PEDS Advance Access originally published online on January 31, 2008
Protein Engineering Design and Selection 2008 21(4):233-239; doi:10.1093/protein/gzm047
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Development of fructosyl amine oxidase specific to fructosyl valine by site-directed mutagenesis
1Department of Biotechnology, Graduate School of Engineering, Tokyo University of Agriculture and Technology, 2-24-16 Naka-cho, Koganei-shi 185-8688, Japan
2 To whom correspondence should be addressed. E-mail: sode{at}cc.tuat.ac.jp
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and methodsResultsDiscussionReferences |
|---|
Docking models of fructosyl amine oxidase (FAOD) from the marine yeast Pichia N1-1 (N1-1 FAOD) with the substrates fructosyl valine (f-Val) and fructosyl-
N-lysine (f-Lys) were produced using three-dimensional protein model as reported previously (Miura et al., 2006, Biotechnol. Lett., 28, 1895-1900). The residues involved in recognition of substrates were proposed, particularly Asn354, which interacts closely with f-Lys, but not with f-Val. Substitution of Asn354 to histidine and lysine simultaneously resulted in an increase in activity of f-val and a decrease in activity of f-Lys and thus, increasing the specificity for f-Val from 13- to 19-fold. In addition to creating two mutant FAODs with great potential for the measurement of glycated hemoglobin, we have provided the first structural model of substrate binding with eukaryotic FAOD, which is expected to contribute to further investigation of FAOD.
Keywords: active site/biosensor/docking/fructosyl amine oxidase/substrate specificity
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionReferences |
|---|
Glycation involves a series of non-enzymatic reactions between reducing sugars and free amino groups in amino acids or proteins. The reaction proceeds through Schiff base intermediate that can undergo a slow Amadori rearrangement to produce a relatively stable product (Thorpe and Bynes, 1982
). Glycation also occurs in blood, where the reaction with glucose results in the formation of Amadori products, hemoglobin A1c (HbA1c) and glycated albumin, which are recognized as important glycemic control indicators of diabetes mellitus (Bookchin and Gallop, 1968; Klenk et al., 1982; Johnson et al., 1983; Iberg and Fluckiger, 1986). Amadori products are also formed during the processing and storage of food products and have a significant influence on the taste, appearance and overall quality of the food stored. Because of the important implications for both medical diagnosis and quality control during food processing, there is a great importance placed on the detection of glycated proteins. A number of methods are available for the detection of glycated compounds, such as HPLC and mass spectrometry. However, because these methods are generally complicated and expensive, there remains a strong demand for a simple and economical detection system that may be applied for point-of-care or home monitoring of diabetes patients or on-site quality control in the food industry.
The development of an enzyme-based sensor using fructosyl amine oxidase (FAOD), sometimes called Amadoriase, is an approach with great potential for satisfying such a demand. FAODs catalyze the oxidation of fructosyl amino acids to produce glucosone, the unglycated amino acid and hydrogen peroxide. FAODs have been isolated and characterized from various microorganisms, including filamentous fungi (Yoshida et al., 1995, 1996; Takahashi et al., 1997; Hirokawa et al., 2003; Akazawa et al., 2004), marine yeasts (Sode et al., 2001) and bacteria (Horiuchi et al., 1989; Ferri et al., 2005). We have already isolated an FAOD from the marine yeast strain N1-1, and succeeded in the cloning of the N1-1 FAOD gene and its recombinant production in Escherichia coli (Ferri et al., 2004). We have also developed a number of N1-1 FAOD-based biosensor systems capable of effectively measuring f-Val (Tsugawa et al., 2000, 2001; Ogawa et al., 2002).
HbA1c results from the glycation of N-terminal valine residue of the hemoglobin β subunit. The abundance and long lifetime of hemoglobin in blood has made HbA1c a major glycemic indicator for diabetes patients, where its level reflects the average blood glucose concentration over the past two months. Glycated albumin, which results from the glycation of the amino group on the side-chain of internal lysine residues, has also become an important indicator for diabetes. Because N1-1 FAOD is unable to measure the glycated proteins directly, the enzymatic measurement of HbA1c or glycated albumin must be preceded by a proteolytic digestion step to liberate f-Val or f-Lys, respectively. A biosensor would ideally use an FAOD enzyme, specific for the corresponding glycated amino acid, for the measurement of either HbA1c or glycated albumin.
