Oxidation-reduction properties of Escherichia coli thioredoxin reductase altered at each active site cysteine residue.

Thioredoxin is a small oxidation-reduction (redox) mediator protein. Its reduction by NADPH is catalyzed by the flavoenzyme thioredoxin reductase. Site-directed mutagenesis has provided forms of the reductase in which Cys135 and Cys138 have each been changed to a serine residue (Prongay, A. J., Engelke, D. R., and Williams, C. H., Jr. (1989) J. Biol. Chem. 264, 2656-2664). Cys135 and Cys138 form the redox-active disulfide in the oxidized enzyme. The redox properties of the two altered forms of Escherichia coli thioredoxin reductase have been determined from pH 6.0 to 9.0. Photoreduction of TRR(Ser135,Cys138) produces the blue, neutral semiquinone species, which disproportionates (Kf = 0.73) to an apparent maximum of 29% of the total enzyme as the semiquinone. In contrast, the semiquinone formed on TRR(Cys135,Ser138) during a photoreductive titration does not disproportionate and 70% of the enzyme is stabilized as the semiquinione. Reductive titrations have demonstrated that 1 mol of sodium dithionite (2 electrons)/mol of FAD is required to fully reduce TRR(Ser135,Cys138) whereas 2 mol of dithionite/mol of FAD are required to fully reduce TRR(Cys135,Ser138). The oxidation-reduction midpoint potentials for the 1-electron and 2-electron reductions of TRR(Ser135,Cys138) have been determined by NADH/NAD+ titrations in the presence of a mediator, benzyl viologen. The midpoint potential for the 2-electron reduction of TRR(Ser135,Cys138) is -280 mV, at pH 7.0 and 20 degrees C. Thus, the redox potential is similar to that of the FAD/FADH2 couple in the dithiol form of wild type enzyme, -270 mV (corrected to 20 degrees C) (O'Donnell, M. E., and Williams, C. H., Jr. (1983) J. Biol. Chem. 258, 13795-13805). The delta Em/delta pH is -57.1 mV, which corresponds to a proton stoichiometry of 2 H+/2 e-.A maximum of 19% of the enzyme forms a stable semiquinone species during the titration, and the potentials for the oxidized enzyme/semiquinone couple, E2, and the semiquinone/reduced enzyme couple, E1, are -306 and -256 mV, respectively, at pH 7.0 and 20 degrees C. These studies provide evidence that the residue at position 138 exerts a greater effect on the FAD than does the residue at position 135.

Thioredoxin is a small oxidation-reduction (redox) mediator protein. Its reduction by NADPH is catalyzed by the flavoenzyme thioredoxin reductase. Site-directed mutagenesis has provided forms of the reductase in which Cys'" and C~S ' '~ have each been changed to a serine residue (Prongay, A. J., Engelke, D. R., and Williams, C. H., Jr. (1989) J. Biol. Chem. 264, 2656-2664. CYS"~ and C~S "~ form the redox-active disulfide in the oxidized enzyme. The redox properties of the two altered forms of Escherichia coli thioredoxin reductase have been determined from pH 6.0 to 9.0. Photoreduction of TRR(Ser'35,Cys138) produces the blue, neutral semiquinone species, which disproportionates ( K t = 0.73) to an apparent maximum of 29% of the total enzyme as the semiquinone. In contrast, the semiquinone formed on T R R ( C~S '~~, S~~"~) during a photoreductive titration does not disproportionate and 70% of the enzyme is stabilized as the semiquinione.
Reductive titrations have demonstrated that 1 mol of sodium dithionite (2 electrons)/mol of FAD is required to fully reduce TRR(Ser'3S,Cys138) whereas 2 mol of dithionite/mol of FAD are required to fully reduce T R R ( C~S ' " , S~~'~~) .
The oxidation-reduction midpoint potentials for the l-electron and 2-electron reductions of TRR(Ser'3S,Cys'3s) have been determined by NADH/ NAD+ titrations in the presence of a mediator, benzyl viologen. The midpoint potential for the 2-electron reduction of TRR(Ser136,Cys138) is -280 mV, at pH 7.0 and 20 "C. Thus, the redox potential is similar to that of the FAD/FADHz couple in the dithiol form of wild type enzyme, -270 mV (corrected to 20 "C) (O'Donnell, M. E., and Williams, C. H., Jr. (1983) J. Biol. Chem. 258,13795-13805). The AE,,,/ApH is -57.1 mV, which corresponds to a proton stoichiometry of 2 H+/2 e-. A maximum of 19% of the enzyme forms a stable semiquinone species during the titration, and the potentials for the oxidized enzyme/semiquinone couple, Ez, and the semiquinone/reduced enzyme couple, E l , are -306 and -256 mV, respectively, at pH 7.0 and 20 "C. These studies provide evidence that the residue at position 138 exerts a greater effect on the FAD than does the residue at position 135.
* This work was supported by the the Health Services and Research Administration of the Department of Veterans Affairs and by Grant GM21444 from the National Institute of General Medical Sciences, National Institutes of Health. 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 to indicate this fact. $ Present address: Dept. of Biological Sciences, Purdue University, West Lafayette, IN 47907.
