Human Placenta Thioredoxin Reductase

Human thioredoxin reductase is a pyridine nucleotide-disulfide oxidoreductase closely related to glutathione reductase but differing from the latter in having a Cys-SeCys (selenocysteine) sequence as an additional redox center. Because selenoproteins cannot be expressed yet in heterologous systems, we optimized the purification of the protein from placenta with respect to final yield (1–2 mg from one placenta), specific activity (42 units/mg), and selenium content (0.94 ± 0.03 mol/mol subunit). The steady state kinetics showed that the enzyme operates by a ping-pong mechanism; the value of k cat was 3330 ± 882 min−1, and the K m values were 18 μm for NADPH and 25 μm forEscherichia coli thioredoxin. The activation energy of the reaction was found to be 53.2 kJ/mol, which allows comparisons of the steady state data with previous pre-steady state measurements. In its physiological, NADPH-reduced form, the enzyme is strongly inhibited by organic gold compounds that are widely used in the treatment of rheumatoid arthritis; for auranofin, the K i was 4 nm when measured in the presence of 50 μmthioredoxin. At 1000-fold higher concentrations, that is at micromolar levels, the drugs also inhibited human glutathione reductase and the selenoenzyme glutathione peroxidase.

In a previous study (7), we had investigated the reductive half-reaction of the enzyme. In brief, it was shown that the reduction of E ox , the disulfide-containing form of human TrxR, by its substrate NADPH leads to a series of transient enzyme species characterized by charge transfer complexes involving oxidized flavin, reduced flavin, and reoxidized flavin, respectively. The reactions result in a stable TrxR species containing reoxidized flavin, the active site pair Cys-57/Cys-62 as a dithiol, and an additional reduced redox active group, probably the Cys-495/SeCys-496 center. The nascent thiolate of Cys-62 forms a charge transfer complex with the flavin, which has a typical absorbance at 540 nm. Thus, human thioredoxin reductase mechanistically resembles glutathione reductase and is distinct from bacterial TrxR (7,11,12). Employing steady state kinetics, we have now continued investigating the catalytic mechanism of human thioredoxin reductase.
Studies with the gold compound aurothioglucose on human glutathione peroxidase (13) and human iodothyronine deiodinase type 1 (14), as well as preliminary studies on thioredoxin reductase in rat liver cytosol (15), indicate a specific inhibition of selenoenzymes by this drug. We therefore analyzed the susceptibility of isolated human thioredoxin reductase to organic gold compounds. Gold and its derivatives have been used as therapeutics in the history of mankind for ages. Most preparations and indications described were based on mystic principles and are obsolete today (16). In rheumatoid arthritis, however, a serious disease affecting 1-2% of the world's population, organic gold compounds like auranofin and aurothioglucose are still first choice therapeutics. As discussed below, the results presented here strongly suggest that gold compounds exert at least some of their pharmacologic effects by inhibiting the selenoenzyme thioredoxin reductase.

Materials
Frozen placentas were kindly provided by Dr. J. Wacker (Department of Obstetrics and Gynecology, Heidelberg University). Purification of human thioredoxin reductase from placenta is delineated below. Recombinant Escherichia coli TrxS 2 with an ⑀ 280 nm of 13.7 mM Ϫ1 cm Ϫ1 (17,18) and human glutathione reductase with an ⑀ 463 nm of 11.3 mM Ϫ1 cm Ϫ1 (19) were produced and isolated as described. Human glutathione peroxidase was purchased from Sigma.
Auranofin was obtained from ICN, and aurothioglucose, thioglucose, gold(III)chloride, and British Anti-Lewisite (BAL) ( were from Sigma. Precast gels (12% polyacrylamide) and the protein dye assay were from Bio-Rad, and molecular weight standards were from Amersham Pharmacia Biotech. All reagents were of the highest available purity.

