Modulation of activity of NADH oxidase from Thermus thermophilus through change in flexibility in the enzyme active site induced by Hofmeister series anions Gabriel Z ˇ olda ´ k 1 , Mathias Sprinzl 2 and Erik Sedla ´ k 1 1 Department of Biochemistry, Faculty of Sciences, P. J. S ˇ afa ´ rik University, Kos ˇ ice, Slovakia; 2 Laboratorium fu ¨ r Biochemie, Universita ¨ t Bayreuth, Germany The conformational dynamics of NADH oxidase from Thermus thermophilus was modulated by the Hofmeister series of anions (H 2 PO 4 – , SO 4 2– , CH 3 COO – , Cl – , Br – ,I – , ClO 4 – , SCN – ) in the concentration range 0–3 M . Both cha- otropic and kosmotropic anions, at high concentration, inhibit the enzyme by different mechanisms. Chaotropic anions increase the apparent Michaelis constant and decre- ase the activation barrier of the reaction. Kosmotropic ani- ons have the opposite effect. Anions from the middle of the Hofmeister series do not significantly affect the enzyme acti- vity even at high concentration. We detected no significant changes in ellipticity of the aromatic region in the presence of the anions studied. There is a decreased Stern–Volmer quenching constant for FAD fluorescence quenching in the presence of kosmotropic anions and an increased quench- ing constant in the presence of chaotropic anions. All of this indicates that active site flexibility is important in the function of the enzyme. The data demonstrate that both the high rigidity of the active site in the presence of kosmotropic anions, and its high flexibility in the presence of chaotropic anions have a decelerating effect on enzyme activity. The Hofmeister series of anions proved to be suitable agents for altering enzyme activity through changes in flexibility of the polypeptide chain, with potential importance in modulating extremozyme activity at room temperature. Keywords: activation; conformational dynamics; flavopro- teins; NADH oxidase; Thermus thermophilus. The native conformation of an enzyme is produced by the complex interaction of van der Waals interactions, hydro- gen bonds and ionic interactions. These interactions produce stability of the enzyme under physiological condi- tions and prevent deleterious conformational changes from perturbations in the environment that would cause deacti- vation. These interactions, however, must not result in protein rigidity because the enzyme active site requires flexibility for optimal catalytic function. The balance of these two tendencies is sensitively adjusted for the physio- logical conditions at which the enzyme works. Examples of such adjustments are enzymes from hyperthermophiles and psychrophiles which have optimal activity at high (> 80 °C) and low (< 20 °C) temperatures, respectively [1,2]. Enzymes from thermophiles are almost inactive at room temperature because of polypeptide and side chain rigidity induced by higher-order interactions within secon- dary and tertiary structures. Psychrophilic enzymes are inactive at room temperature because the high flexibility of their polypeptide and side chains results in partial/local or complete unfolding of the tertiary structure. Modulation of the balance between the rigidity and flexibility of the polypeptide and side chains can be achieved by changing the solvent properties. Stabilization of psychrophilic enzymes without affecting their activity, or activation of thermophilic enzymes without affecting their stability, is interesting for both basic and applied protein chemistry. The use of chaotropic agents (urea, guanidinium hydrochloride) to activate different enzymes has been reported in several papers [3–8]. The change in activity resulted from conformational changes in the tertiary and secondary structure of the enzymes studied. We have shown recently that it is possible to activate NADH oxidase from Thermus thermophilus with urea without affecting the global stability of the enzyme at room temperature [8a]. NADH oxidase (EC 1.6.99.3) from T. thermophilus is a dimeric flavoprotein containing one molecule of FMN or FAD per 25 kDa monomer, which catalyzes hydride transfer from NADH to an acceptor such as FAD, ferricyanide and oxygen [9]. It belongs to the flavin reductase/nitroreductase family, which has similar broad substrate specificity, fold and quaternary structure [10,11]. Localization of the active site of NADH oxidase at the edge of the dimeric interface (Fig. 1) is in agreement with the fact that the active sites of enzymes are usually the most labile part of the enzyme structure [12]. Perturbation of either the static or dynamic state of the active site may lead to significantly changed activity. Previous studies in our laboratory have indicated that activation of NADH oxidase is not achieved by conformational change but is a result of the increased dynamics of the polypeptide/side chain in the enzyme active site. To substantiate these observations and analyze the role of dynamics in enzyme Correspondence to E. Sedla ´ k, Department of Biochemistry, Faculty of Sciences, P. J. S ˇ afa ´ rik University, Moyzesova 11, 041 54 Kos ˇ ice, Slovakia. Fax: + 421 55 622 21 24, Tel.: + 421 55 622 35 82, E-mail: sedlak_er@saske.sk Enzyme: NADH oxidase (EC 1.6.99.3). (Received 26 August 2003, revised 8 October 2003, accepted 28 October 2003) Eur. J. Biochem. 271, 48–57 (2004) Ó FEBS 2003 doi:10.1046/j.1432-1033.2003.03900.x activity, we have investigated the effect of the Hofmeister series of anions on the activity of NADH oxidase from T. thermophilus. The crystal structure provides information about the flexibility of a given structure by comparison of temperature B factors. Temperature B factors are atomic mean square displacements obtained from the intensity of the diffractive spots [13]. The absolute value of the B factor is dependent on the refinement method and the conditions of crystalliza- tion [14]. It is therefore only correct to compare B factors within a particular structure, although such data must also be handled with caution. Data from the crystals are averaged over crystal space and time, therefore they reflect crystal defects, static disorders and other parameters [15]. NADH oxidase has an overall low temperature factor for the whole structure (% 23 A ˚ 2 ) [9] that is in accordance with the high stability of the protein conformation (Fig. 1). The flavin moiety, with a low B factor, indicates rigidity and strong binding to the protein matrix. Trp47, the only tryptophan residue located in close proximity to the flavin cofactor (within 10 A ˚ ), is almost parallel to the isoalloxazine ring, but the elevated temperature factor indicates