THE JOURNAL OF BIOI~ICAI. CXEM~~IZY Vol. 2.53, No. 14, Issue of July 25, pp. 4971-4979, 1978 PrintedinUSA.

Determination of Glucose Oxidase Oxidation-Reduction Potentials and the Oxygen Reactivity of Fully Reduced and Semiquinoid Forms*

(Received for publication, September 29,1977, and in revised form, February 2, 1978)

Marian T. Stankovich,$ Lawrence M. Schopfer, and Vincent Massey From the Department of Biological Chemistry, The University of Michigan, Ann Arbor, Michigan 48109

The oxidation-reduction potential values for the two electron transfers to glucose oxidase were obtained at pH 5.3, where the neutral radical is the stable form, and at pH 9.3, where the anion radical is the stable form. The midpoint potentials at 25’ were:

$3 6.3 EF&. + e- + H+ + EF’lH’ EmI = -0.063 f 0.011 V

EFlH’ + e- + H+ = El%& E,, = -0.066 + 0.007

pH 9.3 EN.,. + e- e El+‘- Em,,= -0.200 f O.OlOV

V

EF’- + e- + H* ti EFldH- Emr = -0.240 f 0.005 V

All potentials were measured uersus the standard hy- drogen electrode (SHE).

The potentials indicated that glucose oxidase radi- cals are stabilized by kinetic factors and not by ther- modynamic energy barriers. The pK for the glucose oxidase radical was 7.26 from dead time stopped flow measurements and the extinction coefficient of the neu- tral semiquinone was 4140 M-’ cm-’ at 570 nm.

Both radical forms reacted with oxygen in a second order fashion. The rate at 25’ for the neutral semiqui- none was 1.4 x lo4 M-’ a-‘; that for the anion radical was 3.5 x 10” M-’ s-l. The rate of oxidation of the neutral radical changed by a factor of 9 for a temperature difference of 22”. For the anion radical, the oxidation rate changed by a factor of 6 for a 22’ change in temperature.

We studied the oxygen reactivity of the a-electron reduced form of the enzyme over a wide wavelength range and failed to detect either oxygenated flavin derivatives or semiquinoid forms as intermediates. The rate of reoxidation of fully reduced glucose oxidase at pH 9.3 was dependent on ionic strength.

Few oxidation-reduction potentials of flavoprotein oxidases have been determined. It has been postulated that the oxi- dases may have less negative potentials than the dehydrogen- ases and that they stabilize red anionic radicals (1). Glucose oxidase, which contains 1 mol of FAD/mol of protein, is unusual among flavoproteins, exhibiting both the blue neutral and the red anionic semiquinones, depending on the pH of observation (pK approximately 7.5) (2).

* This work was supported by Grant GM-11106 from the United States Public Health Service. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely&o indicate this fact.

$ Present address, Chemistry Department, University of Massa- chusetts, Amherst, Mass. 01002.

Because of slow disproportionation, it is difficult to keep glucose oxidase quantitatively in its blue semiquinone form. On the other hand, the anionic form of the semiquinone is more stable. Quantitative conversion to the semiquinone an- ion can be accomplished by the EDTA-light reaction; this reaction proceeds only very slowly beyond the semiquinoid stage for glucose oxidase at pH 9.3. Flavodoxin is an example of a flavoprotein where the blue neutral semiquinone is sta- bilized (3). For flavodoxin, there is a large separation in oxidation-reduction potential between the first and second electron transfer (the second electron transfer has a very negative potential). This large thermodynamic barrier ac- counts for the stability of the semiquinoid form of flavodoxin. In order to determine whether the observed behavior of the two semiquinoid forms of glucose oxidase is due to thermo- dynamic or kinetic stabilization, we determined the oxidation- reduction potentials for each half-reaction (EFl,,/EFlH’, and EFlH’/EFl,,d)’ at low and high pH values, where the neutral and anionic semiquinoid species can be observed without significant contamination with each other.

The oxygen reactivity of the blue neutral flavin radical has been questioned (3-6). Free flavin radicals (pK 8.5) have vastly different oxygen reactivities, the neutral radical react- ing at a rate of 3 X lo4 M-l s-’ or less (4-6) and the anion radical reacting at a rate of 3 x 10’ M-’ s-’ (4, 6). The difference in rates is so great that it is difficult to know if the neutral radical is reacting with oxygen or if the reactivity could be accounted for by small amounts of anion radical present in equilibrium with the neutral form at low pH values. We decided to investigate the oxygen reactivity of the glucose oxidase radicals, over the pH range of 3.8 to 10, in order to determine whether the blue radical reacts with oxygen.

No oxygenated intermediates of fully reduced glucose oxi- dase have been observed in previous studies (7-9). Flavin peroxide adducts, substituted at L,, having distinct spectral properties (maxima at approximately 380 to 390 nm with molar extinction coefficients of 8000 to 9060) have been ob- served with flavin-dependent hydroxylases (10, 11) and with bacterial luciferase (12). Because of their spectral properties we considered it possible that such intermediates may have been overlooked in previous rapid reaction studies with glu- cose oxidase. Accordingly, we measured. the oxygen reaction of reduced glucose oxidase at IO-nm intervals from 300 to 580 nm. No evidence was found for the existence of any interme- diate, either a flavin peroxide or a semiquinoid form, in the conversion of fully reduced enzyme to its oxidized state.

’ The abbreviations used are: EFL, oxidized glucose oxidaae; EFlH’, neutral blue radical, EFi-, anionic red radical, EFldH-, fully reduced glucose oxidaae, unprotonated; EFl.dHz, fully reduced glu- coae oxrdase, protonated, EF& fully reduced glucose oxidase, with- out designation of protonation state.

4971

by guest on July 15, 2018http://w

ww

.jbc.org/D

ownloaded from

4972 Oxidation-Reduction and Oxygen Reactivity of Glucose Oxidase

EXPERIMENTAL PROCEDURES

Materials

Glucose oxidase (EC 1.1.3.4) was prepared from a concentrate of Aspergillus niger obtained from Miles Laboratories, Elkhardt, Ind. (DEE0 concentrate), by the method of Swoboda and Massey (13). Pyrophosphate (0.05 M) was used in all of the experiments measuring potentials at pH 9.3; it was made by adjusting the pH of sodium pyrophosphate (Mallinckrodt analytical reagent) with 6 N HCl (Baker analytical reagent). The pH 9.3 and 10.0 buffers used in the stopped flow experiments were 0.01 M glycine (Matheson, Coleman and Bell, 99.5%) or L-leucine (Nutritional Biochemicals) brought to the appro- priate pH with sodium hydroxide (Mallinckrodt analytical reagent). The pH 5.3 buffers were made from either sodium acetate (Baker analyzed reagent) and acetic acid (Mallinckrodt analytical reagent) or sodium phosphate, monobasic and dibasic (Baker analyzed re- agent), and citric acid (Matheson, Coleman and Bell, reagent), ad- justed to pH 5.3 with sodium hydroxide.

The ethylenediaminetetraacetic acid used in the light reactions was from Mallinckrodt (analytical reagent); sodium oxalate was from Allied Chemical (primary standard). Lumiflavin 3-acetate, tetraace- tylriboflavin, and deazariboflavin were the generous gifts of Dr. Peter Hemmerich (University of Konstanz). Sodium dithionite (purified) was from the Baker Chemical Co., and D-(+)-mannose (chemically pure) was from Pfansteihl Laboratories, Inc.

Glucose oxidase products, glucono-d-lactone, and hydrogen perox- ide were purchased from Mann Hesearch (chemically pure grade) and Baker Chemical Co. (certified ACS reagent), respectively. Sodium gluconate was purchased from Aldrich (reagent).

