Essential Role of Selenium in the Catalytic Activities of Mammalian Thioredoxin Reductase Revealed by Characterization of Recombinant Enzymes with Selenocysteine Mutations*

Mammalian thioredoxin reductases (TrxR) are dimers homologous to glutathione reductase with a selenocysteine (SeCys) residue in the conserved C-terminal sequence -Gly-Cys-SeCys-Gly. We removed the selenocysteine insertion sequence in the rat gene, and we changed the SeCys 498 encoded by TGA to Cys or Ser by mutagenesis. The truncated protein having the C-termi-nal SeCys-Gly dipeptide deleted, expected in selenium deficiency, was also engineered. All three mutant enzymes were overexpressed in Escherichia coli and purified to homogeneity with 1 mol of FAD per monomeric subunit. Anaerobic titrations with NADPH rapidly gen-erated the A 540 nm absorbance resulting from the thio- late-flavin charge transfer complex characteristic of mammalian TrxR. However, only the SeCys 498 3 Cys enzyme showed catalytic activity in reduction of thioredoxin, with a 100-fold lower k cat and a 10-fold lower K m compared with the wild type rat enzyme. The pH optimum of the SeCys 498 3 Cys mutant enzyme was 9 as opposed to 7 for the wild type TrxR, strongly suggesting involvement of the low p K a SeCys selenol in the enzyme mechanism. Whereas H 2 O 2 was a substrate for the wild type enzyme, all mutant enzymes lacked hydroperoxi-dase activity. Thus selenium is required for the catalytic activities of TrxR Proteins were separated on SDS-polyacrylamide PhastGel with a gradient 8–25 and transferred to a nitrocellulose membrane and probed with rabbit antisera raised against rat TrxR (1: 8,000 dilution). For immunoblot detection, the membranes were incu- bated with alkaline phosphatase-labeled anti-rabbit IgG and developed with bromochloroindolyl phosphate/nitro blue tetrazolium substrates. Assays of Enzyme Activity— Enzyme activities of thioredoxin reductase were examined with a Shimadzu UV160U spectrophotometer us- ing previously developed methods (3, 4). Reduction of 5 m M DTNB was measured in 100 m M potassium phosphate buffer, 1 m M EDTA, pH 7.0, 0.1 m M NADPH, and 0.1 mg/ml bovine serum albumin. The reaction was followed by the increase of absorbance at 412 nm. Thioredoxin (5 m M ) reduction was followed by coupling to insulin (160 m M ) disulfide reduction (3, 27). For other substrates, the reaction mixtures contained 50 m M potassium phosphate, 2 m M EDTA, pH 7.0, and 0.2 m M NADPH. redoxin reductases as a function of pH. Activity was assayed by Trx-dependent reduction of insulin. The reaction mixture contained 100 m M potassium phosphate, 5 m M EDTA, 0.1 m M insulin, 0.2 m M NADPH, and 5 m M human Trx (C63S/C72S) in a volume of 120 m l. The reaction was started by addition of the enzyme. After incubation 10 min at 37 °C, the reaction was broken by 500 m l of 8 M guanidine hydrochloride, 1 m M DTNB in 50 m M Hepes buffer, pH 7.6. The activity was calculated from the net absorbance at 412 nm as the formation of SH groups in insulin. The 100 m M potassium phosphate was adjusted to the indicated pH values with NaOH. Activities are expressed here as percentage of the highest activity.

Thioredoxin reductase (TrxR) 1 is a dimeric ubiquitous enzyme that catalyzes reduction of the active site disulfide of thioredoxin by NADPH (1,2). Mammalian thioredoxin reductases have long been known to be very different from the well conserved enzymes present in bacteria, yeast, or plants (3,4). The mammalian enzymes are much larger with two subunits of M r 57,000 compared with 35,000 for the well characterized Escherichia coli enzyme. In particular, mammalian TrxR shows a wider substrate specificity reducing not only Trx from different species but also nondisulfide substrates such as vitamin K (3), alloxan (5), sodium selenite (6), selenocystine (7), selenodiglutathione (8), or S-nitrosoglutathione (9). Also arachidonic acid hydroperoxides such as hydroperoxyeicosatetraenoic acid are reduced to the corresponding alcohol hydroxyeicosatetraenoic acid demonstrating a lipid hydroperoxide reductase activity of the human and bovine enzymes (10).
