Proton Stoichiometry in the Reduction of the FAD and Disulfide of Escherichia coli Thioredoxin Reductase EVIDENCE FOR A BASE AT THE ACTIVE SITE*

The oxidation-reduction midpoint potentials, E,, of the FAD and active site disulfide couples of Escherichia coli thioredoxin reductase have been determined from pH 5.5 to 8.5. The FAD and disulfide couples have similar E, values and thus a linked equilibrium of four microscopic enzyme oxidation-reduction states exists. The binding of phenylmercuric acetate to one enzyme form could be monitored which allowed solving the four microscopic E, values. The E,,, values at pH 7.0 and 12 “C of the four couples of thioredoxin reductase are: (S),-enzyme-PAD/FADH2 = -0.243 V, (SH),-en- zyme-FAD/FADH2 = -0.260 V, (FAD)-enzyme-(S)z/ (SH), = -0.254 V, and (FADH,)-~~z~~~-(S)~/(SH)~ = -0.271 V. Thus, at pH 7.0, the FAD and disulfide moieties have a 0.017-V negative interaction and E, values which are different by 0.011 V. The AE,,,/ApH of the FAD couples E; and E; are about 0.060 V/pH throughout the pH range studied, showing an approx- imately 2-proton stoichiometry of reduction of the enzyme FAD. The AE,/ApH of the disulfide couples E; and E% are about 0.052 V/pH from pH 5.5 to 8.5, showing

containing an oxidation-reduction active disulfide (Zanetti and Williams, 1967;Moore et al., 1964). Thioredoxin reductase contains an FAD and an oxidation-reduction active site disulfide (Zanetti and Williams, 1967;Moore et al., 1964;Ronchi and Williams, 1972). It is thought that the electrons flow sequentially from NADPH to the FAD, from the FAD to the disulfide, and from the dithiol on thioredoxin reductase to the disulfide on thioredoxin. In a 4-electron reduction of thioredoxin reductase at pH 7.6, there is a gradual bleaching of the flavin absorbance throughout the titration (Zanetti and Williams, 1967). Thus, at pH 7.6, the FAD and disulfide appear to have similar oxidation-reduction potentials. The reduction of thioredoxin reductase by NADH is described by the linked equilibria of four enzyme microforms shown in Scheme 1. Use of the nonphysiological pyridine nucleotide avoids the complexes characteristic of reduction with NADPH and is faster than reduction with dithionite.
The two thiols generated upon reduction of the active site disulfide are expected to carry out different tasks by analogy with lipoamide dehydrogenase Williams, 1976a and1976b) and glutathione reductase (Arscott et al., 1981). Presumably, FADH, reduces the active site disulfide by transfer of electrons to the disulfide via the sulfur which is proximal to the FAD (referred to as the flavin thiol), and the electrons in the active site dithiol are passed to the disulfide of thioredoxin via the other active site thiol, referred to as the interchange thiol (Scheme 2).
Model studies show thiol-disulfide interchange to be initiated by attack of a thiol anion on the disulfide (Foss, 1961). Rapid reduction of the disulfide bond in thioredoxin by thioredoxin reductase proceeds at pH 7.0, well below the pK value of model thiols (Moore et al., 1964). Thus, the environment at the active site of thioredoxin reductase must lower the pK of the interchange thiol. There is substantial evidence for a low pK value of an active site cysteinyl thiol (pK = 3-5) in four enzymes: papain (Polgar, 1973;Lewis et al., 1976), glyceraldehyde-3-phosphate dehydrogenase (Polgar, 1973), lipoamide dehydrogenase (Matthews et d., 1977), and glutathione reductase (Arscott et al., 1981). A thiol-base ion pair in a relatively apolar milieu is hypothesized to explain the low pK value of the thiol in each of these enzymes. The resulting thiolate in lipoamide dehydrogenase and glutathione reductase is the donor in a charge transfer complex with the flavin (Scheme 2) (Kosower, 1966;Massey and Ghisla, 1974;Searls and Sanadi, 1961). Because the oxidation-reduction potentials of the two couples are widely separated, two-electron reduced enzyme can be considered a single species in which the sulfurs are reduced and the FAD is not reduced (in the ground state). No such charge transfer complex has been observed in thioredoxin reductase and the oxidation-reduction potentials of the two couples appear to be similar at pH 7.6 (Scheme 1). In order to understand the differences and similarities in the mechanisms of the three pyridine nucleotide-disulfide oxidoreductases, the proton stoichiometry of reduction of the FAD and disulfide couples of thioredoxin reductase (Scheme 1) were determined by measuring the E, values of the couples over a range of pH values.

