Interactions of Quinones with Thioredoxin Reductase

Mammalian thioredoxin reductases (TrxR) are important selenium-dependent antioxidant enzymes. Quinones, a wide group of natural substances, human drugs, and environmental pollutants may act either as TrxR substrates or inhibitors. Here we systematically analyzed the interactions of TrxR with different classes of quinone compounds. We found that TrxR catalyzed mixed single- and two-electron reduction of quinones, involving both the selenium-containing motif and a second redox center, presumably FAD. Compared with other related pyridine nucleotide-disulfide oxidoreductases such as glutathione reductase or trypanothione reductase, the kcat/Km value for quinone reduction by TrxR was about 1 order of magnitude higher, and it was not directly related to the one-electron reduction potential of the quinones. A number of quinones were reduced about as efficiently as the natural substrate thioredoxin. We show that TrxR mainly cycles between the four-electron reduced (EH4) and two-electron reduced (EH2) states in quinone reduction. The redox potential of the EH2/EH4 couple of TrxR calculated according to the Haldane relationship with NADPH/NADP+ was –0.294 V at pH 7.0. Antitumor aziridinylbenzoquinones and daunorubicin were poor substrates and almost inactive as reversible TrxR inhibitors. However, phenanthrene quinone was a potent inhibitor (approximate Ki = 6.3 ± 1 μm). As with other flavoenzymes, quinones could confer superoxide-producing NADPH oxidase activity to mammalian TrxR. A unique feature of this enzyme was, however, the fact that upon selenocysteine-targeted covalent modification, which inactivates its normal activity, reduction of some quinones was not affected, whereas that of others was severely impaired. We conclude that interactions with TrxR may play a considerable role in the complex mechanisms underlying the diverse biological effects of quinones.

dithiol motif substitutes for the role of the selenolthiol in the mammalian enzyme (3,6).
Many factors make mammalian TrxR an important target of drugs and xenobiotics, as recently reviewed (10). The broad substrate specificity of mammalian TrxR is expected to provide antioxidant function at several levels, coupled to reduction of cytosolic lipoate, selenium compounds, hydroperoxides, and ubiquinone-10 (see ; this would be perturbed if TrxR becomes inhibited in cells. Another important antioxidant system that would suffer from TrxR inhibition is that of the mammalian peroxiredoxins, which all require functional Trx for regeneration (16). Furthermore, cytokine functions of extracellular Trx, as well as increased Trx and TrxR levels in some tumor cell lines and in synovial fluid or tissue of patients suffering from rheumatoid arthritis (10,17), argue for targeting of TrxR as a possible therapeutic approach in certain diseases. In fact, TrxR is inactivated by several compounds in clinical use, including 1,3-bis-(2-chloro-ethyl)-1-nitrosourea (BCNU) (18), anticancer platinum compounds (19), antiarthritic gold compounds (20,21), and immunostimulatory dinitrohalobenzenes (22). Modification of TrxR by dinitrohalobenzenes yields covalently linked dinitrophenyl groups that seem to act as mediators between the enzyme-bound FAD and oxygen, thus enhancing the superoxide-producing NADPH oxidase activity of TrxR, while inhibiting the normal function of the enzyme (22). This may play a role in the provoked inflammation seen upon topical application of dinitrohalobenzenes on skin, as discussed elsewhere (2).
Quinones are a widespread group of oxygen-substituted, of-ten biologically active, aromatic compounds. They are found as naturally occurring substances, as synthetic drugs for clinical use, or as components of environmental pollutants. The antitumor, cytotoxic, and antiparasitic activities of clinically used quinones are mainly believed to stem from effects due to redox cycling of their free radical or hydroquinone states, derived from reduction by flavoenzymes, or from covalent modifications of DNA and other cellular nucleophiles (see Refs. [23][24][25]. Reduction of quinones by pyridine nucleotide-disulfide reductases related to TrxR, i.e. glutathione reductase, lipoamide dehydrogenase, and trypanothione reductase, has been studied in significant detail (26 -28). The interactions with quinones may also concomitantly inhibit the normal reactions of these enzymes (29). Previously, studies of mammalian TrxR regarding interactions with quinones include the demonstrated reduction of ubiquinone-10, 2-methyl-1,4-naphthoquinone, and alloxan (14,30). Inactivation of reduced TrxR by antitumor aziridinylbenzoquinones and anthracyclines has also been reported (31). However, no systematic analysis of the reactions of quinones with mammalian TrxR has yet been performed. Here we examined the interactions of recombinant rat TrxR (32) with a number of structurally diverse quinones, including aziridinylbenzoquinones DZQ, RH1, MeDZQ, BZQ (25), anthracycline daunorubicin ( Fig. 1), and a number of model compounds for both partially and fully substituted quinones. We show that significant quinone reduction seems to occur not at the FAD but at the selenolthiol motif, which is unique in comparison to the quinone reduction by non-selenoprotein pyridine nucleotide-disulfide oxidoreductases. Furthermore, we show that cer- tain specific quinone compounds, e.g. juglone, may bypass this selenolthiol motif yet be efficiently reduced by the enzyme. Finally, we show that the four-electron reduced form of TrxR participates in the quinone reduction, and we have determined the previously unknown redox potential of the enzyme.