Cloning of the genes for human (11), rat (12), and bovine TrxR 2 and the discovery of selenium in the enzyme isolated from human tumor cells (14,15) and bovine liver (12) demonstrated that mammalian TrxRs are homologous to glutathione reductase and have a C-terminal elongation containing a conserved selenocysteine residue in the penultimate position ( Fig.  1). The SeCys residue has been implicated in the enzyme mechanism since more than 4 electrons per subunit are required to reduce completely the FAD of the oxidized enzyme (16). Furthermore, the SeCys residue is alkylated with loss of activity only after reduction by NADPH (12,17,18), and free SeCys is released by carboxypeptidase digestion with loss of catalytic activity only from the NADPH-reduced enzyme (12). The SeCys residue is also the target of the irreversible inhibitor 1-chloro-2-nitrobenzene (19) as shown by peptide analyses (17).
The cDNAs for human (20) and rat (12) thioredoxin reductase contain an in frame TGA codon corresponding to the penultimate SeCys residue in the protein and a conserved stemloop structure folded as an SECIS motif about 200 base pairs downstream in the 3Ј-untranslated region (Fig. 1). The SECIS element is required for decoding the mRNA UGA codon, which otherwise confers termination, to incorporate selenocysteine by a species-specific mechanism (21)(22)(23). Attempts to express a putative human placenta thioredoxin reductase in E. coli (11) before it was discovered that the TGA encodes SeCys (20) rather than translation termination resulted in a protein with an identical subunit size as the native enzyme (55 kDa) on SDS-PAGE but inactive since it lacked FAD (11). Truncation by UGA acting as a stop codon, therefore, suggested that the C-terminal SeCys-Gly dipeptide may have a function in protein folding or FAD binding (24).
In this paper we have used site-directed mutagenesis of the rat enzyme to examine the role of the selenocysteine residue. We have replaced the SeCys by either the chemically similar Cys or the redox-inactive Ser, and we also engineered the truncated protein DesSeCys 498 Gly 499 resulting from the UGA acting as a stop codon. We have purified all three mutant proteins to homogeneity in high yield from E. coli. Physicochemical and enzymatic analysis demonstrated that only the SeCys 498 3 Cys enzyme showed activity in reduction of thioredoxin with a major loss of k cat . Native rat TrxR reduced hydrogen peroxide with a high K m value for H 2 O 2 , but this activity was absent in the Cys mutant. However, addition of thiore-doxin and free selenocystine strongly stimulated activity lowering the K m value for H 2 O 2 dramatically. Our results demonstrate that selenium is essential for the catalytic activities of TrxR and directly involved in the mechanism of the enzyme. A preliminary report of this work has been published (25).
Expression of Recombinant Rat Thioredoxin Reductase in COS-7 Cells-The desired region of 1.8 kilobase pairs (Fig. 1b) starting at nt 113 of the rat TrxR cDNA previously cloned and sequenced (12) was amplified by PCR using 5Ј-TCG AAA GCT AGC AAT GAA TGA-3Ј as the forward primer and 5Ј-AAC AAG ATC CAC ATT ACA TAG CTT GAA GGC-3Ј as the reverse primer. The translation start methionine codon is shown in boldface. The primers contained NheI or BamHI restriction site, respectively, to facilitate subcloning of the amplified fragment into a pcDNA3.1/zeo(Ϫ) vector at the corresponding sites. The construct was transferred into COS-7 cells by using LipofectAMINE TM reagent for transient expression. In 75-cm 2 plastic flasks, 2.5 ϫ 10 6 cells were plated and incubated in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum and 1ϫ penicillin/streptomycin (complete medium) at 37°C. When cells were ϳ70% confluent they were incubated with the transfection mixture composed of 5 g of DNA, 12 l of LipofectAMINE TM reagent (2 mg/ml), and serum-free medium. After 5 h, the mixture was replaced with complete medium with or without 1 M sodium selenite as indicated (Fig. 2) and incubated at 37°C for another 35 h before harvesting cells.

