Advertisement

Mechanistic Studies on a Novel, Highly Potent Gold-Phosphole Inhibitor of Human Glutathione Reductase*

Open AccessPublished:March 24, 2005DOI:https://doi.org/10.1074/jbc.M412519200
      The homodimeric flavoprotein glutathione reductase (GR) is a central player of cellular redox metabolism, connecting NADPH to the large pool of redox-active thiols. In this work, the inhibition of human GR by a novel gold-phosphole inhibitor (GoPI) has been studied in vitro. Two modes of inhibition are observed, reversible inhibition that is competitive with GSSG followed by irreversible inhibition. When ∼1 nm GoPI is incubated with NADPH-reduced GR (1.4 nm) the enzyme becomes 50% inhibited. This appears to be the most potent stable inhibitor of human GR to date. Analyzing the monophasic oxidative half-reaction of reduced GR with GSSG at pH 6.9 revealed a Kd(app) for GSSG of 63 μm, and a k(obs)max of 106 s-1 at 4 °C. The reversible inhibition by the gold-phosphole complex [{1-phenyl-2,5-di(2-pyridyl)phosphole}AuCl] involves formation of a complex at the GSSG-binding site of GR (Kd = 0.46 μm) followed by nucleophilic attack of an active site cysteine residue that leads to covalent modification and complete inactivation of the enzyme. Data from titration spectra, molecular modeling, stopped-flow, and steady-state kinetics support this theory. In addition, covalent binding of the inhibitor to human GR was demonstrated by mass spectrometry. The extraordinary properties of the compound and its derivatives might be exploited for cell biological studies or medical applications, e.g. as an anti-tumor or antiparasitic drug. Preliminary experiments with glioblastoma cells cultured in vitro indicate an anti-proliferative effect of the inhibitor in the lower micromolar range.
      The antioxidant enzyme glutathione reductase (GR)
      The abbreviations used are: GR, glutathione reductase; (h)GR, (human) glutathione reductase; Eox, enzyme in oxidized state; EH2, enzyme in a two-electron reduced state; CTC charge-transfer complex; I0.5, inhibitor concentration for 50% inhibition; GoPI, gold-phosphole inhibitor; DMF, N,N-dimethylformamide; PPh3AuCl, chloro(triphenylphosphine)gold(I); SELDI-TOF MS, surface-enhanced laser desorption/ionization time-of-flight mass spectrometry; BCNU, 1,3-bis(2-chloroethyl)-1-nitrosourea; GSNO, S-nitrosoglutathione.
      1The abbreviations used are: GR, glutathione reductase; (h)GR, (human) glutathione reductase; Eox, enzyme in oxidized state; EH2, enzyme in a two-electron reduced state; CTC charge-transfer complex; I0.5, inhibitor concentration for 50% inhibition; GoPI, gold-phosphole inhibitor; DMF, N,N-dimethylformamide; PPh3AuCl, chloro(triphenylphosphine)gold(I); SELDI-TOF MS, surface-enhanced laser desorption/ionization time-of-flight mass spectrometry; BCNU, 1,3-bis(2-chloroethyl)-1-nitrosourea; GSNO, S-nitrosoglutathione.
      plays an essential role in the cellular redox metabolism of most organisms, catalyzing the regeneration of reduced GSH (
      • Becker K.
      • Rahlfs S.
      • Nickel C.
      • Schirmer R.H.
      ). For this reason the mechanism and structure of GRs from several organisms including man (
      • Nordhoff A.
      • Bücheler U.S.
      • Werner D.
      • Schirmer R.H.
      ,
      • Krauth-Siegel R.L.
      • Arscott L.D.
      • Schönleben-Janas A.
      • Schirmer R.H.
      • Williams Jr., C.H.
      ,
      • Pai E.F.
      • Schulz G.E.
      ,
      • Karplus P.A.
      • Schulz G.E.
      ,
      • Savvides S.N.
      • Scheiwein M.
      • Böhme C.C.
      • Arteel G.E.
      • Karplus P.A.
      • Becker K.
      • Schirmer R.H.
      ), Escherichia coli (
      • Rietveld P.
      • Arscott L.D.
      • Berry A.
      • Scrutton N.S.
      • Deonarain M.P.
      • Perham R.N.
      • Williams Jr., C.H.
      ), and the malaria-causing parasite Plasmodium falciparum (
      • Savvides S.N.
      • Scheiwein M.
      • Böhme C.C.
      • Arteel G.E.
      • Karplus P.A.
      • Becker K.
      • Schirmer R.H.
      ,
      • Böhme C.C.
      • Arscott L.D.
      • Becker K.
      • Schirmer R.H.
      • Williams Jr., C.H.
      ,
      • Sarma G.N.
      • Savvides S.N.
      • Becker K.
      • Schirmer M.
      • Schirmer R.H.
      • Karplus P.A.
      ), have been extensively studied. GSH is present in most eukaryotic cells at millimolar concentrations. It is involved in redox homeostasis, formation of deoxyribonucleotides, and detoxification of peroxides, 2-oxoaldehydes, and xenobiotics. Cells that are exposed to oxidative stress as well as rapidly proliferating cells particularly depend on the regeneration of GSH. Thus, selective, highly potent inhibitors of GR (
      • Becker K.
      • Herold-Mende C.
      • Park J.J.
      • Lowe G.
      • Schirmer R.H.
      ,
      • Davioud-Charvet E.
      • Delarue S.
      • Biot C.
      • Schwobel B.
      • Boehme C.C.
      • Mussigbrodt A.
      • Maes L.
      • Sergheraert C.
      • Grellier P.
      • Schirmer R.H.
      • Becker K.
      ,
      • Becker K.
      • Gui M.
      • Schirmer R.H.
      ,
      • Becker K.
      • Schirmer R.H.
      ,
      • Karplus P.A.
      • Krauth-Siegel R.L.
      • Schirmer R.H.
      • Schulz G.E.
      ) are promising lead compounds for the development of novel anti-tumor or antiparasitic drugs (
      • Becker K.
      • Rahlfs S.
      • Nickel C.
      • Schirmer R.H.
      ,
      • Böhme C.C.
      • Arscott L.D.
      • Becker K.
      • Schirmer R.H.
      • Williams Jr., C.H.
      ,
      • Sarma G.N.
      • Savvides S.N.
      • Becker K.
      • Schirmer M.
      • Schirmer R.H.
      • Karplus P.A.
      ,
      • Becker K.
      • Herold-Mende C.
      • Park J.J.
      • Lowe G.
      • Schirmer R.H.
      ,
      • Davioud-Charvet E.
      • Delarue S.
      • Biot C.
      • Schwobel B.
      • Boehme C.C.
      • Mussigbrodt A.
      • Maes L.
      • Sergheraert C.
      • Grellier P.
      • Schirmer R.H.
      • Becker K.
      ,
      • Becker K.
      • Gui M.
      • Schirmer R.H.
      ,
      • Becker K.
      • Schirmer R.H.
      ,
      • Karplus P.A.
      • Krauth-Siegel R.L.
      • Schirmer R.H.
      • Schulz G.E.
      ,
      • Irmler A.
      • Bechthold E.
      • Davioud-Charvet E.
      • Hofman V.
      • Réau R.
      • Gromer S.
      • Schirmer R.H.
      • Becker K.
      ).
      As a member of the pyridine nucleotide-disulfide oxidoreductase family of homodimeric flavoenzymes, each subunit of GR contains one FAD, two substrate binding sites, and, in oxidized GR(Eox), an intramolecular disulfide. During the reductive half-reaction electrons are transferred rapidly from NADPH on the re side of the flavin to the cystine disulfide on the si side of the flavin. In the oxidative half-reaction the final electron acceptor GSSG is reduced to 2 GSH, regenerating the disulfide at the active site of GR. Under physiological, reducing conditions GR(EH2) presumably occurs as the principal species of the enzyme in vivo (
      • Böhme C.C.
      • Arscott L.D.
      • Becker K.
      • Schirmer R.H.
      • Williams Jr., C.H.
      ). The flow of electrons during a catalytic cycle can be monitored by specific spectral changes, using the flavin as an inherent redox indicator. The reductive half-reaction is characterized by a loss of absorbance at 460 nm coupled to a slight blue shift of the local maximum. In addition, the absorbance at 540 nm increases because of the formation of a thiolate-FAD charge-transfer complex (CTC), formed between the flavin and the proximal or charge-transfer thiolate of Cys63 in hGR. Cys58 in hGR provides the interchange or distal thiol that forms a mixed disulfide with glutathione as an intermediate. Reoxidation of the enzyme, which is coupled to the formation of GSH during the oxidative half-reaction, reverses the spectral changes (
      • Krauth-Siegel R.L.
      • Arscott L.D.
      • Schönleben-Janas A.
      • Schirmer R.H.
      • Williams Jr., C.H.
      ,
      • Rietveld P.
