Engineered recombinant factor VII Q217 variants with altered inhibitor specificities.

Recombinant factor VII with residue 217 (chymotrypsinogen numbering system) converted to alanine (VIIQ217A), glutamic acid (VIIQ217E), or glycine (VIIQ217G) was characterized. In a prothrombin time assay, VIIQ217E demonstrated 100%, VIIQ217A 15%, and VIIQ217G <1% clotting activities relative to wild-type VII. Binding of VIIQ217A and VIIQ217G to TF was comparable to that of wild-type VII to TF. All the variants were readily activated by factor Xa. Autoactivation in the presence of TF was efficient with VIIQ217E, slow with VIIQ217A, but undetected with VIIQ217G. Relative to wild-type VII added at the same concentration, VIIQ217E had no effect on the PT of normal plasma, whereas VIIQ217A slightly and VIIQ217G dramatically prolonged the clotting time in a dose-dependent manner. Activation of macromolecular substrates paralleled this functional inhibition. The k(cat)/K(M) values for factor X activation in the presence of TF were 2.4 for VIIaQ217E as compared to 1.9 (M(-)(1) s(-)(1) x 10(7)) for wild-type VIIa, 1.57 for VIIaQ217A, and 0.05 with VIIaQ217G. In comparison to wild-type VIIa, VIIaQ217E cleaved the chromogenic substrate S2765 (Z-D-Arg-Gly-Arg-pNA) with 10-fold higher k(cat). Analysis of the interactions with the inhibitors TFPI and antithrombin III demonstrated that VIIaQ217A but not VIIaQ217E or VIIaQ217G was inhibited less efficiently by TFPI either in the presence or in the absence of factor Xa. In contrast, VIIaQ217A association with antithrombin III in the presence of heparin was the fastest among the variants with a second-order rate constant of 2.31 (x10(3) M(-)(1) min(-)(1)), as compared to 0.47 and 1.47 for VIIaQ217E and wild-type VIIa, respectively. Our results demonstrate that residue Q(217) is important in regulating substrate and, more importantly, inhibitor recognition by VIIa.     

Cloning and sequence analysis of the Escherichia coli metH gene encoding cobalamin-dependent methionine synthase and isolation of a tryptic fragment containing the cobalamin-binding domain

A gene encoding cobalamin-dependent methionine synthase (EC 2.1.1.13) has been isolated from a plasmid library of Escherichia coli K-12 DNA by complementation to methionine prototrophy in an E. coli strain lacking both cobalamin-dependent and -independent methionine synthase activities (RK4536:metE, metHH). Maxicell expression of a series of plasmids containing deletions in the metH structural gene was employed to map the position and orientation of the gene on the cloned DNA fragment. A 6.3-kilobase EcoRI-SalI fragment containing the gene was cloned into the sequencing vector pGEM3B for double-stranded DNA sequencing; the MetH coding region consists of 3372 nucleotides. The enzyme was purified from an overproducing strain of E. coli harboring the recombinant plasmid, in which the level of methionine synthase was elevated 30- to 40-fold over wild-type E. coli. Recombinant enzyme is a protein of 123,640 molecular weight and has a turnover number of 1,450 min-1 in the standard assay. These values are to be compared with previously reported values of 133,000 for the molecular weight and 1,240-1,560 min-1 for the turnover number of the homogenous enzyme purified from a wild-type strain of E. coli B (Frasca, V., Banerjee, R. V., Dunham, W. R., Sands, R. H., and Matthews, R. G. (1988) Biochemistry 27, 8458-8465). Limited proteolysis of the native enzyme with trypsin resulted in loss of enzyme activity but retention of bound cobalamin on a peptide fragment of 28,000 molecular weight. This fragment has been shown to extend from residue 643 to residue 900 of the 1124-residue deduced amino acid sequence.     