Construction of TrxR Mutant Proteins-
The open reading frame of the rat TrxR cDNA from nt 123 to 1649 was amplified by PCR. The sense primer was 5Ј-TCA ACC ATG GAT GAC TCT AAA GAT GCC CCT-3Ј (methionine start codon in bold type). The antisense mutant primer 1 was 5Ј-CAA CAG GAT CCA CAC TGG GGC TTA ACC GCA GCA GC-3Ј; the antisense mutant primer 2 was 5Ј-CAA CAG GAT CCA CAC TGG GGC TTA ACC TGA GCA GC-3Ј, and the antisense mutant primer 3 was 5Ј-CAA CAG GAT CCA CAC TGG GGC TTA ACC TTA GCA GC-3Ј. The mutated sites are underlined. The corresponding Se-Cys codon (TGA) was thus altered to Cys (TGC), Ser (TCA), or stop (TAA) codon, respectively (Fig. 1c). The amplified PCR products were subcloned into pGEM-T vector and sequenced to confirm the expected mutagenesis by an ALF sequencer (Amersham Pharmacia Biotech) as described (12). Note that resequencing the original cDNA corrected the sequence by 3 additional nt, so that position 1646 in the old sequence now corresponds to 1649 (see below).
Overexpression of TrxR Mutant Proteins in E. coli-The cDNA insert was excised from the pGEM-T vector at the introduced NcoI and BamHI site and ligated into a pET-3d vector. The resulting plasmids were used to transform E. coli BL21(DE3)pLysS cells. Cells were grown in LB medium containing carbenicillin (100 g/ml) and chloramphenicol (34 g/ml) at 37°C and induced with 0.4 mM isopropyl-1-thio-␤-D-galactopyranoside at ϳ0.6 A 600 nm for 4 h. The cells were harvested by centrifugation and washed with 50 mM phosphate buffer, pH 7.5, 2 mM EDTA, and 1 mM PMSF and stored as cell pellets at Ϫ80°C until used.
Purification of Recombinant TrxR Mutant Proteins-E. coli cells were resuspended in 5 volumes of cold 50 mM potassium phosphate, 2 mM EDTA, pH 7.5, 1 mM PMSF, and 0.1% Triton X-100 and sonicated. Following centrifugation for 30 min at 12,000 rpm at 4°C in a Sorvall RC5 centrifuge, the supernatant fraction was loaded on a column of DEAE-Sephacel (2.5 ϫ 15 cm). TrxR was eluted with a linear gradient from 0 to 0.5 M NaCl in 50 mM potassium phosphate, pH 7.5 and 1 mM EDTA (1000 ml each). Fractions with DTNB reducing activity were pooled and dialyzed against 50 mM potassium phosphate, pH 7.5, 1 mM EDTA and applied to a column of 2Ј,5Ј-ADP-Sepharose (1.5 ϫ 7 cm) equilibrated with this buffer. The bound enzyme was eluted with a linear gradient of 0.05-0.3 M potassium phosphate buffer, pH 7.5, containing 1 mM EDTA (400 ml each). Fractions containing pure mutant TrxR were dialyzed against 50 mM potassium phosphate, pH 7.5, 1 mM EDTA and stored frozen at Ϫ80°C.