To whom correspondence should be addressed Medical Research Service, 151, Veterans Affairs Medical Center, 2215 Fuller Rd., Ann Arbor, MI 48105.
The transfer of electrons between NADPH and thioredoxin is catalyzed by thioredoxin reductase (Moore et al., 1964;Zanetti and Williams, 1967). Thioredoxin is the reductant in a number of cellular processes (Holmgren, 1985). Escherichia coli thioredoxin reductase, a dimer of two identical subunits, contains two oxidation-reduction centers per subunit: an FAD prosthetic group and a redox-active disulfide (Moore et al., 1964;Zanetti and Williams, 1967). The enzyme can accept a total of 4 electrons. Reductive titration of the enzyme results in a gradual bleaching of the flavin absorbance without the appearance of an unique spectral intermediate at the 2-electron reduced stage (Zanetti and Williams, 1967;Thelander, 1968). The midpoint potentials of the disulfide/dithiol and FAD/FADH2 couples of this enzyme are separated by only 11 mV. Consequently, throughout a reductive titration, an equilibrium of four microforms of the enzyme is established: oxidized enzyme (microform I); 2 forms of 2-electron reduced enzyme, FAD/dithiol (microform 11) and FADH2/disulfide (microform 111); and 4-electron reduced enzyme (microform IV) (O'Donnell and Williams, 1983). The two 2-electron reduced forms are present at near-equimolar concentrations at pH 8.0, and the pH profiles of the ratios (Kl/Kz and K2/K1) of the microscopic equilibrium constants for the microform I/ microform I1 couple, K,, and the microform I/microform I11 couple, K,, reveal pK, values at 7.0 and 7.6, respectively. These pK, values have been assigned to the ionization of an active site thiol and an active site base, respectively (O'Don-ne11 and Williams, 1983).
Only small amounts of l-electron reduced and &electron reduced semiquinone forms appear during titrations with reduced pyridine nucleotides and this has been attributed to l-electron reduction of trace oxygen by FADH,. In reductions promoted by photoirradiation however, the two semiquinone species can be produced in high yield (Zanetti and Williams, 1967;Zanetti et al., 1968;O'Donnell and Williams, 1983).

Redox
Properties of Mutated Thioredoxin Reductase and confirm these conclusions (Kuriyan et al., 1991). The effects of replacing these cysteine residues with serines on the oxidation-reduction properties of t h e FAD are presented.
Absorbance spectra were recorded with a Cary 118C spectrophotometer interfaced to an IBM-AT personal computer using a DATA Translation 2800 multipurpose board and the ASYST software package purchased from the Macmillian Software Co., and adapted by Glenn Piot and L. D. Arscott, Department of Veterans Affairs Medical Center, Ann Arbor, MI.
Anaerobiosis-All titrations were performed under anaerobic conditions at 20 "C, except where noted. Anaerobiosis was performed as previously described (Williams et al., 1979;O'Donnell and Williams, from the denaturation of TRR(Ser'35,Cys'38) and T R R ( C~S '~' , S~~'~' ) 1983). In an attempt to minimize the amount of turbidity resulting during the anaerobic process, the buffer containing either a redox mediator (3,10-dimethyl-5-deaza-isoalloxazine, methyl viologen, or benzyl viologen) or NAD+ was made anaerobic by 15 alternating cycles of vacuum of 10-mm Hg and oxygen-free nitrogen or argon gas a t 4 "C with vigorous vortexing. The volume loss was quantified by changes in absorbance at wavelengths below 300 nm. Enzyme solutions (20-100 pl, 1-4 mM) were added under positive pressure of either nitrogen or argon gas to the anaerobic buffer through an injection port in the anaerobic cell and the solution subjected to five additional alternating cycles of 10-mm Hg vacuum and nitrogen or argon gas at 4 "C, without vortexing, but with the vessel on its side in order to create maximal surface area for gas exchange. The solutions were scrupulously protected from light. Hamilton syringes containing anaerobic titrants were fitted to the anaerobic cuvette under positive pressure of either nitrogen or argon gas (Williams et al., 1979).