      • Arscott L.D.
      • Berry A.
      • Scrutton N.S.
      • Deonarain M.P.
      • Perham R.N.
      • Williams Jr., C.H.
      ,
      • Böhme C.C.
      • Arscott L.D.
      • Becker K.
      • Schirmer R.H.
      • Williams Jr., C.H.
      ).
      Phospholes are phosphacyclopentadienes that have very limited aromatic character (for an example, see the central unit of the compound shown in Fig. 1) and a nucleophilic phosphorus atom, which makes them good reagents for chemical modifications, for example, with thiols (
      • Hay C.
      • Hissler M.
      • Fischmeister C.
      • Rault-Berthelot J.
      • Toupet L.
      • Nyulaszi L.
      • Réau R.
      ). Thus, they are interesting candidates for inhibitor studies on disulfide reductases such as thioredoxin reductase and GR because they exhibit inhibitor concentrations for 50% inhibition (I0.5) in the lower micromolar range (
      • Irmler A.
      • Bechthold E.
      • Davioud-Charvet E.
      • Hofman V.
      • Réau R.
      • Gromer S.
      • Schirmer R.H.
      • Becker K.
      ). Some palladium-containing phospholes even possess I0.5 values in the nanomolar range for thioredoxin reductase (
      • Irmler A.
      • Bechthold E.
      • Davioud-Charvet E.
      • Hofman V.
      • Réau R.
      • Gromer S.
      • Schirmer R.H.
      • Becker K.
      ). Several organic gold compounds, which have high affinity for sulfur- and selenium-containing ligands (
      • Shaw III, C.F.
      ), e.g. proteins with activated cysteine or selenocysteine residue(s), are also potential agents for medical applications.
      Here, we describe a novel, highly potent gold-phosphole inhibitor (GoPI) of hGR. Both the mode of action of the inhibitor on hGR as well as the oxidative half-reaction of hGR during catalysis are studied in detail. We propose a mechanism for the reaction of GoPI with hGR based on studies using mass spectrometry, titrations, stopped-flow kinetics, and steady-state kinetics. The potential of GoPI as a lead compound for the development of novel anti-tumor drugs was confirmed by first experiments with glioblastoma cells cultured in vitro.

      EXPERIMENTAL PROCEDURES

      Materials—Recombinant hGR was produced as described (
      • Nordhoff A.
      • Bücheler U.S.
      • Werner D.
      • Schirmer R.H.
      ). The protein concentration of purified hGR was determined using ϵ463 nm(λmax) = 11.3 mm-1 cm-1 for Eox (FAD-containing subunit, see Ref.
      • Krauth-Siegel R.L.
      • Arscott L.D.
      • Schönleben-Janas A.
      • Schirmer R.H.
      • Williams Jr., C.H.
      ). The ratio A274 nm/A463 nm of purified enzyme was 6.9. The specific enzymatic activity for hGR was measured as described (
      • Krauth-Siegel R.L.
      • Arscott L.D.
      • Schönleben-Janas A.
      • Schirmer R.H.
      • Williams Jr., C.H.
      ). All titrations, steady-state, and rapid reaction experiments were performed in GR buffer containing 47 mm potassium phosphate, 200 mm KCl, and 1 mm EDTA, pH 6.9, as described previously (
      • Krauth-Siegel R.L.
      • Arscott L.D.
      • Schönleben-Janas A.
      • Schirmer R.H.
      • Williams Jr., C.H.
      ,
      • Rietveld P.
      • Arscott L.D.
      • Berry A.
      • Scrutton N.S.
      • Deonarain M.P.
      • Perham R.N.
      • Williams Jr., C.H.
      ,
      • Böhme C.C.
      • Arscott L.D.
      • Becker K.
      • Schirmer R.H.
      • Williams Jr., C.H.
      ). NADPH and GSSG were dissolved in GR buffer, and their concentrations were determined spectrophotometrically or in enzymatic assays, respectively. Solutions of GoPI (see below) in N,N-dimethylformamide (DMF) were freshly prepared before each experiment.
      GoPI Synthesis—The synthesis of [1-phenyl-2,5-di(2-pyridyl)-phosphole}AuCl] was carried out in analogy to the synthesis of [{1-phenyl-2,5-di(2-thienyl)phosphole}AuCl] (
      • Fave C.
      • Cho T.Y.
      • Hissler M.
      • Chen C.W.
      • Luh T.Y.
      • Wu C.C.
      • Réau R.
      ,
      • Fave C.
      • Hissler M.
      • Kárpáti T.
      • Rault-Berthelot J.
      • Deborde V.
      • Toupet L.
      • Nyulászi L.
      • Réau R.
      ) and [{1-phenyl-2,5-di(2-pyridyl)phosphole}W(CO)5] (
      • Hay C.
      • Le Vilain D.
      • Deborde V.
      • Toupet L.
      • Réau R.
      ). The reactant 1-phenyl-2,5-di(2-pyridyl)phosphole was synthesized as described previously (
      • Hay C.
      • Hissler M.
      • Fischmeister C.
      • Rault-Berthelot J.
      • Toupet L.
      • Nyulaszi L.
      • Réau R.
      ). Au-Cl(tetrahydrothiophene) (26 mg, 81 μmol) was added to a solution of 10 ml of CH2Cl2 containing 30 mg (81 μmol) of 1-phenyl-2,5-di(2-pyridyl)phosphole. This mixture was stirred for 3 h at room temperature. Then all the volatile materials were removed in vacuo. The precipitate was washed with pentane (4 × 10 ml) and the gold-phosphole complex was obtained as an air-stable green-yellow solid (34 mg, 70% yield). Structure and purity of GoPI (600.4 g/mol) were confirmed using high resolution mass spectrometry (HR-MS), as well as 1H, 13C[1H], and 31P[1H] NMR spectroscopy. Selected spectroscopic data: 13C[1H] NMR (CDCl3; 75.46 MHz): δ22.7 (s, = CCH2CH2), 30.1 (s, = CCH2), 122.8 (s, C5 pyridyl), 124.5 (d, J(P,C) = 5.5 Hz, C3 pyridyl), 129.3 (d, J(P,C) = 12.4 Hz, Cm), 132.2 (s, Cp), 134.5 (d, J(P,C) = 14.0 Hz, Co) 137.2 (s, C4 pyridyl), 149.4 (s, C6 pyridyl), 152.2 (d, J(P,C) = 13.9 Hz, C2 pyridyl), 153.6 (m, Cβ), Cα and Cipso not observed. 31P[1H] NMR (CDCl3, 121.5 MHz): δ +39.9 (s). High resolution mass spectrometry (mNBA, FAB): (m/z) 601.0859 [M + H]+. C24H22PClN2Au was calculated as 601.0875.
      C24H22PCIN2AuCalculated:C47.98H3.52Found:C48.10H3.62
      (Eq. 1)


      UV-visual (CH2Cl2) λmax (nm) ϵ (m-1 cm-1): 271 (7850) and 383 (8300). Emission (CH2Cl2) λem (nm) was 495.
      Steady-state Kinetics and Inhibition Studies—Steady-state kinetics were monitored spectrophotometrically at 25 °C using a Hitachi U-2001 or a Beckmann DU 650 UV-visual spectrophotometer. All experiments were performed in the presence of the same concentration of DMF without inhibitor as a control. The type of inhibition was characterized by analyzing Km values for GSSG in the presence of GoPI or PPh3AuCl, and for NADPH in the presence of GoPI. All reactions were started by the addition of NADPH unless otherwise described.
      To determine the half-maximal inhibition of hGR by GoPI in comparison with other disulfide reductase inhibitors (I0.5), 1.4 or 2.8 nm hGR was incubated for 10 min with 100 μm NADPH and 0.5–10 nm GoPI or 10–100 nm PPh3AuCl, respectively. Then, 100 μm GSSG was added, and the residual activity of hGR was measured. Alternatively, hGR was incubated with GoPI for 10 min in the absence of NADPH, followed by addition of NADPH and GSSG.
      Time-dependent inactivation of hGR by GoPI was analyzed under two sets of conditions: first, as described by Kitz and Wilson (
      • Kitz R.
      • Wilson I.B.
      ), was to preincubate 0.1 μm NADPH-reduced hGR with 1–3-fold excess GoPI at 4 °C. Higher concentrations of GoPI lead to complete and very rapid inactivation of hGR, which prevented measurement. Aliquots of 5 μl were taken after 0.5–10 min and used for measuring the residual activity at 25 °C in a 500-μl assay mixture. Second, 1.7 nm NADPH-reduced hGR at 25 °C was incubated with inhibitor (1.7–5.0 nm). After various intervals (0.5–10 min) of incubation, the enzymatic activity was measured by adding 1 mm GSSG to the mixture and compared with that of an untreated control. Based on this second approach a second-order rate constant of the reaction between enzyme and inhibitor was determined from the equation, d[E]/dt = k[E][I] (
      • Davioud-Charvet E.