Western Blotting-Proteins were separated on SDS-polyacrylamide PhastGel with a gradient 8 -25 and transferred to a nitrocellulose membrane and probed with rabbit antisera raised against rat TrxR (1: 8,000 dilution). For immunoblot detection, the membranes were incubated with alkaline phosphatase-labeled anti-rabbit IgG and developed with bromochloroindolyl phosphate/nitro blue tetrazolium substrates.
Assays of Enzyme Activity-Enzyme activities of thioredoxin reductase were examined with a Shimadzu UV160U spectrophotometer using previously developed methods (3,4). Reduction of 5 mM DTNB was measured in 100 mM potassium phosphate buffer, 1 mM EDTA, pH 7.0, 0.1 mM NADPH, and 0.1 mg/ml bovine serum albumin. The reaction was followed by the increase of absorbance at 412 nm. Thioredoxin (5 M) reduction was followed by coupling to insulin (160 M) disulfide reduction (3,27). For other substrates, the reaction mixtures contained 50 mM potassium phosphate, 2 mM EDTA, pH 7.0, and 0.2 mM NADPH.
FIG. 1. a, schematic protein structure of rat thioredoxin reductase based on its homology to glutathione reductase (12). b, native rat TrxR cDNA with TGA codon for selenocysteine and selenocysteine insertion sequence (SECIS) element in the 3Ј-untranslated region (3Ј-UTR). c, engineered cDNA for expression of seleniumdeficient TrxR mutants in E. coli. Structures are not drawn to scale.
For H 2 O 2 reduction only the initial velocity during the 1st min was used for calculation of data. Reactions were initiated by adding enzyme to the sample cuvette and the same amount of buffer to the blank cuvette. Reaction rates were followed by the decrease in absorbance at 340 nm, resulting from oxidation of NADPH. The activity was calculated using a molar extinction coefficient of 6,200 M Ϫ1 cm Ϫ1 for NADPH at 340 nm. Apparent K m values were calculated from Lineweaver-Burk plots of Protein Analyses-Thioredoxin reductase concentration was determined either by measuring the absorbance of flavin at 460 nm using 11.3 mM Ϫ1 cm Ϫ1 or using the absorbances at 280 nm and subtracting the absorbance at 310 nm, using a molar extinction coefficient of 100,900 M Ϫ1 cm Ϫ1 . Protein sequence analysis employed generation of tryptic peptides which were purified by high pressure liquid chromatography on a RPC C2/C18 SC 2.1/10 column and sequenced on a PROCISE protein sequencer (12). Absorbance spectra were recorded on a Shimadzu UV 160U UV-visible spectrophotometer.
Anaerobic Titration with NADPH-Solutions of thioredoxin reductase were placed in cuvettes covered with a rubber septum. The NADPH solution was placed in a separate glass vial covered with a rubber septum. The content was bubbled with argon, purified through an O 2 scrubber consisting of 2% sodium dithionite and 0.1 M NaOH. The argon was introduced through a needle penetrating the rubber septum, and through another needle air was evacuated. This anaerobic treatment lasted for 30 min. To the TrxR samples aliquots from the NADPH solution were added via a gas-tight Hamilton syringe. Spectra were recorded after each addition using the spectrophotometer at 25°C.

Expression of Active Rat Thioredoxin Reductase in COS-7
Cells-We previously cloned and sequenced a 2.2-kilobase rat TrxR cDNA containing the open reading frame and a 3Ј-untranslated region of 560 nucleotides containing a SECIS element (12). To show that our rat TrxR cDNA was functional, it was cloned in the mammalian expression vector pcDNA 3.1/ zeo(Ϫ). In transiently transfected COS-7 cells, the total TrxR activity was increased 70% compared with the control cells transfected with the empty vector. By Western blotting, extracts of the cells transfected with rat TrxR expression plasmid showed a new positive band (Fig. 2), which corresponded to TrxR at 55 kDa, plus another positive band with higher molecular weight of unknown origin. This experiment proved that the rat TrxR cDNA directed synthesis of full-length and active enzyme. However, only trace amounts of rat enzyme in a background of COS-7 cell TrxR was obtained from this expression system, which precludes analysis of the effects of mutation by enzyme kinetics and other methods requiring pure proteins in milligram quantities.