RESULTS
The FAD and disulfide centers in thioredoxin reductase are reduced in parallel, generating significant quantities of both microforms I1 and I11 during a titration (Scheme 1). The stoichiometry of reduction of individual microforms may differ. Thus, a full description of the reduction of thioredoxin reductase requires determining the concentrations of each enzyme microform at the different levels of reduction in a titration, as well as the concentrations of NAD' and NADH. These values are obtained from the combined results of two experiments. An NADH titration gives the concentrations of NAD' and NADH and the macroscopic quantities, enzyme-FAD, enzyme-FADHP, enzyme-(S), (enzyme containing a disulfide), and enzyme-(SH), (enzyme containing a dithiol). Addition of PMA,' a reagent which binds tightly to thiols, to a two-electron reduced solution of thioredoxin reductase pulls the intramolecular equilibrium between microforms I1 and I11 to the microform 11. PMA complex as monitored by the rapid jump in absorbance at 456 nm. A PMA addition experiment resolves the macroscopic quantities of oxidized and reduced ' Portions of this paper (including "Materials and Methods" and Table I) are presented in miniprint at the end of this paper. Miniprint is easily read with the aid of a standard magnifying glass. Full size photocopies are available from the Journal of Biological Chemistry, 9650 Rockville Pike, Bethesda, MD 20814. Request Document No. 83M-777, cite authors, and include a check or money order for $7.20 per set of photocopies. Full size photocopies are also included in the microfilm edition of the Journal that is available from Waverly Press.

~'nroredoxln fieductme
FAD and disulfide into the concentrations of the four enzyme microforms. The separate E,,, values of each of the four enzyme couples (Scheme 1) are calculated from the concentrations of the four enzyme microforms in equilibrium with the oxidized and reduced species of titrant of known E, using the Nernst relationship. The titrants used in this study were NADH and an analog which has a more positive E,, APADH, in order to avoid the complexes formed between the enzyme and its physiological pyridine nucleotide substrate.
The detailed explanation of the measurements of the equilibrium concentrations of the oxidized and reduced macroscopic enzyme species (FAD/FADH,, (S)2/(SH)z) and titrant species is given in the Miniprint. Briefly, the concentrations of enzyme-FAD and enzyme-FADH, at each titration point were calculated from absorbance measurements at 456 nm; the concentration of NADH (APADH) in equilibrium with thioredoxin reductase was obtained from absorbance increases at an isosbestic point for thioredoxin reductase reduction (= 347 nm); the concentration of NAD' (APAD+) in equilibrium with the enzyme was the difference between the titrant NADH added and the NADH in equilibrium with the enzyme; the concentration of the reduced active site disulfide (enzyme-(SH),) was the difference between the amount of titrant oxidized by the enzyme and the amount of enzyme-FADH,; and the concentration of the oxidized active site disulfide (enzyme-(S),) was the difference between the total enzyme concentration and the enzyme-(SH), concentration.