it has high flexibility. The indole ring is stabilized through hydrophobic interactions (side chains of Ala46, Leu49, Phe120, Ala121, Ala122, Met123) from helix F. The crystal structure of a homologous nitroreductase in various states shows that binding of the substrate (nicotinic acid) is accompanied by the induced fit of helix F and helix E [10]. Rearrangement of helix F during the binding event results in a change in the torsional angle of several residues. Remarkably, the residues that are involved in substrate binding through changes in their dihedral angles are those with the highest temperature factor, and are mostly from helix F. Similarly, the high B factors of helix F indicate that it is highly flexible in NADH oxidase (Fig. 1). Stabilization or destabilization of this helix would affect interactions with Trp47 and thus the opening of the active site, which is necessary for activity (unpublished observation). This would indicate a mechanism of interaction of NADH substrate with the enzyme common to this flavoenzyme family. The Hofmeister series of anions were chosen as suitable candidates for stabilization/destabilization of this part of NADH oxidase. There are numerous reports on the effect of the Hofmeister series of salts on folding and stability of proteins [16–18] and enzyme activity in both aqueous solutions [19–26] and organic solvents [27]. It is generally accepted that the effect of these salts on protein results from interactions of the salt with the polypeptide chain (enthalpic contribution) and, indirectly, through effects on the water structure (entropic contribution) [28–36]. For our study, we chose the Hofmeister series of anions: H 2 PO 4 – , SO 4 2– , CH 3 COO – , Cl – , Br – ,I – , ClO 4 – , SCN – (ordered from kosmotropic to chaotropic). Anions are more efficient than cations in affecting the properties of polypeptide chains. The anion–water interaction is stronger than the cation–water interaction, thus anions have a greater effect on water ordering. The explanation for this is the asymmetry of the charge in a water molecule, with the negative end of the molecule’s dipole being nearer the center than the positive end [34,36]. The modulation of the conformational dynamics of the enzyme by the Hofmeister anions enabled us to show that both stabilization and destabilization of the active site of NADH oxidase by kosmotropic and chaotropic anions, respectively, inhibits its activity. Application of the Hof- meister series of anions may be a suitable approach to modifying properties of enzymes from extremophiles. The work presented is the result of a continuation of our interest in understanding the role of flexibility for catalytic efficiency of enzymes. NADH oxidase from T. thermophilus is a good candidate for such a study because the flexibility of its polypeptide chain is adjusted to the harsh conditions of thermophilic bacteria. Therefore, the addition of chaotropic agents at room temperature will not significantly perturb the enzyme’s global structure [8a] but will modulate the flexibility of most of its labile parts, i.e. the part of the protein structure where the active sites are usually located [9]. Experimental procedures Analytical-grade biochemicals were obtained from Merck (Germany). Urea (high purity grade) was purchased from Sigma. Urea concentrations were determined from refract- ive index measurements using an Abbe Refractometer AR3- AR6. The pH values of the solutions were measured with a Sensorex glass electrode before and after measurement at room temperature. Only the measurements at which the pH change was less than 0.2 pH unit were used. Protein expression and purification The NADH oxidase from T. thermophilus was overpro- duced in Escherichia coli JM 108. The purification proce- dure for the overproduced NADH oxidase was described Fig. 1. Homodimeric structure of NADH oxidase from T. thermophilus colored according to temperature B factor. Low B factor (< 15 A ˚ 2 rigid structure) is shown by a dark blue color, intermediate B factor (30–45 A ˚ 2 ) by green/yellow, and high B factor (> 60 A ˚ 2 ) by red. Flavin cofactor and the closest tryptophan, Trp47, are shown. The thin line indicates dimeric interface. The isoalloxasine ring of flavin is localized in the rigid part of the homodimer, and Trp47 is localized on the most flexible a-helix of the protein structure, helix F (shown within elliptical traces). Ó FEBS 2003 Anion effect on activity of NADH oxidase (Eur. J. Biochem. 271)49 previously [37]. The final product provided a single band on a SDS/polyacrylamide gel stained with Coomassie Brilliant Blue. Before use, the protein was dialyzed against 5 m M phosphate buffer, pH 7.0, in the absence of FAD. The specific activity of NADH oxidase is % 1.9 UÆmg )1 in 5 m M phosphate buffer, pH 7.0. Steady-state kinetics All kinetic measurements were performed on a Shimadzu UV3000 spectrophotometer. The kinetic parameters were determined from the initial decrease in NADH absorbance at 340 nm (e 340nm ¼ 6220 M )1 Æcm )1 ), at 20 °C. Measure- ments were performed after incubation (12 h) in 120 n M NADH oxidase, 5 m M phosphate, pH 7.0, containing 0.12 m M FAD and different concentrations of salts. The reaction was started by the addition of 180 l M NADH. To determine the K m value the concentration of NADH was varied in the range 5–200 l M . It is not possible to use NADH at higher concentrations because of its large absorbance. The data were fitted to the Michaelis equation: m ¼ V max ½NADH ½NADHþK NADH m;app ð1Þ where, K m, app is the apparent Michaelis constant and the apparent V max is the maximum velocity for the catalytic reaction. The experimental data were also plotted according to the Lineweaver–Burk equation and analyzed by linear regression. Similar results were obtained using both meth- ods. The apparent k cat was determined as V max /[E] 0 , where [E] 0 is the total concentration of NADH oxidase in solution. Determination of the Michaelis–Menten parameters has not been possible in the presence of some concentrations of iodine anions because of a spectral overlap of iodine (product of the peroxide and iodide) and NADH. At high concentrations of rhodanide, perchlorate, sulfate and phos- phate, the activity of NADH oxidase is very low and determination of the Michaelis–Menten constants has large errors. Temperature dependence of enzyme activity Enzyme activity was determined in 5 m M phosphate buffer containing 0.12 m M FAD and 120 n M NADH oxidase. Reactions were started by the addition of NADH to achieve a final concentration of 0.18 m M NADH. Initial velocities