MATERIALS AND METHODS
Enzymes and Proteins-Recombinant rat TrxR1 was essentially prepared as described (32), with slight methodological improvements to be reported elsewhere. The enzyme was pure as judged by Coomassiestained SDS-PAGE and had a specific activity of 22 units/mg in the model DTNB assay (9). The enzyme concentration was determined from the absorbance of FAD, ⑀ 463 ϭ 11.3 mM Ϫ1 cm Ϫ1 (8). Chlamydomonas reinhardtii thioredoxin was prepared as described (33), and its concentration was determined using ⑀ 280 ϭ 10.9 mM Ϫ1 cm Ϫ1 (34).
Experimental Procedures-All experiments were carried out in 0.1 M potassium phosphate buffer (pH 7.0), containing 1 mM EDTA, at 25°C. Rapid reaction studies were performed using a DX.17MV stopped-flow spectrophotometer (Applied Photophysics) under aerobic conditions. Steady-state reaction rates were monitored spectrophotometrically, using a Hitachi-557 spectrophotometer. Typically, 10 -100 M NADPH was used, and the rate of DTNB reduction (0.15-1.5 mM) by TrxR was monitored following the increase in absorbance at 412 nm (⌬⑀ 412 ϭ 13.6 mM Ϫ1 cm Ϫ1 ), considering that one molecule of DTNB (and NADPH) produces two TNB anions (9). Steady-state NADPH-quinone reductase activity of TrxR was monitored following the rate of NADPH oxidation (⌬⑀ 340 ϭ 6.2 mM Ϫ1 cm Ϫ1 ) using 100 M NADPH. In separate experiments, 50 M cytochrome c was added into the reaction mixture, and its quinone-mediated reduction was monitored at 550 nm (⌬⑀ 550 ϭ 20 mM Ϫ1 cm Ϫ1 ). The reaction rate was corrected for the background (direct) reduction of cytochrome c by TrxR. In separate experiments, this was found to be 0.33 mol of cytochrome c reduced per s per mol TrxR subunit under the utilized assay conditions. When C. reinhardtii Trx (10 -80 M) was the electron acceptor, reaction rates were monitored by following NADPH oxidation. In the reverse reaction of TrxR, reduction of NADP ϩ (50 -500 M) was monitored at 340 nm in the presence of Trx (10 -80 M) and 8 mM DTT as reducing agent (38). This activity was corrected for the background reaction (direct reduction of NADP ϩ by TrxR in the presence of 8 mM DTT but in the absence of Trx), which was 0.2 mol NADP ϩ /s/mol subunit. The kinetic parameters of reactions, i.e. the catalytic constant (k cat ) and the bimolecular rate constant (k cat /K m ), correspond to the reciprocal intercepts and slopes of the Lineweaver-Burk plots, where v is reaction rate, [E] and [S] are the enzyme and substrate concentrations, respectively. The k cat corresponds to molecules of NADPH oxidized (or NADP ϩ reduced) by enzyme-active site/s, i.e. the dimeric TrxR holoenzyme has twice the activity.
The TrxR inhibition experiments were performed in two ways. In studies of reversible inhibition, substrates and inhibitor were introduced into the spectrophotometer cell first, at which point reaction was started by TrxR addition. The inhibition constants were calculated from Cleland plots, plotting the dependence of reciprocal k cat /K m values at variable substrate versus inhibitor concentration (K i(slopes) ), or the dependence of plot intercept with y axis versus inhibitor concentration (K i(intercepts) ). In the irreversible inhibition experiments, TrxR (0.6 -6.0 M) was incubated in the presence of NADPH (200 M) and inhibitor at 25°C. After incubation for the indicated period, an aliquot of the reaction mixture was introduced into a spectrophotometer cell (with a 100 -200-fold factor of dilution), and the rate of DTNB reduction was monitored in the presence of 50 M NADPH and 1.5 mM DTNB. Alternatively, quinone reductase activity of TrxR was monitored by following the oxidation rate of 50 M NADPH in the presence of quinone.