The profiles of enzyme-FADH2 formed during anaerobic titrations by NADH are shown in Fig. 1 for pH values of 6.0 and 8.1. At pH 6.0, the E,,, values of the two FAD couples E; and E; are higher (less negative) than the E, values of the two disulfide couples E!,, amd p,, respectively, leading to greater concentrations of enzyme-FADH2 relative to enzyme-(SH), throughout the titration (e.g. approximately 70% FADH, and 30% dithiol at 1 equivalent of reduction). At pH 8.0, the enzyme-FADH, titration profile is a straight line, showing that the potentials of the two FAD couples E; and E; equal the potentials of the two disulfide couples E,!,, and P,, respectively. Since the E,,, of the FAD couple increases relative to the E,,, of the disulfide couple with decreasing pH, the disulfide couple must have a lower H' stoichiometry of reduction than the FAD couple.
The unequal values of E, for the FAD and disulfide couples at pH 6.0 afforded a unique opportunity to measure the rate of electron transfer among individual microforms. This is important since it is essential to the calculation of E,,, values that the oxidized and reduced enzyme species are at equilibrium when the absorbance measurements are made. The FAD of 1-equivalent reduced thioredoxin reductase is 70% reduced at pH 6.0. An anaerobic solution of oxidized enzyme was added to an equal amount of fully reduced enzyme (stoichiometric sodium dithionite) to give a solution initially of 50% enzyme-FADH,, which increased to a concentration of 70% upon equilibrating to a mixture of the four enzyme microforms shown in Scheme 1. The approach to equilibrium was monitored a t 456 nm and found to occur with a half-life of about 8 min at pH 6.0. The addition of a 5-fold excess (over total enzyme) of PMA to the equilibrating mixture slowed the approach to equilibrium over loo-fold, implying that thioldisulfide interchange mediates the equilibration of microform I with microform IV to give a microforms I1 and 111. At pH values much above 6.0, the concentration of enzyme-FADH, is approximately equal to the concentration of enzyme-(SH)n, thus precluding measurements of equilibration rates among these microforms. However, if electron transfer between microforms I and IV occurs via thiol-disulfide interchange, the rate of equilibration of these microforms should increase with increasing values of pH. The rate of equilibration between microforms I1 and I11 is an intramolecular electron transfer and a catalytic step and is therefore rapid on the time scale of the experiments in this study. To ensure that the criterion of equilibrium among enzyme microforms was fulfilled, absorbance measurements were not recorded until the absorbance changes had stabilized to the slow, steady changes due to the comproportionation of FAD and FADH, enzyme species to form semiquinone as described in the Miniprint.
Calculation of the apparent E,,, value for the macroscopic FAD/FADH2 couple and for the macroscopic (S),/(SH), couple a t each titration point showed their apparent E, , , values decreased steadily throughout the titration. For example, calculations of the macroscopic FAD/FADH, couple of thioredoxin reductase at pH 7.0 shows a difference in E, of 0.007 V between the fractional FADH2 levels of 0.38 and 0.78. The difference in E,,, for the macroscopic (S),/(SH), couple in the same experiment is 0.010 V between the fractional (SH), levels of 0.19 and 0.53. Furthermore, for the enzyme-FADH, titration profiles a t all pH values (e.g. Fig. 1 for pH 6.0 and 8.1), a line drawn through the first few data points intersects with a line drawn through the last few data points at 1.0 & 0.1 Equivalents (see Miniprint for explanation of Equivalents). Thus, both of the macroscopic couples (FAD/FADH, and (S)*/(SH)J tend to have a slightly lower apparent value of E,,, throughout a titration. This may be explained by a small intramolecular negative interaction between the FAD and disulfide moieties of the enzyme. Specifically, in Scheme 1, the E & value for the FAD couple where the disulfide is oxidized in these microforms is higher than the E : value for the FAD couple where the microforms contain a dithiol. In other words, reduction of the disulfide lowers the E, value of the FADJFADH, couple and, conversely, reduction of the FAD lowers the E, , , value of the (S),/(SH), couple. In Scheme 1, an interaction between the FAD and disulfide is described by the ratio K1/K4 (= K2/K3). This ratio is 1.0 for the case of no interaction between the FAD and disulfide and is about 4 for a negative interaction of the magnitude observed in this study. A non-Nernstian behavior could also be expected if the enzyme solution contained enzyme that was partially denatured. Two different preparations of thioredoxin reductase exhibited the same titration behavior, making this explanation less likely.
