expression patterns were also investigated in these strains. Relevant alterations were observed at the P-enol-pyruvate/pyruvate and acetyl coenzyme A metabolic nodes, and the formation of reduced fermentation metabolites, such as succinate, d-lactate, and ethanol, was favored in the mutant strains to different extents compared to the wild-type strain. These phenotypic traits were associated with altered levels of the enzymatic activities operating at these nodes, as well as with elevated NADH/NAD⁺ ratios. Thus, targeted modification of global regulators to obtain different metabolic flux distributions under anoxic conditions is emerging as an attractive tool for metabolic engineering purposes.     

Synthesis of l-(+)-Tartaric Acid from l-Ascorbic Acid via 5-Keto-d-Gluconic Acid in Grapes

5-Keto-l-idionic acid (5-keto-d-gluconic acid, d-xylo-5-hexulosonic acid) was found as a metabolic product of l-ascorbic acid in slices of immature grapes, Vitis labrusca L. cv `Delaware'. Specifically labeled compounds, recognized as metabolic products of l-ascorbic acid in grapes, were fed to young grape tissues to investigate the metabolic pathway from l-ascorbic acid to l-(+)-tartaric acid.     

Secondary Cell Wall Polymers of Enterococcus faecalis Are Critical for Resistance to Complement Activation via Mannose-binding Lectin

The complement system is part of our first line of defense against invading pathogens. The strategies used by Enterococcus faecalis to evade recognition by human complement are incompletely understood. In this study, we identified an insertional mutant of the wall teichoic acid (WTA) synthesis gene tagB in E. faecalis V583 that exhibited an increased susceptibility to complement-mediated killing by neutrophils. Further analysis revealed that increased killing of the mutant was due to a higher rate of phagocytosis by neutrophils, which correlated with higher C3b deposition on the bacterial surface. Our studies indicated that complement activation via the lectin pathway was much stronger on the tagB mutant compared with wild type. In concordance, we found an increased binding of the key lectin pathway components mannose-binding lectin and mannose-binding lectin-associated serine protease-2 (MASP-2) on the mutant. To understand the mechanism of lectin pathway inhibition by E. faecalis, we purified and characterized cell wall carbohydrates of E. faecalis wild type and V583ΔtagB. NMR analysis revealed that the mutant strain lacked two WTAs with a repeating unit of →6)[α-l-Rhap-(1→3)]β-d-GalpNAc-(1→5)-Rbo-1-P and →6) β-d-Glcp-(1→3) [α-d-Glcp-(1→4)]-β-d-GalpNAc-(1→5)-Rbo-1-P→, respectively (Rbo, ribitol). In addition, compositional changes in the enterococcal rhamnopolysaccharide were noticed. Our study indicates that in E. faecalis, modification of peptidoglycan by secondary cell wall polymers is critical to evade recognition by the complement system.     

Non-repair Pathways for Minimizing Protein Isoaspartyl Damage in the Yeast Saccharomyces cerevisiae

The spontaneous degradation of asparaginyl and aspartyl residues to isoaspartyl residues is a common type of protein damage in aging organisms. Although the protein-l-isoaspartyl (d-aspartyl) O-methyltransferase (EC 2.1.1.77) can initiate the repair of l-isoaspartyl residues to l-aspartyl residues in most organisms, no gene homolog or enzymatic activity is present in the budding yeast Saccharomyces cerevisiae. Therefore, we used biochemical approaches to elucidate how proteins containing isoaspartyl residues are metabolized in this organism. Surprisingly, the level of isoaspartyl residues in yeast proteins (50–300 pmol of isoaspartyl residues/mg of protein extract) is comparable with organisms with protein-l-isoaspartyl (d-aspartyl) O-methyltransferase, suggesting a novel regulatory pathway. Interfering with common protein quality control mechanisms by mutating and inhibiting the proteasomal and autophagic pathways in vivo did not increase isoaspartyl residue levels compared with wild type or uninhibited cells. However, the inhibition of metalloproteases in in vitro aging experiments by EDTA resulted in an ∼3-fold increase in the level of isoaspartyl-containing peptides. Characterization by mass spectrometry of these peptides identified several proteins involved in metabolism as targets of isoaspartyl damage. Further analysis of these peptides revealed that many have an N-terminal isoaspartyl site and originate from proteins with short half-lives. These results suggest that one or more metalloproteases participate in limiting isoaspartyl formation by robust proteolysis.     