Reactions of TrxR with Disulfide
Substrates-Before analysis of quinone reduction catalyzed by the recombinant mammalian TrxR preparation used throughout this study, we probed its reduction of the two model disulfide substrates DTNB and Trx. As reported before (21), at fixed NADPH concentrations (6 -60 M) and varied DTNB (0.15-1.5 mM), the Lineweaver-Burk plots showed a series of parallel lines, thus pointing to a "ping-pong" scheme (data not shown). The same ping-pong patterns were observed using C. reinhardtii Trx instead of DTNB as substrate (oxidant). The catalytic constants (k cat ) and the steady-state bimolecular rate constants of oxidative and reductive half-reactions (k cat /K m ) as determined here are given in Table I. The k cat values of both reactions were marginally lower than those catalyzed by native purified mammalian TrxR using DTNB (4000 min Ϫ1 for dimeric native TrxR, i.e. 33 s Ϫ1 per active site) or mammalian (3300 min Ϫ1 ; 27.5 s Ϫ1 ) and E. coli (3000 min Ϫ1 ; 25 s Ϫ1 ) Trx, as summarized from the literature (9). It can be noted that C. reinhardtii Trx with K m ϭ 19 M was a slightly more efficient oxidant for the recombinant enzyme than E. coli Trx with K m ϭ 35 M for native mammalian TrxR (9), which in the latter case gives k cat /K m ϭ 0.714 M Ϫ1 s Ϫ1 to be compared with 1.06 M Ϫ1 s Ϫ1 for C. reinhardtii Trx found here (Table I). We also assessed the reverse reaction of TrxR, i.e. reduction of NADP ϩ by C. reinhardtii reduced Trx utilizing 8 mM DTT as reducing agent. This reaction also followed ping-pong kinetics (data not shown) with a k cat close to that of the forward reaction (see Table I).
By using the data presented in Table I, the redox potential of TrxR can be calculated. According to the Haldane relationship, the ratio of the bimolecular rate constants of forward and reverse reactions gives the equilibrium constant of the reaction (K), which in turn is related to the difference in the standard redox potential of the reactants (⌬E 0 7 ϭ 29.5 mV ϫ log K for a two-electron transfer). This relation, based on the rates of enzyme reactions with NADP(H) (E 0 7 ϭ Ϫ0.320 V), has been used for calculating E 0 7 of yeast (39) and P. falciparum glutathione reductases (40). For Reaction 1, the ratio of k cat /K m for NADPH oxidation and NADP ϩ reduction (Table I)  Alternatively, for Reaction 2, the ratio of k cat /K m for reduced and oxidized Trx (Table I) Trx-(SH) 2 ϩ TrxR OX 7 Trx-S 2 ϩ TrxR red REACTION 2 gives K ϭ 2.04 Ϯ 0.2. Because the E 0 7 of C. reinhardtii Trx has been determined to be Ϫ0.290 Ϯ 0.01 V (41), this would give We next analyzed the effect of NADP ϩ on the NADPH-dependent reduction of DTNB by TrxR. At fixed concentrations of DTNB and varied NADPH, the reaction product NADP ϩ acted as a mixed inhibitor with respect to NADPH, increasing both slopes and intercepts in Lineweaver-Burk plots ( Fig. 2A). The same mixed inhibition pattern was observed using fixed NADPH and varied DTNB concentrations ( Fig. 2B). At [NADPH] ϭ 150 M, the K i of NADP ϩ determined according to the slopes of Lineweaver-Burk plots (K i(slopes) ) was equal to 270 Ϯ 40 M, whereas the K i value determined according to the plot intercepts with the y axis (K i(intercepts) ) was equal to 1200 Ϯ 200 M. This inhibition of TrxR by NADP ϩ may possibly play a physiological role in cases of decreased intracellular NADPH/ NADP ϩ ratios, e.g. in hepatocytes upon ethanol intoxication (42).
Quinone Reduction by TrxR-Pyridine nucleotide-disulfide reductases like lipoamide dehydrogenase, glutathione reductase, and trypanothione reductase, as well as low M r -type TrxR of Arabidopsis thaliana, perform mixed single-and two-electron reduction of quinones with reactivity typically increasing with an increase in the single-electron reduction potential (E 1 7 ) of the quinone substrate (26 -28, 38). The E 1 7 values of the quinones examined in the present study are given in Table II. In reactions with mammalian TrxR, we found that 9,10phenanthrene quinone and 5-hydroxy-1,4-naphthoquinone (juglone or walnut toxin), which have low redox potentials (compounds 6 and 8, Table II), led to excess NADPH oxidation over quinone reduction, whereas 1,4-benzoquinone, which has a high redox potential (compound 1, Table II), oxidized more or less a stoichiometric amount of NADPH (Fig. 3A). This showed that juglone and 9,10-phenanthrene quinone participated in redox cycling with TrxR. The reduction of these quinones could be coupled to reduction of cytochrome c, added in separate experiments. The juglone-and phenanthrene quinone-mediated cytochrome c reduction rates by TrxR were 2-and 1.5-fold higher than the NADPH oxidation rates, respectively, and both were decreased by 20 -25% upon addition of 30 g/ml superoxide dismutase. This strongly suggested that the reduction of quinones by TrxR was accompanied by an aerobic redox cycling with formation of superoxide. Such reaction could, however, be consistent with both single-and two-electron reduction mechanisms, because the hydroquinone forms of these quinones may either reduce cytochrome c directly (with transient formation of semiquinones), or may rapidly autoxidize with formation of superoxide that in turn reduces cytochrome c (43,44). Quantitatively, the percentage of single-electron flux may be estimated using 1,4-benzoquinone as substrate, because the hydroquinone form of this quinone cannot reduce cytochrome c at pH Յ 7.2, whereas the benzosemiquinone derivative rapidly reduces cytochrome c (k ϭ 1 M Ϫ1 s Ϫ1 ) (45). Making use of this property, we found that the percentage of single-electron flux in the TrxR-catalyzed reduction, expressed as the ratio of the doubled rate of cytochrome c reduction to NADPH oxidation rate at the expense of 1,4-benzoquinone (44), was 6.5 Ϯ 1%. Hence, TrxR reduces 1,4-benzoquinone mainly with a twoelectron reduction mechanism.
