Effect of mutations at Glu160 and Val198 on the thermostability of lactate oxidase Hirotaka Minagawa 1 , Jiro Shimada 1 and Hiroki Kaneko 2 1 Fundamental Research Laboratories, NEC Corp., Tsukuba, Japan; 2 Department of Applied Physics, College of Humanities and Sciences, Nihon University, Tokyo, Japan We have obtained two types of thermostable mutant lactate oxidase – one that exhibited an E-to-G point mutation at position 160 (E160G) through error-prone PCR-based random mutagenesis, and another that exhibited an E-to-G mutation at position 160 and a V-to-I mutation at position 198 (E160G/V198I) through DNA shuffling-based random mutagenesis – both of which we have previously reported. Our molecular modeling of lactate oxidase suggests that the substitution of G for E at position 160 reduces the electro- static repulsion between the negative charges of E160 and E130 in the (b/a) 8 barrel structure, but a thermal-inactiva- tion experiment on the five kinds of single-mutant lactate oxidase at position 160 (E160A, E160Q, E160H, E160R, and E160K) showed that the side-chain volume of the amino acid at position 160 mainly contributes to the thermostability of lactate oxidase. We also produced V198I single-mutant lactate oxidase through site-directed mutagenesis, and ana- lysed the thermostability of wild-type, V198I, E160G, and E160G/V198I lactate oxidase enzymes. The half-life of E160G/V198I lactate oxidase at 70 °C was about three times longer 2 than that of E160G lactate oxidase, and was about 20 times longer 3 than that of wild-type lactate oxidase. In con- trast, the thermostability of the V198I lactate oxidase was almost identical to that of wild-type lactate oxidase. This indicates that the V198I mutation alone does not affect lactate oxidase thermostability, but does affect it when combined with the E160G mutation. Keywords: lactate oxidase; thermostability; site-directed mutagenesis; random mutagenesis. Lactate oxidase is widely used in biosensors to measure the lactate concentration in blood [1], and increasing the thermostability of lactate oxidase is a major factor in prolonging the life of lactate oxidase-based lactate sensors. One way to achieve this is to use naturally thermostable enzymes isolated from thermophilic bacteria; however, it is not always possible to obtain a desired enzyme (e.g. lactate oxidase) from such bacteria. An alternative method is to enhance protein stability through protein engineering [2], and many researchers have investigated the relationships between protein structure and thermostability [3,4], with some proposing a mutation strategy based on knowledge of the specific enzyme, such as ribonuclease H [5] or lysozyme [6,7], to improve enzyme stability by means of rational design. The thermostable thermolysin-like protease created through rational design by van den Burg et al. [8], was one of the most successful cases, but we still need to obtain a naturally thermostable counterpart thermolysin and deter- mine its 3D structure. The difficulty in trying to improve enzyme stability through rational design is that this requires solving the 3D structure of the enzyme, and, as the 3D structure of the lactate oxidase has not yet been clarified, the rational approach is therefore not applicable. An alternative approach to increase the enzyme thermostability is to screen mutants produced by random mutagenesis, such as error- prone PCR or DNA shuffling [9–11], for thermostable enzymes. The technique of DNA shuffling was introduced by Stemmer [9], and has become a powerful tool for protein engineering. We have already produced two types of thermostable mutant lactate oxidase – E160G [12] and E160G/V198I [13] – and have used E160G lactate oxidase to develop a long-life lactate sensor [14]. However, while we have shown that increasing the thermostability of lactate oxidase prolongs the life of lactate sensors, the thermo- stability of lactate oxidase must be further improved to enhance the practicality of its application to lactate oxidase sensors because the effect of enzyme stability tends to be weaker when the enzyme is immobilized on a membrane rather than in solution [14]. We have constructed a 3D structure of lactate oxidase through homology modeling [15]. In the case of E160G lactate oxidase, our model suggests that the Glu residue at position 160 in wild-type lactate oxidase is located in an a 3 helix, constituting part of the (b/a) 8 barrel, and