We have recently proposed a three-dimensional model of N1-1 FAOD, identifying the residues located at the putative active site (Miura et al., 2006). Although single amino acid substitution of most of these putative active site residues abolished activity, the substitution of Asn354 had a pronounced effect on substrate specificity and retained enzymatic activity. These results validated our proposed three-dimensional model, encouraging us to investigate the possibilities of using the model to improve further the enzymatic properties of FAOD for future analytical applications. We accomplished this by creating docking models of N1-1 FAOD with its substrates f-Val and f-Lys. The resulting models were then used to design new mutants with greatly improved specificity for f-Val, which have considerable potential for the measurement of HbA1c.
|
Materials and methods |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionReferences |
|---|
Materials
Glycated amino acids were prepared according to the method proposed by Keil et al. (1985). 4-Aminoantipyrine was obtained from Kanto Kagaku (Tokyo, Japan). N,N-Bis(4-sulfobutyl)-3-methylaniline, disodium salt (TODB) was obtained from Dojin (Kumamoto, Japan). Horseradish peroxidase was obtained from Amano Enzymes (Tokyo, Japan). All other chemicals were of reagent grade.
Three-dimensional model construction
The N1-1 FAOD three-dimensional model was previously constructed by homology modeling using the Molecular Operating Environment (MOE, Chemical Computing Group, Inc.: http://www.chemcomp.com/) (Miura et al., 2006). The template used for modeling was the crystal structure of monomeric sarcosine oxidase (MsOX) from Bacillus sp. B-0618 (PDB ID: 1EL5), which shows high homology to the N1-1 FAOD. The three-dimensional models of f-Val and f-Lys were constructed using MOE, with the sugar moieties in the β-pyranose form, because Amadori products have been found to be 
61% in that form at room temperature (Monnier and Wu, 2003).
The active site of N1-1 FAOD was predicted using the Alpha Site Finder function of the MOE program, which identifies pockets that are stabilized by hydrophobic and hydrophilic probes and assigns dummy atoms to occupy the van der Walls volume of the cavity. Docking simulations were then carried out using the AS-Dock module of the MOE program on N1-1 FAOD with its substrates. This program generates initial substrate conformations of the predicted active site and determines the conformations with the greatest overlapping van der Waals volume with the dummy atoms. The potential energy minimizations are ranked as the conformations are optimized further. AS-Dock was used with the standard default settings: minimization including protein atoms within 4.5 Å from the 
site, potential cut off is 6.5 Å out of the site, initial conformation is 1000 and RMS gradient is 1. Distance restraints were imposed according to MsOX structural information, as described in the Results section.
The FAOD expression vector pTN1 was previously constructed based on pTrc99A (Ferri et al., 2004). Site-directed mutagenesis was accomplished using the QuikChange Site-Directed Mutagenesis Kit (Stratagene, La Jolla, CA, USA) according to the manufacturer’s instructions. The oligonucleotides used were Asn354Lys (5'-CGGGTGACTCTGGAAAATCGTTCAAGATC-3'), Asn354His (5'-CGGGTGACTCTGGACACTCGTTCAAGATC-3') and Asn354Val (5'-CGGGTGACTCTGGAGTGTCGTTCAAGATC-3') together with their corresponding complementary oligonucleotides. The mutations (italicized) were confirmed by sequencing with the ABI Prism BigDye Terminator cycle sequencing kit v3.0 on an ABI Prism 310 Genetic Analyzer (Applied Biosystems, Foster City, CA, USA).