The oxidation-reduction dyes indigo disulfonate (recrystallized, gift from Dr. F. Guengerich, Vanderbilt University), benzyl viologen (K & K Laboratories, Inc.), and rosinduline G (K & K Laboratories, Inc.) were used without further purification.

Methods

Oxidation-Reduction Potential Measurements-An Orion pH me- ter was used for potentiometric measurements and a temperature- controlled Cary 118 spectrophotometer was used for spectral meas- urements. All potentiometric experiments were performed at 25” in an anaerobic spectrophotometric cell fitted with a platinum indicator electrode, a silver-silver chloride reference electrode, and a Hamilton syringe (14). All voltages are expressed relative to the standard hydrogen electrode.

Two procedures for the determination of the oxidation-reduction potentials for the two l-electron steps in the reduction of glucose oxidase were used, one procedure for pH 5.3 and another for pH 9.3. This was necessitated by the fact that oxidized glucose oxidase forms a complex with sulfite, the oxidation product of dithionite. At pH 5.3, the dissociation constant of this complex was 1.2 x 10m4 M (15). which was low enough to cause 30 to 50% of the glucose oxidase to be in the complexed state during the course of a dithionite reduction. Thus, a direct dithionite reduction was precluded at pH 5.3 and a dye reduc- tion procedure was used.

Midpoint Potentials of Dyes at pH 5.3 and 9.3 Determined by Potentiometric Dithionite Titrations-Dithionite was used as the reductant in preliminary potentiometric titrations of the dyes used to determine the oxidation-reduction potential values of glucose oxidase. In this way, the oxidation-reduction properties of rosinduline G, benzyl viologen, and indigo disulfonate were determined. E,,, values were measured and the dye properties were checked for accordance with the Nernst equation.

Measurement of Oxidation-Reduction Potential of Glucose Oxi- daae at pH 9.3-At pH 9.3, the dissociation constant for the glucose oxidase-sulfite adduct was greater than 2.3 X 10 ’ (15). Under our experimental conditions (3 x 10 ’ M enzyme) less than 2% of the glucose oxidase would be in the form of a complex with sulfite. Therefore, at pH 9.3, dithionite could be used to reduce the enzyme without complications arising from formation of sulfite adducts.

Glucose oxidase in 0.05 M p.yrophosphate buffer, pH 9.3, was titrated with 3 mM dithionite in the anaerobic cell described above. Indigo disulfonate (E, at pH 9.3 = -0.188 V) and benzyl viologen (E, at pH 9.3 = -0.370 V) were used as indicators and mediators to the platinum electrode.

The potentials quoted above were measured experimentally in this work. Indigo disulfonate obeyed the Nernst equation. Its log (ox/red) slope was 30 mV. Oxidized indigo disulfonate had an absorbance maximum at 610 nm, with an extinction of 19,000 Mm’ cm-‘. Under the experimental conditions, reduced indigo disulfonate did not ab-

sorb at wavelengths greater than 510 nm. The dye had an isosbestic point between its oxidized and reduced forms at 460 nm, at which wavelength changes in absorbance due to the oxidized enzyme could be measured without complication from the dye. In addition, the contribution of the dye to the total absorbance at any wavelength could be calculated from the dye absorbance at 610 nm, where no other species absorb.

For the determination of the potential of the 1st electron trans- ferred, a catalytic amount (lo-” M) of benzyl viologen was added to act as a mediator in the system. At this concentration, the absorbance spectrum of neither the oxidized nor reduced forms of benzyl viologen interfered with the spectrum of glucose oxidase. Since only enzyme semiquinone and oxidized indigo disulfonate absorbed at 530 nm, the absorbance of enzyme semiquinone was calculated by correcting the absorbance at this wavelength for the contribution of dye and dividing the resultant absorbance by the measured extinction coefficient of 2700 M-’ cm-’ for the enzyme semiquinone at 530 nm. The result was compared to the absorbance at 460 nm to determine whether all of the enzyme was in fully oxidized and semiquinone forms or if some had become fully reduced. The concentration of enzyme was 3.55 X lo-” M and the concentration of indigo disulfonate was 4.3 X lo-” M. The oxidation-reduction potential was measured both potentiomet- rically and from the dye ratio. The values agreed within 15 mV.

The most accurate values of oxidation-reduction potential were obtained from the fit 50% of the dithionite titrations, where the potential of the system was closest to the midpoint potential of the dye and where the oxidized form of the enzyme and the radical were the major enzyme species.

The oxidation-reduction potential of the second electron reduction step was determined by dithionite titration and equilibration with benzyl viologen, the only dye present. Benzyl viologen is colorless in the oxidized state; in the reduced state it has an absorption maximum at approximately 600 nm and a minimum at approximately 450 nm.

The benzyl viologen used had an impurity which enabled the potential to be stabilized at values as positive as -0.260 V. The main part of the E’:’ versus log (ox/red) plot of the dye had a slope of 50 mV from log (ox/red) = +l to -1. The impurity communicated well between the enzyme and the electrode, since the electrode potential was well poised at a potential of -0.260 V. More than 90% of the benzyl viologen remained in the oxidized state when the glucose oxidase semiquinone was over 50% reduced. The potential measure- ments were taken with the electrodes in the presence of enzyme and mediator and were compared to the values obtained for the dye alone. The presence of the enzyme did not shift the potential of the dye since the spectrophotometric values and potentiometric values agreed within 5 to 15 mV. The oxidation-reduction potential of glucose oxidase was calculated from the Nernst equation at various points in the titration.

Measurement of Oxidation-Reduction Potential of Glucose Oxi- dase at pH 5.3-The buffer used in these experiments was a mixture of citric acid (0.055 M), NaH1P04 (0.13 M), and EDTA (0.025 M) adjusted to pH 5.3 with NaOH. Rosinduline G (azocarmine G) was the dye used to measure glucose oxidase potentials at this pH value. The experimentally determined midpoint potential of rosinduline G (-0.087 V) was more positive than the published value (-0.180 V) (16) at pH 5.3. The potential uersus log (ox/red) plot for the dye deviated from the expected 60 mV slope for values of log (ox/red) greater than 0.8. The slope was greater than 60 mV for this part of the curve. The potentiometric titration curve of the dye was used as a standard curve to calculate cell potential when the dye was equili- brated with glucose oxidase.

Direct electrochemical measurements of cell potential of the rosin- duline G-glucose oxidase equilibrated system could not be made, since the time for equilibration of the system was from 4 to 24 h and the cell containing the electrodes could not be maintained anaerobic over such extended equilibration times. Instead, the experiment was per- formed in an anaerobic spectrophotometric cell. Rosinduline G (1.14 x lo-” M, its maximum solubility) in the above buffer containing 5 X lo-’ M lumiflavin 3-acetate was placed in the main body of the cell. A concentrated glucose oxidase solution was stored in the side arm. The cell was first made anaerobic by 10 cycles of evacuation and equilibration with purified nitrogen. The cell was then irradiated with visible light, causing the rosinduline G to be photoreduced via the lumiflavin-catalyzed EDTA-light reaction (17). The oxidized glucose oxidase was then mixed with the reduced dye. The final concentra- tions of glucose oxidase and dye were equimolar (1.14 X lo-” M).

In order to calculate the concentration of the various species in solution at equilibrium the following criteria were employed. 1) From

by guest on July 15, 2018http://w

ww

.jbc.org/D

ownloaded from

Oxidation-Reduction and Oxygen Reactivity of Glucose Oxidase 4973

the preliminary titration of rosinduline G alone, the extinction of both the oxidized and reduced forms was established at all wavelengths. Since the dye was impure, the measured extinction did not correspond to the literature values and are not reported. The titration, however, established standard extinctions for this batch of dye which were used in subsequent calculations. 2) The known extinction coefficients for the various forms of glucose oxidase were used: for oxidized enzyme at 455 nm, 14,000 M-’ cm-‘; for fully reduced enzyme at 455 nm, 1,200 Mu’ cm-‘; for blue semiquinone at 610 nm, 4,150 M-’ cm-‘; at 530 nm, 3,725 M-’ cm-‘; at 455 nm, 4,150 M-’ cm-’ (18).