were measured in the range 20–40 °C. Temperature during measurements was kept constant by temperature controlled water circulation around the cuvette. Temperature depend- encies were analyzed with a simple Arrhenius equation lnk cat ¼ E a RT þ C 1 ð2Þ where, R is the gas constant (8.314 JÆK )1 Æmol )1 ), E a is the activation energy for the observed reaction, and C 1 is a temperature-independent constant. At least five values were plotted as ln (k cat )vs.T )1 and analyzed by linear regression. Coefficients of linearity were typically higher than 0.98. From comparison of the Arrhenius equation and the transition state theory, the enthalpy (DH*) and entropy (DS*) of activation were calculated DH Ã ¼ E a À RÁT ð3Þ T Á ln k cat T  ¼ T Á DS Ã R þ C 2 ð4Þ C 2 is the temperature-independent constant. Thefreeenergyofactivation(DG*) was calculated from the equation: DG Ã ¼ DH Ã À TDS Ã ð5Þ Fluorescence emission spectroscopy The fluorescence steady-state measurements were per- formed on a Shimadzu RF5000 spectrofluorophotometer. The fluorescence spectrum of tryptophan residues was obtained on excitation at 295 nm. The cuvette contained 5m M sodium phosphate, pH 7.0, with various concentra- tions of salts and 2.4 l M dimeric protein in a total volume of 2.5 mL. Fluorescence measurements were performed at 20 °C. Temperature was kept constant (± 0.3 °C) by temperature controlled water circulation. Quenching of FAD fluorescence The fluorophores in NADH oxidase make it possible to perform fluorescence quenching experiments to investigate the dynamics of the environment near the fluorophore and the accessibility of the fluorophores to solvent. Tryptophan moieties are widely used in quenching experiments. NADH oxidase contains four tryptophans at different positions, which complicates a detailed analysis. The flavin cofactor is another fluorophore that could be used as an intrinsic probe quenched by externally added quenchers, e.g. iodide and rhodanide anions. The commonly used noncharged quen- cher acrylamide is not an efficient quencher of FAD fluorescence. The FAD fluorescence is not affected even at relatively high (0.2 M ) concentrations of acrylamide. Fluor- escence quenching of the FAD was performed using iodide anions (KI). Stock solution (4 M KI in 5 m M phosphate buffer, pH 7.0) was freshly prepared to avoid oxidation of iodide [38]. Sodium dithionite could not be used in the stock solution of KI (inhibition of iodine formation) because of concomitant changes in the redox state of the flavin. As it is a single population of the FAD, it is possible to use a simple Stern–Volmer equation: F 0 F ¼ 1 þ K sm ½KIð6Þ where, K SV is the Stern–Volmer quenching constant. Comparison of the values of K SV allows us to assess the accessibility of the FAD cofactor and, indirectly, the dynamics of its environment. Fluorescence was monitored at 525 nm after excitation by 450 nm in the absence (F 0 )and presence of various concentrations of KI (F). The linearity of the experimental data (coefficient of linearity r % 0.99) confirms the validity of the simple model (Eqn 6). CD measurements CD measurements were performed on a Jasco J-810 spectropolarimeter (Jasco, Tokyo, Japan) at 20 °Cwith 50 G. Z ˇ olda ´ k et al.(Eur. J. Biochem. 271) Ó FEBS 2003 27.4 l M NADH oxidase in 50 m M sodium phosphate, pH 7.0, and at different concentrations of salt. A 1 cm path- length cuvette was used for the aromatic region. Each spectrum was an accumulation of 10 consecutive scans. Results The parameters characterizing the activity of NADH oxidase, i.e. the apparent rate constant, k cat , and the apparent Michaelis constant, K m , strongly depend on the ionic strength of the solution. Increasing the ionic strength from 5m M to 50 m M potassium phosphate results in a sixfold increase in the k cat value, from 1.1 to 6.6 s )1 , and a slight decrease in the K m value, from 8.5 to 5.2 l M . In Table 1, it can be seen that NADH oxidase is nonspecifically activated by increased ionic strength, as all the salts studied at 0.5 M induced an increase in the k cat value of the enzyme. However, a further increase in ionic strength enabled us to distinguish the effects of the different anions. Anions from the middle part of the Hofmeister series, Br – , Cl – , CH 3 COO – , without significant chaotropic or kosmotropic properties did not affect the value of k cat even at high concentrations. On the other hand, both chaotropic and kosmotropic anions caused adecreaseinthek cat value with increased concentration. As confirmed by parallel experiments with KCl and NaCl that provided identical results within the margin of error, cations do not have an effect on the kinetic parameters of NADH oxidase. Figure 2 shows the relative activity of the enzyme in the presence of 1 M and 2 M salt concentrations. Whereas the apparent k cat decreased in the presence of both chaotropic and kosmotropic anions, the apparent K m significantly increased in the presence of chaotropic anions and decreased in the presence of kosmotropic anions (Table 1). It should be noted that the real k cat (k cat real ) is underestimated when the substrate concentration is lower than 10 · K m .Fromthe Michaelis–Menten equation (Eqn 1), we know that in the presence of [S] ¼ 10 · K m the apparent k cat is related to the real catalytic rate, as k cat real ¼ 11/10 · k cat .Inthe presence of high concentrations (> 0.5 M ) of chaotropic salt, the substrate (NADH) concentration [S] is related to K m as [S] ffi 3K m (Table 1). In this case, k cat real is related to k cat as k cat real ffi 4/3k cat . However, at high concentrations of chaotropic salt, the absolute value of k cat is % 7 times lower than in the presence of neutral salt. Thus k cat real in chaotropic salts is related to k cat real in neutral salts as: k cat real chaotrop / k cat real neutral ¼ (4/3) · (1/7), i.e. significantly less than 1. Therefore, even if K m increases by % 2–3-fold, the bell shape of k cat real (the relative values of k cat real chaotrop /k cat real neutral ) will not be significantly affected. As the result of decreased conformational dynamics, enzymes from thermophiles have very low activity at low temperatures [39]. The protein dynamics and thermal stability are inversely related to each other [40,41]. The dependence of the enzyme activity on temperature in the presence of the salts was investigated to assess how the conformational dynamics of the active site is dependent on the type of salt present (Fig. 2). Figure 3 shows the temperature dependence of the relative rate constant k cat in the presence of 2 M salt. For each salt, k cat