the mutation of Glu160 to Gly stabilized the entire protein structure by eliminating the negative charge repulsion between Glu160 and Glu130. To confirm this hypothesis, five kinds of single-mutant lactate oxidase (E160A, E160Q, E160R, E160K, and E160H) were constructed by site-directed mutagenesis, and their thermo- stability was compared with that of wild-type and E160G mutant lactate oxidases. We also produced V198I single- mutant lactate oxidase through site-directed mutagenesis, and compared the thermostability and Michaelis constant (K m ) values of wild-type, V198I, E160G, and E160G/V198I Correspondence to H. Minagawa, Fundamental Research Laborat- ories, NEC Corp., Miyukigaoka 34, Tsukuba 305-8501, Japan. Fax: + 81 29 856 6136, Tel.: + 81 29 850 2613, E-mail: h-minagawa@ab.jp.nec.com Abbreviation: IPTG, isopropyl thio-b- D -galactoside. (Received 7 April 2003, revised 30 June 2003, accepted 14 July 2003) Eur. J. Biochem. 270, 3628–3633 (2003) Ó FEBS 2003 doi:10.1046/j.1432-1033.2003.03751.x lactate oxidases to investigate the effect of an E160G/V198I double mutation 4 on thermostability and enzyme activity. Materials and methods Construction of the mutant lactate oxidase gene The five types of single-mutant lactate oxidase at position 160 (E160A, E160Q, E160H, E160R, and E160K) were constructed by PCR-based site-directed mutagenesis using the Quick Change Site-Directed Mutagenesis Kit (Strata- gene). According to the kit protocol, the plasmid containing the mutant lactate oxidase gene was amplified by PCR with 125 ng of each mutagenic primer (forward and reverse) and 50 ng of pLODwt [12], as a template, under the following conditions: after heat denaturation at 95 °Cfor30s,we applied 16 treatment cycles, each consisting of 30 s at 95 °C, 1 min at 55 °C, and 12 min at 68 °C. The amplified plasmids were cooled on ice for 2 min and mixed with DpnIat37°Cfor1h.Then1lL of plasmid was used to transform Escherichia coli JM109 through electropora- tion, and the electroporated JM109 cells were cultured overnight at 37 °C on an L-plate (1% tryptone, 0.5% yeast extract, 0.5% NaCl, 1.5% agar, and 100 lgÆmL )1 ampicil- lin). Several transformant plasmids were sequenced and the plasmid which had the intentional mutation was selected as mutant lactate oxidase. The E160G/V198I double-mutant lactate oxidase was created by DNA shuffling [13], and the V198I single-mutant lactate oxidase was created by site- directed mutagenesis, as described above. Lactate oxidase purification and activity measurements The wild-type and all mutant lactate oxidases were purified by the method previously reported [12]. E. coli JM109 cells, harboring a plasmid containing lactate oxidase genes, were grown overnight in L-broth (1% tryptone, 0.5% yeast extract, 0.5% NaCl, and 100 lgÆmL )1 ampicillin) at 37 °C. Isopropyl thio-b- D -galactoside (IPTG) was then added to a final concentration of 1 m M and the cells were cultured for a further 2 h. The cells were then harvested and suspended in a 50-m M potassium-phosphate buffer (pH 7.0) containing 1mgÆmL )1 (w/v) lysozyme, and then stirred at room temperature for 1 h. After ultrasonic disruption of the cells, a crude extract was obtained by centrifugation. Phenyl- methanesulfonyl fluoride and EDTA were added to the supernatant as proteinase inhibitors, at final concentrations of 1 m M . 5 After precipitation by ammonium sulfate (60/80% saturation) at room temperature, the lactate oxidase was purified by stepwise column chromatography using Q-Sepharose FF, Phenyl Sepharose 6FF, and Superdex pg200 (all from Pharmacia) in the FPLC system (Pharma- cia). The purity was verified by SDS/PAGE, using PhastGel and the PhastSystem (Pharmacia). Lactate oxidase activity was determined by a peroxidase-coupled spectrophotomet- ric method [12] using H 2 O 2 as the standard. Assays were started by adding lactate oxidase to an activity assay mixture containing 1.5 m M 4-aminoantipyrine, 3.3 m M phenol, 100 m ML -lactate (lithium salt), 40 m M Hepes, and 6 U of horseradish peroxidase, in a final volume of 2 mL at pH 7.3, and then the absorbance (A) at 500 nm was monitored at room temperature. Protein concentrations were measured with a bicinchoninic Protein Assay Regent Kit (Pierce) using purified bovine serum albumin as the standard. Irreversible enzyme inactivation was measured at a protein concentration of 50 lgÆmL )1 in a 40 m M Hepes buffer, pH 7.3. Samples were