Proteases and Peptidases of Castor Bean Endosperm: Enzyme Characterization and Changes during Germination

The endosperm of castor bean seeds (Ricinus communis L.) contains two —SH-dependent aminopeptidases, one hydrolyzing l-leucine-β-naphthylamide optimally at pH 7.0, and the other hydrolyzing l-proline-β-naphthylamide optimally at pH 7.5. After germination the endosperm contains in addition an —SH-dependent hemoglobin protease, a serine-dependent carboxypeptidase, and at least two —SH-dependent enzymes hydrolyzing the model substrate α-N-benzoyl-dl-arginine-β-naphthylamide (BANA). The carboxypeptidase is active on a variety of N-carbobenzoxy dipeptides, especially N-carbobenzoxy-L-phenylalanine-l-alanine and N-carbobenzoxy-l-tyrosine-l-leucine. The pH optima for the protease, carboxypeptidase, and BANAase acivities are 3.5 to 4.0, 5.0 to 5.5, and 6 to 8, respectively.     

Systems-level analysis of genome-scalein silico metabolic models using MetaFluxNet

The systems-level analysis of microbes with myriad of heterologous data generated by omics technologies has been applied to improve our understanding of cellular function and physiology and consequently to enhance production of various bioproducts. At the heart of this revolution residesin silico genome-scale metabolic model. In order to fully exploit the power of genome-scale model, a systematic approach employing user-friendly software is required. Metabolic flux analysis of genome-scale metabolic network is becoming widely employed to quantify the flux distribution and validate model-driven hypotheses. Here we describe the development of an upgraded MetaFluxNet which allows (1) construction of metabolic models connected to metabolic databases, (2) calculation of fluxes by metabolic flux analysis, (3) comparative flux analysis with flux-profile visualization, (4) the use of metabolic flux analysis markup language to enable models to be exchanged efficiently, and (5) the exporting of data from constraints-based flux analysis into various formats. MetaFluxNet also allows cellular physiology to be predicted and strategies for strain improvement to be developed from genome-based information on flux distributions. This integrated software environment promises to enhance our understanding on metabolic network at a whole organism level and to establish novel strategies for improving the properties of organisms for various biotechnological applications.     

A Nonproteolytic "Trypsin-like" Enzyme: Purification and Properties of Arachain

By the use of the proteolytic substrates benzoyl-dl-arginine-p-nitroanilide and benzoyl-l-arginine ethyl ester the enzyme arachain has been purified 325-fold from acetone powders of ungerminated peanuts. The pH optimum for the hydrolysis of benzoyl-dl-arginine-p-nitroanilide was 8.1 in tris buffer, and for benzoyl-l-arginine ethyl ester was 7.5 using N - 2 - hydroxyethylpiperazine - N′ - 2 - ethanesulfonic acid buffer. The purest fraction showed one main band with one to three minor bands on disc gel electrophoresis. The major protein component had an S_20,w of 6.20. The energy of activation for the hydrolysis of benzoyl-dl-arginine-p-nitroanilide was calculated to be 16 kilocalories. The Michaelis constant for benzoyl-dl-arginine-p-nitroanilide was 10 micromolar and for benzoyl-l-arginine ethyl ester was 110 micromolar. The enzyme showed essentially no activity with casein, dimethyl casein, or bovine serum albumin as substrates. A large number of peptides were hydrolyzed by the enzyme, only l-leucyl-l-tyrosine being resistant of the peptides tested. The results suggest that arachain is not a “trypsin-like” protease but is a peptide hydrolase.     

One-Pot Production of l-threo-3-Hydroxyaspartic Acid Using Asparaginase-Deficient Escherichia coli Expressing Asparagine Hydroxylase of Streptomyces coelicolor A3(2)

We developed a novel process for efficient synthesis of l-threo-3-hydroxyaspartic acid (l-THA) using microbial hydroxylase and hydrolase. A well-characterized mutant of asparagine hydroxylase (AsnO-D241N) and its homologous enzyme (SCO2693-D246N) were adaptable to the direct hydroxylation of l-aspartic acid; however, the yields were strictly low. Therefore, the highly stable and efficient wild-type asparagine hydroxylases AsnO and SCO2693 were employed to synthesize l-THA. By using these recombinant enzymes, l-THA was obtained by l-asparagine hydroxylation by AsnO followed by amide hydrolysis by asparaginase via 3-hydroxyasparagine. Subsequently, the two-step reaction was adapted to one-pot bioconversion in a test tube. l-THA was obtained in a small amount with a molar yield of 0.076% by using intact Escherichia coli expressing the asnO gene, and thus, two asparaginase-deficient mutants of E. coli were investigated. A remarkably increased l-THA yield of 8.2% was obtained with the asparaginase I-deficient mutant. When the expression level of the asnO gene was enhanced by using the T7 promoter in E. coli instead of the lac promoter, the l-THA yield was significantly increased to 92%. By using a combination of the E. coli asparaginase I-deficient mutant and the T7 expression system, a whole-cell reaction in a jar fermentor was conducted, and consequently, l-THA was successfully obtained from l-asparagine with a maximum yield of 96% in less time than with test tube-scale production. These results indicate that asparagine hydroxylation followed by hydrolysis would be applicable to the efficient production of l-THA.