A full description of the stoichiometry of reduction of the FAD and disulfide centers of thioredoxin reductase requires separating the macroscopic measurements of FAD/FADH2 and (S),/(SH), into the four microscopic enzyme forms of Scheme 1. These quantities allow the E, values for the four microscopic enzyme couples to be calculated a t different pH values and the H' stoichiometry of reduction of each microform to be obtained from the slope AEJApH. Quantitation of the four enzyme microforms at 1 equivalent of reduction was accomplished in a separate experiment at each pH value using PMA as a probe of the equilibrium concentrations of enzyme microforms.
The organic mercurial p-chloromercuriphenylsulfonate binds tightly to the dithiol of reduced thioredoxin reductase (Zanetti and Williams, 1967). Addition of PMA to a partially reduced solution of thioredoxin reductase results in a rapid change in the absorbance at 456 nm. The absorbance change is due to PMA binding to microform I1 which pulls microform I11 into the microform II.PMA complex via the rapid intramolecular electron transfer equilibrium between microform I11 and microform I1 (catalytic step). Maximal absorbance changes upon adding PMA were obtained with a 1.6-fold excess of PMA over total enzyme. A &fold excess of PMA over total enzyme was used routinely in PMA addition experiments.
The concentration of each of the four enzyme microforms at any level of reduction before PMA addition can be calculated from a PMA addition experiment. These calculations are detailed in the Miniprint. Briefly, the concentration of microform I11 before PMA addition is calculated from the increase in absorbance at 456 nm upon adding PMA. The concentration of microform IV before PMA addition is equal to the residual FADH, after the addition of PMA, the only microform containing FADH, after PMA addition. The concentration of microform I1 before PMA addition is the product of the enzyme concentration and the number of equivalents used to reduce the enzyme minus the sum of the concentrations of microforms I11 and IV. The concentration of microform I before PMA addition is the difference between the total enzyme concentration and the sum of the concentrations of the other microforms.
Sodium dithionite is used to reduce thioredoxin reductase in PMA addition experiments since pyridine nucleotide catalyzes electron transfer between the microform IV . PMA complex and oxidized enzyme (microform I) to produce the microform II.PMA complex, resulting in anomolously high estimates for the concentration of microform 111. Anaerobic reductions of thioredoxin reductase by sodium dithionite are not quantitative except at pH values below 6.5. Hence, the number of sodium dithionite reducing equivalents in thioredoxin reductase before PMA addition is obtained by calculating the percentage of enzyme-FADH, from the absorbance at 456 nm and comparing this value with the enzyme-FADH, titration profile ( Fig. 1) at the same pH value. Sodium dithionite reductions of thioredoxin reductase take about 1 h, ensuring that the enzyme forms are at equilibrium and validating the comparison with an NADH titration.
The equilibrium concentrations of the four enzyme microforms of Scheme 1 at one level of reduction allows the important ratios, K J K , and K1/K4 to be calculated.  [IV]) is a quantitative measure of the strength of the interaction between the FAD and disulfide centers. Hence, these ratios fully describe the equilibrium distribution of the enzyme microforms of Scheme 1 a t any level of reduction (derived in the Miniprint). For example, at pH 6.0, the concentrations of enzyme microforms are: microform I = 2.75 pM, microform II = 2.15 pM, microform 111 = 8.88 pM, and microform IV = 1.82 pM. The ratios calculated from the concentrations of enzyme microforms are: K1/K2 = 0.242 and K,/K4 = 3.82. The distribution of enzyme microforms throughout a titration calculated from these ratios is displayed in Fig. 2. The percentage of enzyme-FADH2 is the sum of the percentages of microforms 111 and IV. Thus, the ratios obtained from a PMA addition experiment allow simulation of the enzyme-FADH2 titration profile as shown by the solid lines in Fig. 1 for the ratios obtained from PMA addition experiments at pH 6.0 and 8.1. The ratios K1/K2 and KJK4 obtained from PMA experiments at the values of pH and concentrations of thioredoxin reductase used in this study are given in Table 11. It can be seen from Fig. 2  addition experiment using 15.6 @M enzyme in Buffer A at pH 6.0, 1 2 "C.