      • McLeish M.J.
      • Veine D.M.
      • Giegel D.
      • Arscott L.D.
      • Andricopulo A.D.
      • Becker K.
      • Müller S.
      • Schirmer R.H.
      • Williams Jr., C.H.
      • Kenyon G.L.
      ).
      Titration and Rapid Reaction Experiments—Titrations were carried out at 25 °C and stopped-flow measurements at 4 °C. To produce hGR(EH2) in the absence of NADP(H), NaBH4 was added at a ∼50-fold molar excess over hGR(Eox) (
      • Rietveld P.
      • Arscott L.D.
      • Berry A.
      • Scrutton N.S.
      • Deonarain M.P.
      • Perham R.N.
      • Williams Jr., C.H.
      ,
      • Böhme C.C.
      • Arscott L.D.
      • Becker K.
      • Schirmer R.H.
      • Williams Jr., C.H.
      ). NaBH4 was dissolved in NaOH, pH 10 (where it is quite stable), prior to each experiment. Salt concentrations and pH of the GR buffer do not change significantly after addition of the NaBH4-containing solution (<4% v/v); excess NaBH4 is degraded rapidly at pH 6.9 (
      • Rietveld P.
      • Arscott L.D.
      • Berry A.
      • Scrutton N.S.
      • Deonarain M.P.
      • Perham R.N.
      • Williams Jr., C.H.
      ,
      • Böhme C.C.
      • Arscott L.D.
      • Becker K.
      • Schirmer R.H.
      • Williams Jr., C.H.
      ,
      • Davis R.E.
      • Swain C.G.
      ). Absorption spectra were recorded on a Varian Cary 3 or a Hewlett-Packard 8453 UV-visual spectrophotometer.
      Titration of Oxidized and Reduced hGR with GoPI—Oxidized enzyme was titrated (in duplicate) with 0–10 eq of GoPI under aerobic conditions (initial concentration of enzyme was 9.6 or 9.7 μm, the final concentration of DMF was 1.8 or 1.6%, v/v). For each experiment GR buffer was titrated with inhibitor as a reference. In a control experiment hGR was titrated with DMF without GoPI, and the spectra and specific enzymatic activity were not affected significantly up to a concentration of 2.7% (v/v). NaBH4-reduced hGR(EH2) was titrated with 0–1.1 eq of inhibitor under anaerobic conditions (initial concentration of enzyme was 16.6 μm, final concentration of DMF was 2.7%, v/v). At the end of the titration, the protein solution was washed repeatedly in a Centricon 30 (Millipore) and assayed for specific activity to test the reversibility of the inhibition.
      Reduction of hGR(Eox) after Preincubation with GoPI—Oxidized enzyme (8.2 μm) was incubated aerobically with 10 eq of inhibitor, followed by titration with 0–11.3 eq of NADPH (0–88 μm) under anaerobic conditions. At the end of the titration, the protein solution was washed repeatedly in Centricon 30 to attempt to remove the inhibitor. Alternatively, 16.3 μm hGR(Eox) was incubated with 1.25 eq of GoPI, followed by a reduction with 0–20 eq of NaBH4 under aerobic conditions. GR buffer containing only the inhibitor was treated identically and served as a spectral control.
      Reactivity of GoPI with Dithiothreitol, GSH, and Bovine Serum Albumin—To analyze spectral changes of GoPI in the presence of different thiols, solutions of 15–20 μm inhibitor in GR buffer were incubated for 10 min with 10 eq of GSH or dithiothreitol. In addition, GoPI was added to a solution containing 48 μm (2.4 eq) bovine serum albumin, and incubated for 10 min.
      Rapid Reaction Kinetics—Rapid reaction kinetics of the oxidative half-reaction were measured under anaerobic conditions in the diode array mode in a Hi-Tech SF-61DX2 stopped-flow photometer (Hi-Tech Scientific, Salisbury, Wiltshire, United Kingdom). The reaction was initiated by mixing equal volumes of substrate and NaBH4-reduced enzyme (preincubated with or without GoPI) in the stopped-flow instrument: in a series of control experiments the first syringe was loaded with 28.1 μm hGR(EH2), and the second syringe was loaded with GR buffer containing 0, 30, 60, 120, or 600 μm GSSG. Afterward, 0.9 eq of GoPI was added to hGR(EH2) in the first syringe (0.86% DMF, v/v) and mixed again with 0, 30, 60, 120, or 600 μm GSSG. Finally, GoPI was added to the enzyme-containing solution at an excess over hGR (∼2.4 eq, ∼3% DMF, v/v) and mixed with 600 μm GSSG. The data from kinetic traces at selected wavelengths were fitted to single exponential functions using the software KinetAsyst3 (Hi-Tech Scientific).
      Molecular Modeling—GoPI was manually built with GaussView (Gaussian, Inc.). The geometry of the molecule was subsequently refined by molecular mechanics methods using the UFF force field. GoPI was manually fitted to the active site of hGR (
      • Karplus P.A.
      • Schulz G.E.
      ) with the program O (
      • Jones T.A.
      • Zou J.Y.
      • Cowan S.W.
      • Kjeldgaard M.
      ) and refined with the program CNS (
      • Brünger A.T.
      • Adams P.D.
      • Clore G.M.
      • DeLano W.L.
      • Gros P.
      • Grosse-Kunstleve R.W.
      • Jiang J.S.
      • Kuszewski J.
      • Nilges M.
      • Pannu N.S.
      • Read R.J.
      • Rice L.M.
      • Simonson T.
      • Warren G.L.
      ).
      SELDI-TOF MS—Purified hGR was analyzed on NP20 Protein-Chip® Arrays (normal phase, hydrophilic properties; Ciphergen Biosystems, Inc.) to detect masses of untreated hGR(Eox) compared with enzyme after pre-reduction with NADPH and incubation with GoPI. Each sample was prepared in triplicate. GR (85 μm) was reduced with 50 eq of NADPH and then incubated with 1 or 10 eq of GoPI for 30 min at room temperature. The resulting solution was diluted 1:100 in double-distilled water and 1 μl of each sample (0.85 pmol) was applied directly to NP20 ProteinChip® Array spots, which were prewetted with 1 μl of double-distilled water. The array surface was allowed to dry for 30 min, and was washed twice with 5 μl of double-distilled water to remove residual salt. Saturated sinapinic acid solution (1 μl 50%) dissolved in 50% acetonitrile, 0.5% trifluoroacetic acid was added and air-dried twice. Subsequently, mass analysis of bound hGR was performed using the ProteinChip Reader (PBS IIc, Ciphergen Biosystems, Inc.). The ProteinChip Arrays were analyzed by averaging 100 laser shots collected in the positive ion mode. The accelerating potential was +20 kV and the extraction delay time was set to 1620 ns. Deflector settings were used to filter out peaks with m/z <8000. Calibration was performed with the following proteins: bovine β-lactoglobulin A (18,363 Da), horseradish peroxidase (43,240 Da), bovine serum albumin (66,433 Da), and chicken conalbumin (77,490 Da).
      Proliferation Assays—Cell cultures of glioblastoma cells NCH37, NCH82, and NCH89 were established in our laboratory (
      • Herold-Mende C.
      • Steiner H.H.
      • Andl T.
      • Riede D.
      • Buttler A.
      • Reisser C.
      • Fusenig N.E.
      • Mueller M.M.
      ) and cultured routinely in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum and antibiotics at 37 °C, 5% CO2, and 95% air in a humidified incubator with medium changes twice a week. After reaching confluence, cells were harvested by a brief incubation with a trypsin/EDTA solution (Viralex, PAA, Linz, Austria) and seeded into a fresh 75-cm2 plastic tissue culture flask. Proliferation assays were performed as described earlier (
      • Herold-Mende C.
      • Steiner H.H.
      • Andl T.
      • Riede D.
      • Buttler A.
      • Reisser C.
      • Fusenig N.E.
      • Mueller M.M.
      ) using the BrdU Labeling and Detection Kit III (Roche Diagnostics, Mannheim, Germany). Cells were seeded in 8 replicas in 96-well plates into Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum and antibiotics, each well receiving 7 × 103 cells. After 24 h GoPI was added at different concentrations (1, 5, 10, and 20 μm). 48 h later, 5′-bromo-2′-deoxyuridine was added to the wells at a final concentration of 100 μm. The 5′-bromo-2′-deoxyuridine incorporation assay was processed according to the manufacturer's instructions. The mean value of the absorbance of the control samples containing no GoPI was defined as 100% proliferation rate. All values are means of at least two independent experiments (each comprising 8 replicas).

      RESULTS

      Chemical Properties of GoPI—GoPI is a symmetric aromatic phosphole-gold complex (Fig. 1), which is stable at room temperature. The phosphorus atom behaves as a classical 2-electron donor toward the gold(I) atom (
      • Fave C.