Expression and Purification of Mutant Thioredoxin Reductase in E. coli-When the 2,156-nucleotide fragment from the rat TrxR cDNA was cloned in an expression plasmid and trans-formed E. coli cells were induced by isopropyl-1-thio-␤-D-galactopyranoside, a strong 20-kDa protein, which was positive in Western blots using an antibody to rat thioredoxin reductase, was generated. The 20-kDa protein was obviously inactive and may have arisen from the folding of the mRNA (12) which is incompatible with the E. coli translation system causing premature termination (21)(22)(23). We reasoned that shortening the cDNA maximally by eliminating the SECIS element and modifying the TGA codon of the gene might lead to overexpression of full-length TrxR mutant proteins in E. coli. As outlined in Fig. 1c and "Experimental Procedures," the TGA was modified to either TGC (cysteine), TCA (serine), or a TAA (termination) codon. Our assumption was correct since high level expression of the three mutant proteins was obtained. Each of the three mutant enzymes were purified by affinity chromatography on a column of 2Ј,5Ј-ADP-Sepharose demonstrating that they were folded and had a functional NADPH-binding site. From a 1-liter culture typically 10 -20 mg of pure enzyme was obtained. The enzymes were homogeneous, with a subunit of M r 57,000 as judged from SDS-PAGE, and all showed a strong positive reaction in Western blotting (Fig. 3). The primary structures of each mutant protein was confirmed by sequencing a number of peptides including the C-terminal tryptic peptide (Fig. 4). In addition a peptide with the internal sequence 450 QGFAAAL 456 was isolated, which corrects the previously reported sequence 450 QALQPL 455 of the rat cDNA clone (12) and demonstrates total identity with the human TrxR as well as a close homology to that of glutathione reductase (Fig. 4). The sequencing results also showed a Trp residue at position 53 rather than Gly.
Spectral Properties of Mutant Enzymes-All mutant enzymes were yellow and had the same visible absorption spectrum and FAD content as wild type selenium-containing rat or calf liver thioredoxin reductase (Fig. 5 and Table I). Even the truncated protein contained FAD, excluding a function of the SeCys residue in binding this cofactor as had been suggested (24). All enzymes rapidly reacted with NADPH to give rise to the characteristic long wavelength band (Fig. 6) arising from the charge transfer complex between FAD and an redox-active thiolate anion (16). This demonstrates that the mutant enzymes catalyzed the intramolecular electron transfer from bound NADPH via FAD to the redox-active disulfide identical in thioredoxin reductase and glutathione reductase (Fig. 4) as described previously for human placenta TrxR (16).
Catalytic Activities of Mutant Enzymes-The availability of milligram quantities of the pure mutant proteins allows us to examine their catalytic properties with different substrates for mammalian thioredoxin reductase. As shown in Table II, only the SeCys 498 3 Cys TrxR showed activity with Trx-S 2 in reducing insulin disulfides, whereas the other mutant proteins were inactive. DTNB which is used as a substrate to assay the enzyme activity (3, 4) showed 4.7% activity with the SeCys 498 3 Cys enzyme compared with calf liver TrxR and with a similar K m value (data not shown). The Cys mutant enzyme showed 9.1, 2.7, and 2.1% activity of wild type TrxR in selenite, lipoic acid, or selenocystine reduction reactions, respectively. The other two mutants had very low activity with all these substrates. Despite the presence of an N-terminal redox-active disulfide in TrxR identical to that of glutathione reductase (Fig.  4), neither the native enzyme nor any of the mutant proteins reduced GSSG (data not shown). 498 3 Cys TrxR and human erythrocyte glutathione reductase (GR) based on primary and secondary structure. 1st line is secondary structure of the rat TrxR predicted using the program "nnpredict"; 2nd line is the rat TrxR sequence deduced from cDNA sequence; 3rd line is human glutathione reductase sequence, and 4th line is secondary structure of human glutathione reductase taken from Ref. 13. The nomenclature is used as follows: (e) ␤-sheet and (H) ␣-helix. The three functional regions FAD-binding motif, redox-active disulfide, and NADP(H)-binding motif are boxed. The amino acid sequences determined by Edman degradation are underlined. The C-terminal extension of the rat TrxR includes a substituted Cys 498 for SeCys. This C-terminal cysteine pair is accessible and is not linked by disulfide in the native enzyme since both the cysteine residues were alkylated by 4-vinylpyridine and released as separate phenylthiohydantoin-cysteine from nonreduced C-terminal peptide, as shown in the lower panel.