One should note that the rate of TrxR-catalyzed NADPH oxidation by 1,4-benzoquinone sharply decreased after 15-20 s (Fig. 3A), and the same feature was observed in reactions with other partially substituted benzoquinones (data not shown). In contrast, reactions with juglone ( Fig. 3A) or other partially substituted naphthoquinones (data not shown) proceeded at the same or even an increased rate with time. We hypothesized that covalent modification of reduced TrxR by partially substituted quinones, known as potent alkylating agents targeting thiol groups and other nucleophiles (24), may over time affect the activity when the enzyme becomes exposed to quinones. Indeed, incubation of reduced TrxR with 50 M 1,4-naphthoquinone or its 5-hydroxy-and 5,8-dihydroxy-derivatives even for as little as 15 s resulted in a 50 -60% decrease of the DTNB reduction rate. Addition of 50 M 1,4-benzoquinone or 2,3dichloro-1,4-naphthoquinone decreased the activity by 80 -85%. Superoxide dismutase and catalase (30 g/ml) could not protect TrxR from this inactivation. In contrast, activity of reduced TrxR was unchanged after incubation with 50 M of the fully substituted 9,10-phenanthrene quinone for at least 3-4 min. Due to the rapid modification of reduced TrxR by partially substituted quinones, it became important to assess whether TrxR was able to reduce those compounds before modification occurred. We therefore monitored the quinone-mediated cytochrome c reduction using stopped-flow spectrophotometry, which revealed that both juglone and 9,10-phenanthrene quinone reduction began immediately after mixing, demonstrating that both compounds were substrates of the unmodified enzyme (Fig. 3B). The kinetic parameters of juglone reduction determined for the 0 -5-s time interval were k cat ϭ 4.5 Ϯ 0.2 s Ϫ1 and k cat /K m ϭ 1.6 Ϯ 0.1 M Ϫ1 s Ϫ1 , which was close to the same parameters determined under steady-state conditions (Table II). These results indicated that reduced TrxR could reduce partially substituted quinones before becoming modified and, notably, that the modified enzyme retained the ability to reduce quinones in solution (Fig. 3A). This property resembles the reactivity of mammalian TrxR modified with dinitrohalobenzenes (22), and it may be of significance for the mechanisms of cytotoxicity of quinones, as is further discussed below.
In Table II, the k cat and k cat /K m values of the different quinones analyzed here for reduction by mammalian TrxR are given. It can be noted that several quinones were highly efficient substrates for the enzyme. It should also be noted that although the log k cat /K m of the quinone reduction somewhat increased with an increase in quinone E 1 7 (Fig. 4), the correlation was poor (r 2 ϭ 0.5631) and much less evident than seen in reduction of these quinones by other pyridine-nucleotide disulfide reductases (26 -28, 38).
To probe further the details of the catalysis, we next attempted to identify the redox state(s) of TrxR responsible for quinone reduction. Four-electron reduced (EH 4 ) TrxR has the typical spectrum of the FAD-thiolate charge-transfer complex with ⌬⑀ 540 ϳ 2.8 mM Ϫ1 cm Ϫ1 (8), whereas the two-electron reduced state (EH 2 ) has about 50% of that absorbance. Evidently, in the EH 2 state of TrxR, two electrons are shared between the N-terminal catalytic disulfide and the C-terminal motif having the Cys-Sec couple, which therefore decreases the proportion of enzyme molecules having the FAD-thiolate charge transfer at any given moment. We obtained the EH 2 spectrum by addition of 1 eq of NADPH under aerobic conditions, whereas the EH 4 state spectrum could be obtained by use of an NADPH regeneration system such as 5.0 mM glucose 6-phosphate and 5 g/ml glucose-6-phosphate dehydrogenase, or addition of 3-5 equivalents of NADPH (data not shown). In the presence of the NADPH-regenerating system under aerobic conditions, however, we failed to observe complete six-electron reduction of TrxR that would be characterized by a disappearance of the 460 -540-nm absorbance (8). By having determined the spectra of the EH 2 and EH 4 species, we subsequently  Table I). c Percent of k cat /K m using C. reinhardtii Trx as oxidant (see Table I). d E 1 7 calculated assuming that aziridine and methyl groups may similarly influence the E 1 7 value, and that the substitution of a methyl group by hydroxymethyl may increase the E 1 7 by 0.05 V (47). e J. Butler, personal communication.