Isolation and characterization of thioredoxin from the cyanobacterium, Anabaena sp

Thioredoxin from Anabaena sp. has been purified 800-fold with an assay based on the reduction of insulin disulfides by NADPH and the heterologous calf thymus thioredoxin reductase. The final material was homogeneous on polyacrylamide gel electrophoresis and had a molecular weight of 12,000; the NH2-terminal residue was serine and the COOH-terminal was leucine. Anabaena thioredoxin-(SH)2 is a hydrogen donor for the adenosylcobalamin-dependent anabaena ribonucleotide reductase and is equally active with the iron-containing ribonucleotide reductase from Escherichia coli. Anabaena thioredoxin-S2 is a good substrate for E. coli thioredoxin reductase. We have compared the structure of Anabaena and E. coli thioredoxins. Clear structural differences between the proteins, compatible with the large evolutionary distance between the organisms, were seen with respect to total amino acid composition, isoelectric point, tryptic peptide maps, and a low immunochemical cross-reactivity. However, both thioredoxins contain a single oxidation-reduction active disulfide bridge with the amino acid sequence: Cys-Gly-Pro-Cys-Lys. The tryptophan fluorescence emission of Anabaena thioredoxin-S2 increases more than 3-fold on reduction to thioredoxin-(SH)2. This behavior is identical with that of E. coli thioredoxin, suggesting a very similar overall folding of homologous molecules.     

A simplified and high-throughput chromogenic assay for testing tissue factor-dependent procoagulant activity

Tissue factor (TF), the primary initiator of the coagulation cascade, plays a critical role in hemostasis and thrombosis, and inhibition of TF activity appears to be an attractive target for the treatment of cardiovascular diseases. However, few selective small-molecule inhibitors of TF are available, and the present assays for measuring TF activity are relatively expensive and complex. The authors present a simple and high-throughput chromogenic assay for screening TF inhibitors based on using commercial human prothrombin complex instead of purified coagulation factors, reducing the dosage, and performing with a one-stage procedure. In the optimized assay, <45 μL cell lysates was incubated with Tris-CaCl(2) buffer (pH 7.3) containing human prothrombin complex at 37°C for 15 min in 96-well or 384-well plates. Tris-EDTA buffer (pH 8.4) containing chromogenic substrate Xa was then added and the absorbance measured at 405 nm. This simplified assay was more sensitive or precise than some reported methods for TF procoagulant activities. Two known active compounds (curcumin and simvastatin) inhibiting TF activity were tested by the simplified assay to validate the screening method. Furthermore, berberine and cryptotanshinone suppressed TF activity induced by lipopolysaccharides in human monocytes by this assay and might be promising new TF inhibitors.     

Cloning and characterization of three fatty alcohol oxidase genes from Candida tropicalis strain ATCC 20336

Candida tropicalis (ATCC 20336) converts fatty acids to long-chain dicarboxylic acids via a pathway that includes among other reactions the oxidation of omega-hydroxy fatty acids to omega-aldehydes by a fatty alcohol oxidase (FAO). Three FAO genes (one gene designated FAO1 and two putative allelic genes designated FAO2a and FAO2b), have been cloned and sequenced from this strain. A comparison of the DNA sequence homology and derived amino acid sequence homology between these three genes and previously published Candida FAO genes indicates that FAO1 and FAO2 are distinct genes. Both genes were individually cloned and expressed in Escherichia coli. The substrate specificity and K(m) values for the recombinant FAO1 and FAO2 were significantly different. Particularly striking is the fact that FAO1 oxidizes omega-hydroxy fatty acids but not 2-alkanols, whereas FAO2 oxidizes 2-alkanols but not omega-hydroxy fatty acids. Analysis of extracts of strain H5343 during growth on fatty acids indicated that only FAO1 was highly induced under these conditions. FAO2 contains one CTG codon, which codes for serine (amino acid 177) in C. tropicalis but codes for leucine in E. coli. An FAO2a construct, with a TCG codon (codes for serine in E. coli) substituted for the CTG codon, was prepared and expressed in E. coli. Neither the substrate specificity nor the K(m) values for the FAO2a variant with a serine at position 177 were radically different from those of the variant with a leucine at that position.     

High expression of the second lysine decarboxylase gene, ldc, in Escherichia coli WC196 due to the recognition of the stop codon (TAG), at a position which corresponds to the 33th amino acid residue of sigma38, as a serine residue by the amber suppressor, supD

Escherichia coli WC196, which was obtained from the strain W3110 by nitrosoguanidine mutagenesis as an overproducer of lysine, produced approximately twenty times more cadaverine than did W3110, and had a twenty fold higher level of rpoS gene product, sigma38, than in W3110. Both WC196 and W3110 had a stop codon (TAG) in rpoS at position which corresponds to the 33th residue of sigma38 protein. In addition, WC196 but not W3110 had a mutation in the gene encoding Ser-tRNA (SerU), called, supD. Analysis of the amino acid sequence of a sigma38 preparation from WC196 showed that the 33th residue of sigma38 is a serine residue. The deltarpoS deltacadA mutant of E. coli W3110 harboring the plasmid containing rpoS, in which the TAG codon was converted to a TCG codon for serine-33 residue of sigma38, expressed a significant amount of Ldc and accumulated a large amount of sigma38. However, the deltarpoS deltacadA mutant of W3110 with the plasmid containing the intact rpoS from W3110 could synthesize neither sigma38 nor Ldc significantly.     