Enzyme Assays
All assays were conducted at 25°C in a total assay volume of 1 ml. Thioredoxin Reductase Activity-For the purification procedure and the inhibition studies, the DTNB reduction assay (4) proved to be sufficiently specific. The enzyme was added to an assay mixture of 100 mM potassium phosphate, 2 mM EDTA, pH 7.4, and 3 mM DTNB (using a 100 mM stock solution in Me 2 SO); after initiating the reaction with the addition of NADPH (200 M final concentration), the increase in absorbance at 412 nm was monitored. 1 enzyme unit is defined as the NADPH-dependent production of 2 mol of 2-nitro-5-thiobenzoate (⑀ 412 nm 13.6 mM Ϫ1 cm Ϫ1 ) per min.
In steady state kinetic studies, the assay mixture contained 100 mM potassium phosphate, 2 mM EDTA, pH 7.4, and five different concentrations both of TrxS 2 (range 5-45 M) and of NADPH (range 5-100 M; ⑀ 340 nm 6.22 mM Ϫ1 cm Ϫ1 ). The reaction was started with thioredoxin reductase (final concentration 4 nM TrxR subunits), and the decrease in absorbance at 340 nm was monitored during the linear phase. 1 enzyme unit is defined as the consumption of 1 mol of NADPH per min. Each combination of [NADPH] and [Trx] was repeated six times, and the mean values were used for computing the kinetic constants.
Glutathione Reductase Activity-Glutathione reductase activity was measured in an assay mixture consisting of 47 mM potassium phosphate, 1 mM EDTA, 200 mM KCl, pH 6.9, and 100 M NADPH; after the addition of glutathione disulfide (1 mM final concentration), the consumption of NADPH was monitored as the decrease in absorbance at 340 nm.
Glutathione Peroxidase Activity-Glutathione peroxidase activity was determined in a GR-coupled assay according to Beutler (20). The assay mixture (100 mM Tris-HCl, 1 mM EDTA, pH 8.0, 4 units/ml glutathione reductase, 2 mM reduced glutathione, 100 M NADPH, and glutathione peroxidase) was equilibrated for 10 min; then the substrate t-butylhydroperoxide (1 mM final concentration) was added and the consumption of NADPH was monitored. We increased the activity of glutathione reductase in the assay from 1 unit/ml (20) to 4 units/ml to assure that this ancillary enzyme was not rate-limiting in the presence of organic gold compounds.
Protein Assay-Protein was determined using the Bio-Rad dye assay with bovine serum albumin as a standard.

Thioredoxin Reductase Purification
Because of the potential risk of infection, laboratory biosafety regulations (3,21) were strictly obeyed in the first steps including acetone precipitation. Unless otherwise stated, all procedures were carried out at 4°C. The TE buffer used throughout the preparation consisted of 50 mM Tris-HCl, 1 mM EDTA, pH 7.6.
Chloroform-1-butanol Extraction- (22)A frozen placenta of approximately 500 g was cut with a stainless steel saw into slices (about 1 ϫ 3 ϫ 10 cm). The slices were cleaned mechanically from debris with a cover slide, weighed out, and transferred to plastic bags. Per 1 g of placenta, 0.6 ml of extraction solution (10 M FAD, 40 M phenylmethylsulfonyl fluoride in TE buffer) was added, and the tissue was thawed by placing the bags into a 40°C water bath. Subsequently, the content of the bags was homogenized in 250-g portions in a Waring blender. Each portion was stabilized with 20 l of 100 mM phenylmethylsulfonyl fluoride. Immediately before treatment with chloroform/1-butanol, the homogenate was titrated to pH 8.3 using 5 M NH 4 OH. The chloroform/ 1-butanol mixture (1:2.5, v/v; Ϫ20°C; 120 l/gram placenta) was added under vigorous stirring. The brownish suspension was rehomogenized in the Waring blender, left for 1 h, and then centrifuged for 90 min at 8000 ϫ g. The supernatant was set aside while the precipitate was taken up in extraction solution (0.4 ml/gram placenta), homogenized, and centrifuged as above. The supernatants were combined, filtered through glass wool (Riedel de Hä en) and adjusted to pH 8.3 using 5 M NH 4 OH. This solution was the chloroform-butanol extract (Table I).