Photoreduction-The enzymes, in 1 ml of 50 mM sodium/potassium phosphate buffer, pH 7.6, containing 30 mM EDTA and 2 nmol of 3,10-dimethyl-5-deaza-isoalloxazine were made anaerobic and reduced at 4 "C by exposure for varying time intervals to bright light from a Sun Gun (Smith-Victor Corp., Griffith, Ind., model Q-1U) at a distance of 7 cm from the sample. The intensity of the illumination was controlled by a rheostat . The absorbance spectra were recorded after each exposure as in Fig. 1. During the photoreduction of T R R ( C~S '~~, S~~'~' ) , the maximum semiquinone formation was 70% of the total enzyme concentration. Therefore, the spectrum of the semiquinone was obtained by extrapolation. The first four spectra recorded during the photoreduction were composites of different ratios of oxidized and semiquinone forms of the enzyme. Therefore, in order to obtain the spectrum of the semiquinone species, the contribution of the oxidized enzyme to the spectra was removed by multiplying the spectrum of the initial oxidized enzyme by a factor representing the fraction of oxidized enzyme in each of the mixed spectra, and subtracting this manipulated spectrum from each of the mixed spectra. The resultant spectra were divided by a factor representing the fraction of semiquinone present in each mixture, and the average of the four extrapolated spectra yielded the extrapolated spectrum of the semiquinone species (Fig. 2). Reduction of the semiquinone by the second electron was more difficult, and complete reduction was not achieved. Therefore, the spectrum of the fully reduced enzyme was also obtained by extrapolation. The last three spectra recorded during the photoreduction were mixtures of semiquinone and reduced forms of the enzyme. Therefore, the contribution of the semiquinone to these mixed spectra was removed by multiplying the averaged extrapolated spectrum of the semiquinone by a factor representing the fraction of semiquinone present in each of the mixed spectra and subtracting this manipulated spectrum from each of the mixed spectra. The resultant spectra were divided by a factor representing the fraction of reduced enzyme present in each of the mixed spectra, and the average of the three extrapolated spectra yielded the extrapolated spectrum of 2-electron reduced enzyme (Figs. 1 and 2).
A blue, neutral semiquinone also forms upon photoreduction of TRR(Ser'35,Cys'3s), but the semiquinone is unstable and slowly disproportionates. Thus, a phototitration was not obtained. The c% of the semiquinone species of wild type thioredoxin reductase and T R R ( C~S '~~, S~~'~' ) are nearly equal. This is true also of c~~~. Therefore, the values for these extinction coefficients calculated for TRR stabilized on TRR(Ser'35,Cys'38).
( C y~'~~, S e r '~~) were used to quantify the amount of semiquinone Dithionite Titrations-Solutions of sodium dithionite were prepared in anaerobic 50 mM sodium pyrophosphate buffer, pH 8.5, and standardized by titrating a solution of lumiflavin-3-acetic acid. In typical titrations of the enzymes, 20 p M TRR(Ser135,Cys'38) and 0.15 p M methyl viologen in 1 ml of 0.1 M sodium/potassium phosphate buffer, pH 7.6, or 40-45 p~ T R R ( C~S '~~, S~~'~' ) and 1.5-2.5 p~ methyl viologen in 2-3 ml of the same buffer were anaerobically reduced with aliquots of sodium dithionite added from a gas-tight Hamilton syringe attached to the anaerobic cell. The progress of the reductions were followed by recording spectra at each stage of the titration (Figs. 3 and 8). The rates of the reductions of these proteins were slow; therefore, a catalytic amount of methyl viologen was included in the titrations. In addition, the appearance of the spectrum of reduced methyl viologen at 395 nm (subtracted in Fig. 3 but prominent in Fig.  8) is an indicator of completion of the titration, since the midpoint potential of the enzymes is considerably more positive than the potential of the methyl viologen. The extinction coefficients for absorbance at 585 and 454 nm of the fully reduced TRR(Ser'35,Cys'38) at pH 6.5, 7.6, and 8.5 were obtained from single additions of excess sodium dithionite. The extinction coefficients for absorbance at 585 and 454 nm of the 2-electron reduced T R R ( C~S '~~, S~~'~' ) were extrapolated from several titrations.
Potentiometric Titrations-The midpoint potentials of TRR(Ser'35,Cys'38) were determined in the pH range 6.0-9.0 at 20 "C by anaerobic NADH titrations. The concentration of the NADH titrant solution was determined by measuring the absorbance at 340 and 259 nm of diluted samples, using the extinction coefficients c~~~ = 6,220 M" cm" and c~5 9 = 14,900 M" cm-'. The concentration of the NAD+ stock solution was determined by measuring the absorbance at 259 nm of diluted samples, using the extinction coefficient ~2 5 9 = 18,000 M-' cm". Anaerobic solutions of TRR(Ser'35,Cys'3s) (60-70 nmol) in 1.5 ml of citrate-phosphate-KC1 or Tris-KC1 buffers of varying pH, containing between 44 nmol and 4.6 pmol of NAD+, were titrated with NADH at 20 "C. Buffer solutions in the pH range 6.0-8.0 were citric acid-dibasic sodium phosphate mixtures; the ionic strength was adjusted to approximately 0.1 by the addition of KC1, and the buffering strength was 32-56 mM. Buffer solutions in the pH range 8.0-9.0 were prepared by mixing together a solution of 50 mM Tris base, 50 mM KC1 with a solution of 100 mM HC1, 100 mM KC1. To compensate for an increasing difference in the midpoint potentials of the enzyme-FAD/FADHp and the NAD+/NADH couples as the pH is lowered, the amount of NAD+ initially present was increased as the pH was lowered, thereby facilitating the quantification of the equilibrium at less than 30% reduction of the enzyme.