Preparation of cell extracts and purification of FAOD wild type and mutants
E. coli DH5 transformed by the mutated pTN1 were grown aerobically at 37°C in LB medium containing 50 µg ampicillin ml–1. After reaching an A660 nm value of 0.5, cells were induced with 0.3 mM isopropyl-β-D-thiogalactoside (IPTG) and incubation was continued at 30°C for 6 h. Cells were harvested by centrifugation and resuspended in 10 mM potassium phosphate buffer (PPB), pH 7.0, and lysed by five passages through a French press (1500 kg cm–2). The lysate was centrifuged at 10 000g at 4°C for 15 min and the supernatant was again centrifuged at 40 000g at 4°C for 90 min. FAOD was then dialyzed against 10 mM PPB, pH 7.0, and applied to the RESOURCE Q column (Amersham Bioscience, Uppsala, Sweden) equilibrated with 10 mM PPB, pH 8.0. The absorbed protein was eluted with a linear gradient of NaCl (0–0.25 M). The active fractions were pooled and flavin adenine dinucleotide (FAD) was added to a final concentration of 100 µM. Ammonium sulfate was added to 35% saturation and the precipitated protein was collected by centrifugation at 15 000g for 20 min. The supernatant was applied to the RESOURCE PHE column (Amersham Bioscience) equilibrated with 10 mM PPB, pH 6.5, containing 35% saturated ammonium sulfate. The absorbed protein eluted with a linear gradient of ammonium sulfate (23.6–16.7% saturated) in 10 mM PPB (pH 6.5). To the pooled active fractions, FAD was added to a final concentration of 100 µM and ammonium sulfate was added to have 80% saturation. After a 30-min incubation at 4°C, the formed precipitate was pelleted by centrifugation at 15 000g for 20 min. The resultant precipitate was dissolved and dialyzed at 4°C against 10 mM PPB, pH 8.0, containing 100 µM FAD and 1% mannose. The sample was subsequently dialyzed against 10 mM PPB, pH 8.0, containing 100 µM FAD. The dialyzed enzyme solution was applied to a RESOURCE Q column (Amersham Bioscience) equilibrated with 10 mM PPB, pH 8.0. The absorbed protein was eluted with a linear gradient NaCl (0.05–0.15 M) in the same buffer. The active fractions were collected and FAD was added to a final concentration of 100 µM. Finally, the purified enzyme solution was dialyzed against 10 mM PPB, pH 7.0, and stored at 4°C. Enzyme purity was confirmed by the appearance of a single band on SDS-PAGE (data not shown) and its concentration was measured using the DC Protein Assay Kit (Bio-Rad, Hercules, CA, USA).
FAOD activity was assayed in 5 or 500 mM PPB, pH 7.0, containing 1.5 mM TODB, 2 U ml–1 horseradish peroxidase and 1.5 mM 4-aminoantipyrine at 25°C in the presence of various concentrations of substrates. The formation of quinoneimine dye was measured spectrophotometrically at 546 nm. One unit is defined as the enzyme quantity that oxidizes one mole of f-Val per minute under the above reaction conditions.
|
Results |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionReferences |
|---|
Docking simulation
Restraints of docking simulation
FAOD has relatively high similarity (31%) with MsOX from Bacillus sp. B-0618, whose tertiary structure has been elucidated (Trickey et al., 1999). A number of residues that play key roles in MsOX, such as FAD binding, substrate recognition and proton relay, are partially conserved in N1-1 FAOD. We previously constructed a three-dimensional model of N1-1 FAOD by homology modeling, using the MsOX crystal structure as template (Miura et al., 2006).
In addition to the structural similarities, FAOD and MsOX share important similarities in their substrates and reaction mechanism also. Both FAOD and MsOX catalyze the oxidation of secondary amines, cleaving their respective substrate at the intermediate’s ketoamine bond (Fig. 1). MsOX catalyzes the oxidation of sarcosine (N-methylglycine) by molecular oxygen and water to form glycine, formaldehyde and hydrogen peroxide. Similarly, FAOD catalyzes the oxidation of fructosyl amino acids to glucosone, amino acid and hydrogen peroxide.