Comparison of the Rat Liver and the Rat SeCys 498 3 Cys Enzyme in Reduction of Human
Thioredoxin-Since the mutant enzyme with the SeCys to Cys replacement was active, we made a detailed comparison with the wild type enzyme prepared from rat liver. As shown in Fig. 7 the apparent K m and k cat values for the wild type enzyme at pH 7.5 were 3.3 M and 2,500 ϫ min Ϫ1 or close to previously reported values (3). In contrast the apparent K m of the SeCys 498 3 Cys mutant was lower than that of the wild type or 0.4 M, whereas the k cat was 14.3 ϫ min Ϫ1 or 0.6% of the selenocysteine containing wild type enzyme.
The activities in thioredoxin-dependent reduction of insulin catalyzed by the wild type and SeCys 498 3 Cys TrxR were determined over the pH range 4.2 to 10.5. The SeCys 498 3 Cys enzyme exhibited a broad pH optimum of 9 (Fig. 8). The sharp drop in activity of the mutant enzyme at high pH may reflect enzyme denaturation. The overall profile was similar to that of wild type TrxR which, however, exhibited a pH optimum of 7 (Fig. 8). This shift in activity to the alkaline side should be related to the replacement of the selenocysteine with a cysteine in a putative active site. At a physiological pH of 7.0, the selenol (nominal pK a of 5.3) should be fully ionized, and the replacement by a thiol (nominal pK a of 8.25) should be partially protonated, suggesting that the ionic state of the SeCys 498 is directly involved in the rate-determining step of catalysis. The results provide a strong argument of the involvement of the selenocysteine residue in the mechanism of electron transfer.
Activity with Hydrogen Peroxide-We examined the activity of rat and human placenta thioredoxin reductase with H 2 O 2 , and the enzymes directly reduced this peroxide (Fig. 9). To measure activity, only the initial rate during the 1st min was used. Assuming Michaelis-Menten kinetic rather than a second order reaction, the apparent K m value for H 2 O 2 was 2.5 mM and the k cat was 100 ϫ min Ϫ1 . In contrast, none of the three mutant proteins showed any peroxide reductase activity (Ͻ1% of the wild type enzyme). As shown in Fig. 9 the SeCys 498 3 Cys mutant enzyme activity with H 2 O 2 was not stimulated by addition of thioredoxin. However, addition of 4.5 M selenocystine together with 5 M Trx strongly stimulated the H 2 O 2 reducing activity. The activity increased to around 20% that of the natural enzyme. Also, most important, the apparent K m value for H 2 O 2 was reduced from 2.5 to 0.1 mM. This effect of free selenocyst(e)ine acting to catalyze the reaction via an unknown intermediate was previously seen with the human enzyme alone (6,10). In this case the SeCys 498 3 Cys enzyme the effect was evident only with thioredoxin present.   (1) and (2)

DISCUSSION
Recent studies have demonstrated multiple isoenzymes of thioredoxin reductase in mammalian tissues including three forms in the cytosol (28). In addition mitochondrial thioredoxin reductase (29 -31) is present as three isoforms probably by differential splicing (32). All these isoforms share the conserved C-terminal sequence -Gly-Cys-SeCys-Gly and the basic glutathione reductase-like structure outlined in Fig. 4, whereas they are different in the N-terminal regions.