Potentiometric titrations were performed on T R R ( C~S '~~, S~~'~' ) as described above. However, the presence of a spectrally silent second reducible center on this protein created a problem of quantifying the various species present at any point during a reductive potentiometric titration. Additionally, the increasing degree of turbidity of the solution throughout the titration decreased the accuracy of spectrally determining the equilibrium concentration of NADH after each addition (see below). These results combined to make meaningful potentiometric titrations of T R R ( C~S '~~, S~~'~~) difficult. Therefore, solutions of T R R ( C y~'~~, S e r '~~) (64 nmol of enzyme in 1.46 ml of 0.1 M sodium/potassium phosphate buffer, pH 7.6, containing 102 and 112 nmol of NAD+) were made anaerobic and reduced by the addition 51 and 219 nmol of NADH. At equilibrium, the spectra were recorded and the macroscopic distributions determined (Table I).
Corrections for Turbidity-The anaerobic process results in denaturation of a small amount of the enzymes, giving rise to turbidity. To correct for the turbidity component in the spectra, the enzyme solution following the titration was centrifuged in an Eppendorf microcentrifuge at 15,000 rpm for 5 min. The precipitate was suspended in 1.5 ml of the buffer and the spectrum recorded. The contribution of turbidity to the spectra recorded during the titrations was subtracted by multiplying the turbidity spectrum by a factor so that the absorbance at 800 nm of the turbidity spectrum equals the absorbance at 800 nm of the enzyme spectrum. The manipulated turbidity spectrum was then subtracted from the enzyme spectrum resulting in zero absorbance at 800 nm. In the worst case, this correction was 25% of the initial absorbance at 454 nm; most titra-tions involved corrections of 8-12% of the initial absorbance at 454 nm.
Quantification of Enzyme Species-After correcting for the turbidity contribution to the spectra, the concentrations of oxidized enzyme, semiquinone and 2-electron reduced enzyme were calculated for each point in the titration. The concentration of semiquinone, E,, formed during the titrations was calculated for TRR(Ser135,Cys138) and T R R ( C~S '~~, S~~I~) from the absorbance at 585 nm using (5% = 3,800 M" cm" (Fig. 1, inset), and this concentration was subtracted from the total enzyme concentration to yield the concentration of oxidized plus 2-electron reduced enzyme (Equation 1).
The contribution of the semiquinone to the absorbance at 454 nm was calculated by multiplying the concentration of the semiquinone by c451 = 3,500 M" cm" (Fig. 1, inset). This value was subtracted from the actual absorbance at 454 nm to yield the absorbance at 454 nm due to oxidized and 2-electron reduced enzyme (Equation 2).
In order to obtain the [E,,] and [&dl, the Aerb4 was applied. The e454 of E,. was pH-dependent in the case of TRR(Ser'35,Cys'38) and was 11,000 M" cm" in the case of T R R ( C~S '~~, S~~'~~) ( 3) The concentration of oxidized enzyme is obtained by difference (Equation 4).
Quantification of NADH and NAD+ at Equilibrium-For accurate determination of the midpoint potential of an enzyme couple using a redox indicator, such as a dye or a pyridine nucleotide, the concentrations of the oxidized and reduced species of the indicator and the enzyme should be easily determined at each titration point. Absorbance spectra of various mixtures of oxidized, semiquinone, and reduced forms of TRR(Ser'35,Cys138) or T R R ( C~S '~~, S~~'~~) corrected for the absorbance of the NADH have apparent isosbestic points at 340 and 342 nm, respectively, as in the dithionite titrations of Figs. 3 and 8. Consequently, the concentration of NADH present at equilibrium can be calculated from the absorbance increases at the isosbestic wavelength during a titration. However, the titrations of these enzymes resulted in increasing amounts of turbidity, which easily cause calculations of erroneous values for the equilibrium concentrations of NADH. Therefore, for each of the titrations the assumption that one-half the concentration of the semiquinone plus the concentration of the reduced enzyme corresponds to the amount of NADH oxidized (Equation 5) was employed.
(Eq. 5) The validity of this assumption was tested by comparing, for a few titrations, the equilibrium concentration of NADH during a titration calculated from the change in absorbance at the isosbestic wavelengths with the values calculated using Equation 5. For the titrations in which negligible turbidity was present the agreement between the two methods ranged between 91 and 98%, whereas when significant turbidity was present the values only agreed in the range of 70-85%. Therefore, the assumption is valid, and the use of Equation 5 is perhaps the more accurate approach in the presence of turbidity.
At pH values below 7, hydrolysis of NADH becomes significant during the prolonged time periods required for stable equilibrium to be reached. Therefore, the amount of hydrolysis was determined during mock titrations performed simultaneously with the enzyme titrations. An amount of NADH equivalent to the amount added to the enzyme at each point of the titration was added to an equal volume of the same buffer used in the titration of the enzyme. The absorbance of this solution was recorded a t the time of addition and after the enzyme solution had reached a stable equilibrium, and the amount of NADH hydrolyzed was determined from the difference in absorbance. The spectra of the NADH titrations of TRR(Ser'35,Cys'38) and T R R ( C~S '~~, S~~I~~) a t pH 7.2, after subtracting the contribution of the equilibrium concentration of NADH, are presented in Figs. 4 and 9, respectively.