at 20 °Cwas Table 1. Apparent rate constant (k cat ), Michaelis constants (K m ) and their ratio r at various concentrations of the salts. Assays were performed using 120 n M enzyme and 0.12 m M FAD in 5 m M potassium phosphate buffer and the given concentration of salts, pH 7.0 at 20 °C. The reaction was started by the addition of 0.18 m M NADH in the absence of salts. Apparent k cat ¼ 1.10 ± 0.11 s )1 , K m ¼ 8.5 ± 0.9 l M and catalytic efficiency k cat /K m ¼ r ¼ 1.30 · 10 5 M )1 Æs )1 . Errors in determination of k cat and K m are within 10%. This value was calculated from several (2–5) independent measurements. ND, Not determined. Anion 0.5 M 1.0 M 1.5 M 2.0 M k cat (s )1 ) K m (l M ) r · 10 )5 ( M )1 Æs )1 ) k cat (s )1 ) K m (l M ) r · 10 )5 ( M )1 Æs )1 ) k cat (s )1 ) K m (l M ) r · 10 )5 ( M )1 Æs )1 ) k cat (s )1 ) K m (l M ) r · 10 )5 ( M )1 Æs )1 ) SCN – 5.36 15.60 3.43 1.00 30.55 0.33 0.78 ND ND 0.49 ND ND ClO 4 – 5.10 15.65 3.26 0.81 44.22 0.18 0.54 ND ND 0.27 ND ND I – 7.18 29.34 2.45 6.97 ND ND 3.86 ND ND 1.29 ND ND Br – 7.61 20.90 3.64 6.75 21.71 3.11 6.22 22.51 2.76 5.36 28.14 1.91 Cl – 5.01 13.25 3.78 7.50 13.67 5.55 6.64 13.72 4.84 6.43 13.40 4.80 CH 3 COO – 4.82 12.86 3.75 4.82 10.05 5.80 5.36 18.94 2.83 4.95 28.12 1.76 SO 2À 4 2.68 11.25 2.40 2.19 7.23 3.03 1.07 ND ND 0.86 ND ND H 2 PO 4 – 3.32 13.67 2.43 2.14 10.45 2.05 1.18 8.94 1.32 0.38 ND ND Fig. 2. Relative activity of NADH oxidase from T. thermophilus in the presence of 1 M (gray histogram) and 2 M (black) sodium or potassium salts of the designated anions, in 5 m M phosphate buffer, pH 7.0, at 20 °C. Activity was initiated by the addition of 0.2 m M NADH. Ó FEBS 2003 Anion effect on activity of NADH oxidase (Eur. J. Biochem. 271)51 taken as the reference value. Figure 3 shows that the slope of the observed dependencies increases according to the position of the anions in the Hofmeister series, in the order from chaotropic to kosmotropic anions. This indicates that, in the presence of chaotropic anions (SCN – ,ClO 4 – )the activation energy is temperature independent, whereas in the presence of kosmotropic anions (SO 4 2– ,H 2 PO 4 – )itis strongly temperature dependent. To determine how activation parameters are affected in the presence of various concentrations of different salts, the temperature dependencies of the rate constants were measured at 20–40 °C (Supplementary material). Figure 4 shows a dependence of DG*, at 20 °C, on the concentration of perchlorate, chloride and sulfate anions. In the range 0.5–1.0 M salt, there is a minimum of this dependence for all anions studied. Whereas in the presence of chloride (neutral) anions, the dependence achieves a local minimum, in the case of both sulfate (kosmotropic) and perchlorate (chao- tropic) anions, the observed minimum is global. It should be noted that, although the observed minima are not pro- nounced, a similar tendency of DG* is observed for all anions, indicating that the observed dependencies are real. A double minimum or wide minimum, in the range 0.5– 2.0 M salt, of DG* vs. concentration is also observed for bromide, iodide and acetate anions, i.e. anions from the middle part of the Hofmeister series. The wide minimum in the case of these anions also supports the relative independ- ence of k cat on the salt concentration (Table 1). Only one minimum and one relatively sharp maximum activity of NADH oxidase is observed for both chaotropic and kosmotropic anions. The DG* and k cat dependencies correlate in this sense that the minimum of DG* is located at a similar (same) concentration range as the maximum of k cat for each given anion. To demonstrate that the observed changes in enzyme activity are related to conformational changes in the active site, we analyzed the CD spectra of the peptide (data not shown) and aromatic regions (Fig. 5). The CD spectrum of NADH oxidase in the aromatic region consists of a positive Fig. 3. Dependence of relative activity of NADH oxidase from T. thermophilus on temperature in the presence of 2 M sodium or potassium salts of the following anions: H 2 PO 4 – (j), SO 4 2– ( ~ ), Cl – (d), Br – (.), I – (e), CH 3 COO – (h), ClO 4 – (,), SCN – (r)in5mM phosphate buffer, pH 7.0, at 20 °C. Fig. 4. Dependence of activation free energy (DG*) of the reaction catalyzedbyNADHoxidasefromT. thermophilus at 20 °Cinthe presence of 2 M NaCl (d), NaClO 4 (,), or Na 2 SO 4 ( ~ ), in 5 m M phosphate buffer, pH 7.0. Fig. 5. CD spectra of NADH oxidase from T. thermophilus in the aromatic region in the presence of 2 M NaCl (dashed line), NaClO 4 (dotted line), or Na 2 SO 4 (thick solid line) and in the absence of salts (thin solid line), in 5 m M phosphate buffer, pH 7.0, at 20 °C. Inset: Nor- malized tryptophan fluorescence (excitation wavelength 295 nm) of NADH oxidase from T. thermophilus in the aromatic region in the presence of 2 M NaCl (dashed line), NaClO 4 (dotted line), ans Na 2 SO 4 (thick solid line) and in the absence of salts (thin solid line), in 5 m M phosphate buffer, pH 7.0, at 20 °C. 52 G. Z ˇ olda ´ k et al.(Eur. J. Biochem. 271) Ó FEBS 2003 band at % 265 nm and negative ellipticity at 286 nm. NADH oxidase contains four tryptophan residues in positions 47, 52, 131, 204. Trp131 and Trp204 are completely exposed to the solution, whereas Trp52 is rigidly embedded in the protein matrix. Trp47 is in a sandwich-like position toward the flavin cofactor at a distance of about 7.7 A ˚ . This is the only tryptophan residue suitably located for interaction with the flavin. Interestingly, this position between Trp47 and the flavin cofactor can be achieved only in the dimeric form of the enzyme [9]. In accordance with previously published CD spectra of the flavin oxidases [42], the pronounced peak at 265 nm may result from an asymmetric environment of the tightly bound flavin cofac- tor and/or Trp47 in the active site of the enzyme, and Trp52. The small negative ellipticity at 286 nm corresponds to the signal of tryptophan residues. The CD spectrum of NADH oxidase in the peptide region is not significantly perturbed, even at high ionic strength (data not shown). Similarly, the spectrum of the enzyme in the aromatic region in the presence of 2 M anions is only slightly affected. A slight decrease in the positive ellipticity at % 265 nm in the presence of perchlorate anions, i.e. a decrease in the asymmetry of the tryptophan residue and/or the flavin cofactor in the active site, may result from dissociation of the flavin cofactor in the presence of nucleophilic agents [42]. A 24 h dialysis of NADH oxidase in the presence of 2 M perchlorate anions did not cause