Acetone Precipitation-Per 1 ml of chloroform-butanol extract, 0.85 ml of acetone was slowly added under stirring. The solution was left for 1 h and then centrifuged for 15 min at 3500 ϫ g. The pellet was taken up in a small volume of TE buffer having a final volume of approximately 100 ml. This sample was dialyzed exhaustively against 2-fold diluted TE buffer and centrifuged (30 min, 25000 ϫ g). The supernatant was set aside while the pellet was resuspended in TE buffer, mixed carefully, and centrifuged as described above. The combined superna-tants were filtered through glass wool and adjusted to pH 8.3 using 5 M NH 4 OH. The resulting solution was referred to as the acetone-treated fraction.
DEAE-52 Cellulose Chromatography-The acetone-treated fraction was applied to a (3.2 ϫ 50 cm) DEAE-52 cellulose column (Whatman), which had been equilibrated with TE buffer before and was operated at room temperature. After washing the column with 1000 ml of TE buffer followed by 500 ml of 50 mM NaCl in TE buffer, thioredoxin reductase activity was eluted with 90 mM NaCl in TE buffer. In this step, the enzyme comigrated with a deep red protein. The pool of active fractions was concentrated and washed with TE buffer in a Centriprep 30 (Amicon). The resulting solution was diluted 2-fold with TE buffer, and the pH was adjusted to 7.6 using 100 mM HCl. This fraction was called the DEAE-cellulose eluate.
2Ј,5Ј-ADP-Sepharose 4B Affinity Chromatography-The above fraction was applied to a 30-ml (1.5 ϫ 17 cm) 2Ј,5Ј-ADP-Sepharose 4B column (Amersham Pharmacia Biotech) in a jacketed chromatography tube. The tube was cooled to 6 Ϯ 1°C, the exact temperature being crucial for the purification success. The column was consecutively washed with 60 ml of TE buffer, 30 ml of 100 mM KCl in TE buffer, 20 ml of 200 mM KCl in TE, 30 ml of 100 mM KCl in TE, 60 ml of 2-fold diluted TE, 60 ml of 500 M NADH in TE, 60 ml of TE, 60 ml of 100 M NADP ϩ in TE, and 30 ml of 300 M NADP ϩ in TE. Finally, TrxR activity was eluted with 750 M NADP ϩ in TE buffer, concentrated, and washed with the buffer in a Centriprep 30. This solution, the 2Ј,5Ј-ADP-Sepharose eluate, contained (on the basis of absorption spectra (7), specific activity, and SDS-polyacrylamide gel electrophoresis analysis) homogeneous (more than 95% pure) thioredoxin reductase.
Sephadex G-200 Gel Filtration-To remove trace impurities, the above fraction may be applied to a Sephadex G-200 column (Amersham Pharmacia Biotech, 1 ϫ 100 cm) equilibrated with TE buffer. Spectroscopically pure fractions were pooled, concentrated, and referred to as the Sephadex G-200 eluate.

Inhibitor Studies
10 mM stock solutions of aurothioglucose and thioglucose in assay buffer and of auranofin in Me 2 SO were prepared immediately prior to use and stored in dark bottles. Dilutions were made in assay buffer. One min after adding the inhibitor to the assay mixture of a given enzyme, the reaction was started with the appropriate substrate. Thioglucose served as a control and had no inhibitory effect on the enzymes within the tested inhibitor concentration range.