After equilibration, the concentration of enzyme semiquinone could be determined at 610 nm, at which wavelength no other species absorbed. From the standard titration of the dye and the known spectra of the various glucose oxidase species, it was found that only oxidized dye and enzyme semiquinone absorbed at 530 nm. The contribution to the 530 nm absorbance by the enzyme semiquinone was subtracted and the resultant absorbance was compared to the dye standardization titration to give the concentration of oxidized dye. The reduced dye concentration was obtained by difference. The concentrations of oxidized and reduced enzyme were calculated at 455 nm. Contributions to the 455 nm absorbance from enzyme semiqui- none and oxidized and reduced dye were calculated and subtracted. The resultant absorbance was due to the sum of the absorbances of the oxidized and reduced forms. As the extinction coefficients of these forms were known, the concentration of oxidized and fully reduced enzyme was readily calculated.

The standard potentiometric titration curve for the dye was used to determine the potential in the cell at equilibrium. Assuming Nerstian behavior for the enzyme, the oxidation-reduction potentials for each of the l-electron transfers in the reduction of glucose oxidase were calculated. The value of AE (E,,, - E,& was obtained by two different methods, from the measured values of E,,,, and Em2 and from the per cent radical at equilibrium.

Additional equilibrations were performed with lumiflavin 3-acetate (&I = -0.150 V, pH 5.3) and tetraacetylriboflavin (E, = -0.110 V pH 5.3). The flavins were photoreduced with EDTA and then equi- molar (3 x LO-’ M) glucose oxidase was added. The spectrum was recorded after equilibration, and then the cell was opened to air and a final fully oxidized spectrum was taken. At equilibrium, four species absorbed at 450 nm: EFlH’, EFl,jHz, free reduced flavin, and free oxidized flavin (we assumed negligible amounts of EFL, were present, see below). The A.& (A~,,II, ux,dirrcj minus Aw,ui~ihrium) observed after opening the cell to air was due to the oxidation of EFIH’ (Ati, = 9950 M-’ cm-‘) plus the oxidation of EFldHZ (Aeam = 12,900 M ’ cm-‘) and the free reduced flavin (AQW = 11,720 M-’ cm-‘). The equilibrium amount of EFlH’ was calculated from its absorbance at 570 run. The equilibrium concentration of EFl,tHz was calculated by difference. The AAdm due to the oxidation of both of these enzyme species was calculated and subtracted from the observed AAw The difference was due to the oxidation of free reduced flavin. The equilibrium concentration of free oxidized flavin was obtained by difference.

The assumption that at equilibrium glucose oxidase existed pri- marily in the semiquinone and fully reduced forms was justified by the observation that initially after mixing glucose oxidase with free dihydroflavin, the amount of enzyme radical formed was high, decay- ing within 1 h to its equilibrium value. However, the absorbance at 450 nm increased only a small amount. This was consistent only with radical being reduced and free dihydroflavin being oxidized.

Production of Enzyme Semiquinone-The deazaflavin-catalyzed photochemical reduction method of Massey and Hemmerich (19) was employed to produce the stable red semiquinone of glucose oxidase.

Concentrated enzyme stock solution was passed over a Sephadex G-25 column to remove any free flavin. The protein was diluted to a final concentration of 2.8 to 3.6 X 10 ’ M in either 0.05 M sodium pyrophosphate containing 0.025 M sodium oxalate (pH 9.3) or 0.01 M glycine-NaOH containing 0.01 M sodium oxalate (pH 9.3). The spec- trum of the enzyme solution was taken, a catalytic amount of deaza- riboflavin (0.05 mol of deazaribotlavin/mol of enzyme flavin) was added, and the spectrum retaken. The solution was put into a stopped flow tonometer with a side arm cuvette, kept in the dark, and made anaerobic by evacuating and equilibrating with purified nitrogen 10 times. The tonometer, in a thermostated water bath, was irradiated with visible light. The progress of semiquinone formation was followed by measuring the absorbance at 530 and 450 nm using the Cary 118. A fmal spectrum of the fully formed semiquinone was taken for comparison with the stopped flow results. Typically, 90 to 98% of the total enzyme was converted to the semiquinone form. Oxalate was

used in place of the traditional EDTA (ethylenediaminetetraacetic acid) because the apparent fit order rate of semiquinone formation was much faster, i.e. reduction under the above conditions with 0.025 M oxalate proceeded at a rate of 0.53 min-‘, whereas with 0.025 M EDTA the rate was 0.07 min-‘. In some experiments, lumiflavin 3- acetate was used in place of deazariboflavin.

Production of Fully Reduced Form-Fully reduced glucose oxidase was produced by either of three methods. At pH 5.3 in 0.05 M acetate buffer, the fully reduced enzyme could be prepared by photoreduction as described above for the semiquinone. Fully reduced enzyme was also generated by adding a lo-fold excess of mannose into a solution of enzyme which had been passed over Sephadex G-25, diluted into water, and made anaerobic. Mannose was used for reduction because its reaction with the enzyme was so slow that it did not interfere with the reoxidation measurements made in the stopped flow apparatus (9). Finally, titration with dithionite at pH 9.3 in the presence of benzyl viologen (0.01 mol of dye/mol of enzyme flavin) was used to produce fully reduced enzyme.

Stopped Flow Experiments--Rapid reaction experiments were car- ried out in a stopped flow apparatus designed for anaerobic work and coupled to a Nova minicomputer system (11). Temperature was controlled at either 3O or 25”. For stopped flow measurements at wavelengths below 350 nm, a Corning filter (CS 7-54) was placed in the light path between the monochrometer and the optical cell in order to eliminate stray light and second order scattering effects.

pH Jump Experiments-The red semiquinone was prepared by photoreduction as described above at pH 9.3 in 0.05 M sodium pyrophosphate buffer containing 0.025 M sodium oxalate and a cata- lytic amount of lumiflavin J-acetate. This enzyme was rapidly mixed, at 25’, in the stopped flow apparatus with an equal volume of various concentrations of acetic acid to give final pH values between 3.8 and 8.5. The acidic solutions were pre-equilibrated with gas mixtures, containing known concentrations of oxygen, before mixing. Certified oxygen-nitrogen gas mixtures were obtained from Matheson Gas Products. The reactions were monitored spectrophotometrically at several wavelengths.

RESULTS

Oxidation-Reduction Potential atpH 5.3-The experimen- tally determined midpoint potential of rosinduline G (-0.087 V) at pH 5.3 was more positive than the published value of -0.180 V for pH 5.3 (16). In addition, there was an impurity in the dye which made the potential readings more positive, at log (ox/red) ratios above 0.8, than would be predicted from the 60 mV slope of a Nernst plot.

Equilibration of the reduced dye with oxidized enzyme at 25” yielded the following E, values: (average of three experi- ments)

E,I EFL,. + e- + H+ * EFlH’ -0.063 f 0.011 V

E,,,s EFlH’ + e- + H+ * EFldHr -0.061 f 0.002 V

The average amount of radical produced at equilibrium was 32% (limits 27 to 38%). According to Clark (20), this corre- sponds to essentially no separation of the two potentials, which agrees with the measured potentials. The relatively large scatter in the above experiments is due to the low concentrations of reactants, imposed by the solubility of the dye.