      • Cho T.Y.
      • Hissler M.
      • Chen C.W.
      • Luh T.Y.
      • Wu C.C.
      • Réau R.
      ). The simplicity of the 13C NMR spectra recorded in solution is in favor of a symmetric structure. However, theoretical calculations and modeling of the structure predict a low energy barrier for the pyridyl rings to rotate around the C-C bonds connecting them to the phosphole ring (
      • Hay C.
      • Hissler M.
      • Fischmeister C.
      • Rault-Berthelot J.
      • Toupet L.
      • Nyulaszi L.
      • Réau R.
      ). It is thus very likely that this dynamic process is rapid in solution at room temperature.
      Steady-state Kinetics—Inhibition of hGR(Eox) by GoPI is competitive for GSSG at high concentrations of inhibitor (Fig. 2): all straight lines in the Lineweaver-Burk plot intersect at the ordinate with almost identical Vmax values, and plots of [GSSG]/ν against [GoPI] lead to predominantly parallel lines when using data points for 0.01, 0.5, 1.0, and 2.0 μm GoPI. The Kic is 0.46 μm (for competitive inhibition of pyridine nucleotide-disulfide oxidoreductases, see for example Ref.
      • Becker K.
      • Herold-Mende C.
      • Park J.J.
      • Lowe G.
      • Schirmer R.H.
      ), calculated using the equation Kic = Km[GoPI]/(Km ′ - Km) for 0.5, 1.0, and 2.0 μm GoPI. Km ′ is the apparent Michaelis constant in the presence of inhibitor obtained from the x axis intercepts of the Lineweaver-Burk plots. The Km for GSSG in the absence of inhibitor is 71 μm. Km for NADPH (4.4–5.6 μm) was not significantly changed in the presence of 0.65–2.5 μm inhibitor (data not shown).
      Figure thumbnail gr2
      Fig. 2Steady-state kinetics of the inhibition of hGR by GoPI at 25 °C. A, Lineweaver-Burk plot; 1/[GSSG] is plotted against 1/V in the presence of: 1, 0.0; 2, 0.01; 3, 0.5; 4, 1.0; and 5, 2.0 μm GoPI. B, plot according to Cornish-Bowden (
      • Bisswanger H.
      ); [GSSG]/V is plotted against [GoPI] for different concentrations of GSSG: 1, 0.2; 2, 0.4; and 3, 1.0 mm. The data were fitted to straight lines by linear regression analysis using data points for 0.01, 0.5, 1.0, and 2.0 μm GoPI (see text).
      In contrast to the competition with GSSG that was observed at high inhibitor concentrations, only ∼1 nm GoPI was sufficient to inactivate 50% of 1.4 nm NADPH-reduced hGR, when hGR was preincubated for 10 min with GoPI. Preincubating oxidized hGR with inhibitor in the absence of NADPH for 10 min requires 6.2 nm inhibitor to achieve 50% inactivation.
      Time-dependent Inactivation—GoPI also inhibits hGR irreversibly in a time-dependent process. The inactivation of GoPI was monitored over a period of 10 min using two different preincubation protocols as described under “Experimental Procedures.” In the first approach enzyme activity decreased with pseudo first-order kinetics (kapp = kinact[I]0/(Ki + [I]0), see also Refs.
      • Kitz R.
      • Wilson I.B.
      and
      • Bisswanger H.
      ), and was dependent on the inhibitor concentration (data not shown). The maximal rate constant for inhibition in the presence of excess inhibitor (kinact) was ∼1.2 min-1, and t½ = ln2/kinact for enzyme inhibition was 0.6 min at 4 °C. When preincubating in the absence of NADPH, less than 20% inhibition was observed. In the second approach, preincubation at 25 °C with a 60-fold lower enzyme concentration resulted in a second-order rate constant of ∼3 × 106 m-1 s-1 (Table I). The change in activity of hGR over time was calculated based on the equation: -d[hGR]/dt = ν = k2[GoPI][hGR] (see also Ref.
      • Davioud-Charvet E.
      • McLeish M.J.
      • Veine D.M.
      • Giegel D.
      • Arscott L.D.
      • Andricopulo A.D.
      • Becker K.
      • Müller S.
      • Schirmer R.H.
      • Williams Jr., C.H.
      • Kenyon G.L.
      ).
      Table ICalculated second-order rate constants of the irreversible inhibition of 1.7 nm NADPH-reduced hGR by GoPI, measured in the reaction with GSSG at 25 °C (second approach, see text)
      Timek2
      1.7 nm GoPI2.5 nm GoPI3.0 nm GoPI5.0 nm GoPI
      minm-1 s-1
      0.52.21 × 1064.42 × 1063.92 × 1065.44 × 106
      1.07.0 × 1063.45 × 1063.1 × 1064.1 × 106
      3.02.48 × 1061.34 × 1061.28 × 1061.97 × 106
      5.01.88 × 1066.79 × 1061.2 × 1061.48 × 106
      10.06.34 × 1062.94 × 1067.85 × 1057.9 × 105
      k2 mean (±S.D.)
      3.2 (±2.0) × 106 m-1 s-1. The mean value ± S.D. was calculated from 29 independent assays
      a 3.2 (±2.0) × 106 m-1 s-1. The mean value ± S.D. was calculated from 29 independent assays
      A stoichiometry for irreversibly bound GoPI was estimated from the time-dependent inactivation by linear regression (
      • Bisswanger H.
      ). Plotting enzyme activity versus inhibitor concentration and extrapolating the curve to the abscissa, leads to a stoichiometry of 1–1.5 eq of GoPI per FAD-containing subunit of hGR (data not shown).
      The structurally related, commercially available substance PPh3AuCl was compared with GoPI as an inhibitor of GR. PPh3AuCl shows a competition with GSSG at higher concentrations (0.1–1 μm) but with similar patterns to those of GoPI as indicated by Lineweaver-Burk as well as Dixon plots (data not shown). However, PPh3AuCl is a much weaker inhibitor than GoPI with a 30-fold higher I0.5 (30 nm).
      Titration of hGR(Eox) and hGR(EH2) with GoPI—The stoichiometry and binding properties of the inhibitor to hGR were characterized by UV-visual absorption spectroscopy: hGR(Eox) was titrated with GoPI aerobically (Fig. 3). Alternatively, hGR(EH2), pre-reduced with NaBH4, was titrated with inhibitor under anaerobic conditions (Fig. 4). Spectra of hGR(Eox) and hGR(EH2) without GoPI are shown for comparison (Fig. 5A). As a control, neither the absorbance nor the specific enzymatic activity of hGR were changed significantly after titration with DMF (the solvent for GoPI). In addition, spectra of controls containing 15–20 μm GoPI in the absence of hGR do not change significantly after incubation with alternative thiol-containing compounds such as 10 eq of dithiothreitol, 10 eq of GSH, or 2.4 eq of bovine serum albumin (data not shown), indicating that spectral changes are specific for the interaction of GoPI with hGR.
      Figure thumbnail gr3
      Fig. 3Titration of hGR(Eox) with GoPI under aerobic conditions at 25 °C. A, selected difference spectra of 9.6 μm hGR(Eox) titrated with GoPI after subtraction of reference spectra (GR buffer titrated with GoPI). Amounts of GoPI with respect to hGR were as follows: 1, 0.10; 2, 0.31; 3, 0.52; 4, 0.83; 5, 1.04; and 6, 4.38 eq of GoPI (1–41 μm). Selected UV-visual absorbance spectra of 9.6 μm hGR(Eox) titrated with GoPI without subtraction of reference spectra are shown in the inset: 0.1, 0.52, 1.25, 2.81, 4.38, 5.94, 7.50, and 10.6 eq of GoPI (1–100 μm). B, plot of corrected absorbance differences from A versus equivalents of GoPI, revealing a stoichiometry of ∼1.25 eq of GoPI (11.9 μm) per FAD containing subunit.
      Figure thumbnail gr4
      Fig. 4Difference spectra of NaBH4-reduced hGR titrated with GoPI under anaerobic conditions at 25 °C. A, spectra of 16.6 μm NaBH4-reduced hGR titrated with GoPI after subtraction of reference spectra (GR buffer titrated with GoPI). Concentrations of GoPI were as follows: 1, 0.08; 2, 0.33; 3, 0.43; 4, 0.64; 5, 0.87; and 6, 1.1 eq of GoPI (1–17.4 μm). B, plot of corrected absorbance differences from A versus equivalents of GoPI.