For inhibition studies on human TrxR, the DTNB reduction assay was used. To verify the effects of an organic gold compound on TrxR under more physiological conditions, we determined the K i for auranofin with TrxS 2 as a substrate. Using two different TrxS 2 concentrations (50 and 75 M together with 100 M NADPH and 1.7 nM TrxR), the inhibitory effects of 5-50 nM auranofin in the assay were observed over 20 min, and initial rates were determined.
To determine the influence of NADPH on TrxR and glutathione reductase inhibition by the gold compounds, 50-l samples containing approximately 600 nM enzyme subunits each were preincubated with 1 M inhibitor in the presence and absence of 200 M NADPH for 20 min at room temperature. Aliquots were taken, diluted, and assayed for residual activity. Control experiments showed that free inhibitor present in diluted aliquots had no effect when added to assays of uninhibited enzyme. To test the reversibility of TrxR inhibition by a gold chelating agent, inhibited samples were exposed to 1 mM BAL for 5 and 20 min before assaying activity. In a complementary protection experiment, 50-l samples of TrxR (700 nM subunits) containing 200 M NADPH and 1 mM BAL were prepared. Then inhibitor was added (1 M), and after 20 min the residual activity was determined.

RESULTS
Enzyme Purification-Because it is not yet possible to produce recombinant mammalian selenoenzymes in heterologous systems, we have optimized the purification of native human thioredoxin reductase. For this purpose, placenta proved to be the organ of choice (5). The first purification steps involve organic solvents (Table I). Apart from their antiseptic effect, these solvents denature the bulk of NADP(H)-dependent enzymes (22), which greatly enhances the efficiency of affinity chromatography used in a later purification step. In comparison with the original report of Oblong et al. (5), we were able to improve the isolation procedure with respect to speed, final yield (1-2 mg of TrxR instead of 0.3 mg from one placenta) and specific activity (Table I). Using atomic absorption spectroscopy, the selenium content of the isolated enzyme was found to be 0.94 Ϯ 0.03 mol/mol of subunit of 55.2 kDa. Approximatly 520 pmol of TrxR subunit exhibits 1 unit of enzymatic activity in the DTNB reduction assay.
Kinetic Studies-The long-standing problem of preparing the substrate thioredoxin in sufficient amounts has led in the past to the use of DTNB as the disulfide substrate or to the use of coupled disulfide systems in TrxR assays (23). Comparisons of these alternative systems with thioredoxin-based assays have been conducted whenever it was crucial (23,24). Luthman and Holmgren found that rat liver, calf liver, and E. coli thioredoxin all gave a k cat of approximately 3000 min Ϫ1 with rat liver thioredoxin reductase; using DTNB as the acceptor substrate, k cat was 4000 min Ϫ1 (24). In the present study on human TrxR, we have used E. coli thioredoxin (TrxS 2 ) and found, as detailed below, a turnover number of 3300 min Ϫ1 .
The kinetic parameters were obtained from secondary plots of the steady state kinetic data shown in Fig. 1. As in the case of the closely related enzyme glutathione reductase (7,12,25,26) the results for TrxR are consistent with a ping-pong mechanism. We determined a k cat of 3330 Ϯ 882 min Ϫ1 and K mvalues of 18 M for NADPH and of 25 M for E. coli thioredoxin. When using the DTNB reduction assay, the K m for NADPH was found to be only 6 M; the K m value of 0.4 mM for DTNB as reported by Oblong et al. (5) was confirmed. Thus, our results obtained with the DTNB reduction assay compare well with a previous study on human placenta TrxR (5) and with the data for TrxR from rat liver (24) and mouse tumor cells (3,27).
It should be noted, however, that at NADPH concentrations above 20 M, substrate inhibition becomes appreciable in both assay systems. This is apparent in Fig. 1 where the data points depart from the straight lines at higher concentrations of NADPH.
The effect of temperature on the rate of the reaction was studied both with TrxS 2 and with DTNB as a substrate; the data for E. coli TrxS 2 are given in Fig. 2. In the range between 5 and 40°C, the activation energy was found to be 53.2 kJ/mol. Above 40°C, human TrxR becomes unstable and is completely inactive at 60°C. Glutathione reductase and other disulfide reductases are known to be fully active at this temperature (6,28,29).