Phosphorylation of CKBBP2/CRIF1 by protein kinase CKII promotes cell proliferation

CKII plays a significant role in cell proliferation and cell cycle control. In this report, yeast two-hybrid assay and pull-down assay demonstrate that CKBBP2/CRIF1 associates with the beta subunit of CKII in vitro and in vivo. Recombinant CKBBP2/CRIF1 is phosphorylated in vitro by purified CKII and by CKII inhibitor apigenin-sensitive protein kinase in HEK293 cell extract. Phosphoamino acid analysis and mutational analysis indicate that CKII phosphorylates serine at residue 221 within CKBBP2/CRIF1. Furthermore, serine to alanine mutation at residue 221 abrogates the phosphorylation of CKBBP2/CRIF1 observed in HEK293 cell extract, indicating that CKII is a major kinase that is responsible for phosphorylation of CKBBP2/CRIF1. As compared with the wild-type CKBBP2/CRIF1 or nonphosphorylatable mutant CKBBP2(S221A) (in which the serine-221 is replaced by alanine), overexpression of CKBBP2(S221E) in COS7 cells promotes cell proliferation. Taken together, the present results suggest that CKII may be involved in cell proliferation, at least in part, through the phosphorylation of serine-221 within CKBBP2/CRIF1.     

Cloning and functional characterisation of a cis-muuroladiene synthase from black peppermint (Menthaxpiperita) and direct evidence for a chemotype unable to synthesise farnesene

Using oligonucleotide primers designed to the known gene sequence of an (E)-beta-farnesene (EbetaF) synthase, two cDNA sequences (MxpSS1 and MxpSS2) were cloned from a black peppermint (Menthaxpiperita) plant. MxpSS1 encoded a protein with 96% overall amino acid sequence identity with the EbetaF synthase. Recombinant MxpSS1 produced in Escherichia coli, after removal of an N-terminal thioredoxin fusion, had a K(m) for FPP of 1.91+/-0.1 microM and k(cat) of 0.18 s(-1), and converted farnesyl diphosphate (FPP) into four products, the major two being cis-muurola-3,5-diene (45%) and cis-muurola-4(14),5-diene (43%). This is the first cis-muuroladiene synthase, to be characterised. MxpSS2 encoded a protein with only two amino acids differing from EbetaF synthase. Recombinant MxpSS2 protein showed no activity towards FPP. One of the two mutations, at position 531 (leucine in MxpSS2 and serine in EbetaF synthase) was shown, by structural modelling to occur in the J-K loop, an element of the structure of sesquiterpene synthases known to be important in the reaction mechanism. Reintroduction of the serine at position 531 into MxpSS2 by site-directed mutagenesis restored EbetaF synthase activity (K(m) for FPP 0.98+/-0.12 microM, k(cat) 0.1 s(-1)), demonstrating the crucial role of this residue in the enzyme activity. Analysis, by GC-MS, of the sesquiterpene profile of the plant used for the cloning, revealed that EbetaF was not present, confirming that this particular mint chemotype had lost EbetaF synthase activity due to the observed mutations.     

Expression, Purification, and Characterization of Recombinant Human Factor X

A system is described for producing recombinant factor X with properties very similar to human plasma factor X. Optimization of the expression system for factor X resulted in the finding that human kidney cells (293 cells) are superior to the widely utilized baby hamster kidney cells (BHK cells) for the expression of functional factor X. It was also determined that production of factor X by 293 cells requires the substitution of the −2 residue (Thr → Arg) which affords the removal of the factor X propeptide. Purification of recombinant and plasma factor X is accomplished using a calcium-dependent monoclonal antibody directed against the gla domain. The proteins are comparable by sodium dodecyl sulfate polyacrylamide gel electrophoresis. The rate and extent of activation by the factor X coagulant protein from Russell's viper venom and by factors IXa and VIIIa are similar; activation of the recombinant protein by VIIa and tissue factor is mildly faster. The activated enzymes have the same activity toward a chromogenic substrate and the biologic substrate, prothrombin. Both enzymes have the same apparent affinity for the activated platelet surface as judged by their ability to activate prothrombin. Finally, inhibition by antithrombin, with or without heparin, and inhibition by the tissue factor pathway inhibitor are equivalent. Recombinant factor X produced by this method is therefore well suited for probing structure–function relationships by mutational analysis.     