the Nernst relationship in Fig. 3 for experiments at pH values of 6.0, 7.0, and 8.1. The solid lines drawn through the data points represent theoretical curves of E,, uersus fractional reduction calculated for a %electron reduction assuming the   Table 11. The E,,, values at pH 8.0 and 8.5 were determined by NADH titrations and PMA addition experiments performed in 0.1 M NaIP207, 0.3 mM NaEDTA-HC1. The E , values at pH 8.3 were determined by an NADH titration and a PMA addition experiment performed in 50 mM Tris, 50 mM KzHPO,, 0.3 mM Na4EDTA-HCl. All other experiments were performed in Buffer A. The solid lines drawn through the data points represent the least squares analyses of the data. The dashed lines represent the dependence of E, on pH for the NAD+/NADH couple. At pH 6.0 and 7.0, the equilibrium between pyridine nucleotide and enzyme greatly favors the oxidation of NADH, resulting in equilibrium concentrations of NADH which are too low to measure until late in the titration. Thus, a t these pH values, NAD+ was added prior to the titration, allowing equilibrium measurements of NADH at lower values of fractional enzyme reduction.
The validity of E , values calculated from measurements of enzyme species in equilibrium with a reference couple rests on the condition that complex formation between enzyme and titrant is negligible, or equal in strength for all enzyme microforms. A long wavelength charge transfer band is associated with complex formation between NADP+ and reduced thioredoxin reductase (Zanetti and Williams, 1967). Charge transfer is not observed between reduced thioredoxin reductase and either NAD+ or APAD' under the conditions employed, suggesting that neither NAD' nor APAD' forms a complex with thioredoxin reductase. To test further for complexation, titrations of the enzyme with NADH were performed using different concentrations of enzyme. Oxidation-reduction equilibria are independent of concentration, whereas complexation equilibria are concentration dependent. Thus, doubling the concentration of enzyme will increase the extent of complexation 4-fold. If significant complexation occurred preferentially between one oxidation-reduction form of thioredoxin reductase and one oxidation-reduction form of pyridine nucleotide, the calculated midpoint potential should shift as the enzyme concentration is varied. The values of E , for the four enzyme couples were determined using different concentrations of enzyme (9-44 p M ) and were within 0.002 V. This difference is within the experimental error of the system.
These data are shown in Fig. 3 for the FAD and disulfide couples (EL and b , , respectively) at pH values of 6.0,7.0, and 8.1.
The E , values for the four microscopic couples of thioredoxin reductase were calculated from pyridine nucleotide titrations and PMA addition experiments at pH values spanning the range from 5.5 to 8.5. The results of the first equivalent of reduction for the FAD couple, E:, and the disulfide couple, E,?,,, are shown in Fig. 4A.3 The results of the second equivalent of reduction for the FAD couple, E,, and the disulfide couple, E%, are shown in Fig. 4B. The variation of E2 with pH has a slope, AEf/ApH, of 0.060 V/pH and a correlation coefficient of 0.997 by a least squares analysis. The AE,/ApH for an oxidation-reduction couple with a 2proton stoichiometry is, theoretically, 0.0566 V a t 12 "C (Clark, 1960, Appendix, Table E). Thus, the results of Fig. 4A show that the reduction of the FAD in thioredoxin reductase containing an oxidized disulfide has a 2.1 -H' stoichiometry throughout the pH range studied. The slope of the disulfide couple AEL/ApH is 0.052 V/pH from pH 5.5 to 8.5 employing a least squares analysis (correlation coefficient = 0.997). Thus, reduction of the disulfide in thioredoxin reductase containing an oxidized FAD is accompanied by a 1.8 -H' stoichiometry from pH 5.5 to 8.5.