Using the Arrhenius plot (Fig. 2), the turnover number at 4°C is 650 Ϯ 17 min Ϫ1 or 10.8 Ϯ 2.9 s Ϫ1 . As discussed below this value is consistent with the slowest step in the reductive half-reaction determined at 4°C in an earlier study of the pre-steady state kinetics (7).
Inhibitor Studies-The results of the studies with aurothioglucose and auranofin in the assay mixture are shown in Fig. 3. Gold-free thioglucose did not inhibit the enzymes in the con-centration range used for the gold compounds. Glutathione reductase and glutathione peroxidase were by at least three orders of magnitude less susceptible to the organic gold compounds than thioredoxin reductase. IC 50 values for the inhibition of all three enzymes by the different inhibitors are given in Table II. It should be noted that glutathione reductase was almost unaffected by aurothioglucose, whereas auranofin had inhibitory effects although only in the upper micromolar range. For glutathione peroxidase, the situation was reversed; that is, auranofin but not aurothioglucose was found to be an inhibitor. Qualitatively speaking, our data agree well with the effects of

FIG. 2. Effect of temperature on the rate of the reaction catalyzed by human thioredoxin reductase presented as an Arrhenius plot.
Duplicate assays were carried out at each temperature. The NADPH concentration was 200 M, and the thioredoxin concentration was 100 M. The data were fitted by linear regression, and the activation energy was determined from the slope using the Arrhenius equation.

TABLE I
Synopsis of the thioredoxin reductase purification procedure The selenium content of the Sephadex G-200 eluate was determined to be 0.94 mol/mol subunit using atomic absorption spectroscopy. In a number of preparations, the specific activity did not increase in the last step. The value of 35.0 units/mg, corresponding to 2 units/nmol subunit, is not exceeded by any known mammalian thioredoxin reductase. 0.2 mg/ml bovine serum albumin in the assay mixture increased the final specific activity to 42 units/mg. Bovine serum albumin, however, was not used in assay mixtures since it interfered with the studies on tight-binding enzyme inhibitors. aurothioglucose on the three enzymes in unfractionated rat cytosol. In this study, TrxR was inhibited 50% by 100-fold less gold thioglucose than needed for 50% inhibition of glutathione peroxidase, whereas GR activity was not affected by submillimolar concentrations of the gold compound (15). Fig. 4 shows the different sensitivity of isolated TrxR and glutathione reductase in their oxidized (E ox ) and NADPH-reduced states toward gold compounds. Whereas in the E ox form the enzymes remained unaffected, NADPH-reduced TrxR was almost completely inhibited (Fig. 4A). The addition of 1 mM BAL reversed the TrxR-inhibition, leading to 50% of initial activity after 5 min and 100% after 20 min. Simultaneous incubation with inhibitor and BAL completely prevented the effect of the gold compounds on TrxR. Glutathione reductase, even in the NADPH-reduced form, was hardly affected by preincubation with organic gold compounds (Fig. 4B). Only inorganic Au(III)Cl 3 led to an inactivation of reduced glutathione reductase; in contrast to the situation with human TrxR, this inhibition was only partially reversible by BAL (less than 15% after 1 h).
In an attempt to describe the inhibition of TrxR by auranofin more precisely, we extended this study to include inhibition steady state kinetics for tight binding enzyme inhibitors that compete with a substrate (30). Initially we tried to analyze progress curves as defined by Morrison and Walsh (30). However, assays with different concentrations of both enzyme and inhibitor did not generate the unique curvature required for such a biphasic analysis. Thus, our rate measurements refer to the first minute of an enzyme assay that showed no subsequent curvature.