|
Because of the structural and functional similarities between FAOD and MsOX, some restraints were imposed in carrying out docking simulations between FAOD and its substrates based on information from the interactions of MsOX with sarcosine. The recognition of sarcosine by MsOX involves hydrogen bonding between the substrate’s carboxylate oxygens and the side-chain nitrogens of Arg52 and Lys348 (Wagner et al., 2000
). To reproduce these interactions, distance restraints were imposed in the minimization step between the substrate’s carboxyl group and the side-chains of His51 and Lys357 (2.8–3.2 Å, 10 kcal), which we previously identified as the corresponding amino acid residues (Miura et al., 2006). Furthermore, the reaction site C–N of the substrate, which forms the intermediate ketoamine bond, is positioned directly above N(5)–C(4a) of the flavin ring in MsOX, thereby facilitating 
-orbital overlap between the flavin ring and the substrate. Similar interactions were also reported for another FAD-dependent enzyme, D-amino acid oxidase, and its substrate (Miura et al., 1997). To reproduce this -orbital overlap in the N1-1 FAOD model, distance restraints were also imposed between the substrate’s reaction site C–N and the C(5a)–N(4) of FAD (3.2–3.8 Å, 10 kcal) (Fig. 2).
|
Docking model of FAOD with f-Val and f-
Lys
Docking simulations were performed on N1-1 FAOD complexed with either f-Val or f-Lys. The final docking models showing the lowest potential energies were selected for each of the two substrates (Fig. 3). In both docking models, some residues, including the residues that had restraints imposed (His51, Gly353 and Lys357), are located within hydrogen-bonding distance to the substrate’s amino acid moiety. Furthermore, many hydrophobic residues are located around the side chain of the substrate. Although many amino acid residues appear to interact with the amino acid moiety of the substrates, only His234 is located within hydrogen-bonding distance to the fructose moiety. Therefore, it appears that FAOD recognizes the amino acid moiety of the substrate more than its fructose moiety. The overall distance between substrate and protein is much shorter in the f-Lys docking model than that in the f-Val docking model, with binding energies of –250 and –150 kcal, respectively.
|
The role of Asn354 in substrate specificity Interestingly, Asn354 is located within hydrogen-bonding distance to the
-amino group of f-Lys, but is not in such proximity to f-Val. This suggests that Asn354 is involved in the recognition of only f-Lys, and offers a great opportunity to improve the enzymes preference for f-Val for its application in HbA1c sensing. Therefore, Asn354 was modified in an attempt to disrupt specifically the recognition of f-Lys, but not that of f-Val. We created the Asn354Lys, Asn354His and Asn354Val single mutants of N1-1 FAOD. Lys and His were selected as they are expected to cause repulsion between the positive charges of the side-chain and -amino group of f-Lys. In contrast, substitution to Val was expected to cause an increase in hydrophobicity at the active site.
Substrate specificity of recombinant FAOD mutants in crude extracts
The three FAOD mutants, together with the previously created Asn354Ala mutant (Miura et al., 2006), were expressed in E. coli and initially characterized in crude cell-free extracts (Table I). All four mutants showed decreased activity when assayed with 10 mM f-Lys, especially Asn354Lys, whose activity could not be detected. Even in the case of Asn354Ala, although its specific activity with 10 mM f-Lys was only slightly affected by the mutation, the Km value for f-Lys was increased >6-fold to 63 mM.
|
All mutants showed activity with 10 mM f-Val that was comparable with wild-type levels, with minor changes possibly reflecting small differences in expression level. Most mutants showed significantly decreased Km values for f-Val, especially Asn354His, which showed a >6-fold decrease. Only Asn354Val showed an increase in Km value (79% increase) for f-Val. All mutants showed improvements in specificity for f-Val, albeit at varying degrees. Because Asn354His showed the greatest decrease in Km value for f-Val and Asn354Lys showed no detectable activity toward f-
Lys, these mutants were selected for purification and further characterization. Characterization of purified Asn354His and Asn354Lys mutants Purification of the FAOD mutants allowed us to characterize their kinetic properties further (Table II). As with the crude extracts, analysis of the purified mutants showed that the Asn354His substitution caused a large decrease in Km value (7.5-fold) for f-Val, whereas Asn354Lys showed only a modest decrease. The Vmax values of f-Val were only slightly affected by the Asn354His and Asn354Lys substitutions, resulting in 4.4- and 1.6-fold increases in VmaxKm–1, respectively.
|
In contrast, the Km value for f-
Lys was unaffected by the Asn354His substitutions, but almost doubled with Asn354Lys. Both mutations resulted in >3-fold decreases in Vmax for f-Lys, resulting in 3- and 9-fold decreases in VmaxKm–1, respectively. Therefore, Asn354His and Asn354Lys substitutions caused 15- and 14-fold increases in specificity for f-Val, respectively, as determined by dividing the VmaxKm–1 for f-Val by that for f-Lys (Table II).