Direct production of proteins with N-terminal cysteine for site-specific conjugation

Proteins with N-terminal cysteine can undergo native chemical ligation and are useful for site-specific N-terminal labeling or protein semisynthesis. Recombinant production of these has usually been by site-specific cleavage of a precursor fusion protein at an internal cysteine residue. Here we describe a simpler route to producing these proteins. Overexpression in E. coli of several proteins containing cysteine as the second amino acid residue yielded products in which the initiating methionine residue had been completely cleaved by endogenous methionine aminopeptidase. While secondary modification of the terminal cysteine was a complicating factor, conditions were identified to eliminate or minimize this problem. Recombinant proteins produced in this way were suitable for site-specific modification of the amino terminus via native chemical ligation technology, as demonstrated by conjugation of a thioester-containing derivative of fluorescein to one such protein. The ability to directly produce proteins with N-terminal cysteine should simplify the application of native chemical ligation technology to recombinant proteins and make the technique more amenable to researchers with limited expertise in protein chemistry.     

The primary structure of Escherichia coli K12 2-deoxyribose 5-phosphate aldolase. Nucleotide sequence of the deoC gene and the amino acid sequence of the enzyme

The sequence of the deoC gene of Escherichia coli K12 and the amino acid sequence of the corresponding protein, deoxyriboaldolase, has been established. The protein consists of 259 amino acids with a molecular weight of 27 737. The purified enzyme may exist both as a monomer and as a dimer. On the basis of amino acid composition, molecular weight and catalytic properties, the enzymes from E. coli and Salmonella typhimurium seem to be almost similar. They belong to the class I aldolases, which form Schiff base intermediates. Using data for the S. typhimurium enzyme, the lysine residue involved in the active site in the E. coli enzyme was tentatively identified.     

Mutation and Mutagenesis of thiol peroxidase of Escherichia coli and a new type of thiol peroxidase family.

A novel thioredoxin-linked thiol peroxidase (Px) from Escherichia coli has been reported previously (M. K. Cha, H. K. Kim, and I. H. Kim, J. Biol. Chem. 270:28635-28641, 1995). In an attempt to perform physiological and biochemical characterizations of the thiol Px, a thiol Px null (tpx) mutant and a functional-residue mutant of thiol Px were produced. The tpx mutant was viable in aerobic culture but grew more slowly than the wild-type cells. The difference in growth rate became more pronounced when oxidative-stress-inducing reagents, such as peroxides and paraquat, were added to the cultures. The viability of the individual tpx mutant under oxidative stress was much lower than that of wild-type cells. tpx mutants growing aerobically respond to paraquat with a sixfold greater induction of Mn-superoxide dismutase than that of the wild-type cells. The deduced amino acid sequence of the thiol Px was found to be from 42 to 72% identical to the sequences of proteins from Haemophilus influenzae (ToxR regulon), Vibrio cholerae (ToxR regulon), and three kinds of streptococci (coaggregation-mediating adhesins), suggesting that they all belong to a new thiol Px family. Alignment of the amino acid sequences of the thiol Px family members showed that one cysteine, which corresponds to Cys-94 in E. coli thiol Px, is perfectly conserved. The substitution of serine for this cysteine residue resulted in complete loss of Px activity. These results suggest that the members of the thiol Px family, including E. coli thiol Px, have a functional cysteine residue and function in vivo as peroxidases.     

Characterization of a membrane-associated serine protease in Escherichia coli.