      Figure thumbnail gr5
      Fig. 5Spectra of hGR(Eox) preincubated with or without GoPI after stepwise addition of NADPH or NaBH4. A, control spectra in the absence of GoPI: 1, hGR(Eox); 2, NADPH-reduced hGR(EH2) generated by adding 3.9 eq of NADPH under anaerobic conditions (ϵ540 nm of 4.0 mm-1 cm-1); 3, NaBH4-reduced hGR(EH2) generated by adding aerobically an ∼50-fold molar excess of NaBH4 over hGR(Eox) (ϵ540 nm of 1.9 mm-1 cm-1). B, difference spectra generated in the reduction by NaBH4 of 16.3 μm hGR(Eox) preincubated with 1.25 eq of GoPI at 25 °C under aerobic conditions. A cuvette, containing GR buffer with the same concentrations of GoPI and NaBH4, was used as a reference. In general, addition of NaBH4 did not change the absorbance in the reference cuvette. Theoretical amounts of NaBH4 were as follows: 1, 1.25; 2, 2.5; 3, 5.0; and 4, 12.5 eq of NaBH4 (0–200 μm). Further addition of NaBH4 did not change the spectrum. C, difference spectra generated in the titration of 8.2 μm hGR(Eox) preincubated with 10 eq of GoPI, after stepwise addition of 0–11 eq of NADPH at 25 °C under anaerobic conditions. The absorbance of a reference cuvette, containing GR buffer with the same concentration of GoPI, was subtracted. Amounts of NADPH were (see inset): 1, 0; 2, 0.4; 3, 1.1; 4, 3.8; and 5, 11.3 eq of NADPH (0–88 μm).
      GoPI itself is a strong chromophore with an absorbance maximum at 397 nm; therefore, the titration spectra of hGR(Eox) in Fig. 3 and hGR(EH2) in Fig. 4 with GoPI were corrected using a reference cuvette containing GoPI in GR buffer leading to Fig. 3A and Fig. 4A, respectively. The limited solubility of GoPI leads to turbidity (data not shown) but only above 15 μm, i.e. not in the concentration range of the crucial data. The absorbance of GoPI in GR buffer is also changed by the presence of oxidized or reduced enzyme, leading to complex spectra in these titrations. The difference spectra of hGR(Eox) titrated with inhibitor shows a strong effect at wavelengths around the absorbance maximum of GoPI, with a minimum at 400–410 nm (Fig. 3A). The titration of hGR(EH2) with GoPI shows a strong hyperchromicity at all wavelengths in the corrected spectra with a local maximum at the flavin absorbance maximum around 450 nm (Fig. 4A).
      Corrected absorbance differences at selected wavelengths allow a more detailed analysis of spectral changes. Thus, for oxidized enzyme, a stoichiometry of 1.25 eq of inhibitor per FAD containing subunit was obtained (Fig. 3B). For the titration of hGR(EH2) with GoPI no clear stoichiometry could be assigned (Fig. 4B).
      Reduction of hGR(Eox) after Preincubation with GoPI— hGR(Eox) preincubated with GoPI was reduced with NADPH under anaerobic conditions or with NaBH4 aerobically (Fig. 5). Addition of NaBH4 causes spectral changes (Fig. 5B, curve 4) comparable with the reduction of hGR in the absence of inhibitor (Fig. 5A, curve 3): a decrease of absorbance at 463 and 380 nm is coupled to blue shifts of the maxima, while the absorbance at 540 nm increases, indicating the generation of the thiolate-FAD CTC.
      Titration of hGR(Eox) with NADPH, after preincubation with GoPI for 1 h, causes unusual spectral changes (Fig. 5C): the first equivalent of added NADPH becomes oxidized, as indicated by the lack of an increase in absorbance at 340 nm. Further addition of NADPH leads to an increase of absorbance at 340 nm, showing a build-up of excess NADPH similar to that seen in reductive titrations of hGR in the absence of inhibitor (Fig. 5A). However, the reduction of hGR in the presence of GoPI does not generate the thiolate-FAD CTC at 540 nm. In contrast, the absorbance decreases slightly, and the local maximum at 495 nm in the difference spectrum is shifted to shorter wavelengths.
      Removal of GoPI—The solution mentioned above, containing 8 μm hGR, 10 eq of GoPI, and 88 μm NADPH, was washed repeatedly under aerobic conditions in a Centricon 30. The spectrum of the recovered enzyme was abnormal (showing a strong absorbance at 366 nm presumably because of the absorbance of bound inhibitor, and a shift of the maximum at 463 to 452 nm); the specific activity of hGR was virtually 0. In a similar experiment with NaBH4-reduced hGR, which was incubated with 1 eq of GoPI, six washing cycles led to a recovery of enzyme with a specific activity from 27 to 43%. The activity between cycles 4 and 6 did not increase significantly.
      Rapid Reaction Kinetics of the Oxidative Half-reaction—The oxidation of hGR(EH2) by GSSG was characterized using NaBH4-reduced enzyme; no NADP(H) was present. Fig. 6A shows the rapid decrease of absorbance at 540 nm, indicating the loss of the thiolate-FAD CTC, whereas the increase in absorbance at 463 nm shows the formation of hGR(Eox). Both kinetic traces reveal one distinct phase as described previously (
      • Krauth-Siegel R.L.
      • Arscott L.D.
      • Schönleben-Janas A.
      • Schirmer R.H.
      • Williams Jr., C.H.
      ) (in the experiment shown in the left-hand panel of Fig. 6A, the enzyme was mixed with GSSG in the absence of GoPI; in the experiment shown in the right-hand panel of Fig. 6A, 0.9 eq of GoPI was added to the reduced enzyme before mixing with GSSG). No additional phases can be observed for either experiment, comprising hGR(EH2) preincubated with or without GoPI. The corresponding rate constant k(obs) is dependent on the concentration of GSSG, but surprisingly, it is independent of the presence or absence of GoPI (Table II).
      Figure thumbnail gr6
      Fig. 6Rapid reaction kinetics of the oxidative half-reaction of NaBH4-reduced hGR with GSSG at 4 °C. A, kinetic traces of the reaction with GSSG after preincubation of reduced enzyme in the absence (left) or presence (right) of GoPI. Comparison of the anaerobic reaction of 300 μm GSSG (after mixing) with 14.1 μm hGR (left), and 13.9 μm hGR preincubated with 0.9 eq of GoPI (right). B, double reciprocal plot of the observed first-order rate constant k1(app) () as a function of the concentration of GSSG revealing -1/Kd(app) for the binding of GSSG to hGR(EH2) and a maximal rate for reoxidation 1/k(obs)max.
      Table IIMeasured and calculated parameters of the oxidative half-reaction of NaBH4-reduced hGR with GSSG at 4 °C Reduced enzyme was preincubated prior to mixing with or without 0.9 eq of GoPI. Each reaction was performed in duplicate. The values for kobs were calculated after fitting the kinetic traces at 540 and 463 nm to a single exponential function. kobs determined at 540 nm did not deviate significantly from kobs determined at 463 nm. Thus, kobs of each reaction was averaged from 4 values.
      GSSGkobs
      hGR (14.1 μm) preincubated without GoPI
      kobs
      hGR (13.9 μm) preincubated with 0.9 eq of GoPI
      Δϵ540 nm
      hGR (14.1 μm) preincubated without GoPI
      Δϵ540 nm
      hGR (13.9 μm) preincubated with 0.9 eq of GoPI
      Δϵ540 nm
      hGR (14.1 μm) preincubated without GoPI
      /Δϵ540 nm
      hGR (13.9 μm) preincubated with 0.9 eq of GoPI
      μms-1mm-1 cm-1
      1519.7 (±2.6)20.7 (±2.7)1.370.682.0
      3037.8 (±6.4)31.3 (±4.5)1.360.612.2
      6057.4 (±3.7)54.7 (±6.8)1.330.622.1
      30078.7 (±15)79.2 (±17)1.520.622.5
      k(obs) max 106 (±10) s-11.390.632.2
      a hGR (14.1 μm) preincubated without GoPI
      b hGR (13.9 μm) preincubated with 0.9 eq of GoPI
      The data obtained from rapid reaction experiments can be used under certain conditions to calculate specific rate constants and a dissociation constant based on the procedure described in Ref.
      • Strickland S.
      • Palmer G.
      • Massey V.
      . Thus, an apparent dissociation constant Kd(app) for the binding of GSSG to hGR(EH2) of 63 μm and a maximal rate for reoxidation k(obs)max of 106 (±10) s-1 was determined (Fig. 6B, Table II).
      In contrast to the rate constant determined, the amount of absorbance loss for the thiolate-FAD CTC during the reaction with GSSG differs markedly between hGR(EH2) preincubated without or with GoPI (Figs. 6A and 7). For hGR preincubated with 0.9 eq of GoPI, the loss of absorbance at 540 nm is ∼45% compared with untreated hGR (Table II). The addition of GoPI at a ∼2.4-fold higher concentration than that of hGR(EH2) leads to a complete block of the reaction (data not shown).