Results from the equilibration of reduced tetraacetyh-ibo- flavin (TARF) with oxidized glucose oxidase yielded a value only for the second electron transfer, since the E, for TARF was -0.110 V at pH 5.3, with a 30 mV log (ox/red) slope. This meant that real equilibrium between EFI,. and TARF could only exist when most of the glucose oxidase was in the semi- quinoid and fully reduced forms and most of the TARF was oxidized. The value obtained, E,,,z = -0.069 V, was slightly more negative than that calculated from the rosinduline G experiments. The largest amount of radical observed was 36%. The same problem was observed with lumiflavin 3-ace- tate. This flavin had E, = -0.150 V at pH 5.3 and the log (ox/ red) had a 30 mV slope. Therefore, upon equilibration with

by guest on July 15, 2018http://w

ww

.jbc.org/D

ownloaded from

4974 Oxidation-Reduction and Oxygen Reactivity of Glucose Oxidase

glucose oxidase, most of the enzyme was in the semiquinoid and fully reduced forms. This gave E,z = -0.072 V with a maximum of 36 to 65% radical. When the red radical was made at pH 9.3 and then the pH of the solution changed by mixing with citric acid to yield pH 5.3, 36 to 42% radical was obtained at equilibrium, corresponding to a 10 to 20 mV separation between the first and second electron reduction steps. The data for pH 5.3 are summarized in Table I.

Oxidation-Reduction Potential at pH 9.3-At pH 9.3 and 25”, the potential of the first electron reduction was deter- mined by dithionite titration of the enzyme in a spectropoten- tiometric cell containing indigo disulfonate and lo-” M benzyl viologen as mediator between the dithionite and the dye. Benzyl viologen decreased the time required for the system to come to equilibrium. Even so, equilibration between dye, enzyme, and dithionite required approximately 2 h after each addition of dithionite. The potential values were taken from the electrode readings during the dithionite titrations. The values for the first electron reduction in glucose oxidase were taken from the first 40% of the dithionite titration. Two values for the oxidation-reduction potential of the first electron re- duction were calculated from this titration. E,, equals -0.200 f .OlO v.

The oxidation-reduction potential of the second electron reduction of glucose oxidase was determined by dithionite titration in the presence of benzyl viologen. The dithionite titration in the presence of this dye was rapid. Less than 15 min was required for equilibration after each dithionite addi- tion. Dithionite titration was carried to completion; the max- imum amount of enzyme radical obtained in this titration was 45 to 50%. In the first 30% of the titration, the electrode potential was not well poised, since the dye was not being reduced. At approximately 40% reduction, the electrode po- tential became well poised. The values of oxidation-reduction potential were calculated from the part of the titration where 60 to 90% of the EFL, was reduced, using the Nernst equation and the electrode readings of the cell potential. The value for the second electron reduction (average of two experiments) was E,z = -0.239 f 0.605 V. The difference in the electrode potential was AE = 39 mV, corresponding to 51% radical (20).

A simple dithionite titration of glucose oxidase at pH 9.3 was also carried out without dye or electrodes. The purpose of the titration was to determine the maximum amount of enzyme radical produced and thus to get a measure of the relative difference in electrode potential (U). From Clark (20), one can determine the relative difference in electrode potential from the amount of radical observed in a dithionite titration. Initially, the titration proceeded fairly slowly (30 to 60 min for equilibration after each addition). When the titra- tion was 60% complete, a maximum of 90% radical was present. Beyond this point in the titration, each addition of dithionite required 2 h for equilibration. According to Clark (20), for- mation of 90% radical in a titration corresponds to a 120 mV separation in oxidation-reduction potential values. This result is considerably different from those obtained in the presence

of dye mediators; its significance will be considered under “Discussion.”

Oxidation of Anionic Semiquinone-The red semiquinone was oxidized by oxygen at pH 9.3 under pseudo-first order conditions (see Fig. 2 for experimental details). The rates of the reactions were such that at least 90% of the reaction could be monitored in the stopped flow trace. Fig. 1 shows that the spectrum of the species at the end-of-flow (3 ms after mixing), compiled from stopped flow traces taken every 10 nm from 300 to 550 nm, was that of the starting red semiquinone. The rate traces were first order for at least 95% of the reaction at all wavelengths. Thus, there was no spectral evidence for intermediates occurring either during the dead time of the stopped flow apparatus or in the interval from the end-of-flow to the end-of-reaction.

When the semiquinone was reacted (at 25”) with various oxygen concentrations and the pseudo-first order rates plotted against oxygen concentration, the resultant line was linear and passed through the origin (Fig. 2). Therefore, the reaction of the red semiquinone with oxygen was strictly second order, with a rate constant (k,,) of 3.5 X 10’ M-l s-l. Most of the rate data were taken from traces at 530 and 450 nm.

I

350 450 550 650

WAVELENGTH (nm )

FIG. I. Comparison of stopped-flow dead time (3 ms after mixing) and static spectra of glucose oxidase semiquinone species. Stopped- flow conditions: temperature was 25’. The enzyme was photoreduced using 0.025 M oxalate and a catalytic amount of deazaflavin (19) in 0.05 M sodium pyrophosphate buffer, pH 9.3. A, the corrected dead time absorbance after mixing with 0.1 M sodium pyrophosphate buffer, pH 9.3, containing 1.27 mM oxygen; 0, the corrected dead time absorbance after mixing with 0.15 M acetic acid (final pH 4.8), con- taining 1.27 mM oxygen. Stopped-flow semiquinone spectra were corrected for 20% oxidized enzyme formed prior to mixing (see text). Static conditions: temperature was 12”. The enzyme was photore- duced using 0.023 M EDTA in 0.026 M sodium pyrophosphate buffer, pH 9.2. -’ ‘-, initial oxidized enzyme; - - -, anionic semiquinone after light irradiation; -, neutral semiquinone immediately after adding citric acid (final pH 5.9). (These spectra were taken from Ref. 18.) For the sake of comparison, all spectra are exhibited in terms of extinction coefficient.

TABLE I

Oxidation-reduction potentials for glucose oridase

Reducing system -

Dithionite titration

Dithionite titration Dye equilibration Dye equilibration Dye equilibration

E “,I En,2 PH Dye used Measured E,,, of

dye

V V -0.200 f 0.10 9.3 Indigo disulfonate (catalytic -0.188

benzyl viologen) -0.239 + 0.005 9.3 Benzyl viologen -0.370”

-0.063 + 0.011 -0.061 f 0.002 5.3 Hosinduline G -0.087 -0.069 5.3 Tetraacetylriboflavin -0.110 -0.072 5.3 Lumiflavin 3-acetate -0.150

a The presence of an impurity resulted in stabilization of electrode potentials at much higher potentials (see text for details).

by guest on July 15, 2018http://w

ww

.jbc.org/D

ownloaded from

Oxidation-Reduction and Oxygen Reactivity of Glucose Oxidase 4975

[O,](rnM)

FIG. 2 (left). The oxidation of glucose oxidase semiquinones by oxygen. Conditions: 30 PM enzyme was photoreduced anaerobically in 0.05 M sodium pyrophosphate buffer, pH 9.3, containing 0.05 M oxalate and a catalytic amount of lumiflavin 3-acetate (17). Reduced enzyme was rapidly mixed at 25’ with oxygen equilibrated in 0.1 M sodium pyrophosphate buffer, pH 9.3, or-O.15 M acetic acid (resultant pH = 4.8).