The @,/ApH and AE$/ApH profiles for the second equivalent of enzyme reduction (Fig. 4B) are similar to the profiles for the first equivalent of enzyme reduction (Fig. 4A The small differences in AE,/ApH profiles between a preliminary study (O'Donnell and Williams, 1981) and the aE,/ApH profiles of Fig. 4, A and B are due to the use of extinction coefficients of enzyme species determined at pH 7.0 for calculating values of E,,, at all pH values and also for not making dilution corrections for the concentration of titrant NADH added. In addition, the E, values at pH 7.0 of the pyridine nucleotide couples and the value of RT/NF were not corrected for temperature in the preliminary study.  Table 11; the multiple determinations at pH 6.0, 7.0, and 8.1 have been averaged. Each solid titration curue is a nonlinear least squares fit to the data.
Values for the EL and E$ couples are calculated from the ratio K J K , and the EL and E& values (explained in the Miniprint). The AEj,/ApH (Fig. 4B) is 0.060 V between pH 5.5 and 8.5 and has a correlation coefficient of 0.996 by a least squares analysis. Thus, reduction of the FAD in thioredoxin reductase containing a reduced disulfide has an approximately 2.1 -H+ stoichiometry throughout the pH range studied. The &,/ApH is 0.052 V between pH 5.5 and 8.5 by a least squares analysis (correlation coefficient = 0.996). Thus, reduction of the disulfide in thioredoxin reductase containing a reduced FAD has an approximately 1.8 -H' stoichiometry between pH 5.5 and 8.5.
The slopes AEJApH for the disulfide couples of thioredoxin reductase show that the proton stoichiometry in the reduction of the disulfide is about 0.2 -H+/pH unit less than theoretical throughout the range of study. This suggests the presence of a base the ionization behavior of which is linked to the oxidation-reduction state of the disulfide. Such a base will alter the intramolecular equilibrium between the FAD and disulfide couples in 2-electron reduced enzyme as shown in Scheme 3. The ionizable group on the dithiol enzyme can be either the same ionizable amino acid side chain as in disulfide enzyme (except with a lowered pK on dithiol enzyme) or one of the nascent active site thiols having a low pK due to an interaction with the protonated base, i.e. a thiolbase ion pair. The formation of a thiol-base ion pair having a thiol anion with a low pK is predicted from the chemistry of thioredoxin reductase catalysis and has a precedent in a number of enzymes including the closely related flavoenzymes lipoamide dehydrogenase and glutathione reductase. The equilibrium constant between microforms I1 and I11 is the ratio K,/K, (Scheme 1). The value of Kl/Kz is obtained from the data of a PMA addition experiment. The intramolecular equilibrium constant Kl/Kz (Table 11) and the inverse K2/Kl are plotted as a function of pH in Fig. 5. The interpretations of these plots are derived in the Miniprint. The pK value from the Kl/Kz plot is that of a group on microform 111 (pKs, Scheme 3) and the pK value from the K2/K, plot is that of a group on microform I1 (pK5, Scheme 3). The values of pKs and pK5 are 7.59 and 6.98, respectively. The values for the intramolecular equilibrium, microform I1 to microform 111, for the fully protonated (KT) and deprotonated (K8) enzyme forms are obtained from the acidic and basic limbs of the theoretical fits to the data of Fig. 5. Clark (1960) develops equations for situations similar to that just described in which the enzyme species that contain a disulfide (111) have an ionization with a pK greater than  6. Variation of the midpoint potential with pH where a pK for oxidized species is less than one pH unit higher than a pK for reduced species. pK on the oxidized species, 7.6; pK on the reduced species, 7.0; observed slope, 0.050 V/pH unit at 12 "C; observed midpoint potential at pH 7.0, -0.254 V; correlation coefficient, 0.998 (Clark, 1960, p. 128).  12, 7.67, 7.05, 6.63, 6.01,5.72. Inset: plot of the per cent enzyme FAD fluorescence uersus pH. Excitation wavelength was 458 nm; emission was at 550 nm. The solid line drawn through the data points is a nonlinear least squares fit of the data to a single ionization (pK = 7.03).