Effect of buffer concentration
Recent studies on N1-1 FAOD have been carried out in 500 mM PPB, as we had noticed that higher buffer concentrations resulted in greatly enhanced specificity for f-Val. We decided to determine whether PPB concentration had a similar effect on the Asn354His and Asn354Lys FAOD mutants, as their specificities had already been greatly enhanced by the substitutions. As expected, decreasing the buffer concentration from 500 to 5 mM PPB changed the substrate specificity of the wild-type enzyme from having a clear preference for f-Val to an essentially equivalent preference for f-Lys (Table II). The VmaxKm–1 value for f-Val slightly decreased with decreasing PPB concentration while that for f-Lys almost tripled.
At 5 mM, each of the wild-type and mutant FAODs has very similar Vmax values with the two substrates (Table II). At 500 mM PPB, the wild-type FAOD has very similar Vmax values with the two substrates, while the two mutants have 2- to 4-fold lower Vmax values with f-Lys than with f-Val. In terms of Km values with f-Val, the changes resulting from the substitutions are comparable under the two PPB concentrations. However, changes caused by the substitutions on the Km values for f-Lys are much more significant at 5 mM PPB than at 500 mM PPB. These results indicate that the mutants are also affected by the PPB concentration, although the relative effects vary, somewhat depending on substrate and enzyme. The Asn354His and Asn354Lys substitutions caused 14- and 15-fold increases in specificity for f-Val, respectively, at 500 mM PPB. Similarly, the Asn354His and Asn354Lys substitutions caused 14- and 19-fold increases in specificity for f-Val, respectively, when assayed in 5 mM PPB.
The HbA1c concentration in whole blood generally ranges from 4 to 12% of total hemoglobin, corresponding to 150 –700 µM f-Val, which would be liberated upon proteolytic digestion. To test the potential of our mutant FAODs for the future measurement of HbA1c, their activities with the two substrates in 500 µM concentrations were compared (Table III). Even in 500 mM PPB, which provides a significant improvement in f-Val specificity, the activity of wild-type FAOD with f-Lys is still 48% of that with f-Val. However, the oxidation of f-Lys by the Asn354His and Asn354Lys mutants is only 2% of that of f-Val, thus providing a great improvement in substrate specificity for the measurement of f-Val liberated from the proteolytic digestion of HbA1c in whole blood.
|
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionReferences |
|---|
Docking models with f-Val and f-
Lys
We proposed the mechanism of substrate recognition by constructing docking models of FAOD with f-Val and f-Lys, based on the available structural information of MsOX binding to its substrate. As with MsOX, the interaction of FAOD with the substrate’s carboxylate group is a major part of the substrate recognition. On the basis of the number of residues involved in the recognition, as well as their distances and binding energies, our docking models are consistent with the preference displayed by FAOD for f-Lys over f-Val at 5 mM PPB.
The role of Asn354 in substrate recognition
Our FAOD docking models also suggest that Asn354 plays a much greater role in the recognition of f-Lys than f-Val. On the basis of our docking models, we created Asn354 mutants of FAOD that showed lower reactivity toward f-Lys compared with wild type. This was particularly true for Asn354His and Asn354Lys, whose greatly reduced reactivity toward f-Lys may be because of repulsion between the positive charge of the residue and the positive charge of the -amino group of f-Lys. The Asn354Val substitution caused adverse effects with both substrates, compared with wild-type enzyme. Substitution to a valine residue increased the hydrophobicity of the active site, thus affecting the recognition of both substrates.