incubated at 70 °Candthen transferred to ice at different time-points during incubation. Residual enzyme activity was then measured, as described above. To determine the temperature dependence of the K m value for L -lactate at pH 7.3, L -lactate concentrations were varied over a significant range (78 l M to 10 m M or 0.78 m M to 100 m M ) and assays were conducted at 15, 25, 35, and 45 °C. K m values were calculated by plotting [S]/V vs. [S], where [S] and V are, respectively, the L -lactate concentration and the initial reaction rate. Results and discussion Irreversible enzyme inactivation assay of mutant lactate oxidases at position 160 On the basis of our lactate oxidase model [15], position 160 is located in an a 3 helix constituting part of the (b/a) 8 barrel, which is a common structure and the most basic frame in the functionally important domain of a family of FMN- dependent a-hydroxy acid oxidizing enzymes. The model suggests that even the shortest distance between Glu160 and Glu130 is 6.03 A ˚ and this distance is probably a result of the electrostatic repulsion between the negative charges of these twoglutamicacids.Wehaveexplainedthatthepartial disorder is caused by this negative charge repulsion between Glu160 and Glu130; the mutation of Glu160 to Gly (E160G: which was made by error-prone PCR [12]) might therefore stabilize the entire protein structure by eliminating this charge repulsion [15]. To test this hypothesis, we changed Glu160 to Gln (E160Q: the same side-chain volume and no electric charge), His, Arg, and Lys (E160H, E160R, E160K, respectively: positive electric charge). We also created mutant Ala at position 160 (E160A), which has a side-chain volume intermediate between those of Gln and Gly. The thermal-inactivation curves of wild-type and mutant lactate oxidase at position 160 (E160G, E160A, E160Q, E160H, E160R, and E160K) at 70 °C are shown in Fig. 1: the heat-inactivation curve of the wild-type lactate oxidase exhibited a simple exponential decay, while those of mutant lactate oxidases exhibited biphasic inactivation. This suggests that at least two processes were involved in the thermal inactivation of the mutants. Similar phenomena have been previously observed for other thermostable mutants (N212D and E160G/N212D) [12] and may be explained by the multiple-exponential state in the unfolding kinetics [16]. The relationships between biphasic inactivation of the mutants and the enzyme thermostability are still not fully understood and therefore further study is required. The thermostabilities of E160H, E160R, and E160K were almost the same as that of the wild-type lactate oxidase, and the thermostabilities of E160Q and E160A were slightly higher than that of the wild-type enzyme, but lower than that of the E160G mutant. Substitution with a small, nonelectric amino acid at position 160 seems to affect the increase in lactate oxidase thermostability. In the a-helix, Gly to Ala substitution was reported to stabilize the helix by decreasing the entropy of the unfolding state [17]. In our Ó FEBS 2003 Thermostability of mutant lactate oxidase 1 (Eur. J. Biochem. 270) 3629 case, E160G was more thermostable than E160A, thus the amino acid substitution at position 160 might have affected the a 3 helix flexibility through interaction with other residues, such as Glu130 or Arg203. It is difficult to explain the thermostability mechanisms of the E160G mutation on the basis of this model structure, and the precise molecular structure of lactate oxidase would have to be known to do so. Our group and other researchers [ 6 18] have attempted to determine the 3D structure of lactate oxidase, and prelimi- nary results were obtained; however, the complete detailed structure of lactate oxidase has not yet been determined. Irreversible enzyme-inactivation assay of mutant lactate oxidases at positions 160 and 198 We created V198I single-mutant lactate oxidase by site- directed mutagenesis using a Quick Change Site-Directed Mutagenesis Kit (Stratagene) and purified the lactate oxidase, as described above. Thereafter, we compared the irreversible thermostability and the temperature dependence of the K m value with the wild-type, E160G, and E160G/ V198I lactate oxidase. The thermal-inactivation curve of the E160G/V198I mutant at 70 °C was obviously less steep than that of the E160G mutant; however, there was very little difference in the thermal-inactivation curves of the