Frc,. 3 (center). Determination of the pK for the transition be- tween the neutral (EFlH’) and anionic (EFI-) semiquinones of glucose oxidase. Conditions: temperature was 25”. Glucose oxidase was pho- toreduced as in Fig. 2 and then mixed with various concentrations of

Oxidation of Neutral Semiquinone-The red semiquinone at pH 9.3 was mixed with 0.15 M acetic acid equilibrated with oxygen, yielding a final pH measured to be 4.8. In Fig. 1, it can be seen that the neutral semiquinone was completely formed in the 3 ms dead time of the stopped flow instrument. The subsequent rates of oxidation appeared to be first order at all wavelengths studied. Again there was no spectral evi- dence for an intermediate in oxidation. The rates of reaction were such that at least 90% of the reaction could be monitored in the stopped flow trace. The pseudo-first order rates for the oxidation of the neutral semiquinone at 25” were measured at various oxygen concentrations. When plotted against the ox- ygen concentration, the rates were linear and passed through the origin (Fig. 2). Therefore, the oxidation of the neutral semiquinone with oxygen was also strictly second order, with a rate constant (k,,) of 1.4 X lo4 Me’ s-‘. This is in substantial agreement with the value obtained by Bright and Porter (21) of 1.6 x 10’ M-’ s-’ under similar conditions.

Extinction Coefficient for Blue Semiquinone-The calcu- lation of the extinction coeffkient of the neutral semiquinone was based on the change in absorbance (from end-of-flow to final oxidation) and on the concentration of enzyme. The total concentration of enzyme was calculated from the absorbance of the oxidized form using the extinction coefficient of 14,100 M -1 cm-l at 450 nm. Since some of the semiquinone became oxidized during handling, it could not be assumed that 100% of the enzyme had reached the optical cell of the stopped flow spectrophotometer in the semiquinone form. However, the fraction of enzyme in semiquinone form reaching the optical cell at pH 9.3 could be determined by comparing h4,, be- tween semiquinone and oxidized enzyme measured from spec- tra recorded with the Gary spectrophotometer to the observed AA4.w in the stopped flow experiments. We assumed that in any given experiment, approximately the same concentration of semiquinone reached the optical cell at all pH values. In this experiment, 80% of the total semiquinone reached the optical cell at pH 9.3 (average of several determinations). It was assumed, therefore, that at pH 4.8, the observed AA570 was due to 80% of the total concentration of enzyme. This resulted in a molar extinction coefficient of 4,140 M-’ cm-’ at 570 nm for the blue semiquinone, in excellent agreement with previous estimations from pH adjustment experiments carried out in an anaerobic optical cell assembly (18).

oxygen-equilibrated acetic acid. The ratio of the concentrations of the semiquinone species was calculated from the absorbance at 580 nm, 3 ms after mixing.

FIG. 4 (right). The pH profile for the oxidation of glucose oxidase semiquinone with oxygen. Conditions: the enzyme was photoreduced as in Fig. 2, then rapidly mixed at 25’ with various concentrations of acetic acid equilibrated with 1.27 mM oxygen. Oxidation was followed spectrophotometrically at 450, 530, and 580 nm. The solid line rep- resents a theoretical curve for pK 7.25. The limiting theoretical rates for pH 9.3 and 4.8 were taken from Fig. 2.

TABLE II

Oxidation rates for glucose oxidase semiquinone and fully reduced forms using 0.635 m M oxygen

kobs (s-7 IOlliC

strength PH Radical Fully reduced 64

3” 25” 30 25”

0.070 3.8 8.15 0.005 5.3 1200 0.025 5.3 0.64 8.06 1100 0.120 5.3 0.90 7.70 920 0.320 5.3 0.84 8.15 660 0.025 9.3 5.2 26.7 126 0.120 9.3 5.6 21.0 (47)” 0.250 9.3 17 0.520 9.3 5.3 21.0 13 15 0.005 10.0 20 67 0.175 10.0 17 5.2 0.340 10.0 4.2 20 (5.5)”

D Determined from Fig. 8.

pH Studies of Seniquinone-Since the reaction of both forms of semiquinone with molecular oxygen was strictly second order and the rates differed only by a factor of 2 to 3, we decided to study the rate of reaction as a function of pH. This study provided two sets of data for the determination of the pK for the deprotonation of the neutral to the anionic semiquinone: (a) the rate of reaction of the semiquinone with oxygen at each pH from 3.8 to 10; and (b) the ratio of the semiquinone neutral and anionic forms at the end of the dead time. Absorbance at 580 nm was used to determine the dead time ratio of neutral semiquinone (which absorbs at 580 nm) and the anionic semiquinone (which does not).

Fig. 3 shows the dead time absorbance measurements at 580 nm, expressed as log (EFi-/EFlH’), plotted uersus pH. This plot yielded a pK of 7.28. The apparent first order rate ( kobs) at 0.635 mM oxygen for the oxidation of the semiquinone varied with pH as shown in Fig. 4. The line drawn through the data represents a theoretical curve for pK 7.25. The plot of rates as a function of pH did not fit the theoretical curve perfectly; however, the rates reached a definite plateau at a value of approximately 8 s-’ for pH values below 5 and of approximately 21 S-I for pH values above 9.

Effect of Ionic Strength and Temperature on Semiquinone

by guest on July 15, 2018http://w

ww

.jbc.org/D

ownloaded from

4976 Oxidation-Reduction and Oxygen Reactivity of Glucose Oxidase

Oxidation-Table II shows that there was no effect of ionic 0 35 strength on the reaction of glucose oxidase radical with oxygen 1 at either pH 5.3 or 9.3. Table II and Fig. 5 contain data on the effect of temperature on the rate of semiquinone oxidation at pH 5.3 and 9.3. The red radical (pH 9.3) reacted at a second order rate of 3.5 x lo4 M-’ s-’ at 25” and 5.8 X 10” M-’ s-’ at 3”. The rate, therefore, varied by a factor of 6 for a 22” change of temperature. The blue radical had second order rate con- stants of 1.3 X lo4 Me2 s-l at 25’ and 1.4 X lo3 M-’ 6’ at 3”, differing by a factor of 9 for a 22” change in temperature.

Oxidation of Fully Reduced Enzyme-When fully reduced glucose oxidase was mixed with 0.1 M sodium acetate buffer, pH 5.3, containing 1.27 mM oxygen the oxidation appeared to be first order at all wavelengths examined (370 to 580 nm). The apparent first order rate constant (3”) was 800 to 900 s-l. Since the reaction half-life was only 1 to 2 ms, 56 to 80% of the reaction was over in the dead time of our stopped flow apparatus. For this reason, the dead time spectrum of the reaction mixture could not be determined. However, there were no obvious spectral indications of intermediates at any wavelength and the rate of oxidation was independent of the method of reduction or the wavelength of observation.

At pH values above 9, the oxidation of fully reduced enzyme with oxygen was much slower and 90 to 100% of the reaction could be observed. Thus, these reactions were more amenable to careful study. Oxidation studies were performed at 3” and 25’ in 0.05 M pyrophosphate, pH 9.3, and in 0.005 M leucine, pH 10.0. Under all conditions studied, apparent first order oxidations were observed. There were no spectral indications of intermediates at any wavelength. Furthermore, there were no detectable intermediates formed in the dead time of the stopped flow experiments. Fig. 6 shows the difference spec- trum (pH 9.3, 25”) of species taken at the dead time versus the fully oxidized spectrum (0.8 s after mixing). This is com- pared to the difference spectrum between fully reduced and oxidized species taken with a Cary recording spectrophotom- eter. The two are superimposable.

[02] (mM)

FIG. 5. The temperature dependence of the rate of oxidation of glucose oxidase semiquinones with oxygen. Conditions: the enzyme was photoreduced as in Fig. 2, except deazariboflavin was used in place of lumitlavin 3-acetate. Semiquinone w&s rapidly mixed: A, at 25” in 0.1 M sodium pyrophosphate buffer (resultant pH = 9.3); B, at 25’ in 0.1 M acetic acid (resultant pH = 5.3); C, at 3” in the same buffer as in A, D, at 3” in the same buffer as in B. The solutions were equilibrated either with air or with 100% 02 before mixing with the anaerobic enzyme species.