Surprisingly, the Asn354His and Asn354Lys mutants also showed enhanced activity (VmaxKm–1) toward f-Val, resulting in greatly increased substrate specificity for f-Val. The histidine and lysine residues may play a role in the recognition of the substrate’s carboxylate via electrostatic interactions, thus effectively stabilizing the charged substrate. Our results strongly agree with the important role of Asn354 in f-Lys recognition and help validate our docking models.
The effects of buffer concentration
Increases in buffer concentrations generally suppress electrostatic interactions and emphasize hydrophobic interactions. Wild-type FAOD showed a much greater increase in Km value for f-Lys than for f-Val when the PPB concentration was increased from 5 to 500 mM. These changes in reactivity suggest that recognition of f-Val has a greater tendency to be controlled by hydrophobic interactions, whereas f-Lys recognition is more controlled by electrostatic interactions.
The Asn354His and Asn354Lys mutants showed very similar trends in Km and Vmax values with changing PPB concentration as the wild-type enzyme. A major exception was the Vmax value for f-Lys, which was greater at 500 mM PPB for the wild-type FAOD and was greater at 5 mM PPB for both mutant enzymes. This suggests that Asn354 is involved in the changes in the reactivity toward f-Lys in response to changes in buffer concentration, but not toward f-Val.
Many FAODs have so far been isolated, cloned and characterized. The bacterial and eukaryotic FAODs form two distinct groups based on their primary structures. The eukaryotic FAODs can be further subdivided into three groups based on substrate specificity as follows: (i) enzymes that show a high preference for -fructosyl amino acids (e.g. FAOD from Penicillium janthinellum AKU3413 (Yoshida et al., 1995, 1996), FPOX from Eupenicillium terrenum ATCC 18 547 and FPOX from Coniochaeta sp. NISL 9330 (Hirokawa et al., 2003)); (ii) enzymes that show a high preference for -fructosyl amino acids (e.g. FAOD from Fusarium oxysporum S-1F4 (Sakai et al., 1995), FAOD from Aspergillus terreus GP1 (Yoshida et al., 1996), Amadoriase Ia,b,c from Aspergillus fumigatus (Takahashi et al., 1997) and FAOD-Ao1 from Aspergillus oryza (Akazawa et al., 2004)); (iii) enzymes that oxidize - and -fructosyl amino acids at comparable levels (e.g. Amadoriase II from A. fumigatus (Takahashi et al., 1997), FAOD-Ao2a,b from A. oryza (Akazawa et al., 2004) and N1-1 FAOD).
In this study, we identified the residues that interact with the substrate. Although most of these residues are conserved in eukaryotic FAODs, Asn354 was not conserved (Fig. 4). Comparison of different groups shows that a histidine residue is located at the position corresponding to Asn354 in all three enzymes of the first group, but not in any of the enzymes in the other two groups, which generally have an asparagine or arginine residue at that position (Fig. 4). This is consistent with the large increase in f-Val specificity observed with the Asn354His mutation. However, the presence of the positively charged arginine residue at that position indicates that f-Lys recognition is not only influenced by electrostatic interactions but also influenced by other factors.
|
For the enzymatic measurement of HbA1c or glycated albumin, they must first be subjected to proteolytic digestion to liberate f-Val or f-
Lys, respectively, followed by the determination of liberated fructosyl amino acids by FAOD. An FAOD enzyme specific for each substrate is therefore ideal, as they would each specifically measure their respective target molecule. The potential of the Asn354His and Asn354Lys mutants for the measurement of HbA1c was further confirmed by measuring their activities at 0.5 mM substrate concentrations. Under such conditions, which represent the expected concentrations of glycated amino acids from hydrolyzed whole blood, the activity resulting from f-Lys is only 2% of that of f-Val. Therefore, the mutants have a great potential for the measurement of HbA1c in whole blood, as the interference from glycated albumin would be greatly reduced. In addition to creating two mutant FAODs that are expected to be applied in the measurement of HbA1c, we have provided new and valuable information on substrate recognition of eukaryotic FAOD. This report is expected to contribute to the progress of FAOD investigations and the development of sensing systems for glycated proteins.
|
Footnotes |
|---|
Edited by Hagan Bayley
|
References |
|---|
|
TopAbstractIntroductionMaterials and methodsResultsDiscussionReferences |
|---|
Akazawa S., Karino T., Yoshida N., Katsuragi T., Tani Y. Appl. Environ. Micorobiol. (2004) 70:5882–5890.