In this study we have examined the result of replacing the penultimate selenocysteine residue in rat thioredoxin reductase (12) by cysteine or serine or removing it together with the C-terminal glycine residue. Our results show that folded FAD containing mutant enzymes in high yield were obtained by expression in E. coli. Titration with NADPH under anaerobic conditions demonstrated the appearance of a lower absorbance in the 460-nm region and a new long wave band at 540 nm in all three mutant enzymes. This is consistent with formation of a thiolate-flavin charge transfer complex originally observable but not interpreted in rat liver thioredoxin reductase (3) and extensively studied in the human placenta enzyme (16). The thiolate-flavin charge transfer arises from the flavin and the N-terminal redox-active disulfide which is fully conserved between glutathione reductase and thioredoxin reductase (11-13) (Fig. 4). The result is consistent with an intact first halfreaction for all mutant proteins involving transfer of electrons from NADPH to the N-terminal redox-active disulfide. The results also rule out a role for the C-terminal SeCys-Gly dipeptide in FAD binding or interaction with this coenzyme.
In contrast to the similar behavior of all three mutant enzymes in NADPH titrations, only the SeCys 498 3 Cys mutant protein showed activity in reduction of thioredoxin. The results of the kinetic analysis showed a major 100-fold decrease in k cat with a shift of the pH optimum of 2 pH units. This is a strong argument in favor of the SeCys residue being directly involved in the active site mechanism of reduction of the active site disulfide (Cys-Gly-Pro-Cys) in thioredoxin (33). Electrons to reduce the oxidized SeCys as a selenenylsulfide will come from the reduced N-terminal disulfide in the other subunit of the  dimer arranged in a head to tail structure as in glutathione reductase 3 (1) (Fig. 4). The decrease in k cat of substituting SeCys by Cys by 2 orders of magnitude is similar to results for E. coli formate dehydrogenase H where this substitution also resulted in more than 2 orders of magnitude reduction in catalytic activity (34). Also for rat type 1 iodothyronine deiodinase, a similar effect on catalytic activity by a SeCys to Cys substitution was reported (35). This lack of activity in the Ser mutant or DesSeCys 498 -Gly 499 enzyme is consistent with a catalytic role of the SeCys residue as an electron acceptor and donor. Recently results from a human thioredoxin reductase Cys mutant expressed in baculovirus was published (36). Although the K m value for thioredoxin was reported to be higher for a Cys mutant than the wild type placenta enzyme and was reported as mM rather than M probably due to a printing error, the large effect on k cat we have observed was also seen in this system. The k cat value for the wild type enzyme was low, however (36). We have recently demonstrated that both the SeCys residue and adjacent Cys residue are alkylated by the irreversible inhibitor 1-chloro-2-nitrobenzene (17). Furthermore, we have unpublished data demonstrating a selenenylsulfide linking these two residues in the wild type-oxidized enzyme. 3 This second C-terminal redox center is proposed to be formed in oxidation of the enzyme by thioredoxin. An additional role of the selenocysteine residue would then be struc-tural since bonds between two adjacent Cys or SeCys residues are rare and not favored. The larger radius of the selenium atom in the SeCys residue may then be important. It is of particular significance in this study that both the Cys 497 and SeCys 498 residues were alkylated by vinylpyridine and detected as free phenylthiohydantoin-derivatives in the Edman degradation (Fig. 4). This is consistent with the presence of two thiol groups left unoxidized to a disulfide in the SeCys 498 3 Cys enzyme after purification. 3 All three mutant proteins were obtained in high yield in E. coli. We have recently crystallized the active rat SeCys 498 3 Cys thioredoxin reductase and obtained crystals that diffract to better than 3-Å resolution, 4 and the structure of the mutant enzyme should soon be available. Preparations of the native selenocysteine-containing wild type enzyme yield enzyme with varying specific activities (4,18,36) probably as a result of varying degrees of loss of selenium (28,36,37). Recently, use of E. coli and gene fusion with engineered bacterial-type SECIS elements and coexpression with selA, selB, and selC genes have resulted in production of large quantities of selenocysteinecontaining wild type enzyme with about 25% of the specific activity of the pure enzyme (38). The main protein in this preparation that remains to be separated (38) is the inactive truncated DesSeCys 498 Gly 499 enzyme we have studied.