FIG. 4. Correlation between quinone redox potential and reduction by TrxR.
The correlation between k cat /K m for quinone reduction by TrxR and the quinone single-electron reduction potential (E 1 7 ) is shown in this plot, with values and numbering of compounds as given in Table II. probed the turnover of TrxR upon addition of NADPH and quinone by monitoring the 540-nm absorbance in stopped-flow experiments. Because the solubility of 9,10-phenanthrene quinone, the most efficient quinone oxidant in terms of k cat (Table  II) is limited (50 M) and other fully substituted quinones oxidize TrxR at low rates (Table II), we used juglone in these experiments. The mixing of TrxR with a 5-10-fold excess of NADPH resulted in a rapid rise in the absorbance at 540 nm, which was completed in 20 ms, reaching an amplitude that corresponded to formation of EH 4 (curve 1, Fig. 5A). This signal was stable for at least 200 s due to the low inherent NADPH oxidase activity of TrxR (Ͻ0.005 s Ϫ1 ). In a separate experiment, juglone was added to the syringe containing NADPH. Simultaneous addition of juglone and NADPH to TrxR resulted in a rapid 540-nm absorbance rise closely reaching the level of EH 4 , whereupon the fall in absorbance suggested the presence of enzyme species cycling between the EH 4 and EH 2 states (0.038 Ն ⌬A 540 Ն 0.025, curves 2 and 3, Fig. 5A) until final oxidation to the fully oxidized state occurred upon NADPH exhaustion. Analyzing the final part of the kinetic curves (⌬A 540 Ͻ 0.025, Fig. 5A) using single-exponential fit yielded a pseudo first-order rate constant (k obs ), which exhibited a linear dependence on juglone concentration (Fig. 5B). The calculated second-order rate constant, 0.012 Ϯ 0.002 M Ϫ1 s Ϫ1 , was 2 orders of magnitude lower than the k cat /K m of juglone in the steady-state reaction (Table II). This strongly suggests that complete oxidation into EH ox species is an unlikely event in reactions of TrxR with juglone in the presence of NADPH and, furthermore, that during reduction of juglone TrxR seems to cycle between the EH 4 and EH 2 species, in analogy to its reduction of disulfide substrates.
The TrxR reaction with the most active fully substituted quinone substrate, phenanthrene quinone (Table II), followed a typical ping pong pattern (data not shown) with a k cat /K m for NADPH of 2.8 Ϯ 0.2 M Ϫ1 s Ϫ1 , which was close to the k cat /K m value of NADPH for DTNB and Trx reduction (Table I). At [NADPH] ϭ 150 M, NADP ϩ acted as a mixed inhibitor with regard to phenanthrene quinone, increasing both the slopes and intercepts in Lineweaver-Burk plots (Fig. 6A) with K i(slopes) ϭ 300 Ϯ 30 M and K i(intercepts) ϭ 1100 Ϯ 200 M; this was close to the corresponding parameters obtained for DTNB reduction (see above). This finding further strengthened the notion that mammalian TrxR can reduce quinones with a catalytic mechanism analogous to its reduction of disulfide substrates.
It has been reported that anthracyclines and aziridinylbenzoquinones are competitive inhibitors with regard to DTNB in steady-state reduction by TrxR, and that incubation of reduced TrxR in their presence may lead to covalent binding and irreversible inactivation (31). In view of the rapid TrxR inactivation by partially substituted quinones (see above), we therefore performed more detailed inhibition studies of TrxR using fully substituted quinones, including the anthracycline compound daunorubicin. Phenanthrene quinone was found to inhibit DTNB reduction, with kinetics compatible with a competitive inhibitor having K i ϭ 6.3 Ϯ 1.0 M (Fig. 6B). In contrast, the anticancer quinone compounds tetramethyl-1,4-benzoquinone, MeDzQ, RH1, BZQ, and daunorubicin ( Fig. 1) were surprisingly weak as competitive inhibitors, displaying K i Ն400 M in the DTNB assay. After 15 and 30 min incubation of TrxR in the presence of 200 M NADPH and 50 M quinone, the TrxR activity had decreased by 50 Ϯ 5 and 70 Ϯ 7% (daunorubicin), 39 Ϯ 4 and 48 Ϯ 5% (MeDZQ), 25 Ϯ 3 and 33 Ϯ 4% (RH1), and 58 Ϯ 5 and 72 Ϯ 5% (BZQ), respectively, using 1.5 mM DTNB as electron acceptor. We also found that a 30-min incubation of TrxR with only NADPH in the absence of quinones resulted in a loss of 26 Ϯ 2% of activity. These activity losses could not be prevented by addition of catalase and superoxide dismutase (30 g/ml).