that of an ionization on enzyme species that contain a dithiol (11) (Clark, 1960, pp. 118-130). Following Clark, we have derived an equation relating E , to pH for this case in the Miniprint. If the two pK values were more widely separated, the E,,, versus pH curve would have a slope of 0.0566 V below the acid pK, a slope of 0.0283 V/pH between the two pK values, and a slope of 0.0566 V/pH above the alkaline pK. Since the two pK values are separated by only 0.6 pH unit, the slopes merge as shown for an ideal case in Fig. 6. The actual value of the E , at a pH equal to a pK will be 0.0086 V above or below the theoretical slopes (dashed lines) at 1 2 "C (Clark, 1960, p. 128). Thus, for the case in point, the data will appear to define a straight line. Using the equation relating E , and pH (see Miniprint) and the pK values determined in There is also a base near the FAD in oxidized thioredoxin reductase as indicated by the dependence of the FAD fluorescence on pH shown in Fig. 7 . A theoretical fit to the data for a single ionization yields a pK of 7.03, the line drawn through the data points of Fig. 7 . Further evidence for a pK on the oxidized enzyme comes from close examination of the absorbance spectra revealing a pH-dependent change in the ratio of extinction coefficient of the 380 nm peak to the 455 nm peak from 1.03 at pH 7.6 to 0.99 at pH 6.0.

DISCUSSION
The oxidation-reduction midpoint potentials (E,) of the FAD and disulfide couples in thioredoxin reductase have been determined at pH values spanning the range 5.5-8.5. The proton stoichiometry of the disulfide couple (obtained from the slope of E, versus pH plots) was 1.8 protons while the proton stoichiometry of the FAD couple was 2.1 protons. The proton stoichiometry of the disulfide couple compared to the FAD couple is reflected in the observation of an increased ratio of flavin reduction to disulfide reduction during titrations performed at low pH relative to titrations at high pH. The proton stoichiometry results disclose the presence of a basic amino acid side chain with an ionization behavior that is linked to the oxidation-reduction state of the disulfide.
The ionization of an enzyme base linked to the oxidationreduction state of the disulfide is also found in the flavopro-SCHEME 4 teins lipoamide dehydrogenase and glutathione reductase which contain an active center disulfide. In lipoamide dehydrogenase, an essential base on oxidized enzyme has a pK of less than 5.5 (Matthews et al., 1977). Upon reduction of the active center disulfide in lipoamide dehydrogenase, a thiolbase ion pair forms in which the thiol has a pK of about 4.8,4 and the pK of the base is shifted to 7.8 (Matthews et al., 1977). Thiol-base ion pairs involving thiols of low pK have also been demonstrated in papain (Polgar, 1973;Lewis et aL., 1976), glyceraldehyde-3-phosphate dehydrogenase (Polgar, 1975), and glutathione reductase (Arscott et al., 1981).
The proton stoichiometry of the disulfide couple in thioredoxin reductase is consistent with the formation of a thiolbase ion pair upon reduction of the disulfide. The ion pair hypothesis for thioredoxin reductase is shown in Scheme 4 where the pK of the base is assumed to be the same in enzyme containing a disulfide as in enzyme having a fully protonated dithiol. This assumption is reasonable since the disulfide and the dithiol are both uncharged. The data of the PMA addition experiments yield estimates for the pK values of the group on disulfide enzyme of 7.59 and the group on dithiol enzyme of 6.98. These pK values correspond to the ionization constants of the base (Kc) and the thiol (Ks) on enzyme containing a fully protonated dithiol (form B, Scheme 4). The E,,, versus pH plots for the disulfide couples of thioredoxin reductase (Fig. 4) do not show a break in the slope up to pH 8.5, indicating that more than one proton associates with the enzyme upon disulfide reduction at least up to pH 8.5. Since the enzyme form having a deprotonated base and a thiol anion (form E, Scheme 4) has only one proton, this species must not exist in significant amounts below pH 8.5. Thus, the pK values of the ion pair base (Klo) and the thiol on enzyme with a deprotonated base ( K g ) (forms C and D, respectively, Scheme 4) must be greater than 8.5. For the linked equilibria of Scheme 4, pK, + pKg = pKs + pKio. Thus, for a value of pKg that is greater than 8.5, the value of pKlo must be greater than 9.1. The intramolecular equilibrium constant for transfer of a proton from the thiol to the base is about 4.0 (K5/K6, Scheme 4) and, thus, dithiol enzyme exists mainly as an ion pair at physiological pH.