Calculations of the Oxidation-Reduction Potentials-The potential for the system at each point in the titration was calculated from the Nernst equation (Equation 8a). . This 1-electron reduction character is seen as a perturbation of the Nernst plots in the regions 0-25% and 75-100% reduction (Fig. 6). Therefore, the midpoint potential for the overall 2-electron transfer was calculated from data between 30 and 70% reduction. This value was used to calculate the theoretical line in the log([E,d]/ [E,,]) uersus E,, plot (Fig. 6).
The potentials for the two 1-electron couples, E1 for the Esq/Ered transition and E, for the E,,/E, transition were calculated using Equations 9-12, where Keq is the equilibrium constant for the comproportionation reaction (Clark, 1960).

Photoreduction of T R R ( C~S '~~, S~~'~' )
in the presence of 3,lO-dimethyl-5-deaza-isoalloxazine2 and 30 mM EDTA Hemmerich, 1977, 1978;, as described under "Materials and Methods," results in the formation of a stable 1-electron reduced species having a spectrum of a blue, neutral semiquinone, E,, (Fig. 1). During such a phototitration of T R R ( C~S '~~, S~~'~' ) a maximum of 70% of the enzyme is present as the semiquinone. The absorbance values at 585 and 454 nm of 100% semiquinone were extrapolated from a plot of A5= uersus A454r revealing extinction coefficients of 3,800 M" cm" and 3,500 M-' cm", respectively (Fig. 1, inset). The computer manipulations of spectra to obtain the spectrum of the semiquinone are described under "Materials and Methods." The extrapolated spectrum of the semiquinone confirmed the values of these extinction coefficients (Fig. Z), which are nearly equal to the values of the semiquinone of wild type thioredoxin reductase (Zanetti et al., 1968;O'Donnell and Williams, 1983). The value of c454 for the fully reduced enzyme, 1,900 M" cm", was extrapolated from the same plot, assuming an c585 of zero.
Reduction of TRR(Ser'35,Cys13s) by phototitration has revealed that this enzyme does not stabilize as much of the blue, neutral semiquinone; instead, the semiquinone that is formed initially disproportionates slowly to oxidized and 2-electron This compound is named as an isoalloxazine as in the reference, but note that it lacks the 7-and 8-methyl groups.

FIG. 1. Photoreduction of T R R ( C~S '~' , S~~'~~) .
A solution of 40.9 nmol of T R R ( c y~'~~, s e r '~' ) in 1 ml of 50 mM sodium/potassium phosphate buffer, pH 7.6, 30 mM EDTA, containing 2 nmol of 3,lOdimethyl-5-deaza-isoalloxazine, was photoreduced by exposure to brilliant light at 4 "C (see "Materials and Methods"). Upper panel, spectra of the first phase of the reduction, showing maximum formation of the semiquinone. The spectra are of oxidized enzyme and enzyme photoreduced for 40 s, 70 s, 2 min, 3.5 min, 6 min, 9.5 min, and 19 min. Lower panel, spectra of the second phase of the reduction. The spectra are of enzyme photoreduced for 48 min, 66 min, 1.5 h, 1.9 h, and 3 h. Dashed spectrum, an extrapolation to full reduction of the enzyme. Inset, plot of A585 versus A454 of the photoreduction. The two solid lines are linear regressions of the first few exposures and the last few exposures. The point of intersection is used to calculate the extrapolated values of tm5 = 3,800 M" cm" and c454 = 3,500 M" cm" for the semiquinone species.

FIG. 2. Spectra of the three oxidation-reduction states of T R R ( C~S '~~, S~~'~' ) .
Solid spectrum, oxidized enzyme; dashed spectrum, extrapolated spectrum of the semiquinone species; dasheddotted spectrum, extrapolated spectrum of 2-electron reduced enzyme. reduced enzyme. The time required for the maximum disproportionation to be achieved precluded a complete phototitration given the limited stability of the reduced enzyme under anaerobic conditions. Inasmuch as the extinction coefficients for the semiquinone of wild type thioredoxin reductase and T R R ( C~S '~~, S~~'~~) are nearly equal, the values determined for T R R ( C~S '~~, S~~'~* ) were used to quantify the amount of semiquinone formed during reductive titrations of TRR(Ser'35,Cys'38). The value of t454 for fully reduced TRR(Ser'35,Cys'38) was determined to be 1,800 M" cm" after complete reduction of the enzyme with sodium dithionite (see "Materials and Methods"). When an anaerobic solution of TRR(Ser'35,Cys'38) in 50 mM sodium/potassium phosphate buffer, pH 7.6, 30 mM EDTA was exposed to brilliant light for 45 s (see "Materials and Methods"), approximately 60% of the enzyme was reduced to the semiquinone, which slowly disproportionated with a half-time of 18 min to a mixture of 45% oxidized, 29% semiquinone, and 26% 2-electron reduced enzyme. Thus, T R R ( C~S '~~, S~~'~~) stabilizes 2.4 times as much semiquinone as TRR(Ser'35,Cys'38).