TrxR Inactivation by Gold Thioglucose and BCNU-Reduced mammalian TrxR is readily inactivated by close to stoichiometric concentrations of gold compounds such as gold thioglucose and auranofin, which presumably form an irreversible covalent bond with the Sec residue (21). BCNU, which alkylates thiol groups and presumably Sec, also inhibits the enzyme but at millimolar concentrations (8). In agreement with earlier studies, we found that incubation of reduced TrxR (0.6 M) with a slight excess of gold thioglucose (1.0 M) led to a rapid and almost complete loss of activity with regard to reduction of DTNB. We also found that 9,10-phenanthrene quinone reduction was inhibited with the same pattern as that of DTNB. After a 20-min incubation of reduced TrxR with 1.0 M Authioglucose, the k cat and k cat /K m of phenanthrene quinone had decreased to 0.25 Ϯ 0.04 s Ϫ1 and 0.0056 Ϯ 0.0006 M Ϫ1 s Ϫ1 , respectively. The reduction of 2-methyl-1,4-naphthoquinone was also significantly decreased upon preincubation with gold thioglucose (k cat ϭ 0.18 Ϯ 0.02 s Ϫ1 and k cat /K m ϭ 0.0014 Ϯ 0.0001 M Ϫ1 s Ϫ1 ). Surprisingly, the reduction of some other quinones was much less affected. This was clearly demonstrated in the reduction of juglone by TrxR after gold thioglucose inhibition (k cat ϭ 3.5 Ϯ 0.3 s Ϫ1 , k cat /K m ϭ 2.5 Ϯ 0.2 M Ϫ1 s Ϫ1 ; kinetic parameters close to those obtained with non-inhibited TrxR, as given in Table II). Fig. 7A clearly shows how the reduction by TrxR of DTNB and phenanthrene quinone was inhibited by gold thioglucose, whereas the rate of juglone re-  2 and 3). B, dependence of the pseudo first-order rate constants (k obs ) of TrxR oxidation at ⌬A 540 Ͻ0.025 (see A) on juglone concentration. NADPH concentration in the assay was either 80 (circles) or 160 M (squares). duction was only marginally affected. Both DTNB and juglone reduction by TrxR were, however, inhibited by BCNU, albeit somewhat less in the case of juglone (Fig. 7B). DISCUSSION The results of this work show that quinones in essence can turn mammalian TrxR into a superoxide-producing NADPH oxidase. We found that some quinones were highly efficient substrates for TrxR and could function as competitive inhibitors with respect to other substrates of the enzyme, whereas other quinones were weak substrates and/or inhibitors. Furthermore, separate classes of quinones reacted with TrxR according to different patterns, with their reduction being either inhibited or nearly unaffected by covalent modification of the selenolthiol motif.
Like other related pyridine-nucleotide disulfide oxidoreductases, TrxR could perform mixed single-and two-electron reduction of quinones. The single-electron flux as determined here, 6.5%, was intermediate between the corresponding values for glutathione reductase (3.6% (27)) and trypanothione reductase (40% (26)). However, quinones were significantly more efficient substrates (oxidants) for TrxR than for the other enzymes, because k cat /K m values (Table II) were about 1 order of magnitude higher than in the corresponding reactions catalyzed by glutathione reductase or trypanothione reductase (26,27). Notably, some quinones analyzed here were as efficient as thioredoxin or DTNB as substrates for TrxR (cf . Tables I  and II).