dissociation of the flavin cofactor (data not shown). The CD spectrum of the enzyme in the aromatic regions therefore probably reflects a slight change in either the conformation or the dynamics of the tryptophan residue in the active site. Fluorescence is the other very sensitive method of monitoring changes in the environment close to the fluorophores. As shown in the inset of Fig. 5, the presence of 2 M chloride or 2 M sulfate causes a change in fluores- cence as compared with low ionic strength. The fluorescence of NADH oxidase is decreased by % 40% in the presence of 2 M perchlorate anions. Interestingly, a similar decrease in the fluorescence of NADH oxidase was also observed at the concentration of urea at which activation of the enzyme occurred (unpublished observation). A decrease in fluores- cence further confirms that the flavin cofactor does not dissociate from the enzyme. Close localization of Trp47 and the flavin cofactor causes resonance energy transfer, result- ing in partial quenching of the tryptophan fluorescence; therefore, dissociation of the flavin would be accompanied by an increase in tryptophan fluorescence. The results presented indicate a close relationship between enzyme activity and the stability/conformational flexibility of the active site. The diminished enzyme activity in the presence of a high concentration (> 1 M )of kosmotropic and chaotropic anions probably reflects high stability/rigidity and too much flexibility of the active site, respectively. We observed that NADH oxidase from T. thermophilus at room temperature is activated 2.5-fold in the presence of 1.0–1.5 M urea. This activation is probably caused by increased conformational dynamics of the side chains in the active site in the presence of urea. If this suggestion of a role for flexibility in enzyme activity is correct, NADH oxidase, in the presence of kosmotropic anions (H 2 PO 4 – , SO 4 2– ) should be activated at higher urea concentrations than in the presence of neutral and chao- tropic anions. The experiments presented in Fig. 6 support this suggestion. In the presence of phosphate and sulfate anions, NADH oxidase is more than 2.5 and 3.5 times more active, respectively, in the presence of urea than without urea (Fig. 6). No activation, but relatively strong inhibition, was observed in the presence of the chaotropic anions, ClO 4 – and SCN – , as a result of increased concentrations of urea (Fig. 6). No significant effect of urea (up to 2 M )onthe ellipticity of NADH oxidase in the aromatic region in 2 M sulfate, chloride and perchlorate anions (data not shown) further indicates that changes in NADH oxidase activity are not the result of pronounced conformational change but are probably due to changes in the dynamics of protein structure. Finally, the effect of anion-induced changes in the dynamics of the FAD microenvironment was further studied by FAD fluorescence quenching using KI (Fig. 7). The quenching of FAD fluorescence monitored at 525 nm in the presence of anions of the Hofmeister series strongly indicates a changed flexibility of the flavin cofactor environment. The effect of rhodanide, iodide and bromide anions on the dynamics of the enzyme active site was not investigated because these anions very efficiently quench FAD fluores- cence. As acrylamide is a weak quencher of FAD fluores- cence, we used the efficient quenching property of iodide anions for these measurements. The maximum concentra- tion of KI in quenching experiments was 0.15 M , i.e. the concentration of iodide anions at which no significant conformational change in NADH oxidase was observed. Monitoring FAD fluorescence quenching is more advanta- geous than monitoring tryptophan fluorescence quenching because there is only one flavin cofactor and it is located in the active site of NADH oxidase. The slope of the depend- encies of F 0 /F vs. quencher in the presence of 2 M chaotropic anions is significantly higher than in the presence of 2 M neutral anions. The higher quenching constant (Eqn 6) indicates an increase in the dynamics of the flavin cofactor in the presence of the chaotropic anions. Analogously, a Fig. 6. Dependence of the relative activity of NADH oxidase from T. thermophilus on [urea], in the presence of 2 M NaH 2 PO 4 (j), Na 2 SO 4 ( ~ ), NaCl (d), KI (e), NaClO 4 (,), or KSCN (r)in5m M phosphate buffer, pH 7.0, 20 °C. Ó FEBS 2003 Anion effect on activity of NADH oxidase (Eur. J. Biochem. 271)53 decrease in the slope of the dependencies in the presence of kosmotropic salts, compared with neutralsalts, indicates that the active site of NADH oxidase is more rigid. Discussion NADH oxidase from T. thermophilus has, like other enzymes from thermophiles, low activity at room tempera- ture. We have recently shown that the enzyme is activated in the presence of a relatively low concentration (% 1 M )of chaotropic agents such as urea and guanidinium hydro- chloride (unpublished observation). The observed activation was not due to a conformational change but was a result of increased conformational dynamics in the active site. The tightly bound structural water between Trp47 and the flavin cofactor [9] was probably released in the presence of chaotropic agents, and the active site of the enzyme opened, facilitating the arrival of the substrate and leading to an increased rate constant and an increased Michaelis constant. To test this suggestion, we investigated the effect of anions of the Hofmeister series. The Hofmeister series of anions can be divided into chaotropic anions, which salt-in the peptide groups, and kosmotropic anions, with a tendency to salt-out nonpolar groups [32]. The difference in the effect of chaotropic and kosmotropic anions is also due to a charge density that affects anion interactions with water molecules [34]. The combination of these effects led to the relatively surprising bell shaped dependence of NADH oxidase activity vs. 1 M and 2 M anions, ordered according to the Hofmeister series (Fig. 2). In fact, reports dealing with the effect of the Hofmeister series of anions on enzyme activity usually show a monotone trend, i.e. enzymes are activated by chaotropic or kosmotropic anions and inhibited by the opposite anions [21–24]. Analysis of the bell shaped curve showed that the decrease in the rate constant in the presence of chaotropic anions corresponded to an increase in the apparent K m , whereas the decrease in k cat in the presence of kosmotropic anions corresponded to the decrease in K m (Table 1). The apparent Michaelis constant measures the binding affinity of the enzyme for the substrate and can