Reductive titration of TRR(Ser'35,Cys'38) with sodium dithionite in the presence of a catalytic quantity of methyl viologen at pH 7.0 and 20 "C demonstrated that, as expected, 1 mol of dithionite/mol of FAD was required to fully reduce the enzyme (Fig. 3). The 2-electron reduced enzyme has an absorbance spectrum identical to the spectrum of 4-electron reduced wild type thioredoxin reductase (Zanetti and Williams, 1967;O'Donnell and Williams, 1983), and has an c454 = 1,700-1,900 M" cm". During the course of the titrations the maximum formation of semiquinone is 5-12% of the total enzyme, and the absorbance peak of reduced methyl viologen at 395 nm does not appear until the enzyme is completely reduced. The absorbance spectra of the enzyme at various stages of the titration reveal an isosbestic at 340 nm (Fig. 3).
The midpoint potentials for the 1-electron and 2-electron reductions of TRR(Ser'35,Cys'38) were determined from NADH/NAD' titrations in the pH range 6.0-9.0 at 20 "C. mV. The spectra of a titration performed at pH 7.2 are presented in Fig. 4 (after subtracting the absorbance contribution of the NADH as described under "Materials and Methods"). Profiles of the distributions of the three redox forms of the enzyme reveal that the equilibrium titration of TRR(Ser135,Cys138) results in a pH-independent maximum of 18-20% semiquinone formation (Fig. 5). The values of the potentials for the two 1-electron couples ( E p = Eox/Esq; El = E,,/Ed) have been calculated, and at pH 7.0 and 20 "C these values for TRR(Ser135,Cys'3s) are -306 and -256 mV, respectively. In contrast, during a reductive titration of T R R ( C~S '~' , S~~'~' ) with NADH in the presence of NAD+ and a mediator, benzyl viologen, the maximum amount of semiquinone (l-electron reduced and 3-electron reduced forms combined) formed increases from 30% at pH 8.9 to 53% at p H 6.0. Thus, the relative redox potentials, E2 and E l , and consequently, the amount of semiquinone stabilized, as well as the pH dependence of the relative potentials, are dependent on the polarity of the residue at position 138.
The plots The spectra, recorded after equilibrium was attained, are of oxidized enzyme and enzyme after addition of 11.8, 26.6, 53.3, 101, 189, and 357 nmol of NADH. The absorbance of the NADH at equilibrium was subtracted from the spectra.  Circles, E,; triangles, E2; diamonds, E,.

E l W T I
25% and above 75% reduction reflect the contribution of the l-electron reduction. A plot of the pH profiles of the midpoint potentials for the overall 2-electron reduction, E,, E,,/E,d couple, and the two l-electron reductions, Ez, E,,/E,q couple and E l , E,/E,d couple, for TRR(Ser'35,Cys'38) is presented (Fig. 7). The slope, AE,/ApH, of TRR(Ser'35,Cys'38) has a value of -57.1 mV. For a proton stoichiometry of 2 H'/2 eat 20 "C, this slope has a value of -58.2 mV (Clark, 1960). Thus, the 2-electron reduction of TRR(Ser'35,Cys'38) has a proton stoichiometry of 2.0 H+/2 e-. The slopes, AE2/ApH and AEl/ApH, for the l-electron reductions of TRR(Ser135,Cys'38) have values of -52.4 and -61.7 mV, respectively, indicating proton stoichiometries of 0.9 and 1.1 H+/e-, respectively. There are no breaks in the slopes of these plots, indicating that the reduction of the FAD is not linked to the ionization of any residues in the pH range 6.0-9.0.
Several reductive titrations of T R R ( c y~l~~, S e r '~~) , in the presence of a catalytic quantity of methyl viologen, with sodium dithionite in the pH 6.5-8.5 range revealed that 1.8-

Redox
Properties of Mutated Thioredoxin Reductase 2.0 mol of dithionite/mol of FAD were required to reduce this enzyme fully (Fig. 8). At approximately 85-90% reduction the absorbance peak of reduced methyl viologen at 395 nm appears. Additions of dithionite beyond the appearance of the reduced methyl viologen results in increases in the absorbance at 315 nm, indicating the presence of unreacted dithionite, and suggesting an apparent equilibrium involving dithionite, bisulfite, oxidized and reduced methyl viologen, and the two forms of the enzyme, E,, and Ered. This equilibrium might indicate that the midpoint potential for the ES,/Ered couple is close to the value of the oxidized/reduced couple of methyl viologen (-446 mV). However, the value of the midpoint potential of the E,/E,d couple of T R R ( C~S '~~, S~~'~' ) measured by potentiometric titrations with NADH is more positive than predicted from the results of the dithionite titrations. An explanation of this anomaly is not apparent. A value of 1,980-2,190 M" cm" for e454 of Ered was extrapolated from several titrations.
The requirement of 1 equivalent (eq) of dithionite to fully reduce TRR(Ser'35,Cys'38) is as expected (Fig. 3). The extrapolated requirement of 1.8-2.0 eq to fully reduce T R R ( C~S '~~, S~~'~' ) is unexpected. However, the appearance of reduced methyl viologen and excess dithionite prior to full reduction of the enzyme, and the requirement of greater than 2 eq to obtain full reduction suggest that the titration may involve a kinetically unreached equilibrium of the semiquinone and 2-electron reduced forms of the enzyme, oxidized and reduced methyl viologen, and the dithionite and sulfite couple (Fig. 8).