Three membrane-associated proteolytic activities in Escherichia coli were resolved by DEAE-cellulose chromatography from detergent extracts of the total envelope fraction. On the basis of substrate specificity for the hydrolysis of chromogenic amino acid ester substrates, the first two eluting activities were determined previously to be protease V and protease IV, respectively (M. Pacaud, J. Bacteriol. 149:6-14, 1982). The third proteolytic activity eluting from the DEAE-cellulose column was further purified by affinity chromatography on benzamidine-Sepharose 6B. We termed this enzyme protease VI. Protease VI did not hydrolyze any of the chromogenic substrates used in the detection of protease IV and protease V. However, all three enzymes generated acid-soluble fragments from a mixture of E. coli membrane proteins which were biosynthetically labeled with radioactive amino acids. The activity of protease VI was sensitive to serine protease inhibitors. Using [3H]diisopropylfluorophosphate as an active-site labeling reagent, we determined that protease VI has an apparent molecular weight of 43,000 in polyacrylamide gels. All three membrane-associated serine proteases were insensitive to inhibition by Ecotin, and endogenous, periplasmic inhibitor of trypsin.     

Protease II from Escherichia coli: sequencing and expression of the enzyme gene and characterization of the expressed enzyme.

Protease II gene of Escherichia coli HB101 was cloned and expressed in E. coli JM83. The transformant harboring a hybrid plasmid, pPROII-12, with a 2.4 kbp fragment showed 90-fold higher enzyme activity than the host. The whole nucleotide sequence of the inserted fragment of plasmid pPROII-12 was clarified by the dideoxy chain-terminating method. The sequence that encoded the mature enzyme protein was found to start at an ATG codon, as judged by comparison with amino terminal protein sequencing. The molecular weight of the enzyme was estimated to be 81,858 from the nucleotide sequence. The reactive serine residue of protease II was identified as Ser-532 with tritium DFP. The sequence around the serine residue is coincident with the common sequence of Gly-X-Ser-X-Gly, which has been found in the active site of serine proteases. Except for this region, protease II showed no significant sequence homology with E. coli serine proteases, protease IV and protease La (lon gene), or other known families of serine proteases. However, 25.3% homology was observed between protease II and prolyl endopeptidase from porcine brain. Although the substrate specificities of these two enzymes are quite different, it seems possible to classify protease II as a member of the prolyl endopeptidase family from the structural point of view.     

Molecular cloning of the Lon protease gene from Thermus thermophilus HB8 and characterization of its gene product.

The gene encoding Lon protease was isolated from an extreme thermophile, Thermus thermophilus HB8. Sequence analysis demonstrated that the T. thermophilus Lon protease gene (TT-lon) contains a protein-coding sequence consisting of 2385 bp which is approximately 56% homologous to the Escherichia coli counterpart. As expected, the G/C content of TT-lon was 68%, which is significantly higher than that of the E. coli lon gene (52% G/C). The amino acid sequence of T. thermophilus Lon protease (TT-Lon) predicted from the nucleotide sequence contained several unique sequences conserved in other Lon proteases: (a) a cysteine residue at the position just before the putative ATP-binding domain; (b) motif A and B sequences required for composition of the ATP-binding domain; and (c) a serine residue at the proteolytic active site. Expression of TT-lon under the control of the T7 promoter in E. coli produced an 89-kDa protein with a yield of approximately 5 mg.L-1. Recombinant TT-Lon (rTT-Lon) was purified to homogeneity by sequential column chromatography. The peptidase activity of rTT-Lon was activated by ATP and alpha-casein. rTT-Lon cleaved succinyl-phenylalanyl-leucyl-phenylalanyl-methoxynaphthylamide much more efficiently than succinyl-alanyl-alanyl-phenylalanyl-methoxynaphthylamide, whereas both peptides were cleaved with comparable efficiencies by E. coli Lon. These results suggest that there is a difference between TT-Lon and E. coli Lon in substrate specificity. rTT-Lon most effectively cleaved substrate peptides at 70 degrees C, which was significantly higher than the optimal temperature (37 degrees C) for E. coli Lon. Together, these results indicate that the TT-lon gene isolated from T. thermophilus HB8 actually encodes an ATP-dependent thermostable protease Lon.     

可溶性组织因子突变体的基因构建、表达和性质

血浆中组织因子 (TF)的过量表达与许多病理过程密切相关。TF抑制物有可能防治这些疾病。设计了两种可溶性组织因子 (sTF)的突变体 (MCsTF和MFsTF) ,突变了协同催化的功能域 ,保留与因子VII/VIIa结合的功能域 ,使突变体竞争性抑制野生型TF的功能。用PCR的方法 ,对可溶性组织因子cDNA基因进行了点突变 ,并实现了在大肠杆菌中高效表达。rsTF、rMCsTF和rMFsTF的促凝活性研究表明 ,rMFsTF的激活X因子活性和促凝血作用相当于rsTF的 10 % ,而rMCsTF几乎完全失去了激活X因子活性和促凝血作用。rMCsTF和rMFsTF与VII/VIIa因子形成复合物对激活X因子的催化特异常数 (kcat/Km)分别是FVII/VIIa·rsTF的 2 .0 %和 3.7% ,也说明突变体与VII/VIIa形成的复合物对激活X因子催化活性显著降低。rMCsTF和rMFsTF对rsTF活性抑制动力学研究及体外活性研究表明 ,两种突变体均有抑制rsTF活性和抑制兔脑粉的促凝活性作用 ,抑制作用呈量效关系     