also be used as an indirect measure of either inherent flexibility of an enzyme molecule [43] or the conformational state of the active/binding site. The salts used, even at 2 M , did not significantly affect the CD spectrum of NADH oxidase in the peptide and aromatic regions. This indicates (a) a strong interaction of the flavin cofactor with the protein matrix even in conditions that lead to the dissociation of the cofactor from certain mesophi- lic flavin oxidases [42], and, more importantly, (b) the unchanged conformational state of the enzyme under the conditions studied. The different dynamics of the enzyme active site in the presence of kosmotropic and chaotropic anions is indicated by: (a) a strong dependence of k cat vs. temperature in kosmotropic anions, and a nearly independ- ent k cat vs. temperature in chaotropic anions (Fig. 3) and (b) positive and negative activation entropy in kosmotropic and chaotropic anions, respectively. Moreover, a decrease in tryptophan fluorescence in the presence of perchlorate anions and slight changes in the CD spectra (Fig. 5) indicate increased dynamics of the tryptophan residue in the active site of the enzyme, similar to results in the presence of % 1.0 M urea. An analogous decrease in ellipticity in the aromatic region accompanied by changes in tryptophan fluorescence of the nonhomologous flavoprotein flavodoxin from Desulfovibrio vulgaris, in the presence of phosphate anions, was interpreted as an increase in the dynamics of the tryptophan residue in the vicinity of the flavin cofactor [44]. A stronger temperature dependence of k cat in the presence of kosmotropic anions indicates the presence of an energy barrier, i.e. the difference between the basic and transition states. On the other hand, the near independence of k cat on temperature in the presence of chaotropic anions indicates that the anions have a similar effect for temperature because the energy difference between the basic and transition states is small. This is in agreement with findings that chaotropic anions destroy the natural hydrogen-bonded network of water with effects similar to increased temperature or pressure [31], with a probable effect on the dynamics of the polypeptide/side chains of enzymes. A noteworthy observation is the linear dependence of activation enthalpy on activation entropy, a phenomenon known as entropy/enthalpy compensation, in the reaction catalyzed by NADH oxidase in the presence of salts in the concentration range 0.5–1.5 M (Fig. 8). It is apparent from these data that chaotropic (SCN – , ClO 4 – ,I – ) and kosmo- tropic (SO 4 2– ,H 2 PO 4 – ) anions are localized at opposite ends of the linear dependence, and neutral anions (Br – , Cl – , CH 3 COO – ) are in the middle of the dependence. This also indicates that chaotropic and kosmotropic salts have Fig. 7. Dependence of FAD fluorescence in the presence of 2 M NaH 2 PO 4 (j), Na 2 SO 4 ( ~ ), NaCl (d), CH 3 COONa (h), or NaClO 4 (,) on concentration of iodide anions expressed as dependence of F 0 /F vs. concentration of iodide anions. Fluorescence was monitored at 525 nm after excitation at 450 nm in the absence (F 0 ) and presence of various concentrations of KI (F). Steeper dependence indicates that, in the presence of the given salt, the accessibility of the flavin cofactor or efficiency of quenching of FAD fluorescence is higher than depend- encies with less steep slopes. The numerical value of the slope of the dependencies is an expression of the Stern–Volmer quenching constant (Eqn 6; coefficients of linearity for all of displayed dependencies were r ‡ 0.99). The K SV values for the salts studied were 5.97 ± 0.26 M )1 for NaH 2 PO 4 , 5.96 ± 0.36 M )1 for Na 2 SO 4 , 14.77 ± 0.26 M )1 for NaCl, 11.76 ± 0.16 M )1 for CH 3 COONa, and 25.06 ± 0.60 M )1 for NaClO 4 . All measurements were performed in 5 m M phosphate buffer, pH 7.0, at 20 °C. 54 G. Z ˇ olda ´ k et al.(Eur. J. Biochem. 271) Ó FEBS 2003 different effects on the dynamic state of the enzyme. As the enzyme catalyzes the same reaction in the presence of chaotropic and kosmotropic anions, the negative value of activation entropy in chaotropic salts indicates a higher flexibility of the basic state compared with the transition state. In other words, the difference between the activation parameters of the enzyme in chaotropic and kosmotropic anions is always negative as is the case of the difference in activation parameters between psychrophilic and mesophilic enzymes [45]. Activation of NADH oxidase by urea in the presence of the anions studied is dependent on their position in the Hofmeister series. There is strong activation by urea in the presence of kosmotropic anions, slight activation in the presence of neutral anions, and deactivation in the presence of chaotropic anions. These observations indicate that the active site of NADH oxidase is more stable in the presence of kosmotropic anions than in the presence of chaotropic anions. The quenching experiments of the flavin cofactor fluorescence (Fig. 7) strongly support this interpretation and strongly indicate that the anion-induced changes in the activity of NADH oxidase are due to a change in flexibility oftheenzymeactivesite. These results show that anions nonspecifically activate NADH oxidase at low concentrations (< 0.5 M ). This is in accordance with the positive electrostatic potential from the protein near the flavin cofactor which is a common feature of homologous flavoenzymes [11]. Nonspecific changes in the exact nature of the contacts within related groups by the anions may then change (in our case activate) the enzyme at low (< 0.5 M ) concentrations of salt. At higher concentra- tions of salt, the effect of the anions is different and depends on their position in the Hofmeister series. Whereas anions from the middle of the Hofmeister series do not affect activity, both chaotropic and kosmotropic anions inhibit NADH oxidase. Changes in K m , Stern–Volmer quenching constants, activation entropy and fluorescence, along with slight changes in CD spectra, and localization of the active site in the region with an increased temperature B factor (Fig. 1) strongly suggest that the mechanism of inhibition of chaotropic and kosmotropic anions includes a modulation of flexibility in comparison with the optimal dynamics the active site. Kosmotropic anions stabilize and increase the rigidity of the enzyme active site and thus slow the catalytic rate k cat . On the other hand, chaotropic anions destabilize and increase the flexibility of the enzyme active site. The increased flexibility in the substrate-binding