We find unequivocal evidence that the EH 4 state of TrxR, which performs the reduction of disulfide substrates (3,5,6), is also responsible for the major part of quinone reduction. (i) The transient 540 nm absorbance data (Fig. 5A) demonstrates TrxR cycling between the EH 4 and EH 2 states in juglone reduction. Only after NADPH exhaustion did the oxidation of EH 2 to E ox take place, and it occurred with a rate constant 2 orders of magnitude lower than the juglone k cat /K m value in steadystate reduction (Fig. 5B and Table II). (ii) The NADPH k cat /K m for the reduction of phenanthrene quinone was close to that seen in the reduction of both DTNB and Trx. (iii) NADP ϩ inhibited phenanthrene quinone reduction in a similar manner and with similar K i values as it inhibited DTNB reduction (Figs. 2B and 6A). Thus, based on these findings, we feel secure to conclude that the same redox state of TrxR was responsible for the major part of reduction of DTNB, Trx, as well as of quinone compounds. In this respect, TrxR resembles glutathione reductase and trypanothione reductase, whose EH 2 redox states reduce both their natural disulfide substrates and quinones (26,27). In contrast to such behavior, quinones are reduced by the super-reduced (EH 4 ) states of both mammalian and Mycobacterium tuberculosis lipoamide dehydrogenases (28,46), which cannot reduce lipoamide. One should note, however, that the reaction of quinones with glutathione reductase, trypanothione reductase, and lipoamide dehydrogenase mainly involves the reduced flavin and not the active site dithiol (26,27), as is true for the EH 4 state of lipoamide dehydrogenase (28,46). Quinones may in general preferably oxidize FADH 2 instead of a dithiol motif due to the high potential of single-electron oxidation of thiols, which is usually Ն1.0 V unless affected by neighboring residues (47). Naturally, we had expected that quinones would have accepted electrons mainly from the equilibrium form of the EH 4 state of TrxR containing the FADH 2 . Unexpectedly, however, TrxR treatment with gold thioglucose almost completely suppressed reduction of phenanthrene quinone as well as of DTNB (Fig. 7A). Even more surprising was the finding that juglone reduction was little affected under the same conditions, whereas both DTNB and juglone reductions by TrxR were inhibited by BCNU, albeit somewhat less in the case of juglone. Collectively, these results demonstrate that juglone is strikingly more efficient and also qualitatively different from Trx, DTNB, or 9,10-phenanthrene quinone in receiving electrons from redox-active groups of TrxR other than the selenolthiol motif. Based on these findings, we propose that certain quinones (e.g. phenanthrene quinone) are reduced mainly or solely by the easily accessible C-terminal selenolthiol motif of mammalian TrxR (7), whereas other quinones (e.g. juglone) are reduced in addition, or possibly solely, by the N-terminal active site dithiol motif in complex with the enzyme-bound FAD. These divergent properties of different quinones should probably involve different accessibility of the individual quinones to the FAD. Our proposed mechanism for quinone reduction by mammalian TrxR is shown in Fig. 8 and is explained further in the figure legend. The parallel participation of FADH 2 and the reduced selenolthiol in quinone reduction by TrxR should explain the poorly expressed dependence of k cat /K m on the E 1 7 of quinones (Fig. 4), which clearly contrasts with the parabolic plots of log k cat /K m versus E 1 7 characteristic for the FADH 2 -mediated quinone reduction by lipoamide dehydrogenase, glutathione reductase, as well as trypanothione reductase (26 -28).
Our data on the mixed inhibition exerted by NADP ϩ taken together with the high rate of the reverse reaction of TrxR may provide valuable information on the thermodynamic properties of human TrxR. To our knowledge, this is the first report showing that forward and reverse reactions of TrxR may proceed at similar rates. Because NADP(H) and Trx (or DTNB) interact with spatially separated domains of TrxR (7), the enzyme shares the "hybrid ping-pong" mechanism seen in glutathione reductase (48,49). In accordance with such a mechanism, one would have expected a competitive inhibition of NADP ϩ toward NADPH and an uncompetitive inhibition toward disulfide substrates (48), which contradicts the findings shown in Fig. 2, A and B. In contrast, the reverse reaction of glutathione reductase, i.e. the reduction of NADP ϩ by GSH, is only Յ10% of the forward reaction rate (39). The mixed type NADP ϩ inhibition was probably observed due to the fast TrxR-  1 in this figure) is reduced by NADPH to FADH 2 in species 2. The electrons may then be transferred to the N-terminal disulfide to form a flavin-thiolate charge transfer complex (species 3) and subsequently to the C-terminal selenenylsulfide of the other subunit to form a selenolthiol motif (species 4). Collectively, enzyme species 2-4 form the two-electron reduced EH 2 state of TrxR. A second molecule of NADPH may subsequently reduce the enzyme to the EH 4 state (species 5 and 6). The selenolthiol motif reduces typical substrates of the enzyme, such as Trx and DTNB, thereby forming a selenenylsulfide and converting TrxR from the EH 4 to the EH 2 state. The selenolthiol in species 4 -6 can be covalently modified by electrophilic compounds (X, which could also be a quinone derivative) to form species 7-9, respectively. This yields an inactivated TrxR unable to reduce disulfide substrates. We showed here that quinones (Q) may be reduced by either the selenolthiol or the FADH 2 (or the charge transfer complex) of the different equilibrium forms of EH 4 , the ratios of which depend on the properties of the individual quinone. Some quinones may also be reduced by the NADPH-reduced inactivated enzyme to form species 7; this reaction was particularly fast in the case of juglone and very slow for phenanthrene quinone. Reduction of quinones by EH 2 to form EH ox (conversion of species 2 to species 1) was found to be very slow. Hydroquinones (QH 2 ) are shown as the main products in this scheme, as we found that only a minor part (6.5%) of the quinone reduction by mammalian TrxR proceeded with one-electron transfer mechanisms. This proposed scheme for reactions of quinones with mammalian TrxR is based both upon the results presented herein and on the previously characterized reactions of mammalian TrxR with natural substrates (5,8) and dinitrohalobenzenes (22). See text for further details. catalyzed reoxidation by NADP ϩ (Fig. 2, A and B) resulting from competition between NADP ϩ and disulfide oxidant for the same redox state of enzyme in the reoxidation reaction (39,48,50). The redox potential of the EH 2 /EH 4 couple of TrxR determined here to Ϫ0.294 V was clearly more negative than the redox potential of the E ox /EH 2 couple of yeast glutathione reductase, which is Ϫ0.255 V (39), and slightly more negative than the E ox /EH 2 couple of pig heart lipoamide dehydrogenase, which is Ϫ0.280 V (51).