V198I mutant and wild-type lactate oxidases (Fig. 2). This indicates that the V198I mutation affects the thermostability of the E160G mutant lactate oxidase, but has a minimal effect on the thermostability of the wild-type enzyme. The temperature dependence of the wild-type, E160G, V198I, and E160G/V198I K m values for L -lactate are shown in Fig. 3. The temperature dependence of the K m values of Fig. 1. Thermal inactivation of wild-type and mutant lactate oxidases at position 160. Residual activities of wild-type (d), E160G (h), E160A (n), E160Q (s), E160H (e), E160R (,), and E160K ( ) lactate oxidases are shown, expressed as a percentage of the original activity. Enzymes in 40 m M Hepes buffer (pH 7.3) were incubated at 70 °Cfor different periods of time. Results represent the mean value of experi- ments performed in triplicate ± SD. Fig. 2. Thermal inactivation of wild-type, E160G, V198I, and E160G/ V198I lactate oxidases. Residual activities of wild-type (d), E160G (h), V198I (s), and E160G/V198I (e) lactate oxidases are shown, expressed as a percentage of the original activity. The enzymes, in 40-m M Hepes buffer (pH 7.3), were incubated at 70 °C for different periods of time. Results represent the mean value of experiments performed in triplicate ± SD. Fig. 3. Temperature dependence of the Michaelis constant (K m ) values of wild-type, E160G, V198I and E160G/V198I lactate oxidases. The K m values of wild-type (d), E160G (h), V198I (s), and E160G/V198I (e) lactate oxidases, at 15, 25, 35, and 45 °C, are shown. The concentration of L -lactate varied from 78 l M to 10 m M (wild-type and V198I) or 0.78 m M to 100 m M (E160G and E160G/V198I), and assays were conducted in 40 m M Hepes buffer at pH 7.3. 3630 H. Minagawa et al. (Eur. J. Biochem. 270) Ó FEBS 2003 E160G and E160G/V198I were almost identical, and the K m values of these two mutants were higher than that of the wild-type lactate oxidase at a temperature range of 15–45 °C; thus, the K m value and the temperature depend- enceofthewild-typelactateoxidasewerealmostthesameas for the V198I mutant lactate oxidase. In terms of the K m value, the E160G mutation greatly affected the lactate oxidase activity, but the V198I mutation had no or only a minor effect on the catalytic activity of lactate oxidase with respect to the wild-type and E160G lactate oxidases. The Fig. 4. Stereoview of the model structure of lactate oxidase: a side view of the (b/a) 8 barrel. The side-chains of Val198, Glu160, Arg203, Glu130, a pyruvate molecule and an FMN molecule are shown as a stick model. The labels of the mutation sites, Glu160 and Val198 are colored green. All the atoms of Gly126, Ala127 and Thr128 are displayed by a van der Waals surface (CPK) model. The atom-type colors are as follows: carbon (green), oxygen (red), nitrogen (blue) and phosphate (magenta). The region considered a flexible large loop structure (residues 189–215), on the basis of the structural comparison between glycolate oxidase and flavocytochrome b 2 , is shown as a blue ribbon (see also the thick lines in Fig. 5). Red cylinders and yellow arrows represent a-helixes and b-strands, respectively, with the a 2 , a 3 , a D and a E helices denoted by a2, a3, aD and aE, respectively. The secondary-structure nomenclature is according to Lindqvist [19]. The picture was created using INSIGHT II software (Version 4.3, Accelrys Inc.) on an ONYX2 workstation (Silicon Graphics, Inc.). Fig. 5. Variation of the atomic temperature factors averaged for the main-chain atoms of spinach glycolate oxidase [19]. Thedatawereobtainedfrom the Protein Data Bank (entry ID: 1GOX). The average B-factor of each residue is plotted against the residue number. Residues 172–204, in which even the main-chain structures are not superimposed between glycolate oxidase and the FMN-binding domain of flavocytochrome b 2 ,are emphasized by thick lines. Residues 189–197, which are disordered in the crystal, are shown as dotted lines. Only the a D helix, a E helix and the secondary structures included in the (b/a) 8 barrel are indicated. The left arrow indicates the value for Arg143 (corresponding to Glu160 of lactate oxidase), and the right arrow indicates the value for Lys181 (corresponding to Val198 of lactate oxidase). Ó FEBS 2003 Thermostability of mutant lactate oxidase 1 (Eur. J. Biochem. 