Bookchin R.M., Gallop P.M. Biochem. Biophys. Res. Commun. (1968) 32:86–93.[CrossRef][Web of Science][Medline]
Ferri S., Miura S., Sakaguchi A., Ishimura F., Tsugawa W., Sode K. Mar. Biotechnol. (2004) 6:625–632.[CrossRef][Medline]
Ferri S., Sakaguchi A., Goto H., Tsugawa W., Sode K. Biotechnol. Lett. (2005) 27:27–32.[CrossRef][Web of Science][Medline]
Hirokawa K., Gomi K., Bakke M., Kajiyama N. Arch. Microbiol. (2003) 180:227–231.[CrossRef][Web of Science][Medline]
Horiuchi T., Kurokawa T., Saito N. Agric. Biol. Chem. (1989) 53:103–110.[Web of Science]
Iberg N., Fluckiger R. J. Biol. Chem. (1986) 261:13542–13545.
Johnson R.N., Metcalf P.A., Baker J.R. Clin. Chim. Acta (1983) 127:87–95.[CrossRef][Web of Science][Medline]
Keil P., Mortensen H.B., Christophersen C. Acta. Chem. Scand. B (1985) 39:191–193.[Web of Science][Medline]
Klenk D.C., Hermanson G.T., Krohn R.I., Fujimoto E.K., Mallia A.K., Smith P.K., England J.D., Wiedmeyer H.M., Little R.R., Goldstein D.E. Clin. Chem. (1982) 28:2088–2094.
Miura R., Setiyama C., Nishina Y., Shiga K., Mizutani H., Miyahara I., Hirotsu K. J. Biochem. (1997) 122. Tokyo. 825–833.
Miura S., Ferri S., Tsugawa W., Kim S., Sode K. Biotechnol. Lett. (2006) 28:1895–1900.[CrossRef][Web of Science][Medline]
Monnier V.M., Wu X. Biochem. Soc. Trans (2003) 31:1349–1353.[Web of Science][Medline]
Ogawa K., Stollner D., Scheller F., Warsinke A., Ishimura F., Tsugawa W., Ferri S., Sode K. Anal. Bioanal. Chem. (2002) 373:211–214.[CrossRef][Web of Science][Medline]
Sakai Y., Yoshida N., Isogai A., Tani Y., Kato N. Biosci. Biotech. Biochem. (1995) 59:487–491.[Medline]
Sode K., Ishimura F., Tsugawa W. Mar. Biotechnol. (2001) 3:126–132.[CrossRef][Medline]
Thorpe S.R., Bynes J.W. The Glycoconjugates—Horowitz M.I., ed. (1982) 3. New York: Academic Press. 113–132.
Takahashi M., Pischetsrieder M., Monnier V.M. J. Biol. Chem. (1997) 272:3437–3443.
Trickey P., Wagner M.A., Jorns M.S., Mathews F.S. Structure (1999) 7:331–345.[Medline]
Tsugawa W., Ishimura F., Ogawa K., Sode K. Electrochemistry (2000) 68:869–871.[Web of Science]
Tsugawa W., Ogawa K., Ishimura F., Sode K. Electrochemistry (2001) 69:973–975.[Web of Science]
Yoshida N., Sakai Y., Serata M., Tani Y., Kato N. Appl. Environ. Microbiol. (1995) 61:4487–4489.[Abstract]
Yoshida N., Sakai Y., Isogaki A., Fukuya H., Yagi M., Tani Y., Kato N. Eur. J. Biochem. (1996) 242:499–505.[Web of Science][Medline]
Wagner M.A., Trickey P., Chen Z.W., Mathews F.S., Jorns M.S. Biochemistry (2000) 39:8813–8824.[CrossRef][Web of Science][Medline]
Received February 16, 2007; revised August 3, 2007; accepted August 3, 2007.
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||