/\ l e

1 I I I I I 300 400 500

WAVELENGTH (nm 1

FIG. 6. Comparison of statically (solid line) and kinetically (‘) determined difference spectra for the oxidation of fully reduced enzyme at pH 9.3. Conditions: temperature was 25”. 30 PM enzyme in 0.05 M sodium pyrophosphate buffer, pH 9.3, was reduced anaerobi- cally with a lo-fold excess of mannose and then oxidized by rapidly mixing with the same buffer equilibrated with 1.27 mM oxygen. The static difference spectrum is that of fully oxidized minus fully reduced enzyme recorded on a Cary 118 spectrophotometer. The kinetic difference spectrum is that of fully oxidized enzyme minus the ab- sorbance 3 ms after mixing.

When the apparent first order rate constants ( kOb) for these four experiments were plotted versus the oxygen concentra- tion, the plot was linear only for the experiment in leucine buffer at 25”. Fig. 7 shows the inverse plots (l/kob versus l/ [O,]) for all four experiments. The inverse plot for the case of leucine buffer at 25” had a zero intercept; the second order rate constant was 1.0 X lo5 Me1 s-l. This is in reasonable agreement with the Weibel and Bright (8) value of 1.5 X 10’ M-’ S-l for the same conditions. Non-zero intercepts were found for the plots under the other conditions, indicating the formation of a Michaelis type complex with oxygen prior to oxidation. The limiting rate for both experiments carried out in pyrophosphate buffers was 31 a-‘. Dividing the slope by the intercept yielded dissociation constants (22) for oxygen of 9.2 X lo-’ M at 3” and 7.7 X 10-O M at 25”. The limiting rate and oxygen dissociation constant for the leucine experiment at 3” were 53 s-’ and 1.1 X lop3 M, respectively.

Effect of Ionic Strength and Temperature on Fully Re- duced Enzyme Oxidation--The effect of ionic strength on the oxidation of fully reduced enzyme is shown in Fig. 8. In these experiments, fully reduced enzyme was mixed with buffers of various ionic strength 6) and ionic composition (see Fig. 8 for experimental details). At pH 9.3, a regular decrease in rate was seen from p = 0.02 to 0.20. Beyond p = 0.20, the rate remained constant. Changing the ionic composition of the buffer appeared to have no effect, indicating a nonspecific ionic strength effect rather than a specific ion effect. A similar pattern was found at pH 10.

The temperature dependence of the rate of oxidation of fully reduced enzyme varied with pH (Table II). At pH 5.3, the rate of oxidation was relatively independent of tempera- ture and ionic strength. A factor of 3 difference in rate of oxidation for a 27O change in temperature at pH 5.6 has been reported (7). At high pH, the rate of oxidation depended both on ionic strength and temperature, i.e. for low ionic strength (lo < 0.03), the rate of oxidation varied severalfold for a 22” change in temperature and at high ionic strength (lo = 0.5), the rate of oxidation was almost independent of temperature. In contrast, the oxidation by oxygen of both semiquinoid species was quite temperature dependent, but was influenced only marginally by ionic strength.

by guest on July 15, 2018http://w

ww

.jbc.org/D

ownloaded from

Oxidation-Reduction and Oxygen Reactivity of Glucose Oxidase 4977

(mM) FIG. 7. The oxidation of fully reduced glucose oxidase by oxygen.

Conditions: 30 pM enzyme in water was reduced anaerobically by a IO-fold excess of mannose and then oxidized by rapidly mixing with the following oxygen-equilibrated buffers: A, 0.1 M sodium pyrophos- phate, pH 9.3, at 3”; B, same buffer as in A, at 25”; C, 0.01 M leucine, pH 10.0, at 3”; D, same buffer as in C at 25’.

150

a

Fro. 8. The effect of ionic strength (11) on the rate of oxidation of fully reduced glucose oxidase. Conditions: 30 pM enzyme in water was reduced anaerobically with a IO-fold excess of mannose, then rapidly mixed at 25’ with buffers equilibrated with 1.27 mM oxygen. At pH 9.3, the buffers were: sodium pyrophosphate for p = 0.025, 0.045, 0.3, 0.4, 0.5; 0.17 M leucine for p = 0.085; 0.01 M PP, plus 0.05 M KC1 for p = 0.1; 0.01 M PP, plus 0.10 M KNO:, for p = 0.15; 0.01 M PP, plus 0.15 M KI for ~1 = 0.20. At pH 10.0: 0.01 M leucine for ~1 = 0.005; 0.01 M leucine plus 0.2 M KC1 for ~1 = 0.1; 0.1 M glycine plus 0.025 M oxalate for p = 0.175; 0.01 M leucine plus 0.4 M KC1 for fl = 0.2.

Effect of Reaction Products on Rate of Oxidation of Fully Reduced Enzyme-The products of the reaction had no effect on the rate of oxidation. When mannose-reduced enzyme was reacted with oxygenated buffers containing 2 NIM concentra- tions of gluconolactone, gluconic acid, or hydrogen peroxide, there was no change in the oxidation rate.

DI.SCUS.SION

The oxidation-reduction potentials of the two separate elec- tron transfers of glucose oxidase have been measured. At pH 5.3, the two electron transfers (Em1 = -0.063 V; Em2 = -0.065 V) are separated only by 2 mV; at pH 9.3, the two electron transfers (E,,, = -0.200 V; E,,,z = -0.240 V) are separated by 40 mV. This small difference in potential would not predict generation or stabilization of the radical form of the enzyme in the presence of an equivalent or an excess quantity of reducing agent with a potential more negative than that of glucose oxidase. The values of Em1 would predict a pK for the ionization of the semiquinone in the range pH 7.2 to 8.0. This should be compared with the value of 7.3 determined spectro- photometrically (Figs. 3 and 4).

At pH 5.3, when EFl,,, is anaerobically mixed with equi- molar reduced TARF or lumiflavin 3-acetate, a maximum of 50 to 75% radical is produced. This initial fraction of radical slowly decays to a final equilibrium value of 1 to 2%, as would be expected for a system in which the reductant has a potential 100 mV more negative than either E,, or E,s. These facts suggest the presence of a kinetic barrier against the second electron transfer from free flavin to glucose oxidase. This is demonstrated more forcefully by stopped flow experiments in which reduced free flavins (=-lo-fold excess) were reacted with oxidized glucose oxidase, at pH 5.3 and 25”. The rate of transfer of the 1st electron between reduced lumiflavin 3- acetate and oxidized glucose oxidase was 2.9 x lo” Mm’ s-l, while the rate of electron transfer from reduced flavin to the neutral enzyme radical was lo-fold slower, 3.3 X 10’ M-’ s-‘.

A maximum of 70% neutral enzyme radical was generated (17).

The presence of a kinetic barrier is also indicated at pH 9.3, where there is only a 40 mV separation between the two electron transfers. In spite of this small separation, quantita- tive production of glucose oxidase radical can be obtained by reductants (deazaflavin radicals generated in the photochem- ical system or reduced free flavins) which have oxidation- reduction potentials much more negative than the enzyme. This is most readily interpreted as being due to a kinetic barrier against the bransfer of the 2nd electron. The presence of a kinetic barrier is further illustrated in the dithionite titration, at pH 9.3, in the presence and absence of a catalytic amount of benzyl viologen. In the absence of benzyl viologen, the time for equilibration after each addition of dithionite is different for the first and second electron reductions (30 to 60 min for the former and 2 h for the latter). A maximum of 90% radical is formed at 60% reduction by dithionite, which would correspond to a difference in oxidation-reduction potential between the two forms of 120 mV. When benzyl viologen is added, complete equilibration occurs within 15 min after each addition throughout the titration. The maximum amount of radical is 45 to 52%, which agrees with the 40 mV separation measured potentiometrically. Ninety per cent stable radical occurring in the absence of benzyl viologen is too large to be accounted for by simple thermodynamic constraints. This observation alone suggests a kinetic barrier (for dithionite) against the second electron transfer. The additional finding that catalytic amounts of benzyl viologen increase the rate of equilibration and re-establish Nernst type behavior confirms the conclusion that kinetic, not thermodynamic, barriers to dithionite reduction exist. This also establishes benzyl violo- gen as a mediator, capable of circumventing that barrier.