site leads to the increase in K m , i.e. a decrease in the affinity of theenzymeforthesubstrate.Thedecreaseink cat , however, can be only partially explained by the observed increase in K m . The main reason for a decreased k cat in the presence of chaotropic anions is probably increased dynamics in the active site which perturbs the proper position of the donor/substrate and the acceptor/flavin cofactor in the hydride transfer and/or other side chains with an active role in the catalytic site of NADH oxidase. Modulation of the conformational dynamics by the Hofmeister series of anions therefore offers a simple strategy for activation of enzymes from thermophiles and psychrophiles. Proteins from thermophiles are stabilized by a combina- tion of strategies [46]. An important one is the presence of optimized ionic pairs on the protein surface (electrostatic interaction), i.e. where the active sites of enzymes are localized [46–48]. Perturbation weakens some of the ionic interactions and may affect the mobility of the polypeptide/ side chain on the protein surface. This would have a positive impact on the enzyme activity without a significant effect on the stable hydrophobic protein core. On the other hand, enzymes from psychrophiles contain a highly charged region in order to improve solvent interac- tions with a hydrophilic surface [50]. Shielding of these interactions by a suitably chosen salt from the Hofmeister series, or another osmolyte, may stabilize the protein structure at increased temperature without deleterious effects on enzyme activity. Acknowledgements We thank the Fonds of Chemischen Industrie for financial support. We are also grateful for support through grants no. D/01/02768 from the Deutsche Akademische Austauschdienst (DAAD), and no. 1/8047/01 and 1/0432/03 from the Slovak Grant Agency. We thank Norbert Grillenbeck for his technical assistance. We also thank Linda Sowdal and Dr LeAnn K. Robinson for their invaluable editorial help in preparing the manuscript. References 1. Vieille, C. & Zeikus, G.J. (2001) Hyperthermophilic enzymes: sources, uses, and molecular mechanisms for thermostability. Microbiol. Mol. Biol. Rev. 65, 1–43. 2. Gerday, C., Aittaleb, M., Arpigny, J.L., Baise, E., Chessa, J P., Garsoux, G., Petrescu, I. & Feller, G. (1997) Psychrophilic enzymes: a thermodynamic challenge. Biochim. Biophys. Acta 1342, 119–131. 3. Fan, Y X., Ju, M., Zhou, J M. & Tsou, C L. (1996) Activation of chicken liver dihydrofolate reductase by urea and guanidine hydrochloride is accompanied by conformational change at the active site. Biochem. J. 315, 97–102. 4. Zhang, H J., Sheng, X R., Pan, X M. & Zhou, J M. (1997) Activation of adenylate kinase by denaturants is due to the Fig. 8. Enthalpy/entropy compensation of activation parameters of reaction catalyzed by NADH oxidase in the presence of chaotropic (SCN – , ClO 4 – ,I – ) (dark symbols), neutral (Br – , Cl – , CH 3 COO – )(grey symbols), and kosmotropic (SO 4 2– ,H 2 PO 4 – ) (white symbols) salts. Ó FEBS 2003 Anion effect on activity of NADH oxidase (Eur. J. Biochem. 271)55 increasing conformational flexibility at its active sites. Biochem. Biophys. Res. Commun. 238, 382–386. 5. Narayanasami, R., Nishimura, J.S., McMillan, K., Roman, L.J., Shea, T.M., Robida, A.M., Horowitz, P.M. & Masters, B.S.S. (1997) The influence of chaotropic reagents on neuronal nitric oxide synthase and its flavoprotein module. Urea and guanidine hydrochloride stimulate NADPH–cytochrome c reductase activity of both proteins. Nitric Oxide – Biol. Chem. 1, 39–49. 6. Das, M. & Dasgupta, D. (1998) Enhancement of transcriptional activity of T7 RNA polymerase by guanidine hydrochloride. FEBS Lett. 427, 337–340. 7. Inui, T., Ohkubo, T., Urade, Y. & Hayaishi, O. (1999) Enhance- ment of lipocalin-type prostaglandin D synthase enzyme activity by guanidine hydrochloride. Biochem. Biophys. Res. Commun. 266, 641–646. 8. Deshpande, R.A., Kumar, A.R., Khan, M.I. & Shankar, V. (2001) Ribonuclease Rs from Rhizopus stolonifer: lowering of optimum temperature in the presence of urea. Biochim. Biophys. Acta 1545, 13–19. 8a. Z ˇ olda ´ k, G., S ˇ ut’a ´ k, R., Antalı ´ k, M., Sprinzl, M. & Sedla ´ k, E. (2003) Role of conformational flexibility for enzymatic activity in NADH oxidase from Thermus thermophilus. Eur. J. Biochem. 270, doi:10.1046/j.1432-1033.2003.03889.x. 9. Hecht, H.J., Erdmann, H., Park, H.J., Sprinzl, M. & Schmid, R.D. (1995) Crystal structure of NADH oxidase from Thermus thermophilus. Nat. Struct. Biol. 2, 1109–1114. 10. Lovering, A.L., Hyde, E.I., Searle, P.F. & White, S.A. (2001) The structure of Escherichia coli nitroreductase complexed with nico- tinic acid: three crystal forms at 1.7 A ˚ , 1.8 A ˚ and 2.4 A ˚ resolution. J. Mol. Biol. 309, 203–213. 11. Haynes, C.A., Koder, R.L., Miller, A F. & Rodgers, D.W. (2002) Structures of nitroreductase in three states: effects of inhibitor binding and reduction. J. Biol. Chem. 227, 11513–11520. 12. Shoichet, B.K., Baase, W.A., Kuroki, R. & Matthews, B.W. (1995) A relationship between protein stability and protein func- tion. Proc. Natl Acad. Sci. USA 92, 452–456. 13. Karplus, M. & Petsko, G.A. (1990) Molecular dynamics simula- tions in biology. Nature 399, 631–639. 14. Carugo, O. & Argos, P. (1998) Accessibility to internal cavities and ligand binding sites monitored by protein crystallographic thermal factors. Proteins 31, 201–213. 15. Frauenfelder, H., Petsko, G.A. & Tsernoglou, D. (1979) Tem- perature-dependent x-ray diffraction as a probe of protein struc- tural dynamics. Nature 280, 558–563. 16. Goto, Y. & Aimoto, S. (1991) Anion and pH-dependent con- formational transition of an amphiphilic polypeptide. J. Mol. Biol. 218, 387–396. 17. Jelesarov, L., Du ¨ rr, E., Thoms, R.M. & Bosshard, H.R. (1998) Salt effects on hydrophobic interaction and charge screening in the folding of a negatively charged peptide to a coiled coil (leucine zipper). Biochemistry 37, 7539–7550. 18. Sedla ´ k, E., Z ˇ olda ´ k, G., Antalı ´ k, M. & Sprinzl, M. (2002) Ther- modynamic properties of nucleotide-free EF-Tu from Thermus thermophilus in the presence of low-molecular weight effectors of its GTPase activity. Biochim. Biophys. Acta 1597, 22–27. 19. Ivell, R., Sander, G. & Parmeggiani, A. (1981) Modulation by monovalent and divalent cations of the guanosine-5¢-triphospha- tase activity dependent on elongation factor Tu. Biochemistry 20, 6852–6859. 20. Fasano, O., De Vendittis, E. & Parmeggiani, A. (1982) Hydrolysis of GTP by elongation factor Tu can be induced by monovalent cations in the absence of other effectors. J. Biol. Chem. 257, 3145– 3150. 