The data in Fig. 5 were fitted satisfactorily to either Equation 1, representing classical competitive inhibition, or to Equation 2, which accounts for the change in concentrations of inhibitor and enzyme as a result of the tight binding in the enzyme inhibitor complex (30).
Fitting the experimental data of Fig [I t ] in Equation 2 was assumed to be the total inhibitor concentration. K i Ј is an apparent quantity (30) that is related to K i as given in Equation (3). Concentration-dependent inhibition of 3 nM human glutathione peroxidase, 1.5 nM human glutathione reductase, and 2 nM human thioredoxin reductase, respectively, is given for standard assay conditions. The inserts show the chemical structure of the inhibitors. accounts for the change in concentrations of these quantities during the assay. We were unable to determine whether there is an additional isomerization complex E⅐I*; the reaction sequence E ϩ I 3 E⅐I 3 E⅐I* is often to be considered for quantifying the effects of reversible inhibitors with K i -values in the submicromolar range (30).
The K i values determined above are approximately 10-fold lower than the IC 50 value listed in Table II, which was determined using DTNB as the disulfide substrate. If we assume that DTNB competes with auranofin as is the case for TrxS 2 (Fig. 5), we can apply Equation (4), taking into account the actual DTNB concentration of 3 mM in the assay and its K m of 0.4 mM.
Thus, there is good agreement between the K i values evaluated with TrxS 2 as the substrate (Fig. 5) and the IC 50 value evaluated from the DTNB-based assay.

DISCUSSION
Isolation of Authentic hTrxR-Difficulty in preparing recombinant hTrxR is a major problem in studying the human thioredoxin system. The published purification protocols for placenta hTrxR result in comparatively low yields and require many time consuming steps (5). The efficient purification procedure presented here provides sufficient amounts of thioredoxin reductase for structural studies. We recently succeeded in crystallizing the isolated enzyme by using PEG 8000 in Tris buffer of pH 7.4 as a precipitant.
Kinetic Studies-The steady state kinetic data are consistent with a bi-bi-ping-pong mechanism, a result which further underlines the similarities between human thioredoxin reductase and glutathione reductase (7,12,26). Using the Arrhenius diagram (Fig. 2), the apparent turnover number of hTrxR at 4°C is 650 Ϯ 17 min Ϫ1 or 10.8 Ϯ 2.9 s Ϫ1 . This allows the comparison of the steady state kinetic data with the presteadystate rates determined at 4°C for the reduction of hTrxR by its substrate NADPH (7); the low temperature was necessary because the first two phases of this reaction are very fast. A rapid absorbance increase at 540 nm (110 s Ϫ1 ), for instance, reflected the reduction of the active site disulfide Cys-57/Cys-62 as indicated by a charge transfer complex between the nascent thiolate 62 with FAD. The slowest reaction phase was observed as an absorbance decrease at 540 nm, signaling reformation of the active site disulfide; it occurred at a rate of approximately 5 s Ϫ1 , which is comparable with the turnover number of TrxR at 4°C (Fig. 2). On the basis of these data, it is tempting to speculate that k cat of human TrxR is limited at least in part by redox interchange between the active site Cys-57/Cys-62 pair and the Cys-495/SeCys-496 redox center. The notion that the active site dithiol passes the reducing equivalents on to another redox center is supported by the observation that at least two equivalents of reducing agent (carrying four electrons) are FIG. 4. NADPH dependence of TrxR and GR inhibition by different gold compounds. Only the NADPH-reduced form of human thioredoxin reductase (A) is inhibited by 20 min preincubation with 1 M inhibitor, whereas the enzyme in the E ox form remains stable. In contrast, human glutathione reductase (B) is not affected by aurothioglucose or auranofin under these conditions. However, AuCl 3 does also strongly inhibit the NADPHreduced form of hGR, indicating a different modification caused by Au(III) when compared with organic Au(I)-compounds. needed for the complete reduction of the active site disulfide (7).