It should be noted that all of the reducing systems which give rise to the semiquinoid forms of glucose oxidase are ones known to be capable of single electron transfers (17, 19, 23). The production of concentrations of glucose oxidase semiqui- none greater than that predicted from the two oxidation- reduction equilibria is readily accounted for by kinetic barriers against the second electron transfer.

In contrast to this situation is the lack of any observable intermediate in either the reductive half-reaction with sub- strate (7-9) or the oxidative half-reaction with oxygen (7-9). In view of the observed kinetic barrier to production of fully reduced enzyme with l-electron reductants, we consider this particularly suggestive evidence against radical intermediates in the catalytic reaction. In addition, since the oxidation- reduction potential difference between the two electron trans- fers is small (compared to the 250 mV separation for flavo- doxin (3)), thermodynamic considerations would oppose the stabilization of glucose oxidase radicals and favor a simulta- neous 2-electron transfer mechanism.

by guest on July 15, 2018http://w

ww

.jbc.org/D

ownloaded from

4978 Oxidation-Reduction and Oxygen Reactivity of Glucose Oxidase

Glucose oxidase can stabilize the two spectrally distinct forms of the semiquinone, depending on pH. Protonation of the anionic form is complete in less than 3 ms, the dead time of the stopped flow apparatus. This was firmly established by demonstrating that the blue semiquinone was the spectral species present 3 ms after jumping the pH of a solution of red semiquinone at pH 9.3 down to pH 4.8 (Fig. 1). Using the stopped flow pH jump technique, a pK of 7.28 was established for the protonation reaction. This value thus represents a refinement of the earlier determination of 7.5 by Massey and Palmer (2), who used photoreduction with EDTA at various pH values to determine the relative amounts of neutral and anionic semiquinone. The speed of the stopped flow measure- ments ensures that the concentration of semiquinone at low pH has not been diminished by disproportionation. The dis- proportionation rate at low pH ( tri2 z 6 h at pH 4.8, 12”) is negligibly slow relative to the dead time of the stopped flow instrument. Since the concentration of blue semiquinone at the dead time was unaffected by disproportionation, it was possible to make accurate measurements of the molar extinc- tion coefficients. Values of 4140 Mm’ cm-’ for 570 nm and 4070 Mm’ Cm-’ for 580 nm were obtained. These values are some- what higher than earlier values obtained by dithionite titra- tion at pH 6.3 (2550 Me’ cm-’ at 570 nm) (24) and EDTA- light reduction at pH 5.98 (3100 Me’ cm-’ at 570 nm) (2) where disproportionation probably occurred. They are, however, in excellent agreement with previous studies involving pH ad- justment to pH 5.9 (4150 Mm’ cm-’ at 570 nm) (18).

Most significantly, however, we have demonstrated that the neutral form of the enzyme-bound semiquinone reacts with oxygen. This is in marked contrast to earlier results obtained with free flavin (4-6), flavodoxin (3), and Shethna flavoprotein (25). In each of those cases, the oxygen reactivity at high pH values was several orders of magnitude larger than that ob- served at low pH values. The difference in rates was large enough to support the conclusion that the neutral semiqui- none was oxygen unreactive and that the observed oxidation at low pH was due to the activity of a small equilibrium concentration of the anionic semiquinone. In the case of glucose oxidase, however, the rates of oxidation for the neutral and anionic semiquinones differ by only a factor of 3 (1.4 x IO4 M-’ s-’ and 3.5 X lo4 M-’ s-‘, respectively) at 25”. Such a difference is too small to support the contention that the rates at low pH are due to an equilibrium amount of the anionic species. Thus, the glucose oxidase neutral semiquinone must react directly with oxygen. This conclusion is further sup- ported by the pH and temperature dependence of oxidation. The pH rate profile (Fig. 4) clearly shows that the rates level off at both pH extremes (below pH 5 and above pH 9), thus indicating reaction of two distinct species. The temperature study (Fig. 5) reveals a different temperature dependence for the semiquinone oxidation at pH 9.3 and pH 5.3. For a 22” change in temperature, there is a g-fold change in rate at pH 9.3 and a B-fold change at pH 5.3. Such a difference is most simply explained as being due to the reaction of two different species with oxygen.

The oxidation of the enzyme semiquinone is strictly second order. Variations in temperature, pH, and ionic strength do not alter the second order nature of this reaction. In contrast, at pH values above 9, fully reduced glucose oxidase exhibits second order behavior only at low ionic strength (p = 0.005) and 25”. Under these conditions, oxidation occurs at a rate of 1.0 X lo5 Mm1 s-‘, which agrees reasonably well with Bright’s (8) value of 1.5 X IO” Mm’ s-‘. At higher ionic strength (p = 0.5) or lower temperature (3”), the kinetics of oxidation pro- vide clear evidence for an intermediate between molecular oxygen and fully reduced enzyme (Fig. 7) (22). This interme-

diate is most simply interpreted as a Michaelis-type complex, since it is not associated with any detectable spectral change.

Considerable effort was expended in attempting to obtain evidence for an oxygenated intermediate in the reaction of oxygen with either the semiquinone or fully reduced enzyme. This effort was prompted by recent work on C,,, covalent intermediates in the oxidation of reduced flavin-dependent hydroxylases (10, 11). These Cq, intermediates are character- ized by molar extinctions of 8000 to 9000 Mm’ cm-’ at 380 to 390 nm. Earlier studies had failed to detect any intermediate in the oxidation of glucose oxidase by oxygen (7-9). The thorough search for intermediates in this investigation was similarly unsuccessful. Despite wide ranging variations in pH, ionic strength, oxygen concentration, and temperature, no spectral evidence for an oxygenated intermediate was found at any wavelength between 300 and 580 nm. The oxidation always appeared to be simple first order, for both semiquinone and fully reduced enzyme.

The data developed in the course of these studies, however, provide some interesting additions to the general body of information on the reactivity of glucose oxidase (see Table II). For pH values greater than 9, the rate of oxidation of fully reduced enzyme is markedly dependent on ionic strength (in the range p = 0.02 to 0.20). However, at pH 5.3 for the fully reduced enzyme and at both pH 5.3 and pH 9.3 for the semiquinone, there is little or no effect of ionic strength on the rate. The temperature dependence of the reaction rate for fully reduced enzyme with oxygen at pH 5.3 and pH 9.3 is slight, which is similar to the findings of Gibson et al. (7) for pH 5.6. On the other hand, the semiquinone oxidation exhibits a large temperature dependence at both pH 5.3 and 9.3 (Fig. 6). The pH rate dependence for oxidation of the fully reduced enzyme is similar to that reported by Bright (8,9), the rate at pH 5.3 being 12 times larger than that at. pH 9.3. The pH profile of the semiquinone oxidation rate is just the reverse, the rate at pH 4.8 being approximately 3 times smaller than that at pH 9.3.

Finally, it is possible to extract some mechanistic informa- tion from a comparison of the relative rates of oxidation of semiquinone and fully reduced enzyme. The position (which was initially advanced by Michaelis (26) and which has since been championed by Bruice (27)) that the oxidation-reduction of flavin oxidases occurs in l-electron steps can be addressed in part by this work. One mechanism (which can be rejected) to account for a stepwise l-electron oxidation of fully reduced flavin is shown below:

EFL,.,,HJ + O:, k EFlH’ + Hog’

EFlH’ + 02 ++ EFL,, + HO?’