270) 3631 specific activity of V198I lactate oxidase was approximately the same as that of wild-type lactate oxidase, and that of E160G/V198I mutant lactate oxidase was approximately the same as that of E160G (data not shown). These observations suggest that the V198I mutation only affects the thermostability of the enzyme when it combines with the E160G mutation. Why does the single E160G mutation affect the thermo- stability whereas the V198I mutation does not? According to our lactate oxidase model [15], position 160 is located in an a 3 helix constituting part of the (b/a) 8 barrel (Fig. 4), which is a common structure and the most basic frame in the functionally important domain of a family of FMN- dependent a-hydroxy acid oxidizing enzymes. The atomic temperature factors of the corresponding position (B-value ¼ 25 A ˚ 2 ) in glycolate oxidase suggest that posi- tion 160 is not flexible, whereas the average isotropic B-value for all the main-chain protein atoms is 26.5 A ˚ 2 (Fig. 5) [19]. Position 198, on the other hand, is located in a large loop (residues 189–215) 7 (Fig. 4). Here, we defined this loop as the region corresponding to the place where the main-chain structures of two homologous enzymes – glycolate oxidase [19] and the FMN-binding domain of flavocytochrome b 2 [20] – are not superimposed on the basis of the 3D structural fitting. The loop starts just after the end of the a D helix and terminates halfway along the a E helix, thus it is quite far from the core structure [i.e. the (b/a) 8 barrel]. Moreover, the atomic temperature factors of the corresponding position in glycolate oxidase (Fig. 5) suggest that the loop which includes position 198 is very flexible. We infer from this that position 160 contributes more directly to stabilizing the (b/a) 8 barrel structure than does position 198. Therefore, the mutation at position 160 should have a much greater effect on stability than the mutation at position 198. Why does the V198I mutation further increase the thermostability of the E160G mutant? Glu160 in the wild- type lactate oxidase corresponds to Arg143 in the a 3 helix of glycolate oxidase. From detailed observation of the glyco- late oxidase crystal structure [19], we have learned that Arg143 plays a very important role in maintaining the stability of the enzyme molecule [15]. The residue stabilizes theentire(b/a) 8 barrel structure by forming a strong salt bridge with the Glu114 in the a 2 helix. It also reinforces the interaction between the (b/a) 8 barrel structure and the large loop, mentioned above (residues 172–204 in glycolate oxidase and residues 189–215 in the wild-type lactate oxidase), by forming a hydrogen-bond network between the NH 2 of Arg143 and the main-chain carbonyls of Lys181 and Asn182 directly, and the main-chain carbonyl of Glu184 by way of a water molecule. However, in lactate oxidase, the exchange of the E160G mutation appears to completely eliminate the interaction between the (b/a) 8 barrel structure and the large loop that would otherwise occur through the hydrogen bond network between Glu160 in the barrel and Arg203 in the loop, in return for stabilizing the barrel structure (see Fig. 4). A Val fi Ile mutation at position 198 in the large loop may restore the interaction. Actually, according to our lactate oxidase model (Fig. 4), the extension of the side-chain generated by the Val fi Ile mutation forms van der Waals contacts with the side-chain atoms of Thr128 or the main-chain atoms of Gly126 and Ala127. As residues 126–128 are located just before the a 2 helix, it is probable that the Val fi Ile mutation at position 198 reinforces the interaction between the (b/a) 8 barrel structure and the large loop via van der Waals forces between the Ile side-chain and the side-chains of amino acids near the a 2 helix. The additive effects of mutations on enzyme stability have often been reported [21]. Hence, one strategy to improve enzyme stability is to screen several thermostabi- lized mutant enzymes that have a single amino acid mutation, and then to generate a combined mutant that has two or more amino acid mutations. We have previously applied this Ôadditive strategyÕ to improve the thermostabi- lity of lactate oxidase [12]. In that work we obtained two types of thermostable mutant lactate oxidase (N212D and E160G) through random mutagenesis and constructed a double mutant N212D/E160G lactate oxidase. The thermo- stability of that double mutant, however, was approxi- mately the same as that of E160G lactate oxidase [12]. In this