Gold Compounds as Inhibitors-Aurothioglucose and auranofin were found to be potent inhibitors of human thioredoxin reductase (Table II, Equations 2 and 3). These organic gold compounds are widely used in the treatment of rheumatoid arthritis. The disease is considered to be an autoimmune condition initiated by various agents, the Epstein-Barr virus being the prime candidate (31). Lymphocytes infected with EBV or other viruses have been shown to secrete thioredoxin (32) which, together with our data, suggests the possibility that the thioredoxin redox system plays a prominent role in autoimmune processes. This notion is supported by the finding that in SjögrenЈs syndrome, another autoimmune disease with joint involvement, secreted thioredoxin levels correlate very well with the expression of EBV material (34). With respect to the inhibition studies on TrxR, it should be emphasized that the activity of thioredoxin as a cytokine depends on its reduced dithiol state (32,33).
As shown in Fig. 4, NADPH-reduced human thioredoxin reductase is highly sensitive to gold compounds, whereas the oxidized form of the enzyme, E ox , is not affected. Since the K m value (ϭ K diss value) for NADPH under quasi in situ conditions is 3-5 times lower than the cytosolic NADPH concentration (7), the reduced gold-sensitive forms of TrxR are likely to be predominant in situ. Not only chemical but possibly also steric reasons may account for the different sensitivities of oxidized and reduced TrxR. For the three human iodothyronine deiodinases, all of them being selenoenzymes, it has been shown that the type 1 enzyme is strongly inhibited by aurothioglucose (K i, app ϳ5 nM), whereas the type 2 and type 3 enzymes are 1000fold less sensitive (38).
Several lines of reasoning indicate, that it is indeed the gold content of the compounds which leads to thioredoxin reductase inhibition. First, the thioglucose moiety of aurothioglucose (and auranofin) is not an inhibitor in the concentration range used in our study. Second, the gold-chelating agent BAL is able both to prevent and to reverse the inhibition of TrxR caused by three different compounds that have only the gold moiety in common. Furthermore, selenols exhibit a higher tendency to bind heavy metal ions than thiols do (16). It is therefore tempting to speculate that the C-terminal redox-active Cys-495/SeCys-496 center of thioredoxin reductase is the target of the inhibitors (39,40). This view is supported by the finding that the structurally and mechanistically closely related but selenium-free enzyme glutathione reductase (1,7,9) is far less sensitive to the inhibition by auranofin and aurothioglucose ( Fig. 3 and Ref. 15). Inorganic AuCl 3 (which is not in therapeutic use because of its high toxicity) caused a BAL-resistant inhibition of NADPHreduced GR. This indicates that, at least for GR, the mode of inhibition by inorganic Au(III)compounds is different from drugs like auranofin and aurothioglucose that contain an Au(I)-moiety.
Virtually complete TrxR inhibition in vitro can be achieved with concentrations far below the clinically used plasma levels (e.g. 20 M for auranofin). Under in vivo conditions, the drug is likely to be bound unspecifically to compounds such as glutathione and other thiols which decreases the actual concentration of free inhibitor. This interpretation is supported by the observation that a 10-fold higher aurothioglucose concentration is needed for 50% TrxR inhibition in rat liver cytosol (15) when compared with our data on the isolated human enzyme.
Auranofin has also been successfully tested as an antineoplastic agent which efficiently inhibits DNA synthesis (Ref. 35 and references therein). Because intracellular thioredoxin is a reducing substrate of ribonucleotide reductase, a key enzyme in DNA synthesis, our results can offer a molecular explanation also for this effect. Other studies dealing with pharmacological effects of carmustine (3,7,37), with the abundancy of TrxR activity in tumor cells (2), with NK lysin as a substrate of TrxR (36), or with reduced thioredoxin as a proliferation-promoting cytokine (32,33) indicate as well that inhibition of thioredoxin reductase may be a rational approach to the treatment of certain malignancies.