      Figure thumbnail gr7
      Fig. 7Spectra recorded during the oxidative half-reaction of NaBH4-reduced hGR with GSSG at 4 °C after preincubation of the reduced enzyme with or without GoPI. Comparison of the anaerobic reaction of 300 μm GSSG (after mixing) with: A, 14.1 μm hGR; and B, 13.9 μm hGR preincubated with 0.9 eq of GoPI. Spectra were recorded after: 1, 1.5 ms; 2, 4.5 ms; 3, 18.5 ms; 4, 28.5 ms; and 5, 2.45 s. Inset, the spectral changes at longer wavelengths have been expanded to facilitate comparison of the loss of Cys63-FAD thiolate charge-transfer complex at 540 nm.
      Molecular Modeling of GoPI Bound to the Active Site of hGR—Because GoPI was shown to be a competitive inhibitor with respect to GSSG, we attempted to model the GoPI molecule into the GSSG binding site of hGR. The molecular models indicate that GoPI is small enough to bind to the GSSG-binding site of oxidized and reduced hGR. Because the side chains of Cys58, Cys63, and His467′ are interposed between the GSSG-binding site and the si side of the flavin ring, a direct interaction between the inhibitor and the flavin can be excluded. As it is not clear whether the dissociation of GoPI into Cl- and GoPI+ is likely to occur, both forms have been modeled into the active site of hGR(EH2) and hGR(Eox). In all models the phenyl ring of GoPI or GoPI+ can fit into a hydrophobic pocket, which is mainly composed of Tyr114 and Leu110 (Fig. 8).
      Figure thumbnail gr8
      Fig. 8Molecular models of hGR and GoPI. A, putative formation of the complex hGR(Eox)·GoPI. NH-ϵ of His467′ is associated with the chloride of GoPI and might act as an acid-base catalyst for the formation of HCl (compare with Refs.
      • Pai E.F.
      • Schulz G.E.
      and
      • Rietveld P.
      • Arscott L.D.
      • Berry A.
      • Scrutton N.S.
      • Deonarain M.P.
      • Perham R.N.
      • Williams Jr., C.H.
      ). B, putative bond between GoPI and His467′. C, putative bond between GoPI and Cys58 without loss of Cys63-FAD thiolate charge-transfer complex. D, putative bond between GoPI and Cys63 with loss of Cys63-FAD thiolate charge-transfer complex. A direct interaction between GoPI and FAD can be excluded. The pictures were generated using MOLSCRIPT and Raster 3D (
      • Kraulis P.J.
      ,
      • Merritt E.A.
      • Bacon D.J.
      ).
      The modeled complex of hGR(Eox)·GoPI+ shows a disulfide bridge between Cys58 and Cys63 and indicates that His467′ from the second subunit coordinates the gold atom of GoPI+ (Fig. 8B). In the case of hGR(Eox)·GoPI (Fig. 8A) the undissociated inhibitor has to be shifted away from the active site cysteines because of the chloride atom, compared with the reduced form of hGR (Fig. 8, C and D).
      In the case of hGR(EH2), it is possible to model GoPI+ covalently bound either to Cys58 or Cys63 (Fig. 8, C and D). Under the assumption that GoPI does not dissociate into Cl- and GoPI+ the putative structures of reduced hGR do not change because the chloride atom of GoPI is replaced after a nucleophilic attack by the Sγ-atom of Cys58 or Cys63.
      SELDI-TOF MS—Comparing the mass of hGR(Eox) and hGR pre-reduced with NADPH followed by incubation with 1 or 10 eq of GoPI reveals a significant mass shift for hGR treated with inhibitor (Fig. 9). The intact average mass of GoPI is 601 Da, whereas the dissociated complex GoPI+ has a mass of 565 Da. Measured mass shifts of reduced hGR treated with 1 eq of GoPI in comparison to untreated enzyme fit to the theoretical mass of one dissociated molecule of GoPI bound per monomer (measured average mass shifts of 499 ± 25 and 1105 ± 159 Da per monomer and dimer, respectively, in comparison to theoretical mass shifts of 565 and 1130 Da). Measured mass shifts of reduced hGR treated with 10 eq of GoPI in comparison to untreated enzyme are very similar to the theoretical mass of three dissociated molecules of GoPI bound per monomer (measured average mass shifts of 1712 ± 27 and 3266 ± 82 Da per monomer and dimer, respectively, in comparison to theoretical mass shifts of 1696 and 3392 Da). A control sample of hGR that was reduced with NADPH without further incubation with GoPI showed no mass shift in comparison to untreated enzyme, indicating that the presence or absence of NADPH does not influence the measured mass.
      Figure thumbnail gr9
      Fig. 9Mass spectrometry of hGR with and without GoPI. SELDI-TOF MS measurements of intact masses of 1.8 pmol of hGR(Eox) (upper spectrum) and 1.8 pmol of hGR pre-reduced with NADPH followed by incubation with 1 or 10 eq of GoPI (middle and lower spectrum). The mass shift is illustrated in the gel view: signal intensities are translated into a gray scale.
      Proliferation Assays—GoPI has an anti-proliferative effect on glioblastoma cells cultured in vitro. Concentrations of GoPI required to inhibit cell growth of glioblastoma cell lines (
      • Herold-Mende C.
      • Steiner H.H.
      • Andl T.
      • Riede D.
      • Buttler A.
      • Reisser C.
      • Fusenig N.E.
      • Mueller M.M.
      ) NCH37, NCH82, and NCH89 by 50% were 5.4 ± 0.7, 12.5 ± 0.8, and 10.8 ± 0.8 μm, respectively. Under the same conditions, 1,3-bis(2-chloroethyl)-1-nitrosourea (BCNU), a clinically used cytostatic agent active against brain tumors, exhibited 50% growth inhibition of NCH37, NCH82, and NCH89 glioblastoma cells at 300 ± 12, 385 ± 25, and 615 ± 38 μm BCNU, respectively.

      DISCUSSION

      Steady-state Kinetics—GoPI was shown to be an inhibitor competitive with GSSG and is characterized by a Kic of 0.46 μm (Fig. 2). In contrast, the measured I0.5 for the EH2 form of hGR can be much lower than the Kic. Thus, when using 1.4 nm NADPH-reduced hGR only ∼1 nm GoPI will cause 50% inactivation. It can be concluded that the competitive component of inhibition, which occurs rapidly, has a Kic of ∼0.46 μm. However, when EH2 is generated, an irreversible reaction with GoPI occurs, most likely by a covalent modification of the active site of hGR (see below). Reduction of hGR can be considered as a prerequisite of the irreversible inhibition (see also Ref.
      • Becker K.
      • Schirmer R.H.
      ). This irreversible reaction permits very low concentrations of GoPI to be effective over time in inhibiting the enzyme.
      Binding Stoichiometry of GoPI to hGR—GoPI binds tightly to the GSSG-binding site of hGR(Eox) with a stoichiometry of ∼1.25 eq per flavin (Fig. 3B) and is thus in the range of 1–1.5 eq estimated from the time-dependent inactivation by linear regression. Because of the small window of applicable concentrations for hGR and GoPI (Fig. 3A), and because of small absorbance changes for binding, the value for Kic = Kd = [hGR][GoPI]/[hGR·GoPI] = 0.46 μm determined from the steady-state kinetics could not be validated by titration.
      Using 9.6 μm hGR for the titration, the assumption, that the added inhibitor is tightly bound to the enzyme is justified by the characteristics of the spectral titrations, and determination of the number of binding sites is possible (Fig. 3). To correct the spectral anomalies that arise from the absorbance near 405 nm and the turbidity of GoPI, reference spectra were subtracted; this leads to the observation that the absorbance of GoPI in GR buffer is changed by the presence of oxidized (Fig. 3) or reduced enzyme (Fig. 4) in different ways. Corrected absorbance differences at selected wavelengths reveal a stoichiometry of ∼1.25 eq of GoPI per FAD-containing subunit of hGR(Eox) (Fig. 3B). Many hydrophobic inhibitors of hGR and P. falciparum GR bind to the intersubunit cavity of the dimer with a stoichiometry of 0.5 (
      • Sarma G.N.
      • Savvides S.N.
      • Becker K.
      • Schirmer M.
      • Schirmer R.H.
      • Karplus P.A.
      ). GoPI, however, binds to the active site of each subunit of the homodimer.
      GoPI Reacts with the Active Site of hGR—The differences between the spectra of oxidized (Fig. 3) and reduced enzyme (Fig. 4) titrated with GoPI could be explained by a covalent modification of a residue in the active site of the reduced enzyme (Figs. 8 and 10). A direct interaction between FAD and GoPI is not possible, as is shown by molecular modeling. However, reaction with Cys58 or Cys63 (Figs. 8, C and D, and 10A) and interaction with His467′ could occur (Figs. 8, A and B, and 10).