21. Wolosiuk, R.A. & Stein, M. (1990) Modulation of spinach chloroplast NADP-glyceraldehyde-3-phosphate dehydrogenase by chaotropic anions. Arch. Biochem. Biophys. 279, 70–77. 22. Wondrak, E.M., Louis, J.M. & Oroszlan, S. (1991) The effect of salt on the Michaelis Menten constant of the HIV-1 pro- tease correlates with the Hofmeister series. FEBS Lett. 280, 344–346. 23. Hall, D.L. & Darke, P.L. (1995) Activation of the herpes simplex virus type 1 protease. J. Biol. Chem. 270, 22697–22700. 24. Nishimura, J.S., Narayanasami, R., Miller, R.T., Roman, L.J., Panda, S. & Masters, B.S.S. (1999) The stimulatory effects of Hofmeister ions on the activities of neuronal nitric-oxide synthase. Apparent substrate inhibition by L -arginine is overcome in the presence of protein-destabilizing agents. J. Biol. Chem. 274, 5399– 5406. 25. Fan, Y X., McPhie, P. & Miles, E.W. (2000) Regulation of tryptophan synthase by temperature, monovalent cations, and an allosteric ligand. Evidence from Arrhenius plots, absorption spectra, and primary kinetic isotope effects. Biochemistry 39, 4692–4703. 26. Tanfani, F., Scire ` , A., Masullo, M., Raimo, G., Bertoli, E. & Bocchini, V. (2001) Salts induce structural changes in elongation factor 1alpha from the hyperthermophilic archaeon Sulfolobus solfataricus: a Fourier transform infrared spectroscopic study. Biochemistry 40, 13143–13148. 27. Ru, M.T., Hirokane, S.Y., Lo, A.S., Dordick, J.S., Reimer, J.A. & Clark, D.S. (2000) On the salt-induced activation of lyophilized enzymes in organic solvents: effect of salt kosmotropicity on enzyme activity. J. Am. Chem. Soc. 122, 1565–1571. 28. von Hippel, P.H. & Wong, K Y. (1964) Neutral salts: the gen- erality of their effects on the stability of macromolecular con- formations. Science 145, 577–580. 29. Breslow, R. & Guo, T. (1990) Surface tension measurements show that chaotropic salting-in denaturants are not just water-structure breakers. Proc. Natl Acad. Sci. USA 87, 167–169. 30. Timasheff, S.N.(1993) The control of protein stability and associ- ation by weak interactions with water: how do solvents affect these processes? Annu. Rev. Biophys. Biomol. Struct. 22, 67–97. 31. Leberman, R. & Soper, A.K. (1995) Effect of high salt con- centrations on water structure. Science 378, 364–366. 32. Baldwin, R.L. (1996) How Hofmeister ion interactions affect protein stability. Biophys. J. 71, 2056–2063. 33. Cacace, M.G., Landau, E.M. & Ramsden, J.J. (1997) The Hof- meister series: salt and solvent effects on interfacial phenomena. Q. Rev. Biophys. 30, 241–277. 34. Collins, K.D. (1997) Charge density-dependent strength of hydration and biological structure. Biophys. J. 72, 65–76. 35. Saunders, A.J., Daris-Searles, P.R., Allen, D.L., Pielak, G.J. & Erie, D.A. (2000) Osmolyte-induced changes in protein con- formational equilibria. Biopolymers 53, 293–307. 36. Hribar, B., Southall, N.T., Vlachy, V. & Dill, K.A. (2002) How ions affect the structure of water. J. Am. Chem. Soc. 121, 12302– 12311. 37. Park, H.J., Reiser, C.O.A., Kondruweit, S., Erdmann, H., Sch- mid, R.D. & Sprinzl, M. (1992) Purification and characterization of a NADH oxidase from the thermophile Thermus thermophilus HB8. Eur. J. Biochem. 205, 881–885. 38. Lehrer, S.S. (1971) Solute perturbation of protein fluorescence. The quenching of the tryptophyl fluorescence of model compounds and of lysozyme by iodide ion. Biochemistry 10, 3254–3263. 39. Za ´ vodszky, P., Kardos, J., Svingor, A. & Petsko, G.A. (1998) Adjustment of conformational flexibility is a key event in the thermal adaptation of proteins. Proc. Natl Acad. Sci. USA 95, 7406–7411. 40. Vihinen, M. (1987) Relationship of protein flexibility to thermo- stability. Protein Eng. 1, 477–480. 41. Tsai, A.M., Udovic, T.J. & Neumann, D.A. (2001) The inverse relationship between protein dynamics and thermal stability. Biophys. J. 81, 2339–2343. 56 G. Z ˇ olda ´ k et al.(Eur. J. Biochem. 271) Ó FEBS 2003 42. Zlateva, T., Boteva, R., Filippi, B., Veenhuis, M. & van der Klei, I.J. (2001) Deflavination of flavo-oxidases by nucleophilic reagents. Biochim. Biophys. Acta 1548, 213–219. 43. Fields, P.A. (2001) Review: protein function at thermal extremes: balancing stability and flexibility. Comp.Biochem.Physiol.A129, 417–431. 44. Murray, T.A., Foster, M.P. & Swenson, R.P. (2003) Mechanism of flavin mononucleotide cofactor binding to the Desulfovibrio vulgaris flavodoxin. 2. Evidence for cooperative conformational changes involving tryptophan 60 in the interaction between the phosphate- and ring-binding subsites. Biochemistry 42, 2317– 2327. 45. Lonhienne, T., Gerday, C. & Feller, G. (2000) Psychrophilic enzymes: revisiting the thermodynamic parameters of acti- vation may explain local flexibility. Biochim. Biophys. Acta 1543, 1–10. 46. Kumar, S., Tsai, C J. & Nussinov, R. (2000) Factors enhancing protein thermostability. Protein Eng. 13, 179–191. 47. Xiao, L. & Honig, B. (1999) Electrostatic contributions to the stability of hyperthermophilic proteins. J. Mol. Biol. 289, 1435– 1444. 48. Loladze, V.V., Ibarra-Molero, B., Sanchez-Ruiz, J.M. & Makhatadze, G.I. (1999) Engineering a thermostable protein via optimization of charge–charge interactions on the protein surface. Biochemistry 38, 16419–16423. 49. Spector, S., Wang, M., Carp, S.A., Roblee, J., Hendsch, Z.S., Fairman, R., Tidor, B. & Raleigh, D.P. (2000) Rational modi- fication of protein stability by the mutation of charged surface residues. Biochemistry 39, 872–879. 50. Feller, G., Arpigny, J.L., Narinx, E. & Gerday, Ch (1997) Mole- cular adaptations of enzymes from psychrophilic organisms. Comp. Biochem. Physiol. 118A, 495–499. Supplementary material The following material is available from http://blackwell publishing.com/products/journals/suppmat/EJB/EJB3900/ EJB3900sm.htm Supplementary material S1. (A–H) Activation parameters calculated from the temperature dependencies of activity at various concentrations of salts. Ó FEBS 2003 Anion effect on activity of NADH oxidase (Eur. J. Biochem. 271)57 . Modulation of activity of NADH oxidase from Thermus thermophilus through change in flexibility in the enzyme active site induced by Hofmeister series anions Gabriel Z ˇ olda ´ k 1 ,. and increase the flexibility of the enzyme active site. The increased flexibility in the substrate-binding site leads to the increase in K m , i.e. a decrease in the affinity of theenzymeforthesubstrate.Thedecreaseink cat , however,. strongly indicate that the anion -induced changes in the activity of NADH oxidase are due to a change in flexibility oftheenzymeactivesite. These results show that anions nonspecifically activate NADH oxidase