The quantification of thiols on T R R ( C Y S '~~, S~~'~' ) with DTNB under non-reducing and denaturing conditions and by conversion to cysteic acid (Prongay et al., 1989) have confirmed the replacement of C Y S '~~ by Ser. Oxidation of the thiol group of C Y S '~~ to a higher oxidation state, such as sulfenate would account for the uptake of 2 electrons from dithionite in excess of those required to reduce the FAD. A distantly related enzyme, NADH peroxidase, is thought to cycle in catalysis between the sulfenate and thiol states (Poole and Claiborne, 1989). However, a sulfenate would not be expected to react with DTNB; to our knowledge, this has not been tested due to the tendency of sulfenates to further oxidation (Capozzi and Modena, 1974;Kice, 1980). The thiol of C Y S '~~ is 4.43 a from the nearest point on the isoalloxazine  ring,3 making it unlikely that the flavin would reduce a sulfenate directly. However, a sulfenate ester between C Y S '~~ and SerI3' would be reduced by dithionite accounting for an additional equivalent of dithionite (Snyder and Carlsen, 1977). The putative hydrophobic environment of the active site in the vicinity of the FAD may serve to stabilize a sulfenate ester. The location of the hydroxyl group of SerI3' near the C-4a position of the electron deficient isoalloxazine ring may result in a polarization of the oxygen atom of the hydroxyl such that a sulfenate on C Y S '~~ can form a sulfenate ester with the hydroxyl. Upon denaturation of the protein the polarizing environment would be removed and the ester subject to hydrolysis. The conditions used for crystallization (ammonium sulfate and dithiothreitol) must also lead to the breakdown of the putative ester and reduction of the sulfenate since it is not observed in the structure of T R R ( C~S '~~, S~~'~' ) (Kuriyan et al., 1991). The presence of a second reducible group on T R R ( C~S '~~, S~~'~' ) was confirmed by reduction of the enzyme with NADH in the presence of NAD' ( Table I). The progressive increase in the extent of turbidity formation throughout the course of a reductive titration of T R R ( C~S '~~, S~~'~' ) with NADH decreased the accuracy of measuring the equilibrium concentration of NADH by its absorbance at 340 nm. Therefore, in two separate experiments the enzyme was partially reduced to approximately 20 and 57% total reduction by single additions of NADH to anaerobic enzyme containing NAD' a t pH 7.6 and 20 "C. After equilibrium was attained, the concentration of NADH was determined by measuring the absorbance at 342 nm, a wavelength at which all the enzyme species are isosbestic; the equilibrium concentrations of oxidized, semiquinone and 2-electron reduced FAD on the enzyme were determined as described under "Materials and Methods." The data support the presence of a second reducible center on the enzyme, which appears to require 2 mol of electrons/mol of enzyme (Table I). The midpoint potential for the 2-electron reduction of the FAD on this enzyme was estimated from these experiments to be -325 f 2 mV at pH 7.6 and 20 "C. Assuming a slope of -58.2 mV/pH, this corresponds to -290 mV at pH 7.0 and 20 "C. Thus, reduction of the FAD, especially the addition of the second electron, is more difficult with T R R ( C~S '~~, S~~'~' ) .
The titration of T R R ( C~S '~~, S~~'~' ) by NADH is shown in Fig. 9. The reduction of a putative second redox center at C Y S '~~ by electrons from NADH provides additional evidence that although the electrons are passed from the FAD to CYS'~' during catalysis by the wild type enzyme, the altered enzyme is capable of reducing the second redox center and, presumably, a mixed disulfide with the substrate at CYS'~~, albeit less efficiently (Prongay et al., 1989). A solution of 146 nmol of T R R ( C~S '~~, S~~'~' ) in 3.0 ml of 17.4 mM Na2HP04, 1.3 mM citrate, 54.2 mM KC1 buffer, pH 7.2, containing 1.5 pmol of NAD+, was titrated anaerobically with a 6.7 mM NADH solution at 20 "C. The spectra, recorded after equilibrium was attained, are of oxidized enzyme and enzyme after the addition of 26.8, 53.6, 93.8, 161, 335, and 1219 nmol of NADH. The absorbance of the NADH present at equilibrium was subtracted from the spectra.

DISCUSSION
The three-dimensional structures of T R R ( C~S '~~, S~~'~' ) and TRR(Ser'35,Cys'38) have been solved, and with this information, it is likely that the structure of the wild type enzyme will become available shortly.* The structure of T R R ( C~S '~~, S~~'~' ) was refined at 2-A resolution, with an Rfactor of 17.7%, including the placement of approximately 270 solvent molecules (Kuriyan et al., 1991). The structures confirm the prediction (Prongay et al., 1989) that the active center thiols are juxtaposed almost parallel to the isoalloxazine ring rather than perpendicular to it, as in glutathione reductase and lipoamide dehydrogenase (Schulz et al., 1978;Schierbeek et al., 19!9). The oxygen of Ser13' (Cys in wild type 2nzyme) is 3.05 A from $-4a, and the sulfur of C Y S '~~ is 4.50 A from N-5a and 4.78 A from C-4a, across the re face. The environment of the isoalloxazine is apolar as the spectrum would predict; only 2 solvent molecules are within hydrogen bonding distance of the isoalloxazine ring (Kuriyan et al., 1991).