case, we found that the N212D mutation affects the enzyme stability of wild-type lactate oxidase, but not that of E160G lactate oxidase. Although many researchers have found that combinations of mutations, and the resulting effects of the amino acid substitutions at different positions, are usually additive, our findings for two types of double mutations (N212D/E160G and E160G/V198I) indicate that the cooperative amino acid substitutions also have an important effect on enzyme thermostability. Acknowledgements We thank M. Kitabayashi of NEC’s Fundamental Research Labor- atories for her technical assistance in the sample preparation. References 1. Ito, N., Matsumoto, T., Fujiwara, H., Matsumoto, Y., Kayashima, S., Arai, T., Kikuchi, M. & Karube, I. (1995) Transcutaneous lactate monitoring based on a micro-planar amperometric bio- sensor. Anal. Chim. Acta 312, 323–328. 2. Bishop, B., Koay, C.D., Sartorelli, C.A. & Regan, L. (2001) Reengineering granulocyte colony-stimulating factor for enhanced stability. J. Biol. Chem. 276, 33465–33470. 3. Fa ´ ga ´ in, C.O ´ . (1995) Understanding and increasing protein sta- bility. Biochim. Biophys. Acta 1252, 1–14. 4. Lee, B. & Vasmatzis, G. (1997) Stabilization of protein structures. Curr. Opin. Biotechnol. 4, 423–428. 5. Ota, M., Kanaya, S. & Nishikawa, K. (1995) Desk-top analysis of the structural stability of various point mutations introduced into ribonuclease H. J. Mol. Biol. 248, 733–738. 6. Imoto, T. (1996) Engineering of lysozyme. EXS 75, 163–181. 7. Takano, K., Ota, M., Ogasahara, K., Yamagata, Y., Nishikawa, K. & Yutani, K. (1999) Experimental verification of the Ôstability profile of mutant proteinÕ (SPMP) data using mutant human lysozymes. Protein Eng. 12, 663–672. 8. Van den Burg, B., Vriend, G., Veltman, O.R., Venema, G., Eijsink, V.G. (1998) Engineering an enzyme to resist boiling. Proc. NatlAcad.Sci.USA95, 2056–2060. 9. Stemmer, W.P. (1994) Rapid evolution of a protein in vitro by DNA shuffling. Nature 370, 389–391. 10. Moore, J.C., Jin, H.M., Kuchner, O. & Arnold, F.H. (1997) Strategies for the in vitro evolution of protein function: enzyme evolution by random recombination of improved sequences. J. Mol. Biol. 272, 336–347. 3632 H. Minagawa et al. (Eur. J. Biochem. 270) Ó FEBS 2003 11. Giver,L.,Gershenson,A.,Freskgard,P.O.&Arnold,F.H.(1998) Direct evolution of a thermostable esterase. Proc. Natl Acad. Sci. USA 95, 12809–12813. 12. Minagawa, H., Nakayama, N., Nakamoto, S. & Kaneko, H. (1998) Lactate oxidase thermostabilized by random mutagenesis. Res. Commun. Biochem. Cell Mol. Biol. 2, 149–156. 13. Minagawa, H. & Kaneko, H. (2000) Effect of double mutation on thermostability of lactate oxidase. Biotechnol. Lett. 22, 1131–1133. 14. Minagawa, H., Nakayama, N., Matsumoto, T. & Ito, N. (1998) Development of long life lactate sensor using thermostable mutant lactate oxidase. Biosens. Bioelectron. 13, 313–318. 15. Kaneko, H., Minagawa, H., Nakayama, N., Nakamoto, S., Handa, S., Shimada, J., Takada, T. & Umeyama, H. (1998) Modeling study of a lactate oxidase–ligand complex: substrate specificity and thermostability. Res. Commun. Biochem. Cell Mol. Biol. 2, 234–244. 16. Chan, S.H. & Dill, A.K. (1998) Protein folding in the landscape perspective: Chevron plots and non-arrhenius kinetics. Proteins 30, 2–33. 17. Matthews, B.W., Nicholson, H. & Bechtel, W.J. (1987) Enhanced protein thermostability from site-directed mutations that decrease the entropy of unfolding. Proc. Natl Acad. Sci. USA 84, 6663– 6667. 18. Morimoto, Y., Yorita, K., Aki, K., Misaki, H. & Massey, V. (1998) L -lactate oxidases from Aerococcus viridans crystallized as an octamer. Preliminary X-ray studies. Biochimie 80, 309–312. 19. Lindqvist, Y. (1989) Refined structure of spinach glycolate oxidase at 2 A ˚ resolution. J. Mol. Biol. 209, 151–166. 20. Xia, Z.X. & Mathews, F.S. (1990) Molecular structure of flavo- cytochrome b2 at 2.4 A ˚ resolution. J. Mol. Biol. 212, 837–863. 21. Wells, J.A. (1990) Additivity of mutational effects in proteins. Biochemistry 29, 8509–8517. Ó FEBS 2003 Thermostability of mutant lactate oxidase 1 (Eur. J. Biochem. 270) 3633 . shown that increasing the thermostability of lactate oxidase prolongs the life of lactate sensors, the thermo- stability of lactate oxidase must be further. Effect of mutations at Glu160 and Val198 on the thermostability of lactate oxidase Hirotaka Minagawa 1 , Jiro Shimada 1 and Hiroki Kaneko 2 1 Fundamental