      Figure thumbnail gr10
      Fig. 10Proposed mechanism for the reaction of GoPI with hGR. GoPI is shown as R → Au-Cl. A, in the case of reduction of hGR, Cys63 (a) or Cys58 (b) is covalently modified, depending on the flow of electrons: pathway a, NADPH provides electrons from the re side of the flavin, followed by a reduction of Cys63 on the si side; pathway b, NaBH4 more likely provides electrons to Cys58 on the si side of the isoalloxazine ring. B, putative products of further consecutive reactions or side reactions formed during the nucleophilic attacks of Cys58, Cys63, or His467′.
      GoPI Reacts with Cys63 of hGR—The orientation of GoPI+ in the modeled structure of GoPI covalently bound to Cys63 (Fig. 8D) is very similar to the position of GoPI+ modeled in the oxidized enzyme (Fig. 8A). This would explain the observation that NADPH is still able to reduce the complex of hGR(Eox)·GoPI, resulting in GoPI becoming covalently linked to Cys63 without formation of Cys63-FAD CTC detected at 540 nm (Fig. 5C). In this case GoPI is already non-covalently bound to the enzyme before reduction, and therefore, the reaction with Cys63 right after the transfer of electrons from FADH could be kinetically and sterically favored (Fig. 10A, pathway a). The titration of hGR(Eox)·GoPI with NADPH, showing that NADPH is still consumed, is also in accordance with the steady-state kinetics, and does not indicate a competition for the substrate-binding site between NADPH and GoPI.
      GoPI Reacts with Cys58 of hGR—The thiolate of Cys58 is likely to react with GoPI, replacing a chloride anion liganded to the gold atom of GoPI in analogy to the physiological nucleophilic attack on GSSG (
      • Krauth-Siegel R.L.
      • Arscott L.D.
      • Schönleben-Janas A.
      • Schirmer R.H.
      • Williams Jr., C.H.
      ,
      • Pai E.F.
      • Schulz G.E.
      ). There appears to be little loss of Cys63 thiolate-FAD CTC at 540 nm as observed for the titration of hGR(EH2) with GoPI (Fig. 4A). The hypothesis of a reaction between Cys58 and GoPI is also supported by the reduction of the hGR(Eox)·GoPI complex with NaBH4, which is also coupled to the generation of the Cys63 thiolate-FAD CTC (Fig. 5B, curve 4). In this case, the flow of electrons could occur from the disulfide on the si side of the flavin, suggesting a kinetically favored nucleophilic attack from Cys58, replacing the chloride and generating in parallel the CTC (Fig. 10A, pathway b).
      Further Reactions Might Follow the Bond Formation Between GoPI and Cys58 of hGR—Addition of NADPH to enzyme having Cys58 reacted with GoPI (Fig. 10A, pathway b) might increase the electron density on Cys63, causing a nucleophilic attack of the Cys63-thiolate at the sulfur or the gold atom bound to Cys58. If the attack occurs at the sulfur of Cys58, a reduced form of GoPI is released. This might be an explanation for the partial recovery of enzymatic activity in the case of NaBH4-reduced GoPI after washing several times and testing the activity in an assay.
      If the attack of the Cys63-thiolate occurs at the gold atom, reduced Cys58 could be released (Fig. 10A, product of pathway a). Alternatively, the chromophore (R in Fig. 10) could be released and in this case, the gold atom would be coordinated by two cysteine residues (Fig. 10B, left side). Such a reaction coupled to the loss of the ligand has been proposed by Abrams and Murrer (
      • Abrams M.J.
      • Murrer B.A.
      ) to explain the mechanism of the gold-containing anti-arthritic prodrugs, auranofin, and analogs.
      Because AuI has a higher affinity for Cys-thiolates than for nitrogen-containing ligands, it is assumed a priori, that reactions between gold complexes and His residues could only occur after saturation of the Cys residues. However, the recently solved structure of Cyp-3 modified by the AuI phosphane Au(PEt)3Cl shows the gold atom bound to Nϵ-2 of His133 at the active site despite the presence of four Cys thiol groups (
      • Zou J.
      • Taylor P.
      • Dornan J.
      • Robinson S.P.
      • Walkinshaw M.D.
      • Sadler P.J.
      ). Thus, the possibility of a reaction between GoPI and His467′ should not be discarded (Fig. 10B, right side). The different spectra in Figs. 4A, 5B, and 7B, might also be explained by consecutive reactions, because of the different assay conditions (Fig. 10B).
      Rapid Reaction Kinetics of the Oxidative Half-reaction—The oxidative half-reaction of hGR(EH2) with GSSG is monophasic as described previously (
      • Krauth-Siegel R.L.
      • Arscott L.D.
      • Schönleben-Janas A.
      • Schirmer R.H.
      • Williams Jr., C.H.
      ). Our values for Kd(app) (63 μm) and k(obs)max (106 s-1) are ∼2-fold higher than those reported by Krauth-Siegel et al. (
      • Krauth-Siegel R.L.
      • Arscott L.D.
      • Schönleben-Janas A.
      • Schirmer R.H.
      • Williams Jr., C.H.
      ), because their assay was performed in the presence of NADPH as part of a single turnover, i.e. not under Vmax conditions. The oxidative half-reaction of hGR with GSSG is about 5- and 12-fold slower compared with the ortholog proteins in yeast and E. coli, respectively (see Table II in Ref.
      • Krauth-Siegel R.L.
      • Arscott L.D.
      • Schönleben-Janas A.
      • Schirmer R.H.
      • Williams Jr., C.H.
      ), and Kd(app) is more than 3-fold lower compared with P. falciparum GR (
      • Böhme C.C.
      • Arscott L.D.
      • Becker K.
      • Schirmer R.H.
      • Williams Jr., C.H.
      ).
      The hGR(EH2) reacts efficiently with GoPI. It was found that 0.9 eq of GoPI leads to an enzyme mixture consisting of modified and completely inactive enzyme molecules, and unmodified, fully active molecules. Inactive enzyme molecules therefore do not influence the kinetics of the oxidative half-reaction (Figs. 6 and 7, Table II). The existence of two enzyme populations, fully active and totally inactive enzyme molecules, is also known for the active-site inhibitor S-nitrosoglutathione (
      • Becker K.
      • Gui M.
      • Schirmer R.H.
      ). At a 2.4-fold molar excess of GoPI over hGR(EH2) almost all enzyme molecules are inactivated. Assuming that the inactivation is caused by a covalent bond between GoPI and Cys58, the thiolate, Cys63 might still provide some charge transfer character. Thus, some of the CTC seen in Fig. 7B could be because of modified enzyme.
      SELDI-TOF MS—SELDI-TOF measurements indicate a covalent modification of reduced hGR. Because the masses of untreated FAD-containing enzyme or holoenzyme pre-reduced with NADPH are similar to the theoretical mass of the hGR-apoenzyme, it can be assumed that non-covalently bound molecules are lost during the laser-induced ionization process. Thus, the detected mass shifts are caused by covalently bound GoPI. Incubation of reduced hGR with equimolar amounts of GoPI leads to a clear and reproducible mass shift of ∼1 molecule of GoPI per molecule of hGR. When GoPI was used in 10-fold excess, the mass shift indicated the binding of ∼3 molecules of GoPI per molecule of hGR. These results indicate that GoPI reacts not only with the active site as shown above, but also with two other residues, presumably accessible cysteines, at least after 30 min incubation at higher concentrations of enzyme and inhibitor used for sample preparation (85 μm hGR and 10-fold molar excess of GoPI in comparison with ∼15 μm hGR and 10-fold molar excess of GoPI used for the titrations). One of the accessible cysteine residues might be, for example, Cys3 at the flexible N terminus of hGR. Attempts to localize the modified residues in tryptic digests by matrix-assisted laser desorption ionization TOF, SELDI-TOF, and high performance liquid chromatography ESI-MS were not successful. Identified peptides covered up to 65% of the hGR amino acid sequence but did not include GoPI-modified peptides. This may be because of low ionization efficiency of these modified peptides or partially because of fragmentation of the peptide-bound inhibitor as seen in measurements of pure GoPI (data not shown).
      Comparison of GoPI with Known hGR Inhibitors—The irreversible inhibition of hGR because of a covalent modification of (an) active site residue(s) leads to a total loss of activity by GoPI. In contrast, exclusive nitration of Tyr106 and Tyr114 at the GSSG-binding site of hGR by peroxynitrite “only” leads to a 33-fold lower kcat, and a ∼103-fold lower catalytic efficiency (kcat/Km) (
      • Savvides S.N.
      • Scheiwein M.
      • Böhme C.C.
      • Arteel G.E.
      • Karplus P.A.
      • Becker K.
      • Schirmer R.H.
      ).
      S-Nitrosoglutathione (GSNO) is an NADPH-dependent inhibitor of hGR acting as reversible inhibitor that is competitive with GSSG (Ki ∼0.5 mm) and as an irreversible inhibitor (90% inhibition after 3 h incubation with 1 mm GSNO). The mechanism of inhibition involves modification (nitrosylation) of at least one catalytic site thiol. Only 30% of hGR(EH2) forms the CTC after GSNO treatment indicating a modification of Cys63 (
      • Becker K.