The content of a catalytically active selenocysteine residue in mammalian thioredoxin reductase explains the previously surprising observation of the direct reduction of lipid hydroperoxides (10). In this paper we have also demonstrated that H 2 O 2 is a substrate for human placenta thioredoxin reductase with a k cat of 100 ϫ min Ϫ1 , which is about 3% that for the natural substrate thioredoxin (4). However, the K m for H 2 O 2 was relatively high or 2.5 mM, which may mean that the enzyme under normal conditions is of relatively minor importance for removal of H 2 O 2 compared with enzymes such as glutathione peroxidases or thioredoxin peroxidases (peroxiredoxins) (39). The hydroperoxidase activity may serve to protect the enzyme from self-inactivation by hydroxyl radical ions formed by oxygen univalent reduction. 5 An hypothesis involving the SeCys residue for regulation of activity has also recently been proposed (28). Our results clearly show that the selenocysteine is required for the activity since the SeCys 498 Ser mutant as well as the DesSeCys 498 Gly 499 enzymes were inactive.
The Cys mutant enzyme showed undetectable activity with H 2 O 2 by itself. However, addition of selenocystine, the diselenide amino acid, and thioredoxin resulted in considerable activity in H 2 O 2 reduction (Fig. 9). The apparent K m value for H 2 O 2 was also much lower and potentially in a physiological range as a local concentration. The same stimulatory effect of selenocystine as a charge transfer catalyst after reduction to the selenol selenocysteine was previously observed with the wild type enzyme in the absence of thioredoxin in reduction of lipid hydroperoxides (10). Since cells do not contain a free pool of selenocysteine (23) this reaction is probably of no physiological significance. In fact with selenocyst(e)ine present, it is known that thioredoxin reductase and thioredoxin will give extensive redox cycling with oxygen (7) as is also the case for selenite (6) or selenodiglutathione (8) at (M) concentrations. Thus, thioredoxin reductase with its selenocysteine residue is a major reason for the inability of cells to handle free selenocysteine; the other is obviously random incorporation of selenocysteine in place of cysteine (23).
The essential selenocysteine residue in thioredoxin reductase is almost certainly the target of therapeutic gold com- pounds, such as gold thioglucose or auronofin which are powerful inhibitors of enzyme in vitro (40,41) and in vivo (42). Also a number of chemotherapeutic drugs are also targeted to the enzyme (43). The absolute requirement of selenocysteine in the function of the mammalian thioredoxin system strongly suggests a mechanism for the essential role of this trace element in cell growth. It is well known that selenite is required for the growth of cells in tissue culture (44,45). Furthermore, selenium has anticarcinogenic effects (46 -48). The selenoprotein thioredoxin reductase is a cornerstone of cellular antioxidant defense and regulation of cell growth and differentiation through effects via thioredoxin. It is involved in synthesis of deoxyribonucleotides for DNA synthesis via the role of reduced thioredoxin as an electron donor for essential enzyme ribonucleotide reductase (2,33). Selenite reduction to selenide is catalyzed by thioredoxin reductase (6), and the truncated protein expected in selenium deficiency is inactive in reduction of thioredoxin; it is yet unknown if this has any other biological role or is present in selenium-starved cells.