How important are the reactions of quinones with mammalian TrxR as a mechanism explaining their biological effects? Naturally, the answer to that question should be complex and dependent on quinone species as well as TrxR expression and function in a particular cell. Nonetheless we would like to emphasize some features that may be of importance. Because quinones, as substrates for TrxR, may be reduced in both oneand two-electron redox cycling reactions producing superoxide, this may lead to increased oxidative stress. This may be especially true for quinones with relatively stable semiquinone states and thereby hydroquinone forms that rapidly autoxidize, e.g. hydroxy-1,4-naphtoquinones and phenanthrene quinone (24,43,44). However, quinones with an unstable semiquinone state may also act as antioxidants when regenerated by TrxR, as in reduction of cytosolic ubiquinone that yields ubiquinol with antioxidant properties (14). Whether the redox cycling of a quinone compound with TrxR leads to antioxidant effects or an increased oxidative stress may therefore not be immediately concluded. However, a second effect of quinone interference with the antioxidant functions of human TrxR would derive from their different degrees of inhibition of reduction of other disulfide substrates, notably Trx and following that also the peroxiredoxins. This could contribute to an oxidative stress as part of the cytotoxic effects displayed by many quinone compounds. Importantly, we showed here that partially substituted quinones may rapidly modify reduced TrxR, causing deviations from the expected steady-state course of the NADPH: quinone reductase reactions and thereby suppressing the disulfide reduction capacity. In view of the rapid modification of the selenolthiol motif by alkylating agents such as 1,3-dinitro-4-chlorobenzene (22), this C-terminal motif is the likely candidate for modification by quinones as well. Further studies of this possible mechanism for quinone cytotoxicity should certainly be carried out at a cellular level.
We also need to discuss the possible importance of TrxR inhibition and inactivation by fully substituted quinones, although this type of inhibition cannot directly involve irreversible covalent modification. The efficient competitive inhibition of TrxR by phenanthrene quinone with regard to DTNB (Fig.  6B) should most probably be due to efficient function of the quinone as an alternative oxidant (Table II), thereby diverting electron flux from DTNB. That interpretation is strengthened by the fact that the K i value of phenanthrene quinone was close to its K m value in NADPH oxidation. Such inhibition might be important in the cytotoxicity of the phenanthrene quinone compound, which is a component of exhaust gases and an important environmental pollutant (24). In relation to such inhibition, we find it important to note that we found the anticancer aziridinylbenzoquinones MeDZQ, RH1, and BZQ (23,25) as well as daunorubicin (Fig. 1) to be almost inactive as direct competitive inhibitors, in contrast to the efficient inhibition by these analogues of purified mammalian TrxR reported previously (31). This discrepancy, also observed by others (10), suggests that alternative mechanisms than direct TrxR inhibition may be of importance for the anticancer effects of these quinones (52). We did note a time-dependent inactivation of reduced TrxR by aziridinylbenzoquinones and daunorubicin, but also this property was less efficient than what was reported earlier (31). Because these quinones are unable to react with cysteine or selenocysteine directly, the inactivation should have been due to alkylating reactions of their reduced products, the quinomethide of daunorubicin or aziridinyl-substituted hydroquinones (23,25). These may be formed during the TrxR-catalyzed reduction. Thus, daunorubicin treatment may with time inhibit TrxR through formation of other metabolites, but daunorubicin does not directly inhibit this enzyme.
To conclude, we have here characterized the reactions of quinones with mammalian TrxR and found unique features in these reactions not shared by other related pyridine nucleotidedisulfide oxidoreductases. These reactions involve the selenocysteine residue of TrxR and should profoundly interfere with the important physiological functions that TrxR plays in mammalian cells. The data presented here could therefore provide a molecular basis for further studies aiming at elucidating the possible underlying reactions mediating biological effects of quinone compounds.