Glutamate 636 of the Escherichia coli pyruvate dehydrogenase-E1 participates in active center communication and behaves as an engineered acetolactate synthase with unusual stereoselectivity

The residue Glu636 is located near the thiamine diphosphate (ThDP) binding site of the Escherichia coli pyruvate dehydrogenase complex E1 subunit (PDHc-E1), and to probe its function two variants, E636A and E636Q were created with specific activities of 2.5 and 26% compared with parental PDHc-E1. According to both fluorescence binding and kinetic assays, the E636A variant behaved according to half-of-the-sites mechanism with respect to ThDP. In contrast, with the E636Q variant a K(d,ThDP) = 4.34 microM and K(m,ThDP) = 11 microM were obtained with behavior more reminiscent of the parental enzyme. The CD spectra of both variants gave evidence for formation of the 1',4'-iminopyrimidine tautomer on binding of phosphonolactylthiamine diphosphate, a stable analog of the substrate-ThDP covalent complex. Rapid formation of optically active (R)-acetolactate by both variants, but not by the parental enzyme, was observed by CD and NMR spectroscopy. The acetolactate configuration produced by the Glu636 variants is opposite that produced by the enzyme acetolactate synthase and the Asp28-substituted variants of yeast pyruvate decarboxylase, suggesting that the active centers of the two sets of enzymes exhibit different facial selectivity (re or si) vis à vis pyruvate. The tryptic peptide map (mass spectral analysis) revealed that the Glu636 substitution changed the mobility of a loop comprising amino acid residues from the ThDP binding fold. Apparently, the residue Glu636 has important functions both in active center communication and in protecting the active center from undesirable "carboligase" side reactions.     

Insights into the role of the histidines in the structure and stability of cytochrome c

In this paper we investigate the role played by each histidine in the amino acid sequence of yeast iso-1-cytochrome c (with the exception of H18, the residue axially coordinated to the heme iron) in determining the protein structure and stability. To this end, we have generated and characterized the double mutants H26Y/H33Y, H26Y/H39K and H33Y/H39K obtained from the C102T variant of the protein, which retain only one histidine side chain in the amino acid sequence. In particular, the H39K mutation inserts a lysine at position 39 as in the sequence of equine cytochrome c. The H26Y/H33Y/H39K triple mutant, which lacks all three histidines, was also produced and its spectroscopic properties are compared with those of the double mutants. The data highlight the critical role played by H26 in determining protein stability. Recombinant horse cytochrome c and the corresponding H26Y mutant were also generated and characterized. Since equine cytochrome c exhibits higher stability than the yeast protein, this provides a valuable opportunity to understand the role played by the invariant H26 residue in determining structure and stability.     

Overproduction of microbial transglutaminase in Escherichia coli, in vitro refolding, and characterization of the refolded form

(MTG) The Streptoverticillium transglutaminase gene, synthesized previously for yeast expression, was modified and resynthesized for overexpression in E. coli. A high-level expression plasmid, pUCTRPMTG-02(+), was constructed. Furthermore, to eliminate the N-terminal methionine, pUCTRPMTGD2 was constructed. Cultivation of E. coli transformed with pUCTRPMTG02(+) or pUCTRPMTGD2 yielded a large amount of MTG (200-300 mg/liter) as insoluble inclusion bodies. The N-terminal amino acid residue of the expressed protein was methionine or serine (the second amino acid residue of the mature MTG sequence), respectively. Transformed E. coli cells were disrupted, and collected pellets of inclusion bodies were solubilized with 8 M urea. Rapid dilution treatment of solubilized MTG restored the enzymatic activity. Refolded MTG, purified by ion-exchange chromatography, which had an N-terminal methionine or serine residue, showed activity equivalent to that of native MTG. These results indicated that recombinant MTG could be produced efficiently in E. coli.