The ion pair is an attractive hypothesis because the ionization behavior of both the thiol and base of the ion pair fulfill needed functions predicted by the chemistry of thioredoxin reductase catalysis. Specifically, thiol-disulfide interchange reactions are known to be initiated via attack on the disulfide by a thiol anion (i.e. transfer of electrons between the dithiol of thioredoxin reductase to the disulfide of thioredoxin). Since, in the cell, the direction of electron flow is from NADPH to thioredoxin, thioredoxin reductase must initiate the thiol-disulfide interchange reaction with thiore-Active Center Base in Thioredoxin Reductase doxin. The inherent nucleophilicity of a thiolate is dependent on the pK of the thiol. The higher the pK of the thiol, the greater the nucleophilicity of the thiol anion (Wilson et al., 1977;Shaked et al., 1980). However, since the reactive species is the thiol anion, the pK of the thiol must be low enough to yield a significant concentration of thiol anion at the pH of the reaction (Jencks, 1969). Hence, the value of 6.98 for the pK of the putative ion pair thiol of thioredoxin reductase yields a thiol which is largely in a deprotonated state at physiological pH. In addition, formation of the mixed disulfide between thioredoxin reductase and thioredoxin would be concerted with an increase in the acidity of the protonated base which could function as a proton donor to the nascent thiolate of thioredoxin. Thus, both a nucleophilic thiol anion and a protonated base are required for efficient catalysis as encompassed in the thiol-base ion pair model.
The ion pair of lipoamide dehydrogenase and the putative ion pair of thioredoxin reductase differ in that the thiol anion and base in lipoamide dehydrogenase have pK values that are about 1.5 pH units lower than the corresponding pK values of the putative ion pair thiol anion and base in thioredoxin reductase. Thioredoxin reductase also differs from lipoamide dehydrogenase and glutathione reductase in that the putative thiolate in thioredoxin reductase does not charge transfer to the FAD. The lack of charge transfer in thioredoxin reductase could be due to an incorrect juxtaposition of the thiolate relative to the FAD or a suboptimal ionization potential of the thiolate.
The proton stoichiometry of about 2 for the reduction of the FAD to FADH2 in thioredoxin reductase is consistent with the spectrum of the enzyme-FADHP  and the pH independence of the reduced spectrum. The resolved spectrum of the FAD in thioredoxin reductase reveals that the FAD is in a hydrophobic environment. Although free FADH, has a pK of 6.5, an elevation of this pK value is expected in the hydrophobic milieu of thioredoxin reductase.
The function of the slight negative interaction between the FAD and disulfide moieties is not clear but could be simply to increase the concentrations of microforms I1 and I11 relative to microforms I and IV (Scheme 1). This would be advantageous if a mixture of microforms I1 and I11 are more catalytically competent than microform IV.
The methods applied here to thioredoxin reductase will be applicable to any oxidation-reduction protein containing two reducible moieties provided that a unique property can br exploited to determine the concentration of any one of the oxidation-reduction species. In addition, the plot used in Fig.  5 gives a more accurate alternative to the classical method of Clark (1960) for the determination of pK values of bases where the pK is linked to the oxidation-reduction state.