Thioredoxin reductase is the only member of the pyridine nulcleotide-disulfide oxidoreductase family in which the microscopic redox potentials of the FAD/FADH2 and disulfide/ dithiol couples have been determined. The potential of the disulfide/dithiol couple is 11 mV more negative than is that of the FAD/FADH2 couple (O'Donnell and Williams, 1983). This is in contrast to lipoamide dehydrogenase, where the macroscopic potential for the first 2 electrons is some 66 mV more positive than that of the second 2 electrons (Matthews and Williams, 1976). In determining the potentials in thioredoxin reductase by equilibration with the NADH/NAD+ system, a correction was made for the enzyme present as semiquinone (O'Donnell and Williams, 1983).
The reductive titrations have revealed that both T R R ( C~S '~~, S~~'~' ) a n d TRR(Ser'35,Cys'38) form the blue, neutral semiquinone upon 1-electron reduction in the pH range of 6.0-9.0 ( Fig. 1-5), as is the case for wild type thioredoxin reductase from E. coli (Zanetti et al., 1968). The amount of semiquinone stabilized on T R R ( C~S '~~, S~~'~' ) is 2.5 times the amount stabilized on TRR(Ser'35,Cys138). Previous characterizations of these enzymes have demonstrated that CYS'~' is positioned near the C-4 and C-4a positions of the flavin. The exact position of C y P 5 was not determined in these studies, although it did not appear to be close enough t o interact directly with the C-4 or C-4a positions of the FAD (Prongay et al., 1989;Prongay and Williams, 1990). This is now confirmed by the x-ray crystal structures (Kuriyan et al., 1991). The differential stabilization of semiquinone by these enzymes may be a direct result of the juxtapositioning of Cy~'~'/Ser'~' to the FAD. Crystallographic studies of the oxidized and semiquinone forms of flavodoxin from Clostridium M P have demonstrated that the semiquinone is stabilized by a hydrogen bonding interaction between the carbonyl oxygen of residue 57 and the N-5 position of the FMN (Smith et al., 1977). Since the hydroxyl group has a far greater hydrogen bonding potential than does a thiol group (Crampton, 1974), stabilization of the semiquinone through a hydrogen bonding interaction between the side chain of residue 138 of thioredoxin reductase and the O(4a) position of the FADH. would be more efficient when residue 138 is a serine than when it is a cysteine. The O(4a) is negatively charged in the zwitterionic resonance form of the neutral flavin radicaL4 Since the side chain of residue 135 does not appear to interact directly with the FAD (Prongay et al., 1989;Prongay and Williams, 1990), the contribution of this residue to the stabilization of the semiquinone is expected to be less than that of residue 138. Thus, the greater stabilization of semiquinone by T R R ( C~S '~~, S~~'~' ) t h a n by TRR(Ser'35,Cys'38) provides additional evidence that CYS'~' is positioned nearer to the C-4 and C-4a positions of the FAD than is CYS'~~. It is concluded, therefore, that CysI3' accepts electrons from FADHz in catalysis.
Potentiometric titrations of wild type thioredoxin reductase have shown that at pH 7.0 and 12 "C the disulfide/dithiol and FAD/FADH, couples have midpoint potentials separated by only 11 mV with a 17 mV negative interaction (O'Donnell and Williams, 1983). At pH 7.0 and 12 "C the midpoint potential of the FAD/FADH2 couple of the dithiol form of the enzyme has a value of -260 mV, and the AE,/ApH of this couple is -60 mV/pH, indicating a 2 H+/2 e-stoichiometry (O'Donnell and Williams, 1983). Applying a temperature correction factor of -1.3 mV/1 "C (Clark, 1960) to the E , of this couple yields a value of -270 mV at pH 7.0 and 20 "C. At pH 7.0 and 20 "C the E, values of the FAD/FADH2 couple of T R R ( S~I -'~~, C~S '~~) a n d T R R ( C~S '~~, S~~'~~) are -280 and -290 mV, respectively ( Fig. 6 and Table I). Thus, the replacement of C Y S '~~ or CysI3' with a serine causes a 10 or 20 mV decrease, respectively, in the midpoint potential of the FAD/ FADH2 couple relative to wild type thioredoxin reductase. These results are consistent with the small negative interaction (17 mV) that the dithiol has on the FAD/FADH2 couple relative to the effect of the disulfide on this couple seen with wild type thioredoxin reductase (O'Donnell and Williams, 1983). These results demonstrate that the increased polarity resulting from the replacement of either thiol exerts a similar effect on the redox potential of the FAD/FADH2 couple. On the other hand, the hydroxyl group of Ser13' exerts a far greater effect than Ser135 toward stabilizing the semiquinone species.