      • Gui M.
      • Schirmer R.H.
      ). In this context, it is interesting to mention that Cys63 in apo-(FAD-free)GR is not protected from modification by BCNU (see below) either. After reconstitution of GR with FAD and addition of NADPH there is no CTC formation because of the inability of modified Cys63 to form a thiolate (
      • Becker K.
      • Gui M.
      • Schirmer R.H.
      ). These results of BCNU- as well as GSNO-induced modification of GR seem to be analogous to those observed with GoPI-treated hGR(Eox) in the presence of NADPH. The weak competition of GSNO with GSSG and the fact that nanomolar concentrations of GR and 0.5 mm GSNO leads to 50% occupation of the catalytic sites (
      • Becker K.
      • Gui M.
      • Schirmer R.H.
      ) also reinforces the comparability between GoPI and GSNO. Nevertheless, GoPI is faster and much more effective as an inhibitor in vitro because of its action at nanomolar concentrations. Thus, to our knowledge GoPI is the best inhibitor modifying the active site of hGR to date.
      BCNU (or carmustine) is the best known inhibitor of hGR that also modifies Cys58 of hGR(EH2); however, hGR(Eox)isnot modified by BCNU (
      • Karplus P.A.
      • Krauth-Siegel R.L.
      • Schirmer R.H.
      • Schulz G.E.
      ). The Cys63 thiolate-FAD CTC can still be formed in the presence of the mutagenic and carcinogenic BCNU, and the carbamoylation process by the electrophilic BCNU product 2-chloroethylisocyanate is strongly dependent on pre-reduction of hGR (
      • Becker K.
      • Schirmer R.H.
      ). The inhibition of hGR by GoPI resembles in part the inhibition by BCNU, concerning the need for pre-reduction, target site, potential target residue(s), and CTC interaction. In contrast to BCNU, GoPI is much more efficient (nanomolar versus micromolar I0.5), and stable in solution; for comparison, the reactive decomposition product of BCNU 2-chloroethylisocyanate has a half-life in aqueous solution of less than 1 min. This makes GoPI a good candidate for cytotoxicity tests on tumor cells. Our first experiments in cell culture show that GoPI inhibits proliferation of glioblastoma cells at concentrations in the lower micromolar range, confirming this assumption. Thus, GoPI might be exploited for medical applications, for example, as a lead compound for the development of anti-tumor or antiparasitic drugs.

      Acknowledgments

      The technical assistance of Marina Fischer is greatly appreciated. We thank Sandro Ghisla for reading the manuscript and making suggestions.

      References

        • Becker K.
        • Rahlfs S.
        • Nickel C.
        • Schirmer R.H.
        Biol. Chem. 2003; 384: 551-566
        • Nordhoff A.
        • Bücheler U.S.
        • Werner D.
        • Schirmer R.H.
        Biochemistry. 1993; 32: 4060-4066
        • Krauth-Siegel R.L.
        • Arscott L.D.
        • Schönleben-Janas A.
        • Schirmer R.H.
        • Williams Jr., C.H.
        Biochemistry. 1998; 37: 13968-13977
        • Pai E.F.
        • Schulz G.E.
        J. Biol. Chem. 1983; 258: 1752-1757
        • Karplus P.A.
        • Schulz G.E.
        J. Mol. Biol. 1989; 210: 163-180
        • Savvides S.N.
        • Scheiwein M.
        • Böhme C.C.
        • Arteel G.E.
        • Karplus P.A.
        • Becker K.
        • Schirmer R.H.
        J. Biol. Chem. 2002; 277: 2779-2784
        • Rietveld P.
        • Arscott L.D.
        • Berry A.
        • Scrutton N.S.
        • Deonarain M.P.
        • Perham R.N.
        • Williams Jr., C.H.
        Biochemistry. 1994; 33: 13888-13895
        • Böhme C.C.
        • Arscott L.D.
        • Becker K.
        • Schirmer R.H.
        • Williams Jr., C.H.
        J. Biol. Chem. 2000; 275: 37317-37323
        • Sarma G.N.
        • Savvides S.N.
        • Becker K.
        • Schirmer M.
        • Schirmer R.H.
        • Karplus P.A.
        J. Mol. Biol. 2003; 328: 893-907
        • Becker K.
        • Herold-Mende C.
        • Park J.J.
        • Lowe G.
        • Schirmer R.H.
        J. Med. Chem. 2001; 44: 2784-2792
        • Davioud-Charvet E.
        • Delarue S.
        • Biot C.
        • Schwobel B.
        • Boehme C.C.
        • Mussigbrodt A.
        • Maes L.
        • Sergheraert C.
        • Grellier P.
        • Schirmer R.H.
        • Becker K.
        J. Med. Chem. 2001; 44: 4268-4276
        • Becker K.
        • Gui M.
        • Schirmer R.H.
        Eur. J. Biochem. 1995; 234: 472-480
        • Becker K.
        • Schirmer R.H.
        Methods Enzymol. 1995; 251: 173-188
        • Karplus P.A.
        • Krauth-Siegel R.L.
        • Schirmer R.H.
        • Schulz G.E.
        Eur. J. Biochem. 1988; 171: 193-198
        • Irmler A.
        • Bechthold E.
        • Davioud-Charvet E.
        • Hofman V.
        • Réau R.
        • Gromer S.
        • Schirmer R.H.
        • Becker K.
        Chapman S. Perham R. Scrutton N. Flavins and Flavoproteins2002. Agency for Scientific Publications, Berlin2002: 803-815
        • Hay C.
        • Hissler M.
        • Fischmeister C.
        • Rault-Berthelot J.
        • Toupet L.
        • Nyulaszi L.
        • Réau R.
        Chem. Eur. J. 2001; 7: 4222-4236
        • Shaw III, C.F.
        Chem. Rev. 1999; 99: 2589-2600
        • Fave C.
        • Cho T.Y.
        • Hissler M.
        • Chen C.W.
        • Luh T.Y.
        • Wu C.C.
        • Réau R.
        J. Am. Chem. Soc. 2003; 125: 9254-9255
        • Fave C.
        • Hissler M.
        • Kárpáti T.
        • Rault-Berthelot J.
        • Deborde V.
        • Toupet L.
        • Nyulászi L.
        • Réau R.
        J. Am. Chem. Soc. 2004; 126: 6058-6063
        • Hay C.
        • Le Vilain D.
        • Deborde V.
        • Toupet L.
        • Réau R.
        Chem. Commun. 1999; : 345-355
        • Kitz R.
        • Wilson I.B.
        J. Biol. Chem. 1962; 237: 3245-3249
        • Davioud-Charvet E.
        • McLeish M.J.
        • Veine D.M.
        • Giegel D.
        • Arscott L.D.
        • Andricopulo A.D.
        • Becker K.
        • Müller S.
        • Schirmer R.H.
        • Williams Jr., C.H.
        • Kenyon G.L.
        Biochemistry. 2003; 42: 13319-13330
        • Davis R.E.
        • Swain C.G.
        J. Am. Chem. Soc. 1960; 82: 5949-5950
        • Jones T.A.
        • Zou J.Y.
        • Cowan S.W.
        • Kjeldgaard M.
        Acta Crystallogr. Sect. A. 1991; 47: 110-119
        • Brünger A.T.
        • Adams P.D.
        • Clore G.M.
        • DeLano W.L.
        • Gros P.
        • Grosse-Kunstleve R.W.
        • Jiang J.S.
        • Kuszewski J.
        • Nilges M.
        • Pannu N.S.
        • Read R.J.
        • Rice L.M.
        • Simonson T.
        • Warren G.L.
        Acta Crystallogr. Sect. D. 1998; 54: 905-921
        • Herold-Mende C.
        • Steiner H.H.
        • Andl T.
        • Riede D.
        • Buttler A.
        • Reisser C.
        • Fusenig N.E.
        • Mueller M.M.
        Lab. Investig. 1999; 79: 1573-1582
        • Bisswanger H.
        Enzymkinetik. 3rd Ed. Wiley-VCH, Weinheim2000
        • Strickland S.
        • Palmer G.
        • Massey V.
        J. Biol. Chem. 1975; 250: 4048-4052
        • Abrams M.J.
        • Murrer B.A.
        Science. 1993; 261: 725-730
        • Zou J.
        • Taylor P.
        • Dornan J.
        • Robinson S.P.
        • Walkinshaw M.D.
        • Sadler P.J.
        Angew. Chem. Int. Ed. Engl. 2000; 39: 2931-2934
        • Kraulis P.J.
        J. Appl. Crystallogr. 1991; 24: 946-950
        • Merritt E.A.
        • Bacon D.J.
        Methods Enzymol. 1997; 277: 505-524