Since no spectral evidence of the intermediacy of EFlH’ is found in the oxidation of EFl,dH2, hp would have to be much faster than hl. However, it is clear from Table II that even in the most favorable case (pH 10, p = 0.340), the semiquinone rate (h2) is only slightly greater than the fully reduced rate. This rate difference is too small to support the proposed mechanism, thus eliminating it as a possibility. It is not possible to address directly other possible mechanisms with the data in this study.

On the basis of oxidation studies of reduced glucose oxidase with a nitroxide radical, Chan and Bruice (27) have postulated the following sequence of l-electron transfers:

EFl,,,,H, + 0% * [EFlH’ O,l-] (+H+)

[EFlH’ O,?] (+H+) A EFl,,, + Hz02

by guest on July 15, 2018http://w

ww

.jbc.org/D

ownloaded from

Oxidation-Reduction and Oxygen Reactivity of Glucose Oxidase 4979

The reaction with an obligatory l-electron acceptor, nitroxide radical, to produce the flavoprotein semiquinone as an inter- mediate is clearly not surprising. Whether this reaction is a valid model for reaction with the natural substrate, oxygen, is however another matter. Chan and Bruice (27) considered it was, since the pH rate profile of the oxidation of the enzyme radical by the nitroxide radical (pK = 7.2) was very similar to that of the oxidation of fully reduced enzyme (pK = 7.5) with oxygen, determined by Weibel and Bright (8), and similar to that of the neutral-anionic radical ionization (pK = 7.4 to 7.5, Ref. 2). It should be noted that Chan and Bruice (27) found that rate of enzyme semiquinone production from EFl,.,,rH? and nitroxide radical to be independent of pH. Therefore, the nitroxide radical reaction could be considered as a reasonable model only for the oxygen reaction in the step defined by hz in the scheme above. In order for the similarity in pH rate profiles to have significance, this requires that h2 should be at least partially rate-determining. We consider this to be rather unlikely, since it would require a steady state level of the enzyme semiquinone superoxide complex. Neither in previous work (7-9) nor in the present study has any such complex been detected. Furthermore, it would not be unreasonable to expect some dissociation of superoxide radical from the pos- tulated complex. Previous studies have also failed to detect superoxide radical formation with glucose oxidase and other tlavoprotein oxidases, although such formation was readily detected with a variety of flavoprotein dehydrogenases (28).

The results presented in this paper give clear evidence for kinetic barriers against the transfer of the 2nd electron in reduction studies in which l-electron reductants were em- ployed. It is interesting to note that a similar kinetic barrier appears to exist in the second electron transfer from reduced enzyme species to the nitroxide radical of Chan and Bruice (27) when (especially at high pH values) the rate of conversion of reduced enzyme to its semiquinone state was much faster than the subsequent oxidation of the semiquinone. Although we recognize the danger of extrapolation from limited model studies to the reaction of the enzyme with its physiological substrates, the existence of kinetic barriers to l-electron trans- fers observed in this study and that of Chan and Bruice (27) appears to us to argue against l-electron transfers in the catalytic reaction mechanism of this enzyme.

REFERENCES 1. Massey, V., Miiller, F., Feldberg, R., &human, M., Sullivan, P.

2. 3.

4. 5. 6.

7.

8.

9.

10.

11.

12.

13.

14.

15.

16. 17.

18.

19.

20.

21.

22.

23.

24. 25. 26.

27.

28.

A., Howell, L. G., Mayhew, S. G., Matthews, R. G., and Faust, G. P. (1969) J. Viol. Chem. 244, 3999-4006

Massey, V., and Palmer, G. (1966) Biochemistry 5, 3181-3189 Mayhew, S. G., Foust, G. P., and Massey, V. (1969) J. Biol. Chem.

244,803-810 Vaish, S. P., and Tollin, G. (19’71) Bioenergetics 2,61-72 Holmstrom, B. (1964) Acta. Phys. Polon. 26.419-426 Faraggi, M., Hemmerich, P., and Pecht, I. (1975) FEBS Lett. 51,

47-51 Gibson, Q. H., Swoboda, B. E. P., and Massey,V. (1964) J. Biol.

Chem. 239,3921-3934 Weibel, M. K., and Bright, H. J. (1971) J. Biol. Chem. 246,

2734-2744 Bright, H. J., and Appleby, M. (1969) J. Biol. Chem. 244,

3625-3634 Strickland, S., and Massey, V. (1973) J. Biol. Chem. 248,

2953-2962 Entsch, B., BaUou, D. P., and Massey, V. (1976) J. Biol. Chem.

251.2550-2563 Hastings, J. W., Balny, C., LePeurch, C., and Douzou, P. (1973)

Proc. Natl. Acad. Sci. U. S. A. 70, 3468-3472 Swoboda, B. E. P., and Massey, V. (1965) J. Biol. Chem. 240,

2209-2215 Guengerich, F. P., Ballou, D. P., and Coon, M. J. (1975) J. Biol.

Chem. 250.7405-7414 Swoboda, B. E. P., and Massey, V. (1966) J. Biol. Chem. 241,

3409-3416 Michaelis, L. (1931) J. Biol. Chem. S&369-372 Massey, V., Stankovich, M., and Hemmerich, P. (1978) Biochem-

istry 1’7, 1-8 Massey, V., Matthews, R. G., Foust, G. P., Howell, L. G. Williams,

C. H., Zanetti, G., and Ronchi, S. (1969) in Pyridine Nucleotide- Dependent Dehydrogenases (Sund, H., ed) pp. 393-411, Sprin- ger-Verlag, Berlin

Massey, V., and Hemmerich, P. (1977) J. Biol. Chem. 252, 5612-5614

Clark, W. M. (1960) Oxidation-Reduction Potentials of Organic Systems, pp. 184-203, The Williams & Wilkins Co., Baltimore

Porter, D. J. T., and Bright, H. J. (1977) J. Biol. Chem. 252, 4361-4370

Strickland, S., Palmer, G., and Massey, V. (1975) J. Biol. Chem. 250,4048-4052

Lambeth, D. O., and Palmer, G. (1973) J. Biol. Chem. 248, 6095-6103

Massey, V., and Gibson, Q. H. (1964) Fed. Proc. 23, 18-29 Edmondson, D. E., and Tollin, G. (1971) Biochemistry 10,133-145 Michaelis, L., Schubert, M. P., and Smythe, C. V. (1936) J. Biol.

Chem. 116,587-607 Chan, T. W., and Bruice, T. C. (1977) J. Am. Chem. Sot. 99,

2387-2389 Massey, V., Strickland, S., Mayhew, S. G., Howell, L. G., Engel,

P. C., Matthews, R. G., Schuman, M., and Sullivan, P. A. (1969) Biochem. Biophys. Res. Commun. 36.891-897

by guest on July 15, 2018http://w

ww

.jbc.org/D

ownloaded from

M T Stankovich, L M Schopfer and V Masseyreactivity of fully reduced and semiquinoid forms.

Determination of glucose oxidase oxidation-reduction potentials and the oxygen

1978, 253:4971-4979.J. Biol. Chem. 

  http://www.jbc.org/content/253/14/4971Access the most updated version of this article at

 Alerts:

  When a correction for this article is posted• 

When this article is cited• 

to choose from all of JBC's e-mail alertsClick here

  http://www.jbc.org/content/253/14/4971.full.html#ref-list-1

This article cites 0 references, 0 of which can be accessed free at

by guest on July 15, 2018http://w

ww

.jbc.org/D

ownloaded from